The potential for tracking wildlife health & disease via bioacoustics is great (commentary)

By Jesse Turner 

• Bioacoustics is the passive, non-invasive recording of sounds emitted by a wide range of animals.
• Analysis of this information reveals the presence and behavior of wildlife, and can also be valuable indicator of animal health, which can then be used in ecosystem monitoring.
• “As disease prevalence skyrockets in wildlife, we are desperately in need of tools to remotely monitor ecosystem health,” a new op-ed argues.
• This post is a commentary. The views expressed are those of the author, not necessarily Mongabay.

Vocalizations have the capacity to illuminate valuable information about an animal’s health and wellbeing, if we are willing to put in the effort to listen. The first time I appreciated this potential was in 2018, on the waters of the San Juan Islands in Washington State. An orca named Talequah had recently lost her calf and was keeping it afloat by nudging the corpse towards the surface, a journey that ended up lasting 17 days and 1,000 miles.

As part of a small team researching underwater noise in the area, I collected recordings of the grieving mother to study if orca calls could indicate mourning behavior. That moment transformed my understanding of how animal vocalizations, or bioacoustics, can encode information. To better understand and predict wildlife health, we should focus research on understanding how welfare metrics can be measured through vocalizations.

The use of bioacoustic signatures associated with illness to monitor human health is blossoming, leading to the term ‘acoustic epidemiology.’ Machine learning has enabled the development of automated methods using vocal recordings to detect tuberculosis, Covid-19, and Parkinson’s disease, demonstrating the ability to monitor certain health indices passively. Over the past two decades, the livestock industry has used acoustics to monitor the health of poultry, swine, and cattle due to vocalizations’ proven correlation with welfare and disease states.

Figure 1. Key epidemiological variables are listed (dark grey boxes), below which metrics are given which could be informative to the specific variable and that can be derived from acoustic data. Examples provide disease scenarios where particular epidemiological variables might be useful, listed within the table and illustrated above (colored graphic boxes). All variables could contribute to risk modeling or integrated surveillance systems (green boxes). For each epidemiological variable and associated example disease scenario, example control measures are given (colored arrows). Abbreviation: PPE/personal protective equipment. Reproduced from Johnson et al. 2023.

With wildlife populations, this correlation becomes more difficult to resolve due to limited encounters and unknown health indices of vocalizing individuals. Even so, correlations between vocalizations and wild animal health are being discovered. Studies have attributed vocalization changes in amphibians to chytrid fungal infections and altered acoustic activity in bats to white-nose syndrome, both of which are important diseases to conservation biology. Pneumonia in bighorn sheep, a growing problem in North America, induces coughing and respiratory distress that can be detected acoustically as well.

Acoustic epidemiology can be paired with more traditional uses of acoustic monitoring that aid in assessing the health of wild populations. The quantity, distribution, and change in vocalizations linked to factors that support or weaken a population can be measured. For example, isolated bird populations can develop altered dialects due to cultural drift, impacting mating and territory formation. Additionally, epidemiological variables commonly used to assess the risk of disease transmission can often be obtained using bioacoustic data (figure 1). For example, the presence of long-tailed macaques, a common reservoir for malaria, has been measured acoustically to monitor geographic risk of disease transmission.

In addition to physical health, vocalizations can contain information on an animal’s psychological condition, or affective state, which can indicate welfare. Both valence (positivity/negativity) and arousal (intensity) are key metrics (figure 2) that can shift the call type or change its duration and acoustic structure. Elevated arousal in terrestrial mammals generally causes an increase in vocalization rate, amplitude, and duration. The effect of valence is more variable between species but nonetheless has been found to correlate with changes in call characteristics. In many species, evolution has shaped acoustic cues to signal negative valence, such as distress, as well as positive valence, such as feeding or social behavior.

Pneumonia in bighorn sheep induces coughing, which can be detected acoustically. Image by Noah Israel/Wikimedia Commons.

Of course, there are obstacles and limitations to this approach. A major complexity is obtaining both welfare indices and vocalizations specific to a wild individual. Researchers should focus on contexts in which vocalizations can be isolated to individuals with known welfare states based on biomarkers, or physiological metrics associated with wellbeing, as well as behaviors and situations that correlate with arousal or valence. Prior to examining affective state, care should be taken to understand the natural history of a call to determine how specific vocalizations express arousal and valence in an evolutionary context.

When looking into health, the mechanisms by which certain diseases could alter vocalizations should also be considered, such as the propensity for respiratory diseases to impact the vocal emission pathway, as is the case for pneumonia in bighorn sheep. Calls can be attributed to an individual via observation, animal-borne acoustic recorders, or acoustic localization. If animal-borne recorders are used, biomarkers can be obtained upon capture in the form of tissue samples, or additional on-animal sensors can be incorporated to measure physiological correlates to affective state, such as heart rate and respiratory rate. Machine learning techniques can resolve changes in vocalizations between individuals of different welfare states, ultimately informing a method to passively monitor wildlife populations.

Even so, not all species vocalize, and not all welfare indices will exhibit altered vocalizations in those that do. Additionally, animal-borne sensors require capture, which is an important step in obtaining biomarkers, which also can be a traumatic event. Therefore, passive methods should be prioritized when possible, and invasive techniques such as capture should focus on the contexts described above and be used in combination with other monitoring techniques.

Figure 2. Core affect represented in two-dimensional space. Words in italics indicate possible locations of specific reported affective states (including discrete/basic emotions). Positive affective states are in quadrants Q1 and Q2, and negative states in quadrants Q3 and Q4. Arrows indicate putative biobehavioral systems associated with reward acquisition and the Q3–Q1 axis of core affect (green), and punishment avoidance and the Q2–Q4 axis of core affect (red). Adapted from Whitman and Miller 2024 and reproduced from Mendl et al. 2010
.

As disease prevalence skyrockets in wildlife, we are desperately in need of tools to remotely monitor ecosystem health. Wildlife health poses threats not only to the affected species and overall ecosystem strength, but also to human and livestock health.

Passive acoustic monitoring has a long track record of remotely observing wildlife, and is increasingly being correlated with animal welfare. I’m again reminded of the grieving mother orca, whose calls underscored the power of listening to nature. Decoding the influence of health on vocalizations via bioacoustics, in the contexts of both physical health and affective state, can enhance our efforts to understand and improve wildlife welfare.

• Jesse Turner, a PhD student at Colorado State University, is developing an animal-borne acoustic system for monitoring soundscapes from the perspective of wildlife.

---