Internet Explorer lacks support for the features of this website
please use a modern browser such as Chrome
A .gov website belongs to an official government organization in the United States
As part of the Faces of the Southeast Fisheries Science Center series
and my passion for life in the oceans probably started from observing the local coastal ocean and its inhabitants.
I earned a Bachelor’s of Science in Marine Biology from the Université de La Rochelle
I went to the Université du Québec à Rimouski
and graduated with a Master’s of Science in Oceanography
I got my Doctorate in Biology from the Institut National de la Recherche Scientifique
After receiving my Ph.D. and a postdoc at Fisheries and Oceans Canada
I had the opportunity to expand my skills to the field of passive acoustics
I worked for 9 years as a bio-acoustician in a Canadian company providing consulting services in underwater acoustics
Something was missing in my life and I realized I wanted to go back to research
I got the chance to join the Marine Mammal and Turtle Division’s Passive Acoustic Ecology Program as a postdoctoral research associate.
I am a NOAA affiliate working for the Cooperative Institute for Marine and Atmospheric Studies
My work mainly involves fieldwork preparation
My current research aims to better understand beaked whales’ diving and calling behavior in the Gulf of Mexico using passive acoustic tracking
This knowledge will be used to obtain more reliable and robust population density estimates from passive acoustics data.
I really enjoy working with great colleagues in an interdisciplinary and collaborative environment
It’s very stimulating to engage with scientists who are knowledgeable
after 10 years working on Arctic or subarctic marine mammal species
I particularly appreciate the opportunity of fieldwork in a warm environment.
Our Marine Mammal and Turtle Division Director, Mridula Srinivasan, has previously provided great advice on “how to become a marine mammal scientist.” One thing that resonates particularly with me is about the knowledge transfer across scientific disciplines
My advice is to pursue a discipline that you are genuinely interested in and that can be easily applied to marine mammal occupations.
I am passionate about new technologies and how they can change the way we do research
I developed a dual-drone approach to simultaneously monitor visual and acoustic behavior of marine mammals
I would love to expand this approach (or a similar one) in my research
Another exciting technology that we are currently interested in is uncrewed systems such as ocean gliders equipped with passive acoustic monitoring devices
These new platforms may help us to conduct dolphin and whale assessments in remote regions
Protecting the ocean and conserving its biodiversity is important to ensure a sustainable future. Marine mammals are essential to the balance of marine ecosystems and they are sentinels of changes in the overall health of the ocean.
Almost all marine mammal species face at least one threat
which are caused either directly or indirectly by human actions
One important issue is the lack of public awareness on biodiversity conservation
Scientists are communicating their research in the hope that policy makers make more informed decisions
public attention and interest in the fate of endangered species is a crucial prerequisite for funding of conservation programs
which is the world’s newest whale species and is already one of the most endangered whales in the world
Because this species is not well known to the general public
Public engagement around science can make a difference.
The French oceanographer Jacques (-Yves) Cousteau is my childhood hero
A quote that I like is “Curiosity is the essence of the scientific mind” (Calvin and Hobbes by Bill Watterson)
Curiosity is a powerful tool for great science but also for everyday learning and opportunity.
I like to be with my husband and my two kids
We love to be outdoors whenever possible and spend a lot of time exploring around Miami and learning about its different heritages
Mouy Tang walked out of an adult care facility on Sept
It wasn’t until a decade later that her remains were found just 100 yards away
Now her family wants the state to answer for her death
Tang was walking along a trail that leads from Unique Living assisted care where she lived for 16 years in Lawndale to a mini-mart just up the road that residents were known to visit
On Feb. 17, 2018, rabbit hunters stumbled across the remains of the 46-year-old woman
Tang’s family has filed a lawsuit against the North Carolina Department of Health and Human Services
The suit claims the state’s negligence to impose regulations at Unique Living led to the woman’s death
“We are still going through the lawsuit to get justice
She saw the pain in his eyes before he even uttered a word
the Florida-based family was in Cleveland County looking for their missing family member
“We kept asking the questions over and over again
and no one seemed to want to answer,” Quynh Tang said
Workers at Unique Living were unaware Tang
who suffered from schizophrenia and diabetes
had left because she walked through a door that had no alarm or sign-out process for the care facility
A state inspection of the facility two months before her disappearance showed that multiple doors in the facility had no alarms
along with a host of other violations from no proper dishware for residence to eat off of to improper temperature controls for the water
Industrial Commission that handles lawsuits against the state
The documents show Unique Living had until Aug
but no inspector came back to the facility to ensure those changes had been made by the time Mouy Tang wandered off
It was on a visit to Cleveland County Department of Social Services where the Tang family began learning of the violations and the living conditions at the facility
according to Chris Duggan and William Goldfarb
attorneys representing the Tang family in the suit
“Our whole crux of this case is that the state knew and should have known that this facility had a lot of issues,” Duggan said
Records from the legal proceedings show Megan Lamphere
the office responsible for regulating these facilities
defended the lack of action taken to enforce deadlines for corrective action
Lamphere noted that the ‘date of correction’ was not actually a deadline,” the Industrial Commission wrote in its opinion
“DHSR would typically return 45-60 days after the correction date to follow up to ensure the facility had come into compliance.”
Tang’s family should have done more to address their concerns regarding her deteriorating mental state,” read the opinion
Quynh Tang said she and her family relied on the facility to take care of Mouy Tang
You don’t know that the people trusted to take care of your family isn’t,” Quynh Tang said
and (Unique Living) would make it look like she was fine.”
Duggan and Goldfarb compiled Mouy Tang’s whole life into 20 boxes
starting from her life in Cambodia through becoming a refugee in the United States to her last day at Unique Living
That information coupled with Cleveland County DSS’s documented history of issues at the facility
Industrial Commission to side with the Tangs twice
Awarding the family $500,000 from the state in the death of Mouy Tang
“Everywhere we turn we would think we were getting justice for her
but we just get knocked down again,” Quynh Tang said
The appeal process could take up to another year
Joyce Orlando can be reached at 704-669-3341 or find her on Twitter @Star_J_Orlando
Researchers studying animals living in the Barkley Canyon
a fascinating part of the ocean off Vancouver Island
found something interesting: northern elephant seals are responding to sonar pulses like Pavlov’s dogs
“We suspect the seals have learned to associate sonar noise from the research instrument with the presence of food — a phenomenon known as the ‘dinner bell’ effect,” Héloïse Frouin-Mouy, lead study author from the University of Miami and the University of Victoria, said in a statement
The sonar device is submerged in over 2,000 feet of water
and eight of the elephant seals in the area took special interest in it
The point of the experiment was to examine the effects of light and bait on deep-sea fish and invertebrate behaviors
“The seals appeared to use this sound to locate an area with prey and may take advantage of fish disturbed by the camera lights,” Mouy continued
their preferred meal as seen in the video footage.”
“We became familiar with the mammals and ended up naming them in the paper after members of The Beach Boys to differentiate between the frequency of visits and observed habits,” Frouin-Mouy explained
It wasn’t just the feeding behavior the researchers noticed, either — they also found the elephant seals to be engaging in acts unseen before
like taking power naps on the seafloor for over eight minutes at a time
They would lie completely motionless down there until they were forced to surface for air
Metrics details
Cetaceans spend most of their time below the surface of the sea
highlighting the importance of passive acoustic monitoring as a tool to facilitate understanding and mapping their year-round spatial and temporal distributions
To increase our limited knowledge of cetacean acoustic detection patterns for the east and west coasts of Gwaii Haanas
acoustic datasets recorded off SG̱ang Gwaay (Sep 2009–May 2011)
and Ramsay Island (Aug 2018–Aug 2019) were analyzed
Comparing overlapping periods of visual surveys and acoustic monitoring confirmed presence of 12 cetacean species/species groups within the study region
Seasonal patterns were identified for blue
Killer whale and delphinid acoustic signals occurred year-round on both coasts of Haida Gwaii and showed strong diel variation
were identified in high-frequency recordings on the west coast
Correlations between environmental factors
and cetacean acoustic occurrence off Gwaii Haanas were also examined
This study is the first to acoustically monitor Gwaii Haanas waters for an extended continuous period and therefore serves as a baseline from which to monitor future changes
movements and distributions of cetaceans off southwestern HG remain poorly understood due to the many challenges of offshore visual monitoring
This study highlights the critical role of collaborations among organizations and experts required to implement such cetacean monitoring work
particularly in remote and rugged coastal regions
Hydrophone deployments and retrievals were conducted as a collaboration between Gwaii Haanas
Acoustic analyses were conducted by JASCO Applied Sciences
Recorders at each site varied in recording durations, schedules, and sample frequencies (Table 1)
Recorders were deployed to seafloor depths between 200 and 741 m as a vertical mooring with the hydrophone approximately 10 m above the seafloor for GI and SG
and approximately 50–100 m above the seafloor for RI
AURALs were fitted with an HTI-96-MIN omnidirectional hydrophone (High Tech Inc.
had an analog gain of 16 dB and were set to record at a sampling frequency of 16,384 Hz with a resolution of 16-bit
AMARs were fitted with a GTI M36-V35-100 omnidirectional hydrophone (GeoSpectrum
nominal sensitivity: − 165 ± 3 dB re 1 V/μPa) and were set to record at two different sampling frequencies
The low-frequency (16 kHz) and high frequency (250 kHz) recording channels had a resolution of 24-bit and 16-bit
The SM2M was fitted with an HTI-92-WB hydrophone (Wildlife Acoustics’ ‘low-noise’ hydrophone option)
set to sample at a frequency of 96 kHz on 16-bit and with an analog gain of 0 dB
The SM2M was calibrated before the deployment in the field (an end-to-end calibration was performed)
Analyses were based on 1-min average power spectral density of the data computed from fast Fourier transforms (FFTs) of 1 s of data overlapped by 0.5 s (120 averages)
Acoustic metrics used to quantify the ambient sound were: (1) root-mean-square sound pressure level (rms SPL) and (2) power spectral density (PSD) level
‘Light’ periods were defined as those hours between sunrise and sunset
and ‘dark’ hours were between sunset and sunrise
and environmental and bathymetric conditions between source and receiver
cetacean vocalization SLs (Supplementary Sect
2.2) were obtained from published literature
and ambient noise levels (NLs) recorded at GS and RI (Supplementary Sect
PL was modelled for two sites along four radials in different directions to sample PL characteristics as a function of range and azimuth (Supplementary Sect
The maximum distance a vocalization can be detected is that at which the vocalization’s RL exceeds NL at the recorder in the same frequency band by at least the DT:
The DT used here strictly represents the signal processing DT for automated detectors and is not related to the listening DT of the animals
At a given source depth, the detection range was estimated separately for each frequency band of the vocalization (Supplementary Section 2.1)
and the final detection range was defined as:
Each iteration of the Monte Carlo process provided a probability of detection by the 10th
and 90th percentiles at each range from the hydrophone
Detection ranges could not be estimated for SG due to mooring-related noise
therefore the greatest values by species at GS were used for detection range (circle) visualization and analysis at SG
two observers were stationed on the research vessel deck above the navigation bridge
each scanning continuously on either side of the transect line using 7 × 50 Fujinon binoculars
To determine the position of animals sighted
radial distances to sightings were determined using the binocular’s reticles or estimating distance by eye if animals were close to the vessel
and radial angles were measured using electronic angle boards made from digital protractors
An additional observer used Fujinon 25 × 150 MTM pedestal-mounted binoculars to assist the two primary observers with species identifications and group size counts
Noise generated by the mooring during periods with tidal current was present during all months and all deployments at SG
This noise was responsible for most fluctuations of the broadband SPL
While most mooring noise occurred below 200 Hz and regularly exceeded the limits of prevailing noise in that frequency band (and rendered both blue whale automated detector and fin whale automated detector ineffective during the first deployment; 2009–2010)
it also extended to frequencies up to 8 kHz
mooring noise occurred primarily from September to January
when storms and high wind periods were more frequent
GS recordings were largely free of equipment-related noise
Due to the absence of mooring-related noise
GS was the only site where the contribution of cetacean calls to ambient noise levels was evaluated
particularly in winter due to low sun angle
Missing values were replaced with a median value calculated for each site
All analyses were performed in R Studio (Version 1.4.1106) statistical software platform
Spectrograms of four mysticetes and six odontocetes calls detected in Gwaii Haanas
spectrograms (all: Hamming window) are plotted on different time and frequency scales
Blue whale (Gowgaia Slope): infrasonic (A–B) vocalizations (0.4 Hz frequency resolution
Fin whale (Gowgaia Slope): 20-Hz pulses (0.25 Hz frequency resolution
Gray whale (Ramsay Island): vocalizations; Humpback whales (SG̱ang Gwaay): song vocalizations; Killer whales (SG̱ang Gwaay): vocalizations (all 2 Hz frequency resolution
Sperm whale (Gowgaia Slope): clicks (64 Hz frequency resolution
Cuvier’s and Baird’s beaked whales (Gowgaia Slope): clicks (512 Hz frequency resolution
Porpoises and Dolphin (unidentified delphinid click trains) (Gowgaia Slope): clicks (64 Hz frequency resolution
Gray whale vocalizations were sparsely detected during manual review at RI
No right whale (Eubalaena japonica) vocalizations were found during the systematic manual analysis of the recordings at all locations
although undetected right whale tonals could have been missed during bouts with humpback songs
no minke whale (Balaenoptera acutorostrata) vocalizations were found during the systematic manual analysis of recordings at any location
No sei whale (Balaenoptera borealis) vocalizations could be identified with confidence
These species are not likely to have contributed much to the overall delphinid occurrence
but cannot be ruled out without further analysis
Dolphin whistles could not be identified at SG where the recording frequency could not capture these signals
Blue, fin, humpback, and sperm whales showed strong seasonal acoustic presence (Figs. 4, 5, 6, 8). Humpback whales and delphinids showed strong diel acoustic presence with more activity at night; killer whales showed a similar but more moderate diel pattern (Fig.S4)
Daily and hourly occurrence of blue whale vocalizations at SG̱ang Gwaay and Gowgaia Slope in 1-h bins
Red lines indicate recorder deployment and retrieval dates
Dark grey areas indicate a lack of recordings
Light grey shaded areas indicate hours of darkness
Fin whale vocalizations were most common from September to April at all three hydrophone locations (Fig. 5). The mooring-related noise at SG also limited our ability to detect fin whale vocalizations manually or automatically, therefore the SG fin whale occurrence presented here should also be taken as an underestimate.
Daily and hourly occurrence of fin whale vocalizations at SG̱ang Gwaay
Gowgaia Slope and Ramsay Island in 1-h bins
Daily and hourly occurrence of humpback whale vocalizations at SG̱ang Gwaay
Gray whale vocalizations were sparsely detected at RI on six occasions: one day in January 2019
No gray whale calls were detected in the SG or GS datasets
Daily and hourly occurrence of killer whale vocalizations at SG̱ang Gwaay
Sperm whale clicks were common at both SG and GS throughout the year, with a notable decrease through winter (Fig. 8). Lower recording sampling rate at SG limited manual detection of sperm whale clicks and rendered the automated detector ineffective; therefore, the SG acoustic occurrence is likely underestimated. No sperm whale clicks were detected at RI.
Daily and hourly occurrence of sperm whale vocalizations at SG̱ang Gwaay and Gowgaia Slope in 1-h bins
Daily and hourly occurrence of Baird’s beaked whale vocalizations at Gowgaia Slope in 1-h bins. Blue dots indicate automated detections. Black dots indicate confirmed occurrence. Red lines indicate recorder deployment and retrieval dates. Dark grey areas indicate a lack of recordings. Light grey shaded areas indicate hours of darkness.
Daily and hourly occurrence of porpoise vocalizations at Gowgaia Slope in 1-h bins
Daily and hourly occurrence of delphinid whistles at Gowgaia Slope and Ramsay Island in 1-h bins. Blue dots indicate automated detections. Black dots indicate confirmed occurrence. Red lines indicate recorder deployment and retrieval dates. Dark grey areas indicate a lack of recordings. Light grey shaded areas indicate hours of darkness.
Daily and hourly occurrence of delphinid clicks at Gowgaia Slope and Ramsay Island in 1-h bins
Passive acoustic monitoring data from the Gowgaia Slope hydrophone in 2017–2018 (left) and 2018–2019 (right)
Long-term spectrograms showing blue and fin whale call presence (top)
Proportion of hours each day with blue or fin whale calls (middle)
Sound pressure levels (SPLs) for decaband 10–100 Hz (green) and decaband 20–40 Hz (blue) (bottom)
Summary of the visual sightings (black rectangle) from visual surveys (SG̱ang Gwaay: July/August 2009 and July 2010; Ramsay Island: August 2018; Gowgaia Slope July 2018 and July 2019) and cetacean acoustic presence (black rectangle) in the area of hydrophones over the same time periods
Light brown indicates that this species is outside recorder frequency range
Green indicates that the species was likely observed (e.g.
sei whale) or cannot be acoustically distinguished to species (porpoise or delphinid groups)
and 2010 recordings overlapped only three days of visual surveys (15–17 July 2010); thus
2009 and 2010 visual surveys were included assuming that the species acoustically detected in 2010 might also be recorded in 2009 if the recording period had started in July
the 2018 hydrophone did not get deployed until after the visual surveys were completed
humpback and killer whales were detected in both acoustic and visual surveys
Pacific white-sided and Risso’s dolphin sightings were confirmed in visual surveys; delphinid clicks were acoustically present but could not be identified to species
and high frequency porpoise clicks were outside the hydrophone recording range
humpback and sperm whales were detected in both acoustic and visual surveys
Dolphin and porpoise clicks were present in the acoustic detections that could not be distinguished to species
while one Dall’s porpoise and one Pacific white-sided dolphin sighting were confirmed in visual surveys
and killer whales were acoustically detected
while one Dall’s porpoise sighting was confirmed in visual surveys
and killer whales were acoustically detected along with delphinid clicks not identified to species
no visual sightings of these species were reported
Time series of proportion of hours each day with calls for humpback
and Baird’s beaked whales (light grey line) and as an 8-day average (composite
dark grey line) over the deployment times for each hydrophone; chlorophyll a (chl-a
green line); and sea surface temperature (SST
Note that Baird’s beaked whales have a different y-axis scale due to the lower number of detections
Proportion of hours with whale calls were lowest at GS compared to other locations for humpback whales
Proportion of hours with whale calls were highest at GS compared to other locations for blue
could potentially explain some of the location differences
the proportion of hours with blue whale calls was higher at GS than SG (Dunn’s test; Z = – 2.52
P = 0.011); with fin whale calls was higher at GS than at other locations (GS vs
P = 1.53e–5); with humpback whale calls was higher at GS than at RI (Dunn’s test; Z = 2.47
P = 0.041); and with sperm whale calls was higher at GS than at SG (Dunn’s test; Z = – 6.65
Spatial and temporal patterns of cetacean habitat use vary considerably by species and remain relatively understudied in northern BC
This study represents the most comprehensive acoustic and visual study of all identifiable cetacean species on the east and west coasts of Gwaii Haanas to date
Passive acoustic monitoring provided unique
and geographic distribution that complemented vessel-based visual surveys for the east and west coasts of Gwaii Haanas
Vessel access to exposed open ocean areas like the west coast of HG is challenging
and the ability to monitor cetaceans by visual surveys off the continental shelf-slope and offshore over multiple years is costly and capacity prohibitive
results indicated acoustic and/or visual presence of twelve cetacean species or species groups
with acoustic seasonal patterns identified for five cetaceans
and acoustic diel patterns for two species and one species group
Implementation and analyses of both survey types over 4 years was made possible by Gwaii Haanas management partners working together
Collaborating to complement summer visual surveys with year-round PAM yielded a more fulsome understanding of cetacean use patterns in and adjacent to Gwaii Haanas that can be used as a ‘baseline’ to monitor future changes
Results from acoustic recordings and visual surveys largely complemented each other in documenting cetacean presence and use on the east and west coasts of Gwaii Haanas
Numerous cetacean species were detected by both acoustic monitoring and visual surveys (i.e.
while other species were detected by only one of these methods
Visual surveys resulted in many ‘likely’ and ‘unidentified’ sightings (results not shown here)
just as acoustic monitoring resulted in some calls by species groups that could not currently be identified to species (e.g.
Acoustic detections confirmed the presence and acoustic use patterns of additional species outside visual survey timeframes: blue and Cuvier’s beaked whales were detected at GS; blue whales at SG
some visually-sighted species (minke whales at RI
and potentially a sei whale at SG) were not detected in overlapping PAM datasets
Hydrophone limitations may result in discrepancies between visual and acoustic surveys
the 16 kHz recording frequency could not capture delphinid species
the 96 kHz recording frequency could not capture porpoises and/or Kogia
Since marine mammals spend most of their time below the sea surface
PAM is an important tool that facilitates the study of these animals when they are otherwise visually inaccessible
Visual surveys complemented PAM by identifying species that are sometimes not acoustically detectable or distinguishable
and covering much larger survey areas than within the acoustic detection range of a single recorder for most cetacean species
visual surveys only observed cetaceans present at the surface within the visual horizon of the survey crew
and confirmed identification was only possible with sufficient proximity to the animal when at the surface
The exact location of animals is known in visual surveys
whereas species-specific detection ranges for each hydrophone deployment means that species with smaller detection ranges (i.e.
< 20 km) such as killer whales and dolphins are more likely to be within the boundaries of Gwaii Haanas
whereas those with large detection ranges such as blue
fin and sperm whales could be inside or outside of Gwaii Haanas
all of these species contributed to the Gwaii Haanas underwater soundscape
The absence of blue whales in the summer visual surveys between July 2009 and 2019
and sporadic detection of blue whale calls over the same periods
likely reflects the seasonal movement of blue whales
blue whale call proportions stayed consistently high from mid-August through December
when the decline in blue whale call proportion was correlated with the lowest SST values
blue whale calling behavior on the west coast of Gwaii Haanas showed geographic and seasonal patterns with consistent interannual variability
and may have been present but not detected at SG and RI due to mooring-related or tidal current-induced flow noise
but visual observations of Bigg’s killer whales are the most commonly reported in HG waters (L
An ongoing study with detailed analysis of killer whale call types should provide further information (Ford et al
therefore further work is required to distinguish these species acoustically
Occasional but consistent summer sightings of Risso’s dolphins in Gwaii Haanas and HG west coast are reported
sometimes in large groups of over 100 individuals
including near RI; pods of Risso’s with young juvenile pink-faced individuals have also been reported in early spring on the east coast of Gwaii Haanas (L
Killer whale and delphinid acoustic signals occurred year-round on both the east coast and west coast of Gwaii Haanas and showed strong diel variation
Cuvier’s and Baird’s beaked whale and porpoise clicks
were the main contributors to the winter soundscape on the west coast
The analysis of effects of environmental conditions on whale acoustic detections contributed ecological insights into the presence and acoustic behavior of many cetacean species and groups in these waters
Cetaceans spend most of their time below the ocean surface
highlighting the importance of passive acoustic monitoring
as necessary tools to facilitate understanding and mapping their year-round spatial and temporal distributions
Year-round PAM data complements visual surveys that often occur in summer when weather and sea state are better for cetacean observations; PAM provides insights into seasonal and diel patterns of acoustic use throughout the year
while visual surveys can be used to estimate the density and abundance of animals and confirm spatial distributions
collaborations between management partners and external experts were and continue to be critical to successful implementation of the marine monitoring program
including the use of PAM and visual surveys for monitoring cetacean acoustic use and ocean noise
The acoustic data used for this study is available per request to Dr
The visual data used for this study is available per request to Dr
Thomas Doniol-Valcroze (Fisheries and Oceans Canada
Prioritizing global marine mammal habitats using density maps in place of range maps
Communication and acoustic behavior of dolphins and whales in Hearing by whales and dolphins 156–224 (Springer
Exploring movement patterns and changing distributions of baleen whales in the western North Atlantic using a decade of passive acoustic data
Marine mammal visual and acoustic surveys near the Alaskan Colville River Delta
Integrating passive acoustic and visual surveys for marine mammals in coastal habitats in 176th Meeting of Acoustical Society of America
Marine mammal habitat models come of age: The emergence of ecological and management relevance
Functional mechanisms underlying cetacean distribution patterns: Hotspots for bottlenose dolphins are linked to foraging
Predicting cetacean habitats from their energetic needs and the distribution of their prey in two contrasted tropical regions
Does chlorophyll a provide the best index of phytoplankton biomass for primary productivity studies?
Sea-surface temperature gradients across blue whale and sea turtle foraging trajectories off the Baja California Peninsula
Seasonal occurrence and diel calling behaviour of Antarctic blue whales and fin whales in relation to environmental conditions off the west coast of South Africa
33 (© Council of the Haida Nation and Her Majesty the Queen in Right of Canada
represented by the Chief Executive Officer of Parks Canada
The underwater acoustic environment at SGaan Kinghlas-Bowie Seamount Marine Protected Area: Characterizing vessel traffic and associated noise using satellite AIS and acoustic datasets
A split-step Padé solution for the parabolic equation method
Gaussian beam tracing for computing ocean acoustic fields
Ambient Noise and Killer Whale Monitoring near Port Metro Vancouver’s Proposed Terminal 2 Expansion Site: July–September 2012
(Technical report by JASCO Applied Sciences for Hemmera
Distribution and relative abundance of cetaceans in western Canadian waters from ship surveys
Spatial density models of cetaceans in the Canadian Pacific estimated from 2018 ship-based surveys
Satellite observations of the Northeast Pacific Ocean
Saha, K. et al. NOAA National centers for environmental information. Dataset https://doi.org/10.7289/v52j68xx (2018)
NASA Goddard Space Flight Center, Ocean Ecology Laboratory & Ocean Biology Processing Group. (NASA OB.DAAC, Greenbelt, MD, USA. https://doi.org/10.5067/AQUA/MODIS/L3B/CHL/2018
Fast stable restricted maximum likelihood and marginal likelihood estimation of semiparametric generalized linear models
R Core Team. R: A language and environment for statistical computing. R foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/ (2021)
Ogle, D. H., Wheeler, P. & Dinno, A. FSA: Fisheries Stock Analysis. R package version 0.8.32. https://github.com/droglenc/FSA (2021)
Non-song social call bouts of migrating humpback whales
Geographic variation in the whistling behavior of small odontocetes
Rankin, S., Oswald, J., Barlow, J. P. & Lammers, M. Patterned burst-pulse vocalizations of the northern right whale dolphin, Lissodelphis borealis. J. Acoust. Soc. Am. 121, 1213–1218. https://doi.org/10.1121/1.2404919 (2007)
Arranz, P. et al. Discrimination of fast click-series produced by tagged Risso’s dolphins (Grampus griseus) for echolocation or communication. J. Exp. Biol. 219, 2898–2907. https://doi.org/10.1242/jeb.144295 (2016)
First record of common bottlenose dolphin (Tursiops truncatus) in Canadian Pacific waters
Vocalizations of common minke whales (Balaenoptera acutorostrata) in an eastern North Pacific feeding ground
A preliminary assessment of the status of marine mammal populations and associated research needs for the west coast of Canada
Predictions of critical habitat for five whale species in the waters of coastal British Columbia
Ou, H., Au, W. W. L., Van Parijs, S., Oleson, E. M. & Rankin, S. Discrimination of frequency-modulated Baleen whale downsweep calls with overlapping frequencies. J. Acoust. Soc. Am. 137, 3024–3032. https://doi.org/10.1121/1.4919304 (2015)
An overview of fixed passive acoustic observation methods for cetaceans
Environmental correlates of blue and fin whale call detections in the North Pacific Ocean from 1997 to 2002
The presence of large whale species in Clayoquot Sound and its offshore waters
Acoustic and satellite remote sensing of blue whale seasonality and habitat in the Northeast Pacific
Insights into the population structure of blue whales in the Eastern North Pacific from recent sightings and photographic identification
Satellite chlorophyll off the British Columbia Coast
Factors affecting the seasonal distribution and biomass of E
spinifera along the Pacific coast of Canada: A spatiotemporal modelling approach
Blue whale habitat associations in the Northwest Pacific: Analysis of remotely-sensed data using a Geographic Information System
Review of Baleen Whale (Mysticeti) reproduction and implications for management
Growth of fin whale in the Northern Pacific Ocean
Seasonality and distribution of whale calls in the North Pacific
The 20-Hz signals of finback whales (Balaenoptera physalus)
Seasonal variability and detection range modeling of baleen whale calls in the Gulf of Alaska
Winter Behaviour and Population Structure of Fin Whales (Balaenoptera physalus) in British Columbia inferred from passive acoustic data (University of British Columbia
Patterns of Fin whale (Balaenoptera physalus) Seasonality and Relative Distribution in Canadian Pacific Waters Inferred from Passive Acoustic Monitoring
Updated Marine Mammal Distribution and Abundance Estimates in British Columbia (Springer
Identification of ecologically and biologically significant areas in the Pacific North Coast integrated management area: Phase II: Final report
movements and habitat fidelity patterns of Fin Whales (Balaenoptera physalus) in Canadian Pacific Waters
Information in Support of the Identification of Habitat of Special Importance to Fin Whales (Balaenoptera physalus) in Canadian Pacific Waters
Distribution and movements of fin whales in the North Pacific Ocean
Temporal separation of two fin whale call types across the eastern North Pacific
and sperm whales in British Columbia: an analysis of commercial whaling records
Food talks: 40-hz fin whale calls are associated with prey biomass
Migration and population structure of northeastern Pacific whales off coastal British Columbia: An analysis of commercial whaling records from 1908–1967
Distribution and abundance of marine mammals in the coastal waters of British Columbia
Distribution and relative abundance of humpback whales in relation to environmental variables in coastal British Columbia and adjacent waters
Winn, H. E. & Winn, L. K. The song of the humpback whale Megaptera novaeangliae in the West Indies. Mar. Biol. 47, 97–114. https://doi.org/10.1007/BF00395631 (1978)
Population characteristics and migration of summer and late-season humpback whales (Megaptera novaeangliae) in southeastern Alaska
North Pacific humpback whale songs: A comparison of southeast Alaskan feeding ground songs with Hawaiian wintering ground songs
Acoustic detections of singing humpback whales (Megaptera novaeangliae) in the eastern North Pacific during their northbound migration
Acoustic monitoring on a humpback whale (Megaptera novaeangliae) feeding ground shows continual singing into late spring
Humpback whale song and foraging behavior on an Antarctic feeding ground
Singing through winter nights: Seasonal and diel occurrence of humpback whale (Megaptera novaeangliae) calls in and around the Gully MPA
Information relevant for the identification of critical habitats of north pacific humpback whales (Megaptera novaeangliae) in British Columbia
Humpback Whale (Megaptera novaeangliae) observations in Laskeek Bay
Twenty years of satellite observations describing phytoplankton blooms in seas adjacent to Gwaii Haanas National Park Reserve
Gray whale Eschrichtius robustus population and stock identity
Results of cetacean observations in Laskeek Bay
New insights into the northward migration route of gray whales between Vancouver Island
The use of passive acoustic monitoring as a census tool of gray whale (Eschrichtius robustus) migration
In Physeter macrocephalus in Behavior of Marine Animals (eds Winn
Distribution and relative abundance of sperm whales in relation to key environmental features
squid landings and the distribution of other cetacean species in the Gulf of California
Sperm whale Physeter macrocephalus Linnaeus
Social organization of sperm whales off the Galapagos Islands
In Encyclopedia of Marine Mammals 3rd edn (eds Würsig
Ocean nomads: Distribution and movements of sperm whales in the North Pacific shown by whaling data and Discovery marks
Sperm whale (Physeter macrocephalus) acoustic ecology at Ocean Station PAPA in the Gulf of Alaska-Part 2: Oceanographic drivers of interannual variability
You are what you eat: Foraging specializations and their influence on the social organization and behavior of killer whales
Habitats of special importance to resident killer whales (Orcinus orca) off the West Coast of Canada
Offshore Killer whales in Canadian pacific waters: Distribution
population status and potential for recovery
Seasonal movements and foraging behaviour of northern resident killer whales (Orcinus orca) in relation to the inshore distribution of salmon (Oncorhynchus spp.) in British Columbia
Life History and Population Dynamics of Northern Resident Killer Whales (Orcinus orca) in British Columbia (Canadian Science Advisory Secretariat Ottawa
Nocturnal activity by mammal-eating killer whales at a predation hot spot in the Bering Sea
Dietary specialization in two sympatric populations of killer whales (Orcinus orca) in coastal British Columbia and adjacent waters
The mixed blessing of echolocation: Differences in sonar use by fish-eating and mammal-eating killer whales
The vocal behaviour of mammal-eating killer whales: Communicating with costly calls
Call traditions and vocal dialects of killer whales (Orcinus orca) in British Columbia Ph.D
Information in Support of the Identification of Critical Habitat for Transient Killer Whales (Orcinus orca) off the West Coast of Canada
Species-specific beaked whale echolocation signals
Two records of Berardius bairdi from the coast of British Columbia
Sightings and strandings of beaked whales on the west coast of
Jefferson, T. A. Phocoenoides dalli. Mamm. Spec. https://doi.org/10.2307/3504170 (1988)
Estimation of population size and trends for highly mobile species with dynamic spatial distributions
Abundance and depth distribution of harbor porpoise (Phocoena phocoena) in northern California determined from a 1995 ship survey
Clicking in a killer whale habitat: Narrow-band
high-frequency biosonar cliks of harbour porpoise (Phocoena phocoena) and Dall’s porpoise (Phocoenoides dalli)
Porpoise clicks from a sperm whale nose—Convergent evolution of 130 kHz pulses in toothed whale sonars?
Detection and classification of narrow-band high frequency echolocation clicks from drifting recorders
Prey dynamics affect foraging by a pelagic predator (Stenella longirostris) over a range of spatial and temporal scales
Dusky dolphin (Lagenorhynchus obscurus) foraging in two different habitats: active acoustic detection of dolphins and their prey
Spatial and temporal patterns of Risso’s dolphin echolocation in the Southern California Bight
Spatio-temporal comparison of Pacific white-sided dolphin echolocation click types
The relationship of midwater trawl catches to sound scattering layers off the coast of northern British Columbia
Low-frequency ambient sound in the North Pacific: Long time series observations
Acoustic models of sound production and propagation in Hearing by whales and dolphins 409–469 (Springer
The acoustic calls of blue whales off California with gender data
Source levels of fin whale 20 Hz pulses measured in the Northeast Pacific Ocean
Vihtakari, M. ggOceanMaps: Plot Data on Oceanographic Maps using 'ggplot2'. R package version 1.2.14. https://mikkovihtakari.github.io/ggOceanMaps/ (2022)
Download references
Our deep appreciation to the Gwaii Haanas Archipelago Management Board
Passive acoustic monitoring data collection and analyses were funded by the Gwaii Haanas marine monitoring program
the Parks Canada Species at Risk Action Plan Implementation Funds
Gowgaia Slope recorder calibration support was provided by Caitlin O’Neill and Peter Van Buren of the Ocean Acoustics Program at the Institute of Ocean Sciences
We are grateful for assistance in mooring logistics provided by Thomas Juhasz
Stephen Page and Roger Savoie of the DFO Ocean Sciences Division at the Institute of Ocean Sciences
as well as Ali Bowker of the DFO Cetacean Research Program
Visual surveys were supported by the survey fund of the DFO Center of Expertise in Marine Mammalogy
the Ocean Protection Plan Baseline program
as well as in-kind support by the Science Branch in the Pacific Region
We are grateful to all the marine mammal observers over the years and we thank numerous officers and crews of the Canadian Coast Guard for their dedication to support science work
Thanks to Andrea Hilborn (DFO) for providing the 8-day composite SST and chl-a data
and Julien Delarue from JASCO for reviewing some cetacean detections
We are grateful to the three anonymous reviewers of this manuscript for relevant comments and revision
Present address: University Corporation for Atmospheric Research (UCAR|CPAESS)
under contract to National Oceanic and Atmospheric Administration
Under Contract to National Oceanic and Atmospheric Administration
These authors contributed equally: Héloïse Frouin-Mouy and Xavier Mouy
proposed and conducted the analyses and prepared the manuscript
oversaw collection of the visual survey data and prepared the data for analysis
oversaw all aspects of acoustic data collection and field work
proposed the analyses and funded acoustic data collection and analyses
All authors provided feedback on the analyses and reviewed the manuscript
The authors declare no competing interests
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations
Download citation
DOI: https://doi.org/10.1038/s41598-022-22069-4
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Sign up for the Nature Briefing: Anthropocene newsletter — what matters in anthropocene research, free to your inbox weekly.
Volume 5 - 2024 | https://doi.org/10.3389/frsen.2024.1439995
Many species of fishes around the world are soniferous
The types of sounds fishes produce vary among species and regions but consist typically of low-frequency (<1.5 kHz) pulses and grunts
These sounds can potentially be used to monitor fishes non-intrusively and could complement traditional monitoring techniques
the significant time required for human analysts to manually label fish sounds in acoustic recordings does not yet allow passive acoustics to be used as a viable tool for monitoring fishes
we compare two different approaches to automatically detect fish sounds
One is a more traditional machine learning technique based on the detection of acoustic transients in the spectrogram and the classification using Random Forest (RF)
The other is using a deep learning approach and is based on the classification of overlapping segments (0.2 s) of spectrogram using a ResNet18 Convolutional Neural Network (CNN)
Both algorithms were trained using 21,950 manually annotated fish and non-fish sounds collected from 2014 to 2019 at five different locations in the Strait of Georgia
The performance of the detectors was tested on part of the data from the Strait of Georgia that was withheld from the training phase
The CNN performed up to 1.9 times better than the RF (F1 score: 0.82 vs
the CNN was able to find more faint fish sounds than the analyst and performed well in environments different from the one it was trained in (Miami F1 score: 0.88)
Noise analysis in the 20–1,000 Hz frequency band shows that the CNN is still reliable in noise levels greater than 130 dB re 1 μPa in the Port of Miami but becomes less reliable in Barkley Sound past 100 dB re 1 μPa due to mooring noise
The proposed approach can efficiently monitor (unidentified) fish sounds in a variety of environments and can also facilitate the development of species-specific detectors
an easy-to-use open-source implementation of the CNN detector with detailed documentation
and found that the recorded levels were linearly related to fish density during the peak spawning period
While passive acoustics shows great promise for monitoring fish populations
it is still largely limited by knowledge gaps about the vocal repertoire of many fish species
Many of these studies target particular species and focus on specific regions
there is a need to develop a generic fish sound detector that is species agnostic and can detect individual sounds (i.e.
and be used in a wide variety of environments
The objective of this study is to develop automatic fish sound detectors that can be used to efficiently analyze large passive acoustic datasets
We implement two different methods and evaluate how a deep learning approach performs compared to a more traditional machine learning approach
We quantify the performance of the detector using data from two different marine environments with completely different fish communities: British Columbia
Two different fish sounds detection approaches are investigated. One is based on random forest (RF) classification, a traditional machine learning technique (Section 2.3). The other is using a convolutional neural network (CNN) which is a newer deep learning technique (Section 2.4). Both approaches use the spectrogram representation of the acoustic signal (Section 2.2)
some sounds may have been mislabeled as fish sounds
Data were collected using either SoundTrap STD300 (Ocean Instruments) or AMAR (Autonomous Multichannel Acoustic Recorder
hydrophones were placed near the seafloor (<1 m) and in water depths less than 20 m
except for Delta Node which had a water depth of 150 m
Recorders were set with different sampling frequencies but all acquired data up to a frequency of at least 16 kHz (i.e.
minimum sampling frequency of 32 kHz)
Description of Dataset 1 collected in the Strait of Georgia
Map of the sampling locations for Dataset 1 (black dots) and Dataset 2 (red star) collected in British Columbia
The red rectangle in (A) indicates the zoomed-in area represented in (B)
NC-RCA In and NC-RCA Out indicate recorders deployed inside and outside the Northumberland Channel Rockfish Conservation Area
Recordings used to create this noise dataset were chosen so it would include a large variety of sounds such as noise from vessels
All fish annotations are labelled as such (“fish”), while all non-fish annotations are grouped into the label “noise” (Table 1)
The entire dataset includes 21,032 fish annotations and 7,323 noise annotations
is composed of 670 audio files (each being either 5-min or 30-min long depending on the deployments) and represents a total of 133.75 h of accumulated acoustic recordings
Map of the sampling location for Dataset 3 collected in the port of Miami
The red rectangle in panel (A) indicates the zoomed-in area represented in panel (B)
The acoustic recorder was located near the Coral City Camera
spectrogram calculation and denoising (equalization) is the first processing step
The spectrograms were calculated using 0.064 s long frames
This resolution was selected as it can represent well the different types of fish sounds (i.e.
Given that all fish sounds of interest in this study have frequencies below 1.2 kHz
the spectrogram is truncated to only keep frequencies from 0 to 1.2 kHz
Magnitude values are squared to obtain energy and expressed in decibels
To improve the signal-to-noise ratio of fish sounds and attenuate tonal sounds from vessels
the spectrogram is equalized using a median filter
for each row (frequency) of the spectrogram
where S[t,f] is the original spectrogram and Smed[t,f] is the median spectrogram calculated as:
where the median is calculated on a window centered on the tth sample and has a duration of 2k+1 bins. Figure 3A shows the equalized spectrogram
we choose a median window equivalent to a 3 s duration (k=150)
which removes constant tonal components from vessels without removing the longer grunting sounds from fish
Illustration of the detection process on a recording containing three fish sounds
Red boxes indicate the time and frequency boundaries of each detected event
The first approach implemented to detect fish sounds is based on the RF classification algorithm
It consists of 1) segmenting the spectrogram to detect acoustic transients
2) extracting features for each detected event
and 3) classifying each event using a binary (“fish” vs
Once the spectrogram is calculated and equalized, it is segmented by calculating the local energy variance on a two-dimensional (2D) kernel of size ΔT×ΔF. The resulting matrix Svar (Figure 3B) is defined as
where μ is the mean over the 2D kernel:
Red and black crosses denote the median and peak frequency of each time slice of the spectrogram
The white box indicates the 95% energy area over which the spectrogram features were calculated
Description of the features calculated for each detection
the random subset of features used for each splits was set to 6
no transfer learning) using the training dataset
The ResNet is run on overlapping slices of spectrogram and provides a classification score betweeen 0 and 1 to indicate the probability that the slice analyzed contains a fish sound
the classification is performed for every 0.01 s of recording on 0.2 s-long spectrogram slices (with a frequency band of 0–1.2 kHz)
Spectrogram slices with a classification score exceeding the user-defined threshold are considered fish sound detections
Consecutive detections are merged into a single detection and its final classification score is the maximum score of the merged detections
To ensure that the spectrogram slices presented to the CNN always have the same size (20×78 bins)
all recordings analyzed are first downsampled to 4 kHz (i.e.
Training was conducted using a NVidia A100SXM4 (40 GB memory) graphical processing unit (GPU) and was performed with a batch size of 32 over 50 epochs. Network weights were optimized to maximize the F1 score using the ADAM optimizer (Kingma and Ba, 2014) set with its default parameters (learning rate = 0.001
To increase the quantity and variability of the training samples
This consisted of creating multiple instances of the same selection by stepping in time
both forward and backward from the middle point of the original annotation
Augmented samples were created by shifting 0.2 s long windows by 0.1 s increments and by ensuring that each of the created samples overlapped in time by at least 90% with the original annotation
Random forest and CNN models were trained and tested by dividing annotated sounds of Dataset 1 (Section 2.1.1) into two subsets
One was composed of 75% of the entire dataset and was used to train the classification models
was used to evaluate the performance of the selected model
These two subsets were carefully defined so annotations from each subset were separated by at least 6 hours
had the two classes (fish and noise) equally represented
and had a similar representation of all deployments
Data used for testing the performance of the classification were not used for training the models
In addition to being tested on part of Dataset 1
the detectors were tested on Datasets 2 and 3
Testing performance on Dataset 2 provides information on how well detectors perform on sounds from similar fish species
Testing performance on Dataset 3 provides information on how versatile the detectors are to new environments and quantifies their ability to detect sounds from fish species they were not trained on
The decisions generated from the detectors can be categorized as follows
• True positives (TP): A fish sound correctly classified as a fish sound;
• False positives (FP): Noise classified as a fish sound (i.e.
• False negatives (FN): A fish sound classified as noise (i.e.
are compared with the automated detections
To assess the performance of the detectors
precision (P) and recall (R) metrics are calculated based on the numbers (N) of TPs
where P measures exactness and R measures completeness
a P of 0.9 means that 90% of the detections classified as fish sounds are in fact fish sounds but says nothing about whether all sounds in the dataset are identified
An R of 0.8 means that 80% of all fish sounds in the dataset are correctly classified
but says nothing about how many classifications are wrong
a perfect classifier would have P and R equal to 1
Neither P nor R alone can describe the performance of a detector/classifier on a given dataset; both metrics are required
The F score is also used to quantify classifier performance
The F score measures the accuracy of the detector and varies from 0 to 1
where an F score of 1 corresponds to a perfect detector
where β is the relative weight between the recall and precision
A β of two means the recall has twice the weight of the precision
a β of 0.5 means the recall has half the weight of the precision
it is considered that P and R are equally important
Note that we did not assess the performance for each sound type separately (i.e.
tones) as the annotation datasets were not labelled to the sound type level
All classifiers used in this study provide binary classification results (i.e.
“fish” or “noise”) as well as a confidence of classification between 0 and 1
The latter can be used to adjust the sensitivity of a classifier
Accepting classification results with a low confidence leads to detecting more fish sounds (high recall)
but also generates more false alarms (low precision)
only accepting classification results with a high confidence leads to detecting fewer fish sounds (low recall)
but also results in fewer false alarms (high precision)
The optimum confidence threshold is considered as the one providing the highest F score
It is defined experimentally by iteratively calculating the performance for small increments (here 0.001) of confidence threshold values from 0 to 1
Detector performance is characterized for different signal-to-noise ratios (SNR)
The SNR of an annotated fish sound is defined as the ratio of the signal power (Ps) to the noise power corrupting the signal (Pn)
The SNR compares the level of the desired signal (i.e.
a fish sound) to the level of the background noise
the less obtrusive the background noise is
Ps is the average power of the fish sound over the duration in seconds
of the sound in the frequency band defined by the analyst (i.e.
frequency boundaries of the annotation box); Pn is the average of the power within the same frequency band d/2 seconds before and after the fish sound
Both Ps and Pn are calculated from the waveform filtered in the frequency band defined by the analyst who generated the annotations (10th order Butterworth bandpass filter)
FishSound Finder is documented and includes tutorials for users not familiar with the python language
This section summarizes the performance results of the RF and the CNN on the three different datasets
Increasing the number of noise examples from 20,000 to 40,000 raises the maximum F1 score by 0.03 (from 0.82 to 0.85) but has no noticeable effect when increasing from 40,000 to 198,461 noise examples
The best CNN performance is achieved by the model trained with the entire noise dataset
The best F1 score for that model is reached with a detection threshold of 0.99 and has a precision of P=0.85 and a recall of R=0.84
the best CNN model performs 1.2 times better than the best RF model
Performance of the RF (blue lines) and CNN (black lines) on the Strait of Georgia dataset (Dataset 1)
dashed and full blue lines represent the performance of the RF with 5
dashed and full black lines represent the performance of the CNN trained with 20,000
the CNN can detect very faint fish sounds with a precision of P=0.74 and a recall of R=0.72; faint fish sounds with a precision of P=0.77 and a recall of R=0.72; and loud fish sounds with a precision of P=0.82 and a recall of R=0.83
Performance of the best CNN by SNR interval
(A) Distribution of the SNR of fish sounds in the test dataset
(B) Precision-recall curve for very faint (SNR<3.4 dB)
faint (3.4 dB≤SNR≤ 6.3 dB) and loud (SNR>6.3 dB) fish sounds
(C–E) show example spectrograms of very faint
(A) Precision-recall curves for the CNN on the original dataset (dashed line)
for the CNN on the re-annotated dataset (solid line)
and for RF on the re-annotated dataset (dotted lines)
(B) Spectrogram of a recording containing eight fish sounds
The red box indicates the fish sound initially annotated by the analyst and the black boxes show the fish sounds that were detected by CNN
(C) Example of mooring noise triggering most of the false alarms of the CNN
Performance of the RF on this dataset is substantially lower than on Dataset 1 with a maximum F1 score F1=0.43 (P = 0.38, R=0.51, threshold = 0.74, dotted line in Figure 7)
The difference in performance between the CNN and RF is more pronounced than for Dataset 1
with a best F1 score for the CNN 1.9 higher than for RF
(A) Precision-recall curve of the CNN on Dataset 3
(B) Example of fish sounds detected by the CNN
This example includes 11 fish sounds: nine are detected by the CNN but 2 fainter fish sounds are missed (at t= 5 s and t= 7.9 s)
(C) Example of higher frequency fish pulse trains missed by the CNN (t= 2 s and t= 6.1 s)
Black boxes indicate detections from the CNN
For the Port of Miami dataset, SPLs in the 20–1,000 Hz frequency band range from 100 to 140 dB re 1μPa. Despite higher SPLs than the Barkley Sound dataset, the overall number of fish detections per minute from the CNN correlates well with the number of fish sounds found by the analyst (R2= 0.94, Figure 9D), even for noise levels greater than 120 dB re 1μPa (R2= 0.82, Figure 9F)
We implemented and compared two approaches to detect fish sounds. One is based on RF, which is a traditional classification machine learning method that has been successful in previous bioacoustic classification tasks. The other is based on a deep neural network architecture which is a technique that recently outperformed more traditional classification methods (Shiu et al., 2020)
Methods like RF require defining a set of features that represent the signal of interest and are used to discriminate between the different sound classes (e.g.
This set of features is typically defined (“hand crafted”) by domain experts who understand which features are the most discriminative
These features can be hard to define and may be highly dependent on noise conditions
poorly chosen features will result in poor classification performance
bypass this step and consider the definition of salient signal features as part of the training process
The first convolutional layers of the CNN are responsible for finding the salient features of the signal (filters) that maximize classification success
Both the features definition and the classification are optimized in unison and are learned directly from the data
We found that CNN performs substantially better than RF on all datasets
Increasing the number of trees in the RF model increased the classification performance but not enough to outperform the CNN
The decrease of 27% in F1 score between the Strait of Georgia (where the model was trained) and Barkley Sound indicates that the RF model does not generalize well enough and is not adaptable to new acoustic environments
the CNN had a satisfactory performance on all datasets
Training the CNN using more noise examples than fish sounds (i.e.
imbalanced training dataset) improved the classification performance
Despite being trained on data from the Strait of Georgia
the CNN performed well on the Barkley Sound data
This result indicates that the model generalized well and was reliable in environments with different noise conditions
This is further demonstrated by the consistent performance of the CNN on data from the Port of Miami
which include high noise levels due to intense vessel traffic
While the Strait of Georgia and Barkley Sound share a large number of similar fish species
the Port of Miami has tropical fish species that are very different from the Canadian datasets
The consistent performance of the CNN in the Port of Miami shows that the model not only learned to recognize Canada-specific fish sounds
but also learned general fish sound characteristics which can be applied in different ecosystems
The calculated performance of automatic detectors and classifiers depends strongly on the datasets used to both train and test the algorithms
Evaluating algorithms on small datasets (e.g.
several hundred sounds collected over a few days)
and recording platforms do not change or are very stable and predictable
which may not be representative of how these algorithms would behave when applied to large continuous passive acoustic datasets
The large dataset we use in this work is comprised of more than 53,000 fish and noise sounds collected over eight different sites in both Canada and the United States
providing confidence that the detector characterized in this paper would behave similarly in other areas
this “breaking point” is not always the same and needs to be defined for each dataset analyzed
Fish sounds detections at the Port of Miami
(A) Time series of the number of fish sounds detected per hour on the 8 days of continuous data collected at the Port of Miami (June 7–16
Red horizontal bars indicate the times at which the instrument was deployed and retrieved
(B) Box plot showing the distribution of number of fish sounds for each hour of the day
Because the CNN detector we provide is not species-specific, it can be used to study and discover general fish occurrence patterns in new environments, help annotate fish sounds in tank or in-situ studies, or be deployed on audio-video systems (e.g., Mouy et al., 2023) to help identify new fish sounds
One reason fish sounds are underused in marine conservation is because the analysis tools developed by engineers and scientists are not always made easily accessible to other researchers in the marine conservation field
we implemented the CNN detector in the easy-to use software FishSound Finder
which is released under an open source license and is accompanied by a step by step tutorial showing how to use it
and improved by other researchers in the community
The CNN detector is implemented in the python software FishSound Finder that can be found on GitHub (https://github.com/xaviermouy/FishSound_Finder)
The datasets presented in this study can be made available upon request to the corresponding author
The author(s) declare that financial support was received for the research
This research was funded by Fisheries and Oceans Canada’s (DFO) Competitive Science Research Fund (CSRF) and Strategic Program for Ecosystem-based Research and Advice
Ecosystem Stressors and Aquatic Invasive Species (SPERA)
and the Natural Sciences and Engineering Research Council (NSERC) Canadian Healthy Oceans Network and its Partners: DFO and INREST (representing the Port of Sept-Iles and City of Sept-Iles)
Field expenses and equipment costs were funded by a NSERC Discovery grant
XM was also partly supported by a NSERC Postgraduate Scholarship
We would like to thank Ocean Networks Canada for providing data from the VENUS cabled observatory (Delta Node) and for providing access to the Advanced Research Computing platform from the Digital Research Alliance of Canada which was instrumental in the training of the deep learning models
Thanks to Jason Gedamke (NOAA Fisheries) for sharing his recorder for the Port of Miami deployment and Katrina Nikolich (University of Victoria) for sharing data from her MSc and PhD
and Erik SA (DFO) for manually annotating fish sounds in several of the passive acoustic datasets used in this study
Thanks to David Hannay and Joann Nippard (JASCO Applied Sciences) for their administrative help during this project
Christian Carrera for his help preparing the equipment
and Harald Yurk (DFO) for reviewing an earlier version of this manuscript
The authors declare that the research was conducted in the absence of commercial or financial relationships that could be construed as a potential conflict of interest
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations
Any product that may be evaluated in this article
or claim that may be made by its manufacturer
is not guaranteed or endorsed by the publisher
Automated classification of bird and amphibian calls using machine learning: a comparison of methods
CrossRef Full Text | Google Scholar
Sound production during competitive feeding in the grey gurnard
CrossRef Full Text | Google Scholar
“Fish sounds and mate choice,” in Sound commun
CrossRef Full Text | Google Scholar
Applications of machine learning to identify and characterize the sounds produced by fish
CrossRef Full Text | Google Scholar
“Vocal–acoustic communication: from neurons to behavior,” in Fish bioacoustics
CrossRef Full Text | Google Scholar
Empirical comparison of voting classification algorithms: bagging
CrossRef Full Text | Google Scholar
CrossRef Full Text | Google Scholar
CrossRef Full Text | Google Scholar
Dask Development Team (2016). Dask: library for dynamic task scheduling. Available at: https://dask.org
Google Scholar
Generalization of spectral flatness measure for non-Gaussian linear processes
CrossRef Full Text | Google Scholar
Automatic detection of marine mammals using information entropy
PubMed Abstract | CrossRef Full Text | Google Scholar
Assessing trends in the density of Atlantic croaker (Micropogonias undulatus): a comparison of passive acoustic and trawl methods
Google Scholar
Detection and classification of right whale calls using an edge detector operating on a smoothed spectrogram
Google Scholar
Google Scholar
The plainfin midshipman’s soundscape at two sites around Vancouver Island
CrossRef Full Text | Google Scholar
Automatic detection of fish sounds based on multi-stage classification including logistic regression via adaptive feature weighting
PubMed Abstract | CrossRef Full Text | Google Scholar
PubMed Abstract | CrossRef Full Text | Google Scholar
“Deep residual learning for image recognition,” in Proceedings of the IEEE conference on computer vision and pattern recognition
CrossRef Full Text | Google Scholar
An approach for automatic classification of grouper vocalizations with passive acoustic monitoring
PubMed Abstract | CrossRef Full Text | Google Scholar
“Batch normalization: accelerating deep network training by reducing internal covariate shift,” in International conference on machine learning (pmlr)
Google Scholar
Multiple sound-producing mechanisms in teleost fishes and hypotheses regarding their behavioural significance
CrossRef Full Text | Google Scholar
Soundscape learning: an automatic method for separating fish chorus in marine soundscapes
PubMed Abstract | CrossRef Full Text | Google Scholar
Adam: a method for stochastic optimization
Google Scholar
Ketos—a deep learning package for creating acoustic detectors and classifiers
CrossRef Full Text | Google Scholar
Performance of a deep neural network at detecting North Atlantic right whale upcalls
PubMed Abstract | CrossRef Full Text | Google Scholar
Automatic data selection for validation: a method to determine cetacean occurrence in large acoustic data sets
PubMed Abstract | CrossRef Full Text | Google Scholar
“Agonistic behaviour and acoustic communication,” in Communication in fishes
Moller (United States: Science Publishers)
Google Scholar
On the reliability of acoustic annotations and automatic detections of antarctic blue whale calls under different acoustic conditions
PubMed Abstract | CrossRef Full Text | Google Scholar
Google Scholar
Improving biodiversity assessment via unsupervised separation of biological sounds from long-duration recordings
PubMed Abstract | CrossRef Full Text | Google Scholar
Comparison of passive acoustic soniferous fish monitoring with supervised and unsupervised approaches
PubMed Abstract | CrossRef Full Text | Google Scholar
CrossRef Full Text | Google Scholar
A quantitative inventory of global soniferous fish diversity
CrossRef Full Text | Google Scholar
Looby, A., Riera, A., Vela, S., Cox, K., Bravo, S., Rountree, R., et al. (2021). FishSounds. Available at: https://fishsounds.net
Google Scholar
Identifying sciaenid critical spawning habitats by the use of passive acoustics
CrossRef Full Text | Google Scholar
Macaulay, J. (2021). SoundSort. Available at: https://github.com/macster110/aipam
Google Scholar
PubMed Abstract | CrossRef Full Text | Google Scholar
“Active and passive acoustics to locate and study fish,” in Fish bioacoustics
CrossRef Full Text | Google Scholar
Passive acoustic detection of sounds produced by the damselfish
CrossRef Full Text | Google Scholar
Acoustic measurement of marine mammal sounds in noisy environments
Google Scholar
Recognizing transient low-frequency whale sounds by spectrogram correlation
PubMed Abstract | CrossRef Full Text | Google Scholar
PubMed Abstract | CrossRef Full Text | Google Scholar
Long-term monitoring of captive red drum Sciaenops ocellatus reveals that calling incidence and structure correlate with egg deposition
PubMed Abstract | CrossRef Full Text | Google Scholar
Swimming sounds and the schooling of fishes
CrossRef Full Text | Google Scholar
Mouy, X. (2021). Ecosound bioacoustic toolkit. Available at: https://ecosound.readthedocs.io
Google Scholar
Identification of fish sounds in the wild using a set of portable audio-video arrays
CrossRef Full Text | Google Scholar
JMesh-A scalable web-based platform for visualization and mining of passive acoustic data
CrossRef Full Text | Google Scholar
“Passive acoustic monitoring of marine mammals in the Arctic,” in Detect
Google Scholar
Cataloging fish sounds in the wild using combined acoustic and video recordings
PubMed Abstract | CrossRef Full Text | Google Scholar
Machine learning analysis reveals relationship between pomacentrid calls and environmental cues
CrossRef Full Text | Google Scholar
“Rectified linear units improve restricted Boltzmann machines,” in Proceedings of the 27th international conference on machine learning (Madison
Google Scholar
Quantitative classification of harbor seal breeding calls in Georgia Strait
PubMed Abstract | CrossRef Full Text | Google Scholar
The sources and prevalence of anthropogenic noise in rockfish conservation areas with implications for marine reserve planning
PubMed Abstract | CrossRef Full Text | Google Scholar
Automatic taxonomic classification of fish based on their acoustic signals
CrossRef Full Text | Google Scholar
Geographical variation in sound production in the anemonefish Amphiprion akallopisos
PubMed Abstract | CrossRef Full Text | Google Scholar
Sounding the call for a global library of underwater biological sounds
CrossRef Full Text | Google Scholar
Google Scholar
Google Scholar
real-time object detection,” in Proceedings of the IEEE conference on computer vision and pattern recognition
CrossRef Full Text | Google Scholar
Evolutionary patterns in sound production across fishes
CrossRef Full Text | Google Scholar
Sablefish (Anoplopoma fimbria) produce high frequency rasp sounds with frequency modulation
PubMed Abstract | CrossRef Full Text | Google Scholar
Gaussian mixture model classification of odontocetes in the Southern California Bight and the Gulf of California
PubMed Abstract | CrossRef Full Text | Google Scholar
Random Forest for improved analysis efficiency in passive acoustic monitoring
CrossRef Full Text | Google Scholar
Listening to Fish: applications of passive acoustics to fisheries science
doi:10.1577/1548-8446(2006)31[433:ltf]2.0.co;2
CrossRef Full Text | Google Scholar
Estimating fish abundance at spawning aggregations from courtship sound levels
PubMed Abstract | CrossRef Full Text | Google Scholar
Sound production as an indicator of red hind density at a spawning aggregation
CrossRef Full Text | Google Scholar
Acoustic analysis of big ocean data to monitor fish sounds
CrossRef Full Text | Google Scholar
Deep neural networks for automated detection of marine mammal species
PubMed Abstract | CrossRef Full Text | Google Scholar
A complexity-entropy based approach for the detection of fish choruses
CrossRef Full Text | Google Scholar
“Passive acoustic detection of modulated underwater sounds from biological and anthropogenic sources,” in Ocean
CrossRef Full Text | Google Scholar
Topological structural analysis of digitized binary images by border following
CrossRef Full Text | Google Scholar
Mechanisms for directional hearing in the sea catfish (Arius felis)
PubMed Abstract | CrossRef Full Text | Google Scholar
Automated detection and localization of bowhead whale sounds in the presence of seismic airgun surveys
PubMed Abstract | CrossRef Full Text | Google Scholar
Automatic grunt detector and recognizer for Atlantic cod (Gadus morhua)
PubMed Abstract | CrossRef Full Text | Google Scholar
Applying artificial intelligence methods to detect and classify fish calls from the northern gulf of Mexico
CrossRef Full Text | Google Scholar
Pacific and Atlantic herring produce burst pulse sounds
PubMed Abstract | CrossRef Full Text | Google Scholar
Van Parijs S and Haggarty D (2024) Automatic detection of unidentified fish sounds: a comparison of traditional machine learning with deep learning
Received: 28 May 2024; Accepted: 05 August 2024;Published: 22 August 2024
Copyright © 2024 Mouy, Archer, Dosso, Dudas, English, Foord, Halliday, Juanes, Lancaster, Van Parijs and Haggarty. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use
distribution or reproduction in other forums is permitted
provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited
in accordance with accepted academic practice
distribution or reproduction is permitted which does not comply with these terms
*Correspondence: Xavier Mouy, eGF2aWVyLm1vdXlAb3V0bG9vay5jb20=
Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations
Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher
94% of researchers rate our articles as excellent or goodLearn more about the work of our research integrity team to safeguard the quality of each article we publish
A 10-year-old missing woman case has been solemnly solved
Two rabbit hunters stumbled across human remains off of East Stagecoach Trail on Feb
the State Medical Examiner’s Office in Charlotte identified the remains as Mouy Tang
was last seen walking along a trail that lead from Unique Living assisted care where she was living in Lawndale to a mini-mart just up the road that residents were known to visit
Tang would be reunited safely with her family
that did not happen," Cleveland County Sheriff Alan Norman said
The medical examiner found no evidence of foul play
and the Sheriff's Office is treating this as a accidental death
The initial search for Tang was frantic due to her multiple medical issues
Tang originally came to Hendersonville with her family in 1980
after her family escaped war-torn Cambodia
Tang lived at Unique Living assisted care for 16 years until she wandered off
Other residents of Unique Living went missing from the facility for days at a time
and two died inside the home from reported accidents
He had diabetes and a history of strokes and heart problems and died of natural causes
A month after Whitesides’ body was located
another resident signed himself out and didn’t return
The assisted care operation was closed by the state of North Carolina shortly after her disappearance
Her remains were found less than a mile from the rest home in heavy underbrush
the Tang family has done everything within their power to assist in the return of their loved one
the family was instrumental in the closure of Unique Living Rest Home
which had been the target of many community complaints over the years of its operation," Norman said
"We know that the discovery of her remains will allow her family some closure."
Sign In
Subscribe Now
Local Journalism Initiative Reporter CANADA’S NATIONAL OBSERVER
New technology is allowing researchers to covertly monitor
record and identify the sounds fish make underwater to try to unravel their deepest secrets
a recent PhD graduate at the University of Victoria
and his colleagues have devised a relatively low-cost portable audio-visual system that surreptitiously records the surprising range of acoustics fish produce
Being able to do so is akin to identifying who is attending a noisy party in a dark room
with important outcomes for scientific monitoring
fisheries and conservation in marine environments
It’s no secret that fish as a whole make a bunch of different noises
But there are huge gaps in scientific understanding around the vocal repertoire of individual fish
“If you tell people that you use acoustics as a new way to detect fish in the ocean
that’s not exactly groundbreaking,” Mouy told Canada’s National Observer
“We realized we had to go deeper to know which fish is present
which travels quickly and clearly in water
is an important form of communication for ocean creatures that may not be able to rely on what they can see in aquatic surroundings that often have little or no light
who now works with the National Oceanic and Atmospheric Administration (NOAA) in the U.S
record and study whales is a long-standing practice because they are large
charismatic animals that are more easily identified
It was while researching marine mammals that his interest was piqued by the mysterious racket being generated by fish
“When I was working in coastal British Columbia
I could hear so many fish all over and all year round
but nobody could tell me what species they were,” Mouy said
I decided to develop instrumentation to figure it out.”
Mouy devised three types of mobile systems for use in different habitats equipped with multiple cameras and hydrophones
The acoustics triangulate the location of a fish making sounds and match it with the video data to confirm the species
After testing the technology at four locations in waters around southern Vancouver Island
the team identified sounds made by three fish species — the quillback rockfish
the copper rockfish and lingcod — which hadn’t been identified as soniferous
It’s an exciting find with big implications
The quillback is a threatened species and knowing the noises it makes will go a long way to determining what habitat it prefers and thrives in
has more than 160 rockfish conservation areas
and as the acoustic catalogue for various fish improves
so can scientific evaluation of whether a protected area is meeting its objectives or not
Snagging the sounds lingcod produce is also a great catch
so it will likely be useful for management or conservation.”
The new sounds will soon be added to a publicly accessible “acoustic library” developed by other researchers at UVic called FishSounds.net
The acoustic fish surveillance system he designed has some advantages over other research methods
Recordings of specific species can be done in tanks
but fish often don’t vocalize in captivity
it isn’t necessarily representative of the types of sound they might make in natural environments
the quality of acoustics recorded in a tank is poor
“It’s distorted because the sound bounces on the walls
It’s like you would record someone in a church
some of which can be deployed for two weeks at a time
is also more effective than costly and infrequent diver-based surveys typically used to monitor fish populations
Mouy has published the building instructions for the systems and made all related materials and processing software open source
so other scientists can add to and improve his work and further contribute to global research on fish acoustics
“I want to avoid that someone has to reinvent the wheel,” he said
“If we want fish acoustics to be really useful
we all really need to share and work as a team and identify these sounds worldwide.”
READ MORE: Fish sounds could help scientists understand their ocean world
READ MORE: University of Victoria researcher helps build global inventory of fish grunts, squeals
Dialogue and debate are integral to a free society and we welcome and encourage you to share your views on the issues of the day. We ask that you be respectful of others and their points of view, refrain from personal attacks and stay on topic. To learn about our commenting policies and how our community-based moderation works, please read our Community Guidelines
This website is using a security service to protect itself from online attacks
The action you just performed triggered the security solution
There are several actions that could trigger this block including submitting a certain word or phrase
You can email the site owner to let them know you were blocked
Please include what you were doing when this page came up and the Cloudflare Ray ID found at the bottom of this page
In an ocean full of clicking shrimp and singing whales
fish are often imagined as the silent actors
but they lack the iconic mewl of a cat or trill of a bird
many fish chat constantly to mark their territories or find mates
And all of our noise—from seismic surveys to boat motors—is making it much more difficult for fish to hear one another
The fact that some fish make noise has been known for a while
But few people have been interested in studying the phenomenon
Mouy and his colleagues will gather underwater recordings to determine which fish are making sounds
“Of the 400 species in British Columbia’s waters
“But there are probably a lot more we don’t know about.”
Some use their bones: catfish make sounds by moving the vertebrae in their spine
while seahorses knock the bones of their skull together
quickly contract specialized sonic muscles around their swim bladder to make low-frequency grunting noises
In primitive fish such as herring, the swim bladder is connected to the digestive system, so a sort of buzzing noise is produced when air is forced out the anus. Lawrence Dill, at Simon Fraser University in Burnaby, British Columbia, and his colleagues won the Ig Nobel prize in 2004 for discovering these “fast repetitive ticks,” or FRTs
Dill knew exactly what he was doing when he named them.)
Fish communicating in the Fraser River delta in British Columbia
Dill suggests they use the farts to communicate
and keep their school together in the dark
But other fish may use their noises differently
Mouy is deploying a series of underwater listening stations off Canada’s west coast
The first was installed in the Strait of Georgia in early May
and 14 more will be installed over the next year
The listening stations include high-definition cameras
The combination of sensors will allow Mouy to link the sounds to a particular species and behavior
Mouy says the recordings could eventually help fish researchers keep track of what’s going on under the sea
“We want to see if we can know where the fish are and what they are doing just by listening to the ocean,” he says
Dill isn’t convinced that the noises will be useful for monitoring—they depend too much on a fish’s particular situation
But recording fish communication will help us understand how fish are affected by noise pollution
“Knowing how fish communicate will help us know how we’re disrupting it,” he says
Recordings of fish communication in the Strait of Georgia
Having a deep understanding and appreciation of fish communication could be good for the fish
researchers in Sweden used their awareness of fish farts to defuse a diplomatic row
the Swedish navy had assumed that herring farts were the sound of Russian submarines lurking in Swedish waters
which led Prime Minister Carl Bildt to write a nasty letter to President Boris Yeltsin
But the researchers were able to identify the navy’s suspicious recordings as flatulent herring
Part of the and family
The family of a woman who walked away from a nursing home and whose body was found a decade later deserves $500,000 in compensation
For 12 years, Mouy Tieng Tang lived at Unique Living
an adult care home in Lawndale.
The refugee from Cambodia had schizoaffective disorder and insulin-dependent diabetes
anti-depressant and anti-anxiety medications.
walking a trail to a mini-mart just up the road.
More: Family of Mouy Tang blames state for her death
That was the last time Tang was seen alive
rabbit hunters found her remains just 100 yards away from where she had lived.
Tang’s Florida-based family later filed a lawsuit against the North Carolina Department of Health and Human Services
saying the state’s negligence led to the woman’s death.
In October 2019, the family was awarded damages of $500,000.
‘Appalling conditions’ At the crux of the case
state inspections showed that Unique Living failed to make sure all doors had alarms and cited other issues with the facility which closed its doors permanently.
the Commission said the state did not direct Unique Living “to do anything beyond the promised 30-minute checks on the residents” to address the door alarm issue
which “did not constitute adequate monitoring of the residents under the circumstances.”
Court documents also list the commission as saying that the door alarm issue paired with sanitation and hygienic violations noted in a July 2008 survey by the state “illustrated the overall appalling condition of Unique Living at that time.”
Tang was noted as acting increasingly disoriented and was the subject of 30-minute checks by Unique Living staff
she left Unique Living and was last spotted at 7:36 a.m
Unique Living staff notified local law enforcement of her disappearance at 8:05 a.m.
In response to Tang’s disappearance, the state returned to inspect the facility one week later
The report noted the continuing violations with the exit doors
as well as deficiencies with bedding and other furnishing issues
The facility was issued a violation and mandated that staff members monitor exit doors at all times.
Unique Living’s license was suspended on Sept
11, 2008, and the facility permanently closed.
Tang was officially declared deceased.
The state appealed the ruling in October 2020
Department of Health and Human Services/Division of Health Services Regulation) had the ability and the regulatory authority to take action against Unique Living to prevent harm to its residents but failed to do so.”
The Commission went on to say that the state failed to present evidence of an intervening or superseding cause of Tang’s death
and that “although it is unclear what happened to Ms
[it] can be inferred that while lost and wandering without her medications or supervision
she sustained significant pain and suffering before she died.”
Diane Turbyfill can be reached at 704-669-3334 and dturbyfill@shelbystar.com
Backed by funding from the Climate Risk and Early Warning Systems (CREWS) initiative
WMO is supporting Cambodia and Lao PDR to strengthen effective and risk-informed early warning services
especially against floods and droughts.
held in a Least Developed Country for the first time
agreed to provide a potential additional US$ 7 million in funding for Cambodia and Lao PDR
bringing the total CREWS portfolio in the two countries to US$ 12 million
Six of the principal CREWS funding partners recently visited Cambodia and Lao PDR to gain first-hand experience of WMO pilot projects in community-based flood warnings
The delegation included climate ambassador Stephane Crouzat of France
The CREWS delegation met heads of national meteorological and hydrological services and ministers to understand how CREWS-funded projects are contributing to Early Warnings for All initiative
Cambodia is one of the 30 countries chosen to accelerate the implementation of the UN Secretary-General’s Early Warnings for All initiative.
nearly 80 percent of the population live in rural areas
and 65 percent rely on agriculture for their livelihoods
which means they eat less than the minimum daily requirement of calories
Floods and drought are regular occurrences and their frequency and intensity have increased due to the changing climate
predict and communicate on the occurrence of these extreme events and their impacts is a priority for the authorities
Under the framework of CREWS Cambodia and Lao PDR project
WMO and its partners have supported the roll-out of Early Warnings For All with a national workshop and the ongoing development of a national early warning roadmap.
Progress to date include an assessment of the capacities of the Departments of Meteorology and Hydrology and River Works
and an investment plan to cover key capacity gaps
such as the installation of a database management system and training of staff already allow a better access to global and regional forecasts from the European Center for Medium-Range Weather Forecasts (ECMWF) and the European Center in Vietnam
WMO is assisting the country by piloting flood early warning systems in three basins
visualization of the risk and flood events and communication and preparedness by the local communities
disaster impacts have increased and there is need for immediate action to prevent or manage risks
Being climate researchers and weather forecasts
I would like to continue participating and supporting various activities of the CREWS project which is strengthening the preparedness and resilience capacities of Cambodia on early warning and early action,” said Seth Vannareth
Ministry of Water Resources and Meteorology and Permanent Representative of Cambodia to WMO
Mr Viengxai Manivong Deputy Director General of the Department of Meteorology and Hydrology (DMH) thanked CREWS for its support on advancing the development of Early Warning Systems.
The risk profile with probabilistic risk assessment provides potential impacts of flood and drought events in current and projected climate scenarios
The workshops presented the results of risk profiles and gather recommendations for applying the results of risk profiles in sustainable development planning and disaster risk reduction measures.
Given the yearly flood and drought disasters
these risk profiles provide in-depth understanding of potential risks and impacts in both countries
The risk profiles are crucial in supporting national agencies and ministries in identifying measures for preventing and mitigating the disaster risks.
‘’Working with national institutions to provide forecasts and risk profiles that allow the population to understand the potential impacts of floods and droughts and measures that need to be taken are key outputs for Early Warnings for All initiative in Cambodia and Lao PDR which WMO is leading with UNDRR following the UN Secretary General mandate to cover everyone on this planet with early warning systems in the next five years,’’ commented Jason Watkins
Welcome to IPE. This site uses cookies. Read our policy
IPE magazine July/August 2023
By IPE Staff2023-01-19T09:03:00+00:00
M&G Asset Management – Joseph Pinto has been appointed as M&G’s next chief executive officer
He will have accountability for all investment capabilities including the equity
private and alternative asset strategies alongside distribution
operations and proposition management across the asset management business
With 30 years of experience in asset management
Pinto joins from Natixis Investment Managers where he has served as a head of distribution and investment solutions for EMEA
Previously at AXA Investment Managers for 13 years
global head of markets and investment strategy and head of business development for South Europe and the Middle East
Pinto joins in March 2023 and will become a member of M&G’s executive committee
announced his intention to retire following 21 years with the business
The appointment is subject to regulatory approval
AP4 – The Swedish national pensions buffer fund has announced three new hires – Mikael Anveden as head of portfolio analysis
Lina Berwick in the role of senior legal counsel
and Per Hammarlund as chief investment strategist
All three started work at the Stockholm-based fund at the beginning of this year
Anveden previously worked at the bank SEB for more than 15 years
most recently as head of institutional advisory
having spent the last nine years as the bank’s chief emerging markets strategist
Berwick was senior legal counsellor at Finanstilsynet
the Norwegian Financial Supervisory Authority
Pollination – John Morton has rejoined global climate change investment and advisory firm Pollination as a member of the firm’s executive committee and as managing director
He returns to Pollination after serving for two years as the US Treasury Department and secretary Janet Yellen’s inaugural climate counsellor
he managed and coordinated the Department of Treasury’s historically ambitious climate change strategy and policies
Morton’s unique breadth of experience spans the private sector
Over the course of his three-decade career
Morton’s work has focused on expediting the transition to a low carbon economy
and he has been at the forefront of designing and structuring some of the most innovative platforms and investment vehicles to mobilize private capital flows into fast-growing global markets
Morton served as the president’s White House senior director for energy and climate change
and previously as the chief of staff and COO at the Overseas Private Investment Corporation (now the US Development Finance Corporation
Morton was formerly managing director of economic policy at The Pew Charitable Trusts
a private equity investor with the Global Environment Fund
TPT Retirement Solutions – The workplace pension scheme providing services to more than 425,000 members has appointed David Lane as its CEO
He had been serving as interim CEO since June 2022
Prior to this he held the post of COO from 2019
Lane joined TPT as chief finance officer in June 2013 having held finance
commercial and operational roles at board level in private and listed businesses spanning a range of industry sectors
including chair of trustees for the Stanley Tools UK Pension Scheme
and a member-nominated trustee for the Jacuzzi UK Retirement Benefits Scheme
Lane will undertake a strategic review of TPT’s business to shape its next phase of growth
LCP – Ruth Ward has joined LCP as a principal from Mercer
She brings over 16 years of de-risking experience including buy-ins and buyouts
She joins LCP as part of a wider expansion of LCP’s longevity de-risking practice which has grown by over 10 people in the past year through internal and external joiners taking the headcount to 60 experienced individuals
which LCP believes is the largest team in the market dedicated to buy-ins/outs
This is in response to the rapid growth in de-risking activity anticipated in 2023 due to improved buy-out funding levels
Pictet Alternative Advisors (PAA) – The firm has strengthened its direct private equity capabilities with the hiring of five experienced investment professionals
global co-head of private equity and head of direct private equity
who was recruited in April 2022 from THI Investments to build the direct franchise
Buckley brings over 20 years of experience investing in the European mid-market with a strong track record and depth of experience across a broad range of sectors and geographies
Nikolaus Hubmann has joined PAA as head of DACH and is responsible for research
due diligence and oversight of investment opportunities within the region
he was a principal at Sun Capital Partners
Hubmann is a specialist in the DACH middle-market
with 12 years’ experience originating and executing both buy-side and sell-side transactions in Germany and the wider DACH region
with a focus on the Industrials and B2B services sectors
Sean Howard has been hired as an investment manager and will support Buckley and the regional heads in the origination and oversight of investment opportunities and portfolio companies
He was previously in Ares Management’s private equity group
where he focused on UK buyout investments across B2B services
Jan Dreesen has joined as an investment associate
Dreesen was an investment professional at Aurelius Group
deal execution and strategic support to portfolio companies in the European middle-market
Hugo Hickson has been hired as a value creation principal
He brings over 10 years’ experience in helping European businesses improve and grow
primarily built up over his tenure at McKinsey & Company
The experienced team will continue to grow through 2023
and will draw on Pictet’s global infrastructure and investment resources including ESG
the team will expand its activities once the UK entity has obtained its regulatory authorisation
Aon – William Engelhardt has been appointed as an associate partner in Aon’s corporate consulting and member options teams
where he provided strategic counsel on UK pension arrangements to a portfolio of trustee and corporate clients
gaining wide experience of funding and investment strategy
He has advised a number of pension scheme trustee boards and companies and is a fellow of the Institute of Actuaries
Engelhardt has joined Aon’s corporate consulting team based in St Albans
where he will divide his time between advising companies on pension issues and leading projects for the member options team
Mercer – The consultancy has appointed Tom Austin as a director in its growing UK covenant advisory team headed by partner Darren Masters
The new hire is in addition to Kriti Chandnani
Brandy Mukasa and Michael Fenton who all joined in the last year
taking the total size of the team to 15 covering the whole of the UK
a qualified accountant with almost two decades of professional services experience
more latterly working at Interpath Advisory
He brings to Mercer a depth of experience of working with trustees and corporates across a range of situations including valuations
Austin’s clients have ranged in size from small owner managed businesses where scheme assets are less than £10m
to multi-national listed groups with multi-employer schemes that have assets of more than £10bn
Arkwright – Mats Langensjö has joined management consultancy Arkwright as a partner
The firm said the Swedish pensions expert would will work with fund managers
and pension companies to achieve their strategic goals and business objectives
Langensjö has already been working with Arkwright for several years
even before he formed the firm Secoria in 2019 as an Arkwright partner company
Having been inquiry chair and adviser to the Swedish Ministry of Finance during his career so far
Arkwright said Langensjö had been key part of its team working with the Swedish Pensions Agency (Pensionsmyndigheten) and on the establishment of the Swedish Fund Selection Agency (Fondtorgsnämnden) – the new procurement agency for funds to be offered on the revamped Premium Pension System platform
Langensjö said in a statement from the consultancy: “I’m really excited about joining Arkwright at such important and formative times.”Martin Tarmet
said Langensjö would be invaluable as the firm continued to grow and develop its business
XPS Pensions Group – Dave Barratt has been hired as a senior consultant in the firm’s growing risk settlement team as client demand continues to rise
Barratt joins from Aon and has more than 15 years of specialist experience
having advised on buy-in and buyout projects for schemes of all sizes
Barratt will continue to focus on leading on buy-in and buyout transactions as well as helping clients with strategic end-game journey planning
Allspring Global Investments – The asset manager has announced the opening of its new office in Hong Kong and the appointment of Linda Luk as head of sales for North Asia
Luk joins Allspring with almost 30 years of experience working in the investment industry
and marketing activities across the region
She will be responsible for executing Allspring’s sales strategy and servicing clients in Hong Kong
she served as a managing director and chief distribution officer at Income Partners Asset Management
most recently serving as a managing director and head of Hong Kong distribution
Franklin Templeton – Marie Keil-Mouy has been appointed as head of institutional sales Germany on 1 January 2023
In her new position she will lead the business development efforts related to Franklin Templeton’s institutional business in Germany
Keil-Mouy will lead a senior team of Frankfurt-based institutional sales professionals and work in close coordination with Franklin Templeton’s regional distribution teams and alternative sales specialists to engage clients on the firm’s full range of asset management capabilities
Keil-Mouy worked as deputy head institutional sales Germany and Austria at Robeco
where she led the institutional client and consultant coverage
she worked as an executive director at DekaBank where she focused on asset management strategies for pension funds and insurances
Natixis Investment Managers – Fabrice Chemouny has been appointed head of international distribution
overseeing client and development activities for EMEA
He was previously head of Asia Pacific at Natixis IM and has more than 20 years of experience in asset management
chief administration officer for asset and wealth management
will oversee post-sales support activities for international distribution
as well as Natixis IM Solutions activities
in addition to his existing responsibilities for global operations and technology
human resources and corporate social responsibility strategy
Fabrice Chemouny and Christophe Lanne will both report to Tim Ryan
head of asset and wealth management within Groupe BPCE’s Global Financial Services and will continue to serve on the management committee of asset and wealth management
Federated Hermes – EOS at Federated Hermes has announced the hire of Ross Teverson to lead its engagement team for Asia and global emerging markets
Teverson will be responsible for leading the development and delivery of EOS’s Engagement Plan for corporate and public policy
and market best practices across Asia and global emerging markets
EOS engaged with 270 companies across these regions
Teverson has deep experience in asset management and stewardship
with more than 20 years’ investing in and engaging with companies across the Asia and emerging market regions
He was previously head of strategy for global emerging market equities at Jupiter Asset Management where he held the additional roles of member of the stewardship committee and leadership group
he held various investment-focused positions at Standard Life Investments
Federated Hermes now has over 45 professionals working across its stewardship and responsibility teams
making it one of the largest stewardship resources in the world
BNP Paribas Asset Management (BNPP AM) – The firm continues to strengthen its ETF distribution capabilities with the appointment of John Altas in the newly-created role of ETF specialist for the Nordic region
he joined on 16 January and reports functionally to Claus Hecher
regional ETF head of sales for Northern Europe
head of distribution for BNPP AM in the Nordic region
Altas joins from BlackRock and has 10 years’ asset management industry experience
as a relationship manager and sales support manager
At BNPP AM he will focus on developing the company’s ETF client base in the Nordic region and providing clients with specialist expertise in sustainable ETFs and index investment solutions
Copyright © 1997–2025 IPE International Publishers Limited
Site powered by Webvision Cloud
Get the best experience and stay connected to your community with our Spectrum News app. Learn More
-- The Cleveland County Sheriff's Office says the remains of a woman who disappeared 10 years ago have been found
The remains of Mouy Tang were discovered in a heavily wooded area off Philadelphia Road near the now-closed Unique Living Rest Home
Sheriff Allen Norman believes her death was an accident
He also says her family was instrumental in the closure of the rest home
which had been the target of numerous complaints
Get the latest news, sports and weather delivered straight to your inbox. Click here to sign up for email and text alerts.
A University of Victoria PhD student is appealing to beachcombers and boaters to keep their eyes out for some scientific equipment lost at sea
Acoustics student Xavier Mouy is working on a project to catalog fish sounds along the B.C
monitoring fishes over long time periods and geographic areas
to create a non-intrusive tool for marine conservation and fisheries management
RELATED: Feds pump $3.5M into UVic climate research centre
A key piece of that monitoring – a PVC frame with hydrophones and camera – was deployed in Roberts Bay
22 and when Mouy returned a week later to retrieve it
attached to a line with two orange Scotsman surface buoys
was deployed on the ocean floor by Graham’s rock off Roberts Bay in Sidney
“We circled around the location with an echo sounder and submersible but couldn’t find it,” said Mouy
“We came back with divers too and they couldn’t find it.”
RELATED: Breakthrough in brain health thanks to UVic researchers
The lost scientific instrument is crucial for his PhD research and will set his project back
“I’d have to find some money to buy another hydrophone
The camera is designed by myself; it is not the most expensive but it is the most time consuming,” Mouy said
If found please call 250-885-2347 or email xaviermouy@uvic.ca
there have been several unsolved cases in the county
but the Cleveland County Sheriff’s Office doesn’t see them as cold cases
“All of these cases are continually under investigation,” Sheriff Alan Norman said
The sheriff’s office keeps the cases active because they are working to help bring closure to the families that have been affected
maybe we can bring these people some closure,” Norman said
There are currently two missing person cases and eight unsolved homicides in Cleveland County
Asha Degree was 9 years old when she went missing on Feb
her family has continued to hold out hope for her safe return
the Federal Bureau of Investigation put up $25,000 and the sheriff’s office and people from Cleveland County have put together another $20,000 toward a reward in the hopes of enticing someone to come forward with information on the case
“Someone in the community knows something,” Public Affairs Specialist with the FBI Shelley Lynch said on Feb
disappeared after leaving an assisted living facility in Lawndale
at the Interstate 85 rest stop in Cleveland County
Porter was working at the rest stop as part of his employment with the N.C
He was found bound with a gunshot wound to the chest in the rest stop utility room
witnesses reported seeing two young black men drive away toward I-85 South in a dark Volkswagen
Sketches of the two men were circulated after the murder
in a wooded area on Selkirk Road in Lawndale
He had been reported missing by his family before his body was found
Ricky Mauney’s body was found in the road in the 2900 block of Philadelphia Church Road in Lawndale on Dec
Kimberly Francis Pagano was found dead Jan
at her residence at 306 Compact School Road in Kings Mountain
She was last seen alive earlier that evening by her boyfriend
at his residence at 1226 Mary’s Grove Church Road in Kings Mountain
his fiancé found his body on their front lawn
and the sheriff’s office follows up on every lead
according to a statement from the sheriff’s office
The sheriff’s office has received and continues to receive evidence in each case
No information has been or will be released in any of the cases as to maintain the integrity of the cases and not jeopardize any future investigations
Anyone with information pertaining to any of these cases is asked to call sheriff’s office at 704-484-4888 or Crime Stoppers at 704-481-8477
division manager of 9Base Urban Development Solution
Cambodian music-lovers can now set up accounts with Apple’s iTunes Store
the technology company announced yesterday
which opened direct services to 12 countries in the region
could be limited in Cambodia because of a lack of local content and a high volume of pirated products
The authorised Apple service centre iOne said Mac users had lobbied for iTunes access for years
More popular than iTunes would be the App Store
to which Cambodians had long sought access
“Apple agreed with our suggestion and decided to launch the app store,” Serey Vuth said
“We have been trying to convince them for a long time.”
Apple won’t be the only company marketing digital music in Cambodia
With vast selections of music and albums priced as low as US$0.75
pirated-music vendors would outprice Apple
Ouk Serey Vuth admitted that piracy was a problem
but said Cambodians would choose the legal option that backed up music files if lost
Apple would have to boost access to local content if it wants to increase its market share in Cambodia
“I don’t think it’s going to make any difference,” Hello Mobile chief executive Simon Perkins said yesterday
“I think it’s only going to be relevant when it offers products for the Cambodian market.”
Many Apple users in Cambodia probably had iTunes accounts abroad
The announcement yesterday did not indicate if Cambodians would be able to upload applications designed in their own language
said yesterday using Khmer apps could be challenging
but it’s not easy to put them on the Apple Store
Lots of Apple users in Cambodia jailbreak their phones to put on Khmer applications,” he said
To contact the reporter on this story: Don Weinland at [email protected]
Sign in
Join now, it's FREE!
God called Dad home peacefully in hospital at the age of 90
Loving husband of Geertruida Bos. Father to John (Nellie)
Wilma (Peter) and Larry (Jacky). Opa to Joanna (deceased)
Jessica. Dad was a proud “Opa” to 9 great grand children – Maya
James. Also remembered by nieces and nephews in Canada and in Europe
Last September we had the wonderful opportunity to celebrate Dad’s 90th birthday
Dad and Mom celebrated their 65th wedding anniversary
We give thanks to God for the many blessings and His faithfulness during Dad’s life on earth
his love of family and for his steadfast example of his faith in God
We would also like to thank the staff of Mill Creek Care Centre Floor 3 west for the love and care for Dad the past few months. This was greatly appreciated
In lieu of flowers donations can be made to Heart and Stroke Foundation of Ontario.A private family burial will take place prior to the visitation and a celebration of Dad’s life to be held at the First Christian Reformed Church
2019 with visitation at 1:30 pm followed by a Celebration of LIfe at 2:00 pm and a time of refreshments and fellowship after the service
Isaiah 40:31 – But those who hope in the Lord will renew their strength. They will soar on wings like eagles; they will run and not grow weary
PHNOM PENH: Thousands of Cambodians working for a Hong Kong-listed casino company have gone on strike for higher pay and better working conditions
About 3,000 workers joined the strike in front of the NagaWorld hotel and casino complex
They demanded a pay increase to US$300 a month for hotel staff and $500 a month for those on the gambling floor
The striking workers said wages now start at $150 to $250 a month
They are also seeking the reinstatement of their union president
who has been working at the casino for 19 years
said the company had warned employees in a memo they would be fired if they went on strike
we need to pay rent and food prices nowadays have also gone up," By Kunthea said
Card dealer Chan Maiya Mouy complained of poor working conditions including abusive behaviour
"We can't go to the toilet if there is no one to fill in for us," Chan Maiya Mouy said
they curse us and throw water bottles at us."
which has the exclusive licence to operate in Phnom Penh
reported an estimated $1.8 billion in revenue last year
It posted a 9-month net gaming revenue of $616.3 million last year
deputy director of Labour Ministry's labour dispute department
NagaWorld did not respond to requests for comment
By subscribing, you accept the terms and conditions in our privacy policy.