Quantifying harbour porpoise foraging behaviour in CPOD data: identification, automatic detection and potential application

Abstract

Harbour porpoise foraging behaviour 2.3 Gemini Offshore Wind Farm 2.4 CPOD data collection 3 Method 3.1 Click train based methods 3.2 Click based methods 4 Results and discussion 4.1 Identification of foraging behaviour 4.1.1 Click train based methodology 4.1.2 Click based methodology 4.2 Methods for automated detection of foraging/feeding events 4.2.1 Methodology comparison 4.2.2 Buzz clicks detection algorithm 4.3 Application to Gemini data 4.3.1 Changes over time and between stations 4.3.2 Relation to pile driving activities 4.4 Limitations 5 Conclusions and recommendations 5.1 Conclusions 5.2 Recommendations 6 Acknowledgments 7 Quality Assurance References Annex I: additional results Justification Wageningen Marine Research report C039/19| 4 of 41 Summary Harbour porpoises (Phocoena phocoena) are regularly monitored to assess how they are impacted by the construction and operation of offshore wind farms. A suitable method to do this is passive acoustic monitoring (PAM), and in particular using specific stationary hydrophones called CPODs. These devices provide information on click activity, which can then be analysed to investigate habitat use over time, differences between areas and the impact of human activities. Due to their small size and high metabolism porpoises are thought to need a more or less constant supply of prey to survive. Prey occurrence is thus considered one of the main drivers in porpoise distribution. And successful feeding is vital to the fitness and survival of individual porpoises. Information on foraging behaviour, however, is difficult to obtain in the field, in particular as animals feed under water. Recently the tagging of animals has provided new insights into porpoise behaviour, but it has been done for a limited number of individuals and for short times only. CPOD data have been used in Dutch waters to monitor harbour porpoise habitat use and behaviour before, during and after the construction of wind farms. The analyses have focussed on using a number of parameters that can be derived from the data, such as porpoise positive minutes, hours or days, encounter and waiting times. From other studies, primarily in captivity, we know that during foraging porpoises produce a characteristic pattern of clicks, starting with an approach phase and ending with a so-called "terminal buzz". Aim of our study was to investigate if we could quantify foraging behaviour from CPOD data, and we were able to use an existing data set of harbour porpoise click activity from the Gemini wind park (June 2015 to February 2016). The study consisted of three phases. First, the different existing methods were applied to a sample set of data to determine the most suitable approach to identify foraging behaviour. The results indicate that re-classification of clicks following the method developed by Pirotta (2014 a,b) to identify terminal buzzes provides the best results. Second, an algorithm was written to allow the automated analyses of CPOD data following this method. Finally, this analytical tool was applie...

Extracting foraging behavior from passive acoustic monitoring data to better understand harbor porpoise (Phocoena phocoena) foraging habitat use

“…This buzz:click ratio was used as a measure of foraging intensity. Comparable with Berges et al (2019), the buzz:click ratios, R b , were defined as:…”

“…Foraging buzzes are described as click sequences less than 10 ms long and indicative of a prey capture attempt (DeRuiter et al, 2009; Miller, 2010; Verfuß et al, 2009). Few studies using acoustic detection have actively differentiated occurrence from feeding behavior in wild porpoises (Berges et al, 2019; Nuuttila et al, 2013, 2018; Pirotta, Brookes, et al, 2014; Sørensen et al, 2018; Zein et al, 2019). There are principally two methods for identifying foraging behavior using passive acoustic monitoring; the click‐train based method or the click‐based method (Berges, 2019).…”

“…Few studies using acoustic detection have actively differentiated occurrence from feeding behavior in wild porpoises (Berges et al, 2019; Nuuttila et al, 2013, 2018; Pirotta, Brookes, et al, 2014; Sørensen et al, 2018; Zein et al, 2019). There are principally two methods for identifying foraging behavior using passive acoustic monitoring; the click‐train based method or the click‐based method (Berges, 2019). The click‐train based method examines entire click‐trains to identify distinct phases in the ICI time series typically indicative of foraging events (Miller, 2010; Schaffeld et al, 2016).…”

Feasibility of a Passive Acoustic Monitoring (PAM) network for Harbour Porpoises on the Dutch Continental Shelf

“…Linnenschmidt et al, 2013;Wisniewska et al, 2016). Berges et al (2019) compared different analytical options to identify and quantify these final buzzes when found in CPOD data collected in the GEMINI wind farm area in the Netherlands. The click-based approach by Pirotta et al (2014a,b), in which an ICI <10ms is used to classify buzzes, was deemed the most suitable for the data analysed.…”

“…Amundin, 1991), but currently cannot be distinguished from feeding buzzes in the CPOD data. Berges et al (2019) used the proportion of buzz clicks relative to the total number of clicks, the buzz click ratio, as an index of feeding behaviour. For the time being they assumed that the occurrence of non-foraging buzzes are a constant or at least random source of potential false positives.…”