The role of vocal individuality in conservation

1. What they did

Many animals can be told apart by their calls, because each individual’s voice varies little from call to call but differs from other individuals. The authors reviewed how this “vocal individuality” can be used in conservation, as a way of recognising individuals without catching or marking them.

They:

• Set out why identifying individuals matters for conservation: better population models, spotting individuals that matter more for breeding, catching errors in counts, and avoiding the harm that capture and marking can cause.
• Distinguished discrimination (telling individuals apart at one point in time) from identification (recognising the same individual again later).
• Gave practical advice on recording and measuring calls, and compared the main ways of analysing them, by ear and eye and with statistics.
• Described three bird case studies (European bittern, corncrake and owls), discussed the main limitations, and considered the potential in mammals.

2. What they found

Why individuals matter

• Population viability analysis (PVA) models usually treat all individuals as equal, which can make predicted extinction risk overly pessimistic. Knowing individuals improves the estimates of survival and breeding.
• A few individuals often produce most of the young: 17% of a common buzzard population produced 50% of the next year’s fledglings, and 8% of cheetah matrilineages produced over 50% of the population over 20 years.
• Counts can be biased. Playback drew responses from only 10 to 40% of Yuma clapper rail males present. Mobile floating male bitterns were counted repeatedly, so a population of 15 (nine residents and six floaters) was counted as 25.
• Capture and marking can harm animals and also only mark part of a population. Radio-tags often last a year or less, while stable calls could allow longer monitoring.

Methods

• Microphone choice depends on the call: low-pitched bittern booms (100 to 200 Hz) need a shotgun microphone rather than a parabola. Compressed recording formats (such as MiniDisc) can distort important parts of a call.
• Spectrograms are useful for looking at sounds but should not be used for direct measurements; the authors suggest waveforms for timing and power spectra for frequency.
• Experienced field researchers can often identify individuals by ear, but this has rarely been tested as a census method; one bittern study found it depended on experience and worked only for small numbers. Comparing spectrograms by eye is common and sometimes beats statistics (for example with bottlenose dolphin signature whistles), but it is subjective, so observer repeatability and inter-observer reliability need checking.
• Standard discriminant function analysis (DFA) can only assign calls to individuals already known, so it cannot detect newcomers. Similarity techniques and some neural networks can recognise new individuals. Classification scores without validation on new data are “almost meaningless”.

Case studies

• European bittern: the only species the authors knew of that is routinely counted and monitored by voice, in the United Kingdom for over 10 years. A semi-automatic method separated more than 40 bitterns, but the main feature, the boom’s dominant frequency, changed between years in four radio-tagged males, sometimes by more than the differences between birds.
• Corncrake: census rules assumed about 75% of males call each night. Voice identification showed only 41% did, so the population was underestimated by up to 30%. Males in poorer habitat moved further.
• Owls: calls are individually distinctive in many species and seem stable. European scops owl territories had 55 to 78% turnover between years; African wood owls, where both sexes call, showed 19.3% turnover in males and 13.7% in females over a 12-year study. These owl studies assumed vocal stability without independently marked birds.

Limitations

Four main problems: proving individuality and its stability needs intensive study, ideally with independently marked animals; it needs sound analysis skills and equipment; it only covers the animals that call (often territorial males); and identifications always carry some uncertainty. Females of many bird species are quiet, although contact calls (as in Steller sea lion mothers and pups) may help.

3. What they concluded

• Vocal individuality is surprisingly under-used. Most studies show that individuality exists, not that it works as a conservation tool.
• To test it as a tool: choose an easily recorded call (usually long-range advertising calls); test with as many individuals as will be monitored (studies often use only 10 to 20); choose measures that survive poor recordings; prefer similarity techniques; record locations, times and other calling birds; test with blind, simulated census trials; and give clear guidelines to the people who will collect and analyse recordings.
• Mammals have potential but are harder: their calls are more complex and can change with social context (chimpanzees, marmosets, dolphins). The swift fox is a notable exception where it has been applied.
• It is not a cure for every monitoring problem, but where a species is sensitive to disturbance or hard to monitor, its voice can be an effective conservation tool.

6. Other technologies, skills or innovations

• Shotgun and parabolic microphones, analogue, digital and compressed recorders, calibration tones, high-pass and low-pass filters.
• Waveform, power spectrum and spectrogram measurements.
• Automated identification with neural networks, which the authors had used to count and identify corncrakes in simulated census and monitoring tasks with a set of 30 individuals.

7. Implications for conservation (from the paper)

• In the authors’ view, non-invasive techniques such as vocal individuality should be the default, with invasive marking used only when its benefits outweigh possible harm.
• Recognising individuals by voice can correct census errors (floaters, birds that call less than expected) and give survival, turnover and habitat-use data for secretive, nocturnal or low-density species.
• Vocal stability over time and how to monitor females are the most important current limits.