Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Mar 3.
Published in final edited form as: Curr Biol. 2026 Feb 18;36(5):1291–1299.e4. doi: 10.1016/j.cub.2026.01.042

Sequencing of distinct wing behaviors during Drosophila courtship

Xinping Li 1, Kyle Thieringer 1,2, Yiqin Gao 1,3, Mala Murthy 1,4
PMCID: PMC12952713  NIHMSID: NIHMS2149983  PMID: 41713408

Summary

Some behaviors, like biting followed by chewing and then swallowing, unfold in stereotyped sequences, while others, such as limb movements during defensive maneuvers, can be flexibly combined as needed. During courtship, male Drosophilid flies produce a series of actions, including orientation, tapping, singing, licking, and copulation, that follow an ordered but temporally variable sequence1,2. At shorter timescales, however, individual actions remain highly dynamic. For example, courtship songs are composed of variable sequences of distinct syllables, with their patterning and amplitude actively shaped by female cues3–5. Leveraging recent advances in behavioral quantification6, we characterize a distinct courtship wing behavior that we term “waggling”, defined by specific kinematic features. This behavior is present across multiple Drosophila species and characterized by rhythmic, anti-phase wing movements. We identify an intermediate level of stereotyped behavioral structure: a directional three-part motif where males and females first decelerate to near-complete stillness, followed by male-initiated waggling, which then transitions into courtship song. Wing kinematics during waggle bouts are predictive of wing choice in subsequent songs, suggesting waggling may serve as a preparatory behavior. We then focus on P1/pC1 neurons, known to promote courtship5,7–11. Optogenetic activation of specific P1/pC1 neuron subsets in solitary males, without any female cues, is sufficient to recapitulate the entire stillness-to-waggling-to-singing progression. These findings reveal a new layer of stereotyped structure within a flexible courtship display and demonstrate that P1/pC1 neurons can orchestrate multi-action behavioral programs through internal dynamics.

eTOC Blurb:

Li et al. characterize “waggling,” a rhythmic, anti-phase wing behavior in courting Drosophila males that forms part of a stereotyped three-step sequence: stillness, waggling, then singing. Optogenetic activation of P1/pC1 neuronal subsets recapitulates the entire progression in solitary males, revealing that internal dynamics can generate the multi-action behavioral sequence.

Results

Drosophila males produce two distinct wing behaviors during courtship - singing and waggling

We used quantitative pose estimation (via SLEAP6) of a large dataset of Drosophila melanogaster male-female interactions (32 pairs, 13.4 hours of videos), along with audio recordings, to identify a previously uncharacterized non-acoustic wing behavior during fly courtship, which we term wing “waggling”. During waggling, D. melanogaster males extend both wings (at reduced angles relative to song12), in contrast to the characteristic unilateral wing extension of singing (Figure 1A–B). The defining feature of waggling is the rapid, anti-phase oscillation of the wings at a median frequency of approximately 10.6 Hz (Figure 1C), resembling the motion of windshield wipers (Videos S1). This frequency is intermediate between the recently described low-frequency thoracic vibrations13,14 and the higher-frequency wing vibrations that generate courtship song12,15,16. While previous studies since Sturtevant (1915) have described a bilateral wing movement termed ‘scissoring’15,17–20 that may overlap with waggling, scissoring has never been quantitatively defined. The term has been applied inconsistently to a variety of wing movements that include ‘slowly scissored open and shut’, ‘opened and closed a few times’, and ‘rapidly and repeatedly crossed and re-crossed’. And, a more recent study explicitly defined ‘scissoring’ in D. nepalensis and D. trilutea as bilateral wing spreading21, which is distinct from the kinematic signature of waggling. To avoid ambiguity, we use the distinct term “waggling” and define it as a rapid, anti-phase oscillatory wing movement.

Figure 1. Drosophila melanogaster males produce two distinct wing behaviors during courtship.

Figure 1.

Open in a new tab

A. An example waggling bout (top) and courtship song bout (bottom). Snapshots (with SLEAP6 skeletons overlaid) indicate discrete moments in each behavior during male/female interactions. Male wing extension angles are different during waggling versus singing (the dominant wing is defined as the wing with greater mean amplitude throughout the bout); waggling involves an oscillatory behavior between the two wings. Waggling does not produce a detectable sound on the microphones (see Methods).

B. Wing extension angles. Polar histograms plot the distribution of maximum wing extension angles for waggling (green, 1 857 bouts) and singing (purple, 5087 bouts), the radial axis shows probability density, and dots outside each plot mark the per-fly medians of these peak angles.

C. Frequency-domain analysis. Left two panels (green): Mean power spectra (± SEM) of dominant and non-dominant wings during waggling (left) and mean absolute dominant/non-dominant wing phase difference (right); insets show the distribution of each individual fly’s median values. Right-most panel (purple): Mean power spectra (± SEM) of dominant and non-dominant wings during singing.

D. Gantt-style timelines plot all waggling (green) and singing (purple) bouts detected in three representative recordings; each horizontal line is one fly pair, and colored bars mark the start-to-end time of individual waggle or song bouts (see Methods). Waggling and singing both occur throughout courtship.

E. Waggle and song bout durations are comparable (paired t-test, p=0.901; ns, not significant).

F. Males spend a larger fraction of the recording singing versus waggling (Wilcoxon signed-rank test, *** p < 0.001).

G. RMS (Root Mean Square) microphone amplitude is highest during song bouts, intermediate at baseline, and lowest during waggling bouts (Friedman test with Wilcoxon post-hoc, *** p < 0.001).

H. Spatial geometry of the courting male relative to the female (see Methods).

I. Female-centered map of male location during waggling (green) or singing (purple). The orange triangle marks the female thorax and her orientation. Males stay close in both behaviours but, whereas singing occurs mostly directly behind the female, waggling is produced all around the female.

J. Female-centered quiver plot of male orientation during waggling (green) or singing (purple). Males orient towards the female during both behaviors.

K-M. Distributions of male–female spatial parameters (see H). Probability-density functions show (K) distance, (L) body angle, and (M) target angle during waggling (green), singing (purple) and control (cyan; frames that precede the first waggle or song frame in each recording).

N=32 males across panels B-C and E-M. For E-G, box plots show median ± IQR, whiskers extend to 1.5×IQR. Dots mark the individual per-fly medians.

See also Figure S1 and Videos S1–S2.

Similar to singing, waggling is never observed in solitary males and occurs in bouts throughout male-female interactions. While waggling occurs less frequently than singing under our recording conditions, it represents a significant component of male courtship behavior (Figure 1D–F). However, waggling does not generate an audible signal (Figure 1G). These properties are conserved across species. Using similar methods for behavioral quantification, we observed waggling in both D. simulans and D. sechellia (Videos S2, Figure S1); both species are separated from the D. melanogaster lineage only about 3 million years ago22. While D. sechellia exhibited greater wing extension angles and lower oscillation frequencies compared to D. melanogaster and D. simulans, the presence of anti-phase oscillatory wing movements across species highlights that waggling is likely an important courtship behavior.

While we do not know if waggling serves a communicative purpose, it only occurs when males are close to females (within 3–4 mm) and are oriented toward them (target angle ~0°), suggesting it may be driven by cues from the female. One key difference with singing is that while song is produced mostly behind the female (often while chasing her3), waggling is produced all around the female (Figure 1H–M). Below we investigate the coordination between singing and waggling.

Waggling is part of a behavioral sequence during courtship

Song production is linked to locomotor dynamics - male and female speed can predict both the timing and patterning of song3. Is the same true for waggling? We found that waggling and singing showed distinct, and often opposite, associated locomotor dynamics (Figure 2A–B). The onset of a waggling bout is preceded by marked deceleration of both males and females, who both remain stationary prior to and throughout the behavior (Figure 2C–E). This contrasts with singing, which is typically marked by acceleration prior to a song bout, as well as continued motion of both male and female throughout the song bout. These findings further establish waggling as a distinct behavioral state with specific locomotor requirements. It is worth noting that the stillness phase (when both males and females remain stationary) can potentially overlap with abdominal vibrations, which are used as a communication signal13,14, but not measured here.

Figure 2. Locomotor dynamics differ around waggle versus song bouts.

Figure 2.

Open in a new tab

A. Parameters analyzed: the distance between the male and female thorax, the speed of the male or female, along with their heading directions.

B. Bout-triggered averages (mean ± SEM) of male–female distance (left), female speed (middle), and male speed (right) aligned to bout start and end. Green: waggle bouts; purple: song bouts. Waggle occurs at larger male-female distances and slower speeds relative to song.

C. Male and female trajectories during individual waggle (male: navy; female: orange) and song bouts (gray). All trajectories are centered on the bout start position.

D. Z-scored speeds around waggle bouts. Bout-triggered averages after z-scoring speeds within individual animal recordings (same data as B but z-scored).

E. Scatter plot of male versus female speeds during pre-waggling (brown, 1 s before bouts), waggling (green, entire bout averages), and all other times (gray, 1 s bins). Boxed region marks the “stillness” zone (0 – 0.67 mm/s male, 0 – 0.3 mm/s female; defined in Figure S2).

N=32 male-female pairs in B-E.

See also Figure S2.

In addition, we observed that waggling often precedes song but not the other way around (Figure S2B, also see Figure 3G–H). We therefore examined the relationship between waggle bouts and the song bouts that follow. We found that waggling bouts often transition directly into song bouts, a sequence we term a “linked” waggle (Figure 3A–B). These linked transitions have distinct dynamic signatures compared to “unlinked” waggles that are not immediately followed by song. Specifically, these linked bouts are associated with closer male-female proximity, lower female speed both before and after waggle offset, and higher male speed immediately following the waggle (Figure 3C).

Figure 3. Sequencing of stillness, waggling, and singing.

Figure 3.

Open in a new tab

A. A waggle bout can either be directly followed by song (top: “linked” waggle) or not (bottom: “unlinked” waggle). In either case, males can use the same wing as dominant during waggling and subsequent singing (shown at top) or they can switch the dominant wing between the two behaviors (shown at bottom).

B. Distribution of gap durations between the end of waggle bouts and the start of the subsequent song bout. The red dashed line marks the cutoff (0.15 s) separating linked and unlinked transitions (see Methods).

C. Locomotor dynamics around waggle bout ends for linked (solid lines) and unlinked (dashed lines) transitions to song. Bout-triggered averages (mean ± SEM) of male–female distance (left) and speeds (female: middle; male: right) aligned to waggle bout ends. Linked waggles involve a steep drop in male-female distance as males transition from waggling to singing, as well as a sharp increase in male and female speeds.

D. Wing choice in song bouts following waggling. Left panel: Fraction of song bouts where the dominant wing matches the dominant wing of the preceding waggle bout for unlinked and linked transitions. Dots mark individual animals (paired t-test, **p < 0.01). Right panel: Same data binned by waggle–song gap duration (log-scale bins: 0.007–0.046 s, n=947; 0.046–0.312 s, n=145; 0.312–2.138 s, n=290; 2.138–14.630 s, n=393; 14.630–100.133 s, n=78). Longer gap durations significantly reduce the predictability of wing choice from the dominant waggle side (Spearman ρ = −0.14, p < 0.001).

E. Receiver Operating Characteristic (ROC) curves comparing the power of female location at the start of the song (see Figure S2E) to predict the male’s dominant wing choice. Female location is a worse predictor for wing choice during linked song bouts (solid line, AUC = 0.748) than during unlinked song bouts (dashed line, AUC = 0.840).

F. ROC curves comparing the predictive power of two classifiers for wing choice during linked song bouts only. The dominant wing of the preceding waggle bout (light purple, “waggle wc”) is a better predictor (AUC = 0.888) than the female’s location (pink, “female location”) (AUC = 0.748).

G-H. Bout-to-bout behavioral state transitions. Network diagram (G) and transition probability matrix (H) showing probabilities of transitioning between three different kinds of ‘bouts’: stillness, waggling, and singing. Only three states are shown for clarity (see Figure S2D for full four-state matrix that includes “other” for all other behaviors produced during courtship); probabilities do not sum to 1 as transitions to the “other” state are excluded. Solid arrows and asterisks indicate statistically significant transitions (one-tailed permutation test, ***p < 0.001). See Methods for stillness, waggling and singing bout definitions.

I. Three-gram transition probabilities. Bars show probability of each three-bout sequence, with counts above. Word cloud shows the most frequent three-gram sequences, with font size scaled to frequency.

J. Representative sequence from stillness to waggling and to singing. Wing angles (dominant in black and non-dominant in gray), audio recordings, and male (blue) and female (orange) speeds during a single recording (see Video S3).

K. Conceptual models of behavioral sequencing during courtship. Top: Self-transitory model: male’s internal dynamics drive behavioral transitions, with female sensory cues modulating the probability of staying in or exiting each state. Bottom: Cue-dependent model: female sensory cues directly drive transitions from one behavior to the next.

N=32 male-female pairs in B-I.

See also Figure S2 and Video S3.

Furthermore, wing choice for the subsequent song bout is strongly correlated with the dominant wing used during the preceding waggle, especially in linked transitions (Figure 3D). The shorter the time gap between waggling and singing, the stronger the correlation. Remarkably, this coupling effect persists for over 14 seconds before eventually decaying back to chance levels (Figure 3D). Since male wing choice during song is known to be strongly biased by female position23,24,this coupling could, in principle, be an artifact of this shared external cue. However, we found that female location (Figure S2E) is a worse predictor for wing choice during linked song bouts (those immediately following a waggle) than during unlinked song bouts (Figure 3E). When directly comparing predictors for these linked song bouts, the dominant wing used during the preceding waggle is a significantly better classifier of subsequent song bout wing choice than the female’s location (Figure 3F). This suggests that the waggle-song wing choice coupling is not merely a correlation and likely reflects an internal functional linkage.

We next analyzed behavioral transitions between stillness (periods when both males and females remain stationary), waggling, and singing, which revealed a clear, directional structure (Figure 3G–H, Figure S2D). The transition probability from waggling to singing is exceptionally high, and the most frequent three-state sequence was stillness → waggling → singing (Figure 3I–J, Video S3). This progression suggests a stereotyped behavioral sequence within courtship interactions.

Two mechanistic models, both supported by existing literature3,25, could explain this behavioral sequence (Figure 3K): a “self-transitory” model where male internal dynamics drive the sequence progression with female cues possibly gating the transitions, or a “cue-dependent” model where female sensory inputs directly drive each behavior in the sequence. Distinguishing these requires examining behavior in the absence of female cues.

P1 neuron subset orchestrates the behavioral sequence by modulating locomotion

We next used optogenetics to uncover the cell types that promote the waggling and the courtship sequence described above (Figure 4A–B). P1/pC1 neurons are a sexually dimorphic group of roughly 60–78 neurons per hemisphere that are capable of driving a persistent arousal state5,7,26,27 and that promote a variety of social behaviors, including both courtship and aggressive actions8,10,28–30. For simplicity, we will refer to P1/pC1 as P1 throughout the text. Within this population, activation of a male-specific subset of 6–8 P1 neurons per hemisphere (called P1a) can drive acute locomotor arrest12,13,26. In addition, P1a activation, in solitary male flies, does not drive courtship singing directly, but rather a persistent state of elevated song production5. Because of this, we hypothesized that P1 neurons might orchestrate the behavioral program of stillness leading to waggling and then singing.

Figure 4. P1 neuronal subsets drive behavioral sequences that include waggling.

Figure 4.

Open in a new tab

A. Experimental design. Optogenetic activation of P1 neuronal subsets to investigate the production of slowing, waggling, and singing states in the absence of female cues.

B. Optogenetic stimulation protocol (see Methods). A one second light pulse (of varying amplitude) delivered every 30 seconds across the stimulation block.

C. Behavioral responses across 11 different genotypes targeting different P1 neuronal subsets. Left: Anatomical expression patterns. Right: Mean ± SEM walking speed (blue) aligned to LED light onset (red bar), with waggling (green) and singing (purple) probabilities. Each trace represents the average response across all 5 stimulation levels. Genotypes with red labels exhibit robust activation of the full behavioral repertoire (slowing, waggling, and singing).

D. Summary of optogenetically-induced speed responses across genotypes. Violin plots show distribution across animals for baseline speed (top; prior to the start of optogenetic stimulation block), block speed (middle; speed during the entire optogenetic block), and percentage change (bottom; calculated as (during-pre)/pre, comparing speed during optogenetic pulse (when red LED is on) to pre-pulse speed).

E. Latency from optogenetic onset to behavioral responses in genotypes that drive the full behavioral repertoire. Violin plots show distribution across trials for peak deceleration (blue), waggling onset (green), and singing onset (purple). Friedman test within each genotype (***p < 0.001).

F. Bout-to-bout transition probability matrix for behaviors occurring after optogenetic onset in the effective driver lines (pooled data from r71_dsx, split1, split2, split3, split6).

G. Network of genetic driver combinations for P1 neurons and functional outcomes. Connections indicate genetic driver combinations; those producing full behavioral repertoire are indicated in red, others in gray. See Table S1 for full information on genetic driver lines.

N = 17 (dsx_fru), 15 (np_dsx), 14 (np_fru), 19 (r71_dsx), 15 (r71_fru), 19 (split1/split P1a), 18 (split2), 19 (split3), 19 (split4), 19 (split5), 18 (split6) males. Violin plots show median and quartiles (black lines, in D and E) and means (white circles, in E).

See also Table S1 and Figures S2–S4.

P1 neurons contain both Doublesex and Fruitless expressing subsets30–33, comprising multiple genetically and functionally heterogeneous subtypes8,33–37. While most studies of P1 have so far focused on P1a, we instead selected and tested multiple genetic lines known to label broader groupings of P1 neurons26,35,36 (Figure 4C, Figure S3A, Table S1). These lines each target a different subset of P1 neurons. By activating these P1 sub-populations in solitary males (without the presence of female sensory cues), we could test whether internal neural dynamics alone can drive the behavioral progression from stillness to waggling to singing (Figure 4A). We chose an optogenetic paradigm in which we interleaved brief (1 second) stimuli with long intervals (Figure 4B), in order to uncover behavioral sequences that followed stimulation - this is in contrast with prior work that used higher duty cycles for P1 neuron activation5,12.

Screening of 11 genetically defined P1 subsets revealed two distinct response classes based on acute locomotor effects (Figure 4C–D). Activation of some P1 subsets (split2, split 6, split1 (also known as P1a), split3, r71_dsx, np_dsx) showed acute deceleration during optogenetic stimulation but elevated overall locomotor speed, while the other group (dsx_fru, r71_fru, split4, np_fru, split5) showed either acceleration or no change during acute stimulation but exhibited suppressed overall locomotor speed. In addition, the acute locomotor effects, regardless of direction, were dose-dependent and influenced by the initial locomotor state of the fly at stimulation onset (Figure S3B). Only drivers inducing acute slowing (r71_dsx, split1/P1a, split2, split3, split6) reliably triggered both waggling and singing (Figure 4C, Figure S4A). The kinematics of this optogenetically-evoked waggling, including its anti-phase wing movements, closely matched those of the natural behavior (Figure S4C). The latency from optogenetic onset to peak slowing, then waggling, then singing followed a consistent temporal order that mirrored the natural progression (Figure 4E, Figure S4B). To confirm that P1 activation recapitulates the entire behavioral sequence observed in natural courtship, and not simply a probabilistic increase in independent behaviors, we performed the same state-transition analysis used for the wild-type data. There was a clear, directional progression from waggling to singing (Figure 4F), and the most frequent three-state sequence was stillness → waggling → singing (Figure S4D).

In addition, locomotor state modulated sequence probability—flies that slowed down sufficiently were more likely to complete the sequence, suggesting locomotion itself acts as an internal context that shapes behavioral state transitions (Figure S4E). For instance, np_dsx P1 activation drove very little waggling and singing following optogenetic activation - males of this genotype had a higher overall speed, and so the slowing induced by optogenetic activation did not cause enough of a reduction in speed to reliably lead to the subsequent behaviors (Figure 4C).

Taken together, our results strongly support the self-transitory model (Figure 3K): external sensory input is not necessary to transition from stillness to waggling to singing; instead P1 neurons (and specifically the R71G01 and dsx subsets (Figure 4G) drive slowing (until the fly reaches stationary), which then enables the production of waggling followed by singing. While roles for P1 neurons, in solitary males, in both acute modulation of locomotion and persistent singing have been reported previously5,12, through quantification of wing movements, we now additionally reveal that activation of particular subsets of P1 promotes a reliable behavioral sequence that includes a new behavior, waggling.

Discussion

Drosophila courtship has been characterized as a probabilistic sequence of distinct actions that unfold over minutes2, with individual actions like courtship song being dynamically shaped by sensory cues on hundred-millisecond to second timescales3,4. Our findings reveal an intermediate level of organization: the stillness → waggling → singing sequence represents a stereotyped behavioral ‘chunk’, that is deployed within the otherwise variable flow of the courtship sequence. This suggests a hierarchical system where the nervous system can call upon structured sub-modules as needed, rather than generating every action de novo in response to external cues.

While we do not yet know if waggling serves a communicative function, our finding that the dominant wing during waggling reliably predicts which wing the male will extend for subsequent singing (Figure 3D), suggests waggling acts as motor priming, a “warm-up” that facilitates wing selection and extension required for effective courtship singing. An alternative explanation for this correlation is that wing choice for both behaviors is independently driven by the same external cue, as female position and motion are known to predict song wing choice23,24. However, the sensory context for waggling is distinct: it occurs when the female is stationary (Figure 2E) and positioned directly in front of the male (Figure S2E), a posture where there is little difference between visual input to the left and right eyes.

Consistent with this, female location (the bias in her head position relative to the male’s center line) is a worse predictor of the dominant wing during waggling than for wing choice during song (Figure S2F). In this context, the dominant wing during preceding waggle bout is a better predictor of song wing choice than female location (Figure 3F). And, the strong waggle-song wing choice correlation persists in both blind (both eyes painted) males as well as in solitary males where the sequence was induced optogenetically (Figure S2G). The persistence of this motor coupling in the absence of female sensory input supports that waggling serves a motor preparatory role.

The mechanism underlying the P1-driven slowing → waggling → singing module contrasts with another recently described Drosophila courtship sequence. McKellar et al.25 identified an “engagement motif” (proboscis extension, abdominal bending, and foreleg lifting) controlled by the aSP22 descending neuron pair, where increasing spike counts trigger cumulative, overlapping actions. Our sequence operates differently: actions are transitional rather than cumulative, and control is orchestrated by P1 in the central brain, which likely acts through distinct descending neuron pathways37. While we haven’t identified the specific neurons downstream of P1 that control waggling, activation of known song-production neurons (pIP10, pMP2, dPR1, TN1)9,38–40 never elicited waggling (data not shown), indicating P1 coordinates distinct parallel pathways for waggling and singing. These can now be analyzed with the newly available male CNS connectome41.

Our results contribute to the growing body of evidence for functional heterogeneity within the P1 neuron population. Two very recent studies use either single-cell RNA sequencing (Allen et al. bioRxiv 2025) or connectomics (Rubin et al. bioRxiv 2025) to map the heterogeneity within the P1 population33,37. Rubin et al. (2025), in particular, define 48 distinct P1/pC1x cell types based on morphology and connectivity in the newly completed male brain connectome41, with many types comprising only 1–3 neurons per hemisphere. The 11 P1 genotypes we tested do not represent distinct cell types, therefore, but rather different, partially overlapping subpopulations within the P1 neural space, allowing us to dissect functional domains. While we do not know how these lines overlap precisely, previous studies have shown that some of these drivers label largely separate populations35,36. Importantly, the Rubin et al. study (bioRxiv 2025) defined the cell types labeled in the split1/P1a driver line: types 4a,b and 12a. When these types are activated, they drive persistent song production following the optogenetic stimulus, similar to our results for R71G01-containing lines (splits 1, 2, 3, and r71_dsx). For this subset, our results further point to a clear split in the dsx-expressing and fru-expressing P1 subsets: dsx+ P1 neurons drive both waggling and singing, while fru+ P1 neurons do not. This molecular dissociation aligns with previous studies demonstrating functional specializations for dsx+ versus fru+ P1 neurons34,35. Since our optogenetic paradigm likely activates multiple subtypes simultaneously, the absence of the full behavioral sequence in some lines could result from co-activation of neurons with opposing functions.

Finally, P1 neurons appear to function as a context-dependent switch, orchestrating distinct motor programs based on the presence or absence of female motion cues. When females are moving, R71G01/P1 activation increases the gain of visual processing pathways to enable chasing and pursuit42. In contrast, our work shows that in the absence of such female motion cues, R71G01/P1 activation initiates an entirely different sequence: acute locomotor slowing, followed by waggling, and then a transition to song. This suggests P1 neurons translate the sensory context into adaptive actions, engaging a “chasing” program when the female is moving and a “slowing → waggling → singing” program when she is still. In both of these contexts, P1 neurons likely are ‘aware’ of the female’s presence through detection of pheromonal cues34. Once singing begins (following waggling), locomotor speed increases regardless of whether the sequence is naturally occurring (Figure 2B) or optogenetically induced (Figure 4C), readying the male for the next phase of the dynamic courtship interaction.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to the lead contact, Mala Murthy (mmurthy@princeton.edu).

Materials availability

This study did not generate new transgenic reagents. Fly strains and lines used in this study are listed in the key resources table.

Key resources table

REAGENT or RESOURCE SOURCE IDENTIFIER
Antibodies
Chicken anti-GFP Immunology Consultants Laboratory Cat#CGFP-45A-Z
Mouse mAb anti-Bruchpilot (nc82) Developmental Studies Hybridoma Bank Cat#nc82; RRID: AB_2314866
Alexa Fluor 488-goat anti-chicken SouthernBiotech Cat#6100-30; RRID: AB_2796165
Alexa Fluor 633-goat anti-mouse Thermo Fisher Scientific Cat#A-21050; RRID: AB_2535718
Experimental models: Organisms/strains
D. melanogaster: wild type NM91 Coen et al.3; obtained from Peter Andolfatto N/A
D. sechellia: strain sec07 Gift from Richard Benton DSSC: 14021-0248.07
D. sechellia: strain sec28 DSSC: 14021-0248.28
D. simulans: strain sim04 Gift from Richard Benton DSSC: 14021-0251.04
D. simulans: strain sim196 DSSC: 14021-0251.196
D. melanogaster: w[1118] P{y[+t7.7] w[+mC]=20XUAS-IVS-CsChrimson.mVenus}attP18;; Bloomington Drosophila Stock Center RRID: BDSC_55134
D. melanogaster: w[1118]; P{y[+t7.7] w[+mC]=R15A01-p65.AD}attP40; Bloomington Drosophila Stock Center RRID: BDSC_68837
D. melanogaster: w[1118];; P{y[+t7.7] w[+mC]=R71G01-GAL4.DBD}attP2 Bloomington Drosophila Stock Center RRID: BDSC_69507
D. melanogaster: w[1118]; P{y[+t7.7] w[+mC]=R17D06-p65.AD}attP40; Bloomington Drosophila Stock Center RRID: BDSC_68843
D. melanogaster: w[1118];;P{y[+t7.7] w[+mC]=R17D06-GAL4.DBD}attP2 Bloomington Drosophila Stock Center RRID: BDSC_68951
D. melanogaster: w[1118]; P{y[+t7.7] w[+mC]=R22D03-p65.AD}attP40; Bloomington Drosophila Stock Center RRID: BDSC_69869
D. melanogaster: w[1118];; P{y[+t7.7] w[+mC]=R22D03-GAL4.DBD}attP2 Bloomington Drosophila Stock Center RRID: BDSC_69827
D. melanogaster: w[1118];;P{y[+t7.7] w[+mC]=GMR71G01-GAL4}attP2 Bloomington Drosophila Stock Center RRID: BDSC_39599
D. melanogaster: ;;NP2631 Yu et al.39 and Ishii et al.35; Gift from Kenta Asahina N/A
D. melanogaster: ;20xUAS>stop>CsChrimson.mVenus; Gift from Vivek Jayaraman N/A
D. melanogaster: w+;;dsxGal4 Gift from Stephen Goodwin N/A
D. melanogaster: ;;fruFLP Gift from Barry Dickson N/A
D. melanogaster: ;;dsx-LexA, 8xLexAop2-FLP Deutsch & Clemens et al.43 N/A
Software and algorithms
Matlab R2024a https://www.mathworks.com/ RRID: SCR_001622
Inkscape 1.3.2 https://inkscape.org/ RRID: SCR_014479
Fiji https://fiji.sc/ RRID: SCR_002285
SLEAP https://sleap.ai/ RRID: SCR_021382
Open in a new tab

Data and code availability

Data reported in this paper will be shared by the lead contact upon request. Code for detecting and annotating waggle bouts is on Github at github.com/murthylab and will be publicly available upon publication. Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

STAR Methods

Experimental model and study participant details

Male Drosophila melanogaster, D. simulans, and D. sechellia flies were reared at 25°C on 12:12 light:dark cycles. D. melanogaster and D. simulans were maintained on standard cornmeal medium; D. sechellia on Formula 4–24 medium (Carolina) mixed with noni juice. Virgin males were collected within 8 hours of eclosion, aged 3–7 days, and single-housed except for optogenetic experiments (2–8 flies per vial). For courtship assays, individual virgin males were paired with age-matched virgin females that were also single-housed prior to experiments. For experiments with blind male flies, wild-type males were briefly anesthetized on ice. A small dab of black paint was applied to both compound eyes to occlude vision. These males were allowed to recover for one day before behavioral recording. For optogenetic experiments, males were transferred to ATR-supplemented food (1mM) within 8 hours of eclosion. Genotype details are in the Key Resource Table and Table S1.

Methods details

Courtship behavioral assays

Courtship interactions were recorded in a custom-fabricated behavioral chamber consisting of a 30 mm × 30 mm 3D-printed base (Formlabs Form 2, Black resin) and a clear vacuum-molded PETG dome. The arena was side-illuminated with 850 nm infrared LEDs, and videos were captured from above at 150 frames per second with 1,024 × 1,024 pixel resolution (30.3 pixels/mm) using a FLIR Blackfly S camera equipped with an 850 nm longpass filter. Acoustic signals were simultaneously recorded through microphones embedded in the chamber floor. Video files were compressed in real-time using GPU-accelerated H264 encoding to produce nearly lossless videos with independently seekable frames. This behavioral apparatus and recording protocol are described in detail in Pereira et al. (2022)6. Part of the wild-type D. melanogaster courtship dataset used in this study was previously published in Pereira et al. (2022)6. All D. simulans and D. sechellia data were newly collected for this manuscript.

Virgin males and females (3–7 days post-eclosion) were single-housed prior to experiments. All recordings began within 2 hours of incubator lights turning on to capture peak courtship activity. Individual male-female pairs were gently aspirated into the chamber and recorded for up to 30 minutes or until copulation occurred. For D. sechellia experiments, paper soaked with noni juice was placed under the chamber floor to provide species-appropriate olfactory cues. Only recordings containing courtship interactions were included in analyses; pairs showing no courtship behavior were excluded.

Optogenetics

To activate specific P1 neuron subsets, we used eleven genotypes expressing CsChrimson under different genetic drivers (see Table S1 for complete genotype list and Figure S3A for anatomical expression patterns). Males were collected within 8 hours of eclosion and transferred to food supplemented with 1 mM all-trans-retinal (ATR) for at least three days prior to experiments. To prevent ATR degradation while maintaining circadian rhythms, vials were housed in custom-made blue acrylic light-filtering boxes.

Optogenetic experiments were performed using the same behavioral apparatus as courtship assays. Solitary males were individually placed in recording chambers and allowed to acclimate for 5 minutes before recording onset. Red light stimulation (627 nm) was delivered through programmable LEDs (Luxeon) mounted around the chamber.

The stimulation protocol consisted of 30-second trial blocks, each containing a 1-second light pulse followed by a 29-second inter-stimulus interval (Figure 4B). Five LED intensities (1, 5, 25, 125, and 205 μW/mm2) were presented in randomized order, with each intensity repeated 10 times, resulting in a total experimental duration of 25 minutes per fly. Light intensity was calibrated using a power meter positioned at the chamber floor. Video and audio acquisition followed the same procedures as for courtship assays.

Anatomy

To visualize anatomical expression patterns of the eleven P1 neuron subsets, we performed whole-mount immunohistochemistry on adult male brains. Brains were dissected and stained following standard protocols (detailed protocols available at https://www.janelia.org/projectteam/flylight/protocols). Primary antibodies included mouse anti-Bruchpilot (nc82, 1:30) for neuropil counterstaining and chicken anti-GFP (1:1000) to detect mVenus expression. Secondary antibodies were goat anti-chicken Alexa Fluor 488 (1:250) and goat anti-mouse Alexa Fluor 633 (1:250). All antibodies are listed in the Key Resources Table.

Serial optical sections were obtained at 1 μm intervals on a Zeiss 700 confocal microscope with a Plan-Apochromat 20×/0.8 NA objective. Confocal stacks were processed into maximum intensity projections using Fiji. For Figure S3A, original projections were paired with masked versions where non-specific expression outside the central brain was digitally removed to highlight P1 neuron populations. Figure 4C shows the masked anatomical patterns for clarity.

Behavioral data analysis

Pose tracking

All videos were tracked using SLEAP6, and then manually proofread. For each fly, we tracked four body parts: head, thorax, left wing tip, and right wing tip. The thorax position served as the primary reference point for fly location throughout the recording. Tracking data were exported at the original 150 fps temporal resolution for subsequent analyses. Tracked coordinates were first processed to reduce noise and tracking errors. Wing tip positions were smoothed by interpolating missing values using nearest-neighbor interpolation. Thorax coordinates underwent additional smoothing with a 3-frame moving average filter followed by a Savitzky-Golay filter.

Audio processing

Acoustic signals were recorded through 9 microphones embedded in the chamber floor and digitized at 10 kHz for courtship assays and 5 kHz for optogenetic experiments. Audio channels were first baseline-subtracted then passed through a band-pass filter (50–1000 Hz). The three channels with highest amplitude at each time point were selected and averaged to produce a single audio trace. Audio and video streams were synchronized post-hoc using camera shutter signals recorded.

Feature extraction

The body axis for each fly was defined as the vector from thorax to head. Wing angles were calculated as the angle between each wing (thorax to wing tip vector) and the body axis, with 0° representing wings folded against the body and 90° representing full extension. The dominant wing was defined as the wing with greater mean amplitude throughout each bout.

Locomotor metrics were derived from thorax positions. Speed was calculated as the displacement along the heading direction between consecutive frames. All speed traces shown in figures were smoothed with a median filter. For z-scored speeds shown in figures, z-scoring was performed on the entire recording for each animal.

To quantify male-female interactions, we computed three key spatial relationships (Figure 1H): (1) the Euclidean distance between male and female thoraxes, (2) the body angle between the flies’ body axes, and (3) the target angle between the male’s body axis and the vector pointing from male to female thorax. For spatial visualization, we generated female-centric plots. The location heatmap (Figure 1I) was created by converting male positions relative to the female into Cartesian coordinates based on thorax-to-thorax distance and the angle from the female’s body axis. These positions were binned into a hexagonal grid (0.10 mm edge length). The orientation quiver plot (Figure 1J) displayed male body orientations at different positions around the female, with data grouped into 14 × 14 spatial bins (approximately 0.86 mm per bin). Each bin’s vector base represents the average male position, its direction shows the average relative body angle, and its length scales with the number of observations in that bin.

Additionally, for wing choice analyses (Figure 3E–F, Figure S2E–F), we quantified the female’s head position relative to the male’s head. This relationship was defined by two components: (1) the Euclidean distance between the male and female head nodes, and (2) the signed angle between the male’s body axis (thorax-to-head vector) and the vector pointing from the male’s head to the female’s head. For the spatial visualization in Figure S2E, we generated a male-centric heatmap by converting these relative female head positions into Cartesian coordinates. These positions were then binned into a hexagonal grid (0.10 mm edge length).

Waggle and song bout detection

Waggling and courtship song bouts were identified based on distinct kinematic and acoustic signatures. For courtship song bouts detection, we used a custom written script that identified periods of substantial wing extension accompanied by elevated acoustic amplitude. Bout boundaries were defined as the frames where wing angles exceeded and subsequently returned to baseline. All detected song bouts were manually proofread. For waggling detection, different approaches were used depending on the dataset. In courtship assay datasets (D. melanogaster, D. simulans, and D. sechellia), waggle bouts were manually annotated by identifying periods of rhythmic, anti-phase wing movements lacking acoustic signals. For optogenetic experiments, we used a custom written waggling detector. Wing angles were band-pass filtered (8–18 Hz), and the Hilbert transform was applied to extract instantaneous phase and amplitude. Waggling was identified when wings oscillated in anti-phase with sufficient amplitude above baseline. Waggle bout boundary was set from the onset to the offset of the continuous oscillatory wing movement. Wing oscillations must persist for at least 3 cycles to be counted as waggling. The automated detector achieved 89.5% precision and 83.6% recall when validated against manually annotated data. All automatically detected waggling bouts were subsequently manually proofread.

Stillness definition

Stillness was defined as periods when both male and female flies remained nearly stationary. To determine appropriate speed thresholds, we applied a knee-point algorithm to the cumulative distribution of speeds pooled across all recordings. This algorithm identifies the point of maximum curvature by calculating the perpendicular distance from each point on the normalized CDF to a line connecting the distribution endpoints. The resulting inflection points distinguished stationary from mobile behavior: 0.67 mm/s for males and 0.30 mm/s for females (Figure S2A). Periods were classified as stillness only when both flies simultaneously maintained speeds below their respective thresholds.

Temporal linking of waggle bouts

To examine coupling between waggling and singing, we measured the gap duration from each waggle bout offset to the subsequent song bout onset (Figure 3A). We classified transitions as “linked” or “unlinked” using k-means clustering (k=2) on log-transformed gap durations, with 0.15 seconds as the dividing threshold. Waggle bouts not followed by song within the recording period were excluded from this analysis.

ROC analysis

We used Receiver Operating Characteristic (ROC) analysis to quantify the performance of different variables in classifying binary behavioral choices (e.g., left vs. right dominant wing). For a given classifier, such as the female’s head angle or the wing angle difference from a preceding waggle, we swept a threshold across its range. At each threshold, we computed the True Positive Rate (TPR, fraction of correctly predicted wing choice) and the False Positive Rate (FPR, fraction of incorrectly predicted wing choice). The Area Under the Curve (AUC) of the resulting TPR vs. FPR plot was used to measure classifier performance, where AUC=0.5 signifies chance-level classification.

Behavioral state transitions

Four mutually exclusive behavioral states were defined during natural courtship: waggling, singing, stillness, and other. Each frame of the recording was assigned to one of these states based on a hierarchical classification scheme. First, stillness frames were identified based on the speed criteria described above. Next, waggling and singing bouts were assigned to their respective frames, overwriting any stillness assignments. This ensured that periods where flies were stationary while waggling or singing were correctly classified by their active behavior rather than as stillness. All remaining frames were classified as “other.”

To focus analysis on meaningful behavioral periods, minimum bout duration thresholds were established for each state. For stillness and other states, bout duration thresholds were determined using the knee-point algorithm applied to the distribution of bout durations, yielding thresholds of 0.71 s and 0.72 s, respectively (Figure S2C). Waggling and singing bouts retained their detection-based durations as described above. Bouts shorter than these thresholds were excluded from subsequent analyses.

State transition probabilities were computed by tallying all bout-to-bout transitions across recordings. To assess statistical significance, we performed permutation tests (10,000 iterations) where bout labels were randomly shuffled within each recording while preserving the frequency and duration of each state. This null model tested whether observed transition probabilities exceeded chance expectations (one-tailed test for enrichment). P-values were corrected for multiple comparisons using the Benjamini-Hochberg false discovery rate (FDR) procedure. Figure 3H presents the 3×3 transition probability submatrix for the three courtship-relevant states (stillness, waggling, singing), while the complete 4×4 transition matrix including all transitions to and from “other” states is shown in Figure S2D. For three-gram analysis, we examined all sequences of three consecutive behavioral bouts, excluding sequences containing “other” states or self-transitions.

For optogenetic experiments, behavioral state transition and three-gram analyses were performed using the same methods described above. We focused on the 15-second window following optogenetic onset and included only trials that contained both waggling and singing.

Quantification and statistical analysis

All statistical analyses were performed using custom scripts in MATLAB R2024a. Statistical tests were selected based on data distribution and experimental design. The Shapiro-Wilk test was used to assess normality of data distributions, which determined whether parametric or non-parametric tests were applied.

Tests used in this study included: paired t-tests for normally distributed paired comparisons; Wilcoxon signed-rank tests for non-normally distributed paired data; one-way ANOVA with Tukey’s post-hoc correction for normally distributed data across multiple groups; Kruskal-Wallis tests with Dunn-Šidák post-hoc correction for non-normally distributed multi-group comparisons; Friedman tests for non-parametric repeated measures analysis when comparing multiple conditions within subjects; Spearman rank correlation for testing monotonic relationships between continuous variables; and one-tailed permutation tests (10,000 iterations) with Benjamini-Hochberg false discovery rate correction for state transition probabilities.

Statistical details for each comparison are provided in figure legends, including sample size, test used, and p-values. Data are presented as mean ± SEM for bout-triggered averages and time series. Box plots display median (center line), interquartile range (box), and 1.5×IQR (whiskers), with individual data points overlaid. Violin plots show full data distributions with median and quartiles marked. Significance levels are indicated as ns (not significant), * (p < 0.05), ** (p < 0.01), or *** (p < 0.001).

Supplementary Material

1
2
Download video file (4.3MB, mp4)
3
Download video file (4.9MB, mp4)
4
Download video file (17.2MB, mp4)

Document S1. Figures S1–S4, Tables S1, and supplemental references

Video S1. Male Drosophila melanogaster performing singing and waggling towards a female, related to Figure 1.

The video shows side-by-side comparison of singing (left) and waggling (right) behaviors. Videos were recorded at 150 frames per second (fps) and are played back at 30 fps.

Video S2. Male Drosophila simulans and Drosophila sechellia performing waggling towards females, related to Figure 1 and Figure S1.

The video shows side-by-side comparison of waggling behavior in D. simulans (left) and D. sechellia (right). Videos were recorded at 150 frames per second (fps) and are played back at 30 fps.

Video S3. The stereotyped stillness-waggling-singing courtship sequence in Drosophila melanogaster, related to Figure 3.

The sequence begins as both flies decelerate to a state of mutual stillness. The male then initiates waggling followed by song. The video was recorded at 150 frames per second (fps) and is played back at 30 fps.

Highlights:

  • Drosophila males produce “waggling,” distinct rhythmic anti-phase wing oscillations.

  • Waggling forms a sequence: stillness to waggling to singing during courtship.

  • P1/pC1 neuron activation triggers the full sequence in solitary males.

  • P1/pC1 neurons show functional diversity in locomotor control.

Acknowledgments

We thank Yoon Woo Park for help with data annotation. We thank Junyu Li and Talmo Pereira6 for collecting and proofreading the wild type D. melanogaster courtship data. We thank Richard Benton, Kenta Asahina, Vivek Jayaraman, Barry Dickson, and Stephen Goodwin for gifting flies. We thank Jan Clemens and the entire Murthy lab for helpful discussions. This research was supported by the following grants to MM: HHMI Faculty Scholar award, NIH NINDS R35 Research Program Award, and NIH BRAIN R01 NS104899.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Declaration of interests

The authors declare no competing interests.

References

  • 1.Anderson DJ, and Perona P (2014). Toward a science of computational ethology. Neuron 84, 18–31. 10.1016/j.neuron.2014.09.005. [DOI] [PubMed] [Google Scholar]
  • 2.Yamamoto D, and Koganezawa M (2013). Genes and circuits of courtship behaviour in Drosophila males. Nat. Rev. Neurosci 14, 681–692. 10.1038/nrn3567. [DOI] [PubMed] [Google Scholar]
  • 3.Coen P, Clemens J, Weinstein AJ, Pacheco DA, Deng Y, and Murthy M (2014). Dynamic sensory cues shape song structure in Drosophila. Nature 507, 233–237. 10.1038/nature13131. [DOI] [PubMed] [Google Scholar]
  • 4.Coen P, Xie M, Clemens J, and Murthy M (2016). Sensorimotor transformations underlying variability in song intensity during Drosophila courtship. Neuron 89, 629–644. 10.1016/j.neuron.2015.12.035. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Roemschied FA, Pacheco DA, Aragon MJ, Ireland EC, Li X, Thieringer K, Pang R, and Murthy M (2023). Flexible circuit mechanisms for context-dependent song sequencing. Nature 622, 794–801. 10.1038/s41586-023-06632-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Pereira TD, Tabris N, Matsliah A, Turner DM, Li J, Ravindranath S, Papadoyannis ES, Normand E, Deutsch DS, Wang ZY, et al. (2022). SLEAP: A deep learning system for multi-animal pose tracking. Nat. Methods 19, 486–495. 10.1038/s41592-022-01426-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Inagaki HK, Jung Y, Hoopfer ED, Wong AM, Mishra N, Lin JY, Tsien RY, and Anderson DJ (2014). Optogenetic control of Drosophila using a red-shifted channelrhodopsin reveals experience-dependent influences on courtship. Nat. Methods 11, 325–332. 10.1038/nmeth.2765. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Hindmarsh Sten T, Li R, Hollunder F, Eleazer S, and Ruta V (2025). Male-male interactions shape mate selection in Drosophila. Cell 188, 1486–1503.e25. 10.1016/j.cell.2025.01.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.von Philipsborn AC, Liu T, Yu JY, Masser C, Bidaye SS, and Dickson BJ (2011). Neuronal control of Drosophila courtship song. Neuron 69, 509–522. 10.1016/j.neuron.2011.01.011. [DOI] [PubMed] [Google Scholar]
  • 10.Kohatsu S, and Yamamoto D (2015). Visually induced initiation of Drosophila innate courtship-like following pursuit is mediated by central excitatory state. Nat. Commun 6, 6457. 10.1038/ncomms7457. [DOI] [PubMed] [Google Scholar]
  • 11.Kimura K-I, Hachiya T, Koganezawa M, Tazawa T, and Yamamoto D (2008). Fruitless and doublesex coordinate to generate male-specific neurons that can initiate courtship. Neuron 59, 759–769. 10.1016/j.neuron.2008.06.007. [DOI] [PubMed] [Google Scholar]
  • 12.Clemens J, Coen P, Roemschied FA, Pereira TD, Mazumder D, Aldarondo DE, Pacheco DA, and Murthy M (2018). Discovery of a new song mode in Drosophila reveals hidden structure in the sensory and neural drivers of behavior. Curr. Biol 28, 2400–2412.e6. 10.1016/j.cub.2018.06.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Steinfath E, Khalili A, Stenger M, Schultze BL, Ravindran SN, Alizadeh K, and Clemens J (2024). A neural circuit for context-dependent multimodal signaling in Drosophila. bioRxiv. 10.1101/2024.12.04.625245. [DOI] [Google Scholar]
  • 14.Fabre CCG, Hedwig B, Conduit G, Lawrence PA, Goodwin SF, and Casal J (2012). Substrate-Borne Vibratory Communication during Courtship in Drosophila melanogaster. Curr. Biol 22, 2180–2185. 10.1016/j.cub.2012.09.042. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Ewing AW, and Bennet-Clark HC (1968). The courtship songs of Drosophila. Behaviour 31, 288–301. 10.1163/156853968x00298. [DOI] [Google Scholar]
  • 16.Bennet-Clark HC, and Ewing AW (1968). The wing mechanism involved in the courtship of Drosophila. J. Exp. Biol 49, 117–128. 10.1242/jeb.49.1.117. [DOI] [Google Scholar]
  • 17.Manning A (1960). The sexual behaviour of two sibling Drosophila species. Behaviour 15, 123–145. 10.1163/156853960x00133. [DOI] [Google Scholar]
  • 18.Cobb M, Connolly K, and Burnet B (1985). Courtship behaviour in the melanogaster species sub-group of Drosophila. Behaviour 95, 203–230. 10.1163/156853985x00136. [DOI] [Google Scholar]
  • 19.Welbergen PH, Scharloo W, and Van Dijken FR (1987). Collation of the Courtship Behaviour of the Sympatric Species Drosophila Melanogaster and Drosophila Simulans. Behaviour 101, 253–274. 10.1163/156853987x00017. [DOI] [Google Scholar]
  • 20.Sturtevant AH (1915). Experiments on sex recognition and the problem of sexual selection in Drosoophilia. J. Anim. Behav 5, 351–366. 10.1037/h0074109. [DOI] [Google Scholar]
  • 21.Mo W-Z, Li Z-M, Deng X-M, Chen A-L, Ritchie MG, Yang D-J, He Z-B, Toda MJ, and Wen S-Y (2022). Divergence and correlated evolution of male wing spot and courtship display between Drosophila nepalensis and D. trilutea. Insect Sci. 29, 1445–1460. 10.1111/1744-7917.12994. [DOI] [PubMed] [Google Scholar]
  • 22.Garrigan D, Kingan SB, Geneva AJ, Andolfatto P, Clark AG, Thornton KR, and Presgraves DC (2012). Genome sequencing reveals complex speciation in the Drosophila simulans clade. Genome Res. 22, 1499–1511. 10.1101/gr.130922.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Ribeiro IMA, Drews M, Bahl A, Machacek C, Borst A, and Dickson BJ (2018). Visual projection neurons mediating directed courtship in Drosophila. Cell 174, 607–621.e18. 10.1016/j.cell.2018.06.020. [DOI] [PubMed] [Google Scholar]
  • 24.Ning J, Li Z, Zhang X, Wang J, Chen D, Liu Q, and Sun Y (2022). Behavioral signatures of structured feature detection during courtship in Drosophila. Curr. Biol 32, 1211–1231.e7. 10.1016/j.cub.2022.01.024. [DOI] [PubMed] [Google Scholar]
  • 25.McKellar CE, Lillvis JL, Bath DE, Fitzgerald JE, Cannon JG, Simpson JH, and Dickson BJ (2019). Threshold-Based Ordering of Sequential Actions during Drosophila Courtship. Curr. Biol 29, 426–434.e6. 10.1016/j.cub.2018.12.019. [DOI] [PubMed] [Google Scholar]
  • 26.Hoopfer ED, Jung Y, Inagaki HK, Rubin GM, and Anderson DJ (2015). P1 interneurons promote a persistent internal state that enhances inter-male aggression in Drosophila. Elife 4. 10.7554/eLife.11346. [DOI] [Google Scholar]
  • 27.Jung Y, Kennedy A, Chiu H, Mohammad F, Claridge-Chang A, and Anderson DJ (2020). Neurons that function within an integrator to promote a persistent behavioral state in Drosophila. Neuron 105, 322–333.e5. 10.1016/j.neuron.2019.10.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Stockinger P, Kvitsiani D, Rotkopf S, Tirián L, and Dickson BJ (2005). Neural circuitry that governs Drosophila male courtship behavior. Cell 121, 795–807. 10.1016/j.cell.2005.04.026. [DOI] [PubMed] [Google Scholar]
  • 29.Pan Y, Meissner GW, and Baker BS (2012). Joint control of Drosophila male courtship behavior by motion cues and activation of male-specific P1 neurons. Proc. Natl. Acad. Sci. U. S. A 109, 10065–10070. 10.1073/pnas.1207107109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Koganezawa M, Kimura K-I, and Yamamoto D (2016). The neural circuitry that functions as a switch for courtship versus aggression in Drosophila males. Curr. Biol 26, 1395–1403. 10.1016/j.cub.2016.04.017. [DOI] [PubMed] [Google Scholar]
  • 31.Rideout EJ, Dornan AJ, Neville MC, Eadie S, and Goodwin SF (2010). Control of sexual differentiation and behavior by the doublesex gene in Drosophila melanogaster. Nat. Neurosci 13, 458–466. 10.1038/nn.2515. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Robinett CC, Vaughan AG, Knapp J-M, and Baker BS (2010). Sex and the single cell. II. There is a time and place for sex. PLoS Biol. 8, e1000365. 10.1371/journal.pbio.1000365. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Allen AM, Neville MC, Nojima T, Alejevski F, and Goodwin SF (2025). A role for exaptation in sculpting sexually dimorphic brains from shared neural lineages. bioRxiv. 10.1101/2025.06.04.657833. [DOI] [Google Scholar]
  • 34.Coleman RT, Morantte I, Koreman GT, Cheng ML, Ding Y, and Ruta V (2024). A modular circuit coordinates the diversification of courtship strategies. Nature 635, 142–150. 10.1038/s41586-024-08028-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Ishii K, Wohl M, DeSouza A, and Asahina K (2020). Sex-determining genes distinctly regulate courtship capability and target preference via sexually dimorphic neurons. Elife 9. 10.7554/eLife.52701. [DOI] [Google Scholar]
  • 36.Zhang W, Guo C, Chen D, Peng Q, and Pan Y (2018). Hierarchical control of Drosophila sleep, courtship, and feeding behaviors by male-specific P1 neurons. Neurosci. Bull 34, 1105–1110. 10.1007/s12264-018-0281-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Rubin GM, Managan C, Dreher M, Kim E, Miller S, Boone K, Robie AA, Taylor AL, Branson K, Schretter CE, et al. (2025). Networks of sexually dimorphic neurons that regulate social behaviors in Drosophila. bioRxiv. 10.1101/2025.10.21.683766. [DOI] [Google Scholar]
  • 38.Shirangi TR, Wong AM, Truman JW, and Stern DL (2016). Doublesex regulates the connectivity of a neural circuit controlling Drosophila male courtship song. Dev. Cell 37, 533–544. 10.1016/j.devcel.2016.05.012. [DOI] [PubMed] [Google Scholar]
  • 39.Yu JY, Kanai MI, Demir E, Jefferis GSXE, and Dickson BJ (2010). Cellular organization of the neural circuit that drives Drosophila courtship behavior. Curr. Biol 20, 1602–1614. 10.1016/j.cub.2010.08.025. [DOI] [PubMed] [Google Scholar]
  • 40.Lillvis JL, Wang K, Shiozaki HM, Xu M, Stern DL, and Dickson BJ (2024). Nested neural circuits generate distinct acoustic signals during Drosophila courtship. Curr. Biol 34, 808–824.e6. 10.1016/j.cub.2024.01.015. [DOI] [PubMed] [Google Scholar]
  • 41.Berg S, Beckett IR, Costa M, Schlegel P, Januszewski M, Marin EC, Nern A, Preibisch S, Qiu W, Takemura S-Y, et al. (2025). Sexual dimorphism in the complete connectome of the Drosophila male central nervous system. bioRxiv. 10.1101/2025.10.09.680999. [DOI] [Google Scholar]
  • 42.Hindmarsh Sten T, Li R, Otopalik A, and Ruta V (2021). Sexual arousal gates visual processing during Drosophila courtship. Nature 595, 549–553. 10.1038/s41586-021-03714-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Deutsch D, Clemens J, Thiberge SY, Guan G, and Murthy M (2019). Shared song detector neurons in Drosophila male and female brains drive sex-specific behaviors. Curr. Biol 29, 3200–3215.e5. 10.1016/j.cub.2019.08.008. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2149983-supplement-1.pdf (8.4MB, pdf)
2
Download video file (4.3MB, mp4)
3
Download video file (4.9MB, mp4)
4
Download video file (17.2MB, mp4)

Data Availability Statement

Data reported in this paper will be shared by the lead contact upon request. Code for detecting and annotating waggle bouts is on Github at github.com/murthylab and will be publicly available upon publication. Any additional information required to reanalyze the data reported in this paper is available from the Lead Contact upon request.

RESOURCES