Developing predictive models for aquatic microbial populations requires very fine spatial and temporal resolution of data that traditional monitoring techniques are typically incapable of providing. The need for continuous presence in an environment combined with the desire for directed (intelligent) sampling has prompted the development of a sensor network to address these needs. The network incorporates low-energy demand, and highly adaptable sensors which exploit recent advances in computer networking and robotics to process sensor data and ensure high data fidelity. The coordination of stationary sensor nodes and mobile sensing using a sampling robot allow for efficient collection of samples from features of interest, exemplified in studies of a recent cyanobacterial bloom in Lake Fulmor, California.