Periodic Reporting for period 1 - RAIN (Responses of cetAceans to natural and anthropogenIc Noise)
Reporting period: 2023-03-01 to 2025-02-28
The ocean ambient soundscapes, against which cetaceans must detect and decipher signals of interest is dynamic and complex. Where it was previously dominated by spatio-temporally loud signals from natural and biological sources, the post-industrial era has seen rapid influxes in acoustic signals from human-made sources (Hildebrand, 2009), which have been linked to negative effects on cetacean behaviour and physiology, with potential implications for individual fitness that may culminate in population-level consequences (Weilgart, 2007). However, although many natural noise sources can have similar acoustic characteristics to those produced by human-made sources, there is a lack of data on how cetaceans hear, respond to and cope with changes in natural noise within their environment, for which they have been subjected to over evolutionary timescales (Gomes et al. 2021). This ultimately has limited our understanding of how these species are potentially coping with the rapidly changing ocean soundscape of the Anthropocene. RAIN therefore aimed to study the responses of cetaceans to natural noises to better inform and contextualise our understanding of how these species respond to acoustically similar anthropogenic underwater noises such as from shipping.
To address this, we used harbour porpoises (Phocoena Phocoena) as a model species to understand how a marine mammal may hear and respond to rain noise as a dominant natural noise source in marine environments, and compared this to their responses to vessel noise. Harbour porpoises were chosen as they are the most common cetacean species in European shelf waters. Although harbour porpoise best hearing (10 to 140 kHz; Kastelein et al. 2002). is considerably above the peak of vessel noise (~125 Hz), exposure to the higher frequency components of vessel noise (up to and beyond 100 kHz; Hermannsen et al., 2014) has been found to illicit behavioural changes (e.g. reduced foraging) in tagged individuals from this species (Wisniewska et al. 2018). These high-frequency components of continuous vessel noise are similar to the noises of rain underwater, which is a principal source of natural abiotic noise, which produces a continuous, broadband (1-50 kHz) and high amplitude signal (Ma et al., 2005). However, to date, the audibility of rain and responses made by cetaceans to this natural noise source underwater has received no attention in the scientific literature.
Accordingly, to investigate the aim of RAIN and address these identified knowledge gaps, we initially proposed to answer three overall research objectives:
1. Quantify the absolute contributions of rain and vessel noise to local marine soundscapes experienced by cetaceans using long-term Passive Acoustic Monitoring (PAM).
2. Assess the behavioural and acoustic responses of individual harbour porpoises to rain and vessel noise events using state-of-the-art acoustic tag technology (DTAGs) to determine whether these responses are the same or differ depending on the sound source.
3. Investigate whether rain and/or vessel noise impacts the performance of harbour porpoises during an echolocation discrimination task.
Through RAIN we aimed to derive impact via the provision of new knowledge and data on the sounds of rain underwater, and the determine whether this noise source is audible to harbour porpoises and effects their behaviour. Via RAIN we aimed to improve awareness and understanding around how cetaceans respond to natural noise sources to better contextualise how they respond to human-made noise sources. Thus supporting and informing how best to reduce impacts to marine mammals and manage underwater noise in Europe to ensure we are able to sustainably use marine environments to meet blue economy and green transitional needs without compromising biodiversity.
To address this, we used harbour porpoises (Phocoena Phocoena) as a model species to understand how a marine mammal may hear and respond to rain noise as a dominant natural noise source in marine environments, and compared this to their responses to vessel noise. Harbour porpoises were chosen as they are the most common cetacean species in European shelf waters. Although harbour porpoise best hearing (10 to 140 kHz; Kastelein et al. 2002). is considerably above the peak of vessel noise (~125 Hz), exposure to the higher frequency components of vessel noise (up to and beyond 100 kHz; Hermannsen et al., 2014) has been found to illicit behavioural changes (e.g. reduced foraging) in tagged individuals from this species (Wisniewska et al. 2018). These high-frequency components of continuous vessel noise are similar to the noises of rain underwater, which is a principal source of natural abiotic noise, which produces a continuous, broadband (1-50 kHz) and high amplitude signal (Ma et al., 2005). However, to date, the audibility of rain and responses made by cetaceans to this natural noise source underwater has received no attention in the scientific literature.
Accordingly, to investigate the aim of RAIN and address these identified knowledge gaps, we initially proposed to answer three overall research objectives:
1. Quantify the absolute contributions of rain and vessel noise to local marine soundscapes experienced by cetaceans using long-term Passive Acoustic Monitoring (PAM).
2. Assess the behavioural and acoustic responses of individual harbour porpoises to rain and vessel noise events using state-of-the-art acoustic tag technology (DTAGs) to determine whether these responses are the same or differ depending on the sound source.
3. Investigate whether rain and/or vessel noise impacts the performance of harbour porpoises during an echolocation discrimination task.
Through RAIN we aimed to derive impact via the provision of new knowledge and data on the sounds of rain underwater, and the determine whether this noise source is audible to harbour porpoises and effects their behaviour. Via RAIN we aimed to improve awareness and understanding around how cetaceans respond to natural noise sources to better contextualise how they respond to human-made noise sources. Thus supporting and informing how best to reduce impacts to marine mammals and manage underwater noise in Europe to ensure we are able to sustainably use marine environments to meet blue economy and green transitional needs without compromising biodiversity.
Below the work performed and the main achievements from each objective of RAIN, and the subsequent plan for publication of this work are detailed.
Objective 1
Work Performed:
To address Objective 1, we deployed two PAM stations in the Bay of Aarhus, Denmark (Locations: (1) 56˚09.509’N, 010˚16.276’E and (2) 56˚06.035’N, 010˚15.967’E). At location 1 we deployed a SoundTrap ST500 (OceanInstruments, NZ) at approximately 9 m depth from the 4th December 2023 until the 7th February 2024. At location 2 we deployed a SoundTrap ST600HF (OceanInstruments, NZ) at approximately 14 m depth from the 11th July 2024 until the 31st August 2024. Both devices were secured 1 m above the seafloor and were moored to the bottom using a 50 kg biodegradable stone bag attached to an acoustic release.
Sound recordings from both deployments were analysed as 1 s decidecade sound pressure levels (Hann window and 0% overlap) with centre frequencies from 25 Hz to 80 kHz (ST500) or 160 kHz (ST600HF) using the standardise software PAMGuide (Merchant et al. 2015). Sound pressure levels were then auditory frequency weighted for very high-frequency (VHF) cetacean (harbour porpoise) hearing (Southall et al. 2019), and an energy summation was used to convert weighted sound pressure levels to broadband levels (25 Hz up to 160 kHz depending on deployment). The ambient level of each PAM deployment was assumed to be the 90th exceedance level across the entire deployment period. All sound analysis for PAM deployments was conducted in MATLAB.
For each PAM deployment, all high-amplitude events which exceeded the ambient sound level by at least 6 dB for a period of 30 s and did not occur within 30 s of the previous event were detected and were subsequently classified as ‘Rain’, ‘Vessel’ or ‘Ambient’. To aid the classification process, data on the hourly precipitation from the nearest meteorological station, and composite radar reflectivity in the 500 m2 pixel containing the PAM station, were extracted from the Danish Meteorological Institute (DMI) Open Data Application Programming Interface (API) using custom scripts written in MATLAB. Additionally, data on vessel activity within 5 km of the PAM stations were also obtained from Automatic Identification System (AIS) data taken from the Danish Maritime Authority (DMA). High-amplitude events were then visually classified using time-series plots created around the peak of the event ± 5 minutes. An event was classified as ‘Rain’ if visual inspection indicated (1) the presence of consistent broadband noise in frequencies between either 13-25 kHz or 1-50 kHz; (2) if a sharp onset or offset in noise was observed in the decidecade spectrogram; (3) if rain was recorded in the same hour at the local meteorological station and/or if composite radar reflectivity data indicated rainfall; and (4) if a clear increase in the broadband VHF cetacean weighted sound pressure level could be detected. Events were classified as ‘Vessel’ if an AIS vessel was present during the event or if manual auditing indicated the presence of a Lloyd’s Mirror signature. All other events were classified as ‘Ambient’ and were removed from the subsequent analysis.
Following the initial analysis, it was realised that rain data from the PAM deployment at location 1 was contaminated by a lot of noise from human-made sources (e.g. vessels, fast ferries and pumps), due to the proximity of this deployment to the Port of Aarhus. It was therefore decided that the data from this deployment and the subsequent analysis would not be included in the final publication. Nonetheless this data, has since been re-purposed for PhD projects looking at human-made noise sources in Aarhus Bay being undertaken within the same lab group and under the supervision of Professors Peter Teglberg Madsen and Jakob Tougaard at Aarhus University. Consequently, although this data was not used within RAIN, I was able to contribute towards other projects within the same lab group. All analysed data discussed below are therefore solely based on the data collected and analysed for location 2.
Broadband VHF cetacean weighted sound pressure levels of rain noise were compared to the local ambient sound level to assess audibility to harbour porpoises. Effects of exposure to rain noise on harbour porpoises based on PAM data included assessing: (1) temporary noise induced hearing loss and (2) auditory masking. The potential for noise from rain events to result in temporary noise induced hearing loss in harbour porpoises was assessed by comparing broadband cumulative sound exposure level computed over the duration of each rain event to published non-impulsive thresholds for Temporary Threshold Shifts (TTS; a temporary noise induced hearing loss) in VHF cetaceans such as harbour porpoises (153 dB re 1 µPa2s; Southall et al. 2019). The potential for auditory masking was quantified as the percentage reduction in listening space (see Pine et al. 2020). Whereby the unweighted sound pressure levels for rain events in each decidecade frequency band (25 Hz to 160 kHz) were compared to the harbour porpoise audiogram (Kastelein et al. 2002) or 50th/90th exceedance level in the same decidecade band (whichever was greater) of the PAM deployment without rain events included.
Main achievement:
In total, 1216 h of data were analysed from the PAM deployment at location 2 and 2276 high amplitude events were detected which ranged in duration from 30 s to approximately 3 h. Of these, 127 were classified as ‘Rain’, and 1142 were classified as ‘Vessel’ based on visual inspection and comparison to meteorological and AIS data.
When rain events were compared to the ambient noise level of 84 dB re 1 µPa RMS, rain sounds exceeded the ambient by an average of 10 dB (94 dB re 1 µPa RMS) and a maximum of 30 dB (114 dB re 1 µPa RMS). For vessel events, the median level was similar to that of rain at 91 dB re 1 µPa RMS, but a higher maximum broadband level was measured at 140 dB re 1 µPa RMS. These results indicated that rain and vessels were both clearly audible to harbour porpoises.
Based on the measurements of rain collected at the PAM station, the cumulative sound exposure levels from rain in this assessment did not reach levels expected to result in a TTS in harbour porpoise hearing.
With regards to auditory masking, results from the PAM data indicated that within the main frequencies of rain noise (1-50 kHz) harbour porpoises would experience a 65% reduction in listening space, which increased to a 90% reduction between 2 and 40 kHz. In frequencies relevant for echolocation (100-125 kHz) a 40% reduction in listening space was found. However, this result is likely an underestimate due to self-noise limitations of the hydrophone used which was unable to accurately capture harbour porpoise hearing abilities at these much higher frequencies. Our results therefore demonstrate that rain is a natural auditory masking source in frequencies between 1 and 50 kHz and may have masking potential at higher frequencies used for echolocation, but this requires further investigation.
As a result of the PAM deployments and subsequent analyses we have been able to quantify the contributions of rain and vessel noise to local marine soundscapes experienced by cetaceans addressing Objective 1 of RAIN.
Objective 2
Work Performed:
Data from 31 wild harbour porpoises, tagged with non-invasive high resolution sound- and movement-recording tags (DTAGs) in inner Danish waters of Kattegat, the Belt seas and Skaggerak, was used to assess behavioural responses to rain noise underwater. The tag data was initially obtained from the EU Horizon 2020 SATURN project which aimed to quantify responses of harbour porpoises to vessel noise (Grant Agreement No. 101006443).
Sound data when the porpoise was at least 0.5 m below the surface was analysed as decidecade sound pressure levels centred at 16 kHz to minimise flow noise in sound recordings (Wisniewska et al. 2018). Initially, sound recordings were subdivided into 1 ms bins over 0.5 s intervals and the 10th exceedance level in each bin was removed and a trimmed mean of the power during the remaining 1 ms bins was calculated over each 0.5 s. The sound pressure level in the 16 kHz band was then calculated over 10 s intervals (50% overlap and taking the 75th exceedance level of all 0.5 s intervals within each 10 s). As this approach needed to adequately capture both vessel and rain noise, extensive discussions were had with collaborators within the SATURN project on how to analyse noise data from tagged individuals to ensure we were able to accurately quantify noise levels and exposures from vessels and rain to make comparisons across these sources. Due to the dependence between these projects and requirements to be able to use this data, the extensive discussions resulted in considerable delays to the analysis of tag data required by this project. This ultimately had a knock on effect with regards to project time available for completing Objective 3 (see below). The approach described above was ultimately decided upon by all collaborators as it was able to minimise flow noise during powerful swimming exertions while adequately capturing high-amplitude noise peaks.
Rain and vessel events were detected and classified in tag acoustic recordings, by detecting all high-amplitude events (>90 dB) in the 16 kHz band and assuming the next event did not occur within a 3 minute period of the previous event. All high-amplitude events were initially classified as part of the SATURN project, and events which were classified as ‘Vessel’ for SATURN were then excluded from the analysis within RAIN. All non-vessel events were then re-assessed using time-series plots around the event peak ± 2 minutes. An event was classified as rain if visual inspection indicated the presence of consistent broadband noise in frequencies between either 13-25 kHz or 1-50 kHz and/or if a sharp onset or offset in noise was observed, similar to the approach employed for the PAM data in Objective 1. To aid classification of the tag data, aural inspection for the sound of ‘running water’ was also used to detect the presence of rain.
To convert the noise level at 16 kHz measured by the tags to the representative broadband sound pressure levels experienced by harbour porpoises, data from the PAM station (Objective 1) was used to derive a conversion factor. This was done by calculating the difference between the 16 kHz decidecade sound pressure levels and broadband (25 Hz to 160 kHz) VHF cetacean weighted sound pressure levels of all detected rain events in the PAM data and adding the median difference (7.6 dB) to the measured levels at 16 kHz recorded by the tag.
Tag data was used to assess changes in harbour porpoise behaviour (e.g. buzz and respiration rate, dive depth (m), and minimum specific acceleration (m/s2)) during all high-amplitude rain events which occurred at least one hour after the initial tag deployment (to reduce impacts of animal handling on results). Buzzes and respirations from tag deployments were detected using the approach described by Rojano-Doñate et al. (2024), and pressure, magnetometer and accelerometer measurements from tag data were converted to depth (m), magnetic field (µT) and acceleration (m/s) to assess changes in dive behaviour and movement. Changes in harbour porpoise behaviour were assessed using a period of 5 minutes before and 15 minutes after the high amplitude event threshold was crossed to create ‘rain exposures’. Buzz rate (a proxy for energy acquisition) and respiration rate (a proxy for energy expenditure) were calculated as a function of diel period and treatment (control or rain exposure) to derive number of buzzes and respirations in each period and for each individual and all animals combined. A Generalized Linear Mixed-Model was then used to assess statistical changes in buzz and respiration rate as a function of diel period and control or rain exposure, including animal ID as a random intercept using R software.
Main achievement:
814 h of data from 31 harbour porpoise tag deployments were analysed for the presence of vessels (SATURN) and rain. Of these individuals, 17 were exposed to rain events, with a total of 99 events detected and 95% of these events occurred for a duration of 5 s to 20 min.
After tag recordings were corrected for VHF cetacean hearing, rain events were clearly audible to tagged harbour porpoises. The median broadband VHF cetacean weighted sound pressure level of all rain events detected on tagged individuals was 99 dB re 1 µPa RMS and the maximum was 129 dB re 1 µPa RMS. Measured levels varied between individuals, and the loudest rain event recorded was experienced by a juvenile male harbour porpoise (hp13_170a).
Assessments of behavioural responses to rain noise indicated that in contrast to what was found during exposure to vessel noise in the same tagged individuals (SATURN project), exposure to rain did not significantly alter the foraging behaviour of wild harbour porpoises. However, a significant reduction in respiration rate (6%) was reported during rain exposures, which additional post-hoc analysis indicated could be due to porpoises conducting deeper dives (~2.2 m deeper) during periods with rain. These results indicating that porpoises may use changes in dive depth as an evolutionary coping mechanism to reduce auditory load.
As a result of the tag deployments and subsequent analysis performed within RAIN and the associated SATURN project, we were therefore able to assess the behavioural responses of individual harbour porpoises to rain noise events and compare these responses to those made by the same individuals during exposure to vessel noise, in turn determining that responses made by harbour porpoises to these two noise sources differ, addressing Objective 2 of RAIN.
Objective 3
Due to the unexpected and extensive discussions on how to analyse noise data from tagged harbour porpoises, we were unable to undertake the work initially proposed to address Objective 3. However, through work conducted as part of Objective 1, we were able to begin to address this objective by quantifying the auditory masking potential of rain exposures on harbour porpoises at frequencies used for echolocation by this species (i.e. 100-125 kHz). However, given the self-noise limitations of the hydrophone used in the PAM deployment, we do wish to conduct additional work following the completion of RAIN to better address how rain noise could be affecting echolocation performance in this species.
Publication plan:
The work undertaken within RAIN and the results of Objectives 1 and 2 have been combined into a scientific publication which is currently in preparation to be submitted to Science Advances. This publication is titled: “Does loud natural noise from rain affect harbour porpoises” and has been produced as a collaborative effort between myself (Charlotte R. Findlay) and five co-authors from Aarhus University (Departments of Biology and Ecoscience) and the University of Veterinary Medicine Hannover (Institute for Terrestrial and Aquatic Wildlife Research). We plan to submit this publication for review after the project end date.
Additionally, work which is complementary to Objective 2 of RAIN on the energetic responses of the tagged wild harbour porpoises during exposure to vessel noise undertaken as part of the SATURN project which I am a co-author on will also be submitted for publication following the end date of RAIN.
Objective 1
Work Performed:
To address Objective 1, we deployed two PAM stations in the Bay of Aarhus, Denmark (Locations: (1) 56˚09.509’N, 010˚16.276’E and (2) 56˚06.035’N, 010˚15.967’E). At location 1 we deployed a SoundTrap ST500 (OceanInstruments, NZ) at approximately 9 m depth from the 4th December 2023 until the 7th February 2024. At location 2 we deployed a SoundTrap ST600HF (OceanInstruments, NZ) at approximately 14 m depth from the 11th July 2024 until the 31st August 2024. Both devices were secured 1 m above the seafloor and were moored to the bottom using a 50 kg biodegradable stone bag attached to an acoustic release.
Sound recordings from both deployments were analysed as 1 s decidecade sound pressure levels (Hann window and 0% overlap) with centre frequencies from 25 Hz to 80 kHz (ST500) or 160 kHz (ST600HF) using the standardise software PAMGuide (Merchant et al. 2015). Sound pressure levels were then auditory frequency weighted for very high-frequency (VHF) cetacean (harbour porpoise) hearing (Southall et al. 2019), and an energy summation was used to convert weighted sound pressure levels to broadband levels (25 Hz up to 160 kHz depending on deployment). The ambient level of each PAM deployment was assumed to be the 90th exceedance level across the entire deployment period. All sound analysis for PAM deployments was conducted in MATLAB.
For each PAM deployment, all high-amplitude events which exceeded the ambient sound level by at least 6 dB for a period of 30 s and did not occur within 30 s of the previous event were detected and were subsequently classified as ‘Rain’, ‘Vessel’ or ‘Ambient’. To aid the classification process, data on the hourly precipitation from the nearest meteorological station, and composite radar reflectivity in the 500 m2 pixel containing the PAM station, were extracted from the Danish Meteorological Institute (DMI) Open Data Application Programming Interface (API) using custom scripts written in MATLAB. Additionally, data on vessel activity within 5 km of the PAM stations were also obtained from Automatic Identification System (AIS) data taken from the Danish Maritime Authority (DMA). High-amplitude events were then visually classified using time-series plots created around the peak of the event ± 5 minutes. An event was classified as ‘Rain’ if visual inspection indicated (1) the presence of consistent broadband noise in frequencies between either 13-25 kHz or 1-50 kHz; (2) if a sharp onset or offset in noise was observed in the decidecade spectrogram; (3) if rain was recorded in the same hour at the local meteorological station and/or if composite radar reflectivity data indicated rainfall; and (4) if a clear increase in the broadband VHF cetacean weighted sound pressure level could be detected. Events were classified as ‘Vessel’ if an AIS vessel was present during the event or if manual auditing indicated the presence of a Lloyd’s Mirror signature. All other events were classified as ‘Ambient’ and were removed from the subsequent analysis.
Following the initial analysis, it was realised that rain data from the PAM deployment at location 1 was contaminated by a lot of noise from human-made sources (e.g. vessels, fast ferries and pumps), due to the proximity of this deployment to the Port of Aarhus. It was therefore decided that the data from this deployment and the subsequent analysis would not be included in the final publication. Nonetheless this data, has since been re-purposed for PhD projects looking at human-made noise sources in Aarhus Bay being undertaken within the same lab group and under the supervision of Professors Peter Teglberg Madsen and Jakob Tougaard at Aarhus University. Consequently, although this data was not used within RAIN, I was able to contribute towards other projects within the same lab group. All analysed data discussed below are therefore solely based on the data collected and analysed for location 2.
Broadband VHF cetacean weighted sound pressure levels of rain noise were compared to the local ambient sound level to assess audibility to harbour porpoises. Effects of exposure to rain noise on harbour porpoises based on PAM data included assessing: (1) temporary noise induced hearing loss and (2) auditory masking. The potential for noise from rain events to result in temporary noise induced hearing loss in harbour porpoises was assessed by comparing broadband cumulative sound exposure level computed over the duration of each rain event to published non-impulsive thresholds for Temporary Threshold Shifts (TTS; a temporary noise induced hearing loss) in VHF cetaceans such as harbour porpoises (153 dB re 1 µPa2s; Southall et al. 2019). The potential for auditory masking was quantified as the percentage reduction in listening space (see Pine et al. 2020). Whereby the unweighted sound pressure levels for rain events in each decidecade frequency band (25 Hz to 160 kHz) were compared to the harbour porpoise audiogram (Kastelein et al. 2002) or 50th/90th exceedance level in the same decidecade band (whichever was greater) of the PAM deployment without rain events included.
Main achievement:
In total, 1216 h of data were analysed from the PAM deployment at location 2 and 2276 high amplitude events were detected which ranged in duration from 30 s to approximately 3 h. Of these, 127 were classified as ‘Rain’, and 1142 were classified as ‘Vessel’ based on visual inspection and comparison to meteorological and AIS data.
When rain events were compared to the ambient noise level of 84 dB re 1 µPa RMS, rain sounds exceeded the ambient by an average of 10 dB (94 dB re 1 µPa RMS) and a maximum of 30 dB (114 dB re 1 µPa RMS). For vessel events, the median level was similar to that of rain at 91 dB re 1 µPa RMS, but a higher maximum broadband level was measured at 140 dB re 1 µPa RMS. These results indicated that rain and vessels were both clearly audible to harbour porpoises.
Based on the measurements of rain collected at the PAM station, the cumulative sound exposure levels from rain in this assessment did not reach levels expected to result in a TTS in harbour porpoise hearing.
With regards to auditory masking, results from the PAM data indicated that within the main frequencies of rain noise (1-50 kHz) harbour porpoises would experience a 65% reduction in listening space, which increased to a 90% reduction between 2 and 40 kHz. In frequencies relevant for echolocation (100-125 kHz) a 40% reduction in listening space was found. However, this result is likely an underestimate due to self-noise limitations of the hydrophone used which was unable to accurately capture harbour porpoise hearing abilities at these much higher frequencies. Our results therefore demonstrate that rain is a natural auditory masking source in frequencies between 1 and 50 kHz and may have masking potential at higher frequencies used for echolocation, but this requires further investigation.
As a result of the PAM deployments and subsequent analyses we have been able to quantify the contributions of rain and vessel noise to local marine soundscapes experienced by cetaceans addressing Objective 1 of RAIN.
Objective 2
Work Performed:
Data from 31 wild harbour porpoises, tagged with non-invasive high resolution sound- and movement-recording tags (DTAGs) in inner Danish waters of Kattegat, the Belt seas and Skaggerak, was used to assess behavioural responses to rain noise underwater. The tag data was initially obtained from the EU Horizon 2020 SATURN project which aimed to quantify responses of harbour porpoises to vessel noise (Grant Agreement No. 101006443).
Sound data when the porpoise was at least 0.5 m below the surface was analysed as decidecade sound pressure levels centred at 16 kHz to minimise flow noise in sound recordings (Wisniewska et al. 2018). Initially, sound recordings were subdivided into 1 ms bins over 0.5 s intervals and the 10th exceedance level in each bin was removed and a trimmed mean of the power during the remaining 1 ms bins was calculated over each 0.5 s. The sound pressure level in the 16 kHz band was then calculated over 10 s intervals (50% overlap and taking the 75th exceedance level of all 0.5 s intervals within each 10 s). As this approach needed to adequately capture both vessel and rain noise, extensive discussions were had with collaborators within the SATURN project on how to analyse noise data from tagged individuals to ensure we were able to accurately quantify noise levels and exposures from vessels and rain to make comparisons across these sources. Due to the dependence between these projects and requirements to be able to use this data, the extensive discussions resulted in considerable delays to the analysis of tag data required by this project. This ultimately had a knock on effect with regards to project time available for completing Objective 3 (see below). The approach described above was ultimately decided upon by all collaborators as it was able to minimise flow noise during powerful swimming exertions while adequately capturing high-amplitude noise peaks.
Rain and vessel events were detected and classified in tag acoustic recordings, by detecting all high-amplitude events (>90 dB) in the 16 kHz band and assuming the next event did not occur within a 3 minute period of the previous event. All high-amplitude events were initially classified as part of the SATURN project, and events which were classified as ‘Vessel’ for SATURN were then excluded from the analysis within RAIN. All non-vessel events were then re-assessed using time-series plots around the event peak ± 2 minutes. An event was classified as rain if visual inspection indicated the presence of consistent broadband noise in frequencies between either 13-25 kHz or 1-50 kHz and/or if a sharp onset or offset in noise was observed, similar to the approach employed for the PAM data in Objective 1. To aid classification of the tag data, aural inspection for the sound of ‘running water’ was also used to detect the presence of rain.
To convert the noise level at 16 kHz measured by the tags to the representative broadband sound pressure levels experienced by harbour porpoises, data from the PAM station (Objective 1) was used to derive a conversion factor. This was done by calculating the difference between the 16 kHz decidecade sound pressure levels and broadband (25 Hz to 160 kHz) VHF cetacean weighted sound pressure levels of all detected rain events in the PAM data and adding the median difference (7.6 dB) to the measured levels at 16 kHz recorded by the tag.
Tag data was used to assess changes in harbour porpoise behaviour (e.g. buzz and respiration rate, dive depth (m), and minimum specific acceleration (m/s2)) during all high-amplitude rain events which occurred at least one hour after the initial tag deployment (to reduce impacts of animal handling on results). Buzzes and respirations from tag deployments were detected using the approach described by Rojano-Doñate et al. (2024), and pressure, magnetometer and accelerometer measurements from tag data were converted to depth (m), magnetic field (µT) and acceleration (m/s) to assess changes in dive behaviour and movement. Changes in harbour porpoise behaviour were assessed using a period of 5 minutes before and 15 minutes after the high amplitude event threshold was crossed to create ‘rain exposures’. Buzz rate (a proxy for energy acquisition) and respiration rate (a proxy for energy expenditure) were calculated as a function of diel period and treatment (control or rain exposure) to derive number of buzzes and respirations in each period and for each individual and all animals combined. A Generalized Linear Mixed-Model was then used to assess statistical changes in buzz and respiration rate as a function of diel period and control or rain exposure, including animal ID as a random intercept using R software.
Main achievement:
814 h of data from 31 harbour porpoise tag deployments were analysed for the presence of vessels (SATURN) and rain. Of these individuals, 17 were exposed to rain events, with a total of 99 events detected and 95% of these events occurred for a duration of 5 s to 20 min.
After tag recordings were corrected for VHF cetacean hearing, rain events were clearly audible to tagged harbour porpoises. The median broadband VHF cetacean weighted sound pressure level of all rain events detected on tagged individuals was 99 dB re 1 µPa RMS and the maximum was 129 dB re 1 µPa RMS. Measured levels varied between individuals, and the loudest rain event recorded was experienced by a juvenile male harbour porpoise (hp13_170a).
Assessments of behavioural responses to rain noise indicated that in contrast to what was found during exposure to vessel noise in the same tagged individuals (SATURN project), exposure to rain did not significantly alter the foraging behaviour of wild harbour porpoises. However, a significant reduction in respiration rate (6%) was reported during rain exposures, which additional post-hoc analysis indicated could be due to porpoises conducting deeper dives (~2.2 m deeper) during periods with rain. These results indicating that porpoises may use changes in dive depth as an evolutionary coping mechanism to reduce auditory load.
As a result of the tag deployments and subsequent analysis performed within RAIN and the associated SATURN project, we were therefore able to assess the behavioural responses of individual harbour porpoises to rain noise events and compare these responses to those made by the same individuals during exposure to vessel noise, in turn determining that responses made by harbour porpoises to these two noise sources differ, addressing Objective 2 of RAIN.
Objective 3
Due to the unexpected and extensive discussions on how to analyse noise data from tagged harbour porpoises, we were unable to undertake the work initially proposed to address Objective 3. However, through work conducted as part of Objective 1, we were able to begin to address this objective by quantifying the auditory masking potential of rain exposures on harbour porpoises at frequencies used for echolocation by this species (i.e. 100-125 kHz). However, given the self-noise limitations of the hydrophone used in the PAM deployment, we do wish to conduct additional work following the completion of RAIN to better address how rain noise could be affecting echolocation performance in this species.
Publication plan:
The work undertaken within RAIN and the results of Objectives 1 and 2 have been combined into a scientific publication which is currently in preparation to be submitted to Science Advances. This publication is titled: “Does loud natural noise from rain affect harbour porpoises” and has been produced as a collaborative effort between myself (Charlotte R. Findlay) and five co-authors from Aarhus University (Departments of Biology and Ecoscience) and the University of Veterinary Medicine Hannover (Institute for Terrestrial and Aquatic Wildlife Research). We plan to submit this publication for review after the project end date.
Additionally, work which is complementary to Objective 2 of RAIN on the energetic responses of the tagged wild harbour porpoises during exposure to vessel noise undertaken as part of the SATURN project which I am a co-author on will also be submitted for publication following the end date of RAIN.
RAIN is the first project to demonstrate that the noise from rainfall is clearly audible to harbour porpoises and can potentially interfere with their sensory performance when conducting biologically important tasks. Rain as a distributed noise source is a significant contributor to marine soundscapes and when present exceeds noise levels of distant shipping. However, when present at the same time as a nearby ship, the noise from rain is likely to become obscured to harbour porpoises or other receivers.
Rain recorded at the PAM station in our study did not reach cumulative sound exposure levels expected to result in temporary noise induced hearing loss in harbour porpoises based on the latest criteria (153 dB re 1 µPa2s; Southall et al. 2019). Theoretically, rain sounds could reach these levels if persistent at high amplitudes for prolonged durations. However, we feel that additional research on how cetacean auditory systems can cope with prolonged exposure to low levels of continuous noise is still needed to accurately determine risks of noise induced hearing loss during exposure to noise from vessels and rain.
In RAIN we clearly demonstrate that auditory masking is a natural phenomenon which can occur even in the absence of human-made noises. Consequently, it can be expected that harbour porpoises and other cetacean species will have developed mechanisms to deal with periods of auditory masking and can tolerate occasional masking events by human-made sources with few negative effects. However, given that these signals are not independent in the soundscape, and due to global increases in anthropogenic activities at sea and the increasing frequency, intensity and duration of rainfall due to climatic changes, our results indicate that cetaceans may experience greater levels and durations of auditory masking into the future if human-made noise is not reduced.
Through RAIN and associated work as part of the SATURN project we have demonstrated that harbour porpoises show different responses to vessels and rain, suggesting that individuals may be using contextual information or other factors to distinguish between these noise sources when deciding how to respond. Furthermore, we have highlighted diving deeper a potential mechanism which harbour porpoises use to cope with increased auditory load from exposure to rain noise underwater. Further research is however needed to better understand how sounds from rain change with increasing depth from the surface and in areas with complex oceanographic conditions, and/or how prey behaviour changes during rainfall as a strong driver of porpoise behaviour.
In conclusion, RAIN has yielding important findings, expanding our understanding of the responses of cetaceans to natural noise sources which will help to contextualise their responses to acoustically similar anthropogenic sources such as shipping noise. These insights therefore contribute to the broader field of underwater noise impacts to marine wildlife and provide support for future research into the responses of cetaceans to both natural and anthropogenic noise sources.
Rain recorded at the PAM station in our study did not reach cumulative sound exposure levels expected to result in temporary noise induced hearing loss in harbour porpoises based on the latest criteria (153 dB re 1 µPa2s; Southall et al. 2019). Theoretically, rain sounds could reach these levels if persistent at high amplitudes for prolonged durations. However, we feel that additional research on how cetacean auditory systems can cope with prolonged exposure to low levels of continuous noise is still needed to accurately determine risks of noise induced hearing loss during exposure to noise from vessels and rain.
In RAIN we clearly demonstrate that auditory masking is a natural phenomenon which can occur even in the absence of human-made noises. Consequently, it can be expected that harbour porpoises and other cetacean species will have developed mechanisms to deal with periods of auditory masking and can tolerate occasional masking events by human-made sources with few negative effects. However, given that these signals are not independent in the soundscape, and due to global increases in anthropogenic activities at sea and the increasing frequency, intensity and duration of rainfall due to climatic changes, our results indicate that cetaceans may experience greater levels and durations of auditory masking into the future if human-made noise is not reduced.
Through RAIN and associated work as part of the SATURN project we have demonstrated that harbour porpoises show different responses to vessels and rain, suggesting that individuals may be using contextual information or other factors to distinguish between these noise sources when deciding how to respond. Furthermore, we have highlighted diving deeper a potential mechanism which harbour porpoises use to cope with increased auditory load from exposure to rain noise underwater. Further research is however needed to better understand how sounds from rain change with increasing depth from the surface and in areas with complex oceanographic conditions, and/or how prey behaviour changes during rainfall as a strong driver of porpoise behaviour.
In conclusion, RAIN has yielding important findings, expanding our understanding of the responses of cetaceans to natural noise sources which will help to contextualise their responses to acoustically similar anthropogenic sources such as shipping noise. These insights therefore contribute to the broader field of underwater noise impacts to marine wildlife and provide support for future research into the responses of cetaceans to both natural and anthropogenic noise sources.