Revealing year-round activity of cave-dwelling insectivorous bats with a sonotype classifier in data-deficient areas.

This study demonstrates the first long-term acoustic monitoring of cave-dwelling insectivorous bats in the Republic of Congo by utilizing a novel sonotype classifier to reveal year-round activity patterns and reproductive phenology in a previously data-deficient region.

The Gist

The Big Picture: Listening to the "Bat City"

Imagine you are standing outside a massive, ancient apartment building (a cave) in the Republic of Congo. Inside, thousands of tiny tenants (bats) live their lives. But here's the problem: you can't see inside, and you don't speak their language. You have no idea who lives there, when they leave for work, when they come home, or if they are raising a family.

This study is like installing a high-tech, 24-hour security camera and voice recorder at the front door of two of these "bat apartments" (Mont Belo and Boundou caves). The goal? To figure out the daily and yearly schedule of these mysterious neighbors without ever having to kick down the door and disturb them.

The Challenge: The "Dictionary" Problem

Usually, to identify a bat by its sound, scientists need a dictionary of known bat calls. But in Central Africa, this dictionary was missing. It was like trying to identify a song on the radio when you don't know the names of any bands or singers.

The Solution: The researchers used a clever "sonotype classifier." Think of this as a smart music shuffler. Instead of needing to know the exact name of every song, it groups sounds by their "shape" and "pitch."

• Group A: Sounds like a high-pitched squeak.
• Group B: Sounds like a low, rumbling buzz.

To make this useful, the team did a little bit of "field work." They gently caught a few bats at the cave entrance, recorded their voices, and took notes on who they were. They used these few samples to "teach" the smart shuffler which group of sounds belonged to which type of bat. Suddenly, the mystery was solved: they could now tell the difference between the "Miniopterus" group and the "Hipposideros" group just by listening.

What They Discovered: The Bat Daily Routine

Once they could listen and identify the groups, they mapped out the bats' lives over 19 months. Here is what the "security footage" revealed:

1. The Commute (Daily Activity)
Most bats follow a strict commute schedule, just like humans.

• Evening Rush Hour: Around sunset (6:00 PM), there is a massive burst of activity. The bats fly out to hunt insects.
• Morning Rush Hour: Around sunrise (5:00 AM), they all come back home to sleep.
• The "Double Shift": For most of the year, the bats do two shifts: go out, come back, go out again, come back.

2. The "Nursery" Effect (Seasonal Changes)
The routine changes drastically during the rainy season (specifically around October to December).

• The Analogy: Imagine a school bus that usually runs on a tight schedule. But during "Parenting Season," the bus stops running on a schedule and starts making constant, erratic trips.
• What happened: During the rainy season, when baby bats are being born and fed, the activity becomes evenly spread out all night long. Instead of one big rush, the mothers are constantly flying out to catch bugs, coming back to feed their babies, and flying out again. They are too busy parenting to stick to a strict schedule!

3. The "Vacation" (Dry Season)
During the long dry season (June to August), the caves are much quieter.

• The Analogy: It's like a ghost town. The food (insects) is scarce, so the bats either leave the area or stay very quiet inside the cave to save energy. The "commute" slows down significantly.

Why This Matters: Protecting the "Apartments"

The researchers found that these caves aren't just random holes in the ground; they are critical real estate.

• Mont Belo Cave seems to be the "Premium Condo." It hosts more species and is a major hub for raising babies (maternity colonies). It needs high-level protection.
• Boundou Cave is more like a "Stable Apartment Complex." It has fewer species but is used consistently year-round.

The Takeaway:
If you want to protect these bats, you can't just protect the cave; you have to protect the timing.

• If you disturb the cave during the dry season, it might be annoying, but the bats are less active.
• If you disturb the cave during the rainy season (when moms are feeding babies), you could crash the whole family's ability to survive.

The Bottom Line

This paper is a blueprint for how to study animals in places where we know very little. By combining smart sound software with gentle field checks, the researchers proved you don't need to be an expert with a dictionary to understand a forest's secrets. You just need the right tools to listen.

They showed us that even in data-poor regions, we can learn exactly when bats are eating, sleeping, and raising families, giving us the power to protect their homes at the exact moments they need it most.

Technical Summary

1. Problem Statement

• Data Deficiency: Despite bats being the second most diverse mammalian order and crucial for ecosystem services (insect control, pollination), ecological data on their populations and phenology is scarce, particularly in Central Africa.
• Conservation Gap: Cave-dwelling bats are highly vulnerable to habitat degradation, mining, and tourism. Effective conservation requires understanding annual phenology (when and how long bats occupy caves), but this data is often missing for African cave systems.
• Methodological Limitations: Traditional monitoring (visual counts, capture) is invasive, time-consuming, and prone to observer bias. Passive Acoustic Monitoring (PAM) is a non-invasive alternative, but its application in the Republic of Congo is hindered by a lack of local reference call libraries, making species identification in massive datasets impossible using standard classifiers.

2. Methodology

The study employed a hybrid approach combining passive acoustic monitoring with targeted field captures to overcome the lack of local acoustic libraries.

• Study Sites: Two caves in the Republic of Congo (Mont Belo and Boundou) located in a karstic savannah/forest region.
• Duration: 19 months (September 2021 – March 2023).
• Data Collection:
  • Passive Acoustic Monitoring (PAM): SM4BAT recorders with U1 microphones were placed at cave entrances. Recordings were taken 30 minutes before sunset to 30 minutes after sunrise.
  • Capture Sessions: 11 sessions using harp traps at cave entrances to capture bats for species identification, biometric data, and reproductive status assessment (pregnancy, lactation, testes size).
• Data Processing & Classification:
  • Sonotype Classifier: The study utilized a pre-existing sonotype classifier (Roemer et al., 2021; modified by Labadie et al., 2025b) to group calls based on shape and frequency without prior species labels.
  • Local Calibration: Captured bats provided reference calls to map "sonotypes" to specific species or species groups (e.g., Hipposideros caffer, Miniopterus group, Rhinolophus groups).
  • Activity Metric: Bat activity was quantified as "bat passes" (detection of one or more calls within a 5-second interval).
• Statistical Analysis:
  • Generalized Linear Mixed Models (GLMM) using the glmmTMB package in R to model bat passes as a function of the month, with random effects for month/year.
  • Density functions were used to analyze the distribution of activity throughout the night.
  • Bonferroni corrections were applied to control for Type I errors.

3. Key Contributions

• First Long-Term Acoustic Study in Congo: This is the first study to provide a 19-month continuous acoustic record of cave bat communities in the Republic of Congo.
• Methodological Framework for Data-Deficient Regions: The paper demonstrates a replicable workflow for studying bat ecology in regions lacking reference libraries: using a generic sonotype classifier augmented by local reference calls collected via targeted capture.
• Linking Acoustics to Reproductive Phenology: The study successfully correlates acoustic activity patterns with reproductive cycles (gestation, parturition, lactation) derived from capture data, validating PAM as a proxy for biological status in the absence of direct observation.

4. Key Results

• Species Composition: Seven species were identified across the two caves, grouped into five acoustic clusters:
  • Hipposideros caffer
  • Miniopterus group (M. cf. minor, M. cf. inflatus)
  • Rhinolophus denti
  • Rhinolophus landeri-alcyone
  • Rhinolophus – Macro – Tria group (found only in Boundou).
• Nocturnal Activity Patterns:
  • Bimodal Pattern: Most months showed two peaks: emergence shortly after sunset (18:00) and return before sunrise (05:00–06:00).
  • Tri/Quadri-modal Pattern: During the long rainy season (December), activity became evenly distributed throughout the night. This shift is attributed to lactating females making repeated foraging trips to nurse young, a behavior consistent with energy constraints during reproduction.
• Annual Phenology:
  • High Activity Periods: Activity peaked during the rainy seasons and the short dry season. This correlates with lactation, juvenile independence, and mating periods.
  • Low Activity Periods: The long dry season (June–August) showed reduced activity, likely due to resource scarcity.
  • Reproductive Synchrony:
    • Hipposideros caffer and Miniopterus groups showed evidence of potentially two reproductive cycles or cryptic species complexes.
    • Mating generally occurred in the long dry season (June–July), with births and lactation peaking in the long rainy season (October–December) to coincide with insect abundance.
• Site Differences: Mont Belo appeared to function primarily as a maternity/breeding site with distinct reproductive peaks, while Boundou showed more stable, year-round occupation by multiple species.

5. Significance and Implications

• Conservation Prioritization: The study provides the data necessary to calculate vulnerability indices for these caves. Mont Belo is identified as having higher biotic value due to its role as a maternity site, suggesting it requires stricter protection against disturbance during reproductive peaks.
• Scalability: The methodology proves that quantitative bat monitoring is feasible in data-deficient regions without pre-existing acoustic libraries. This offers a pathway for global bat conservation in the tropics.
• Ecological Insight: The research confirms that cave-dwelling insectivorous bats in Central Africa synchronize their reproductive cycles with seasonal resource availability (rainfall/insect abundance), highlighting the vulnerability of these populations to climate change or habitat disruption that alters these seasonal cues.
• Future Directions: The authors suggest future integration of non-invasive hormonal monitoring (via urine collection) to further refine reproductive phenology without the need for frequent captures.