bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

1 Acoustic community structure and seasonal turnover in tropical South Asian 2 Anand Krishnan1 3 4 5 6 1- Indian Institute of Science Education and Research (IISER) Pune, Pashan Road, Pune 7 411008, India. Email: [email protected] 8 9 10 Running title: Seasonal dynamics in tropical birdsong 11 Keywords: song, bioacoustics, seasonal dynamics, community turnover, phylogenetic 12 diversity, acoustic niche 13 14 15 16 17 18 19 Abstract 20 Birds produce diverse acoustic signals, with coexisting occupying distinct ‘acoustic 21 niches’ to minimize masking, resulting in overdispersion within acoustic space. In tropical 22 regions of the world, an influx of migrants from temperate regions occurs during winter. The 23 effects of these migrants on acoustic community structure and dynamics remain unstudied. 24 Here, I show that in a tropical urban bird community, the influx of winter migrants is 25 accompanied by a turnover of the acoustic community. However, in spite of this turnover, the 26 acoustic community remains overdispersed in acoustic niche space. The winter acoustic 27 community additionally exhibits lower frequency-band diversity, consistent with species singing 28 less continuously, as well as lower phylogenetic diversity. My data thus suggests that acoustic 29 niches and community structure are stable across seasons in spite of species turnover. 30 Migrants occupy similar regions of acoustic space as residents, and are relatively closely related 31 to some of these species. Their arrival therefore leads to greater phylogenetic clustering in the 32 winter, and thus lower phylogenetic diversity, although the acoustic community remains 33 overdispersed. Studying seasonal dynamics of acoustic communities thus provides valuable 34 insight into assembly processes, as well as a potential framework for long-term monitoring of 35 urban ecosystems. 36 37 38 39 40 41 42 43 44 bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

45 Introduction 46 The diverse communication functions of the songs of breeding birds include territorial 47 advertisement and attracting mates (1). In the more well-studied temperate regions of the world, 48 many birds sing during the spring and summer breeding season, migrating southward during the 49 winter months (2). Within communities of vocal species (or acoustic communities), simultaneous 50 vocalizations may mask each other, posing a barrier to efficient communication (3–6). As a 51 result, acoustic signals of a community of birds (as well as other diverse ) may exhibit 52 overdispersion (7–9), occupying distinct regions of acoustic parameter space (or “acoustic 53 niches”, for example singing at different times or frequencies) to minimize overlap (10–17). The 54 development and expansion of passive acoustic recording and monitoring techniques has 55 enabled study of the acoustic space of entire bird communities (10,18–20), however, much 56 remains to be understood about the dynamics of these communities. In particular, the influence 57 of seasonal dynamics (21,22) on singing activity and acoustic community structure has received 58 relatively little study. 59 Studies of the seasonal dynamics of bird song are particularly relevant in tropical landscapes, 60 which are home to the largest proportion of the world’s biodiversity. Owing to the high species 61 diversity of the tropics, and impending threats from deforestation and climate change, tropical 62 acoustic communities provide valuable insights informing both our understanding of behavioral 63 ecology and conservation (23). In Neotropical forest bird communities, signals of coexisting 64 birds may diverge as a result of competition, and the overall temporal patterns of singing activity 65 are driven by community composition (7,8). In addition to the highly diverse resident breeding 66 avifauna, tropical regions also receive a large influx of winter migrants from temperate regions, 67 which may increase the local species diversity (24,25). Many of these migrants remain highly 68 vocal on their wintering grounds, and their effect on the structure and dynamics of tropical 69 acoustic communities has not, to my knowledge, been studied. The influx of winter migrants into 70 acoustic niche space, many of which are close relatives of tropical resident species, could result 71 in more dense ‘packing’, and thus clustering in acoustic space (26,27). Alternatively, resident 72 species may drop out of the acoustic community (this means an absence of vocalizations, either 73 implying local movements or becoming silent), thus resulting in species turnover of the acoustic 74 community. In this latter scenario, overdispersion of the acoustic community, and thus acoustic 75 niche structure, may be predicted to persist within the winter. However, these patterns and 76 seasonal dynamics remain poorly known. 77 Here, I quantify seasonal community dynamics of avian vocal activity in an urban dry deciduous 78 scrub-grassland in peninsular India. This habitat is highly biodiverse, generally bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

79 underrepresented in studies of tropical ecology (28), and under severe threat due to extensive 80 destruction and land-use change (29). Using recordings of the acoustic community across wet 81 (monsoon) and dry (winter) seasons, I quantify whether the acoustic community exhibits 82 overdispersion consistent with the presence of acoustic niches, and whether the influx of winter 83 migrants alters this community structure (i.e. the distribution of species within acoustic space) or 84 instead results in turnover of the acoustic community. Finally, I examine whether the arrival of 85 migrants alters the phylogenetic (and thus species) diversity of the acoustic communities. 86 Because scrub habitats possess a varied and conspicuous avifauna, most of which breed in the 87 monsoon (30), and also receives a large influx of winter migrants (31), studying this habitat 88 provides valuable insight into seasonal patterns in biodiversity. 89 90 Materials and Methods 91 Study Site 92 I conducted this study in the Vetal Tekdi Biodiversity Park, a small (<80ha) remnant patch of 93 tropical dry-deciduous scrub-grassland mosaic within the city of Pune, Maharashtra, India. The 94 typical vegetation of this area includes native vegetation such as Acacia and Anogeissus, and a 95 largely open savanna-type habitat structure with a grassy understory (29) (Figure 1). The study 96 site is a public park at the top of a hill at about 790m asl. The slopes of this hill have been 97 planted over by non-native trees, and native vegetation now only remains in higher areas. This 98 habitat exhibits pronounced seasonality, with the Southwest monsoons between June and 99 September, and also undergoes annual burning of grass (32,33). Using multiple microphones, it 100 was possible to collect simultaneous and comprehensive information on bird vocalizations 101 across the landscape. Additionally, it was possible to identify vocalizing species and assign calls 102 to them relatively easily in this open habitat, which is key to identifying species vocalizations in 103 large passively recorded datasets. The avifauna of this park includes several species 104 characteristic of scrub landscapes across India, for example Francolinus sp, Lanius sp and 105 Pericrocotus erythropygius (31,34) (see Supplementary Data), and may thus be considered 106 representative of these habitats. 107 108 Recordings 109 In order to quantify seasonal changes in bird singing activity during the dawn chorus, I 110 conducted recordings during August 2017 (the monsoon breeding season for many resident 111 birds), and November-December 2017 (winter migrant season). The recording apparatus I used 112 consisted of four Sennheiser (Wedemark, Germany) ME62 omnidirectional microphones bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

113 connected to a Zoom H6 (Tokyo, Japan) portable recorder for simultaneous, synchronized 114 recordings. The four microphones were connected to the recorder using 10m long XLR cables, 115 each leading off to the north, south, east and west of the recorder. The microphones were then 116 placed on the ground to record singing activity. Using four sensitive, low-noise omnidirectional 117 microphones enabled me to maximize detection of high-frequency species with soft calls, and 118 thus simultaneous, comprehensive coverage of all bird species vocalizing within this landscape. 119 Additionally, this allowed me to minimize differences in detection due to microphone placement, 120 or due to differences in temperature and humidity across seasons, because four microphones 121 increase the likelihood of detecting species whose signals might be affected by these factors. 122 Because I did not observe spatial heterogeneity in singing activity within this rather open habitat, 123 I performed pooled analyses of all four microphone channels except where otherwise indicated. 124 On each sampling day, I recorded 45 minutes of bird singing activity between 630 and 730AM 125 (the hour of sunrise, referred to as the early morning recording), and another 45 minutes 126 between 730 and 830AM (referred to as the late morning recording) (broadly following a similar 127 sampling strategy to (8)). In total, the primary dataset consisted of six sampling days in the 128 monsoon and six in the winter. At 90 minutes a day, this resulted in a total of 18 sampling hours 129 of bird song activity, or 72 hours of audio data in total across four microphones. Before and after 130 the data collection periods, I made recordings and surveys of bird vocalizations at the site using 131 a single microphone, both to aid identification of calls detected during the census and as a 132 reference library to calculate call parameters. 133 134 Species activity and seasonal turnover 135 To determine relative activity patterns of each species in the acoustic community across 136 seasons, I divided each 45-minute recording into five-minute segments. Within each five-minute 137 segment, I identified all species of vocalizing bird across all four microphone channels (pooled) 138 both by ear and by spectrographic visualization of calls in Raven Pro 1.5 (Cornell Laboratory of 139 Ornithology, Ithaca, NY, USA). Most bird species in this habitat are familiar and common Indian 140 birds, including in urban environments, and are thus easily identifiable by voice, as well as by 141 comparison to reference recordings I made earlier. Therefore, although there is a remote 142 possibility of mis-detection in any census using a human observer, this is very unlikely to alter 143 patterns of presence-absence, particularly of common species. I constructed a presence- 144 absence matrix where a 1 indicated presence of a species’ vocalizations and 0 their absence in 145 each 5-minute sample. Once this had been carried out for the entire dataset (12 45-minute 146 recordings per season, and four microphone channels per recording), I proceeded to examine bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

147 seasonal patterns for species that were present in 10% or more of total 5-minute samples in 148 each season (8) (which I henceforth refer to as the acoustic community), to avoid bias 149 introduced by vocally “rare” species. For these species, I determined what I henceforth refer to 150 as an acoustic “abundance index”. In order to account for non-independence of consecutive 5- 151 minute samples, I drew one 5-minute sample at random for each species from each 45-minute 152 recording (thus a total of 12 per species per season, six from early morning recordings and six 153 from late morning recordings), and calculated the percentage of 1’s (or presences) in this 154 random draw (schematic of workflow in Figure 2A). The process was repeated 10000 times for 155 each species for each season. The average percentage of presences across 10000 random 156 draws is the acoustic abundance index, which represents the probability of detecting a species’ 157 vocalizations in a randomly-drawn 5-minute acoustic sample. A species that is vocally 158 ubiquitous would score close to a 1 in abundance index, and a rarer species would score a low 159 value. This avoids potential bias introduced by a species being highly vocal on one day and 160 silent on another, by ensuring that samples from each recording day are given equal weight. To 161 calculate species turnover and diversity accounting for abundance, I calculated Jaccard 162 similarity indices between the two communities, and Shannon-Weiner alpha-diversity indices for 163 both monsoon and winter acoustic communities using the vegan (35) package in R (36). 164 165 Acoustic space and community structure 166 To further examine seasonal patterns in the diversity of acoustic communities, I used Raven Pro 167 to identify clean examples of calls for each of the species present in the monsoon and winter 168 acoustic communities. I used recordings made outside of the study period from the same site 169 wherever clean examples were not present within the recordings, and for three species, 170 supplemented this with recordings made from peninsular India (as close to Pune as possible, 171 analyses suggested no differences in the parameters calculated) and archived on the Xeno- 172 canto (https://www.xeno-canto.org) and AVoCet (http://avocet.zoology.msu.edu) bird song 173 databases. Using these vocalizations (at least 10 vocalizations per species where possible), I 174 calculated six temporal and frequency parameters in Raven Pro: average peak frequency, 175 maximum and minimum peak frequency, frequency bandwidth (90%), note duration and relative 176 time of peak frequency. I performed a principal components analysis on the correlation matrix of 177 this data to reduce dimensionality. To compare acoustic niche space and structure across 178 seasons, I performed a MANOVA using the manova1 function in MATLAB (Mathworks, Inc, 179 Natick, MA, USA), and one-sample Kolmogorov-Smirnov tests to test for overdispersion of each 180 seasonal community compared to a uniform distribution (9). bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

181 182 Acoustic diversity indices 183 Using the first 20 minutes of each 45-minute recording sample (making sure they were free of 184 rain or loud anthropogenic noise), I calculated three indices of acoustic or frequency band 185 diversity (20) using the soundecology (37) package in R to determine overall levels of singing 186 activity. These were the acoustic diversity index (ADI), acoustic evenness index (AEI, where 187 lower values indicate more spread across the frequency spectrum, and thus more evenness), 188 and the bioacoustic index (BI), a measure of acoustic diversity that is more robust against 189 abiotic noise (20,38,39). To quantify changes in singing activity for 10 common species across 190 seasons, I selected the first five minutes from each of six 45-minute recordings for each season 191 (12 in total, free from anthropogenic or weather-related noise). I determined the total percentage 192 of time spent vocalizing across seasons from these five-minute samples (Supplementary Data). 193 194 Phylogenetic diversity indices 195 Finally, to estimate the phylogenetic diversity of the acoustic communities across seasons as a 196 comparison to their species diversity and seasonal dynamics, I downloaded a phylogeny pruned 197 to contain all the 53 species (including shared species) across both communities from the avian 198 Tree of Life (40). This meta-tree provides 100 different possible phylogenetic hypotheses of the 199 relationships between selected species, and all indices calculated were estimated for each of 200 these 100 trees, thus giving a distribution of 100 index values. I further pruned these trees down 201 from 53 to 40 and 43 species respectively, to calculate separate indices for monsoon and winter 202 acoustic communities. Using these values and the abundance indices for each species, I 203 calculated three commonly measured indices of phylogenetic diversity using the picante (41) 204 package in R: Faith’s phylogenetic diversity (PD) (42), the Mean pairwise phylogenetic distance 205 between species in a community (MPD), and the mean nearest-neighbor phylogenetic distance 206 between species (MNTD) (43) (100 values for each). For the last two, I calculated both raw 207 values and abundance-weighted diversities for both monsoon and winter acoustic communities. 208 To compare phylogenetic diversity values across seasonal acoustic communities, I used paired 209 statistical tests in MATLAB, as each pair of values were calculated using one of the 100 210 possible phylogenetic trees. For consistency with the phylogenetic analyses, I follow the 211 adopted in this tree throughout. Because the acoustic community here is composed 212 of species across multiple families, the broad tree topology does not change much across the 213 100 possible phylogenies. The distributions of index values reported here are thus robust for 214 broad comparisons across clades. bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

215 216 Results 217 Species diversity of the avian acoustic community 218 Across the entire dataset, I identified vocalizations of 85 bird species (list of species in 219 Supplementary Data) (44). Of these, 56 species were recorded during the monsoon sampling 220 and 70 species were recorded during the winter, suggesting higher overall species richness 221 during the winter (Figure 1). During the monsoon, early and late morning censuses both 222 recorded an average of 39 species (rounded off) respectively (differences between early and 223 late morning were not statistically significant in a paired t-test: t=0.2571, dF=5, p=0.8074), 224 whereas winter censuses recorded an average of 41 and 44 species in the early and late 225 morning respectively (again not significantly different from each other: t=1.4, dF=5, p=0.2204), 226 per census. Across seasons, I performed unpaired t-tests on early and late morning censuses 227 respectively to determine if more species were recorded in winter. Early morning censuses 228 detected similar numbers of species across seasons (t=1.2677, dF=10, p=0.2336). However, 229 late morning censuses in winter detected significantly more species than in the monsoon 230 (t=2.6451, dF=10, p=0.0245), providing some support for higher overall species richness in the 231 winter. For abundance analyses of the acoustic communities (Figure 2A), however, I selected 232 only species present in >10% of 5-minute samples in each season (see Methods). This resulted 233 in estimates of abundance indices for 40 species in the monsoon and 43 species in the winter 234 (therefore, a total of 53 species in subsequent phylogenetic analyses). Thus, although overall 235 species diversity was higher in the winter, the numbers of regularly vocalizing species 236 comprising the acoustic community were comparable across seasons. 237 238 Seasonal turnover of the acoustic community 239 The distributions of abundance indices for 40 bird species comprising the monsoon acoustic 240 community and 43 bird species comprising the winter acoustic community are shown in Figure 241 2B and 2C. Although the number of species in both communities were comparable, this pattern 242 belies a considerable degree of turnover in species composition across seasons. In all, 30 243 species were shared between the two seasonal communities, resulting in a Jaccard similarity of 244 0.56. However, if we only consider species above the mean abundance index for each season 245 (roughly 0.5 in both seasons), Jaccard similarity dropped to 0.41, suggesting that 59% of the 246 most abundant species exhibited seasonal changes in acoustic abundance. This is apparent 247 from Figure 2D and 2E, where a number of frequently recorded species (in red) decreased 248 abundance by more than one standard deviation in the winter, whereas others (in blue) bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

249 increased by more than one standard deviation. I detected resident species such as Francolinus 250 pictus, Cuculus varius, Cacomantis passerinus and Ploceus philippinus frequently in the 251 monsoon but not at all in the winter, whereas others such as Stigmatopelia senegalensis, Pavo 252 cristatus and Eudynamys scolopaceus were detected considerably less frequently in the winter. 253 Four monsoon species with low abundance indices (Vanellus indicus, Stigmatopelia chinensis, 254 Hirundo concolor and Psittacula cyanocephala) were either absent or detected too infrequently 255 in the winter to determine an abundance index. I also list them here under species that 256 decrease abundance in the winter (red dashed lines in Figure 2D). Species increasing in 257 abundance index in the winter included residents such as Saxicoloides fulicatus and Lonchura 258 punctulata that were also recorded during the monsoon, other residents such as Lanius schach 259 and Chrysomma sinense that were not, and long-distance migrants such as Phylloscopus 260 trochiloides, Phylloscopus humei and Ficedula parva. Several other species, including the 261 residents Copsychus saularis and Lonchura malabarica, and the migrants Sylvia curruca, 262 Anthus trivialis and Anthus campestris, had relatively low winter abundance indices but were 263 either unrecorded or recorded too infrequently to determine an index in the monsoon. They are 264 therefore listed here as species that increase abundance in the winter (blue dashed lines in 265 Figure 2D). Shannon-Weiner diversity indices that account for the abundance (45) (in this case, 266 the abundance indices) of species were comparable for both acoustic communities, at 3.541 for 267 the monsoon acoustic community and 3.559 for the winter, indicating relative stability in acoustic 268 community diversity (i.e. regularly detected species) despite the turnover in species 269 composition. 270 271 Monsoon and winter acoustic communities are overdispersed in acoustic space 272 The first two principal components of 6 acoustic parameters explained approximately 74% of 273 variation (eigenvalues of 3.35 and 1.07 respectively, see Supplementary Data for values). PC1 274 exhibited the strongest positive loadings on frequency parameters and weak negative loadings 275 on temporal parameters, whereas PC2 loaded strongly and positively on relative time of peak, 276 and to a lesser extent on note duration. The results of multivariate statistical analysis on these 277 principal components indicate that the overall acoustic parameter (or niche) space occupied by 278 vocalizing species was not significantly different across seasons (Figure 3). MANOVA tests both 279 on the measured parameters and their first three principal components failed to reject the null 280 hypothesis that vocalizations of both monsoon and winter acoustic communities were drawn 281 from the same distribution (On parameters: Wilk’s lambda=0.9331, dF within groups=81, dF 282 between groups-1, total dF=82, p=0.4933, On PCs: Wilk’s lambda=0.9431, p=0.0959). Further, bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

283 both communities were statistically indistinguishable from a uniform distribution spanning the 284 same range of values in PC1 space (representing the bulk of variation in frequency), which 285 suggests that simultaneously vocalizing birds were overdispersed across acoustic parameter 286 space (One-sample Kolmogorov-Smirnov tests, monsoon: D=0.1268, p=0.5007; winter: D= 287 0.1673, p=0.1606). This is consistent with species occupying distinct acoustic niches (8,9,46), 288 and with this niche structure remaining stable across seasons, even with the arrival of winter 289 migrants. 290 291 Lower frequency band diversity in the winter 292 Next, I investigated whether turnover and change in species composition was represented by 293 acoustic diversity indices that quantify the relative amount of energy in the frequency band 294 spectrum, and are thus representative of overall levels of singing activity (Figure 4, also see 295 Supplementary Data). All three measures indicate that the monsoon community exhibited 296 greater diversity and evenness than the winter community (Average ADI: Monsoon: 1.4, Winter: 297 0.59, Average AEI: Monsoon, 0.65, Winter: 0.85, lower values indicate greater evenness, 298 Average BI: Monsoon: 26.65, Winter: 17.11, Wilcoxon signed-rank test, ADI and BI: W=78, 299 N=12/group, P<0.001, AEI: W=0, P<0.001). These values were consistent within each season 300 across recording days, which further indicates robustness across seasons, and that the lower 301 acoustic diversity and evenness observed in winter is genuine. An examination of the relative 302 energy across frequency bands from 0-10 KHz reveals that this reduction appears to be driven 303 by a reduction in activity between roughly 3-7KHz, indicating lower singing activity in the 304 winter(47) (Figure 4B). This suggests two possibilities or a combination of both: firstly, that the 305 species dropping out of the acoustic community in the winter were highly vocal during the 306 monsoon, and their absence drove the lower acoustic indices. Indeed, several species 307 mentioned earlier, with high monsoon abundance indices, were detected less frequently or were 308 absent during the winter. A second contributing factor may be that species that did not change 309 in abundance (and were therefore still frequently vocal) may have been singing shorter bouts 310 outside of their breeding season, reducing the amount of sound in those frequency bands. 311 To test this second hypothesis, I examined changes in percentage of time spent singing for 10 312 such species over 6 five-minute samples recorded on different days (including all types of 313 vocalizations). For seven of these species, the differences were not statistically significant, but 314 three showed statistically significant (2.59

317 abundance indices across seasons, thus suggesting that they, together with the previously 318 mentioned monsoon vocalizing species, drove the changes in acoustic diversity indices 319 observed here. 320 321 Lower phylogenetic diversity in the winter acoustic community 322 Using a publicly available phylogeny (Figure 5A) (40), I estimated phylogenetic diversity to 323 understand whether the arrival of winter migrants and seasonal turnover influenced the 324 phylogenetic structure of acoustic communities. Three measures of phylogenetic diversity (PD, 325 MPD and MNTD) were all lower for the winter community than for the monsoon (Averages: PD: 326 Monsoon: 1544.83, Winter: 1510,16, MPD: Monsoon: 139.29, Winter: 118.28, MNTD: Monsoon: 327 50.24, Winter: 48.9). Weighting MPD and MNTD for abundance resulted in slightly lower 328 diversity values across communities but the seasonal trend remained the same (Averages: 329 MPD: Monsoon: 130.92, Winter: 102.13, MNTD: Monsoon: 45.32, Winter: 41.27) (Figure 5B, 330 5C, also see Supplementary Data). This effect was consistent and statistically significant across 331 all 100 possible phylogenetic trees (PD: paired t-test: t=18.7524, dF=99, p<0.001, effect 332 size=0.38, MPD unweighted: t= 144.18, dF=99, p<0.001, effect size= 3.22, MPD weighted: 333 t=133.44, dF=99, p<0.001, effect size=4.75, MNTD unweighted: t=12.19, dF=99, p<0.001, effect 334 size=0.38, MNTD weighted: t=35.48, dF=99, p<0.001, effect size=1.17), with larger effect sizes 335 after weighting MPD and MNTD for abundance. Thus, although species diversity was 336 comparable in the winter (43 species versus 40 in the monsoon), phylogenetic diversity was 337 lower, indicating more clustering at certain points in the phylogenetic tree of the community. 338 339 340 Discussion 341 To summarize, my study of the avian acoustic community in an urban tropical scrub-grassland 342 habitat uncovers overdispersion of acoustic signals, consistent with the presence of acoustic 343 niches. Although there is considerable seasonal turnover in the species composition of the 344 acoustic community, this overdispersion is maintained across seasons. Additionally, overall use 345 of frequency band space is lower in the winter, putatively driven by a relative lack of continuous 346 vocalizations outside the breeding season (even though detection probabilities of at least some 347 species do not change). Finally, although species diversity in these seasonal acoustic 348 communities is comparable (and late morning winter recordings detected more bird species on 349 average), phylogenetic diversity of the winter acoustic community is actually lower than that of 350 the monsoon. In spite of this increased phylogenetic clustering, species remain overdispersed in bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

351 acoustic space. Taken together, the data indicate that migrants are close relatives of resident 352 species, and fit into the same acoustic niche space, resulting in a stable, overdispersed acoustic 353 community structure across seasons. 354 355 Bioacoustics and the seasonal dynamics of tropical avian communities 356 Species in the tropics exhibit considerable seasonal activity changes and local movements, 357 particularly where differences between dry and wet seasons are more pronounced. In semiarid 358 regions and savanna ecotopes of Asia (29), the habitat is relatively open and seasonally dry 359 (32,33). Most resident birds in this habitat breed during the southwest monsoon, and are very 360 vocal between June and October (31). During the winter season, many of the monsoon 361 community’s most vocal resident (‘resident’ is used here to imply birds that do not undertake 362 long-distance migrations) birds fell completely silent, and were also not observed visually during 363 recordings or otherwise (for example, Cuculus varius, Cacomantis passerinus, Francolinus 364 pictus and Ploceus philippinus), indicating possible local movements out of the study area. On 365 the other hand, some resident birds (Lanius schach and Chrysomma sinense, for example) 366 were recorded during the winter only, suggesting local movements into the study area. Other 367 monsoon-breeding birds (Pavo cristatus and Eudynamys scolopaceus) were recorded during 368 the winter and seen frequently, but were much less vocal. The converse was true of 369 Saxicoloides fulicatus, which was detected more frequently in the winter. For these latter 370 species, this may indicate either some local movement, relative silence during one season, or a 371 combination of both. Local movements may be driven by seasonal resources, for example, the 372 seeding of grass (in the case of seed-eaters such as Lonchura punctulata and Lonchura 373 malabarica) (30,31,34). 374 In addition, several long-distance migrants were commonly detected in winter and frequently 375 vocal, including Acrocephalus dumetorum and Hippolais rama (identified visually; it was not 376 possible to distinguish these species acoustically, and they have been treated as a single unit 377 for calculating abundance indices), Phylloscopus trochiloides, Phylloscopus humei, and 378 Ficedula parva. Thus, a combination of local movements, changes in singing activity and the 379 arrival of winter migrants appears to drive species turnover in the avian acoustic community. 380 This turnover, and the frequent detection of common winter migrants, suggests that acoustic 381 monitoring may provide a valuable tool to detect and monitor the annual influx of winter 382 migrants, as well as to elucidate patterns of local movement in resident species (10,48–50). 383 These local movements remain poorly understood for many bird species, including relatively bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

384 common ones. Using acoustic methods may provide valuable insight into their movement 385 ecology and behavioral dynamics. 386 387 Stable acoustic community structure across seasons 388 In spite of the seasonal turnover in species composition, my data suggest that the overall niche 389 structure of bird vocalizations within the community does not change, and the winter community 390 remains overdispersed in acoustic space. This suggests that community structure within 391 acoustic space is stable across seasons, and that long-distance migrants fit into the same 392 acoustic space occupied by resident birds (9,51). This stability across seasons may be a result 393 of migrant species occupying acoustic niches left vacant by the local movements or silence of 394 resident birds. However, the overall use of higher-frequency bands declines in the winter, and 395 indices of frequency-band diversity suggest lower overall singing activity. In addition to the 396 absence of a number of vocal monsoon-breeders, I present preliminary evidence that several of 397 the most vocal resident species (in particular, Cisticola and Prinia warblers), sing for a greater 398 percentage of time in the monsoon breeding season. A number of birds are known to exhibit 399 higher singing activity coinciding with the breeding season, correlated with hormonal changes 400 such as an increase in male testosterone levels (52). Thus, although the acoustic community 401 remains overdispersed in the winter, niche separation between simultaneously vocalizing 402 species is likely to be of greater importance in the monsoon breeding season, when their 403 vocalizations are more likely to overlap. However, the abundance indices of these species do 404 not exhibit a seasonal change. Therefore, the use of presence-absence abundance indices over 405 5-minute time blocks may prove useful in quantifying relative abundance regardless of seasonal 406 changes in vocalization. These seasonal changes may, however, result in different acoustic 407 diversity and frequency-band index values even though species diversity is comparable. 408 Acoustic indices (48,49) should therefore be combined with abundance indices to understand 409 community-level seasonal changes in singing activity, at least in tropical bird communities. 410 411 Acoustic niches in tropical bird communities 412 The acoustic niche hypothesis states that in order to minimize masking or the overlap of 413 sounds, species in a habitat or community may exhibit acoustic signals occupying distinct 414 regions of acoustic space to minimize overlap, and thus masking (10,11,14). In terms of 415 community structure, this is predicted to result in overdispersion of acoustic signals such that 416 they approximate a uniform distribution across acoustic parameter space. Multiple studies have 417 found evidence both for and against this hypothesis in diverse organisms (8,9,11–13,51,53). I bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

418 find that both monsoon and winter communities are overdispersed in frequency space (PC1 of 419 measured acoustic signal parameters), which accounts for the largest proportion of variation in 420 acoustic signals. This suggests that acoustic niche structure in bird communities is stable 421 across seasons, although, as mentioned above, niche separation may be of greater importance 422 in the monsoon when species vocalize in longer, continuous bouts. 423 Most migrants recorded during this study, for example, warblers and flycatchers, belong to 424 families that are also represented by resident species (44). All indices of phylogenetic diversity 425 are significantly lower in the winter, even though species diversity is comparable or slightly 426 higher. This indicates that the influx of winter migrants results in more phylogenetic clustering 427 within the acoustic community, and concomitant species turnover while still remaining 428 overdispersed in acoustic space. It is possible that maintaining an overdispersed acoustic niche 429 structure may support the coexistence of multiple closely related species during the non- 430 breeding season. This, however, requires a detailed behavioral study on focal species to 431 address. I note here that my study focuses only on the avian acoustic community within a single 432 urban habitat, to comprehensively understand their seasonal dynamics. However, it is likely that 433 these patterns are general to other habitats receiving an influx of winter migrants as well. 434 435 Conclusions 436 My study uncovers evidence that bird acoustic communities in tropical deciduous scrubland 437 exhibit distinct niches in acoustic space, such that signals of coexisting species are 438 overdispersed. This acoustic niche structure is stable across seasons to the influx of local and 439 winter migrants, although species composition exhibits turnover within this space, consistent 440 with migrants occupying a similar acoustic space to resident species. Finally, phylogenetic 441 diversity declines in the winter, suggesting that the arrival of migrants results in more 442 phylogenetically clustered communities. A reduction in singing (a proxy of territoriality) by 443 resident birds, and maintaining an overdispersed community structure, may enable their 444 coexistence with multiple closely related migrants, a topic requiring further study. Knowledge of 445 acoustic space and turnover patterns is invaluable to comprehensive assessment, in particular 446 in urban habitats of peninsular India, which are undergoing rapid land-use change. Acoustic 447 monitoring thus serves as a framework for long-term, non-invasive study of the responses of 448 avian communities to land-use and climate change (21,22,49,50). 449 450 451 bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

452 Acknowledgments 453 I thank Rohit Chakravarty and Rashim Malhotra for help with data collection, Raghav Rajan and 454 his lab group for discussions and feedback, Samira Agnihotri and Krishnapriya Tamma for 455 helpful discussions. 456 457 Funding 458 My research is funded by an INSPIRE Faculty Award from the Department of Science and 459 Technology, Government of India and an Early Career Research (ECR/2017/001527) Grant 460 from the Science and Engineering Research Board (SERB), Government of India. 461 462 References 463 1. Bradbury JW, Vehrencamp SL. Principles of Communication. 2nd ed. Sinauer 464 Associates; 2011. 465 2. Catchpole CK, Slater PJB. Bird Song: Biological themes and Variations. 2nd Ed. 466 Cambridge, UK: Cambridge University Press; 2008. 467 3. Bee M a. Finding a mate at a cocktail party: spatial release from masking improves 468 acoustic mate recognition in grey treefrogs. Anim Behav. 2008;75:1781–91. 469 4. Bee MA, Micheyl C. The “Cocktail Party Problem”: What is it? How can it be solved? And 470 why should animal behaviorists study it? J Comp Psychol. 2008;122(3):235–51. 471 5. Hart PJ, Hall R, Ray W, Beck A, Zook J. Cicadas impact bird communication in a noisy 472 tropical rainforest. Behav Ecol. 2015;26(3):839–42. 473 6. Wong S, Parada H, Narins PM. Heterospecific acoustic interference: Effects on calling in 474 the frog Oophaga pumilio in Nicaragua. Biotropica. 2009;41(1):74–80. 475 7. Luther DA. Signaller: receiver coordination and the timing of communication in 476 Amazonian birds. Biol Lett [Internet]. 2008;4(6):651–4. Available from: 477 http://rsbl.royalsocietypublishing.org/cgi/doi/10.1098/rsbl.2008.0406 478 8. Luther D. The influence of the acoustic community on songs of birds in a neotropical rain 479 forest. Behav Ecol. 2009;20(4):864–71. 480 9. Cardoso GC, Price TD. Community convergence in bird song. Evol Ecol. 481 2010;24(2):447–61. 482 10. Farina A, Lattanzi E, Malavasi R, Pieretti N, Piccioli L. Avian soundscapes and cognitive 483 landscapes: Theory, application and ecological perspectives. Landsc Ecol. 484 2011;26(9):1257–67. 485 11. Henry CS, Wells MM. Acoustic niche partitioning in two cryptic sibling species of bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

486 Chrysoperla green lacewings that must duet before mating. Anim Behav [Internet]. 487 Elsevier Ltd; 2010;80(6):991–1003. Available from: 488 http://dx.doi.org/10.1016/j.anbehav.2010.08.021 489 12. Schmidt AKD, Römer H, Riede K. Spectral niche segregation and community 490 organization in a tropical cricket assemblage. Behav Ecol. 2013;24(October):470–80. 491 13. Krishnan A, Tamma K. Divergent morphological and acoustic traits in sympatric 492 communities of Asian barbets. R Soc Open Sci [Internet]. 2016;3(8):160117. Available 493 from: http://rsos.royalsocietypublishing.org/lookup/doi/10.1098/rsos.160117 494 14. Krause BL. Wild Soundscapes: discovering the voice of the natural world. Berkeley, CA, 495 USA: Wild Sanctuary Books; 2002. 496 15. Fleischer RC, Boarman WI, Cody ML. Asynchrony of song series in the Bewick’s wren 497 and wrentit. Anim Behav. 1985;33(1969):674–6. 498 16. Brumm H. Signalling through acoustic windows: Nightingales avoid interspecific 499 competition by short-term adjustment of song timing. J Comp Physiol A Neuroethol 500 Sensory, Neural, Behav Physiol. 2006;192:1279–85. 501 17. Popp J, Ficken R, Reinartz J. Short-Term Temporal Avoidance of Interspecific Acoustic 502 Interference among Forest Birds. Auk [Internet]. 1985;102(4):744–8. Available from: 503 http://www.jstor.org/stable/10.2307/4086711 504 18. Pijanowski BC, Farina A, Gage SH, Dumyahn SL, Krause BL. What is soundscape 505 ecology? An introduction and overview of an emerging new science. Landsc Ecol. 506 2011;26(9):1213–32. 507 19. Gasc A, Francomano D, Dunning JB, Pijanowski BC. Future directions for soundscape 508 ecology: The importance of ornithological contributions. Auk [Internet]. 2017;134(1):215– 509 28. Available from: http://www.bioone.org/doi/10.1642/AUK-16-124.1 510 20. Pijanowski BC, Villanueva-Rivera LJ, Dumyahn SL, Farina A, Krause BL, Napoletano 511 BM, et al. Soundscape Ecology: The Science of Sound in the Landscape. Bioscience. 512 2011;61(3):203–16. 513 21. Buxton RT, Brown E, Sharman L, Gabriele CM, McKenna MF. Using bioacoustics to 514 examine shifts in phenology. Ecol Evol. 2016;6(14):4697–710. 515 22. Sanders CE, Mennill DJ. Acoustic monitoring of nocturnally migrating birds accurately 516 assesses the timing and magnitude of migration through the Great Lakes. Condor 517 [Internet]. 2014;116(3):371–83. Available from: 518 http://www.bioone.org/doi/10.1650/CONDOR-13-098.1 519 23. Du Toit JT, Walker BH, Campbell BM. Conserving tropical nature: Current challenges for bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

520 ecologists. Trends Ecol Evol. 2004;19(1):12–7. 521 24. Kimura K, Yumoto T, Kikuzawa K. Fruiting phenology of fleshy-fruited plants and 522 seasonal dynamics of frugivorous birds in four vegetation zones on Mt. Kinabalu, Borneo. 523 J Trop Ecol. 2001;17(6):833–58. 524 25. Katuwal HB, Basnet K, Khanal B, Devkota S, Rai SK, Gajurel JP, et al. Seasonal 525 changes in bird species and feeding guilds along elevational gradients of the Central 526 Himalayas, Nepal. PLoS One. 2016;11(7):1–17. 527 26. Cavender-Bares J, Ackerly DD, Baum DA, Bazzaz FA. Phylogenetic overdispersion in 528 Floridian oak communities. Am Nat [Internet]. 2004;163(6):823–43. Available from: 529 http://www.ncbi.nlm.nih.gov/pubmed/15266381 530 27. Pigot AL, Trisos CH, Tobias JA. Functional traits reveal the expansion and packing of 531 ecological niche space underlying an elevational diversity gradient in birds. 532 Proc R Soc B Biol Sci [Internet]. 2016;283(1822):20152013. Available from: 533 http://rspb.royalsocietypublishing.org/content/283/1822/20152013 534 28. Burke A. Conserving tropical biodiversity: the arid end of the scale. Trends Ecol Evol. 535 2004;19(5):225–6. 536 29. Ratnam J, Tomlinson KW, Rasquinha DN, Sankaran M. Savannahs of Asia: Antiquity, 537 biogeography, and an uncertain future. Philos Trans R Soc B Biol Sci. 2016;371(1703). 538 30. Ali S, Ripley SD. Handbook of the Birds of India and Pakistan. 2nd Editio. Oxford 539 University Press; 1999. 540 31. Rasmussen PC, Anderton JC. Birds of South Asia: The Ripley Guide. Lynx Edicions, 541 Barcelona; 2005. 542 32. Nerlekar AN, Kulkarni DK. The Vetal Hills: an Urban Wildscape in Peril. Taprobanica 543 [Internet]. 2015;07(02):72–8. Available from: 544 https://www.researchgate.net/profile/Ashish_Nerlekar2/publication/283344427_THE_VET 545 AL_HILLS_AN_URBAN_WILDSCAPE_IN_PERIL/links/5634d22f08aeb786b702c0f9.pdf 546 33. Punalekar S, Mahajan DM, Kulkarni DK. Impact of exotic tree species on the native 547 vegetation of Vetal Hill, Pune. Indian J For. 2010;33(4):549–54. 548 34. Grimmett R, Inskipp C, Inskipp T. Birds of the Indian Subcontinent. 1st Editio. A&C Black; 549 1998. 550 35. Jari Oksanen, F. Guillaume Blanchet, Michael Friendly, Roeland Kindt, Pierre Legendre, 551 Dan McGlinn, Peter R. Minchin, R. B. O’Hara, Gavin L. Simpson, Peter Solymos, M. 552 Henry H. Stevens, Eduard Szoecs HW. vegan: Community Ecology Package. R package 553 version 2.5-2. https://CRAN.R-project.org/package=vegan. 2018. bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

554 36. R Core Team . R: A language and environment for statistical computing. [Internet]. 555 Vienna, Austria: R Foundation for Statistical Computing,; 2013. Available from: 556 http://www.r-project.org/ 557 37. Villanueva-Rivera LJ, Pijanowski BC. soundecology: Soundscape Ecology. R package 558 version 1.3.3. https://CRAN.R-project.org/package=soundecology. 559 38. Boelman NT, Asner GP, Hart JP, Martin RE. Multi-trophic invasion resistance in Hawaii: 560 Bioacoustics, field surveys and airborne remote sensing. Ecol Appl. 2007;17(8):2137–44. 561 39. Villanueva-Rivera LJ, Pijanowski BC, Doucette J, Pekin B. A primer of acoustic analysis 562 for landscape ecologists. Landsc Ecol. 2011;26(9):1233–46. 563 40. Jetz W, Thomas GH, Joy JB, Hartmann K, Mooers a. O. The global diversity of birds in 564 space and time. Nature. 2012;491(7424):444–8. 565 41. Kembel SW, Cowan PD, Helmus MR, Cornwell WK, Morlon H, Ackerly DD, et al. Picante: 566 R tools for integrating phylogenies and ecology. Bioinformatics. 2010;26(11):1463–4. 567 42. Faith. D.P. Conservation evaluation and phylogenetic diversity. Biol Conserv [Internet]. 568 1991;61:1–10. Available from: http://schistocerca.org/BSC5937/PDF/Faith 1992 569 (Biological Conservation).pdf 570 43. Webb CO, Ackerly DD, McPeek MA, Donoghue MJ. Phylogenies and Community 571 Ecology. Annu Rev Ecol Syst. 2002;33:475–505. 572 44. del Hoyo J, Collar NJ, Christie DA, Elliott A, Fishpool LDC. HBW and BirdLife 573 International Illustrated Checklist of the Birds of the World. Lynx Edicions BirdLife 574 International; 2014. 575 45. Hill MO. Diversity and evenness: A unifying notation and its consequences. Ecology. 576 1973;54(2):427–32. 577 46. Chek A a., Bogart JP, Lougheed SC. Mating signal partitioning in multi-species 578 assemblages: A null model test using frogs. Ecol Lett. 2003;6:235–47. 579 47. Farina A, Ceraulo M, Bobryk C, Pieretti N, Quinci E, Lattanzi E. Spatial and temporal 580 variation of bird dawn chorus and successive acoustic morning activity in a 581 Mediterranean landscape. Bioacoustics. 2015;24(3):269–88. 582 48. Buxton RT, Agnihotri S, Robin V V., Goel A, Balakrishnan R. Acoustic indices as rapid 583 indicators of avian diversity in different land-use types in an Indian biodiversity hotspot. J 584 Ecoacoustics [Internet]. 2018;2:1–17. Available from: 585 https://www.veruscript.com/a/GWPZVD/ 586 49. Buxton RT, McKenna MF, Clapp M, Meyer E, Stabenau E, Angeloni LM, et al. Efficacy of 587 extracting indices from large-scale acoustic recordings to monitor biodiversity. Conserv bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

588 Biol. 2018;32(5):1174–84. 589 50. Lellouch L, Pavoine S, Jiguet F, Glotin H, Sueur J. Monitoring temporal change of bird 590 communities with dissimilarity acoustic indices. Methods Ecol Evol. 2014;5(6):495–505. 591 51. Tobias JA, Planque R, Cram DL, Seddon N. Species interactions and the structure of 592 complex communication networks. Proc Natl Acad Sci [Internet]. 2014;111(3):1020–5. 593 Available from: http://www.pnas.org/cgi/doi/10.1073/pnas.1314337111 594 52. Ball GF, Castelino CB, Maney DL, Appeltants D, Balthazart J. The Activation of Birdsong 595 by Testosterone. Ann N Y Acad Sci [Internet]. 2003;1007(1):211–31. Available from: 596 http://onlinelibrary.wiley.com/doi/10.1196/annals.1286.021/abstract 597 53. Narins PM. Frog Communication. Sci Am. 1995;273:78–83. 598 599 Figure legends 600 Figure 1: The study site, Vetal Tekdi Biodiversity Park, and some birds found in the study area. 601 All images by Vivek Kannadi, Science Media Center, IISER Pune. The graph represents total 602 number of species recorded in each season, and each point represents the number of species 603 recorded in each individual 45- minute recording. 604 605 Figure 2: (A) Calculating abundance indices for each species per season. Each 45-minute 606 recording was broken down into 9 5-minute samples, and presence-absence of each species’ 607 vocalizations was indicated by a 1 or 0. One of these 9 samples was drawn at random from 608 each recording, resulting in 12 values per season (six early morning and six late morning 609 recordings). The percentage of 1’s in the resulting 12 values is the acoustic abundance index (in 610 this case 0.5). I repeated this random sampling procedure 10,000 times for each species in 611 each season to obtain the final average value of the abundance index. (B,C) Distributions of 612 abundance indices for 40 monsoon (A) and 43 winter (B) species comprising the respective 613 seasonal acoustic communities. (D,E) Turnover in species composition of the acoustic 614 community, represented by the seasonal change in abundance index. Species decreasing 615 abundance index in winter by >1 standard deviation are in red solid lines (D) and bold red text 616 (E), and species increasing by the same in winter are the blue solid lines (D) and bold blue text 617 (E). Species that do not exhibit this change are in black in both (D) and (E). Species 618 represented by dashed lines in (D) and regular blue or red text in (E) are those that were 619 detected often enough to calculate an abundance index (although low) in one season but not 620 the other. I also consider these species as exhibiting seasonal turnover. 621 bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

622 Figure 3: Biplot of first and second canonical variables from MANOVA on the principal 623 components of 6 acoustic parameters, for the monsoon (red) versus winter (blue) acoustic 624 communities. 625 626 Figure 4: Acoustic diversity indices decline in winter. (A) Acoustic diversity index declines in the 627 winter, indicating less diversity of frequency band space. (B) The amount of energy across 628 frequency bands suggests that this is driven by a decline in higher frequency activity in winter. 629 (C) The Acoustic Evenness Index is higher in winter, indicating lower evenness and more 630 clustering in frequency band space. (D) The Bioacoustic Index also supports a decline in higher 631 frequency bands, putatively driven by a decline in continuous singing activity (also see 632 Supplementary Figure). 633 634 Figure 5: Phylogenetic diversity of the winter acoustic community is lower. (A) Maximum clade 635 credibility consensus tree of the 100 possible trees used in the analysis, presented here for 636 illustrative purposes only. Although node support for the consensus tree is low at the tips, the 637 broad species relationships change very little across the 100 trees. Species are colored 638 according to Figure 2 to represent whether they increase or decrease abundance index in the 639 winter. (B) Faith’s PD (phylogenetic diversity) is lower in winter than in monsoon, thus lying 640 away from the dotted line, which represents equal values of both. (C) Plots of MNTD (squares) 641 and MPD (circles) for monsoon (red) and winter (blue) acoustic communities. The X axis 642 represents unweighted values, and the Y axis the values weighted for abundance. Weighting 643 reduces values of both indices, but the winter community exhibits consistently lower MNTD and 644 MPD than the monsoon community. bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

80 70 Early AM 60 Late AM 50 40 30

No. of species 20 10 0 Monsoon Winter bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Absolute change in abundance index across seasons A B D 1

15 0.9 1 0 0 1 1 0 1 0 0 Distributions of abundance indices: Monsoon 0.8

0.7 10 0.6 0.5

No. of species 5 0.4

1 x12 Abundance index 0.3 0.2 0 0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Abundance index 0

Monsoon Winter

1 0 0 1 1 1 0 0 0 0 1 1 C E Cisticola juncidis Pycnonotus cafer 15 Corvus macrorhynchos Distributions of abundance indices: Winter Francolinus pictus Prinia socialis Saxicoloides fulicatus Ploceus philippinus Prinia inornata Lonchura punctulata Pavo cristatus Prinia hodgsonii Anthus rufulus 10 Stigmatopelia senegalensis Prinia sylvatica Amandava amandava Acrocephalus dumetorum/ Zosterops palpebrosus 0.5 Orthotomus sutorius Cyornis tickelliae Hippolais rama Eudynamys scolopaceus Turdoides malcolmi Centropus sinensis Phylloscopus trochiloides Cuculus varius Pellorneum ruficeps Phylloscopus humei

No. of species Psittacula krameri 5 Cacomantis passerinus Psittacula eupatria Chrysomma sinense Pericrocotus cinnamomeus Acridotheres tristis/fuscus Lanius schach Aegithina tiphia Merops orientalis Ficedula parva Tephrodornis pondicerianus Copsychus saularis Nectarinia asiatica 0 Vanellus indicus Sylvia curruca Nectarinia zeylonica 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Stigmatopelia chinensis Anthus trivialis Repeat 10,000 times and take Francolinus pondicerianus Hirundo concolor Anthus campestris Abundance index Dendrocitta vagabunda the average for each species in each season Psittacula cyanocephala Lonchura malabraica Megalaima haemacephala Dicrurus macrocercus/leucophaeus Parus major bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

2

Monsoon 1.5 Winter

1

0.5

0

-0.5

-1 Second canonical variable (MANOVA) Second canonical -1.5

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 First canonical variable (MANOVA) bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.

Monsoon early AM B A Frequency band diversity Frequency band differences in acoustic activity 2 Monsoon late AM 1 Winter early AM Winter late AM 1.8 0.9 Monsoon early AM Monsoon late AM 1.6 0.8 Winter early AM Winter late AM 1.4 0.7 1.2 0.6 ADI 1 0.5 0.8 0.4 0.6 0.3 0.4 0.2 0.2 0.1 Proportion of signal above -50dBFS Proportion 0 0 0-1 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 1 2 3 4 5 6 Frequency band (KHz) Measurement day (1-6) C D Frequency band evenness Bioacoustic index 1 30 28 0.9 26 24 0.8 22 BI AEI 20 0.7 18 16 0.6 14

0.5 12 10 1 2 3 4 5 6 1 2 3 4 5 6 Measurement day (1-6) Measurement day (1-6) Nectarinia zeylonica Lonchura punctulata Lonchura malabarica

Amandava amandava A bioRxiv preprint doi: https://doi.org/10.1101/518985; this version posted January 13, 2019. The copyright holder for this preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. Ploceus philippinus

Anthus rufulus

Nectarinia asiatica Megalaima haemacephala Anthus trivialis Merops orientalis Cuculus varius

Anthus campestris Cacomantis passerinus Eudynamys scolopaceus Centropus sinensis Ficedula parva Vanellus indicus Cyornis tickelliae Stigmatopelia chinensis Stigmatopelia senegalensis Copsychus saularis Francolinus pondicerianus Saxicoloides fulicatus Francolinus pictus Pavo cristatus Acridotheres tristis Psittacula eupatria Pycnonotus cafer Psittacula krameri 20.0 Orthotomus sutorius Psittacula cyanocephala Tephrodornis pondicerianus Prinia socialis Aegithina tiphia Prinia hodgsonii Pericrocotus cinnamomeus Corvus macrorhynchos Prinia sylvatica Dendrocitta vagabunda Prinia inornata Lanius schach Dicrurus macrocercus

Parus major Cisticola juncidis Phylloscopus trochiloides Phylloscopus humei

Hirundo concolor

Hippolais rama

Sylvia curruca Chrysomma sinense

Pellorneum ruficeps

Turdoides malcolmi

Zosterops palpebrosus

B C 1850 160 MNTD and MPD 1800 Phylogenetic Diversity 140 1750 1700 120 1650 100 MPD monsoon 1600 MPD winter 1550 80 Winter PD Winter 1500 1450 60

1400 MNTD monsoon 40 MNTD winter

1350 by relative abundance Weighted 1300 20 1300 1350 1400 1450 1500 1550 1600 1650 1700 1750 1800 1850 20 40 60 80 100 120 140 160 Monsoon PD Unweighted