TY - JOUR T1 - Bemovi, software for extracting Behaviour and Morphology from Videos JF - bioRxiv DO - 10.1101/011072 SP - 011072 AU - Frank Pennekamp AU - Nicolas Schtickzelle AU - Owen L. Petchey Y1 - 2014/01/01 UR - http://biorxiv.org/content/early/2014/11/07/011072.abstract N2 - Microbes are critical components of ecosystems and vital to the services they provide. The essential role of microbes is due to high levels of functional diversity, which are, however, not always mirrored in morphological differentiation hampering their taxonomic identification. In addition, the small size of microbes hinders the measurement of morphological and behavioural traits at the individual level, as well as interactions between individuals.Recent advances in microbial community genetics and genomics, flow cytometry and digital image analysis are promising approaches, however they miss out on a very important aspect of populations and communities: the behaviour of individuals. Video analysis complements these methods by providing in addition to abundance and trait measurements, detailed behavioural information, capturing dynamic processes such as movement, and hence has the potential to describe the interactions between individuals.We introduce bemovi, a package using R - the statistical computing environment - and the free image analysis software ImageJ. Bemovi is an automated digital video processing and analysis work flow to extract abundance and morphological and movement data for numerous individuals on a video, hence characterizing a population or community by multiple traits. Through a set of functions, bemovi identifies individuals present in a video and reconstruct their movement trajectories through space and time, merges measurements from all treated videos into a single database to which information on experimental conditions is added, readily available for further analysis in R.We illustrate the validity, precision and accuracy of the method for experimental multi-species communities of protists in aquatic microcosms. We show the high correspondence between manual and automatic counts of individuals and illustrate how simultaneous time series of abundance, morphology and behaviour are constructed. We demonstrate how the data from videos can be used in combination with supervised machine learning algorithms to automatically classify individuals according to the species they belong to, and that information on movement behaviour can substantially improve the predictive ability and helps to distinguish morphologically similar species. In principle, bemovi should be able to extract from videos information about other types of organism, including “microbes”, so long as the individuals move relatively fast compared to their background. ER -