Daily and seasonal use of vocalizations by nesting black-tailed godwits

Ground-nesting shorebirds must balance the need for acoustic communication at the nest with the constant threat posed by predators. Although it may seem likely that their calls are adapted to minimize detection by predators, little is known about how these birds communicate at the nest or whether they employ cryptic strategies to avoid predation. Using passive acoustic devices and software to analyse extensive acoustic data, we quantified and categorised the calls of black-tailed godwits (Limosa limosa limosa) recorded throughout the whole incubation at eight nests at a dairy farm in the Netherlands in March-June 2021. While incubating, godwits frequently use five main call types, with distinct diurnal patterns and high variation in the number of calls between breeding pairs. Birds used two quiet calls, one for communication at the nest and a second without an easily suggested meaning. Three loud calls were presumably used for predator alert, territory establishment and long-range communication. Interestingly, although nests were close to each other and exposed to the same aerial predators, the involvement of incubating birds in predator alert calling consistently differed. Furthermore, we described the relationship between the number of predator alert calls and the probability of a godwit flying off the nest. Our findings show that incubating godwits predominantly use loud vocalizations during the day, with only a few calls at night, which were more frequent on nights with a full moon. These descriptive findings for a single godwit community should now be expanded to other contexts, experimental situations, and shorebird species.

Methods
The study was conducted in 2021 at a dairy farm near Wommels, NW Netherlands, focusing on breeding black-tailed godwits. Eight godwit nests were recorded using SM2 recorders with microphones placed near the nests to capture vocalizations during the incubation period (24–29 days). Video footage from five nests complemented the audio recordings. Recordings were processed into 3-second fragments and analyzed for vocalizations using spectrograms, which were categorized into call types through manual and machine-learning techniques. We detected 77 325 3 s fragments that contained calls at eight nests during 183 fully recorded days. To better represent an individual’s decision to call, we used ‘acoustic events’, whereby all calls of the same type that are uttered with pauses of maximally 30 seconds are considered one event. A total of 15,078 acoustic events were identified. Statistical analysis, including general additive models (GAMs), was used to analyze daily and seasonal variations in vocal activity. For details, see the methods section in the published article.