Bacterial community segmentation facilitates the prediction of ecosystem function along the coast of the western Antarctic Peninsula

Bacterial community structure can be combined with observations of ecophysiological data to build predictive models of microbial ecosystem function. These models are useful for understanding how function might change in response to a changing environment. Here we use five spring-summer seasons of bacterial community structure and flow cytometry data from a productive coastal site along the western Antarctic Peninsula to construct models of bacterial production (BP), an ecosystem function that heterotrophic bacteria provide. Through a novel application of emergent self-organizing maps we identified eight recurrent modes in the structure of the bacterial community. A model that combined bacterial abundance, mode and the fraction of cells belonging to the high nucleic acid population (fHNA; R-2 = 0.730, P<0.001) best described BP. Abrupt transitions between modes during the 2013-2014 spring-summer season corresponded to rapid shifts in fHNA. We conclude that parameterizing community structure data via segmentation can yield useful insights into microbial ecosystem function and ecosystem processes.