Replication Data for: Acoustic occurrence of deep-diving cetaceans in the southern Adriatic Sea

Goose-beaked whale (Ziphius cavirostris) and sperm whale (Physeter macrocephalus) are two deep-diving odontocete species known to occur in the southern Adriatic Sea, based on intermittent stranding and sighting data. To get a better understanding of their presence in this region, we studied the occurrence of echolocation clicks produced by these species in passive acoustic recordings. Passive acoustic data were collected offshore Dubrovnik, Croatia. Data were collected from October 2018 to December 2019 and from June 2020 to December 2020 using a High-frequency Acoustic Recording Package (HARP) deployed at approximately 1,000 m depth. To determine the occurrence of echolocation clicks, a two-step process was used: automated detection followed by manual annotation of the detected clicks. Automated detection was implemented using a low signal-to-noise ratio threshold and a broad range of parameters to detect as many clicks as possible. Two click clustering algorithms were evaluated for performance on this data set: an unsupervised clustering algorithm and a filtering algorithm. The performance of the algorithms was compared to the manual annotations to determine a time-effective method for future monitoring. Both species were found to be acoustically active in the southern Adriatic Sea. Goose-beaked whales were detected throughout the recording period, with echolocation clicks peaking during the winter of 2018–2019 and in the spring of 2019. Conversely, sperm whales were more acoustically active during short, irregular windows, mostly during the summer of 2019. These results suggest that goose-beaked whales are residents, and, due to the irregularity of their detection, sperm whales are likely occasional visitors to this region. The unsupervised clustering algorithm performed better than the tested filtering algorithm, with F-scores of 0.74 and 0.46, respectively. This study provides knowledge that can help with effective conservation efforts and further contribute to our understanding of cetacean occurrence in the region.

基于间断性搁浅与目击数据，已知鹅喙鲸（Goose-beaked whale, Ziphius cavirostris）与抹香鲸（sperm whale, Physeter macrocephalus）这两种深潜齿鲸类会出没于亚得里亚海南部海域。为进一步明晰二者在该区域的分布情况，我们针对被动声学记录中这两类物种所产生的回声定位咔嗒声的出现情况开展了研究。被动声学数据采集于克罗地亚杜布罗夫尼克近海，数据采集时段涵盖2018年10月至2019年12月，以及2020年6月至2020年12月，所用设备为部署于约1000米水深的高频声学记录包（High-frequency Acoustic Recording Package, HARP）。为确定回声定位咔嗒声的出现情况，本研究采用了两步流程：先实施自动检测，再对检测到的咔嗒声开展人工标注。自动检测环节采用低信噪比阈值与宽泛的参数范围，以尽可能多地捕捉到目标咔嗒声。本数据集针对两种咔嗒声聚类算法的性能展开了评估：分别为无监督聚类算法与过滤算法。我们将两种算法的检测结果与人工标注结果进行比对，以筛选出适用于未来鲸类监测的省时高效方案。监测结果显示，两种鲸类在亚得里亚海南部海域均存在声学活动。鹅喙鲸在整个记录周期内均被检测到，其回声定位咔嗒声的出现峰值集中在2018-2019年冬季与2019年春季。与之相反，抹香鲸的声学活动则集中在短促且不规则的时段内，主要见于2019年夏季。上述结果表明，鹅喙鲸为该区域的定居物种；而由于抹香鲸的检测结果呈现出明显的不规则性，其大概率为该区域的偶访物种。无监督聚类算法的表现优于本次测试的过滤算法，二者的F分数（F-score）分别为0.74与0.46。本研究所得成果可为该区域鲸类的有效保护工作提供参考，同时也有助于深化我们对该区域鲸类分布情况的认知。