Helicopter Noise in Hawaiʻi's Protected Natural Areas

Home , Japanese bush warbler, Myadestes, Songbird

HELICOPTER NOISE IN HAWAIʻI’S PROTECTED NATURAL AREAS CHANGES TEMPORAL CHARACTERISTICS OF SONGBIRD VOCALIZATIONS

A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI‘I AT HILO IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE

TROPICAL CONSERVATION BIOLOGY AND ENVIRONMENTAL SCIENCE

DECEMBER 2019

By Karen Gallardo Cruz

Thesis Committee: Patrick Hart, Chairperson Kristina Paxton Matthew Knope Bryan Pijanowski

Keywords: helicopter noise, vocal plasticity, Hawaiʻi, bird vocalizations, noise pollution

ACKNOWLEDGEMENTS

There are many people who I’d like to thank from the bottom of my heart for their support of this project. First of all, my parents, Guadalupe and Frank Gallardo, who have unconditionally supported my dreams and academic pursuits—mi éxito es su éxito. A tremendous thank you to my committee members, all whom gave me critical feedback on my manuscript and helped me improve it: Dr. Patrick Hart for granting me the opportunity to join the LOHE lab, believing in my ability to take on this project, and helping me throughout the process; Dr. Kristina Paxton for patiently answering every question I had, big or small, and for all of her help in statistical analysis; and Dr. Matthew Knope and Dr. Bryan Pijanowski for graciously offering their time, knowledge, and expertise to help ensure the success of my project and quality of my manuscript. A special thank you to Kurt Fristrup and Ashley Pipkin from the National Park Service Natural Sounds and Night Skies Division for initiating the project and for providing background noise data from the Larson Davis 831 Sound Level Meter ANSI Type 1 Acoustic Monitoring system. Thank you to Danielle Foster from Hawaiʻi Volcanoes National Park and Scott Jorgensen from Blue Hawaiian Helicopters, who were instrumental in my site selection. A huge thank you to every person that accompanied me in the field, especially Serafina Gajate and Jeff Wood for their enthusiasm and repeated help. Finally, I am so grateful to my dear friends and family for their emotional support throughout this endeavor. Funding: This work was supported by the United States National Park Service and the University of Hawaiʻi at Hilo (Task Agreement number P17AC00716).

ii ABSTRACT

Anthropogenic noise has adverse effects on birds, including decreased breeding success, increased flushing behavior, and changes in vocalization patterns. The avifauna in Hawaiʻi is among the most threatened in the world, and helicopter noise in Hawaiʻi’s forests could be another stressor native birds face in addition to disease, habitat loss, and non-native species, but its effect on bird vocalizations has never been assessed. My primary objective was to determine if helicopter noise affects the temporal characteristics of songbird vocalizations within protected natural areas. I placed automated acoustic recorders in three forested areas that are subjected to helicopter traffic from air tours, two in Hawaiʻi Volcanoes National Park and one in the Upper Waiākea Forest Reserve on the Island of Hawaiʻi. I found that songbirds change their vocalization time in response to the power (dB) of approaching helicopter noise. These results indicate that birds are using temporal shifts in vocalizations to mitigate masking effects from helicopter noise. Additionally, I found that the strength and direction of the response is species- specific. Warbling White-eye increased vocalization time, and ‘Apapane and Japanese Bush- warbler decreased vocalization time as helicopter noise power increased, suggesting differences in resilience to helicopter noise between species. Furthermore, results of this study suggest that birds respond the strongest to helicopter noise in areas with very loud and frequent helicopter traffic. My results demonstrate impacts of anthropogenic noise on native bird habitat and may serve as the foundation of an air tour management plan that considers reducing the number of helicopter overflights over protected natural areas to 4 or less helicopters per hour and enforcing a higher flight altitude to decrease power levels of overflights to 191 dB or less.

iii TABLE OF CONTENTS

Acknowledgements...... ii Abstract...... iii Table of contents...... iv List of tables...... v List of figures...... vi Introduction...... 1 Methods...... 5 1. Study sites...... 5 2. Study species...... 5 3. Field recording methodology...... 6 4. Recording and statistical analysis...... 7 4.1 Bird vocalizations and noise power level...... 8 4.2 Bird vocalizations and noise period...... 9 4.3 Soundscape measurements and noise period...... 10 Results...... 12 1. Helicopter abundance and power...... 12 2. Bird vocalizations and noise power level...... 12 3. Bird vocalizations and noise period...... 12 4. Soundscape measurements and noise period...... 13 Discussion...... 14 Tables...... 19 Figures...... 20 References...... 29

iv LIST OF TABLES Table 1. Table depicting species used in each analysis, varying by site...... 19

v LIST OF FIGURES Figure 1. Study sites...... 20

Figure 2. Predicted values of total vocalization time as a function of relative maximum power of helicopter noise for three species at Nāpau Crater...... 21

Figure 3. Interval plots depicting the total vocalization times for “before,” “during,” and “after” periods of helicopter noise for the most prominent bird species at three sites...... 22

Figure 4. Interval plots depicting the total vocalization time of three bird species during the “before,” during,” and “after” periods of helicopter noise...... 23

Figure 6. Interval plot showing the relative maximum helicopter noise power levels (dB) per site...... 25

Figure 7. Interval plot showing helicopter abundance per site...... 26

Figure 8. Predicted values of total vocalization time based on the generalized linear mixed model for five species at Escape Rd...... 27

Figure 9. Predicted values of total vocalization time based on the generalized linear mixed model for five species at UWFR...... 28

vi INTRODUCTION

Human invasion of natural habitats is often thought of in terms of infrastructure and urban development, but humans also invade natural habitats by introducing noise into the natural soundscape. Human-produced noise, or anthropogenic noise, is prominent in protected natural areas (Buxton et al. 2017) and can have severe consequences for the species that live within those areas. Anthropogenic noise negatively impacts species from many taxonomic groups and can lead to increases in heart rate and activity in land mammals (Weisenberger et al. 1996; Krausman et al. 1998; Maier et al. 1998), acoustic trauma in cephalopods and cetaceans (André et al. 2011; D. Ketten, Woods Hole Oceanographic Institution Report 2004), and behavioral changes in arthropods, anurans, cetaceans, and birds (Patenaude et al. 2002; Schmidt et al. 2014; DeRose-Wilson et al. 2015; Orci et al. 2016; Roca et al. 2016; but see Costello and Symes 2014). Specifically, in birds, anthropogenic noise can cause shifts in stress hormone levels and reduce fitness in adults and young (Kleist et al. 2018). Noise can also alter bird communities and affect predator-prey interactions. For example, Francis et al. (2009) found that areas close to gas well compressor noise had lower bird species richness than areas further from the noise. However, nesting success was higher in noisier areas due to noise intolerance by predatory birds.

In addition to the direct impacts of anthropogenic noise on species richness, bird physiology, and fitness, noise can impede bird communication. Birds use acoustic cues to send and receive crucial information between mates and conspecifics and between parents and offspring (Marler 2004, Gil and Brumm 2014). Additionally, birds vocalize to ward off predators and intruders and to communicate danger associated with predators (Marler 2004). Although birds use vocalizations in many important ways, the two main functions of bird vocalizations are territory defense and mate attraction, particularly for songbirds that have developed song to attract mates and repel rivals (Catchpole and Slater 2008). High noise levels in the environment have been linked to decreased pairing and reproductive success of songbirds due to a decreased ability for mating partners to vocally communicate (Habib et al. 2007; Halfwerk et al. 2011). Moreover, a 10 dB(A) increase in noise above background levels translates into a 90% decrease in listening area (Barber et al. 2010; Buxton et al. 2017). This reduced distance at which vocalizations by a signaller can be detected and translated by a receiver can in turn lead to reduced territory size and higher levels of territory intrusion (Nemeth and Brumm 2010).

1 To mitigate negative effects of anthropogenic noise on communication, some bird species have developed mechanisms to increase the chance of successfully transmitting their vocalizations in the presence of anthropogenic noise. Several songbird species have shown a shift in the frequency range of their vocalizations away from the fequency of the noise immediately upon being confronted with anthropogenic noise (Potvin and Mulder 2013; Roca et al. 2016). Other species sing louder or more repetitively in the presence of anthropogenic noise, presumably to increase their chances of being heard (Pieretti and Farina 2013; Potvin and Mulder 2013). Alternatively, some bird species adjust the temporal characteristics of their vocalizations (Brumm and Zollinger 2013). For example, common nightingales (Luscinia megarhynchos) preferentially sing during silent periods between periods of noise (Brumm 2006), while bird communities near commercial airports with early flights begin the dawn chorus earlier than bird communities living farther from the same airports (Gil et al. 2015).

Anthropogenic noise is prominent in United States protected natural areas and can impact wildlife (Buxton et al. 2017). Noise from air tours, especially helicopter tours, is common in U.S. national parks, with 95% of all air tours nationwide occurring in ten of the fifty-eight national parks (B. Lignell, National Park Service Report 2018). Not only does aircraft noise negatively impact visitor experience in national parks (Gramann 1999) and reduce the aesthetic appeal of natural landscapes (Mace et al. 2003), but it can also have negative consequences on birds. For instance, increased stress and agonistic behavior in response to aircraft noise has been found in ducks, sea birds, hawks, and owls (Andersen et al. 1989; Brown 1990; Conomy et al. 1998; Delaney et al. 1999; Goudie and Jones 2004). Furthermore, bald eagles show flight and alert responses to aircraft overflights, responding more strongly to helicopters in comparison to jets and light planes (Grubb and Bowerman 1997). However, though exposure to aircraft noise can initially lead to changes in behavior, birds can also become habituated and respond less to aircraft noise with increased exposure. For example, an experimental study exposing American black ducks (Anas rubripes) to jet plane noise found that their reaction to the noise decreased over time (Conomy et al. 1998). Similarly, red-tailed hawks (Buteo jamaicensis) nesting at sites with novel helicopter traffic exhibited stronger avoidance behavior in response to helicopter overflights than those nesting at sites that had been experiencing helicopter traffic for three decades longer (Andersen et al. 1989). These past studies have provided valuable information regarding how birds respond to aircraft noise, but no studies have explored the effect of aircraft

2 noise on songbirds despite their ubiquity in natural areas that are constantly exposed to aircraft noise. Understanding how different species react to aircraft noise would allow land managers to create effective management strategies for the groups of wildlife in the areas they manage.

Hawaiʻi Volcanoes National Park (HAVO) experiences the second highest air tour activity, mostly by helicopters, in the nation (B. Lignell, National Park Service Report 2018), with only Grand Canyon National Park being greater. HAVO is also home to eleven bird species endemic to Hawaiʻi, including six species listed as endangered and two listed as vulnerable by the IUCN Red List as of 2018. Hawaiʻi’s native bird species are already facing many stressors such as introduced predators, habitat loss, and disease-transmitting mosquitoes (Atkinson and Lapointe 2009; Pratt et al. 2009; LaPointe et al. 2010). The presence of helicopter noise in the protected natural habitat of endemic Hawaiian birds could be an additional stressor that is decreasing the fitness of Hawaiian birds by impairing their ability to communicate vocally with one another.

Past studies examining the effect of noise on bird vocalizations have focused mainly on constant noise coming from sources such as roads and gas well compressors. Noise coming from helicopters provides the opportunity to understand how birds respond to a noise source that is sporadic, relatively short-lived, and can be so loud that bird song modifications such as frequency shifts, repetitiveness, or increases in power are unlikely to overcome masking (Gil et al. 2015). The most adaptive response by song birds to this type of noise may be through temporal changes in vocalizations, but this question has yet to be explored. The objective of my study is to examine if helicopter noise affects the vocalizing behavior of wild songbirds in protected natural areas. Because some species show greater sensitivity to anthropogenic noise than others (Francis et al. 2009; Roca et al. 2016), I predict that the level of response to helicopter noise will differ by species. Additionally, birds may not be able to overcome masking by helicopter noise and vocalizing requires energy expenditure. Therefore, my first hypothesis is that the amount of time birds spend vocalizing will change as the helicopter noise changes in power and frequency range. I predict that birds will vocalize less with increasing power and frequency range of helicopter noise. Similarly, my second hypothesis is that the vocalization time of each songbird species will differ between the periods before, during, and after the helicopter noise. I predict that birds will vocalize more before the onset of helicopter noise, when

3 no masking by helicopter noise occurs, than during the helicopter noise, when masking of vocalizations may be the highest. If vocalization time indeed decreases during the noise, I also predict there will be less vocalizations in the period immediately after the helicopter noise because it may take birds time to recover from the effects of the noise.

Apart from observing species-level responses, I am also interested in measuring how the entire community of biological sounds changes in response to helicopter noise. Sounds are ecological properties of landscapes (Schafer 1977). The collection of sounds in a landscape is known as the soundscape and consists of anthropogenic sound, non-biological ambient sound, and biological sound (Pijanowski et al. 2011). In the field of soundscape ecology, the entire community of biological sounds is known as biophony, and includes any sounds produced by animals such as arthropods, birds, and mammals (Pijanowski et al. 2011), and can be measured through the use of soundscape indices such as the Bioacoustic Index (Boelman et al. 2007) and the Acoustic Complexity Index (Farina et al. 2011). My third hypothesis is that soundscape indices will reflect vocalization differences at the community level between the periods before, during, and after helicopter noise. Finally, I hypothesize that birds in areas of loud and frequent helicopter traffic will react the strongest to noise. Conversely, because some birds are capable of habituating to helicopter noise (Andersen et al. 1989; Grubb and Bowerman 1997; Conomy et al. 1998), an alternative hypothesis is that birds in sites with frequent helicopter traffic will react the least strongly to helicopter noise. Results from this study will provide valuable information on how helicopter noise affects the vocalizing behavior of different songbird species and will help to inform the creation of effective management strategies for air tours over protected natural areas that are home to a diverse array of species.

4 METHODS

1. Study sites

I conducted this study in three protected natural areas on Hawaiʻi Island that experienced various degrees of traffic from helicopter tours: Escape Rd. and Nāpau Crater in HAVO, and the Upper Waiākea Forest Reserve (UWFR). Escape Rd. and Nāpau Crater experience chronic levels of helicopter traffic due to their proximitiy to a popular air tour destination and have been experiencing helicopter overflights since at least 1983. UWFR did not regularly experience helicopter traffic until the volcanic eruption of the Kīlauea Volcano in May 2018, when the airspace above UWFR became an easy route for helicopter tours to access the eruption zone. All three sites are dominated by 10-15 m tall native ‘ōhi‘a (Metrosideros polymorpha) trees. However, the HAVO sites are on 200 to 400 year-old substrate (D. Sherrod et al., U.S. Geological Survey Open-File Report 2007) with an open-canopy forest and a native staghorn fern (Dicranopteris linearis) understory, while the substrate at the Upper Waiākea Forest Reserve is approximately 3,000 to 5,000 years old (D. Sherrod et al., U.S. Geological Survey Open-File Report 2007) and consists of a closed canopy forest with an understory of native shrubs and ferns. Native and non-native birds occur at all three study sites. All sites recorded are completely rural and are not subjected to light pollution. Additionally, HAVO sites are not regularly exposed to noise other than that from helicopters and fixed wing aircraft, but some road noise is audible at UWFR despite a distance of approximately 2.25 km from the nearest highway. Ancecdotally, helicopters fly at a higher altitude at Escape Rd. and UWFR than at Nāpau Crater.

2. Study species

All species studied are passerines (order Passeriformes). At all three sites, the most common species were ʻŌmaʻo (Myadestes obscurus; OMAO; family Turdidae) and ʻApapane (Himatione sanguinea; APAP; family Fringillidae), both native species, and Warbling White-eye (Zosterops japonicus; WAWE; family Zosteropidae), a non-native species. In addition, Japanese Bush-warbler (Horornis diphone; JABW; family Cettiidae) occurred at both HAVO sites, while Hawaiʻi ʻAmakihi (Chlorodrepanis virens; HAAM; family Fringillidae) only occurred at Escape Rd. and UWFR. Iʻiwi (Drepanis coccinea; IIWI; family Fringillidae) are abundant only at UWFR. ʻŌmaʻo are listed as Vulnerable under the IUCN Red List and Iʻiwi are listed as threatened under the Endangered Species Act. 5 3. Field recording methodology

I conducted soundscape recordings at HAVO between March and May 2018 and again between September and December 2018 following a period of disruption due to hazards from the Kīlauea Volcano eruption from May to August 2018. At UWFR, I conducted soundscape recordings between September and December 2018. Recording months were during the potential breeding season for native Hawaiian forest birds, when birds are the most vocal (Lepson and Freed 1995).

I recorded the soundscape at 11 stations at each site (Fig.1). At each recording station, I deployed one weatherproof passive acoustic recorder with an omnidirectional microphone (mono-L, models SM2 and SM2+ from Wildlife Acoustics), which was programmed to record for nineteen minutes with a one-minute break between recordings during peak helicopter tour hours, 7 am to 5 pm. I tied recorders to a tree at a height of 1.5 to 2 m above the ground and programmed them to record for ten days. Recordings were saved onto 16 GB Secure Digital cards in .wav file format at 44100 Hz sample rate with 32 bits of digitization and gain of 36 dB, which maximized the amplitude of bird song in the recordings while avoiding clipping of recordings from the helicopter noise. Due to a limited number of recorders, I rotated ten recorders between recording stations throughout the recording months until each station had at least six days of recordings.

I conducted sensitivity tests on the recorders and microphones to determine the distance between recorders that would ensure independence in acoustic recordings. From a portable speaker, I emitted pure tones and vocalizations of the most common bird species at a constant amplitude of 70 dB, the average maximum power of bird song observed in these forests. Then, using the recorder settings previously described, I recorded the emitted sounds at several distances from the recorder. I determined from these tests that 90 meters was sufficient distance between recorders to ensure independent recording samples.

Past literature reviews on the effects of noise on wildlife suggest that ambient noise levels be reported in publications (Shannon et al. 2016). Ambient noise levels were measured at Nāpau Crater and Escape Rd. using a Larson Davis 831 Sound Level Meter ANSI Type 1 Acoustic Monitoring system, calibrated yearly. Due to limited access to this acoustic monitoring system, I measured background noise levels at UWFR using methods similar to Potvin and Mulder (2013). 6 Ambient noise levels were measured using the slow response measurement with A weighting at five recording stations using a sound level meter (Sinometer, model JTS1357 from Hisonic International Inc.). Measurements of ambient noise were taken on the same day between 8 am and 2 pm for one minute at each station, and I rotated between stations to get a total of two measurements at each. I used the two measurements to calculate mean ambient noise at each station and averaged the means from all stations to get a final calculation of ambient noise at UWFR.

4. Recording and statistical analysis

I examined the frequency of helicopter overflights that occur at each site based on nine hours of recordings. To examine the frequency of helicopter overflights occurring at Escape Rd. and Nāpau Crater, I chose recordings that overlapped in date and time at these two recoding stations during April 2018, counted the number of helicopters recorded per hour, and used these values to calculate mean number of helicopter overflights per hour at each station. In the same manner for UWFR, I calculated the mean number of helicopters per hour from recordings taken at the same recording station from September and October 2018. I conducted a Kruskal-Wallis test to examine differences in helicopter frequency between sites.

Recordings were analyzed in the Listening Observatory for Hawaiian Ecosystems (LOHE) Bioacoustics Lab at the University of Hawaiʻi at Hilo using Raven Pro acoustic analysis software (1.5 Beta Version, 2013) with a Frequency Fourier Transform of 512. Helicopter noise recordings considered for analysis did not contain rain or strong wind. The presence of distant aircraft was allowed in recordings due to constant helicopter traffic during some recording days, especially at Nāpau Crater. However, all recordings had a buffer of at least ten minutes with no helicopter noise above 1 kHz prior to the start of the helicopter noise considered for the analysis. Because some recording stations had a limited number of helicopter noise recordings from which to choose, I also allowed the presence of a second helicopter noise occurring after the helicopter noise to be analyzed.

For each recording station, I randomly chose four recordings from different days that fit the aforementioned criteria. However, two stations in UWFR and one at Escape Rd. only had three helicopter recordings each due to poor weather conditions and a limited number of noise recordings during the recording period. Therefore, I analyzed 43 recordings at Escape Rd., 42 7 recordings at UWFR, and 44 recordings at Nāpau Crater. Each helicopter recording was given a unique identifier to be used as a random factor in later statistical analysis. For all bird vocalization data, I only used bird vocalizations that were at least 5 dB higher than the average power of the background noise in the recording. Background noise was measured during the first second of the helicopter noise that did not contain any vocalizations or harmonics, between 8-9.5 kHz. For all analyses, I only included species that were detected at least 20 times in the recordings (Table 1).

4.1 Bird vocalizations and noise power level

All statistical analyses were done in R (R Core Team 2016) at an alpha level of 0.05. I addressed the effect of helicopter noise on total vocalization time of birds in two ways. First, to address the question of whether total vocalization time changes as helicopter noise increases in power and frequency range, and if the response differs by species (Question 1), I measured the maximum power (dB) and frequency range (kHz) of helicopter noise during ten-second periods for each of the randomly selected recordings per station. Measurements were taken every ten seconds, starting at the peak frequency of the helicopter noise and working backwards in the recording to the start of the helicopter noise. During the same ten-second periods of helicopter noise analyzed, I measured the total vocalization time (in seconds) of each bird species. Total vocalization time was calculated separately for each species and defined as the sum of all vocalization times during the ten-second period. In cases when the peak frequency of helicopter noise was higher than 3.5 kHz, I began measurements at 3.5 kHz instead of at the peak helicopter noise to reduce the chance of missing bird vocalizations masked by the helicopter noise. Data for frequency range and maximum power of helicopter noise was positively correlated (Pearson’s correlation; r = 0.786, n = 286, p < 0.001). Therefore, I only included maximum power for statistical analysis, as the power of helicopter noise can be more easily measured in the field than frequency range and may thus be easier to use in land management practices. I tested the relationship between total vocalization time (response variable) and the interaction between maximum power of the helicopter noise and bird species (explanatory variables) using a mixed effects negative bionomial model to account for over-dispersion in the data (R package lme4; Bates et al. 2015). Helicopter nested in recording station was included in the model as a random factor. I re-scaled vocalization time data by rounding to the nearest hundredth and multiplying by

8 100, then centered and scaled maximum power data so these two measurements would be statistically comparable. The analysis was done separately by site to separate potential site-level differences. For significant models, I conducted a post-hoc Tukey test (R package lsmeans; Lenth 2016), and calculated p-values using the R package afex (Singmann et al. 2019). Power values are inflated due to gain settings in recorders, and thus are presented in the results as relative maximum power (dB).

4.2 Bird vocalizations and noise period

To determine whether different bird species vocalized more before the onset of helicopter noise than during and after (Question 2), I utilized the same recordings from the previous analysis, but excluded recordings that contained noise from a second helicopter after the noise to be analyzed to avoid measuring potential effects from the second helicopter noise on bird vocalizations. Furthermore, recordings from Escape Rd. and UWFR that contained additional noise below 1 kHz from distant helicopters were removed from the analysis. This led to a total of 39 recordings analyzed from Escape Rd., 41 recordings from UWFR, and 36 recordings from Nāpau Crater. Although some recordings from Nāpau Crater were excluded from the analysis, the majority of the analyzed recordings contain some helicopter noise from distant helicopters due to constant presence of helicopter traffic. For each recording, I measured total vocalization time for 10 seconds immediately before the helicopter noise began, for 10 seconds immediately after the noise ended, and for 10 seconds during the helicopter noise, measured immediately after the peak frequency of the helicopter noise. Similar to the previous analyses, in cases when the peak helicopter noise was higher than 3.5 kHz, I began measurements at 3.5 kHz instead of at the peak helicopter noise. I conducted the analysis separately for each site and, due to high variability in vocalization time data, calculated mean vocalization time of each species for each recording station. Then, I tested the relationship between mean vocalization time (response variable) and the interaction between period of helicopter noise and bird species (explanatory variables) using a mixed effects negative bionomial model to account for over-dispersion in the data (R package lme4; Bates et al. 2015). Recording station, helicopter, and maximum power of the entire helicopter noise were included as random factors for Nāpau Crater and Escape Rd. data. Maximum power was excluded as a random factor for UWFR data due to lack of convergence in the model. Similar to the previous analysis, I re-scaled vocalization time and

9 centered and scaled maximum power, calculated p-values using the R package afex (Singmann et al. 2019), and conducted a post-hoc Tukey test for significant models to determine differences between periods. To detect further differences in vocalizations between helicopter noise periods, I noted the vocalization type for three species that have easily distinguishable songs and calls, Warbling White-eye, Hawaiʻi ʻAmakihi, and ʻŌmaʻo. Then, combining data from all three sites due to limited data points from each site, but separating by species, I tested the relationship between total vocalization time (response variable) and the interaction between vocalization type and period (explanatory variables) using a mixed effects negative binomial model to account for over-dispersion in the data (R package lme4; Bates et al. 2015).

4.3 Soundscape measurements and noise period

To answer whether the entire community of biological sounds differed between the periods before, during, and after the helicopter noise, I used a soundscape index. Several soundscape indices have been developed to measure characteristics of biophony such as complexity, diversity, and entropy among others, and many of these indices correlate with biodiversity (Boelman et al. 2007; Sueur et al. 2008; Villanueva-Rivera et al. 2011; Buxton et al. 2018). For this study, I used the Bioacoustic Index (BI), which is a function of the sound pressure level and the number of frequency bands being occupied in a recording (Boelman et al. 2007). The BI also has been shown to correlate with avian diversity in Hawaiʻi (Boelman et al. 2007). For the analysis, I used the same recordings as the previous analysis, but excluded recordings that had helicopter noise with frequency exceeding 4 kHz. This resulted in a sample size of 36 recordings from Escape Rd., 39 recordings from UWFR, and 19 recordings from Nāpau Crater. For each recording, I saved 10-second wav files before, during, and after the helicopter noise, and calculated the Bioacoustic Index (BI; R package soundecology; Villanueva- Rivera et al. 2011) of each file between 4 – 8 kHz, which preserves the frequency range for all of the bird and cricket species in my sites but ensures that helicopter noise is not included in the calculation of the BI. I combined data from the three sites which included all 11 recording stations each at Escape Rd. and UWFR, but only 9 recording stations at Nāpau Crater due to lack of recordings with helicopter noise below 4 kHz at two stations. I calculated the mean BI for each recording station and tested the relationship between mean BI (response variable) and time

10 period (explanatory variable) using a linear mixed model (R package lme4; Bates et al. 2015). Recording station was a nested random factor within site.

11 RESULTS 1. Helicopter abundance and power

Sites significantly differed in the number of overflights they experience per hour (H = 13.769, df = 2, p = 0.001). Nāpau Crater had a higher number of overflights per hour (16.22 +/- 10.17 sd) than Escape Rd. (3.66 +/- 3.12 sd) and UWFR (0.66 +/- 1.65 sd), but Escape Rd. and UWFR did not significantly differ in number of overflights per hour (Figure 7). Helicopter noise at Nāpau Crater was louder than helicopter noise at UWFR, but helicopter noise at Escape Rd. was not significantly different from Nāpau Crater nor UWFR in maximum power (ANOVA,

F2,29.54 = 4.41, p = 0.02; Figure 6). Ambient noise levels at Nāpau Crater were 34.8 dB(A) and 39.3 dB(A) at Escape Rd. Ambient noise levels at UWFR were 45.83 dB(A).

2. Bird vocalizations and noise power level

At Nāpau Crater, I found that helicopter noise significantly affected the vocalizing behavior of native and non-native birds (F2,422.3 = 6.37, p < 0.01). The total time birds spent vocalizing changed as helicopter noise increased in power, but the strength and direction of the response differed by species (Fig. 2). Per ten second period, Warbling White-eye increased total time spent vocalizing by 0.59 s for every 10 dB increase in helicopter noise. Conversely, ʻApapane and Japanese Bush-warbler decreased total time spent vocalizing by 0.54 s and 0.47 s, respectively, for every 10 dB increase in helicopter noise. Mean vocalization time during a 10 second period of no helicopter noise at Nāpau was 4.14 s (sd = 2.63) for Warbling White-eye, 3.08 s (sd = 2.20) for ʻApapane, and 1.64 s (sd = 2.04) for Japanese Bush-warbler. Although there were small changes in vocalization time for other bird species at Escape Rd. (Figure 8) and UWFR (Figure 9), there was not a significant relationship between the power of helicopter noise and vocalization time at these sites (Escape Rd.: F4, 429.10 = 0.94, p = 0.44; UWFR: F4,807.68 = 1.61, p = 0.17). These results indicate that the response to helicopter noise is strongest at sites with loud and frequent helicopter traffic.

3. Bird vocalizations and noise period

Contrary to my prediction that birds would decrease vocalizations during and after the helicopter noise in comparison to the period before the noise, I did not find a significant difference in total vocalization time between the periods before, during, and after the helicopter

12 noise (Fig. 3; Nāpau: F4, 141.87 = 0.87, p = 0.48; Escape Rd.: F2, 106.62 = 1.73, p = 0.18; UWFR: F 4,

208.02= 0.48, p = 0.75). Furthermore, I found that there is no difference in vocalization time of songs or calls between noise periods for ʻŌmaʻo (Fig.4a; F2, 79.63 = 3.31 p = 0.04), Warbling

White-eye (Fig. 4b; F2, 73.27 = 1.01, p = 0.37) or Hawaiʻi ʻAmakihi (Fig 4c; F2, 54.23 = 0.48, p = 0.62).

4. Soundscape measurements and noise period

When examining changes within the entire soundscape before, during, and after helicopter noise, I found that the Bioacoustic Index differs between periods when all three sites are combined (F2, 60 = 3.99, p = 0.02). Bioacoustic Index values were higher during the helicopter noise than after the helicopter noise, while Bioacoustic index values before helicopter noise did not differ between the other two time periods (Fig. 5). These findings indicate a higher diversity of sounds during the helicopter noise in comparison to after the noise.

13 DISCUSSION

I predicted that birds in areas of loud and frequent helicopter traffic would react the strongest to noise. Within this study, Nāpau Crater experienced the most frequent and some of the loudest helicopter noise, and likewise birds exhibited the strongest response to helicopter noise at this site. My results suggest that birds at this site may have developed mechanisms to effectively overcome any masking effects of noise on their vocalizations given that helicopter noise has become a regular part of the soundscape at Nāpau Crater. Results demonstrate that ʻApapane, Warbling White-eye and Japanese Bush-warbler actively changed the amount of time they vocalized in relation to helicopter noise, suggesting the presence of vocal plasticity in these species (Brumm and Zollinger 2013). Noise from an approaching helicopter begins with low power and frequency range. Though distant noise can still incite a behavioral or physiological response, it may not be strong enough to mask vocalizations (Dooling and Blumenrath 2013). However, as distance between a bird and a noise source decreases, the noise can mask signals and elevate the bird’s hearing threshold (Dooling and Blumenrath 2013). Because bird signals generally contain more energy in the higher frequency bands, masking effects from a distant helicopter noise are not very likely because helicopter noise is primarily at low frequency bands, and birds are likely to still be able to recognize and discriminate signals (Dooling and Blumenrath 2013). However, when a helicopter flies directly above the recorder, especially at lower altitudes, helicopter noise reaches power and frequency range levels that mask bird signals (personal observations).

Because approaching helicopter noise is concentrated in the lower frequency bands, the frequency range of a bird’s vocalizations may influence their response to helicopter noise. I predicted that the response to helicopter noise would differ by species and that increasing noise power would cause a greater masking effect for all species and lead to decreases in vocalization time. However, Warbling White-eye vocalized more as helicopter noise increased in power. A similar observation has been made in serins (Serinus serinus) with highway noise (Díaz et al. 2011) and silvereyes (Zosterops lateralis), a close relative of Warbling White-eye, which increased call duration and amplitude during periods of high and low frequency noise (Potvin and Mulder 2013). Increasing call duration or amplitude in the presence of noise would only be an effective mechanism to increase the chance of being heard if noise frequency does not

14 completely mask vocalizations. Warbling White-eye vocalize at frequencies above approaching helicopter noise, and thus may still be able to transmit and detect signals between conspecifics even as helicopter noise increases in power and frequency range (Dooling and Blumenrath 2013). Furthermore, silvereyes have the ability to immediately adjust the frequency of their vocalizations to reduce overlap with noise (Potvin and Mulder 2013). If Warbling White-eye share in this trait, there is a higher likelihood that any vocalizations they emit will be successfully transmitted even in periods of noise. Conversely, to Warbling White-eye and as predicted, ʻApapane and Japanese Bush-warbler decreased their vocalization time as the helicopter noise increased in power. ʻApapane and Japanese Bush-warbler vocalize at lower frequencies and may experience a greater masking effect from approaching helicopter noise, thus it may be more beneficial for these two species to avoid or reduce masking effects by reducing vocalizations during periods of noise (Brumm and Zollinger 2013). The ability of Warbling White-eye to effectively vocalize during periods of noise could give them an advantage over ‘Apapane and Japanese Bush-warbler because they may still be able to communicate with mates and young and maintain territories even in areas of frequent and loud helicopter noise. Warbling White-eye, an invasive species, are already the most abundant land-bird in Hawaiʻi (Van Riper 2000), and their potential resilience to anthropogenic noise could aid in their ability to thrive in a diversity of environments.

I did not find a strong relationship between vocalization time and maximum power of helicopter noise at areas with infrequent helicopter traffic and mid-level maximum power. Past studies have found that the response of birds to noise is dependent on the amplitude of noise (Brumm et al. 2009; Dowling et al. 2011; Potvin and Mulder 2013). Helicopter overflights fly less frequently and at a higher altitudes at Escape Rd. and UWFR. Therefore, the levels of high- amplitude helicopter noise at these two sites may not be severe enough to mask bird song and incite a strong response by birds and may explain why I did not find an effect of noise on bird vocalizations at these two sites. Differences in forest structure may also partially explain differences in the response to helicopter noise between Nāpau Crater and UWFR. Vocalizing from a higher perch improves sound transmission and can mitigate masking effects of helicopter noise (Brumm and Slabbekoorn 2005). Thus, the availability of higher perches at UWFR may be abating the effect of helicopter noise on birds. However, the forest structure between Nāpau Crater and Escape Rd. is very similar, suggesting that differences between these two sites in the

15 frequency and altitude levels of helicopter overflights primarily explain differences in bird response to helicopter noise. My findings imply that if helicopter overflights were managed to fly at higher altitudes and less frequently at Nāpau Crater, we may see a decreased effect of helicopter noise on the vocalizing behavior of birds.

I predicted that soundscape indices would reflect vocalization differences at the community level between the periods before, during, and after helicopter noise. My study revealed community-level responses to helicopter noise, with a higher diversity of sounds produced by birds and crickets during the helicopter noise than after the helicopter noise. This was a surprising result given that two of three bird species decreased their vocalizations as helicopter noise increased at Nāpau Crater. However, because soundscape index results are averages of each recording station from all three sites, they may be revealing a broader pattern of how birds react to helicopter noise. Furthermore, I found that there is no difference in the Bioacoustic Index (BI) between the periods before and after helicopter noise, suggesting that birds recover relatively quickly from any effects of the noise. Increased BI values during helicopter noise was not due to the presence of helicopter noise in the recording because I only calculated BI values between 4 – 8 kHz and only included recordings with peak helicopter noise up to 4 kHz. All of the bird and cricket species at each site have a vocalization range that reaches at least 4.5 kHz, and thus their presence in the recording still contributes to the overall BI measurement. However, I may have found a stronger effect of noise period on BI if I had been able to take into account the vocalizations of bird species that primarily vocalize below 4 kHz. Although results revealed community-level differences in the biotic soundscape between periods of helicopter noise, I did not find significant differences in total vocalization time or vocalization type between periods of noise when I examined species individually. This finding was contrary to my prediction that birds would vocalize more before the onset of helicopter noise than during and after the noise.

Studying the effects of anthropogenic noise on wildlife from multiple perspectives allows for a broader understanding of how this relatively new addition to the natural soundscape is impacting wildlife and can lead to more effective management strategies. Birds depend on vocal cues to find mates and young, locate food, maintain territories, and communicate danger (Marler 2004; Habib et al. 2007; Catchpole and Slater 2008; Nemeth and Brumm 2010; Halfwerk et al.

16 2011). Consequently, disruptions in communication caused by anthropogenic noise can reduce fitness. Most studies examining the effect of noise on bird vocalizations focus on changes in amplitude and frequency of bird song in relation to noise (Dowling et al. 2011; Potvin and Mulder 2013; Roca et al. 2016). Others that have studied the effects of noise on temporal aspects of bird vocalizations have focused on changes in song rate or the cessation of song during noise exposure (Brumm 2006; Brumm et al. 2009; Halfwerk and Slabbekoorn 2009), or the relative start times of singing during periods of reduced ambient noise (Gil et al. 2015). Measuring temporal changes of bird vocalizations in response to a potential stressor can help us understand whether the stressor affects behavior, communication, and fitness. To my knowledge, my study is the first to examine temporal changes in songbird vocalizations during a power gradient of the same noise source. However, future studies following my methodology should control for time of day in their study design. Birdsong rates naturally fluctuate throughout the day (Van Riper III and Scott 1979), but I was not able to control for this within the limited scope of my study, which may have caused the high variability in my vocalization time data. Despite this variability, however, I was able that demonstrate that helicopter noise does indeed affect songbird vocalizations, and that the response to helicopter noise varies by species and degree of noise exposure. To expand on the results of this study, future studies could explore what changes in vocalizing behavior led to the observed changes in vocalization time with increasing power of noise. Are there changes in signal duration, rate, or redundancy? Additionally, future studies could explore whether the song structure of individuals in an area constantly exposed to noise differs from that of individuals in an area with no noise exposure. This could reveal the evolution of song traits that are adapted to provide release from masking effects, such as changes in song length, frequency and power.

In summary, I found that helicopter noise induces a temporal shift in songbird vocalizations, but this shift only occurs in areas with very loud and frequent helicopter traffic, as was hypothesized. Another hypothesis was that the amount of time birds spend vocalizing will change as helicopter noise changes in power and frequency range. I found that, at the species level, birds change the amount of time they spend vocalizing as helicopter noise increases in power, a behavior that may be an adaptation to mitigate effects of masking. This change in vocalization time provides further evidence that anthropogenic noise can cause birds to temporally adjust vocalizations (Brumm 2006; Halfwerk and Slabbekoorn 2009; Gil et al. 2015)

17 and suggests the presence of vocal plasticity in ‘Apapane, Warbling White-eye, and Japanese Bush-warbler. However, differences in vocal plasticity between species may also indicate differences in resilience to noise. I also hypothesized that the total vocalization time of each songbird species will differ between the periods before, during, and after the helicopter noise, and that soundscape indices will reflect vocalization differences at the community level between the same noise periods. Results showed that vocalization time did not differ between the helicopter noise periods at the species level, but at the community level, more species vocalized during the helicopter noise than after the noise. Additionally, community-level results showed that birds may recover relatively quickly from any effects of noise. However, helicopter noise could still be detrimentally affecting physiology, pairing and breeding success, and territory size of birds by limiting communication between individuals (Habib et al. 2007; Nemeth and Brumm 2010; Halfwerk et al. 2011; Kleist et al. 2018). These effects could have a greater impact on Hawaiian endemics, which already face a number of stressors (Atkinson and Lapointe 2009; Pratt et al. 2009; LaPointe et al. 2010), than on introduced species. To prevent further impacts of noise on Hawaiian endemic birds, my findings suggest that helicopter overflights should be managed to fly at higher altitudes and less frequently to see a decreased effect of helicopter noise on vocalizing behavior of birds in HAVO, especially at Nāpau Crater. More broadly, anthropogenic noise is abundant in United States protected natural areas, including areas considered critical habitat for U.S. endangered species (Buxton et al. 2018). These protected natural areas will not be fully completing their mission until anthropogenic noise is properly managed. Land managers can monitor the altitude and frequency of helicopter overflights over their lands to ensure they are at levels that will have the least amount of impact on wildlife. Results from my study suggest that an overflight frequency of less than 4 overflights per hour and mean power levels of 191 dB or less are sufficiently low to avoid detecting an effect of helicopter noise on bird vocalizations. However, the only way to ensure the absence of an effect is by completely prohibiting helicopter overflights above protected natural areas.

18 TABLES Table 1. Table depicting species used in each analysis, varying by site. Question 1 addressed whether total vocalization time (s) changed with increasing power of helicopter noise (dB), and Question 2 addressed whether total vocalization time (s) differed between ten-second periods “before,” “during,” and “after” helicopter noise. Question 1 Sites Question 2 Sites Species Nāpau Escape Rd. UWFR Nāpau Escape Rd. UWFR OMAO X X X X APAP X X X X X HAAM X X IIWI X X WAWE X X X X JABW X X X X

19 FIGURES

Figure 1. Study sites: Left) Hawaiʻi Island map indicating recording sites and Hawaiʻi Volcanoes National Park Boundary in white. Nāpau Crater is indicated by a red circle, Escape Rd. is indicated by a gold triangle, and UWFR is indicated by a green square. A) Recording stations at Nāpau Crater. B) Recording stations at Escape Rd. C) Recording stations at UWFR.

Figure 2. Predicted values of total vocalization time as a function of relative maximum power of helicopter noise for three species at Nāpau Crater. Predicted values are based on the generalized linear mixed model. Power values are inflated due to gain settings in recorders. Letters next to each line indicate species that are significantly different from one another based on a Tukey’s post-hoc analysis.

Figure 3. Interval plots depicting the total vocalization times for “before,” “during,” and “after” periods of helicopter noise for the most prominent bird species at a) Nāpau Crater, b) Escape Rd., and c) UWFR. Circles represent mean vocalization time and bars represent standard error.

Figure 4. Interval plots depicting the total vocalization time of a) Hawaiʻi ʻAmakihi, b) Warbling White-eye, and c) ʻŌmaʻo during the “before,” during,” and “after” periods of helicopter noise. Circles represent mean vocalization time and bars represent standard error. Letters above each bar indicate periods that are significantly different from one another based on a Tukey’s post-hoc analysis. Photo credits: Lars Petersson, Ann Tanimoto-Johnson, and Etienne Artigau.

Figure 5. Interval plot depicting mean Bioacoustic Index values of all sites, calculated for each recording station for “before,” “during,” and “after” periods of helicopter noise. Circles represent the mean of mean Bioacoustic Index values and bars represent the standard error. Letters above each bar indicate periods that are significantly different from one another based on a Tukey’s post-hoc analysis.

Figure 6. Interval plot showing the relative maximum helicopter noise power levels (dB) per site. Dots represent mean noise power levels and bars represent the standard error of the mean. Letters above each bar indicate sites that are significantly different from one another based on a Tukey’s post-hoc analysis. Power values (dB) are inflated due to gain settings of recorders.

Figure 7. Interval plot showing helicopter abundance per site. Dots represent mean helicoper abundance and bars represent the standard error of the mean. Letters above each bar indicate sites that are significantly different from one another based on a Tukey’s post-hoc analysis.

Figure 8. Predicted values of total vocalization time based on the generalized linear mixed model for five species at Escape Rd., a) ʻApapane, b) Warbling White-eye, c) Japanese Bush- warbler, d) Hawaiʻi ʻAmakihi, e) ʻŌmaʻo. Noise power values are inflated due to gain settings in recorders.

Figure 9. Predicted values of total vocalization time based on the generalized linear mixed model for five species at UWFR, a) ʻApapane, b) Warbling White-eye, c) Iʻiwi, d) Hawaiʻi ʻAmakihi, e) ʻŌmaʻo. Noise power values are inflated due to gain settings in recorders.

28 REFERENCES Andersen DE, Rongstad OJ, Mytton WR. 1989. Response of nesting red-tailed hawks to helicopter overflights. Condor. 91:296–299. André M, Solé M, Lenoir M, Durfort M, Quero C, Mas A, Lombarte A, Van der Schaar M, López-Bejar M, Morell M, et al. 2011. Low-frequency sounds induce acoustic trauma in cephalopods. Front Ecol Environ. 9(9):489–493. Atkinson CT, Lapointe DA. 2009. Introduced avian diseases, climate change, and the future of Hawaiian honeycreepers. Source J Avian Med Surg. 23(1):53–63. Barber JR, Crooks KR, Fristrup KM. 2010. The costs of chronic noise exposure for terrestrial organisms. Trends Ecol Evol. 25(3):180–189. Bates D, Mächler M, Bolker B, Walker S. 2015. Fitting Linear Mixed-Effects Models Using lme4. J Stat Softw. 67(1):1–48. Blumstein DT. 2014. Attention, habituation, and antipredator behaviour: implications for urban birds. In: Gil D, Brumm H, editors. Avian urban ecology. 1st ed. Oxford: Oxford University Press. p. 41–68. Boelman NT, Asner GP, Hart PJ, Martin RE. 2007. Multi-trophic invasion resistance in Hawai’i : bioacoustics, field Surveys, and airborne remote sensing. Ecol Appl. 0017(8):2137– 2144. Brown AL. 1990. Measuring the effect of aircraft noise on seabirds. Environ Int. 16:587–592. Brumm H. 2006. Signalling through acoustic windows: nightingales avoid interspecific competition by short-term adjustment of song timing. J Comp Physiol A. 192(12):1279–1285. Brumm H, Schmidt R, Schrader L. 2009. Noise-dependent vocal plasticity in domestic fowl. Anim Behav. 78:741–746. Brumm H, Slabbekoorn H. 2005. Acoustic communication in noise. Adv Study Behav. 35:151– 209. Brumm H, Zollinger SA. 2013. Avian vocal production in noise. In: Brumm H, editor. Animal communication and noise. 2nd ed. Heidelberg (Germany): Springer. p. 187–227. Buxton R, McKenna MF, Clapp M, Meyer E, Stabenau E, Angeloni LM, Crooks K, Wittemyer G. 2018. Efficacy of extracting indices from large-scale acoustic recordings to monitor biodiversity. Conserv Biol. 32(5):1174–1184. Buxton RT, McKenna MF, Mennitt D, Fristrup K, Crooks K, Angeloni L, Wittemyer G. 2017. Noise pollution is pervasive in U.S. protected areas. Science. 356(6337):531–533. Catchpole CK, Slater PJ. 2008. Bird song: biological themes and variations. 2nd ed. Cambridge, UK: Cambridge University Press. Conomy JT, Dubovsky JA, Collazo JA, Fleming WJ. 1998. Do black ducks and wood ducks habituate to aircraft disturbance? Source J Wildl Manag. 62(3):1135–1142.

29 Costello RA, Symes LB. 2014. Effects of anthropogenic noise on male signalling behaviour and female phonotaxis in Oecanthus tree crickets. Anim Behav. 95:15–22. Delaney DK, Grubb TG, Beier P, Pater LL, Reiser MH. 1999. Effects of helicopter noise on mexican spotted owls. Source J Wildl Manag. 63(1):60–76. DeRose-Wilson A, Fraser JD, Karpanty SM, Hillman MD. 2015. Effects of overflights on incubating Wilson’s plover behavior and heart rate. J Wildl Manage. 79(8):1246–1254. Díaz M, Parra A, Gallardo C. 2011. Serins respond to anthropogenic noise by increasing vocal activity. Behav Ecol. 22(2):332–336. Dooling RJ, Blumenrath SH. 2013. Avian sound perception in noise. In: Brumm H, editor. Animal communication and noise. 2nd ed. Heidelberg (Germany): Springer. p. 229–250. Dowling JL, Luther DA, Marra PP. 2011. Comparative effects of urban development and anthropogenic noise on bird songs. Behav Ecol. 23:201-209x. Farina A, Pieretti N, Piccioli L. 2011. The soundscape methodology for long-term bird monitoring: A Mediterranean Europe case-study. Ecol Inform. 6(6):354–363. Francis CD, Ortega CP, Cruz A. 2009. Noise pollution changes avian communities and species interactions. Curr Biol. 19:1415–1419. Gil D, Honarmand M, Pascual J, Pérez-Mena E, Macías Garcia C. 2015. Birds living near airports advance their dawn chorus and reduce overlap with aircraft noise. Behav Ecol. 26(2):435–443. Goudie RI, Jones LI. 2004. Dose-response relationships of harlequin duck behaviour to noise from low-level military jet over-flights in central Labrador. Environ Conserv. 31(4):289–298. Gramann J. 1999. The effect of mechanical noise and natural sound on visitor experiences in units of the national park system. NPS Soc Sci Res Rev. 1(1):1–16. Grubb TG, Bowerman WW. 1997. Variations in breeding bald eagle responses to jets, light planes and helicopters. J Raptor Res. 31(3):213–22. Habib L, Bayne EM, Boutin S. 2007. Chronic industrial noise affects pairing success and age structure of ovenbirds Seiurus aurocapilla. J Appl Ecol. 44(1):176–184. Halfwerk W, Holleman LJM, Lessells CkM, Slabbekoorn H. 2011. Negative impact of traffic noise on avian reproductive success. J Appl Ecol. 48(1):210–219. Halfwerk W, Slabbekoorn H. 2009. A behavioural mechanism explaining noise-dependent frequency use in urban birdsong. Anim Behav. 78(6):1301–1307. Kleist NJ, Guralnick RP, Cruz A, Lowry CA, Francis CD. 2018. Chronic anthropogenic noise disrupts glucorticoid signaling and has multiple effects on fitness in an avian community. PNAS. 115(9):E648-657. Krausman PR, Wallace MC, Hayes CL, DeYoung DW. 1998. Effects of jet aircraft on mountain sheep. J Wildl Manage. 62(4):1246.

30 LaPointe DA, Goff LM, Atkinson CT. 2010. Thermal constraints to the sporogonic development and altitudinal distribution of avian malaria plasmodium relictum in Hawai’i. J Parasitol. 96(2):318–324. Lenth R V. 2016. Least-Squares Means: The R Package lsmeans. J Stat Softw. 69(1):1–33. Mace BL, Bell PA, Loomis RJ, Haas GE. 2003. Source attribution of helicopter noise in pristine national park landscapes. J Park Recreat Admi. 21(3):97–119. Maier JAK, Murphy SM, White RG, Smith MD. 1998. Responses of caribou to overflights by low-altitude jet aircraft. J Wildl Manage. 62(2):752. Marler P. 2004. Bird calls: a cornucopia for communication. In: Marler P, Slabbekoorn H, editors. Nature’s Music: The Science of Birdsong. 1st ed. Elsevier, San Diego: Academic Press. p. 132–176. Nemeth E, Brumm H. 2010. Birds and anthropogenic noise: are urban songs adaptive? Am Nat. 176(4):445–475. Orci KM, Petróczki K, Barta Z. 2016. Instantaneous song modification in response to fluctuating traffic noise in the tree cricket Oecanthus pellucens. Anim Behav. 112:187–194. Patenaude NJ, Richardson WJ, Smultea MA, Koski WR, Miller GW, Wursig B, Greene CR. 2002. Aircraft sound and distrubance to bowhead and beluga whales during spring migration in the Alaskan Beaufort Sea. Mar Mammal Sci. 18(2):309–335. Pieretti N, Farina A. 2013. Application of a recently introduced index for acoustic complexity to an avian soundscape with traffic noise. J Acoust Soc Am. 134(1):891–900. Pijanowski BC, Villanueva-Rivera LJ, Dumyahn SL, Farina A, Krause BL, Napoletano BM, Gage SH, Pieretti N. 2011. Soundscape ecology: the science of sound in the landscape. Bioscience. 61(3):203–216. Potvin DA, Mulder RA. 2013. Immediate, independent adjustment of call pitch and amplitude in response to varying background noise by silvereyes (Zosterops lateralis). Behav Ecol. 24(6):1363–1368. Pratt TK, Atkinson CT, Banko PC. 2009. Conservation biology of Hawaiian forest birds: implications for island avifauna. Yale University Press. R Core Team. 2016. R: A Language and Environment for Statistical Computing. R Found Stat Comput. Van Riper III C, Scott JM. 1979. Observations on distribution, diet, and breeding of the Hawaiian Thrush. Condor. 81:65–71. Van Riper SG. 2000. Japanese White-eye (Zosterops japonicus). In: Poole A, Gill F, editors. Birds of North America. Ithaca, NY: Cornell Lab of Ornithology. Roca IT, Desrochers L, Giacomazzo M, Bertolo A, Bolduc P, Deschesnes R, Martin CA, Rainville V, Rheault G, Proulx R. 2016. Shifting song frequencies in response to anthropogenic noise: a meta-analysis on birds and anurans. Behav Ecol. 27(5):1269–1274.

31 Schafer MR. 1977. The tuning of the world: toward a theory of soundscape design. Philadelphia, PA: University of Pennsylvania Press. Schmidt R, Morrison A, Kunc HP. 2014. Sexy voices e no choices: male song in noise fails to attract females. Anim Behav. 94:55–59. Shannon G, McKenna MF, Angeloni LM, Crooks KR, Fristrup KM, Brown E, Warner KA, Nelson MD, White C, Briggs J, et al. 2016. A synthesis of two decades of research documenting the effects of noise on wildlife. Biol Rev. 91(4):982–1005. Singmann H, Bolker B, Westfall J, Aust F. 2019. afex: analysis of factorial experiments. Sueur J, Pavoine S, Hamerlynck O, Duvail S. 2008. Rapid acoustic survey for biodiversity appraisal. PLoS One. 3(12). Villanueva-Rivera LJ, Pijanowski BC, Doucette J, Pekin B. 2011. A primer of acoustic analysis for landscape ecologists. Landsc Ecol. 26(9):1233–1246. Weisenberger ME, Krausman PR, Wallace M, DeYoung DW, Maughan EO. 1996. Effects of simulated aircraft noise on heart-rate and behavior of desert ungulates. J Wildl Manage. 60(1):52–61.