Effects of duty cycle on passive acoustic monitoring metrics: The case of odontocete vocalizations

Passive acoustic monitoring is critical for long-term odontocete monitoring using autonomous recording devices. However, technical constraints, such as storage capacity and data processing limitations, often require temporal subsampling. This study investigates how varying duty cycles (50%-10%) and listening periods (1 min to 6 h) affect the detection of delphinid whistles and clicks, and harbor porpoise clicks. Two types of instruments were used: broadband recorders for whistles and F-PODs for clicks. As each device offers different configuration options, subsampling schemes were tailored to each signal type. The impact of duty cycles on seasonal patterns was evaluated using daily detection positive minutes and hours and diel patterns were assessed using hourly positive minutes and daily detection positive minutes ratios. Results indicate that higher duty cycles (50%) better preserve temporal pattern representations, particularly in high-activity sites, across both instruments and signal types. Lower duty cycles reduce the quality of data representation, especially in low-activity areas. Short listening periods (5-30 min) most closely approximate metrics from continuous recordings. These findings highlight the importance of adapting subsampling strategies to instrument capabilities and the overall level of acoustic activity, which varies across taxa and sites, to obtain an accurate representation of odontocete acoustic presence.