Prey and predator overlap at the edge of a mesoscale eddy: fine-scale, in-situ distributions to inform our understanding of oceanographic processes

Big data as generated from environmental modelling, genomics, and novel instrumentation necessitates analytical techniques that are beyond our typical statistical tools. Machine learning can reach broadly across these applications, as evidenced by its rapid rise across multiple disciplines and industries. The availability of machine learning allows for unique pattern/relationship seeking in model outputs, as well as for the rapid resolution of complex instrument outputs such as those from imaging systems or other observational sensors. Here we used a convolutional neural network (CNN) to automate the analysis of over a 100 million images generated by the In situ Ichthyoplankton Imaging system (ISIIS) during a study that adaptively sampled a mesoscale eddy (ME) in the Straits of Florida. Eddies are ubiquitous features that affect primary- and secondary production as well as population connectivity. The aim of the study was to investigate the physical and biological drivers of zoo- and ichthyoplankton aggregations at the eddy edge, a location of high interest due to frontal processes in this area. Undulating ISIIS tows generated fine-scale distribution data for plankters and revealed that only four days before the ME dissipated, larval fish and Oithona spp. copepod concentrations within the eddy were significantly higher than outside of the eddy in the Florida Current. While larval fishes are known predators of Oithona and these coupled distributions suggest potential predator-prey interactions, Random Forests models indicated that distributions of both taxa were primarily driven by variables such as current speed and direction that signify the physical footprint of the ME. These results suggest that eddy-related advection leads to largely passive overlap between predator and prey, a positive, energy-efficient outcome for predators at the expense of prey.