Modularity in the evolution of visual signals associated with aggressive displays

Interactions between conspecifics commonly involve the use of 9 stereotyped display movements, which can vary markedly between species. Theoretically, 10 sexual selection by female choice can lead to large differences between species, but 11 sexual selection by male competition may result in more limited diversification. Here, we 12 evaluate display evolution in the aggressive signals of 10 leaf warbler species. Using 13 high-speed videography of territorial behavior, we quantify differences in wing motion 14 intensity and form. We find that both the rate of wing motion and the form of the display 15 remain similar across species, which we attribute to an effective signal maintained 16 through multiple speciation events. Differences among species arise though discrete 17 additions to the behavioral repertoire (three species), loss of display (one species) and the 18 presence of a pale patch on the wing. While some habitats differ discretely and 19 dramatically in light intensity, this cannot account for all the differences in display 20 behavior. We conclude that display evolution proceeds largely in a modular fashion. The 21 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint basic conventional signal is maintained across species, enabling modifications to appear 22 without loss of efficacy. 23 24 MAIN TEXT 25

26 Displays to conspecifics are employed in a wide range of social situations including, 27 during the breeding season, to attract mates (Mitoyen et al. 2019) and repel competitors 28 (van Staaden et al. 2011). Spectacular displays are associated with sexual selection by 29 female choice, and such displays often differ dramatically among closely related species. 30 In birds, this is exemplified by the striking secondary sexual traits and associated displays 31 of males in polygynous taxa, such as the Birds of Paradise (Scholes 2008, Scholes et al. 32 2017, Ligon et al. 2018, Miles & Fuxjager 2018), manakins (Prum 1990, Anciães & 33 Prum 2008), and hummingbirds (Clark et al. 2018, Simpson & McGraw 2019). Models 34 of diversification in such cases include co-evolution of male and female trait following 35 slight displacements from equilibrium (Lande 1981, Kirkpatrick 1982) and mutation-36 order selection, whereby different attractive mutations arise in different populations 37 (Price 2002, Mendelson et al. 2014). The implication from both the bewildering diversity 38 of secondary traits in such groups, and from these models, is that display behaviors may 39 shift in arbitrary, unpredictable directions, with little connection to differences among 40 environments (Fig. 1A). 41 In contrast to displays used to attract mates, we might expect aggressive displays 42 between males to be simple to transmit an unambiguous signal of intent (Morris 1957, 43 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Cullen 1966, Hurd & Enquist 2001). Further, such displays should often be short, so that 44 individuals can display sequentially, and thereby assess each other (Catchpole 1980). 45 While male competition can drive evolution through a number of mechanisms 46 (Tinghitella et al. 2018), short simple displays may restrict possibilities for divergence. 47 Instead, if the display is optimized to efficiently communicate between conspecifics, it 48 may be passed little changed through descendants, and exhibit substantial stasis (Fig. 49 1B). For example, in Anolis lizards, territorial displays directed to males are highly 50 stereotyped within species and differ in relatively small ways between species (Ord and 51 Martins 2006). 52 Spectacular divergence in arbitrary directions and extreme stasis lie at ends of a 53 continuum. Modifying both processes are effects of the environment (Fig. 1C). For 54 example, despite large differences among related Birds of Paradise in the form of their 55 displays, species found on the forest floor have larger display repertoires than those in the 56 canopy (Ligon et al. 2018, Miles & Fuxjager 2018). And despite the small differences 57 among Anolis lizards Ord and Martins (2006) were able to rank those differences (e.g., in 58 dewlap pulse rate) to show how different features of the display correlate with shade vs. 59 sunny habitats, number of sympatric species, occupancy of the canopy, and sexual size 60 dimorphism. This and other studies of intrasexual aggression (Jenssen 1977, Fleishman 61 1992, Clark et al. 2015) suggest that although diversification may be quite limited, a 62 relatively high fraction of the diversity may be a result of adaptation to different 63 environments, associated with selection for efficient communication. Consequently, the 64 evolution and adaptive significance of aggressive displays may be investigated using the 65 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint comparative method, which seeks correlations between display and a species morphology 66 and ecology (Fig. 1C). 67 The best known hypothesis that relates environmental differences to features of 68 display is that of sensory drive (Endler 1992, Cummings et al. 2018). In this hypothesis, 69 environmental differences impact transmission and perception, resulting in the evolution 70 of traits involved in communication. Sensory drive has been most often applied to the 71 evolution of color and color patterns, with one of the clearest examples being that of 72 plumage brightness among Phylloscopus warblers breeding along an elevational gradient 73 in the west Himalaya. Marchetti (1993) showed the brightness of an unpigmented patch 74 on the wing (termed the wing-bar) correlates positively with darkness of habitat, which 75 she inferred to result from evolutionary adjustments to maintain a certain level of 76 conspicuousness. Phylloscopus males compete for territories (Marchetti 1998, Scordato 77 2018), utilizing short-range aggressive displays composed of rapid wing movements, 78 which expose the wing bar (Marchetti 1993). Rapid wing movements are one of the most 79 commonly used motions during avian threats (Andrew 1956, 1961, 2008, Tinbergen 80 1960, Kenyon & Martin 2022), presumably because the wings are so easily moved, with 81 altered rates and overall form of wing motion potentially indicating aggressive 82 motivation, increasing conspicuousness, or a combination of factors. We set out to ask if 83 sensory drive has affected evolution of wing movements in the Phylloscopus, as 84 brightness is expected to impact perception of motion-based signals as well as colors 85 (Warrant 1999, Boström et al. 2016). 86 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint We address the following questions: 87 (1) Is there substantial conservation of display form (Fig. 1B)? 88 (2) Have discrete elements of display been added or lost across species (Fig. 1A)? 89 (3) Is there any evidence for diversification in display associated with habitat 90 (Fig. 1C)? For example, broader, more exaggerated wing movements may be 91 present in darker habitats to enhance visibility. 92 We find evidence in support of stasis of the main display, punctuated by unique 93 evolutionary events that result in the addition of a novel display or complete display loss. 94 The presence of discretely different behaviors which have evolved just once precludes the 95 strong use of the comparative method and suggests that a common mode of display 96 evolution is modular, with addition of unique elements. We draw on environmental 97 differences between species, plus previous work on the color patterns of these species 98 (Marchetti 1993), to make some adaptive hypotheses about display evolution. 99 100

101 STUDY SYSTEM 102 The genus Phylloscopus (leaf warblers) contains 76 species (Alstrom et al. 2018), 103 which vary in mass from 5-12 g. In all species, individuals spend much of their time 104 foraging for insects in trees and bushes (e.g., in a non-breeding season study, P. 105 trochiloides, consumed one arthropod every 14 seconds throughout the day (Price 1981)). 106 We studied 10 species that breed along a limited elevational gradient (2,000-4,000m; 107 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Price et al. 1997; Fig. 2) in the west Indian Himalayan state of Himachal Pradesh (Price 108 et al. 2003). During the breeding season, species partition themselves along this 109 elevational gradient in association with dominant plant species (Price 1991), resulting in 110 occupancy of distinct primary breeding habitats (Fig. 2). All species are partial or 111 complete migrants, spending the non-breeding season at lower elevations and latitudes. 112 All species have similar plumages, possessing greenish-olive to brown upperparts 113 and pale underparts. Many species have a light stripe of unmelanized feather keratin 114 across the tip of the greater covert feathers, producing a wing bar. Wing bar size varies 115 between species (Fig. 2, (Price & Pavelka 1996) and for the one species where it has been 116 studied (P. humei) wing-bar size is about 10% larger in males than females (Scordato et 117 al. 2012). Color of the sexes is similar, as assessed spectrophotometrically (unpublished 118 data). Plumages do not vary seasonally, except for feather wear, which can reduce the 119 size of wing bars over time (Scordato et al. 2012). 120 We conducted simulated territorial intrusions using playback experiments and 121 filmed species responses in the breeding season. To quantify variation in wing motion, we 122 applied a geometric morphometric approach to compare wing trajectory shapes within a 123 morphometric space. To assess the influence of the light environment we measured 124 habitat illuminance in the primary breeding habitats across the elevational gradient. 125 126 FIELD METHODS 127 Author 1 studied warbler behaviors during the breeding season in the Manali 128 Wildlife Sanctuary, Himachal Pradesh, India (32.25oN, 77.17oE, spanning 2000m - 129 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint 3600m, between April 22 –June 22, 2019 and April 25 – May 30, 2022) and at Nain 130 Gahar village, Himachal Pradesh, India (32.73 oN, 76.86 oE between June 11-21, 2019 131 and June 15 – July 8, 2022). They also visited two sites in Arunachal Pradesh (26.97 oN, 132 92.92 oE and 27.06 oN, 93.03 oE) and one site in Andhra Pradesh (17.81 oN, 82.49 oE) 133 during the non-breeding season (December 22, 2021 – January 22, 2022). We collected 134 two sets of data: the first on rate of wing movement during foraging and territorial 135 intrusions, and the second on form of the display (Table 1). Rate of wing movement was 136 collected because it was qualitatively apparent that movement increases during 137 aggressive responses (call note rate similarly increases, (Wheatcroft 2015) and because 138 variation in display rates may indicate differences in aggressive motivation and condition 139 (Yasukawa 1978, Clutton-Brock & Albon 1979, Ord & Evans 2003, Barnett et al. 2014). 140 Display form was collected to quantitatively assess whether displays vary using a 141 geometric morphometric approach and to explicitly address the question of how displays 142 have been modified across species. 143 To document the use of wing movements during foraging, we opportunistically 144 filmed individuals. When a species of interest was detected, we used a single Sony RX10 145 DSC III camera mounted on a tripod set at 60 frames per second (fps) and filmed the 146 individual for as long as possible. Birds were identified to species at the time of filming 147 or when reviewing the video footage through call notes, songs, plumage, or a 148 combination of these traits. 149 Most of the data comes from territorial playback experiments in the breeding 150 season. Author 1 located singing males between 0500-1100. Experiments consisted of 151 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint two parts. During Part I we established a filming focal point and observed the territory 152 owner’s response. We placed a Bluetooth speaker (Ultimate Ears WONDERBOOM) in 153 the target individual’s territory and played that species song to simulate a territorial 154 intrusion. If the territory owner responded by singing back and approaching the speaker, 155 we continued to play the song for 5 minutes to observe where the individual would perch 156 in the territory. This was to maximize the likelihood that the camera setup would capture 157 the behaviors of interest. In total, we attempted 306 behavioral trials. Of those, 43% (n = 158 132) were terminated during Part I due to either poor filming conditions or lack of 159 response from the territory owner. 160 Part II: Once a consistent focal point was established, we set up two camera 161 teams. The first team consisted of a single person with a camera (Sony RX10 DSC III) 162 mounted on a tripod, filming the target bird at 60 fps continuously during the trial. This 163 allows for an extended view of the display and was used to calculate wing flick rate and 164 record all the motions present in the species display repertoire. The second camera team 165 operated a high-speed camera assembly, which consisted of three Sony RX10 DSC III 166 cameras mounted on tripods, each equipped with a Ziv TRS-10 Timer Remote set to the 167 same channel, allowing for simultaneous remote triggering. We used three cameras to 168 increase the chances of capturing displays during which the bird is oriented laterally 169 (defined as the line from beak to tail running perpendicular to the camera lens). The 170 cameras were placed at least 3 m. from the focal point, in an arc with each camera 171 separated by 45o from the next one. The cameras were set to film at 480 fps on a delayed 172 trigger. This mode continuously films until the trigger is pressed, at which point the prior 173 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint two seconds of footage are written to the SD card, thereby allowing the capture of display 174 behaviors without a response delay from the observer. Both setups were left undisturbed 175 for 10 minutes before the trial started. During the trial the two observers remained 8 176 meters away from the focal point. 177 During the behavioral trials, Author 1 played a target species song for 10 minutes. 178 In 34% of trials (n = 174 total trials) the bird did not respond, at which point the trial was 179 deemed unsuccessful and ended (n = 59 terminated trials). For clarification, this differs 180 from the termination described in part I; here the individual responded during the pre-trial 181 period (Part I) but then ceased responding after the camera array was set up and trials 182 began. Because of set-up time and the 10-minute undisturbed period, approximately 20-183 25 minutes could elapse between song playbacks. 184 If the bird responded by calling or singing back or by approaching the speaker 185 during this time, the trial continued. The high-speed camera assembly was triggered by 186 Author 1 when the bird perched near the focal point, within the camera’s frame of view. 187 The single camera team filmed the bird for as long as possible until sight of the bird was 188 lost, at which point camera recording was paused. The playback trial continued until the 189 end of the 10-minute period, accumulating as many 2 second videos as possible when the 190 bird displays at the focal point. If the bird continuously displayed throughout the time of 191 the first trial, we would leave the camera array in place and begin a 5-minute pause 192 period. After the pause period, the trial was repeated up to a maximum of three times. In 193 total, we had 115 aggressively responding males distributed across 10 species and were 194 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint able to record high speed (480 fps) videos of 73 males and 60 fps videos of 68 males 195 (Table 1). 196 After the behavioral trials, we calibrated the filming area. We placed a 3’’x 5’’ 197 checkerboard and XRite ColorChecker Passport close to the area where the bird perched. 198 We then moved the checkerboard through the filming area, pointing to each of the camera 199 views. These tools were used to provide a scale visible from any camera angle. 200 201 LIGHT MEASUREMENTS 202 To assess the influence of the light environment on wing displays, we measured 203 habitat illuminance in the primary breeding habitats across the elevational gradient. In 204 both breeding season field sites (Manali, Himachal Pradesh and Nain Gahar, Himachal 205 Pradesh), we deployed Onset light and temperature loggers (HOBO Pendant MX2202 206 Temperature/Light Data Logger), resulting in 5-13 samples per habitat across the two 207 field locations (Table S1). Light loggers were mounted horizontally with the light sensor 208 facing the sky on either horizontal branches in woodlands (birch, conifer, rhododendron, 209 oak) or on PVC pipes staked in the ground in understory and open habitats. The 210 placement of loggers reflected where the associated species were commonly found; 211 canopy or midstory in wooded areas and low-lying shrubs or the ground in open or 212 understory areas. Loggers were configured to record light and temperature every minute. 213 Data from the loggers were downloaded to the HOBOconnect app via Bluetooth at the 214 time of collection. 215 216 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint ANALYSIS 217 WING MOTION 218 We quantified variation in wing motion using two measurements: rate (from the 219 60 fps videos) and form (from the 480 fps videos). From the single 60 fps camera, we 220 measured wing flick rate in two contexts: foraging and territorial response. Videos were 221 selected for foraging analysis if the individual could be seen actively searching for or 222 capturing prey. Territorial responses were filmed at the time of simulated territorial 223 intrusion experiments. We calculated rate as the number of wing flicks an individual 224 performs divided by the total amount of time the bird is present on screen. Using the trim 225 function on QuickTime Player (Apple Computer) with video playback at half speed, we 226 analyzed the video frame by frame to quantify wing flicks. To separate flicks from wing 227 movements used in locomotion we only counted wing flicks when the bird hopped less 228 than one body length during the observation sequence. Total display time is the difference 229 in the timestamp when the individual leaves the frame of view and the timestamp when 230 the focal individual first appears in the frame. We only selected individuals that 231 completed at least 5 wing flicks during the time on video. For individuals that left the 232 frame and reappeared over the course of a video recording we took the first display 233 sequence in which the bird performed at least 5 wing flicks. 234 To quantify the form of the display during the territorial playback experiments, 235 we used the high frame rate videos in which the bird is in a lateral orientation. We 236 determined lateral orientation visually and then confirmed orientation by viewing the 237 angle from the other two cameras. This resulted in a sample size of high-quality 238 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint recordings from 37 individuals from 10 species, reduced from a total high-frame rate 239 dataset of 73 individuals. We extracted frames using the extractFrames function in 240 StereoMorph (Olsen & Westneat 2015). We isolated each individual wing flick, defined 241 by the frame in which the wing begins the upstroke to when it returns to the starting 242 position after the downstroke. We numbered the frames from 1 (initiation) to the end of 243 the flick; the number of frames varied from 35-45. To compare the variation in the form 244 of the display while accounting for variation in duration, we described the shape of each 245 wing flick using 15 time points (Fig. 3). First, we identified three specific wing positions 246 from each video: initiation, end and maximum wing extension. The initiation and end of 247 the wing flick were assigned time points 1 and 15, respectively. The point of maximum 248 wing extension, when the wing transitions from the upstroke to the downstroke was 249 assigned time point 7 (Fig. 3). From these we added 12 additional time points by 250 extracting frames uniformly dispersed between 1-7 (5 points) and 8-15 (7 points). 251 All statistical analyses were conducted in R, version 4.2.2 (R Core Team 2022). 252 We used the labelFrames function in StereoMorph to place a landmark on the tip of the 253 8th primary feather for each of the 15 frames. Each wing flick is then described using 15 254 “homologous” landmarks, creating a shape capturing the trajectory of wing motion (Fig. 255 3). The landmarks describing the trajectory shapes were scaled and aligned using the 256 Generalized Procrustes Analysis (GPA) in the R package borealis (Angelini 2022) to 257 remove variables of size, rotation and orientation, leaving a set of aligned coordinates 258 that capture variation in shape. We performed a principal component analysis (PCA) on 259 the correlation matrix of the aligned coordinates and visualized the location of each 260 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint individual’s trajectory by plotting the first two PCA axes. We plotted convex hulls around 261 the data points in the morphospace generated by PC1 and PC2, the most significant axes 262 of variation. We used lme4 (Bates et al. 2015) to calculate percent variance between and 263 within species display components for both PC1 and PC2. 264 To visualize how the shape of wing movement changes along PC1 and PC2, we 265 back transformed the PCscores to their relative positions in the morphospace (Olsen 266 2017). The points along the outside of each backtransformed shape correspond to the 15 267 landmarks used to describe the trajectory. 268 269 LIGHT MEASUREMENTS 270 We extracted the lux values from the logger files and log-transformed the data. To 271 standardize for longer day lengths as the season progressed, we filtered the data to 272 include times between 0600 -1800, which is a time interval occurring after sunrise and 273 before sunset through the entirely of the breeding season. We then took the average lux 274 measurement per logger per day. Results from habitats present in both locations were 275 similar and we combined them. We fitted a linear mixed-effects model with habitat as a 276 fixed effect and location and logger ID as nested random effects using lme4. We then 277 performed a post-hoc pairwise comparison of the habitats using the R package emmeans 278 (Lenth 2023). 279 280

281 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint We observed three discrete behaviors (Fig. 3). One behavior, which we term the 282 double wing flick is shared between all species. This behavior resembles motions used 283 during takeoff, where both wings are moved simultaneously through rotation at the elbow 284 and shoulder joints during the upstroke. During the downstroke, the humerus is extended 285 horizontally from the body, resulting in the extension of the distal portion of the wing 286 until it folds back to rest near the starting position. One species (P . pulcher) commonly 287 conducts a double shiver flick, distinguished from the double flick by both shape (Fig. 288 5B) and rate (Fig. S1). During this behavior the wings are extended horizontally from the 289 body at the shoulder and undergo a series of rapid rotations at the elbow and wrist joint, 290 resulting in a shivering motion. During territorial displays, this behavior is repeated and 291 rarely interspersed with double wing flicks. Finally, two related species (P . occipitalis and 292 P . reguloides) conduct a single wing flick, whereby one wing is raised vertically from the 293 body at the elbow joint, but the wing does not undergo a horizontal extension. Instead, 294 the wing is placed back to the starting position before alternating with the other. Both 295 species regularly intersperse single wing flicks with double wing flicks during both 296 foraging and territorial displays (Table S2). 297 298 WING FLICK RATE 299 P. affinis does not flick its wings at all in response to aggressive playback, 300 although it does so when foraging (Fig. 4). All other species flick their wings 301 significantly faster during territorial displays than during foraging (Fig. 4, Table S3). All 302 species, including P . affinis, have similar foraging wing flick rates (F8,24 = 0.49, P = 0.8). 303 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Once P . affinis is excluded, species do not differ significantly in territorial display wing 304 flick rate (F7,14 = 0.44, P = 0.9, Table S4). Shiver flicks are the dominant behavior used 305 by P . pulcher during territorial contexts, comprising 95% of the motions used during a 306 display. Shiver wing flick rates are significantly faster than double wing flick rates (Fig. 307 S1). Shiver flicks are only used during territorial interactions, and we observed no 308 instances of this behavior during foraging. Within the foraging context, single wing flick 309 rates do not differ significantly from double wing flick rates in P . occipitalis and P . 310 reguloides. Single wing flicks used during territorial displays are significantly faster than 311 those used during foraging (Fig. S2). 312 313 WING TRAJECTORY 314 We first analyzed double wing flicks by the 9 species that use them during 315 territorial displays. The primary axis of variation (PC1: 52% of the variance explained) 316 describes reduced horizontal movement of the wing (Fig. 5A). Individuals with low 317 values of PC1 move their wings more elliptically while those with high values of PC1 318 move it along a more constrained vertical axis. PC1 scores do not vary significantly 319 among species (ANOV A: F8,19 = 2.03, P = 0.10, table S6). The second axis of variation 320 (PC2: 15% of the variance explained) corresponds to an offset of the starting and ending 321 points. Individuals with low values of PC2 tend to place their wings close to the point at 322 which they initiate their wing flick, while those with high values of PC2 have more 323 variability with where the final downstroke points land relative to where they start. 324 Species did not vary significantly in their PC2 scores (ANOV A: F8,19 = 1.75, P = 0.15, 325 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Table S6). Most of the variance is within versus between species (75% within for PC1 326 scores, 79% for PC2 scores). 327 Next, we included all three wing flick types in a single analysis, where we 328 combined observations from the 9 species to compare the three types. The three motions 329 differ significantly along both PC1 (ANOV A: F2, 37 = 9.24, P = 0.001, a posteriori tests 330 are in Table S8) and PC2 (ANOV A: F2,37 = 7.98, P = 0.001, a posteriori tests are in Table 331 S8). Shiver and single wing flicks have higher PC1 scores relative to double wing flicks 332 (Fig. 5) because the upstroke and downstroke landmarks are closer together, with less 333 horizontal motion. Both shiver and single wing flicks have lower PC2 scores than double 334 wing flicks, reflecting more consistency in placing the wing tip back in same region it 335 started. Although shiver flicks (n = 8) occupy a larger area of the morphospace along PC2 336 (Fig. 5B) than single wing flicks (n = 4) this may be due to a larger sample size capturing 337 more individual variation. 338 339 HABITAT LIGHT 340 Habitats differ in illuminance (Fig. 6); a posteriori pairwise tests indicate that 341 open is brighter than all other habitat types and understory is significantly darker than all 342 other habitat types (Table S10). Conifer, birch, oak, and rhododendron did not vary 343 significantly in brightness between each other. 344 345

346 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Within species, individuals often communicate using specific and repetitive 347 patterns of ritualized movements. In this study, we investigated the use of wing motion in 348 the aggressive displays of Phylloscopus warblers to assess the extent of display variation 349 and whether it is associated with habitat. We find that the primary display used in 350 response to territorial playback, the double wing flick, remains conserved in both rate and 351 overall form across species. Despite the widespread use of this display and its expected 352 efficacy in close-range interactions, three species have modified their behavioral 353 repertoires through the addition of novel, discrete behaviors: the shiver wing flick (P. 354 pulcher) and the single wing flick (P. occipitalis and P. reguloides) and one species, P. 355 affinis, has lost the display. This repertoire of wing flick behaviors align with descriptions 356 in earlier studies (Marchetti 1993), and our analysis introduces a novel quantitative lens, 357 offering a comparative perspective across species. These modifications are the result of 358 three evolutionary events. Further, a pale wing-bar has been independently added twice 359 during the divergence of these species from their common ancestor (Price and Pavelka 360 1996, Fig. 2), and is present in 7 of the species. We first evaluate why the primary display 361 remains similar across species, and then investigate each of the modifications in turn. 362 Finally, we evaluate the relationship between display and habitat. 363 364 WING DISPLAY DIVERSIFICATION AND FUNCTION 365 Behaviors used in an aggressive signaling context may be under greater selective 366 pressure to remain simple and consistent compared to behaviors used in courtship 367 (Tinbergen 1960, Irwin 1996), which are shaped primarily by sexual selection through 368 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint female choice Our results largely support this hypothesis, as the primary display used 369 across species, the double wing flick, is found in 9 of 10 species in response to territorial 370 playback. We applied a geometric morphometric approach to quantify variation in the 371 shape of the wing display to test for subtle and continuous differences in display form. 372 Instead, we largely found that most of the variation in the double wing flick is within 373 species, not between, with 75% and 79% of the variation occurring among individuals 374 within species than among species for PC1 and PC2 shape scores, respectively (Fig. 5). 375 Species also do not vary significantly in their rate of wing flicking in territorial contexts 376 (Fig. 4). This conservation in the double wing flick suggests that this behavior may be 377 under little selective pressure to diversify. 378 Within a species, wing displays are variable across two measures – rate and form 379 (Figs. 3 and 5). By comparing a signaling (aggression) and non-signaling (foraging) 380 context, we showed that an aggressive stimuli (song playback) induces an increase in 381 wing flicking rate, except for one species which drops the display altogether (Fig. 4). 382 Observations of aggressive interactions between conspecifics confirm the use of a high 383 wing flicking rate in territorial disputes, which, if unresolved, then escalate to chasing 384 behaviors and can end in physical fights (Price 1981). An increase in rate compared to a 385 nonsignaling context and an association with attack escalation imply the wing 386 movements are likely used as threat displays (Számadó 2003). Display rate is associated 387 with levels of aggressive motivation in many other taxa (Deag & Scott 1999, Lange & 388 Leimar 2003, Ord & Evans 2003, Castro et al. 2006, Elwood et al. 2006, Brown et al. 389 2007, Crothers & Cummings 2015). 390 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Potentially receivers may be assessing subtle information in form (Byers et al. 391 2010, Barske et al. 2011) but it is conceivable wing motion may have no information 392 content but rather serve to amplify other traits, such as color patches (Hasson 1991, 393 Bókony et al. 2006). Indeed, Marchetti (1993, 1998) showed the wing-bar of one 394 Phylloscopus species (P . humei) functions in aggressive interactions. Motion is one of the 395 most effective ways to capture attention (Abrams & Christ 2003, Franconeri & Simons 396 2003, Rushton et al. 2007) and in birds, wing movement is perhaps the simplest way to 397 increase visibility. Further work is required to elucidate what specific features 398 competitors are assessing during aggressive interactions, which should increase our 399 understanding of why features of the display are evolutionarily constrained. 400 401 ADDITIONAL DISPLAYS 402 In addition to the baseline double wing-flick, a subclade within the Phylloscopus 403 have a display in which species alternate single wings (see Table S2 for example 404 sequence) (del Hoyo et al. 2020a, b). The clade includes both P . reguloides and P . 405 occipitalis from this study. The single wing flicks are often intermingled with double 406 wing flicks during both foraging and aggressive displays (Table S2) and it is used by both 407 males and females during foraging throughout the year. These species form large flocks 408 in the winter (Macdonald & Henderson 2008, Hariharan et al. 2022), where wing motions 409 may serve to facilitate flock cohesion or communication. The rate of the single wing 410 flicks increases during male aggressive interactions (Fig. S2), so the display functionally 411 operates in the same way as the double wing flick. However, the single wing flick is 412 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint distinctive in its trajectory, where the distal portion of the wing remains unextended and 413 most of the motion is concentrated in lifting the wing at the wrist joint. This results in 414 relatively less horizontal movement than the double flick (Fig 5B). 415 The other display, employed by P . pulcher is different. This display, the shiver 416 double flick, is used only in the breeding season and largely replaces that of the double 417 wing flick although the double flick is still occasionally used during display bouts. The 418 shiver flick is characterized by faster movement (Fig. S1), achieved through a series of 419 rotations at the wrist joint (Fig. 3). This also results in relatively less horizontal motion 420 than the double wing flick (Fig. 5B), although the motion itself is different from the 421 single wing flick. It resembles the motions used by young birds of all species when they 422 are begging for food (Supp. Video 1), but the reasons why it has been established as an 423 aggressive display in this species alone remain obscure. 424 The addition of these discrete behavioral elements to aggressive displays rather 425 than a replacement of the shared form mirrors results from aggressive contexts in other 426 taxa. Comparative analyses of Anolis have documented the use of display modifiers, 427 which are additional movements that are added to shared core displays (Jenssen 1977, 428 Ord et al. 2002). Multiple forms of threat display may have evolved to reflect different 429 escalatory steps (Andersson 1976, Hurd & Enquist 2001) or to overcome reduced 430 reliability in the original signal (Andersson 1980). Alternatively, different display 431 behaviors may mediate species recognition (Macedonia & Stamps 1994, Clark et al. 432 2015), as has been suggested for variation in other visual signals such as color 433 (Couldridge & Alexander 2002, Klomp et al. 2017, Dyson et al. 2020). In our case, 434 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint transmission of species identity seems unlikely as the displays are used at close range 435 once a challenger has been identified as a conspecific. Further, the two closely related 436 species with single and double wing flicks in their repertoire, are exceptionally similar in 437 their plumage and morphology. Hence, they would be expected to have diverged in the 438 display if it was evolved in species recognition. 439 440 SENSORY DRIVE 441 In Kashmir, Marchetti (1993) observed that Phylloscopus species with wing-bars 442 inhabited darker environments than those without wing-bars, which she attributed to 443 sensory drive. She argued that in darker habitats, species maintain visibility by becoming 444 brighter in appearance. P. affinis is the one species without wing-bars held in common 445 between that study and ours. This species breeds above tree line in high alpine juniper 446 (Price 1991), which has substantially higher illuminance than all other habitats we 447 studied (Fig. 6), reflecting the open composition of this habitat with little to no tree cover. 448 This species does not flick its wings in display (Marchetti 1993) suggesting a role for 449 sensory drive in not only affecting plumage, but also display. The lack of wing flicking in 450 this species is not a consequence of reduced aggressive responses. Indeed, in our 451 experiment, the territory owner responded very aggressively, singing back, approaching, 452 and even attacking the speaker, but it never flicks its wings. The signaling environment 453 may provide some clues as to why it has dropped wing movements. A distinctive feature 454 of P . affinis is the prominent yellow (carotenoid based) underparts (Grimmett et al. 2012 455 p. 340 plate 151). We suggest that detection and assessment in this species may be 456 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint achieved by display of the underparts. It bears noting that two other species without 457 wing-bars both flick their wings. One, P . xanthoschistos, has yellow underparts but lives 458 in woodland, and the second, P . sindianus lives in relatively open habitats, whose light 459 environment we were not able to measure but is likely to be intermediate between that of 460 open habitat and dense woodland. 461 In dim light conditions the tradeoff between temporal and spatial resolution 462 becomes exacerbated (Lythgoe 1979). We predicted that poor motion discrimination in 463 dark environments can lead to pressures to exaggerate critical features of a motion-based 464 display, which may lead to interspecific variation in the use of wing movement. 465 However, with the exception of P. affinis’s habitat light intensity was similar across the 466 habitat types occupied by other species. This differs from the results found in Kashmir by 467 Marchetti (1993) and suggests that there may be geographic variation in habitat features 468 across these species distributions. However, light intensity is only one axis upon which 469 the sensory environment can vary. The spatial organization of the background (Hulse et 470 al. 2020) and its motion (Ord et al. 2007, Peters 2013) are other variables that should 471 affect the perception of a visual display and remain to be assessed in this system. 472

473 Our ability to dissect display movements using high speed video coupled with a novel use 474 of a morphometric approach to study bird displays shows that rather than diversify across 475 species, the form of the primary aggressive display has been largely conserved. 476 Aggressive displays are expected to be simple in form in order to convey an 477 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint unambiguous message (Hurd & Enquist 2001), e.g. through changes in rate. Once such a 478 display efficiently conveys a message it may be carried through subsequent speciation 479 events. Nevertheless, we find that the primary display has been built on to generate 480 differences among species, through either its complete loss, by the addition or subtraction 481 of color patches on the wing, or by addition of qualitatively different displays. The effect 482 of the signaling environment is weak, with the only possibility we highlight being the 483 loss of display in one species. Given that environments vary in many ways other than 484 light intensity, such as foliage structure and background color, we anticipate that future 485 detailed studies of habitat will further our understanding of the origin of qualitatively 486 different displays, and their link to color patch evolution. At present, however, we 487 consider aggressive displays to have evolved through a mix of strong stabilizing selection 488 on some elements, “arbitrary” addition of an effective display in some lineages, and mild 489 influences of the habitat (Fig. 1). 490 491

492 Abrams, R.A. & Christ, S.E. 2003. Motion onset captures attention. Psychol. Sci. 14: 493 427–432. 494 Alstrom, P., Rheindt, F.E., Zhang, R., Zhao, M., Wang, J., Zhu, X., Gwee, C.Y., Hao, 495 Y., Ohlson, J., Jia, C., Prawiradilaga, D.M., Ericson, P.G.P., Lei, F. & Olsson, 496 U. 2018. Complete species-level phylogeny of the leaf warbler (Aves: 497 Phylloscopidae) radiation. Molecular Phylogenetics and Evolution 126: 141–152. 498 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Anciães, M. & Prum, R.O. 2008. Manakin display and visiting behaviour: a 499 comparative test of sensory drive. Anim. Behav. 75: 783–790. 500 Andersson, M. 1976. Social Behaviour and Communication in the Great Skua. 501 Behaviour 58: 40–77. 502 Andersson, M. 1980. Why are there so many threat displays? J. Theor. Biol. 86: 773–503 781. 504 Andrew, R.J. 1956. The aggressive and courtship behavior of certain Emberizines. 505 Behaviour 10: 255–308. 506 Andrew, R.J. 1961. The displays given by Passerines in courtship and reproductive 507 fighting: a review (Part A). Ibis 103a: 549–579. 508 Andrew, R.J. 2008. The displays given by passerines in courtship and reproductive 509 fighting: A review. Ibis 103a: 315–348. 510 Angelini, D. 2022. borealis: Tools for reproducible geometric morphometric analysis. 511 Apple Computer. n.d. QuickTime Player. Cupertino, California. 512 Barnett, C.A., Sakaluk, S.K. & Thompson, C.F. 2014. Aggressive displays by male 513 House Wrens are composed of multiple components that predict attack. J. Field 514 Ornithol. 85: 56–62. 515 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Barske, J., Schlinger, B.A., Wikelski, M. & Fusani, L. 2011. Female choice for male 516 motor skills. Proc. Biol. Sci. 278: 3523–3528. 517 Bates, D., Mächler, M., Bolker, B. & Walker, S. 2015. Fitting Linear Mixed-Effects 518 Models Using lme4. J. Stat. Softw. 67: 1–48. 519 Bókony, V ., Lendvai, Á.Z. & Liker, A. 2006. Multiple cues in status signalling: The role 520 of wingbars in aggressive interactions of male house sparrows. Ethology 112: 521 947–954. 522 Boström, J.E., Dimitrova, M., Canton, C., Håstad, O., Qvarnström, A. & Ödeen, A. 523 2016. Ultra-Rapid Vision in Birds. PLoS One 11: e0151099. 524 Brown, W.D., Chimenti, A.J. & Siebert, J.R. 2007. The payoff of fighting in house 525 crickets: Motivational asymmetry increases male aggression and mating success. 526 Ethology 113: 457–465. 527 Byers, J., Hebets, E. & Podos, J. 2010. Female mate choice based upon male motor 528 performance. Anim. Behav. 79: 771–778. 529 Castro, N., Ros, A.F.H., Becker, K. & Oliveira, R.F. 2006. Metabolic costs of 530 aggressive behaviour in the Siamese fighting fish,Betta splendens. Aggress. 531 Behav. 32: 474–480. 532 Catchpole, C.K. 1980. Sexual selection and the evolution of complex songs among 533 European warblers of the genus Acrocephalus. Behaviour 74: 149–166. 534 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Clark, C.J., McGuire, J.A., Bonaccorso, E., Berv, J.S. & Prum, R.O. 2018. Complex 535 coevolution of wing, tail, and vocal sounds of courting male bee hummingbirds. 536 Evolution 72: 630–646. 537 Clark, D.L., Macedonia, J.M., Rowe, J.W., Stuart, M.A., Kemp, D.J. & Ord, T.J. 538 2015. Evolution of displays in Galápagos lava lizards: comparative analyses of 539 signallers and robot playbacks to receivers. Anim. Behav. 109: 33–44. 540 Clutton-Brock, T.H. & Albon, S.D. 1979. The roaring of Red Deer and the evolution of 541 honest advertisement. Behaviour 69: 145–170. 542 Couldridge, V .C.K. & Alexander, G.J. 2002. Color patterns and species recognition in 543 four closely related species of Lake Malawi cichlid. Behav. Ecol. 13: 59–64. 544 Crothers, L.R. & Cummings, M.E. 2015. A multifunctional warning signal behaves as 545 an agonistic status signal in a poison frog. Behav. Ecol. 26: 560–568. 546 Cullen, J.M. 1966. E. Ritualization of animal activities in relation to phylogeny, 547 speciation and ecology: Reduction of ambiguity through ritualization. Philos. 548 Trans. R. Soc. Lond. 251: 363–374. 549 Cummings, M.E., Endler, J.A. & Handling editor: Rebecca C. Fuller. 2018. 25 Years 550 of sensory drive: the evidence and its watery bias. Curr. Zool. 64: 471–484. 551 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Deag, J.M. & Scott, G.W. 1999. “Conventional” Signals in Avian Agonistic Displays: 552 Integrating Theory, Data and Different Levels of Analysis. J. Theor. Biol. 196: 553 155–162. 554 del Hoyo, J., Collar, N. & Christie, D. 2020a. Claudia’s Leaf Warbler (Phylloscopus 555 claudiae). In: Birds of the World (J. del Hoyo, A. Elliott, J. Sargatal, D. A. 556 Christie, & E. de Juana, eds). Cornell Laboratory of Ornithology. 557 del Hoyo, J., Collar, N., Christie, D. & Kirwan, G.M. 2020b. Hartert’s Leaf Warbler 558 (Phylloscopus goodsoni). In: Birds of the World (J. del Hoyo, A. Elliott, J. 559 Sargatal, D. A. Christie, & E. de Juana, eds). Cornell Laboratory of Ornithology. 560 Dyson, M.L., Perez, D.M., Curran, T., McCullough, E.L. & Backwell, P.R.Y. 2020. 561 The role of claw color in species recognition and mate choice in a fiddler crab. 562 Behav. Ecol. Sociobiol. 74: 116. 563 Elwood, R.W., Pothanikat, R.M.E. & Briffa, M. 2006. Honest and dishonest displays, 564 motivational state and subsequent decisions in hermit crab shell fights. Anim. 565 Behav. 72: 853–859. 566 Endler, J.A. 1992. Signals, signal conditions, and the direction of evolution. Am. Nat. 567 139: S125–S153. 568 Fleishman, L.J. 1992. The influence of the sensory system and the environment on 569 motion patterns in the visual displays of anoline lizards and other vertebrates. Am. 570 Nat. 139: 36–61. 571 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Franconeri, S.L. & Simons, D.J. 2003. Moving and looming stimuli capture attention. 572 Percept. Psychophys. 65: 999–1010. 573 Grimmett, R., Inskipp, C. & Inskipp, T. 2012. Birds of India, Pakistan, Nepal, 574 Bangladesh, Bhutan, Sri Lanka, and the Maldives, 2nd ed. Princeton University 575 Press. 576 Hariharan, P., Bangal, P., Sridhar, H. & Shanker, K. 2022. Habitat use by mixed-577 species bird flocks in tropical forests of the Western Ghats, India. J. Trop. Ecol. 578 38: 393–400. 579 Hasson, O. 1991. Sexual displays as amplifiers: practical examples with an emphasis on 580 feather decorations. Behav. Ecol. 2: 189–197. 581 Hulse, S.V ., Renoult, J.P. & Mendelson, T.C. 2020. Sexual signaling pattern correlates 582 with habitat pattern in visually ornamented fishes. Nat. Commun. 11: 2561. 583 Hurd, P.L. & Enquist, M. 2001. Threat display in birds. Can. J. Zool. 79: 931–942. 584 Irwin, R.E. 1996. The Phylogenetic Content of Avian Courtship Display and Song 585 Evolution. , doi: 10.1093/oso/9780195092103.003.0008. 586 Jenssen, T.A. 1977. Evolution of Anoline Lizard Display Behavior. American Zoologist 587 17: 203–215. 588 Kenyon, H.L. & Martin, P.R. 2022. Aggressive signaling among competing species of 589 birds. PeerJ 10: e13431. 590 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Kirkpatrick, M. 1982. SEXUAL SELECTION AND THE EVOLUTION OF FEMALE 591 CHOICE. Evolution 36: 1–12. 592 Klomp, D.A., Stuart-Fox, D., Cassidy, E.J., Ahmad, N. & Ord, T.J. 2017. Color 593 pattern facilitates species recognition but not signal detection: a field test using 594 robots. Behav. Ecol. 28: 597–606. 595 Lande, R. 1981. Models of speciation by sexual selection on polygenic traits. Proc. Natl. 596 Acad. Sci. U.S.A. 78: 3721–3725. 597 Lange, H. & Leimar, O. 2003. The function of threat display in wintering great tits. 598 Anim. Behav. 65: 573–584. 599 Lenth, R.V . 2023. Estimated Marginal Means, aka Least-Squares Means [R package 600 emmeans version 1.8.5]. Available at: https://CRAN.R-601 project.org/package=emmeans. 602 Ligon, R.A., Diaz, C.D., Morano, J.L., Troscianko, J., Stevens, M., Moskeland, A., 603 Laman, T.G. & Scholes, E., 3rd. 2018. Evolution of correlated complexity in the 604 radically different courtship signals of birds-of-paradise. PLoS Biol. 16: 605 e2006962. 606 Lythgoe, J.N. 1979. The Ecology of Vision. Clarendon Press. 607 Macdonald, D.W. & Henderson, D.G. 2008. Aspects of the behaviour and ecology of 608 mixed-species bird flocks in Kashmir. Ibis 119: 481–493. 609 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Macedonia, J.M. & Stamps, J.A. 1994. Species recognition in Anolis grahami (sauria, 610 iguanidae): Evidence from responses to video playbacks of conspecific and 611 heterospecific displays. Ethology 98: 246–264. 612 Marchetti, K. 1993. Dark habitats and bright birds illustrate the role of the environment 613 in species divergence. Nature 362: 149–152. 614 Marchetti, K. 1998. The evolution of multiple male traits in the yellow-browed leaf 615 warbler. Anim. Behav. 55: 361–376. 616 Mendelson, T.C., Martin, M.D. & Flaxman, S.M. 2014. Mutation-order divergence by 617 sexual selection: diversification of sexual signals in similar environments as a first 618 step in speciation. Ecol. Lett. 17: 1053–1066. 619 Miles, M.C. & Fuxjager, M.J. 2018. Synergistic selection regimens drive the evolution 620 of display complexity in birds of paradise. J. Anim. Ecol. 87: 1149–1159. 621 Mitoyen, C., Quigley, C. & Fusani, L. 2019. Evolution and function of multimodal 622 courtship displays. Ethology 125: 503–515. 623 Morris, D. 1957. “Typical Intensity” and Its Relation To the Problem of Ritualisation. 624 Behaviour 11: 1–12. 625 Olsen, A.M. 2017. Feeding ecology is the primary driver of beak shape diversification in 626 waterfowl. Funct. Ecol. 31: 1985–1995. 627 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Olsen, A.M. & Westneat, M.W. 2015. StereoMorph: An R package for the collection of 628 3D landmarks and curves using a stereo camera set-up. Methods Ecol. Evol. 6: 629 351–356. 630 Ord, T.J., Blumstein, D.T. & Evans, C.S. 2002. Ecology and signal evolution in lizards. 631 Biol. J. Linn. Soc. Lond. 77: 127–148. 632 Ord, T.J. & Evans, C.S. 2003. Display Rate and Opponent Assessment in the Jacky 633 Dragon (Amphibolurus muricatus): An Experimental Analysis. Behaviour 140: 634 1495–1508. 635 Ord, T.J., Peters, R.A., Clucas, B. & Stamps, J.A. 2007. Lizards speed up visual 636 displays in noisy motion habitats. Proc. Biol. Sci. 274: 1057–1062. 637 Peters, R.A. 2013. Noise in Visual Communication: Motion from Wind-Blown Plants. 638 In: Animal Communication and Noise (H. Brumm, ed), pp. 311–330. Springer 639 Berlin Heidelberg, Berlin, Heidelberg. 640 Price, T. 1991. Morphology and Ecology of Breeding Warblers Along an Altitudinal 641 Gradient in Kashmir, India. J. Anim. Ecol. 60: 643–664. 642 Price, T. & Pavelka, M. 1996. Evolution of a colour pattern: history, development, and 643 selection. J. Evol. Biol. 9: 451–470. 644 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Price, T., Zee, J., Jamdar, K. & Jamdar, N. 2003. Bird Species Diversity Along the 645 Himalaya: a Comparison of Himachal Pradesh with Kashmir. J. Bombay Nat. 646 Hist. Soc. 100: 394–410. 647 Price, T.D. 1981. the Ecology of the Greenish Warbler Phylloscopus Trochiloides in Its 648 Winter Quarters. Ibis 123: 131–144. 649 Price, T.D. 2002. Domesticated birds as a model for the genetics of speciation by sexual 650 selection. Genetica 116: 311–327. 651 Price, T.D., Helbig, A.J. & Richman, A.D. 1997. Evolution of breeding distributions in 652 the Old World leaf warblers (Genus Phylloscopus). Evolution 51: 552–561. 653 Prum, R.O. 1990. Phylogenetic analysis of the evolution of display behavior in the 654 Neotropical manakins (Aves: Pipridae). Ethology 84: 202–231. 655 R Core Team. 2022. R: A Language and Environment for Statistical Computing. 656 Rushton, S.K., Bradshaw, M.F. & Warren, P.A. 2007. The pop out of scene-relative 657 object movement against retinal motion due to self-movement. Cognition 105: 658 237–245. 659 Scholes, E., Gillis, J.M. & Laman, T.G. 2017. Visual and acoustic components of 660 courtship in the bird-of-paradise genus Astrapia (Aves: Paradisaeidae). PeerJ 5: 661 e3987. 662 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Scholes, E., Iii. 2008. Evolution of the courtship phenotype in the bird of paradise genus 663 Parotia (Aves: Paradisaeidae): homology, phylogeny, and modularity. Biol. J. 664 Linn. Soc. Lond. 94: 491–504. 665 Scordato, E.S.C. 2018. Male competition drives song divergence along an ecological 666 gradient in an avian ring species. Evolution 72: 2360–2377. 667 Scordato, E.S.C., Bontrager, A.L. & Price, T.D. 2012. Cross-generational effects of 668 climate change on expression of a sexually selected trait. Curr. Biol. 22: 78–82. 669 Simpson, R.K. & McGraw, K.J. 2019. Interspecific Covariation in Courtship Displays, 670 Iridescent Plumage, Solar Orientation, and Their Interactions in Hummingbirds. 671 Am. Nat. 194: 441–454. 672 Számadó, S. 2003. Threat displays are not handicaps. J. Theor. Biol. 221: 327–348. 673 Tinbergen, N. 1960. Comparative Studies of the Behaviour of Gulls (Laridae): a 674 Progress Report. Behaviour 15: 1–69. 675 Tinghitella, R.M., Lackey, A.C.R., Martin, M., Dijkstra, P.D., Drury, J.P., Heathcote, 676 R., Keagy, J., Scordato, E.S.C. & Tyers, A.M. 2018. On the role of male 677 competition in speciation: a review and research agenda. Behav. Ecol. 29: 783–678 797. 679 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint van Staaden, M.J., Searcy, W.A. & Hanlon, R.T. 2011. Chpater 3 - Signaling 680 Aggression. In: Advances in Genetics (R. Huber, D. L. Bannasch, & P. Brennan, 681 eds), pp. 23–49. Academic Press. 682 Warrant, E.J. 1999. Seeing better at night: life style, eye design and the optimum 683 strategy of spatial and temporal summation. Vision Res. 39: 1611–1630. 684 Wheatcroft, D. 2015. Repetition rate of calls used in multiple contexts communicates 685 presence of predators to nestlings and adult birds. Anim. Behav. 103: 35–44. 686 Yasukawa, K. 1978. Aggressive tendencies and levels of a graded display: Factor 687 analysis of response to song playback in the redwinged blackbird (Agelaius 688 phoeniceus). Behav. Biol. 23: 446–459. 689 Yu, G., Smith, D.K., Zhu, H., Guan, Y. & Lam, T.T.-Y. 2017. GGTREE: an R package 690 for visualization and annotation of phylogenetic trees with their covariates and 691 other associated data. Methods in Ecology and Evolution 8: 28–36. 692 693 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint TABLES AND TABLE LEGENDS 694 695 Species Foraging videos (n) Behavioral trials with aggressively responding males (n) Territorial 60 fps video recordings used in rate analysis (n) Territorial high- speed (480 fps) video recordings (n) High-speed videos with a lateral orientation used in shape analysis P . affinis 3 (2 b, 1 w) 11 5 5 NA P . chloronotus 3 b 16 5 11 7 double P . humei 4 b 17 4 13 5 double P . magnirostris 0 5 0 3 3 double P . occipitalis 3 b 15 2 10 3 double, 3 single P . pulcher 4 b 18 2 12 2 double, 7 shiver P . reguloides 5 (3 b, 2 w) 12 2 7 2 double, 1 single P . sindianus 2 b 4 2 3 1 double P . trochiloides 5 (1 b, 4 w) 6 2 2 2 double P . xanthoschistos 4 b 11 3 6 3 double 696 Table 1: Sample sizes from the video data. Foraging videos (column 2) were taken 697 opportunistically through the breeding season (April – July, labeled with b) and a few 698 from the nonbreeding winter season (December – January, labeled with w). Behavioral 699 trials were filmed using a single 60 fps camera (column 4) and a 480 fps high-speed 700 camera array (column 5) and were all conducted during the breeding season. Many 701 individuals readily performed territorial wing displays (column 3) but only a subset of 702 these were captured on video used in data analysis. Double, single and shiver wing flicks 703 in column 6 refer to three qualitatively different displays, as described further in the 704

section. 705 706 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint FIGURES 707 708 Figure 1: We consider three possible processes affecting the evolution of sexually 709 selected traits. Anticlockwise from bottom left. (A) A collection of quite different 710 displays accumulates among species, exemplified by sexual selection through female 711 choice in polygynous species, with little influence of environmental differences. 712 (B) Because traits used in competition at short range are expected to be optimized to be 713 unambiguous and striking, all species inherit a similar display through their common 714 ancestor. (C) The form of the display is modified according to environmental conditions 715 (e.g. through sensory drive). The drawing indicates our findings. First, we show a basic 716 display is conserved across species, as in (A). Second, we show some species have added 717 unique components to this basic display, including some which can be related to 718 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint environment (plumage patch) and others (a second display or loss of display), whose 719 origins are less clear. 720 721 722 Figure 2. Left Phylloscopus phylogeny (Alstrom et al. 2018), with the 10 species in this 723 study highlighted in black. Note that the species we study span the root of the tree. Right. 724 Size of the point at the tip corresponds to each species wing bar size (Price & Pavelka 725 1996) and is color coded by primary habitat during the breeding season; note the two 726 species pairs which show convergent evolution in habitat. Constructed using ggtree (Yu 727 et al. 2017) 728 729 730 Psibilatrix Pbonelli Porientalis Pmaculipennis PinornatusPsubviridisPyunnanensis PforrestiPkansuensisPproregulusPtytleriPschwarziParmandiiPgriseolusPoccisinensis PgrammicepsPmontisPcalciatilisPrickettiPcantator PgoodsoniPclaudiae Psarasinorum Pmakirensis Pamoenus Pmaforensis Ptrivirgatus Ppresbytes Pdavisoni PogilviegrantiPhainanus Ppulcher Phumei Pchloronotus PreguloidesPoccipitalis Pxanthoschistos Psubaffinis PcollybitaPcanariensis Pibericus Ptrochilus PneglectusPfuscatus Pfuligiventer PintermediusPburkii Ptephrocephalus Ppoliogenys Pvalentini Pwhistleri PomeiensisPsoror Pruficapilla Pumbrovirens Pbudongoensis Pherberti Plaurae Plaetus Polivaceus Pcebuensis Pcoronatus Pijimae Pcastaniceps PemeiensisPnitidus PplumbeitarsusPborealoidesPtenellipesPborealis Pexaminandus Pxanthodryas Psindianus Paffinis Ptrochiloides Pmagnirostris pulcher humei chloronotus sindianus affinis reguloides occipitalis xanthoschistos trochiloides magnirostris Habitat Birch Broadleaf Conifer High alpine scrub Oak Rhododendron Riverine Woodland Wing.Bar.Size 0 1 2 3 4 5 A. pulcher humei chloronotus sindianus affinis reguloides occipitalis xanthoschistos trochiloides magnirostris Habitat Birch Broadleaf Conifer Oak Open (high alpine scrub) Rhododendron Riverine Woodland Wing Bar Size (mm) 0 10 20 30 40 50 B. (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint 731 Figure 3. Left Trajectories for the displays observed in the warblers studied. Plotted is the 732 distance moved in 15 equal time intervals. The dots correspond to 3 time points. L1 733 (green): when the wing flick begins. L5 (orange): when the wing reaches the 734 upstroke:downstroke transition. L13 (purple): A sample landmark showing how this 735 position can differ depending on the type of motion used in display. Right Video frames 736 for each labeled point. From the top, species are Phylloscopus chloronotus, P. reguloides, 737 P. pulcher. Video examples of these behaviors can be found in the supplementary data. 738 739 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint 740 Figure 4: Species territorial wing flick rate plotted against foraging wing flick rate, with 741 standard error. The black dashed line is the line of equality; 8 species flick wings faster in 742 territorial interactions. Among the 9 species foraging and territorial rate are not correlated 743 (r = 0.43, P = 0.2). For sample sizes for each species, see Table 1. 744 745 0 1 2 3 0 1 2 3 Foraging wing flick rate (wing flicks/sec) Territorial wing flick rate (wing flicks/sec) Species P. affinis P. chloronotus P. humei P. occipitalis P. pulcher P. reguloides P. sindianus P. trochiloides P. xanthoschistos (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint 746 Figure 5. (A) Double wing flick shapes for individual males. The colored lines 747 circumscribe the convex hull for each species (note that some species have a sample size 748 of n = 2, so are connected by a line). Background (in grey) illustrates the trajectories; 749 upstroke landmarks (points 1-6) are in green, the upstroke:downstroke transition (point 7) 750 is shown in orange and downstroke landmarks (points 8-15) are in purple. PC1 (52% of 751 the variance) describes a restriction in horizontal movement of the wing and PC2 (15%) 752 represents an offset in the starting and ending points of the wing. (B) Principal 753 components were conducted on the entire dataset for the three wing flick types. The 754 convex hulls enclose all species for each wing flick type (double, as in the left plot, 755 single: reguloides, occipitalis, shiver pulcher.) PC1 (46%) describes a restriction in 756 horizontal movement of the wing and PC2 (13%) an offset in the starting and ending 757 points of the wing. 758 759 760 A. B. PC1: Restricted horizontal movement PC2: Offset in starting and ending points A. B. PC1: Restricted horizontal movement PC2: Offset in starting and ending points (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 22, 2024. ; https://doi.org/10.1101/2024.02.22.581650doi: bioRxiv preprint Figure 6. Brightness for 6 different habitats in the western Himalaya (mean + standard 761 deviation). For statistical tests see table S9. Pairwise Tukey tests indicate open and 762 understory habitats are significantly brighter and darker, respectively, from birch, conifer, 763 oak and rhododendron, which are not different from each other. 764 5 6 7 8 9 BirchConiferOakOpenRhododendronUnderstoryHabitat Illuminance (lux, on log scale) HabitatBirchConiferOakOpenRhododendronUnderstory 90015003000500030000 5 6 7 8 9 BirchConiferOakOpenRhododendronUnderstoryHabitat Illuminance (lux, on log scale) HabitatBirchConiferOakOpenRhododendronUnderstory