Ecological inferences about marine mammals from passive acoustic data.

Monitoring on the basis of sound recordings, or passive acoustic monitoring, can complement or serve as an alternative to real-time visual or aural monitoring of marine mammals and other animals by human observers. Passive acoustic data can support the estimation of common, individual-level ecological metrics, such as presence, detection-weighted occupancy, abundance and density, population viability and structure, and behaviour. Passive acoustic data also can support estimation of some community-level metrics, such as species richness and composition. The feasibility of estimation and certainty of estimates is highly context dependent, and understanding the factors that affect the reliability of measurements is useful for those considering whether to use passive acoustic data. Here, we review basic concepts and methods of passive acoustic sampling in marine systems that often are applicable to marine mammal research and conservation. Our ultimate aim is to facilitate collaboration among ecologists, bioacousticians, and data analysts. Ecological applications of passive acoustics require one to make decisions about sampling design, which in turn requires consideration of sound propagation, sampling of signals, and data storage. One also must make decisions about signal detection and classification and evaluation of the performance of algorithms for these tasks. Investment in the research and development of systems that automate detection and classification, including machine learning, are increasing. Passive acoustic monitoring is more reliable for detection of species presence than for estimation of other species-level metrics. Use of passive acoustic monitoring to distinguish among individual animals remains difficult. However, information about detection probability, vocalisation or cue rate, and relations between vocalisations and the number and behaviour of animals increases the feasibility of estimating abundance or density. Most sensor deployments are fixed in space or are sporadic, making temporal turnover in species composition more tractable to estimate than spatial turnover. Collaborations between acousticians and ecologists are most likely to be successful and rewarding when all partners critically examine and share a fundamental understanding of the target variables, sampling process, and analytical methods.

Performance metrics for marine mammal signal detection and classification.

Automatic algorithms for the detection and classification of sound are essential to the analysis of acoustic datasets with long duration. Metrics are needed to assess the performance characteristics of these algorithms. Four metrics for performance evaluation are discussed here: receiver-operating-characteristic (ROC) curves, detection-error-trade-off (DET) curves, precision-recall (PR) curves, and cost curves. These metrics were applied to the generalized power law detector for blue whale D calls [Helble, Ierley, D'Spain, Roch, and Hildebrand (2012). J. Acoust. Soc. Am. 131(4), 2682-2699] and the click-clustering neural-net algorithm for Cuvier's beaked whale echolocation click detection [Frasier, Roch, Soldevilla, Wiggins, Garrison, and Hildebrand (2017). PLoS Comp. Biol. 13(12), e1005823] using data prepared for the 2015 Detection, Classification, Localization and Density Estimation Workshop. Detection class imbalance, particularly the situation of rare occurrence, is common for long-term passive acoustic monitoring datasets and is a factor in the performance of ROC and DET curves with regard to the impact of false positive detections. PR curves overcome this shortcoming when calculated for individual detections and do not rely on the reporting of true negatives. Cost curves provide additional insight on the effective operating range for the detector based on the a priori probability of occurrence. Use of more than a single metric is helpful in understanding the performance of a detection algorithm.

Improve automatic detection of animal call sequences with temporal context.

Many animals rely on long-form communication, in the form of songs, for vital functions such as mate attraction and territorial defence. We explored the prospect of improving automatic recognition performance by using the temporal context inherent in song. The ability to accurately detect sequences of calls has implications for conservation and biological studies. We show that the performance of a convolutional neural network (CNN), designed to detect song notes (calls) in short-duration audio segments, can be improved by combining it with a recurrent network designed to process sequences of learned representations from the CNN on a longer time scale. The combined system of independently trained CNN and long short-term memory (LSTM) network models exploits the temporal patterns between song notes. We demonstrate the technique using recordings of fin whale (Balaenoptera physalus) songs, which comprise patterned sequences of characteristic notes. We evaluated several variants of the CNN + LSTM network. Relative to the baseline CNN model, the CNN + LSTM models reduced performance variance, offering a 9-17% increase in area under the precision-recall curve and a 9-18% increase in peak F1-scores. These results show that the inclusion of temporal information may offer a valuable pathway for improving the automatic recognition and transcription of wildlife recordings.

The Lombard effect in singing humpback whales: Source levels increase as ambient ocean noise levels increase.

Many animals increase the intensity of their vocalizations in increased noise. This response is known as the Lombard effect. While some previous studies about cetaceans report a 1 dB increase in the source level (SL) for every dB increase in the background noise level (NL), more recent data have not supported this compensation ability. The purpose of this study was to calculate the SLs of humpback whale song units recorded off Hawaii and test for a relationship between these SLs and background NLs. Opportunistic recordings during 2012-2017 were used to detect and track 524 humpback whale encounters comprised of 83 974 units on the U.S. Navy's Pacific Missile Range Facility hydrophones. Received levels were added to their estimated transmission losses to calculate SLs. Humpback whale song units had a median SL of 173 dB re 1 µPa at 1 m, and SLs increased by 0.53 dB/1 dB increase in background NLs. These changes occurred in real time on hourly and daily time scales. Increases in ambient noise could reduce male humpback whale communication space in the important breeding area off Hawaii. Since these vocalization changes may be dependent on location or behavioral state, more work is needed at other locations and with other species.

Publisher Correction: Deep neural networks for automated detection of marine mammal species.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Lombard effect: Minke whale boing call source levels vary with natural variations in ocean noise.

Minke whales were acoustically detected, localized, and tracked on the U.S. Navy's Pacific Missile Range Facility from 2012 to 2017. Animal source levels (SLs) were estimated by adding transmission loss estimates to measured received levels of 42 159 individual minke whale boings. Minke whales off Hawaii exhibited the Lombard effect in that they increased their boing call intensity in increased background noise. Minke whales also decreased the variance of the boing call SL in higher background noise levels. Although the whales partially compensated for increasing background noise, they were unable or unwilling to increase their SLs by the same amount as the background noise. As oceans become louder, this reduction in communication space could negatively impact the health of minke whale populations. The findings in this study also have important implications for acoustic animal density studies, which may use SL to estimate probability of detection.

Deep neural networks for automated detection of marine mammal species.

Deep neural networks have advanced the field of detection and classification and allowed for effective identification of signals in challenging data sets. Numerous time-critical conservation needs may benefit from these methods. We developed and empirically studied a variety of deep neural networks to detect the vocalizations of endangered North Atlantic right whales (Eubalaena glacialis). We compared the performance of these deep architectures to that of traditional detection algorithms for the primary vocalization produced by this species, the upcall. We show that deep-learning architectures are capable of producing false-positive rates that are orders of magnitude lower than alternative algorithms while substantially increasing the ability to detect calls. We demonstrate that a deep neural network trained with recordings from a single geographic region recorded over a span of days is capable of generalizing well to data from multiple years and across the species' range, and that the low false positives make the output of the algorithm amenable to quality control for verification. The deep neural networks we developed are relatively easy to implement with existing software, and may provide new insights applicable to the conservation of endangered species.

Migratory behavior of eastern North Pacific gray whales tracked using a hydrophone array.

Eastern North Pacific gray whales make one of the longest annual migrations of any mammal, traveling from their summer feeding areas in the Bering and Chukchi Seas to their wintering areas in the lagoons of Baja California, Mexico. Although a significant body of knowledge on gray whale biology and behavior exists, little is known about their vocal behavior while migrating. In this study, we used a sparse hydrophone array deployed offshore of central California to investigate how gray whales behave and use sound while migrating. We detected, localized, and tracked whales for one full migration season, a first for gray whales. We verified and localized 10,644 gray whale M3 calls and grouped them into 280 tracks. Results confirm that gray whales are acoustically active while migrating and their swimming and acoustic behavior changes on daily and seasonal time scales. The seasonal timing of the calls verifies the gray whale migration timing determined using other methods such as counts conducted by visual observers. The total number of calls and the percentage of calls that were part of a track changed significantly over both seasonal and daily time scales. An average calling rate of 5.7 calls/whale/day was observed, which is significantly greater than previously reported migration calling rates. We measured a mean speed of 1.6 m/s and quantified heading, direction, and water depth where tracks were located. Mean speed and water depth remained constant between night and day, but these quantities had greater variation at night. Gray whales produce M3 calls with a root mean square source level of 156.9 dB re 1 µPa at 1 m. Quantities describing call characteristics were variable and dependent on site-specific propagation characteristics.

Swim track kinematics and calling behavior attributed to Bryde's whales on the Navy's Pacific Missile Range Facility.

Time difference of arrival methods for acoustically localizing multiple marine mammals have been applied to recorded data from the Navy's Pacific Missile Range Facility in order to localize and track calls attributed to Bryde's whales. Data were recorded during the months of August-October 2014, and 17 individual tracks were identified. Call characteristics were compared to other Bryde's whale vocalizations from the Pacific Ocean, and locations of the recorded signals were compared to published visual sightings of Bryde's whales in the Hawaiian archipelago. Track kinematic information, such as swim speeds, bearing information, track duration, and directivity, was recorded for the species. The intercall interval was also established for most of the tracks, providing cue rate information for this species that may be useful for future acoustic density estimate calculations.

Automated acoustic localization and call association for vocalizing humpback whales on the Navy's Pacific Missile Range Facility.

Time difference of arrival (TDOA) methods for acoustically localizing multiple marine mammals have been applied to recorded data from the Navy's Pacific Missile Range Facility in order to localize and track humpback whales. Modifications to established methods were necessary in order to simultaneously track multiple animals on the range faster than real-time and in a fully automated way, while minimizing the number of incorrect localizations. The resulting algorithms were run with no human intervention at computational speeds faster than the data recording speed on over forty days of acoustic recordings from the range, spanning multiple years. Spatial localizations based on correlating sequences of units originating from within the range produce estimates having a standard deviation typically 10 m or less (due primarily to TDOA measurement errors), and a bias of 20 m or less (due primarily to sound speed mismatch). An automated method for associating units to individual whales is presented, enabling automated humpback song analyses to be performed.

Calibrating passive acoustic monitoring: correcting humpback whale call detections for site-specific and time-dependent environmental characteristics.

This paper demonstrates the importance of accounting for environmental effects on passive underwater acoustic monitoring results. The situation considered is the reduction in shipping off the California coast between 2008-2010 due to the recession and environmental legislation. The resulting variations in ocean noise change the probability of detecting marine mammal vocalizations. An acoustic model was used to calculate the time-varying probability of detecting humpback whale vocalizations under best-guess environmental conditions and varying noise. The uncorrected call counts suggest a diel pattern and an increase in calling over a two-year period; the corrected call counts show minimal evidence of these features.

Site specific probability of passive acoustic detection of humpback whale calls from single fixed hydrophones.

Passive acoustic monitoring of marine mammal calls is an increasingly important method for assessing population numbers, distribution, and behavior. A common mistake in the analysis of marine mammal acoustic data is formulating conclusions about these animals without first understanding how environmental properties such as bathymetry, sediment properties, water column sound speed, and ocean acoustic noise influence the detection and character of vocalizations in the acoustic data. The approach in this paper is to use Monte Carlo simulations with a full wave field acoustic propagation model to characterize the site specific probability of detection of six types of humpback whale calls at three passive acoustic monitoring locations off the California coast. Results show that the probability of detection can vary by factors greater than ten when comparing detections across locations, or comparing detections at the same location over time, due to environmental effects. Effects of uncertainties in the inputs to the propagation model are also quantified, and the model accuracy is assessed by comparing calling statistics amassed from 24,690 humpback units recorded in the month of October 2008. Under certain conditions, the probability of detection can be estimated with uncertainties sufficiently small to allow for accurate density estimates.

A generalized power-law detection algorithm for humpback whale vocalizations.

Conventional detection of humpback vocalizations is often based on frequency summation of band-limited spectrograms under the assumption that energy (square of the Fourier amplitude) is the appropriate metric. Power-law detectors allow for a higher power of the Fourier amplitude, appropriate when the signal occupies a limited but unknown subset of these frequencies. Shipping noise is non-stationary and colored and problematic for many marine mammal detection algorithms. Modifications to the standard power-law form are introduced to minimize the effects of this noise. These same modifications also allow for a fixed detection threshold, applicable to broadly varying ocean acoustic environments. The detection algorithm is general enough to detect all types of humpback vocalizations. Tests presented in this paper show this algorithm matches human detection performance with an acceptably small probability of false alarms (P(FA) < 6%) for even the noisiest environments. The detector outperforms energy detection techniques, providing a probability of detection P(D) = 95% for P(FA) < 5% for three acoustic deployments, compared to P(FA) > 40% for two energy-based techniques. The generalized power-law detector also can be used for basic parameter estimation and can be adapted for other types of transient sounds.