Data generating processes for trip-level commercial fishing data feature endogenous targeting of individual fish species and spatially, temporally varying stock conditions that are unobserved by the researcher but partially observed by fishermen. We present a model and identification strategy to address serious challenges that arise in this setting for obtaining consistent estimates of the structural properties of fishing technologies. Our estimation strategy exploits (1) the timing and information available to fishermen when factor inputs and output targeting decisions are made, and (2) exogenous, natural variation in stock abundance at the trip-level spatial scale. In a first stage, estimation methods used by stock assessment scientists are adopted to account for endogenous fishing power and spatially-temporally varying abundance. Parameters of our multiple-species (trip-level) cost function are then estimated through a decomposition of costs into an endogenous targeting component and a component that varies exogenously due to random fluctuations in the marine environment. Nonparametric series estimation methods are used to control for endogenous targeting and unobserved multiple-species abundance. An application to the Gulf of Mexico commercial reef fish fishery is presented to illustrate the model. The approach solves an identification problem that pervades previous empirical analysis of fishing technologies and offers a new tool for managing spatial-temporal fishing.