Most animals – from insects to mammals – use vocal sounds to communicate with each other. But not all of these sounds are audible to humans. Frogs, mice and even some primates can produce noises that are ultrasonic, meaning their frequency is so high they cannot be detected by the human ear. These ‘ultrasonic vocalizations’ are used to relay a variety of signals, including distress, courtship and defense.

To understand the role ultrasonic vocalizations play in social interactions, it is important to work out which animal is responsible for emitting the sound. Current methods have a high error rate and often assign vocalizations to the wrong participant, especially if the animals are in close contact with each other. To solve this issue, Sterling et al. developed the hybrid vocalization localizer (HyVL), a system which detects ultrasonic sounds using two different types of microphones. The tool is then able to accurately locate where an ultrasonic vocalization is emitted from within a precision of millimeters.

Sterling et al. used their new system to study courtship interactions between two to three mice. The experiments revealed that female courtship vocalizations were substantially rarer than previously reported when two mice were interacting. When three mice were together (one female, two males), Sterling et al. found that one of the male mice typically dominated the conversation. This result was also reflected by the male mouse locating themselves anogenitally to the female, as males tend to vocalize more when in this position.

In neuroscience, researchers often measure ultrasonic vocalizations to monitor social interactions between rats and mice. HyVL could provide neuroscientists with a more affordable and easier to use platform for conducting these kinds of experiments, which are important for studying behavior and how the brain develops.

Ultrasonic vocalizations (USVs) fulfill an important role in animal ecology as means of communication or navigation in many rodents (Mahrt et al., 2013; Brudzynski, 2021; Zaytseva et al., 2019; Volodin et al., 2022; Murrant et al., 2013), bats (Schnitzler et al., 2003), frogs (Feng et al., 2006), cetaceans (Mourlam and Orliac, 2017), and even some primates (Bakker and Langermans, 2018; Ramsier et al., 2012). In many of these species, USVs have been shown to be present innately and to have significance at multiple stages of life, from neonates (Kikusui et al., 2011) to adults (Mahrt et al., 2013), often with diverse functions as distress/alarm calls (Kikusui et al., 2011; Litvin et al., 2007), courtship signals (Marconi et al., 2020), territorial defense signals (Rieger and Marler, 2018), private communication (Ramsier et al., 2012), and echolocation (Schnitzler et al., 2003). USVs have been extensively studied in mice, where their communicative significance has been widely demonstrated by their influence on conspecific behavior (Hammerschmidt et al., 2009; Pultorak et al., 2017; Chabout et al., 2015; Musolf et al., 2015; Sugimoto et al., 2011; Tschida et al., 2019; also in line with observational studies; Warren et al., 2020; Nicolakis et al., 2020; Rieger et al., 2021; Petric and Kalcounis-Rueppell, 2013). USVs can be grouped into different types that are highly context-dependent (Chabout et al., 2015; Musolf et al., 2015; Nicolakis et al., 2020; Chen et al., 2021; de Chaumont et al., 2021; Castellucci et al., 2018; Pultorak et al., 2018; Burke et al., 2018; Zala et al., 2017a; Mun et al., 2015; von Merten et al., 2014; Scattoni et al., 2009; Warren et al., 2021; Dou et al., 2018; Hoier et al., 2016; Chabout et al., 2012), and USV syntax itself is predictive of USV sequence (Hertz et al., 2020). Taken together, the current literature suggests USVs convey affective and social information in different behavioral contexts. This is further supported by the modulatory effect that testosterone and oxytocin have on USV production (Kikusui et al., 2021b; Kikusui et al., 2021a; Timonin et al., 2018; Pultorak et al., 2015; Guoynes and Marler, 2021; Tsuji et al., 2021; Tsuji et al., 2020). Importantly, the neuronal circuitry underlying USVs has recently been identified and is being studied extensively (Tschida et al., 2019; Chen et al., 2021; Michael et al., 2020; Gao et al., 2019; Tasaka et al., 2018; Fröhlich et al., 2017; Shepard et al., 2016; Arriaga and Jarvis, 2013; Fujita et al., 2012; Wang et al., 2008).

Because of their social and affective significance and our growing mechanistic understanding, mouse USVs are increasingly being used as a behavioral measure in neurodevelopmental and neurolinguistic translational research (de Chaumont et al., 2021; von Merten et al., 2014; Fröhlich et al., 2017; Yang et al., 2021; Binder et al., 2021; Hepbasli et al., 2021; Agarwalla et al., 2020; Tsai et al., 2012; Hodges et al., 2017). Their manipulation and precise measurement not only provide the basis for tackling many fundamental questions but also pave the way, via advanced animal models, for the discovery of essential, novel drug targets for many debilitating conditions such as autism-spectrum disorder (Tsai et al., 2012; Silverman et al., 2010), Parkinson’s disease (Ciucci et al., 2009), stroke-induced aphasia (Palmateer et al., 2016), epilepsy aphasia syndromes (Erata et al., 2021), progressive language disorders (Menuet et al., 2011), chronic pain (Palazzo et al., 2008), and depression/anxiety disorders (Moskal and Burgdorf, 2018), where ultrasonic vocalizations serve as a biomarker for animal well-being and normal development. Consequently, we expect the scientific importance of mouse USVs to continue to increase in the coming years, highlighting the necessity to advance the methods required for their study. In recent years, substantial advances have been made in USV detection (Coffey et al., 2019; Fonseca et al., 2021; Zala et al., 2017b; Van Segbroeck et al., 2017; Chabout et al., 2017), classification (Coffey et al., 2019; Fonseca et al., 2021; Van Segbroeck et al., 2017; Ivanenko et al., 2020), and localization (Oliveira-Stahl et al., 2023; Heckman et al., 2017; Warren et al., 2018a; Neunuebel et al., 2015).

Localization is of particular importance during social interactions, when most USVs are emitted and any meaningful analysis of USV properties rests on a reliable assignment of each USV to its emitter. This task is complex for multiple reasons: (i) most USVs are emitted at close range, (ii) social behavior often requires free movement of the animals, and (iii) USV production is invisible (Chabout et al., 2012; Mahrt et al., 2016). With reliable assignment, all subsequent analyses can be conducted with substantial confidence concerning each USV’s emitter. Although USVs could in theory be classified and assigned based on their shape (Marconi et al., 2020; Liu et al., 2003; Holy and Guo, 2005; Barnes et al., 2017; Musolf et al., 2010), this approach will depend strongly on different behavioral contexts and strains. Recent advances in acoustic localization (Heckman et al., 2017; Warren et al., 2018a; Neunuebel et al., 2015) have improved the localization accuracy to 11–14 mm; however, close-up snout–snout interactions – which is when a large fraction of USVs are emitted – require an even higher precision.

We have developed an advanced localization system for USVs in which is a high-resolution 'acoustic camera' consisting of 64 ultrasound microphones with an array of four high-quality ultrasound microphones. Both systems can individually localize USVs but exhibit rather complementary patterns of localization errors. We fuse them into a hybrid system that exploits their respective advantages in sensitivity, detection, and localization accuracy. We achieve a median absolute localization error of 3.4–4.8 mm, translating to an assignment rate of ~91%. Compared to the previous state of the art (Oliveira-Stahl et al., 2023; Warren et al., 2018a), the accuracy represents a threefold improvement that halves the proportion of previously unassigned USVs. Given the physical dimensions of the mouse snout (ø ~10 mm), this likely approaches the physical limit of localizability for USVs. We successfully apply it to and analyze dyadic and triadic courtship interactions between male and female mice. The comparison of dyadic and triadic interactions is chosen here as courtship interactions in nature are naturally competitive and this comparison is therefore both scientifically relevant and can benefit from high-reliability assignment of USVs. We demonstrate that the fraction of female vocalizations has likely been overestimated in previous analyses due to a lack of precision in sound localization. Further, in the triadic recordings we find that in competitive male–male–female courtship, one male takes a dominant role, which shows in emitting most USVs and also positioning itself more closely to the female abdomen.

We analyzed courtship interactions of mice in dyadic and triadic pairings. The mice interacted on an elevated platform inside an anechoic booth (see Figure 1A, for details see ‘Recording setup’). Each trial consisted of 8 min of free interaction while movements were tracked with a high-speed camera (see Figure 1B), and USVs were recorded with a hybrid acoustic system composed of four high-quality microphones (i.e., USM4) as well as a 64-channel microphone array (Cam64, often referred to as an acoustic camera; see Figure 1C for raw data samples, green and red dots mark the start and stop times of USVs).

Figure 1 with 1 supplement see all

Download asset Open asset

Most USVs were emitted in close proximity in dyadic and triadic pairings (see Figure 1D). Reliably assigning most USVs to their emitter therefore requires a highly precise acousto-optical localization system. The presently developed Hybrid Vocalization Localizer (HyVL) system is the first to achieve sub-centimeter precision, that is, ~3.4–4.8 mm (see Figure 2 for an overview). This accuracy on the acoustic side is achieved by combining the complementary strengths of the USM4 and Cam64 data. The Cam64 data is processed using acoustic beamforming (Van Veen and Buckley, 1988), which delivers highly precise estimates (median absolute errors [MAE] = ~4–5 mm), but is not sensitive enough for very-high-frequency USVs (see Figure 1—figure supplement 1). The USM4 data is analyzed using the previously published SLIM algorithm (Oliveira-Stahl et al., 2023), which delivers accurate (MAE = ~11–14 mm) and less frequency-limited estimates. The accuracy of SLIM, the previously most accurate ultrasonic localization technique (see ‘Discussion’ for a comparison), is generally lower than that of HyVL, but it makes essential contributions to the overall accuracy of HyVL through the integration of the complementary strength of the two methods/microphone arrays (see Figure 3A and L, shape of errors). The methods exhibit a complementary pattern of localization errors, which predestines them for high synergy when combined (see below).

Figure 2

Download asset Open asset

Figure 3 with 3 supplements see all

Download asset Open asset

For each USV, a choice is made between the USM4/SLIM and Cam64/Beamforming estimates based on a comparison of each method’s USV-specific certainty and the relative position of the mice to the estimates, using an extended, hybrid Mouse Probability Index (MPI; Neunuebel et al., 2015). HyVL is the first system of its kind that exploits a hybrid microphone array to overcome the limitations of each subarray. The positions of the mice are obtained via manual and automatic video tracking using DeepLabCut (Mathis et al., 2018), each of which achieve millimeter precision for localizing the snout.

Overall, 228 recordings were collected from 14 male and 4 female mice (153 dyadic, 67 triadic, and 8 with a single mouse). In 90 recordings, USVs were produced and recorded with Cam64 and USM4 simultaneously (55 dyadic, 28 triadic, and 7 single). The single mouse recordings were also used in a previous publication (Oliveira-Stahl et al., 2023) where only the SLIM accuracy was evaluated. A total of 112 recordings were recorded in a balanced design (four dyadic and four triadic per male mouse paired with all females) and the remaining recordings conducted with good vocalizers to maximize the number of USVs for downstream analysis. In all trials combined, 13714 USVs were detected.

Sex distribution of vocalizations during social interaction

Courtship interactions between mice lead to high rates of vocal production, but are challenging due to the relative proximity, including facial contact. Previous studies using a single microphone have often assumed that only the male mouse vocalized (Rotschafer et al., 2012; Choi et al., 2011; Pomerantz and Clemens, 1981; Nunez et al., 1978), while more recent research has concluded that female mice vocalize as well (Neunuebel et al., 2015; Sangiamo et al., 2020). Female vocalizations were typically less frequent, but constituted a substantial fraction of the vocalizations (11–18%) (Oliveira-Stahl et al., 2023; Heckman et al., 2017; Neunuebel et al., 2015; Warren et al., 2018b). Below, we demonstrate that the accuracy of the localization system can be an important factor for conclusions about the contribution of different sexes to the vocal interaction.

Over all dyadic and triadic trials combined, females produced the minority of vocalizations. Naive estimation without MPI selection using SLIM estimates ~14%, while HyVL tallies it at just 7% (Figure 4A). Applying MPI selection, SLIM estimates only 5.5%, while HyVL arrives at significantly less, just 4.4% (p=0.002, paired Wilcoxon signed-rank test, Figure 4A/B), while reliably classifying 91.1% of all vocalizations.

Figure 4 with 1 supplement see all

Download asset Open asset

Using HyVL instead of SLIM significantly reduces the fraction of female vocalizations, suggesting that less accurate algorithms overestimate the female fraction (only results for MPI-selected USVs shown, Figure 4B). Considering only vocalizations that are emitted when the snouts are >50 mm apart further significantly reduces the fraction to female USVs to 1.1% after MPI selection (p=5.2 × 10^–8, Wilcoxon rank-sum test). Comparing the percentage of female vocalizations between dyadic and triadic trials, no significant differences were found (p=0.22, Wilcoxon rank-sum test, Figure 4D).

Beyond the absolute distance between the mouths of the mice, high-accuracy localization of USVs allows one to position the bodies of the animals relative to one another at the times of vocalization by combining acoustic data with multiple concurrently tracked visual markers. This provides an occurrence density of other mice relative to the emitter (Figure 4E).

Female mice appear to emit vocalizations in very close snout–snout contact, with a small fraction of vocalizations occurring when the male snout is around the hind-paws/ano-genital region (Figure 4F). Male mice emit vocalizations both in snout–snout contact, but also at greater distances, which dominantly correspond to a close approach of the male’s snout to the female ano-genital region (Figure 4G). This was verified separately with a corresponding analysis, where the recipient’s tail-onset was used instead (not shown).

In summary, the combination of high-precision localization and selection using the MPI indicates that female vocalizations may be even less frequent than previously thought. When they vocalize, the mice appear to almost exclusively be in close snout–snout contact. As this is incidentally also the condition that has the highest chance of mis-assignments, even the remaining female vocalizations need to be treated with caution.

Vocalization rate analysis

In dyadic trials, one female and one male mouse interacted, whereas in triadic trials either two males and one female or two females and one male mouse interacted. We first address in dyadic trials, whether there were significant differences in individual vocalization rates between the mice. For the balanced dataset of 14 × 4 dyadic interactions (pairing of all males with all females), we did not find a significant effect of individual on vocalization rates for either male and female mice (see Figure 5—figure supplement 1, p=0.46 and p=0.16, respectively, one-way ANOVA analysis with factor individual, for n = 4 recordings in males and n = 14 recordings in females). For triadic trials, we could not perform the corresponding analysis since the two male/female recordings could not be distinguished reliably in post hoc tracking.

In the balanced dyadic and triadic datasets, only 23/112 recordings contained vocalizations. We collected additional dyadic and triadic recordings for the purpose of maximizing the number of USVs, both for assessing HyVL performance and comparing dyadic and triadic interactions. In this enlarged dataset, a total of 83 recordings (55 dyadic, 28 triadic) were available, which contained USVs. This dataset was still balanced for female mice, but, unbalanced for male mice, that is, although the same mice participated in both dyadic and triadic recordings, however, not with exactly the same number of recordings. While the analysis on the balanced dataset above did not suggest significant differences between individuals, we thus cannot fully exclude that the reported differences below are partially due to individual differences between some male mice.

In the analysis of triadic interactions, we separate competitive and alternative contexts depending on whether a mouse had to compete with another same sex mouse or could interact with two opposite sex mice, respectively. For triadic trials we further separate the same-sex mice into dominant and subordinate, based on who vocalized more.

However, in competitive interactions between males, one male mouse significantly and strongly dominated the 'conversation,' with on average ninefold more vocalizations than the other male mouse (T_D vs. T_s, Figure 5A and B, both comparisons: p<0.005 [Wilcoxon sum of ranks test]) after Bonferroni correction. Specifically, Bonferroni correction was conducted per panel/measured variable on the basis of the number of hypotheses actually tested for, that is, six tests per panel, three for each sex: dyadic vs. triadic; triadic: dominant vs. subordinate; triadic: competitive vs. alternatives. While the present division into dominant and subordinate mouse based on a higher vocalization rate within a recording will always lead to a significant difference, the quantitative difference between them is the striking aspect in this comparison. Overall male vocalization rates were similar in competitive and alternative triadic trials. Female vocalization rates were similar across all compared conditions.

Figure 5 with 1 supplement see all

Download asset Open asset

The mean vocalization energy of dominant males in triadic pairings tends to be higher than those of submissive males in triadic pairings; however, this result did not reach significance in the present dataset (see Figure 5C). No effects of vocalization energy were found in females.

The distance to the closest animal of the opposite sex was found to be even closer during triadic trials (see Figure 5D), driven purely by male vocalizers (p=0.00046, after Bonferroni correction as above, Wilcoxon sum of ranks test): the distance to the closest animal does not change between conditions for vocalizing females (p=0.975, Wilcoxon sum of ranks test). Interestingly, the distance to the closest animal was larger for females at the time of vocalization when they had a same-sex competitor on the interaction platform with them than when they were the only female (T_c vs. T_a, p=0.0068, Wilcoxon sum of ranks test).

Lastly, we investigated whether the division into a dominant and subordinate male based on the vocalization rate was also reflected in the spatial behavior of the male mice relative to the female mouse. For this purpose, we again constructed relative spatial interactions histograms (see Figure 6, analogous to Figure 4), separately for USV-rate-dominant and subordinate males. The results are displayed as the relative location between the male snout and the female abdomen. Dominant males spent more time close to the female abdomen, thus engaging in ano-genital contact (Figure 6A, center), in comparison with subordinate males (Figure 6B). This is highlighted in the difference between the spatial interaction histograms (Figure 6C), where the most salient dominant peak occurs in the center, while the subordinate male spent more time in snout–snout contact, indicated by the blue arc at about one mouse body length from the center (shown in blue here). These differences were significant, in addition to a number of other locations in the spatial interaction histogram. Significance analysis was performed using 100× bootstrapping on the relative spatial positions to estimate p=0.99 confidence bounds around the histograms of the dominant and subordinate, respectively. Significance at a level of p<0.01 highlights multiple relative spatial positions.

Figure 6

Download asset Open asset

In summary, in competitive triadic interactions, one of the male mice took a strongly dominant role, evidenced both in the vocalization rate and the more abundant ano-genital interactions with the female throughout the recordings. In triadic interactions, the female mouse was generally approached more closely by a male mouse, in particular in the alternative condition. The latter could, however, be a consequence of the larger number of male animals on the platform compared to dyadic and triadic competitive (from the perspective of the female).

We have developed and evaluated a novel, hybrid sound localization system (HyVL) for USVs emitted by mice and other rodents. USVs are innately used by rodents to communicate social and affective information and are increasingly being used in neuroscience as a behavioral measure in neurodevelopmental and neurolinguistic research. In the context of dyadic and triadic social interactions between mice, we demonstrate that HyVL achieves a groundbreaking increase in localization accuracy down to ~3.4–4.8 mm, enabling the reliable assignment of >90% of all USVs to their emitter. Further, we demonstrate that this can be combined with automatic tracking, enabling a near-complete and automated analysis of vocal interaction between rodents. The showcased analyses demonstrate the advantages obtained through more precise localization, further discussed below. HyVL is based on an array of high-quality microphones in combination with a commercially available, affordable acoustic camera. With our freely available code, this system can be readily reproduced by other researchers and has the potential to revolutionize the study of natural interactions of mice.

All experimental procedures were approved by the animal welfare body of the Radboud University under the protocol DEC-2017-0041-002 and conducted according to the Guidelines of the National Institutes of Health.

Data analysis

The analysis of the raw data involved multiple stages (see Figure 2): from the audio data, the presence and origin of USVs were estimated automatically. From the video data, mice were carefully tracked by hand at the temporal midpoint of each USV as near-optimal estimates for their acoustically localized origin. To estimate what proportion of our precision would be lost when using a faster and more scalable visual tracking method, we also tracked the mice automatically during dyadic trials. The estimated locations of the mice and USVs were then used to attribute the USVs to their emitter. All these steps are described in detail below.

Audio preprocessing

Request a detailed protocol

Prior to further analysis, acoustic recordings were filtered at different frequencies. USM4 data was band-pass filtered between 30 and 110 kHz before further analysis using an inverse impulse response filter or order 20 in MATLAB (function: designfilt, type: bandpassiir). Cam64 data was band-pass filtered with a frequency range adapted to the frequency content of each USV. Specifically, first the frequency range of the USV was estimated as the 10th–90th percentile of the set of most intense frequencies at each time point. Next, this range was broadened by 5 kHz at both ends, and then limited at the top end to 95 kHz. If this range exceeded 50 kHz, the lower end was set to 45 kHz. This ensured that beamforming was conducted over the relevant frequencies for each USV and avoided the high-frequency regions where the Cam64 microphones are dominated by noise (see Figure 1C, Figure 1—figure supplement 1).

Video preprocessing

Request a detailed protocol

The high-speed camera lens failed to produce perfect rectilinear mapping and was placed off-center with respect to the interaction platform, thereby producing a nonlinear radial-tangential visual distortion. We corrected for the radial distortion with

$x_{rd}=x_{\frac{1}{2}}+\frac{atan\left(r_d/\lambda \right)}{r_d/\lambda }\ast \left(x_{ru}-x_{\frac{1}{2}}\right)\ast Z_x$

$y_{rd}=y_{\frac{1}{2}}+\frac{atan\left(r_d/\lambda \right)}{r_d/\lambda }\ast \left(y_{ru}-y_{\frac{1}{2}}\right)\ast Z_y$

where ${\displaystyle \left[x_{rd},y_{rd}\right]}$ represent the radially distorted image coordinates, ${\displaystyle \left[x_{\frac{1}{2}},y_{\frac{1}{2}}\right]}$ the coordinates of the image center, $r_d$ the Euclidean distance to the radial distortion center, $\lambda$ the distortion strength, ${\displaystyle \left[x_{ru},y_{ru}\right]}$ the radially undistorted coordinates, and $Z_x,Z_y$ axis-specific zoom factors. The tangential distortion, on the other hand, we corrected with

$x_{td}=x_{tu}-\frac{(x_{tu}-a_x)}{|x-a_x|}\ast \frac{{\kappa }_x}{p_y}\ast (y_{tu}-\mathrm{\Delta }{p}_y)\ast Z_x$

$y_{td}=y_{tu}-\frac{(y_{tu}-a_y)}{|y-a_y|}\ast \frac{{\kappa }_y}{p_x}\ast (x_{tu}-\mathrm{\Delta }{p}_x)\ast Z_y$

where ${\displaystyle \left[x_{td},y_{td}\right]}$ represent the tangentially distorted image coordinates, ${\displaystyle \left[x_{tu},y_{tu}\right]}$ the tangentially undistorted coordinates, ${\displaystyle \left[a_x,a_y\right]}$ the coordinates of the tangential distortion center, $\left[x,y\right]$ the size of the image, ${\displaystyle \left[{\kappa }_x,{\kappa }_y\right]}$ the tangential distortion strengths, ${\displaystyle \left[p_x,p_y\right]}$ the size of the interaction platform in the undistorted image, and ${\displaystyle \left[\mathrm{\Delta }{p}_x,\mathrm{\Delta }{p}_y\right]}$ the offset of the platform with respect to the top-left corner of the undistorted image.

Detection of ultrasonic vocalizations

Request a detailed protocol

USVs were detected automatically using a set of custom algorithms described elsewhere (Ivanenko et al., 2020). Detection was only performed on the USM4 data as their sensitivity and frequency range were generally better than for the Cam64 (see Figure 1C, Figure 1—figure supplement 1). A vocalization only had to be detected on one of the four high-quality microphones to be included into the set. In total, we collected 13,406 USVs, out of which 8424 occurred when the Cam64 recordings were active.

Automatic visual animal tracking

Request a detailed protocol

To assess whether we could reliably assign USVs to their emitter in a fast and scalable way, we automatically tracked multiple body parts of interacting mice in all frames — most importantly the snout and head center — for all dyadic trials (using DeepLabCut [Brudzynski, 2021]; see Figure 2) and a subset of triadic trials (using SLEAP [Pereira et al., 2022]; see Figure 6). With this approach, tracking is not temporally restricted to the midpoint of USV production, but can be performed for every frame of the entire recording. This data can be used to establish spatial densities of interaction against which, for example, the spatial density of vocalizations can be compared (Oliveira-Stahl et al., 2023).

For the dyadic recordings, mice were tracked offline using a combination of DeepLabCut (DLC) (Mathis et al., 2018) and extensive pos-processing to maintain animal identity over the entire recording. While the tracking results from DLC were generally quite accurate, we refrained from using them directly because of inaccuracies and identity switches that occurred on many hundreds of occasions in every recording. Instead we adopted a strategy where DLC generated an overcomplete set of candidate locations followed by custom synthesis and tracing of these alternatives in space and time (see Figure 3—figure supplement 1). In short, improved marker locations were generated from marker estimate clouds produced by DLC. Next, these marker positions were assembled into short spatiotemporal threads with the same, unknown identity based on a combination of spatial and temporal analysis. Finally, the thread ends were connected based on quadratic spatial trajectory estimates for each marker, yielding the complete track for both mice. This strategy resulted in reliable, high-quality tracking for all recordings, with a greatly reduced number of manual corrections needed overall (~10 per trial on average). All resulting tracks were visually verified (for a representative example, see Video 1).

Video 1

Download asset

For tracking the triadic interactions with two males, we used the SLEAP (Pereira et al., 2022) tracking system (version 1.3.0). To obtain the frame-by-frame pose estimations, we utilized the SLEAP graphical user interface to train a bottom-up U-net model, which is used to identify the body parts first and then attribute them to separate instances. Initially, we trained the model on the manually annotated frames from the dyadic tracking process. Subsequently, we annotated ~200 additional frames, all in triadic conditions in which the model exhibited poor performance. The extended dataset was then used to retrain the model. To establish the basis for triadic tracking, we employed SLEAP’s tracker to group the predicted instances across frames. The tracker compared instances across the full six-node skeleton and aimed to maximize the overall similarity across the three track assignments using the Hungarian algorithm. To identify candidate instances for comparison, it employed optical flow based on the previous five frames and selected instances based on the 0.95 quantile of similarity scores. We also applied SLEAP’s post-tracking data cleaning techniques to connect any breaks in single tracks. Subsequently, we examined all 14 recordings frame by frame to rectify any identity switches and eliminate inaccurate predictions. For instance, we addressed cases where two instances were detected on a single mouse or when one instance appeared to cover two mice. To further refine the results, we interpolated outlying instances based on velocity jumps.

We compared the accuracy of localization on the basis of manual tracking with that of automatic tracking (N = 5046 USVs, see Figure 3—figure supplement 3). Directly comparing the snout positions between the methods shows a median difference of 3.76 mm. The resulting error for localizing USVs was still superior to other systems, but significantly increased by ~0.9 mm (MAE = 5.71 mm) relative to manual tracking. Both manual and automatic tracking appear to have particular patterns of residual errors, indicated by the fact that the error between the tracking methods is much larger than their difference in USV localization error. The percentage of reliably assignable USVs interestingly increased to 93.6% (HyVL) compared to 92% with manual tracking for the dyadic recordings only. We optimized the mouth location on the snout-to-head-center line, finding an optimal distance of 15% of the snout to head center distance to the front of the animal. This indicated that the automatic tracking tended to place the snout tracking point a bit further into the snout than manual tracking, which might also explain the increase in assignment, due to a slight – but erroneous – increase in the separation between the snouts. While these results suggest that manual tracking is still advantageous, it highlights that completely automatic analysis of dyadic and possibly n-adic social interaction experiments is feasible at slightly reduced accuracy.

Manual visual animal tracking

Request a detailed protocol

To test the maximum precision of HyVL, we manually tracked the spatial locations of all mice during all USVs from the video data to assess the precision of the automatic visual and acoustic tracking. During manual tracking, the observer was presented with a combined display of the vocalization spectrogram and the concurrent video image at the temporal midpoint of each USV (MultiViewer, custom-written, MATLAB-based visualization tool). The display included a zoom function for optimal accuracy as tracking was click-based. Users could also freely scroll in time to ensure consistent animal identities. Only the snout and head center (i.e., midpoint between the ears) needed to be annotated because these points define a vector representing the head location and direction, which was all that was required in subsequent behavioral analyses.

Localization of ultrasonic vocalizations

Request a detailed protocol

USVs were spatially localized using a hybrid approach that integrates SLIM (Oliveira-Stahl et al., 2023) (based on four high-quality microphones) and beamforming (based on the 64-channel microphone array), drawing on the complementary strengths of the two microphone arrays (see Figure 1—figure supplement 1). For example, the Cam64 array provided excellent localization for USVs with energy below ~90 kHz due to the increasing noise floor of the MEMS microphones with sound frequency. Conversely, the four high-quality ultrasonic microphones (USM4) have a rather flat noise level as a function of frequency. On the other hand, USM4 will occasionally have glitches in one of the microphones, which can be compensated for in Cam64-based estimates through the number of microphones. As a consequence, the errors of the two methods show an L-shape (see Figure 3A), which highlights the synergy of a hybrid approach.

Acoustic localization using the Cam64 recordings was performed on the basis of delay-and-sum beamforming (Van Veen and Buckley, 1988). In beamforming, signals from all microphones are combined to estimate a spatial density that correlates with the probability of a given location being the origin of the sound. Specifically, we computed beamforming estimates for a surface situated 1 cm above and co-centered with the interaction platform, extending to 5 cm beyond all edges of the platform (i.e., 50 × 40 cm in total) at a final resolution of 1 mm in both dimensions. We refer to this density of sound origin as $D_{SO}\left(x,y\right)$ where $x$ and $y$ denote spatial coordinates. To prevent noises unrelated to a specific USV from contaminating the location estimate, we limited beamforming to a particular frequency range estimated from the simultaneous data of the USM4 array that enveloped the USV. Spatial density was defined as

${\displaystyle D_{SO}\left(x,y\right)=\sum _{f=F_{min}}^{F_{max}}{D}_{SO}\left(x,y,f\right)=\sum _{f=F_{min}}^{F_{max}}\sum _{m=1}^{64}{e}^{i2\pi fd\left(m,x,y,z\right)}}$

where $d\left(m,x,y,z\right)$ denotes the difference in arrival time at each microphone $m$ for sounds emitted from a location with coordinates $\left(x,y,z\right)$ , where $z$ is omitted in $D_{SO}\left(x,y\right)$ as it is a fixed distance to the plane of the microphone array. Beamforming was performed in the computational cloud backend provided by the Cam64 manufacturer, the so-called Sorama Portal (https://www.sorama.eu/sorama-portal).

The final beamforming estimate was calculated sequentially in two steps: first, a coarse estimate with 1 cm resolution was generated over the entire beamforming surface. Second, a fine-grained estimate with 1 mm resolution was generated over a 30 × 30 mm window centered on the peak location of the coarse estimate (see Figure 2 for an example). This two-step approach was chosen to optimize performance, as an estimate with 1 mm resolution over the entire beamforming surface would be computationally expensive while failing to produce a better result. For USVs of sufficient quality (i.e., containing frequency content below ~90 kHz while being sufficiently intense and long), both the coarse and fine estimates of $D_{SO}\left(x,y\right)$ contained a peak whose height was typically very large compared to the surrounding values at distances greater than a few centimeters. The peak location of the fine-grained estimate was used as the final estimate of the USV’s origin. To assess the quality of this location estimate, we computed a SNR per USV as follows:

$SN{R}_{Cam64}\left(v\right)=\frac{max\left(D_{SO}\left(x,y\right)\right)}{std\left(D_{SO}\left(x,y\right)\right)}$

where $D_{SO}\left(x,y\right)$ is assumed to be calculated for the USV $v$. The inverse, $1/{SNR}_{Cam64}$ was used as a proxy for the uncertainty of localization for a given USV.

Localization from the USM4 recordings was performed using the SLIM method (Oliveira-Stahl et al., 2023). Briefly, SLIM analytically estimates submanifolds (in 2D: surfaces) of a sound’s spatial origin for each pair of microphones and combines these into a single estimate by intersecting the manifolds (in 2D: lines). The intersection has an associated uncertainty that scales with the uncertainty of the localization estimate for a given USV, specifically the uncertainty was defined as the standard deviation of all locations that were >90% times the maximum of the intersection density of all origin curves.

Lastly, for each USV where both Cam64 and SLIM location estimates ${\displaystyle {\dot{X}}_{Cam64}}$ and ${\displaystyle {\dot{X}}_{SLIM}}$ were available, a single estimate ${\displaystyle {\dot{X}}_{HyVL}}$ was computed based on the two estimates, spatial uncertainties and their spatial relation to the mice at the current time (see below).

USV assignment

Request a detailed protocol

The final hybrid location estimate and assignment to a mouse was performed while taking into account the probability of making a false assignment as proposed before (Neunuebel et al., 2015), through the calculation of the mouse probability index $MPI$. While the $MPI$ was previously only used to exclude uncertain assignments (e.g., if two mice are nearly equidistant to the estimated sound location), we also adapted it here to select and combine the location estimates. The ${MPI}_k$ for each mouse k was computed as,

${MPI}_k=\frac{P_k}{{\sum }_{m=1}^n{P}_m}$

Here, $P_k$ is the probability that the USV in question originated from mouse $k$ computed as

${\displaystyle P_k=N\left({\dot{X}}_{Method}-X_{mouth,k},{\sigma }_{Method}^2\right)}$ , where ${\displaystyle {\dot{X}}_{Method}}$ is an estimate of the acoustic origin, ${\displaystyle X_{mouth,k}}$ the position of the mouth of mouse $k$, and ${\sigma }_{Method}^2$ the uncertainty of the estimate, with ${Ẋ}_{Method}$ and ${\sigma }_{Method}^2$ specific to the Method used. $X_{mouth,k}$ was assumed to lie on a line connecting the snout and head-center. For manually tracked recordings, the optimal location on this line was close to the snout (~2% toward the head, where % is relative to the snout-to-head-center tracked distance), while in the automatic tracking it was ahead of the snout tracking point (~15% away from the head). ${\sigma }_{Method}^2$ was computed for each USV as the method’s intrinsic per-USV uncertainty estimate. As these uncertainty estimates only correlate with the absolute uncertainty (i.e., in millimeters), we scaled them such that their average across all USVs matched the residual error of each method in the Far-condition (all animals >100 mm apart, see Figure 3C and Oliveira-Stahl et al., 2023). In this way, the ${MPI}_k$ for individual USVs took into account the uncertainty of each method: if the uncertainty of one method was higher, probabilities across mice would become more similar and the ${MPI}_k$ would reduce.

For a given USV, we computed the ${MPI}_k$ for all mice for both methods. The mouse with the largest ${MPI}_k$ per method, which coincides with the mouse at the smallest distance to the estimate, was denoted as ${MPI}_{Cam64}$ and ${MPI}_{SLIM}$ , respectively. If only one of the two exceeded 0.95, this method’s estimate was selected. If both exceeded 0.95, then the estimate with the smaller distance to the mouse with the highest ${MPI}_k$ was chosen. This combination ensured that only reliable assignments were performed, while minimizing the residual error. Similar to Neunuebel et al., 2015, we also excluded estimates that were too far away from any mouse (50 mm). This distance threshold mainly serves to compensate for a deficiency of the $MPI$: if all mice are far from the estimate, all $P_k$ are extremely small; however, the ${MPI}_k$ will often exceed 0.95. The distance threshold corresponds to setting the individual $P_k=0$ in the ${MPI}_k$ , thus excluding candidate mice that are highly unlikely to be the source of the USV. USVs that had no ${MPI}_k$ > 0.95 for either method were excluded from further analysis. The fraction of included USVs is referred to as selected in the plots. Maximizing this fraction is essential to perform a complete analysis of vocal communication.

We compared the above-described combination strategy to a large number of alternative strategies, including maximum likelihood combination of estimators (Ernst and Banks, 2002), or selecting directly based on the largest ${MPI}_k$ or largest $P_k$ . While all these approaches led to broadly similar results, the described approach achieved the most robust and reliable results (see ‘Discussion’ for additional details).

Audiovisual alignment

Request a detailed protocol

For both microphone sets, precise measurements of their location in relation to the camera’s location were used to position acoustic estimates in the coordinate system of the images provided by the camera. In the final analysis, we noticed for each microphone set small, systematic (0.5–2 mm) shifts in both X and Y. We interpreted these as very small measurement errors in the relative positions of the camera or microphone arrays and corrected these post hoc in the setup definition, followed by rerunning all subsequent analysis steps. This reduced all systematic shifts to near 0.

Spatial vocalization analysis

Request a detailed protocol

To gain insight into the spatial positioning of the interacting mice, we represented the relative animal positions in a polar reference frame centered on the snout of the emitter. In this format, the radial distance corresponded to the snout–snout distance and the radial angle described the relative angle between the gaze direction of the emitter and the snout position of the recipient (i.e., with the line from the head center to the snout of the emitter pointing towards 0°; see also Figure 4E).

The position density of the recipient mouse was collected in cumulative fashion, with the polar coordinate system translated appropriately for each USV based on its temporal midpoint. We assumed that the mice had no preference for relative vocalizations to either side of their snout, so all relative spatial positions were agglomerated in the right hemispace for further analysis. All data points were then binned using a polar, raw-count histogram with bins of 10° and 1 cm.

All code necessary to implement the HyVL system has been deposited at https://github.com/benglitz/HyVL (copy archived at Englitz, 2023) and https://doi.org/10.34973/7kgc-ta72. All data has been made available at https://doi.org/10.34973/7kgc-ta72.

1. 1. Sterling M
   2. Englitz B
   3. Teunisse R

   (2023) Donders Repository

   Ultrasonic vocal interaction resolved with millimeter precision using hybrid beamforming.

   https://doi.org/10.34973/7kgc-ta72

1. 1. Agarwalla S
   2. Arroyo NS
   3. Long NE
   4. O’Brien WT
   5. Abel T
   6. Bandyopadhyay S

   (2020) Male-specific alterations in structure of isolation call sequences of mouse pups with 16p11.2 deletion

   Genes, Brain, and Behavior 19:e12681.

   https://doi.org/10.1111/gbb.12681

   • PubMed
   • Google Scholar
2. 1. Arriaga G
   2. Jarvis ED

   (2013) Mouse vocal communication system: are ultrasounds learned or innate?

   Brain and Language 124:96–116.

   https://doi.org/10.1016/j.bandl.2012.10.002

   • PubMed
   • Google Scholar
3. 1. Bakker J
   2. Langermans JAM

   (2018)

   Handbook of Ultrasonic Vocalization - A Window into the Emotional Brain

   535–544, Ultrasonic components of vocalizations in Marmosets, Handbook of Ultrasonic Vocalization - A Window into the Emotional Brain, Elsevier.

   • Google Scholar
4. 1. Barnes TD
   2. Rieger MA
   3. Dougherty JD
   4. Holy TE

   (2017) Group and individual variability in mouse pup isolation calls recorded on the same day show stability

   Frontiers in Behavioral Neuroscience 11:243.

   https://doi.org/10.3389/fnbeh.2017.00243

   • PubMed
   • Google Scholar
5. 1. Binder MS
   2. Shi HD
   3. Bordey A

   (2021) CD-1 Outbred Mice Produce Less Variable Ultrasonic Vocalizations Than FVB Inbred Mice, While Displaying a Similar Developmental Trajectory

   Frontiers in Psychiatry 12:687060.

   https://doi.org/10.3389/fpsyt.2021.687060

   • PubMed
   • Google Scholar
6. 1. Brudzynski SM

   (2021) Biological functions of rat ultrasonic vocalizations, arousal mechanisms, and call initiation

   Brain Sciences 11:605.

   https://doi.org/10.3390/brainsci11050605

   • PubMed
   • Google Scholar
7. 1. Burke K
   2. Screven LA
   3. Dent ML

   (2018) CBA/CaJ mouse ultrasonic vocalizations depend on prior social experience

   PLOS ONE 13:e0197774.

   https://doi.org/10.1371/journal.pone.0197774

   • PubMed
   • Google Scholar
8. 1. Castellucci GA
   2. Calbick D
   3. McCormick D

   (2018) The temporal organization of mouse ultrasonic vocalizations

   PLOS ONE 13:e0199929.

   https://doi.org/10.1371/journal.pone.0199929

   • PubMed
   • Google Scholar
9. 1. Chabout J
   2. Serreau P
   3. Ey E
   4. Bellier L
   5. Aubin T
   6. Bourgeron T
   7. Granon S

   (2012) Adult male mice emit context-specific ultrasonic vocalizations that are modulated by prior isolation or group rearing environment

   PLOS ONE 7:e29401.

   https://doi.org/10.1371/journal.pone.0029401

   • PubMed
   • Google Scholar
10. 1. Chabout J
    2. Sarkar A
    3. Dunson DB
    4. Jarvis ED

    (2015) Male mice song syntax depends on social contexts and influences female preferences

    Frontiers in Behavioral Neuroscience 9:76.

    https://doi.org/10.3389/fnbeh.2015.00076

    • PubMed
    • Google Scholar
11. 1. Chabout J
    2. Jones-Macopson J
    3. Jarvis ED

    (2017) Eliciting and analyzing male mouse ultrasonic vocalization (usv) songs

    Journal of Visualized Experiments 01:54137.

    https://doi.org/10.3791/54137

    • PubMed
    • Google Scholar
12. 1. Chen J
    2. Markowitz JE
    3. Lilascharoen V
    4. Taylor S
    5. Sheurpukdi P
    6. Keller JA
    7. Jensen JR
    8. Lim BK
    9. Datta SR
    10. Stowers L

    (2021) Flexible scaling and persistence of social vocal communication

    Nature 593:108–113.

    https://doi.org/10.1038/s41586-021-03403-8

    • PubMed
    • Google Scholar
13. 1. Choi H
    2. Park S
    3. Kim D

    (2011) Two genetic loci control syllable sequences of ultrasonic courtship vocalizations in inbred mice

    BMC Neuroscience 12:104.

    https://doi.org/10.1186/1471-2202-12-104

    • PubMed
    • Google Scholar
14. 1. Ciucci MR
    2. Ahrens AM
    3. Ma ST
    4. Kane JR
    5. Windham EB
    6. Woodlee MT
    7. Schallert T

    (2009) Reduction of dopamine synaptic activity: degradation of 50-kHz ultrasonic vocalization in rats

    Behavioral Neuroscience 123:328–336.

    https://doi.org/10.1037/a0014593

    • PubMed
    • Google Scholar
15. 1. Coffey KR
    2. Marx RE
    3. Neumaier JF

    (2019) DeepSqueak: a deep learning-based system for detection and analysis of ultrasonic vocalizations

    Neuropsychopharmacology 44:859–868.

    https://doi.org/10.1038/s41386-018-0303-6

    • Google Scholar
16. 1. de Chaumont F
    2. Lemière N
    3. Coqueran S
    4. Bourgeron T
    5. Ey E

    (2021) LMT USV Toolbox, A Novel Methodological Approach to Place Mouse Ultrasonic Vocalizations in Their Behavioral Contexts-A Study in Female and Male C57BL/6J Mice and in Shank3 Mutant Females

    Frontiers in Behavioral Neuroscience 15:735920.

    https://doi.org/10.3389/fnbeh.2021.735920

    • PubMed
    • Google Scholar
17. 1. Dou X
    2. Shirahata S
    3. Sugimoto H
    4. Abreu-Villaça Y

    (2018) Functional clustering of mouse ultrasonic vocalization data

    PLOS ONE 13:e0196834.

    https://doi.org/10.1371/journal.pone.0196834

    • Google Scholar
18. Software

    1. Englitz B

    (2023) Hyvl - hybrid vocalization Localizer, version swh:1:rev:7ab9f02a38d8ae5b668e4801463bb22367da4637

    Software Heritage.

    https://archive.softwareheritage.org/swh:1:dir:9d9e1598ff96a36cdc3d45cd54c5de47746fe3cd;origin=https://github.com/benglitz/HyVL;visit=swh:1:snp:b6db3bffd9022374a42a8fdb8464bcf19d8bae1e;anchor=swh:1:rev:7ab9f02a38d8ae5b668e4801463bb22367da4637
19. 1. Erata E
    2. Gao Y
    3. Purkey AM
    4. Soderblom EJ
    5. McNamara JO
    6. Soderling SH

    (2021) Cnksr2 Loss in Mice Leads to Increased Neural Activity and Behavioral Phenotypes of Epilepsy-Aphasia Syndrome

    The Journal of Neuroscience 41:9633–9649.

    https://doi.org/10.1523/JNEUROSCI.0650-21.2021

    • PubMed
    • Google Scholar
20. 1. Ernst MO
    2. Banks MS

    (2002) Humans integrate visual and haptic information in a statistically optimal fashion

    Nature 415:429–433.

    https://doi.org/10.1038/415429a

    • PubMed
    • Google Scholar
21. 1. Feng AS
    2. Narins PM
    3. Xu C-H
    4. Lin W-Y
    5. Yu Z-L
    6. Qiu Q
    7. Xu Z-M
    8. Shen J-X

    (2006) Ultrasonic communication in frogs

    Nature 440:333–336.

    https://doi.org/10.1038/nature04416

    • PubMed
    • Google Scholar
22. 1. Fonseca AH
    2. Santana GM
    3. Bosque Ortiz GM
    4. Bampi S
    5. Dietrich MO

    (2021) Analysis of ultrasonic vocalizations from mice using computer vision and machine learning

    eLife 10:e59161.

    https://doi.org/10.7554/eLife.59161

    • PubMed
    • Google Scholar
23. 1. Fröhlich H
    2. Rafiullah R
    3. Schmitt N
    4. Abele S
    5. Rappold GA

    (2017) Foxp1 expression is essential for sex-specific murine neonatal ultrasonic vocalization

    Human Molecular Genetics 26:1511–1521.

    https://doi.org/10.1093/hmg/ddx055

    • PubMed
    • Google Scholar
24. 1. Fujita E
    2. Tanabe Y
    3. Imhof BA
    4. Momoi MY
    5. Momoi T

    (2012) Cadm1-expressing synapses on Purkinje cell dendrites are involved in mouse ultrasonic vocalization activity

    PLOS ONE 7:e30151.

    https://doi.org/10.1371/journal.pone.0030151

    • PubMed
    • Google Scholar
25. 1. Gao SC
    2. Wei YC
    3. Wang SR
    4. Xu XH

    (2019) Medial preoptic area modulates courtship ultrasonic vocalization in adult male mice

    Neuroscience Bulletin 35:697–708.

    https://doi.org/10.1007/s12264-019-00365-w

    • PubMed
    • Google Scholar
26. 1. Guoynes CD
    2. Marler CA

    (2021) An acute dose of intranasal oxytocin rapidly increases maternal communication and maintains maternal care in primiparous postpartum California mice

    PLOS ONE 16:e0244033.

    https://doi.org/10.1371/journal.pone.0244033

    • PubMed
    • Google Scholar
27. 1. Hammerschmidt K
    2. Radyushkin K
    3. Ehrenreich H
    4. Fischer J

    (2009) Female mice respond to male ultrasonic “songs” with approach behaviour

    Biology Letters 5:589–592.

    https://doi.org/10.1098/rsbl.2009.0317

    • PubMed
    • Google Scholar
28. 1. Heckman JJ
    2. Proville R
    3. Heckman GJ
    4. Azarfar A
    5. Celikel T
    6. Englitz B

    (2017) High-precision spatial localization of mouse vocalizations during social interaction

    Scientific Reports 7:3017.

    https://doi.org/10.1038/s41598-017-02954-z

    • PubMed
    • Google Scholar
29. 1. Hepbasli D
    2. Gredy S
    3. Ullrich M
    4. Reigl A
    5. Abeßer M
    6. Raabe T
    7. Schuh K

    (2021) Genotype- and Age-Dependent Differences in Ultrasound Vocalizations of SPRED2 Mutant Mice Revealed by Machine Deep Learning

    Brain Sciences 11:1365.

    https://doi.org/10.3390/brainsci11101365

    • PubMed
    • Google Scholar
30. 1. Hertz S
    2. Weiner B
    3. Perets N
    4. London M

    (2020) Temporal structure of mouse courtship vocalizations facilitates syllable labeling

    Communications Biology 3:333.

    https://doi.org/10.1038/s42003-020-1053-7

    • PubMed
    • Google Scholar
31. 1. Hodges SL
    2. Nolan SO
    3. Reynolds CD
    4. Lugo JN

    (2017) Spectral and temporal properties of calls reveal deficits in ultrasonic vocalizations of adult Fmr1 knockout mice

    Behavioural Brain Research 332:50–58.

    https://doi.org/10.1016/j.bbr.2017.05.052

    • PubMed
    • Google Scholar
32. 1. Hoier S
    2. Pfeifle C
    3. von Merten S
    4. Linnenbrink M

    (2016) Communication at the garden fence--context dependent vocalization in female house mice

    PLOS ONE 11:e0152255.

    https://doi.org/10.1371/journal.pone.0152255

    • PubMed
    • Google Scholar
33. 1. Holy TE
    2. Guo Z

    (2005) Ultrasonic songs of male mice

    PLOS Biology 3:e386.

    https://doi.org/10.1371/journal.pbio.0030386

    • PubMed
    • Google Scholar
34. 1. Ivanenko A
    2. Watkins P
    3. van Gerven MAJ
    4. Hammerschmidt K
    5. Englitz B
    6. Theunissen FE

    (2020) Classifying sex and strain from mouse ultrasonic vocalizations using deep learning

    PLOS Computational Biology 16:e1007918.

    https://doi.org/10.1371/journal.pcbi.1007918

    • Google Scholar
35. 1. Kappel S
    2. Hawkins P
    3. Mendl MT

    (2017) To group or not to group? Good practice for housing male laboratory mice

    Animals 7:88.

    https://doi.org/10.3390/ani7120088

    • PubMed
    • Google Scholar
36. 1. Keesom SM
    2. Finton CJ
    3. Sell GL
    4. Hurley LM

    (2017) Early-life social isolation influences mouse ultrasonic vocalizations during male-male social encounters

    PLOS ONE 12:e0169705.

    https://doi.org/10.1371/journal.pone.0169705

    • PubMed
    • Google Scholar
37. 1. Kikusui T
    2. Nakanishi K
    3. Nakagawa R
    4. Nagasawa M
    5. Mogi K
    6. Okanoya K

    (2011) Cross fostering experiments suggest that mice songs are innate

    PLOS ONE 6:e17721.

    https://doi.org/10.1371/journal.pone.0017721

    • PubMed
    • Google Scholar
38. 1. Kikusui T
    2. Shima Y
    3. Sonobe M
    4. Yoshida Y
    5. Nagasawa M
    6. Nomoto K
    7. Mogi K

    (2021a) Testosterone regulates the emission of ultrasonic vocalizations and mounting behavior during different developmental periods in mice

    Developmental Psychobiology 63:725–733.

    https://doi.org/10.1002/dev.22045

    • PubMed
    • Google Scholar
39. 1. Kikusui T
    2. Sonobe M
    3. Yoshida Y
    4. Nagasawa M
    5. Ey E
    6. de Chaumont F
    7. Bourgeron T
    8. Nomoto K
    9. Mogi K

    (2021b) Testosterone increases the emission of ultrasonic vocalizations with different acoustic characteristics in mice

    Frontiers in Psychology 12:680176.

    https://doi.org/10.3389/fpsyg.2021.680176

    • PubMed
    • Google Scholar
40. 1. Kondrakiewicz K
    2. Kostecki M
    3. Szadzińska W
    4. Knapska E

    (2019) Ecological validity of social interaction tests in rats and mice

    Genes, Brain, and Behavior 18:e12525.

    https://doi.org/10.1111/gbb.12525

    • PubMed
    • Google Scholar
41. 1. Lee NS
    2. Beery AK

    (2019) Neural circuits underlying rodent sociality: A comparative approach

    Current Topics in Behavioral Neurosciences 43:211–238.

    https://doi.org/10.1007/7854_2018_77

    • PubMed
    • Google Scholar
42. 1. Litvin Y
    2. Blanchard DC
    3. Blanchard RJ

    (2007) Rat 22kHz ultrasonic vocalizations as alarm cries

    Behavioural Brain Research 182:166–172.

    https://doi.org/10.1016/j.bbr.2006.11.038

    • PubMed
    • Google Scholar
43. 1. Liu RC
    2. Miller KD
    3. Merzenich MM
    4. Schreiner CE

    (2003) Acoustic variability and distinguishability among mouse ultrasound vocalizations

    The Journal of the Acoustical Society of America 114:3412–3422.

    https://doi.org/10.1121/1.1623787

    • PubMed
    • Google Scholar
44. 1. Mahrt EJ
    2. Perkel DJ
    3. Tong L
    4. Rubel EW
    5. Portfors CV

    (2013) Engineered deafness reveals that mouse courtship vocalizations do not require auditory experience

    The Journal of Neuroscience 33:5573–5583.

    https://doi.org/10.1523/JNEUROSCI.5054-12.2013

    • PubMed
    • Google Scholar
45. 1. Mahrt E
    2. Agarwal A
    3. Perkel D
    4. Portfors C
    5. Elemans CPH

    (2016) Mice produce ultrasonic vocalizations by intra-laryngeal planar impinging jets

    Current Biology 26:R880–R881.

    https://doi.org/10.1016/j.cub.2016.08.032

    • PubMed
    • Google Scholar
46. 1. Marconi MA
    2. Nicolakis D
    3. Abbasi R
    4. Penn DJ
    5. Zala SM

    (2020) Ultrasonic courtship vocalizations of male house mice contain distinct individual signatures

    Animal Behaviour 169:169–197.

    https://doi.org/10.1016/j.anbehav.2020.09.006

    • Google Scholar
47. 1. Mathis A
    2. Mamidanna P
    3. Cury KM
    4. Abe T
    5. Murthy VN
    6. Mathis MW
    7. Bethge M

    (2018) DeepLabCut: markerless pose estimation of user-defined body parts with deep learning

    Nature Neuroscience 21:1281–1289.

    https://doi.org/10.1038/s41593-018-0209-y

    • PubMed
    • Google Scholar
48. 1. Menuet C
    2. Cazals Y
    3. Gestreau C
    4. Borghgraef P
    5. Gielis L
    6. Dutschmann M
    7. Van Leuven F
    8. Hilaire G

    (2011) Age-related impairment of ultrasonic vocalization in Tau.P301L mice: possible implication for progressive language disorders

    PLOS ONE 6:e25770.

    https://doi.org/10.1371/journal.pone.0025770

    • PubMed
    • Google Scholar
49. 1. Michael V
    2. Goffinet J
    3. Pearson J
    4. Wang F
    5. Tschida K
    6. Mooney R

    (2020) Circuit and synaptic organization of forebrain-to-midbrain pathways that promote and suppress vocalization

    eLife 9:e63493.

    https://doi.org/10.7554/eLife.63493

    • PubMed
    • Google Scholar
50. 1. Moskal JR
    2. Burgdorf J

    (2018)

    Handbook of Ultrasonic Vocalization - A Window into the Emotional Brain

    413–421, Ultrasonic vocalizations in rats as a measure of emotional responses to stress: models of anxiety and depression, Handbook of Ultrasonic Vocalization - A Window into the Emotional Brain, Elsevier.

    • Google Scholar
51. 1. Mourlam MJ
    2. Orliac MJ

    (2017) Infrasonic and ultrasonic hearing evolved after the emergence of modern whales

    Current Biology 27:1776–1781.

    https://doi.org/10.1016/j.cub.2017.04.061

    • PubMed
    • Google Scholar
52. 1. Mun HS
    2. Lipina TV
    3. Roder JC

    (2015) Ultrasonic vocalizations in mice during exploratory behavior are context-dependent

    Frontiers in Behavioral Neuroscience 9:316.

    https://doi.org/10.3389/fnbeh.2015.00316

    • PubMed
    • Google Scholar
53. 1. Murrant MN
    2. Bowman J
    3. Garroway CJ
    4. Prinzen B
    5. Mayberry H
    6. Faure PA

    (2013) Ultrasonic vocalizations emitted by flying squirrels

    PLOS ONE 8:e73045.

    https://doi.org/10.1371/journal.pone.0073045

    • PubMed
    • Google Scholar
54. 1. Musolf K
    2. Hoffmann F
    3. Penn DJ

    (2010) Ultrasonic courtship vocalizations in wild house mice, Mus musculus

    Animal Behaviour 79:757–764.

    https://doi.org/10.1016/j.anbehav.2009.12.034

    • Google Scholar
55. 1. Musolf K
    2. Meindl S
    3. Larsen AL
    4. Kalcounis-Rueppell MC
    5. Penn DJ

    (2015) Ultrasonic vocalizations of male mice differ among species and females show assortative preferences for male calls

    PLOS ONE 10:e0134123.

    https://doi.org/10.1371/journal.pone.0134123

    • PubMed
    • Google Scholar
56. 1. Neunuebel JP
    2. Taylor AL
    3. Arthur BJ
    4. Egnor SER

    (2015) Female mice ultrasonically interact with males during courtship displays

    eLife 4:e06203.

    https://doi.org/10.7554/eLife.06203

    • PubMed
    • Google Scholar
57. 1. Nicolakis D
    2. Marconi MA
    3. Zala SM
    4. Penn DJ

    (2020) Ultrasonic vocalizations in house mice depend upon genetic relatedness of mating partners and correlate with subsequent reproductive success

    Frontiers in Zoology 17:10.

    https://doi.org/10.1186/s12983-020-00353-1

    • PubMed
    • Google Scholar
58. 1. Nunez AA
    2. Nyby J
    3. Whitney G

    (1978) The effects of testosterone, estradiol, and dihydrotestosterone on male mouse (Mus musculus) ultrasonic vocalizations

    Hormones and Behavior 11:264–272.

    https://doi.org/10.1016/0018-506x(78)90030-2

    • PubMed
    • Google Scholar
59. 1. Oliveira-Stahl G
    2. Farboud S
    3. Sterling ML
    4. Heckman JJ
    5. van Raalte B
    6. Lenferink D
    7. van der Stam A
    8. Smeets CJLM
    9. Fisher SE
    10. Englitz B

    (2023) High-precision spatial analysis of mouse courtship vocalization behavior reveals sex and strain differences

    Scientific Reports 13:5219.

    https://doi.org/10.1038/s41598-023-31554-3

    • PubMed
    • Google Scholar
60. 1. Palazzo E
    2. Fu Y
    3. Ji G
    4. Maione S
    5. Neugebauer V

    (2008) Group III mGluR7 and mGluR8 in the amygdala differentially modulate nocifensive and affective pain behaviors

    Neuropharmacology 55:537–545.

    https://doi.org/10.1016/j.neuropharm.2008.05.007

    • PubMed
    • Google Scholar
61. 1. Palmateer J
    2. Pan J
    3. Pandya A
    4. Martin L
    5. Kumar S
    6. Ofomata A
    7. Jones TA
    8. Gore AC
    9. Schallert T
    10. Hurn PD

    (2016) Ultrasonic vocalization in murine experimental stroke: A mechanistic model of aphasia

    Restorative Neurology and Neuroscience 34:287–295.

    https://doi.org/10.3233/RNN-150583

    • PubMed
    • Google Scholar
62. 1. Pereira TD
    2. Tabris N
    3. Matsliah A
    4. Turner DM
    5. Li J
    6. Ravindranath S
    7. Papadoyannis ES
    8. Normand E
    9. Deutsch DS
    10. Wang ZY
    11. McKenzie-Smith GC
    12. Mitelut CC
    13. Castro MD
    14. D’Uva J
    15. Kislin M
    16. Sanes DH
    17. Kocher SD
    18. Wang SS-H
    19. Falkner AL
    20. Shaevitz JW
    21. Murthy M

    (2022) Publisher Correction: SLEAP: A deep learning system for multi-animal pose tracking

    Nature Methods 19:486–495.

    https://doi.org/10.1038/s41592-022-01495-2

    • PubMed
    • Google Scholar
63. 1. Petric R
    2. Kalcounis-Rueppell MC

    (2013) Female and male adult brush mice (Peromyscus boylii) use ultrasonic vocalizations in the wild

    Behaviour 150:1747–1766.

    https://doi.org/10.1163/1568539X-00003118

    • Google Scholar
64. 1. Pomerantz SM
    2. Clemens LG

    (1981) Ultrasonic vocalizations in male deer mice (Peromyscus maniculatus bairdi): their role in male sexual behavior

    Physiology & Behavior 27:869–872.

    https://doi.org/10.1016/0031-9384(81)90055-x

    • PubMed
    • Google Scholar
65. 1. Portfors CV

    (2007)

    Types and functions of ultrasonic vocalizations in laboratory rats and mice

    Journal of the American Association for Laboratory Animal Science 46:28–34.

    • PubMed
    • Google Scholar
66. 1. Portfors CV
    2. Perkel DJ

    (2014) The role of ultrasonic vocalizations in mouse communication

    Current Opinion in Neurobiology 28:115–120.

    https://doi.org/10.1016/j.conb.2014.07.002

    • PubMed
    • Google Scholar
67. 1. Pultorak JD
    2. Fuxjager MJ
    3. Kalcounis-Rueppell MC
    4. Marler CA

    (2015) Male fidelity expressed through rapid testosterone suppression of ultrasonic vocalizations to novel females in the monogamous California mouse

    Hormones and Behavior 70:47–56.

    https://doi.org/10.1016/j.yhbeh.2015.02.003

    • PubMed
    • Google Scholar
68. 1. Pultorak JD
    2. Matusinec KR
    3. Miller ZK
    4. Marler CA

    (2017) Ultrasonic vocalization production and playback predicts intrapair and extrapair social behaviour in a monogamous mouse

    Animal Behaviour 125:13–23.

    https://doi.org/10.1016/j.anbehav.2016.12.023

    • Google Scholar
69. 1. Pultorak JD
    2. Alger SJ
    3. Loria SO
    4. Johnson AM
    5. Marler CA

    (2018) Changes in behavior and ultrasonic vocalizations during pair bonding and in response to an infidelity challenge in monogamous california mice

    Frontiers in Ecology and Evolution 6:125.

    https://doi.org/10.3389/fevo.2018.00125

    • Google Scholar
70. 1. Ramsier MA
    2. Cunningham AJ
    3. Moritz GL
    4. Finneran JJ
    5. Williams CV
    6. Ong PS
    7. Gursky-Doyen SL
    8. Dominy NJ

    (2012) Primate communication in the pure ultrasound

    Biology Letters 8:508–511.

    https://doi.org/10.1098/rsbl.2011.1149

    • Google Scholar
71. 1. Rieger NS
    2. Marler CA

    (2018) The function of ultrasonic vocalizations during territorial defence by pair-bonded male and female California mice

    Animal Behaviour 135:97–108.

    https://doi.org/10.1016/j.anbehav.2017.11.008

    • Google Scholar
72. 1. Rieger NS
    2. Monari PK
    3. Hartfield K
    4. Schefelker J
    5. Marler CA

    (2021) Pair-bonding leads to convergence in approach behavior to conspecific vocalizations in California mice (Peromyscus californicus)

    PLOS ONE 16:e0255295.

    https://doi.org/10.1371/journal.pone.0255295

    • PubMed
    • Google Scholar
73. 1. Rotschafer SE
    2. Trujillo MS
    3. Dansie LE
    4. Ethell IM
    5. Razak KA

    (2012) Minocycline treatment reverses ultrasonic vocalization production deficit in a mouse model of Fragile X Syndrome

    Brain Research 1439:7–14.

    https://doi.org/10.1016/j.brainres.2011.12.041

    • PubMed
    • Google Scholar
74. 1. Sangiamo DT
    2. Warren MR
    3. Neunuebel JP

    (2020) Ultrasonic signals associated with different types of social behavior of mice

    Nature Neuroscience 23:411–422.

    https://doi.org/10.1038/s41593-020-0584-z

    • PubMed
    • Google Scholar
75. 1. Scattoni ML
    2. Crawley J
    3. Ricceri L

    (2009) Ultrasonic vocalizations: a tool for behavioural phenotyping of mouse models of neurodevelopmental disorders

    Neuroscience and Biobehavioral Reviews 33:508–515.

    https://doi.org/10.1016/j.neubiorev.2008.08.003

    • PubMed
    • Google Scholar
76. 1. Schnitzler HU
    2. Moss CF
    3. Denzinger A

    (2003) From spatial orientation to food acquisition in echolocating bats

    Trends in Ecology & Evolution 18:386–394.

    https://doi.org/10.1016/S0169-5347(03)00185-X

    • Google Scholar
77. 1. Shemesh Y
    2. Sztainberg Y
    3. Forkosh O
    4. Shlapobersky T
    5. Chen A
    6. Schneidman E

    (2013) High-order social interactions in groups of mice

    eLife 2:e00759.

    https://doi.org/10.7554/eLife.00759

    • PubMed
    • Google Scholar
78. 1. Shepard KN
    2. Chong KK
    3. Liu RC

    (2016) Contrast enhancement without transient map expansion for species-specific vocalizations in core auditory cortex during learning

    eNeuro 3:ENEURO.0318-16.2016.

    https://doi.org/10.1523/ENEURO.0318-16.2016

    • PubMed
    • Google Scholar
79. 1. Silverman JL
    2. Yang M
    3. Lord C
    4. Crawley JN

    (2010) Behavioural phenotyping assays for mouse models of autism

    Nature Reviews. Neuroscience 11:490–502.

    https://doi.org/10.1038/nrn2851

    • PubMed
    • Google Scholar
80. 1. Sugimoto H
    2. Okabe S
    3. Kato M
    4. Koshida N
    5. Shiroishi T
    6. Mogi K
    7. Kikusui T
    8. Koide T

    (2011) A role for strain differences in waveforms of ultrasonic vocalizations during male-female interaction

    PLOS ONE 6:e22093.

    https://doi.org/10.1371/journal.pone.0022093

    • PubMed
    • Google Scholar
81. 1. Tasaka G-I
    2. Guenthner CJ
    3. Shalev A
    4. Gilday O
    5. Luo L
    6. Mizrahi A

    (2018) Genetic tagging of active neurons in auditory cortex reveals maternal plasticity of coding ultrasonic vocalizations

    Nature Communications 9:871.

    https://doi.org/10.1038/s41467-018-03183-2

    • PubMed
    • Google Scholar
82. 1. Timonin ME
    2. Kalcounis‐Rueppell MC
    3. Marler CA
    4. Ebensperger L

    (2018) Testosterone pulses at the nest site modify ultrasonic vocalization types in a monogamous and territorial mouse

    Ethology 124:804–815.

    https://doi.org/10.1111/eth.12812

    • Google Scholar
83. 1. Tsai PT
    2. Hull C
    3. Chu Y
    4. Greene-Colozzi E
    5. Sadowski AR
    6. Leech JM
    7. Steinberg J
    8. Crawley JN
    9. Regehr WG
    10. Sahin M

    (2012) Autistic-like behaviour and cerebellar dysfunction in Purkinje cell Tsc1 mutant mice

    Nature 488:647–651.

    https://doi.org/10.1038/nature11310

    • PubMed
    • Google Scholar
84. 1. Tschida K
    2. Michael V
    3. Takatoh J
    4. Han B-X
    5. Zhao S
    6. Sakurai K
    7. Mooney R
    8. Wang F

    (2019) A specialized neural circuit gates social vocalizations in the mouse

    Neuron 103:459–472.

    https://doi.org/10.1016/j.neuron.2019.05.025

    • PubMed
    • Google Scholar
85. 1. Tsuji C
    2. Fujisaku T
    3. Tsuji T

    (2020) Oxytocin ameliorates maternal separation-induced ultrasonic vocalisation calls in mouse pups prenatally exposed to valproic acid

    Journal of Neuroendocrinology 32:e12850.

    https://doi.org/10.1111/jne.12850

    • PubMed
    • Google Scholar
86. 1. Tsuji T
    2. Mizutani R
    3. Minami K
    4. Furuhara K
    5. Fujisaku T
    6. Pinyue F
    7. Jing Z
    8. Tsuji C

    (2021) Oxytocin administration modulates the complex type of ultrasonic vocalisation of mice pups prenatally exposed to valproic acid

    Neuroscience Letters 758:135985.

    https://doi.org/10.1016/j.neulet.2021.135985

    • PubMed
    • Google Scholar
87. 1. Urbanus BHA
    2. Peter S
    3. Fisher SE
    4. De Zeeuw CI

    (2020) Region-specific Foxp2 deletions in cortex, striatum or cerebellum cannot explain vocalization deficits observed in spontaneous global knockouts

    Scientific Reports 10:21631.

    https://doi.org/10.1038/s41598-020-78531-8

    • PubMed
    • Google Scholar
88. 1. Van Segbroeck M
    2. Knoll AT
    3. Levitt P
    4. Narayanan S

    (2017) MUPET-Mouse ultrasonic profile extraction: A signal processing tool for rapid and unsupervised analysis of ultrasonic vocalizations

    Neuron 94:465–485.

    https://doi.org/10.1016/j.neuron.2017.04.005

    • PubMed
    • Google Scholar
89. 1. Van Veen BD
    2. Buckley KM

    (1988) Beamforming: a versatile approach to spatial filtering

    IEEE ASSP Magazine 5:4–24.

    https://doi.org/10.1109/53.665

    • Google Scholar
90. 1. Volodin IA
    2. Dymskaya MM
    3. Smorkatcheva AV
    4. Volodina EV

    (2022) Ultrasound from underground: cryptic communication in subterranean wild-living and captive northern mole voles (Ellobius talpinus)

    Bioacoustics 31:414–434.

    https://doi.org/10.1080/09524622.2021.1960191

    • Google Scholar
91. 1. von Merten S
    2. Hoier S
    3. Pfeifle C
    4. Tautz D
    5. Rosenfeld CS

    (2014) A role for ultrasonic vocalisation in social communication and divergence of natural populations of the house mouse (Mus musculus domesticus)

    PLOS ONE 9:e97244.

    https://doi.org/10.1371/journal.pone.0097244

    • Google Scholar
92. 1. Wang H
    2. Liang S
    3. Burgdorf J
    4. Wess J
    5. Yeomans J

    (2008) Ultrasonic vocalizations induced by sex and amphetamine in M2, M4, M5 muscarinic and D2 dopamine receptor knockout mice

    PLOS ONE 3:e1893.

    https://doi.org/10.1371/journal.pone.0001893

    • PubMed
    • Google Scholar
93. 1. Warren MR
    2. Sangiamo DT
    3. Neunuebel JP

    (2018a) High channel count microphone array accurately and precisely localizes ultrasonic signals from freely-moving mice

    Journal of Neuroscience Methods 297:44–60.

    https://doi.org/10.1016/j.jneumeth.2017.12.013

    • PubMed
    • Google Scholar
94. 1. Warren MR
    2. Spurrier MS
    3. Roth ED
    4. Neunuebel JP

    (2018b) Sex differences in vocal communication of freely interacting adult mice depend upon behavioral context

    PLOS ONE 13:e0204527.

    https://doi.org/10.1371/journal.pone.0204527

    • PubMed
    • Google Scholar
95. 1. Warren MR
    2. Clein RS
    3. Spurrier MS
    4. Roth ED
    5. Neunuebel JP

    (2020) Ultrashort-range, high-frequency communication by female mice shapes social interactions

    Scientific Reports 10:2637.

    https://doi.org/10.1038/s41598-020-59418-0

    • PubMed
    • Google Scholar
96. 1. Warren MR
    2. Spurrier MS
    3. Sangiamo DT
    4. Clein RS
    5. Neunuebel JP

    (2021) Correction: Mouse vocal emission and acoustic complexity do not scale linearly with the size of a social group

    The Journal of Experimental Biology 224:jeb243045.

    https://doi.org/10.1242/jeb.243045

    • PubMed
    • Google Scholar
97. 1. Yang X
    2. Guo D
    3. Li K
    4. Shi L

    (2021) Altered postnatal developmental patterns of ultrasonic vocalizations in Dock4 knockout mice

    Behavioural Brain Research 406:113232.

    https://doi.org/10.1016/j.bbr.2021.113232

    • PubMed
    • Google Scholar
98. 1. Zala SM
    2. Reitschmidt D
    3. Noll A
    4. Balazs P
    5. Penn DJ

    (2017a) Sex-dependent modulation of ultrasonic vocalizations in house mice (Mus musculus)

    PLOS ONE 12:e0188647.

    https://doi.org/10.1371/journal.pone.0188647

    • PubMed
    • Google Scholar
99. 1. Zala SM
    2. Reitschmidt D
    3. Noll A
    4. Balazs P
    5. Penn DJ
    6. Cooper BG

    (2017b) Automatic mouse ultrasound detector (A-MUD): A new tool for processing rodent vocalizations

    PLOS ONE 12:e0181200.

    https://doi.org/10.1371/journal.pone.0181200

    • Google Scholar
100. 1. Zaytseva AS
     2. Volodin IA
     3. Ilchenko OG
     4. Volodina EV

     (2019) Ultrasonic vocalization of pup and adult fat-tailed gerbils (Pachyuromys duprasi)

     PLOS ONE 14:e0219749.

     https://doi.org/10.1371/journal.pone.0219749

     • PubMed
     • Google Scholar
101. 1. Zhang C
     2. Florencio D
     3. Ba DE

     (2008) Maximum likelihood sound source localization and beamforming for directional microphone arrays in distributed meetings

     IEEE Transactions on Multimedia 10:538–548.

     https://doi.org/10.1109/TMM.2008.917406

     • Google Scholar

Author details

1. Max L Sterling

   1. Computational Neuroscience Lab, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands
   2. Visual Neuroscience Lab, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands
   3. Department of Human Genetics, Radboudumc, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands

   Contribution

   Software, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-2114-2265

2. Ruben Teunisse

   Computational Neuroscience Lab, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands

   Contribution

   Formal analysis, Writing - review and editing

   Competing interests

   No competing interests declared

3. Bernhard Englitz

   Computational Neuroscience Lab, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen, Nijmegen, Netherlands

   Contribution

   Conceptualization, Data curation, Software, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft, Project administration, Writing - review and editing

   For correspondence

   englitz@science.ru.nl

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-9106-0356

• 1,409

  views

• 210

  downloads

• 10

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.