Acoustic Monitoring for Nocturnal and Elusive Wildlife: Setup, Interpretation, and Applications
Passive recorders can reveal wildlife activity that people rarely see. The harder part is knowing what a recording actually proves.
Image: Land & Wildlife Report
Much of wildlife monitoring happens when people are not around to see it. Bats leave their roosts after sunset. Frogs may call after rain. Owls can announce their presence from inside a forest without ever crossing an observer’s line of sight.
Passive acoustic monitoring, usually shortened to PAM, turns those sounds into data. Autonomous recorders can remain in the field for days, weeks, or months, collecting vocal activity across times and weather conditions that would be difficult to cover with people alone.
That makes acoustic monitoring especially useful for nocturnal, elusive, or visually difficult species. It does not make the method foolproof. Poor placement, bad timing, background noise, weak identification models, or a species that simply did not call can all change the result.
The most useful way to think about acoustic monitoring is not as a machine that tells managers what animals are present. It is a sampling method that records sound. Everything after that depends on survey design and interpretation.
Passive acoustic monitoring can detect bats, birds, frogs, insects, and other vocal wildlife without requiring a person to be present when the animal calls. It is especially useful for long-duration monitoring and species active at night.
The method has a built-in limit: a recorder detects sound, not animals. A call can support a species detection, but silence does not automatically mean the species was absent. Weather, season, animal behavior, microphone placement, equipment, and identification software all affect what reaches the final dataset.
Southwestern Indiana already provides a real example. Indiana DNR uses acoustic bat surveys as part of long-term statewide monitoring, and those data now feed into the North American Bat Monitoring Program.
In This Report
- What acoustic monitoring actually measures
- Why it works for elusive wildlife
- Field setup matters
- How Indiana is using acoustic monitoring
- Automated identification and AI
- Why silence does not mean absence
- What monitoring at scale can show
- Acoustic monitoring vs. camera traps
- When acoustics are not enough
- Common Questions
What Acoustic Monitoring Actually Measures
Passive acoustic monitoring uses autonomous recording units to capture sound without requiring a researcher to remain at the site. Depending on the equipment and target species, the system may record ordinary audible sound, ultrasonic bat calls, or both.
Human surveys provide short observations. Acoustic recorders can sample across many nights and repeat the same protocol at multiple locations. That is useful when activity is brief, weather-dependent, seasonal, or clustered during hours when field crews would otherwise need to remain outside.
Any conclusion about species presence, abundance, movement, or behavior is an inference built from recorded vocal activity.
That distinction sounds obvious, but it drives nearly every limitation of the method. A species can be present without calling. Two species can produce similar calls. One microphone may hear an animal that another recorder misses. A noisy site can reduce the usable detection area even when the habitat itself has not changed.
Acoustic monitoring works best when the biological question matches what the recorder can actually observe.
Ultrasonic detectors record echolocation calls and can support distribution, activity, trend, and regulatory surveys.
Vocal species can be sampled repeatedly without requiring an observer to be present for every calling period.
Breeding calls provide useful seasonal detections, although temperature, rainfall, and timing can strongly affect calling.
Why It Works for Nocturnal and Elusive Wildlife
Some animals are simply easier to hear than see. A bat can pass overhead in darkness without being visible. An owl may call from dense cover. Frogs can be obvious acoustically while nearly impossible to locate individually in a wetland.
Recorders also solve a scheduling problem. Wildlife activity does not conveniently happen during a two-hour field visit. Autonomous units can sample the same place across multiple nights, weather conditions, and activity periods.
Long-duration sampling does not remove bias. It changes where the bias appears. Instead of depending mostly on whether an observer happened to be present, the result depends more heavily on microphone sensitivity, recorder settings, placement, recording schedule, environmental noise, and how detections are classified afterward.
Field Setup Can Decide Whether the Data Are Useful
Acoustic monitoring often looks deceptively simple: attach a recorder to a tree, turn it on, and come back later. The equipment may be easy to deploy. Designing a defensible survey is not.
Presence? Activity timing? Long-term trend? Regulatory survey?
Microphone, frequency range, recording mode, batteries and storage must fit the target.
Placement, timing and settings need to be repeatable across sites and survey periods.
Classifier output still needs quality control appropriate to the decision being made.
Placement changes what the recorder hears
Vegetation, height, nearby roads, running water, wind exposure, microphone orientation, and the amount of open space around a recorder can all affect detections. A microphone placed beside dense foliage does not sample the same acoustic space as one facing an open corridor.
There is no universal mounting height or detection radius that works across taxa and habitats. Bat echolocation, frog calls, bird song, equipment sensitivity, weather, and habitat structure operate at different scales. That is why standardized survey protocols matter more than a generic distance estimate.
Timing matters just as much as placement
Recording schedules should match the ecology of the species. Continuous recording can maximize coverage, but it also creates large datasets. Targeted schedules may be more efficient when calling periods are well understood.
Season and weather matter too. Amphibian calling can change quickly with temperature and rainfall. Bat activity changes across the summer. Bird vocal activity varies with breeding stage and time of day.
A perfectly functioning recorder deployed during the wrong biological window can collect excellent audio that answers the wrong question.
What Acoustic Monitoring Looks Like in Indiana
Acoustic monitoring is not an experimental idea sitting somewhere outside the region. Indiana has used it for years to track summer bat populations.
Indiana’s Mobile Acoustic Bat Surveys
Indiana Department of Natural Resources uses ultrasonic detectors and vehicle-mounted microphones to record bat echolocation calls along predetermined routes.
The routes are generally 15–25 miles long and are driven shortly after sunset. Each area is surveyed twice during a roughly six-week period beginning in June and ending in mid-July.
Repeating the same routes allows biologists to compare bat activity through time instead of treating each night’s detections as an isolated observation.
Indiana DNR reports that mobile acoustic surveys were completed in 76 counties from 2011 through 2019. In 2020, the state updated its methods to follow protocols from the North American Bat Monitoring Program, or NABat. Those standardized methods combine mobile and stationary acoustic sampling and allow Indiana data to contribute to broader analyses across North America. [Indiana DNR]
The regional value is fairly direct. White-nose syndrome caused major declines in several cave-hibernating species, while migratory bats face a different set of pressures. Long-running monitoring lets managers look for changes across years instead of relying on isolated observations.
It also shows why acoustic data are stronger when paired with other methods. Indiana uses winter hibernacula counts, summer roost observations, acoustic surveys, and other field information to understand bat populations. One method fills gaps left by another.
Automated Identification Is Getting Better. It Still Needs Context.
A long acoustic survey can produce far more recordings than a person could reasonably inspect one by one. Automated classifiers solve part of that problem by scanning files and assigning likely species identifications or confidence scores.
That makes large monitoring programs possible. It also creates a temptation to treat the software output as the biological result.
A 2026 study tested AI-assisted bird and frog classifiers against more than 17,000 manually validated detections from real-world recordings. Performance was not uniformly mediocre or uniformly excellent. Species tended to fall toward the extremes: some were identified with very high precision, while others performed poorly. Call similarity, background noise, and vocal complexity all affected the results. [Bell et al., 2026]
That is an important correction to a common assumption about AI wildlife monitoring. A classifier can work extremely well for one target species and badly for another inside the same dataset.
The recorder matters too
Software is only analyzing what the microphone captured. A 2025 comparison of commercial bat detectors and lower-cost open-source AudioMoth units found that full-spectrum commercial detectors generally recorded more species and accumulated species detections more quickly. Lower-cost devices could still provide useful data, but matching the stronger equipment sometimes required more units or longer sampling periods. [Perks et al., 2025]
That does not mean inexpensive recorders are poor tools. It means equipment is part of the sampling design. A monitoring program cannot assume that two devices provide interchangeable data simply because both produce audio files.
Software can review enormous datasets quickly. Managers still have to decide whether the equipment, training data, confidence threshold, and verification process are suitable for the decision at hand.
Why Silence Does Not Mean Absence
False positives get attention because they create detections that should not exist. False negatives can be just as important.
A species may be present and remain undetected because it did not vocalize, because the call was too faint, because wind or rain masked it, because the recorder was poorly placed, or because the classifier failed to recognize the signal.
The safest interpretation of an empty recording period is therefore narrow: no usable target vocalization was detected during the sampled period under those conditions.
That is different from proving that the species was not there.
Federal bat surveys show how seriously this is treated
The U.S. Fish and Wildlife Service’s 2026 range-wide survey guidance for Indiana bats and northern long-eared bats includes acoustic surveys as one method for establishing presence or probable absence. The guidance specifies survey effort, timing, equipment, weather, analysis, and reporting requirements rather than treating a few quiet nights as proof that bats are absent. [USFWS, 2026]
The agency also tells surveyors to check the current list of approved automated acoustic-identification software before submitting a 2026 study plan. That is a good example of the larger point: software is part of a survey protocol, not a substitute for one.
What Happens When Acoustic Monitoring Is Repeated at Scale
A single recorder can answer a local question. Hundreds of standardized surveys repeated over years can do something much larger.
The North American Bat Monitoring Program brings together stationary acoustic surveys, mobile acoustic routes, capture surveys, roost counts, and emergence counts. USGS reported that by 2025 the program contained data contributed by more than 480 partner organizations and 2,400 users.
The scale matters because long-term wildlife monitoring is less about collecting isolated detections than finding patterns that persist across space and time. [USGS NABat]
Researchers are also combining data types instead of treating acoustics as a standalone measurement. A 2025 NABat analysis integrated stationary acoustics, mobile acoustic transects, and live-capture data while accounting for imperfect detection and species misclassification.
At the range-wide scale, the model estimated a 24% decline in relative abundance of hoary bats between 2012 and 2023, with substantial uncertainty around the estimate. The analysis did not claim that every state or every local population declined by the same amount. Trends varied geographically. [USGS, 2025]
That is the difference between simply counting recorded calls and using repeated monitoring to estimate a population trend.
Acoustic Monitoring vs. Camera Traps
Acoustic recorders and camera traps are sometimes compared as though one should replace the other. They observe different things.
Acoustic Monitoring
- Actually observes
- Sound produced within the effective acoustic sampling area.
- Best fit
- Vocal species, nocturnal activity, repeated temporal sampling, soundscape questions.
- Main strength
- Can sample long periods without requiring an animal to pass directly in front of a sensor.
- Main weakness
- A silent animal may remain completely invisible to the method.
- Abundance
- Call activity should not automatically be treated as a count of animals.
Camera Traps
- Actually observes
- Animals entering the camera’s physical detection zone.
- Best fit
- Visually identifiable mammals and other animals moving through predictable locations.
- Main strength
- Provides direct visual evidence and may show behavior, condition, group size, or individual markings.
- Main weakness
- An animal outside the detection zone can remain completely invisible to the camera.
- Abundance
- Repeated photographs are not automatically the same as independent animals.
Detection distance, cost, weather sensitivity, and identification accuracy vary too much among species, equipment, habitats, and protocols to reduce the comparison to one universal number.
The better question is simple: which method directly samples the biological signal needed to answer the management question?
Acoustic monitoring is one piece of a much larger technology toolkit. For a broader comparison of drones, thermal imaging, environmental DNA, and other monitoring approaches, see New Tools in Wildlife Management: What Drones, Thermal Imaging, eDNA, and Deterrent Research Actually Deliver.
When Acoustic Monitoring Is Not Enough
Acoustic monitoring is a poor primary method when the target rarely vocalizes, when individuals must be physically identified, when population size requires a stronger abundance estimate, or when the management decision requires information that sound cannot provide.
A recorder may confirm that a bat species is active near a forest edge. It cannot tell managers which tree is being used as a maternity roost. That may require mist-netting, radio telemetry, emergence counts, or direct roost surveys.
A recorder may detect frog calls in a wetland. It does not automatically estimate how many frogs produced those calls.
The same rule applies across monitoring technologies. The tool should follow the question, not the other way around.
What to Remember About Passive Acoustic Monitoring
- PAM detects sound, not animals. Biological conclusions are inferred from recorded vocal activity.
- Long-duration recording is especially useful for nocturnal, elusive, seasonal, or intermittently vocal species.
- Placement, timing, weather, microphone performance, recorder settings, and survey design can all change detection probability.
- Automated classifiers make large datasets manageable, but accuracy can vary sharply among species.
- Silence is not proof of absence. Strong non-detection conclusions require repeated sampling under appropriate conditions.
- Indiana already uses acoustic bat monitoring as part of a broader long-term program, and those data contribute to continent-wide NABat analyses.
Common Questions
Can acoustic monitoring tell exactly how many animals are present?
Usually not by simply counting calls. One animal may vocalize repeatedly, several animals may overlap, and calling rates change with behavior and environmental conditions. Population or density estimates require statistical methods designed around detection probability and the biology of the target species.
Does an acoustic recording prove a species is present?
A correctly identified diagnostic call can provide strong evidence of presence. Confidence depends on recording quality, how distinctive the call is, the classifier or reviewer used, and the consequence of being wrong. High-stakes regulatory detections may require stricter validation.
If nothing is recorded, does that mean the species is absent?
No. It means the survey did not detect a usable target vocalization during the sampled period. The animal may have been silent, outside the effective detection area, masked by noise, or missed because of weather, equipment, timing, or identification error.
Are inexpensive acoustic recorders good enough for wildlife surveys?
They can be. Research shows that lower-cost open-source equipment can produce useful monitoring data, but performance is not always identical to commercial systems. Survey duration, number of devices, sampling rate, microphone quality, habitat, and target species all affect whether a lower-cost system is appropriate.
Why are bats such a common use for acoustic monitoring?
Bats are active at night and many species produce ultrasonic echolocation calls that can be recorded electronically. That lets biologists sample activity without having to see or capture every bat. Acoustic data are commonly combined with mist-netting, roost surveys, emergence counts, winter surveys, and other methods.
References
- Bell, K., Edwards, B., Phan, T. L., & Malerba, M. E. (2026). Promise and pitfalls: Variable performance of AI-assisted passive acoustic monitoring of birds and frogs using BirdNET and VicFrogNET. Ecological Informatics, 94, 103644. https://doi.org/10.1016/j.ecoinf.2026.103644
- Indiana Department of Natural Resources. (2026). Acoustic monitoring of bats. Indiana Division of Fish & Wildlife
- Knight, E. C., Hannah, K. C., Foley, G. J., Scott, C. D., Brigham, R. M., & Bayne, E. M. (2017). Recommendations for acoustic recognizer performance assessment. Wildlife Society Bulletin, 41(2), 231–242. https://doi.org/10.1002/wsb.771
- Marques, T. A., Thomas, L., Ward, J., DiMarzio, N., & Tyack, P. L. (2013). Estimating animal population density using passive acoustics. Biological Reviews, 88(2), 287–309. https://doi.org/10.1111/brv.12001
- Perks, S. J., O’Connell, M. J., & Goodenough, A. E. (2025). Comparing passive acoustic monitoring bat data from commercial and open-source detectors: Evidence to support best practice. Ecological Solutions and Evidence, 6(3), e70103. https://doi.org/10.1002/2688-8319.70103
- Royle, J. A., & Link, W. A. (2006). Generalized site occupancy models allowing for false positive and false negative errors. Ecology, 87(4), 835–841. DOI
- Shonfield, J., & Bayne, E. M. (2017). Autonomous recording units in avian ecological research: Current use and future applications. Avian Conservation and Ecology, 12(1), 14. https://doi.org/10.5751/ACE-00974-120114
- U.S. Fish and Wildlife Service. (2026). Range-wide Indiana bat and northern long-eared bat survey guidelines. U.S. Fish and Wildlife Service
- U.S. Geological Survey. (2025). North American Bat Monitoring Program (NABat). U.S. Geological Survey
- Udell, B. J., Straw, B. R., Davis, H. T., Lohre, B. T., Reichard, J. D., Coleman, J. T. H., & Reichert, B. E. (2025). Results and code from the North American Bat Monitoring Program’s integrated species distribution and trend models for hoary bat and silver-haired bat. U.S. Geological Survey data release. https://doi.org/10.5066/P134REO7