Identifying complex behaviors such as spawning and fine-scale activity is extremely challenging in highly migratory fish species and is becoming increasingly critical knowledge for fisheries management in a warming ocean. Habitat use and migratory pathways have been extensively studied in marine animals using pop-up satellite archival tags (PSATs), but high-frequency data collected on the reproductive and swimming behaviors of marine fishes has been limited by the inability to remotely transmit these large datasets. Here, we present the first application of remotely transmitted acceleration data to predict spawning and discover drivers of high activity in a wild and highly migratory pelagic fish, the mahi-mahi (Coryphaena hippurus). Spawning events were predicted to occur at nighttime, at a depth distinct from non-spawning periods, primarily between 27.5 and 30°C, and chiefly at the new moon phase in the lunar cycle. Moreover, throughout their large-scale migrations, mahi-mahi exhibited behavioral thermoregulation to remain largely between 27 and 28°C and reduced their relative activity at higher temperatures. These results show that unveiling fine-scale activity patterns are necessary to grasp the ecology of highly mobile species. Further, our study demonstrates that critical, and new, ecological information can be extracted from PSATs, greatly expanding their potential to study the reproductive behavior and population connectivity in highly migratory fishes.