temporal and frequency niches during the d awn c horus Nguyen 1 An analysis of temporal and frequency n iches d uring the dawn c horus of birds in Monteverde, Costa Rica Brianne N. Nguyen University of California, Irvine EAP Tropical Biology and Conservation Program, Fall 2017 15 December 2017 A BSTRACT Many animals communicate using acoustic signal s, making frequency bands a limited resource. To deal with this, animals can partition themselves into acoustic frequency niches. Species of i nsects, anurans, and birds have been observed to call at discrete frequencies when sharing space with each other . In this study, I recorded and analyzed the temporal and frequency overlap in bird calls around dawn in Monteverde, Costa Rica over the course of ten d ays. I found very little temporal overlap in the sample I recorded. The calls that did overlap often also overlapped in frequency. My results suggest that frequency niches are not a major method used to improve communication among the birds in Monteverde , at dawn , in late November. This data outli nes the use of frequency bands by birds in Monteverde and is the first step in understanding the bioacousitic ecology of the area. AnÂ‡lisis de los nichos temporales y de frecuencia durante los coros maÂ–aneros de aves en Monteverde, Costa Rica RESUMEN Muchos animales se comunican usando seÂ–ales acÂœsticas, en esta comunicaciÂ—n las bandas de frecuencia pueden ser un recurso limitante. Para lidiar con esto, los animales pueden repartirse en nichos acÂœsticos. En ciertas especies de insectos, de anuros y de aves se ha observado que llaman a frecuencias discretas cuando comparten el espacio entre ellas. En este estudio, registrÂŽ y analicÂŽ la frecuencia temporal y la frecuencia superpuesta de las llamadas de aves alr ededor del amanecer en Monteverde, durante diez dÂ’as. EncontrÂŽ muy poca superposiciÂ—n temporal. Las llamadas que se superpusieron en el tiempo, a menudo tambiÂŽn se superpusieron en la frecuencia. Mis resultados sugieren que los nichos acÂœsticos no son la f orma principal de mejorar la comunicaciÂ—n entre aves al amanecer, en Monteverde a finales de noviembre. Estos datos describen el uso de bandas de frecuencias por ciertas aves en Monteverde, y es el primer paso en el entendimiento de la ecologÂ’a acÂœstica en el Â‡rea. All animals need to communicate. One major method of communicat ion i s making sound s , which takes energy to produce and often requires specialized organs. Still, many organisms use sound to communicate : humans speak, birds sing, frogs croak, dogs bark, etc. The
temporal and frequency niches during the d awn c horus Nguyen 2 problem then becomes, how do you make yourself heard in all the noise? One study conducted in Puerto Rico showed that some species of frogs in the same area all call at the same time, but at diff erent frequencies in order to make their individual calls distinguishable (Villanueva Rivera 2014). These frogs are said to be occupying different frequency niches. Studies have shown that crickets and birds also operate within distinct frequency niches wh en calling in the same time and space (Schmidt et al. 2012 ; Stone 2000 ). Birds call thr oughout the day and can occupy a wide variety of habitats . One time they do sound together is dawn. This phenomen on is called the dawn chorus. There is no definitive answer to why the dawn chorus occurs, but there are many hypotheses. One hypothesis is that in the early morning it is less windy in some places and there is less atmospheric turbulence , therefore allowing for better song or call transmission (Brown and Handford 2003). Another theory is that birds are active at dawn but insects are not, therefore their time is better spent calling than foraging. Other studies that normalize d for foraging strategy and foragi ng height say that eye size (and therefore light sensitivity) is the factor which best predicts dawn chorus timing (Berg et al. 2006 ) . The timing of the dawn chorus has been found to change with ambient light levels (e.g. the moon cycle), weather, and temp erature (Bruni et al. 2014) . In species that have soci al hierarchies, individuals may call at different times based on their rank (Otter et al. 1996). Stanley et al. (2016) found that species whose calls occupied the same space as insect chirping (i.e. crickets and cicadas) delayed their calls until after those insects ceased making noise. Many factors contribute to the eventual timing and distribution of bird calls an d songs. It is important to study bioacoustic ecology, because the ability for an animal to communicate often essential to its ability to survive and reproduce. As human populations grow, so do the sounds they make. This can affect the ability of other or ganisms to communicate. It has been shown that bird species richness and abundance changes with differing amounts of anthropogenic noise (Ar ÂŽ valo and Newhard 2011). Understanding how birds utilize the frequency spectrum is the first step in understanding h ow we as humans can mitigate our effect on the frequency spectrum. I monitored and recorded the different species of birds that call around Monteverde and determi ned the order in which they bega n calling in the morning and the frequency range s their calls inhabit. Monteverde is a unique place of study with many forested areas directly adjacent to developed areas and a wide variety of birds accustomed to tropical and urban environments . I hypothesize d that the use of frequency niches could expla in the calling patterns of tropical bird species in Monteverde. MATERIALS AND METHODS I used a TASCAM DR 40 Linear PCM Recorder to record ambient sounds in Cerro Plano, Monteverde ( 10.3126, 84.8221 ) for an hour each morning starting at 5:00 am for 10 days between 22 November 2017 and 02 December 2017 . The data from t wo days was unusable due to extremely high winds. The study site was a stand of secondary forest adjacent to a mildly trafficked dirt road around 1400 m in elevation. T he recorder w as attached to a tree at
temporal and frequency niches during the d awn c horus Nguyen 3 approximately 1.5 m in hei ght for the duration of the study period using a small tripod and the internal microphone was covered in 1 cm thick pliable foam to protect it from the wind. The sound files were saved in WAV f ormat . I processed the audio files using Audacity 2.0 and Raven Pro 1.5 (Bioacoustics Research Program 2014) . I fi rst amplified the recordings with Audaci ty, and then created spectrograms using Raven. I generated the spectrograms at 16 bit and 44100 Hz . I used the default settings with a frequen cy resolution of 256 samples, a time grid with 50% overlap, and grid spacing of 31.3 Hz using the window Hann function . I isolated and selected all bird calls during the sampling period while looking at the spectrograms and listening to the audio simultaneously (Figure 1) . Raven generated the start time, end time, Figure 1 . A spectrogram from Raven illustrating the vocalization of a White fronted Amazon overlapping with that of the White eared Ground Sparrow. Selection boxes demarking the temporal and frequency ranges are visible as transparent boxes around each vocalization. lowest frequency, highest frequency, and maximum freque ncy for each selection. I parsed the first two minutes of every five minute interval (i.e. 5:00 5:02, 5:05 5:07 , etc . ) in order to get a comprehensive sample of the entire recording . Calls or songs were isolated that were aud ible or visible in the Raven generated spectrogram. Calls or songs were named if they could be defined with a clear beginning and end and were given uniqu e codes so that similar calls could be grouped. I tried to identify calls that occurred often or across multiple recording days to the species level. After the recording period, I surveyed the recording site for half an hour on three days to t ry and get a physical description of the sounding birds to match with the recorded bird calls . I was able to identify birds to the species leve l using physical traits and confirm the song identities with xeno canto.org. The calls of birds that were never seen were also identified using the bird call database at xeno canto.org, by consulting local experts , and by referencing Stiles and Skutch's A Guide to the Birds of Costa Rica . I consulted users on xeno canto forums as well by posting cropped audio files of repeating signals. !"#$% & '()*$%+,-./0)* , !"#$% & %/(%+, 1()2*+,34/(()5 ,
temporal and frequency niches during the d awn c horus Nguyen 4 I then determine d if there were calls that overlapped in both frequency and time. Temporal overlap was determined using a SUMPRODUCT formula on Google Sheets. Frequency overlap was determined by hand. Pairs of overlapping calls were identified and coded as either: overlapping in frequency, not overlapping in frequency or belonging to the same s pecies. If three calls overlapped, the call would get a code for each overlapping incident it was a part of (i.e. if a call overlapped temporally with two other calls it could be both overlapping and not overlapping in frequency range) . Also, it was noted if the member of the overlapping pair was a White fronted Amazon, as those had qualitatively been noted to faci litate many temporal and frequency overlaps. The first call for each of the id entified species was noted for seven of the eight days . On the eighth day, record ing did not start until 5:33. I could not be certain the first call recorded that day was the first for that species and so excluded day eight from the species call progression data. RESULTS I analyzed a total of 2016 calls. These were sorted into 97 groups of identical calls. Of these 97 groups, 34 group s had mo re than 10 calls in them and 10 groups were identified to the species. There were 165 calls that were heard in the recording, but could not be isolated in Rave n due to an excess of background noise (e.g. rain, wind, vehicles, or dogs ) and so did not receive a call code. Table 1 . Numbers of bird vocalizations made by identified and unidentified species. Organized from most frequent to least frequent. Calls were recorded between 22 November 2017 and 02 December 2017 in Monteverde, Costa Rica. Species Number of Calls Brown Jay ( Psilorhinus morio ) 313 Great tailed Grackle ( Quiscalus mexicanus ) 159 Boat billed Flycatcher ( Megarynchus pitangua ) 144 White fronted Amazon ( Amazona albifrons ) 77 Rufous collared Sparrow ( Zonotrichia capensis ) 70 White eared Ground Sparrow ( Melozone leucotis) 49 Social Flycatcher ( Myiozetetes similis ) 27 Streak headed Woodcreeper ( Lepidocolaptes souleyetii ) 17 Orange billed Nightingale Thrush ( Catharus aurantiirostris ) 12 Rufous and white Wren ( Thryophilus rufalbus ) 2 Unidentified calls 981 Uncoded calls 165 Total Calls 2016 There were not many calls from 5:00 5:25 am. Some calls began to be recorded from 5:25 5:35 am . There was an observed peak in bird sounds from 5:35 5:40 am , with calls
temporal and frequency niches during the d awn c horus Nguyen 5 generally decreas ing in abun dance after that point (Figure 2 ). These results a re not s ignificantly different , with much deviation in day to day activity. Figure 2 . Average number of calls over time. The average number of vocalizations occurring in five minute intervals from 5:05 6:05 am over eight days . During this period of time, civil twilight was around ( Â± 2 minutes) 5:15 and dawn was around 5:33 am. (n=1865 , error bars are standard deviation ) Only three species were recorded across all recording days. The Great t ailed Grackle, Boat billed Flycatcher , and White eared Ground Sparrow. There was no clear progression for the first call by species (Figure 3 ) . T he order changes from day to day. The Boat billed Flycatcher's first call is recorded after that of the Great tailed Grackle on all study days but one (Day 2). The Streak headed Woodcreeper's first call is recorded before the Boat billed Flycatcher on 4 out of 5 days recorded and the White eared Ground Sparrow was recorded before the Boat billed Flycatcher 5 out of 7 days . The Brown Jay was recorded first on 4 of the 5 days it was present in the sample . 6, 76, 86, 96, :6, ;6, <6, =6, >6, ?6, -@%(/A%,B2.C%(,)',D/EEF, G#.%,)',H/I,
temporal and frequency niches during the d awn c horus Nguyen 6 Figure 3 . Daily variation in call progression . Each point is the first call recorded for a certain species on a certain day. Time was measured in seconds with time 0 being 5:00 am. Points nearer to the bottom of the graph were recorded earlier in the day. Lines connect the first calls of the same species . There were 8 calls found over multiple days . (n=38) Of the 2016 calls recorded, 370 were overlapping in temporal range (i.e. start time to end time of each call ; Figure 4 ). Of the 370 calls that overlapped temporally, 77% (296 calls) were also overlapping in frequency range. 55% wer e overlap ping calls of different species, w ith 5% being attributed to the White fronted Amazon. 22% of calls with overlapping frequency ranges were from bir ds of the same species (Figure 5 ). 7966, 7>66, 8966, 8>66, 9966, 9>66, H/I,7, H/I,8, H/I,9, H/I,:, H/I,;, H/I,<, H/I,=, 3%J)*+F,/K%(,;L66,/.,MFN, O%J)(+#*A,H/I, P()5*,Q/I, P)/$&C#EE%+,REIJ/$J"%(, 3$(%/S&"%/+%+,!))+J(%%4%(, 1(%/$&$/#E%+,1(/JSE%, !"#$%&%/(%+,A()2*+,F4/(()5, O2')2F&J)EE/(%+,F4/(()5,
temporal and frequency niches during the d awn c horus Nguyen 7 Figure 4 . Number and percent of calls overlapping temporally from 23 November 2017 to 02 December 2017 . "Overlapping" refers to calls that had start times before the end time of a different call. "Not overlapping" refers to calls with start times after the end time of the previous call and end times before the start of the next call. (n=2016) Figure 5 . Of calls that overlapped temporally, n umber and pe rcent of calls with overlapping frequency ranges. "Overlapping" refers to the overlap in the frequency range of the calls. "Overlapping, Same Species" means that the calls overlapped in frequency range and were made by the same species. "Overlapping, T@%(E/44#*AU,9=6U, 7>V, WD-GX1TOY,B-ZX[, T@%(E/44#*AU, W\-]^X[U, W_XODXBG-1X[, T@%(E/44#*AU,7?8U, ;6V, WD-GX1TOY,B-ZX[, T@%(E/44#*AU, W\-]^X[U, W_XODXBG-1X[, T@%(E/44#*AU,3/.%, 34%J#%FU,>:U,88V, T@%(E/44#*AU,!"#$%& '()*$%+,-./0)*FU, 86U,;V,
temporal and frequency niches during the d awn c horus Nguyen 8 White fronted Amazons" refers to calls that were overla pping in frequency and made by White fronted Amazons . (n=370) DISCUSSION I found that birds began to sing around 5:20 am and had a peak in activity around 5:40 am every day (Figure 2 ) . T here was a clear overall trend for the eight days studied. It is possible that the variation among the average number of calls over time and the progression of species ' first calls from day to day is related to the variation in weather every day. For two days during the recording period it was raining lightly, and each day had a varying amount of wind. Birds are known to delay their first calls according to weather (Bruni et al. 2014). It is also important to note that the majority of sounds recorded were calls (single note vocalizations) and not songs (multi note comple x vocalizations). A dawn chorus is normally defined as an abundance of bird songs. I believe that if this study was conducted during the mating season of the species studied , when birds sing more often , a larger difference could be observed. There was not a regular progression of bird calls throughout the course of the study, with many species switching places from day to day (Figure 3 ) . However, there were some patterns that held true for the majority of the study days, like the Great tailed Grackle calli ng before the Boat billed Flycatcher on all days but one. This supports the idea that birds inherently begin calling once sp ecies specific needs are meant, e.g. a certain amount of light (Berg et al. 2006) . There were not many instances of temporal overla p in bird calls (Figure 4 ) . When they did overlap, 77% also overlapped in frequency ranges. This suggests that birds in Monteverde are not using frequency niches to improve communication during the dawn chorus at this time . However, many instances of calls that overlapped could be attributed to individuals of the same species (22%) and 10% involved White front ed Amazon s (5% were made by White fronted Amazons) . This is a notable portion of the overlapping calls as White fronted Amazons were recorded a total of 77 times and were involved in 20 instances of temporal and frequency overlap, meaning that more than 1 in 4 calls by White fron ted Amazons interfered with vocalizations of another species. This is probably due to the fact that their calls occupy a large range in the frequency spectrum and they often call multiple times consecutively (Figure 1 ). This shows there is some separation of vocalizations by frequency in Monteverde, but the majority of separation between vocalizations occurs temporally. It must be said that t here were some false positives recorded in temporal and frequency overlap because of the method used fo r measuring the characteristics, a s calls often utilize different portions of the frequency spectrum at different times in their call , so w hile the total range may overlap, the po rtion in use might not (Figure 1). It is also worth mentioning that the volume at which the birds sounded was not constant, and some overlapping calls had one participant that was very faint which may not have impact ed the ability of the louder individual to communicate. I t was also difficult to hear calls at lower frequencies because of anthropogenic noise pollution and wind. There are also possible false negatives where calls might have overlapped so that one call w as inaudible and indistinguishable from the other in the recording
temporal and frequency niches during the d awn c horus Nguyen 9 and on the sonogram. With this in mind, I think that frequency niches might play a larger role in successful communication when more birds are singing at once, like during the mating season . It is also worth noting that t empora l and frequency overlap can be disproportionately attributed to certain species , e.g. White fronted Amazons . The majority of species in this study used very simple calls, even though the area studied was relatively un populated. Naugler and Ratcliffe (1994) found that individuals of the species Spizella arborea had more complex calls in less populated areas. They attributed this finding to a greater availability of frequency niches and a male bird's ability to seek out a habitat where their song will be most effective. It is possible that the birds in my study area were affected by the anthropogenic noise that occupies the lower range of the frequ ency spectrum throughout the morning . Bird calls and therefore the ability of birds to communicate can easily be masked by outside sources . I experienced firsthand how difficult it can be to differentiate bird calls from anthropogenic noises (including domestic dog vocalizations). It has been found that anthropogenic noise pollut ion can affect the timing of the dawn chorus in birds. Birds in more heavily trafficked areas were found to sing louder and earlier to combat anthropogenic noise sources (Arroyo Solis et al. 2013). Invasive species may also impact normal animal vocalizatio ns. Both and Grant (2012) exposed white banded tree frogs to recordings of invasive american bullfrogs in Brazil and found that the tree frogs immediately shifted their call frequencies higher and maintained these new frequencies even after the removal of the stimulus. Studies have also found that birds with higher frequency calls are more successful in urban areas and some species shift their call frequencies higher in response to urban noise (Hu and Cardoso 2009, Slabbekoorn and Peet 2003 ). However, b irds who shift their call frequencies upward to avoid urban noise pollution could suffer if their possible mates prefer lower frequency calls (Halfwerk et al. 2011). Bird calls often adjust avoid overlap in frequency ranges. Future studies could be conducted in Monteverde on the effect of urban noise on the call frequency of birds in the area. It might also be interesting to look at the different types of calls that birds utilize. While I was measuring and selecting calls, it seemed like the majority of calls could be grouped into two factions : large frequency range and short time or long period of time and small frequency range. It would be interesting to quantify this observation and determine if one morphotype is more popular or successful than another. In conclusion, the birds in this study seemed to be able to partition themselves temporally rather than by frequency. These results suggest that the frequency spectrum in Monteverde is not a limited resource during late November for species singing during the dawn chorus . It is probable that during the breeding season the acoustic area will be more crowded, making the frequency spectrum a more limited resource. The frequency spectrum is one more resource that we share with the na tural wor ld and should study and manage accordingly.
temporal and frequency niches during the d awn c horus Nguyen 10 ACKNOWLEDGEMENTS I would like to thank Federico Chinchilla and Andres Camacho for helping me deve lop and refine my project idea and paper . I would like to thank Edgardo Arevalo for giving me guidance on how to collect my data. I would also like to thank Felix Salazar and Marvin Hidalgo for allowing me to collect preliminary data at the Santuario Ecologico and Estacion Biologica Monteverde. Identifying all the calls would not have been possible without the help of Frank Joyce, Andres Camacho, and all the users on xeno canto.org. Mckenzie Sime helped me revise my paper and Federico Chinchilla helped me translate my abstract . Finally , I would like to thank the Mena Bello family for allowing me to live in and collect data outside their home for the duration of the study. LITERATURE CITED ArÂŽvalo, J. and Newhard, K. 2011. Traffic noise affects forest bird species in a protected tropical forest. Revista de biologia tropical , 59 (2): 969 980. Arroyo Solis, A., Castillo, J., Figueroa, E., Lopez Sanchez, J. and Slabbekoorn, H. 2013. Experimental evidence for an impact of anthropogenic noise on dawn chorus timing in urban birds. Journal of Avian Biology. 44: 288 296. Berg, K., Brumfield, R. a nd Apanius, V. 2006. Phylogenetic and ecological determinants of the neotropical dawn chorus. Proceedings of Royal Society B: Biological Sciences. 273:999 1005. Bioac oustics Research Program. 2014 . Raven Pro: Interactive Sound Analysis Software (Ve rsion 1.5) [Computer software]. Ithaca, NY: The Cornell Lab of Ornithology. Available from http://www.birds.cornell.edu/raven. Both, C. and Grant, T. 2012. Biological invasions and the acoustic niche: the effect of bullfrog calls on the acoustic signal s of white banded tree frogs. Biology Letters. 8:714 716. Brown, T. and Handford, P. 2003. Why birds sing at dawn: the role of consistent song transmission. Ibis. 145:120 129. Bruni, A., Mennill, D. and Foote, J. 2014. Dawn chorus start time variation in a temperate bird community: relationships with seasonality, weather, and ambient light. Journal of Ornithology. 155: 877 890. Halfwerk, W., Bot, S., Buikx, J., van der Velde, M., Komdeur, J., Cate, C. and Slabbekoorn H. 2011. Low frequency songs l ose their potency in noisy urban conditions. PNAS. 108(35):14549 14554. Hu, Y and Cardoso, G. 2009. Are bird species that vocalize at higher frequencies preadapted to inhabit noisy urban areas? Behavioral Ecology. 20:1268 1273. Naugler, C. and Ratcliff e, L. 1994. Character release in bird song: a t est of the acoustic competition hypothesis using american tree s parrows Spizella arborea . Journal of Avian Biology . 25(2): 142 148. Otter, K., Chruszcz, B. and Ratcliffe, L. 1996. Honest advertisement and song output during the dawn chorus of black capped chickadees. Behavioral Ecology. 8(2):167 173.
temporal and frequency niches during the d awn c horus Nguyen 11 Slabbekoorn, H and Peet, M. 2003. Birds sing at a higher pitch in urban noise. Nature. 424: 267. Stanley, C., Walter, M., Venkatraman, M and Wilkinson, G . 2016. Insect noise avoidance in the dawn chorus of Neotropical birds. Animal Behavior. 112:255 265. Stiles, F.G. and Skutch, A.F., 1989. Guide to the birds of Costa Rica. Comistock. Stone, E. 2000. Separating the noise from the noise: a finding in sup port of the "niche hypothesis , " that birds are influenced by human induced noise in natural habitats. Anthrozoos. 13(4):225 231. Villanueva Rivera, L. J. 2014. Eleutherodactylus frogs show frequency but no temporal partitioning: implications for t he acoustic niche hypothesis. PeerJ. 2: e496.