Funding Priorities in Passive Acoustic Applications in Fisheries

by Rodney Rountree and Cliff Goudey

 

Research presented at MIT Sea Grant’s April 2002 workshop - Workshop on the Applications of Passive Acoustics to Fisheries - underscores the great strides that have been made in the application of passive acoustics to fisheries and related issues over the last two decades.  It is clear from this body of work that while passive acoustics is currently underutilized as a research tool, it is rapidly emerging and holds great promise for the future.

Studies presented described the rapid growth of research on passive acoustics applications to fisheries and marine census.  Passive acoustics approaches can provide fish biologists and fishery scientists with a non-destructive sampling tool offering a unique perspective on the biology and ecology of soniferous species of fishes.  However, the workshop participants identified a number of areas in which scientific and technical developments are needed to enhance our understanding of fish and promote future research.  Subsequent to this workshop, the organizers (Rountree, Goudey, Hawkins), together with other leaders in the field (Mann, Luczkovich) have been active in promoting the field.  In August 2003, Luczkovich, Mann and Rountree organized a special symposium at the American Fisheries Society Annual Meeting in Quebec on passive acoustics applications to fisheries that received rave international reviews.  And most significantly, important new initiatives in passive acoustics have begun in many areas of the United States (Massachusetts, New Jersey, North Carolina, and California, and in all 13 National Marine Sanctuaries).  As a result of these more recent activities, we can now provide a more comprehensive description of areas that need to be addressed in order to advance the field of passive acoustics applications to fisheries.  These areas can be categorized as follows: research needs, software needs, hardware needs, and education/outreach needs.

 

I.                   Research Needs - Research needs can be simplistically summarized as the questions: What, When, Where and How Many?

A)    What? - What fishes and invertebrates are making sounds?  To date we know of approximately 700 species of fishes worldwide that produce sounds.  More species are being added everyday, but there are no systematic effort to catalogue vocal fishes.  The landmark studies by Marie Fish and William Mowbray came to an end over 30 years ago.  No one has taken up the gauntlet since then.  Knowledge of what produces underwater sounds is critical to the success of higher level studies.  Currently the number of unidentified underwater sounds attributed to fishes is far greater than those that can be positively identified.  Goudey and Rountree have recently recorded unknown sounds on the commercial fishing grounds off the coast of Massachusetts.  Even more remarkably, Rountree and Juanes and their students have recorded many unknown sounds in areas that have been extensively studies by conventional means (e.g. Woods Hole, MA), or that are in the very heart of the industrial world (the docks on Manhattan Island, New York, NY).  How can it be that we know so little about fishes in these areas that we can't even identify some of the most frequent and widespread fish sounds?  As a biologist, I find the lure of the unknown so close to the doorstep to be a powerful motivator.   Although some progress is being made in this area, below we list the most pressing needs.

1)      An effort to catalogue historic records of known and unknown sounds.  Efforts are now underway to digitize and catalogue sound archives from laboratories across the country and in Europe.  Cornell University will host a public data set containing the complete sound recordings of Fish, Mowbray, Winn and many other scientists.  Unfortunately, many of these historic recordings are poorly documents and were made with long outdated technologies.  More importantly, they are inadequate for the needs of fisheries researchers, as they are only partially complete.  Many of our most important commercial fishes that are vocal have yet to be recorded.

2)      Systematic efforts to identify and validate sound sources.  To expand the catalogue of fish and invertebrate sounds, field and laboratory studies are needed to identify unknown sounds, and audition fishes and invertebrates for sound production.  Systematic efforts to identify sounds in each geographic region, including estuaries and tidal freshwaters are critical to the advancement of passive acoustics.  Proof that a particular sound is made by, and unique to, a given species is of critical importance to passive acoustics.  Lack of, or perceived lack of, such proof is the most common criticism encountered in funding proposals.

3)      Studies to determine the correlation between sound production and specific behaviors.  Under what behavioral conditions do fish/invertebrates vocalize or make sounds?  It is important to realize that incidental sounds made by fishes can be just as important as vocalizations.  For example, feeding sounds can be used to determine foraging times, locations, and consumption rates.

B)     When? -  Temporal patterns in sound occurrence are needed for two main reasons.  First, information on when fishes and invertebrates make sounds reveals important information about their biology and ecology.  Second, we need this information to effectively use passive acoustics as a tool to locate fish, to identify their habitat requirements, and to conduct presence/absence and abundance surveys (see below).  Some of the most important needs here are:

1)      Studies to determine the relationship of sound production to fish size.  Early studies have shown that sound characteristics sometimes change with fish size (e.g., dominant frequency) and hence the relationship between fish size and sound characteristics can be used to determine length frequency data for vocal fishes.

2)      Studies to determine the relationship of sound production to sex and maturity stage.  Do both male and females in a population produce sounds?  Are the sounds the same or different?  Do immature fish make sounds?  Are the sounds made by immature and mature individuals different?  The answers to these questions can provide scientists with valuable information on the temporal and spatial distribution patterns of fishes by sex and maturity stages.  They can also provide data useful in studies of reproductive ecology and studies specifically important for fisheries assessment, such as the quantification of fish fecundity.

3)      Quantification of daily and seasonal patterns in sound production.  If we know why a fish vocalizes then studies of the daily and seasonal patterns of that vocal activity can be used as an index of the daily and seasonal patterns of that behavior.  For example, if it is known that a particular sound is only produced when a fish is spawning, then the determination of the daily pattern in vocal activity can be used to infer the daily pattern of spawning (i.e., what time of day does the fish spawn).  Similarly, seasonal patterns in vocal activity can provide an index of the spawning season period.

 

C)    Where?  Identifying where a fish occurs is one of the most fundamentally important topics in fish ecology and fisheries management.  Its the first step towards identifying essential fish habitat (EFH) as mandated of fisheries managers by the reauthorization of the Magnuson-Stevens Fishery Conservation and Management Act (Oct. 1996).  Essential Fish Habitats are defined as "those waters and substrate necessary for fish for spawning, feeding or growth to maturity."  The lowest level criterion for identification of EFH is simply presence absence.  At the minimum, passive acoustics surveys can be used to identify the presence of fishes in habitats.  If the behavior associated with a particular fish sound is known, then higher levels of identification, such as the identification of spawning habitat, can be achieved.  This does not require the ability to determine abundance.  However, if passive acoustics can be used to additionally provide an index of abundance, then the highest levels of EFH can be determined (see below).  Research needed:

1)      Quantification of sound source detection ranges.  In order to accurately associate underwater sounds with specific habitats, it is necessary to know the range at which the sound can be detected by the survey instruments.  This is the second most commonly sited criticism encountered in funding proposals.   In order to determine sound source detection ranges several types of studies are needed:

(a)   How loud is it?  - Quantification of sound source levels for a given species/stage/environment is the first step.

(b)  Sound Propagation - how far does sound travel under a given set of environmental conditions.  This is the study of tomography. Studies in shallow water are especially needed.

(c)   Fish Shoals - studies are needed to examine how the multiple vocalization of many fishes in a shoal combine.  How do sound pressure levels vary with shoal size and distance from the source?

2)      Passive acoustic mapping of the spatial distribution of sounds.  This is one of the most important potential products of passive acoustics technology.  Studies to map the distribution of fish sounds, and hence the distribution of vocal fishes and their essential fish habitat are strongly needed.  Passive acoustics offers many advantages over traditional methods of mapping fish distribution.  It is non-destructive, non-invasive, and relatively inexpensive over the long run compared to tradition methods such as trawl surveys.  If the "what" and "when" are known, then passive acoustics can be used to map EFH based on presence/absence.  The "when" is important because spatial surveys with passive acoustics are only valid if conducted during the window of time when fish are vocal.  In many cases, fish restrict their vocal activity predominately to narrow windows of time.  For example, the striped cusk-eel calls predominantly during the hour after sunset, so attempts to use passive acoustics to map their distribution would have to be conducted within that time frame.  However, if the "how many" is known, then more advanced EFH determinations based on abundance are possible (see below).

 

D)    How many?  Quantification of fish abundance at a location and time are necessary to establish the highest criterion for EFH.  This is also fundamental to efforts to census marine life (sensu the international program Census of Marine Life).  Once the questions described above have been answered, there are still additional studies needed to allow abundance to be quantified by passive acoustics techniques.  The two most important study areas are described below.

1)      Correlation between the "amount" of vocalization and the abundance of fish.  In the simplest situation, the number of fish calls can be counted and correlated to the true amount of fish present.  This requires the ability to distinguish individual calls.  More rigorously it would also require the ability to distinguish individuals that call repeatedly from those that don't.  Often, however, fish and calls are so numerous that individual calls can not be distinguished.  In these cases, studies are needed to determine the relationship between sound pressure level and fish abundance.

2)      Determination of the proportion of fish that are vocal at a given place and time.  As mentioned above, vocalization in many fish is sex dependent.  Often, only the males call.  In addition, in some cases vocalization may be related to size and maturation stage, so that at a given place and time, the number of fish calling is a function of the size and maturation stage distribution of the population.  However, there can be many reasons why individual fishes fail to vocalize in a given situation.  Studies are needed to allow researchers to estimate the proportion of vocalizing and non-vocalizing fishes over a given time period.

 

II.                 Software needs - In order to carry out the research priorities listed above, software must be developed in several areas.  Program should be packaged to allow biologists to process sound data to the extent possible.  Extensive experience with acoustics and acoustics software programming should not be required.

A)    Automatic processing of sound recordings.  The passive acoustics studies described above generate large amounts of acoustic data that are currently prohibitively time consuming to process to obtain the basic biological data being sought (e.g., temporal activity patterns, spatial distribution, etc.).  Software designed to reduce much of the drudgery of processing acoustic data is critically needed.  Some specific components of automatic processing include:

1)      Automatic signal detection.  Software that can be used to detect the call of a given fish species amid long timeseries of sound recordings with multiple sounds sources and high levels of "noise" is essential to the further development of passive acoustics in fisheries.

2)      Temporal pattern analysis.  Software that can track the temporal occurrence patterns of detected calls within long timeseries. This ability is critical to the efficient determination of the daily and seasonal patterns in sound production.

3)      Call Characterization.  Software to identify individual calls and automate the summary of call characteristics such as duration, pulse rate, pulse repetition, fundamental frequency, pulse width, sound level, etc.

4)      Call classification.  Software that can automate the classification of calls into species (e.g. haddock), behavior (e.g., spawning haddock or feeding haddock), and species groups (e.g. unknown gadid) or classify with the most similar known (e.g. most similar to gadid calls) and unknown calls (e.g., most similar to call "Unidentified A" recorded by Smith, 2004).  Obviously this software would require an extensive collection of known and unknown fish sounds, and rigorous characterizations for each (this has never been done systematically across all known species, though Katz and Lobel have been working on some groups of fishes).

 

B)     Localization of sound sources.  Software, and hardware, developments are needed to simplify the localization of sound sources received by multiple hydrophone arrays.  Automatic signal detection and classification software are important prerequisites.  Currently localization is done in a very laborious manner by manually determining the starting points of the same sound received among multiple hydrophones and then using various statistical techniques such as time-delay differences to triangulate on the calls.  In order to develop the ability to produce real-time GIS plots of vocal fishes overlain on environmental parameters, these steps must be automated.  One important product of localization, other than habitat association mapping, is the determination of true sound source levels under varying environmental conditions.

 

C)    Soniferous Behavior.  Software packages that allow researchers to simultaneously process acoustic and visual data are needed to enhance the study of the correlation between sounds and specific behaviors.  For example, a program that displays the waveform and spectrogram of recorded sounds in one window and video in a second window would be very useful.  Ideally the user could click on a specific point on the waveform/spectrogram, and see the corresponding video frame and visa versa.  Such software would allow researchers to more rapidly and efficiently correlate individual sounds, and sound components with fish behavior and functional morphology.

 

III.               Hardware needs - acoustic hardware technology is rapidly evolving, however, there are several specific needs that would enhance the development of passive acoustics in fisheries.

 

A)    Data storage -  Low cost, high capacity digital recorders are needed to allow field sampling at the rates and intensities necessary to study temporal and spatial patterns of sound production.  Devices that can be programmed to record at varying sampling rates and time settings, including continuous and time lapse would be particularly valuable.  Existing low cost recorders developed by the music industry are rapidly becoming scarce due to the industry shift from wave file storage formats to MP3 file format storage (MP3 is unacceptable because it compresses and distorts the sounds).

B)     Coupled "visualization" technologies.  In order to validate the identification of sound sources in the field, devices that couple passive acoustics with other technologies that can be used to identify a fish are needed.

1)      Coupled acoustic-optic systems - Devices that are capable of recording both underwater video and acoustic data are among the simplest to use and are highly reliable for identification.  Amazingly, most underwater video technology lack acoustic recording capabilities and underwater video makers are reluctant to incorporate acoustic recording technology.  Attempts to mount hydrophones on existing underwater devices such as ROVs are frustrated by the noise generated by these devices, and by the lack of efficient acoustic data transfer.  In most cases, acoustic data is received and transmitted independently.  Acoustic data is transmitted via temporary cables that are subject to frequent failure because the devices are not designed for acoustic data transmission (e.g., one has to run a separate acoustic cable along the ROV tether).  Another difficulty of acoustic optic systems is the need use artificial light sources for the video recording, and resulting dramatic change in fish behavior and probability of detection.  In addition, artificial light provides a limited range of visibility.  Finally studies are needed to understand the effect of light of various wavelengths and intensities on fish behavior in order to develop the optimum optic system.

2)      Coupled passive and active acoustics systems - active acoustics can potentially be used as an alternative method for "visualization" of sound sources.  Active acoustic technology has the advantage of providing relatively long-range observation capabilities compared with video.

3)      Coupled passive acoustics and laser line-scan technology - another method of visualization that should be developed.

 

C)    Portable passive acoustic survey devices - there is a great need for small self contained packages of acoustic gear that allow mobile recording of acoustic and video during shore or small boat based studies.  Currently investigators have to purchase separate components and assemble them into a package themselves.  There are many problems with this approach because components were not manufactured to work together and often numerous in-line adapters must be used to transfer the acoustic signal among components.  Researchers need to be able to drop a hydrophone over the side of a small boat or dock, together with a small underwater video camera, and be able to simultaneously record and monitor video and acoustic data.  Ideally, the date would be automatically converted to digital form for storage.  Availability of low cost devices of this type would go a long way towards making small scale research projects feasible for a larger scientific community.

D)    Portable hydrophone arrays - Hydrophone arrays with separation distances on the order of meters to tens of meters, that can be deployed from small and large ships are needed to allow detailed study of fish vocal behavior and sound source levels under field conditions.  The array should be able to self determine the relative inter-hydrophone spacing and its location relative to a surface buoy with GPS input.   

E)     Additional hardware needs. 

1)      Improved ship-based listening systems are needed, including dangling and towed hydrophones.

2)      Bottom-mounted listening systems are needed for determining the temporal patterns of fish sounds.  Robust, low-cost systems are especially important.

3)      Drifting sonobuoy systems are needed, either storing the data, or telemeter data to ships or shore-based listening stations.

4)      Systems are needed to measure source levels and calibrate listening devices so that researchers can determine the distance to sound sources.

5)      Unmanned, archival acoustic recorders are needed for use on ships of opportunity.

 

IV.              Education and Outreach - most fish bioacousticians are biologists first and engineers second.  They have arrived at fish bioacoustics because it is a powerful tool for studying fish — one unmatched by other approaches.  This means that engineering and signal processing principles must be learned on the job.  Unfortunately, there is no one good source of information about recording and signal processing that is accessible and practical for the fish bioacoustician.  This gap can be bridged both by producing these targeted materials and conducting training workshops, and by attracting engineers with a biological interest to the field.  The workshop participants identified a strong need to educate scientists, managers and the public on the uses of passive acoustics.  Some specific needs include:

A)    Passive acoustic literature. Most acoustic textbooks and educational literature are targeted to an audience of physical oceanographers and acousticians. Manuals and other literature intended to introduce biologists to the field of passive acoustics are badly needed.  David Mann is currently working on one manual, but his work could be speeded up and enhanced with some dedicated funding. 

B)     Training workshops.  Workshops that bring biologists, acousticians and engineers together are vital to the future development of the field as they stimulate the transfer of knowledge and ideas among the disciplines.  The passive acoustics workshop sponsored by MIT Sea Grant and ONR has already generated a significant amount of growth in the field.

C)    Training center.  A central location where students and researchers can attend introductory and advanced training in passive acoustics technologies and applications to fisheries was identified as an important need by participants in the passive acoustics workshop.

D)    Outreach.  It was also recognized that passive acoustics technologies provide a unique public outreach potential.   Scientists and laymen alike are often fascinated by the phenomenon of underwater sounds.  Passive acoustics technologies are amenable to multimedia display via the Internet and have great potential as public education and outreach tools.