Fish Acoustic Detection Algorithm Research (FADAR): A powerful tool to explore the role of call types in fish spawning behavior and trend in spawning population structure

FADAR is a user-friendly deep learning algorithm specifically designed to identify and classify the call types of four Caribbean grouper species. FADAR is capable of processing ten thousand 20-s audio files in less than 2.5 h with an average accuracy of 90% for all four species. Therefore, long term recordings can be analyzed in just a few weeks and provide insight into the ecology of grouper spawning aggregations. FADAR was used to study the reproductive cycle of red hind (Epinephelus guttatus) over a 12-year period at a spawning aggregation site located off the west coast of Puerto Rico, in the Greater Antilles. Red hind produce four distinct call types that are used in combination during spawning aggregations, but mostly two are used in spawning associated courtship. Yearly times series confirmed in captivity observations of call type relative timing in courtship leading to spawning. The interannual variability observed in the relative variation of call type numbers suggests however, a shift in the male to female ratio toward a male dominated population, which could affect reproductive success.