The effect of anthropogenic noise on terrestrial wildlife is a relatively new area of study with broad ranging management implications. Noise has been identified as a disturbance that has the potential to induce behavioral responses in animals similar to those associated with predation risk. This study investigated potential impacts of a variety of human activities and their associated noise on the behavior of elk (Cervus elaphus) and pronghorn (Antilocapra americana) along a transportation corridor in Grand Teton National Park.
We conducted roadside scan surveys and focal observations of ungulate behavior while concurrently recording human activity and anthropogenic noise. Although we expected ungulates to be more responsive with greater human activity and noise, as predicted by the risk disturbance hypothesis, they were actually less responsive (less likely to perform vigilant, flight, traveling and defensive behaviors) with increasing levels of vehicle traffic, the human activity most closely associated with noise. Noise levels themselves had relatively little effect on ungulate behavior, although there was a weak negative relationship between noise and responsiveness in our scan samples. In contrast, ungulates did increase their responsiveness with other forms of anthropogenic disturbance; they reacted to the presence of pedestrians (in our scan samples) and to passing motorcycles (in our focal observations).
These findings suggest that ungulates did not consistently associate noise and human activity with an increase in predation risk or that they could not afford to maintain responsiveness to the most frequent human stimuli. Although reduced responsiveness to certain disturbances may allow for greater investment in fitness-enhancing activities, it may also decrease detections of predators and other environmental cues and increase conflict with humans.
Citation: Brown CL, Hardy AR, Barber JR, Fristrup KM, Crooks KR, Angeloni LM (2012) The Effect of Human Activities and Their Associated Noise on Ungulate Behavior. PLoS ONE 7(7): e40505. https://doi.org/10.1371/journal.pone.0040505
Editor: Matt Hayward, Australian Wildlife Conservancy, Australia
Received: December 7, 2011; Accepted: June 12, 2012; Published: July 10, 2012
Copyright: © 2012 Brown et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This study was funded by the United States National Park Service (RM-CESU Cooperative Agreement # H1200040001). Fellowship support was given to CLB by Colorado State University, Rocky Mountain Goats Foundation, and the National Science Foundation (DGE-0649876). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Anthropogenic noise can impact animals in ways that are only beginning to be explored . Noise is pervasive in both developed and natural areas ,  and can be deleterious to an animal's physiology and behavior. If chronic, it may affect an animal's auditory system , increase cardiac and stress levels , , and impair communication –. Noise can also alter pairing and reproduction , , age structuring , and density and occupancy patterns –.
Noise has also been identified as a disturbance that could induce behavioral responses similar to those associated with predation risk . The risk-disturbance hypothesis predicts that animals exposed to anthropogenic disturbance, such as noise, will exhibit antipredator behavior that takes time and energy away from fitness-enhancing activities . Indeed, prior studies have documented behavioral responses, such as vigilance, avoidance, and flight, to anthropogenic noise for a variety of taxa , –. An increase in vigilance may be costly if it results in a decrease in maintenance activities such as foraging , , and displacement or flight may expend valuable amounts of energy –. Thus, noise can affect habitat selection, foraging patterns, and overall energy budgets , , with potential population-level effects. However, noise may not have lasting negative effects if animals habituate to the disturbance, that is exhibit reduced responsiveness over time after repeated exposure without consequence ; e.g., , –. In some cases animals may even be attracted to and benefit from noisy disturbed areas, for example if they provide shelter from predators –.
Large mammals, such as ungulates, may be particularly sensitive to anthropogenic disturbance , , including human activities associated with recreation, transportation, ecotourism and the noise they produce , –. Recreational activities such as snowmobiling, skiing, biking and hiking can alter the behavior of ungulates , –. Roadways can also induce a range of behavioral responses in ungulates, which in some cases seem attracted to or unaffected by road activity ,  but more commonly exhibit risk-avoidance behavior in response to roads , , , , –. Although the degree to which animals are responding to visual or acoustic disturbances generated by these recreational and transportation activities remains largely unexplored, there is some evidence for the independent effect of noise, reviewed in , , ; but see .
The goal of this research was to quantify the behavioral response of ungulates to a variety of human activities and their associated noise along the primary travel corridor in Grand Teton National Park, USA. We evaluated the effect of human activities and concurrent sound properties on ungulate behavior along this corridor. If, according to the risk disturbance hypothesis , activities of park visitors represent a form of predation risk to ungulates, then we predicted ungulates would display heightened responsive behavior with increasing levels of anthropogenic stimuli, including both noise and human activity. Alternatively, the behavior of ungulates along the travel corridor could be unaffected by the level of noise and human activity if they have habituated to human disturbance over time or if sensitive individuals have been previously displaced from this location .
We conducted the study in summer 2008 along 22 km of Teton Park Road in Grand Teton National Park in northwestern Wyoming, USA (43–50′00″ N, 110–42′03″ W; Figure 1). Teton Park Road is located at the eastern base of the Teton Range and traverses the valley floor from north to south through a predominantly open sage-brush community where large ungulates congregate and visitors often stop to view wildlife. The study area included a stretch of Teton Park Road from its junction with Spalding Bay Drive to its junction with the town of Moose (Figure 1). Our research focused on the two ungulate species most prevalent along the road, elk (Cervus elaphus) and pronghorn (Antilocapra americana). Large numbers of elk (∼2,500–4,500 ) and pronghorn (∼200 ) spend the summer in Grand Teton National Park with the potential to move into and out of our study area. The behavior of both species may be influenced by predation risk in this system given the presence of carnivores within the park, including grizzly bear (Ursus arctos), black bear (Ursus americanus), gray wolf (Canis lupus), and mountain lion (Puma concolor), although these predators were only rarely observed in our study area. These ungulates also have the potential to experience hunting by humans, particularly when they venture outside our study area during the fall archery and rifle hunting seasons.
We recorded the behavior of individuals in ungulate herds through scan sampling at 42 points along Teton Park Road (Figure 1) from 14 June 2008 to 18 October 2008. We selected sampling points every 160 to 650 meters to standardize search efforts over space and time and to maximize visible area from the road in an attempt to include the entire viewshed along this stretch of Teton Park Road. Scan sampling occurred during both daytime and crepuscular hours, with staggered starting times to balance sampling effort across periods, allowing at least twelve hours between surveys.
To conduct scan sampling, we drove along Teton Park Road starting at either the northern or southern end of the study area and stopped at each sampling point to scan for ungulate herds with binoculars and a spotting scope. A herd was defined as ≥1 animal present, and a distance of 100 meters was used to delineate different herds, following Childress and Lung  who described this as the maximum distance at which elk respond to conspecific vocalizations. Once a herd was sighted, we noted the time of day and counted the number of individuals in the herd. We visually estimated whether the herd was clustered, with most individuals within 25 meters of a nearest neighbor, or dispersed, with most individuals greater than 25 meters from a nearest neighbor; we selected this threshold because it was relatively easy to detect visually and it divided our herds roughly evenly into clustered and dispersed categories. We used laser rangefinders to measure the distance to the center of the herd from the road (our vehicle) and the distance to closest vegetation cover, categorized as near or far to cover (using 100 m as the threshold, a distance across which elk vigilance patterns are known to change ).
Once the initial herd data were collected, we recorded behavior only if the herd was within 500 meters of the sampling point to ensure accuracy of behavioral observations. One observer scanned the herd from left to right recording the behavioral category of each individual, following , : feeding, grooming (licking or scratching), bedded, mating (sparring or bugling), traveling (walking), fleeing (running), scanning (standing with head above shoulder level), vigilant (displaying alarm or acute attention toward stimuli), and defensive (kicking, biting, charging). Scan surveys lasted approximately 1 minute. It is important to note that ungulates were not tagged or individually identified in our study area; thus, although we can be confident that we sampled unique individuals within each sampling bout as we moved along Teton Park Road, we cannot rule out the possibility that we observed the same individuals on multiple occasions across our scan and focal (described below) sampling bouts.
While ungulate behavioral data were collected, a second observer simultaneously conducted a scan sample to count different kinds of human activity within 200 meters of the sampling point. Ungulates have been shown to be sensitive to the approach speed and direction of anthropogenic stimuli ; therefore we categorized vehicles as moving versus stopped. Ungulates can also be particularly responsive to the human form ; therefore we also recorded the number of pedestrians along the road. Human activities recorded during scan samples included the number of automobiles (autos) passing, the number of autos stopped (including our own vehicle), and the number of pedestrians at each sampling point. Observers strove to remain in the vehicle to reduce potential observer effects, but on rare occasions when it was necessary to exit the vehicle during a scan observation (e.g., to see a herd that was partially obscured from view), we recorded the observer as a pedestrian to account for our presence and potential influence. We also recorded whether motorcycles, trucks (including recreational vehicles and large commercial and construction vehicles), and bicycles were passing but rarely recorded these activities during our scan samples. Consequently, we did not analyze these three activities separately, but rather grouped passing motorcycles, trucks, and autos into an additional category (total vehicles passing) and omitted passing bicycles from the analyses.
Concurrent with the ungulate and human behavioral observations, we used a portable recording device to sample noise. The recorder (iAudio 7, Cowan America, Irvine, California) was attached to PA3 microphones and a horn lens. The device was mounted on our research vehicle approximately 1.5 meters off the ground and microphones were spaced 2 meters apart pointing in opposite directions. The consistently close proximity of the recorder to the road allowed us to effectively record motorized vehicles, road noise, bicycles, and pedestrians (i.e., human voices). We used a sampling rate of 64 bits per second and recordings were saved as uncalibrated WMA files that could be analyzed for relative metrics of sound. We produced waveforms using SWITCH sound file converter (NCH Software, Canberra, Australia) and spectrograms using RAVEN PRO 1.4 (Cornell University, Ithaca, New York) to quantify relative sound metrics. As the perception of loudness depends on both the amplitude and frequency of sound waves, we measured average power, or the mean relative amplitude over the entire observation, and peak frequency, or the frequency at which the maximum power occurred.
Focal Animal Sampling.
In addition to scan sampling, we conducted extended behavioral observations of individual focal animals. We initiated focal animal sampling opportunistically, between scan sampling events, as well as systematically, during scheduled daytime and crepuscular focal animal sessions. Observers drove the length of the study area searching for ungulate herds. When a herd was sighted within 500 meters of the road, we recorded its dispersion and location. We randomly selected a focal animal within a herd by counting individuals in the herd from left to right until reaching a chosen random number, and we recorded its sex classification (adult male, adult female or adult female with calf, if a female was in close proximity to or seen tending to a calf). The focal animal observer continuously recorded the behavioral state (same categories as described above) and the timing of any changes in behavioral state for up to 50 minutes or until the focal animal bedded or moved out of view. We excluded focal animal samples with a duration less than 3 minutes (following Childress and Lung ) resulting in an average sample duration of 14.6 minutes (SE = 0.8, n = 113).
As with scan samples, we continuously recorded sound for the duration of the focal sample to measure average power and peak frequency. Simultaneously, a second observer alternated between conducting scan samples of behavior for all individuals within the herd and conducting scan samples to count human activities in the vicinity (within 200 meters of the observers). The alternating herd and human activity scans continued throughout the duration of the focal animal sample, with repeated intervals of approximately 45 seconds to 3 minutes; the duration and frequency of scan samples were dependent on herd size and amount of human activity in the vicinity. The herd behavioral scans were conducted for a concurrent study (Hardy, unpublished data); we use only the human activity data here. Anthropogenic activities recorded during focal samples included the number of autos, motorcycles, trucks, and bicycles passing; the number of autos stopped; and the number of pedestrians present.
We developed a candidate set of nonlinear mixed models with a binomial distribution (Proc NLMixed, SAS 9.1) to evaluate if and how acoustic variables and human activities predicted the probability that each individual within a herd was responding or not responding, expressed as a binary, categorical variable. Individuals were classified as ‘responding’ if they were vigilant, if they displayed defensive behavior, or if they were fleeing or traveling , . Although animals may travel for a variety of reasons, human activity has been observed to provoke movement in general ,  and walking in particular , – in a variety of ungulates, including elk within this Greater Yellowstone Ecosystem , .
Our candidate models included all combinations of five acoustic and human activity predictor variables (average power, peak frequency, total vehicles passing, autos stopped, and pedestrians present). Each model additionally included all of the following covariates that have been shown to influence responsive behavior in ungulates , , , , : distance to road, distance to cover, dispersion (clustered versus dispersed), herd size, species (pronghorn or elk), Julian date, and time of day (crepuscular: ≤1 hour after dawn or prior to dusk, or daytime: >1 hour after dawn or prior to dusk, as determined by regional sunrise and sunset tables). We also included the herd ID (a number from 1 to 161 assigned to each scan sample) as a random effect in each model to avoid statistical issues related to pseudoreplication, since an individual's behavior within a scan sample may be correlated with the behavior of the other animals scanned within the same herd. Our candidate model set included an intercept-only model, a covariate model, and models with all subsets of acoustic and human activity predictors in addition to the covariates.
AICc (Akaike Information Criterion adjusted for small sample size)  based on likelihood values were calculated for each model of ungulate herd responsiveness. We reported model weights (wi) and AICc differences (Δ), measuring the information loss between models given the data, to compare model ranking. Because our model set was balanced by including all combinations of acoustic and human activity variables, we were able to calculate relative variable importance weights (sum of model weights for all models containing that specific variable) to determine which of these variables were the strongest predictors of ungulate responsiveness , . For each predictor, we also calculated model-averaged parameter estimates and their associated 95% confidence intervals to account for model selection uncertainty and to provide unconditional estimates not dependent on a single model . However, because model-averaging might not reliably assess the effect of a single predictor variable , , we also reported parameter estimates for the predictors in the top model, which necessarily provide conditional estimates, and we calculated estimates from the relationship between each sole predictor and responsiveness, which produce estimates that are not conditional on other predictors.
Focal animal sampling.
We used linear regressions (Proc Genmod, SAS 9.1) to evaluate the relationship between behavioral budgets of individual animals in the focal observations and acoustic and human activity. For these analyses, the sampling unit was the focal animal and our response variable was the proportion of time spent responding (i.e., vigilant, defensive, fleeing, traveling). Proportionate data was square root arcsine transformed to normalize variance prior to analyses. We calculated overall rates for human activity variables, averaged across all human scans that occurred during a focal observation (i.e., mean number of activities per scan), to adjust for variation in the number of human activity scans conducted while observing focal animals.
To predict focal animal responsiveness, we created candidate models with all combinations of acoustic and human activity predictors (in addition to an intercept-only model and a model with just the covariates), using similar variables as for the scan samples. However, we separated the total passing vehicles into passing autos and motorcycles, and we also included passing bicycles as a distinct predictor, because they were recorded in sufficient frequency in our focal samples due to their longer duration; this resulted in a total of seven acoustic and human activity predictors. All candidate models included the same covariates as in the scan samples, including distance to road, distance to cover, dispersion, herd size, species, Julian date, and time of day. Past studies suggest the sex of an individual may also affect responsiveness , ; thus we additionally included the focal animal's sex classification. As with the scan samples, we reported AICc values, model weights, and parameter estimates and confidence intervals from the top model, from model averaging, and from a model where each variable was the sole predictor; variable importance weights were also calculated to determine which acoustic and human activity variables were the strongest predictors of ungulate responsiveness.
Across 161 scan samples, we observed a total of 334 autos stopped, 265 total vehicles passing (including 245 autos, 11 trucks, 9 motorcycles), 135 pedestrians, and 4 bicycles passing. Our uncalibrated measures of average power during scan samples ranged from 37.8 dB to 80.9 dB (mean = 64.9, SE = 0.9). Peak frequency ranged from 172 to 4307 Hz, falling within the hearing range of ungulates , and averaged 958 Hz (SE = 41), consistent with the low frequency of traffic noise . Of all human activities measured, the number of autos passing was most strongly correlated with average power measurements during scan samples (r = 0.37), further pointing to traffic as a dominant source of noise. Of 1013 ungulates scanned across all scan samples, 234 (23%) were engaged in responsive behavior (14% traveling, 7% vigilant, 2% fleeing, and 0.2% defensive).
When comparing our candidate models predicting ungulate responsiveness, there was some model selection uncertainty (Table 1) with substantial support for the top 3 models (out of 33) that fell within 2.0 ΔAICc ; these top models contained all acoustic and human activity predictors except peak frequency. Based on the magnitude and direction of parameter estimates, ungulates were more likely to respond when there were more pedestrians present and less likely to respond when there were high levels of traffic, with traffic having a greater effect than pedestrians (Table 2). The 95% confidence intervals around the parameter estimates for total vehicles passing and pedestrians did not overlap zero in the top model or from model averaging, further suggesting that they both influenced responsiveness. The parameter estimate for average power was relatively small, and its 95% confidence interval overlapped zero when model averaging but not when average power was the sole predictor, suggesting only a weak negative relationship between noise and responsiveness. The parameter estimates for autos stopped and peak frequency were also small, with confidence intervals overlapping zero both from model averaging and when they were the only predictors (Table 2). Comparing the importance weights of the acoustic and human activity variables confirmed that the number of vehicles passing and pedestrians were relatively more important predictors of ungulate responsiveness than average power, the numbers of autos stopped, and peak frequency (Table 2). Based on the magnitude and directions of parameter estimates for the covariates, ungulates were more likely to respond when herds were dispersed, were closer to the road, and were composed of pronghorn, with at least one confidence interval that did not overlap zero from the top model, model averaging, or the model with a single predictor (Table 2).
We conducted 113 focal samples throughout the field season generating 1,632 minutes of individual observations. We observed 3,275 autos stopped, 3,040 vehicles passing (including 2,786 autos, 171 trucks, 83 motorcycles), 1,047 pedestrians, and 41 bicycles passing summed over 2,172 human activity scans that were concurrent with the 113 focal observations. Our uncalibrated measures of average power during focal samples ranged between 57.0 dB and 77.0 dB (mean = 69.2, SE = 0.4), while peak frequency ranged between 172 and 11,887 Hz (mean = 958, SE = 74.1). Of the human activities measured, the number of autos passing was most strongly correlated with average power during focal samples (r = 0.54), again implicating auto traffic as a major source of noise. On average, focal ungulates spent 25% (SE = 2%) of their time engaged in responsive behavior (13% traveling, 8% vigilant, 4% fleeing, 0.1% defensive).
When comparing our candidate models predicting ungulate responsiveness, there was considerable model selection uncertainty (Table 3), with substantial support for the top 8 models (out of 129) that fell within 2.0 ΔAICc ; these top models contained all acoustic and human activity predictors. Based on the magnitude and direction of parameter estimates in the most strongly supported models, focal animals increased their responsiveness with increasing motorcycle traffic and decreased their responsiveness with increasing auto traffic, with motorcycles having a larger effect size than autos (Table 4). The 95% confidence intervals around the parameter estimates for these two predictors did not overlap zero in the top model, further suggesting they influenced responsiveness. In contrast, the parameter estimates for the other acoustic and human activity variables in the top model (average power and pedestrians) were relatively small, with confidence intervals that overlapped zero (Table 4). All model-averaged parameter estimates of acoustic and human activity variables were smaller than those from the highest-ranking models, with confidence intervals overlapping zero, suggesting that they did not strongly influence responsiveness, though this could be attributed to averaging over many models with high uncertainty, which may reduce the ability to correctly estimate the effect of a single predictor , . Comparing the relative importance weights of the acoustic and human activity predictors revealed that the number of autos passing was the most important predictor of ungulate responsiveness followed by the number of motorcycles passing. Average power, pedestrians, autos stopped, peak frequency, and bicycles passing were relatively less important (Table 4). Further, based on the magnitude and directions of parameter estimates for the covariates (with at least one confidence interval that did not overlap zero from model averaging or the single-predictor model), ungulates were more responsive in smaller herds and during daytime hours, and cows with a calf were more responsive than males or females without a calf (Table 4).
The risk-disturbance hypothesis states that anthropogenic disturbance such as human-related presence, objects, or sounds will elicit antipredator behavior . Thus, we expected ungulates to exhibit heightened levels of responsive behavior in the presence of human activities and noise along Teton Park Road in Grand Teton National Park. The results suggest that human activities can alter responsive behaviors in ungulates. Contrary to our predictions, however, ungulates were not more likely to respond, but rather less likely to respond to increased vehicle traffic, which was the human activity most closely associated with noise. Though noise levels themselves did not have a strong effect on ungulate behavior, there was a weak negative relationship between average power and responsiveness in our scan samples.
One possible explanation for these findings is that ungulates in our study area did not perceive traffic and its associated noise as a form of predation risk, perhaps because individuals sensitive to these stimuli have been displaced over time or because the individuals that remain have habituated over time to these frequent stimuli. Ungulates are known to habituate to regular exposure to noise ,  and other non-lethal human activities  and to display individual variation within populations in their avoidance or tolerance of roads . Elk in particular exhibit behavioral patterns that suggest habituation along roads and other areas disturbed by human activities –. This tolerance would explain a lack of effect of traffic on responsiveness, but does not seem sufficient to explain the finding that increasing traffic caused ungulates to be less responsive.
The decrease in responsiveness with increasing traffic could indicate that passing vehicles provide a refuge from predators, such that ungulates have come to perceive reduced predation risk when traffic and their associated noise levels are high. Previous studies have demonstrated direct benefits of human activity to prey through reduced predator abundance , –, and it is possible that this could also translate to indirect benefits through reduced investment in vigilance and other forms of antipredator behavior. Alternatively, another explanation for our findings is that traffic disturbances are actually perceived as a form of predation risk by ungulates, but they cannot afford to maintain high levels of responsiveness to such a continuous and pervasive form of disturbance. Specifically, the risk allocation hypothesis  suggests that animals will devote a larger proportion of risky intervals to antipredator behavior, when those intervals are brief and infrequent. In contrast, when periods of risk are lengthy and more frequent, animals will devote a reduced proportion of those risky intervals to antipredator behavior in order to avoid the high cost of lost foraging. In the context of anthropogenic disturbance, Miller et al.  found certain human activities, when infrequent and unpredictable, were related to heightened levels of flush distance in ungulates. In our study, auto traffic, with its associated noise, was the most prevalent anthropogenic disturbance; thus, high traffic levels may have reduced responsiveness due to risk allocation decisions. In comparison, pedestrians, a less frequent form of disturbance, were more likely to elicit responsive behavior in our scan samples, consistent with prior studies implicating the human form as an importance source of disturbance for ungulates . Similarly, responsiveness was greater in response to the least common form of disturbance, motorcycle traffic, as would be predicted by the risk allocation hypothesis. Interestingly, bicycles, which are quieter but similar in shape to motorcycles, were not an important predictor of responsive behavior, suggesting that the loud noise generated by motorcycles in particular may be a disturbance stimulus.
Although the goal of this study was to evaluate whether anthropogenic disturbances affected ungulate behavior, we also measured a variety of covariates for inclusion in our models. The directions of their effects on responsiveness supports earlier findings that ungulates were more responsive when they were in smaller herds, when they were dispersed rather than clustered, and when they were closer to roads, further suggesting they were not completely tolerant of human activity , , , . Our results also suggest that ungulates may be more responsive during daytime hours; this adds to prior findings that time of day influences responsiveness, though the direction of the effect varies across ungulate species and populations, including elk , . Pronghorn were more responsive than elk, and females with young were more responsive than adult males and adult females without young, again consistent with prior studies demonstrating the sensitivity of pronghorn , ,  and of females with young , ,  to disturbance.
Understanding the behavioral responses of wildlife to anthropogenic disturbance can have important conservation and management implications –. Our results highlight an interesting effect of human disturbance on behavior. Except in the case of motorcycles, which are relatively infrequent disturbance events, ungulates spent less time responding with increased vehicle traffic and its associated noise, allowing more time for maintenance activities such as feeding. Presumably, increased levels of energy enhancing activities can positively affect fitness, suggesting a benefit of reduced responsiveness to traffic. However, we urge caution with this interpretation, since unresponsive behavior also could have negative implications, for example by reducing their ability to visually detect predators and other cues in the environment, potentially adding to any masking of acoustic cues caused by the anthropogenic noise itself . Reduced responsiveness of ungulates to road traffic could also lead to increased human conflict such as negative encounters with recreationists or collisions with vehicles , , major concerns for park managers . Finally, it is important to emphasize that noise can have negative impacts on fitness and population persistence in ways that may not be reflected by individual behavioral responsiveness . Thus, although anthropogenic noise did not appear to detract from fitness-enhancing behaviors in this system, we suggest continued investigation of possible population-level noise impacts.
We would like to thank Grand Teton National Park, including S. Burson, S. Cain, and S. Dewey, for permission to conduct this research and logistical support. The National Park Service Natural Sounds and Night Skies Division and the Cornell Lab of Ornithology provided recording equipment and valuable assistance with acoustical software and analyses. We also thank M. Scott and K. Van Ort for their help with the collection and organization of data. We are grateful to J. Eickoff for providing statistical guidance. The comments of M. Hayward and two anonymous reviewers improved the manuscript considerably. The National Park Service, the National Science Foundation, Colorado State University and the Rocky Mountain Goats Foundation provided support for this project.
Conceived and designed the experiments: CLB ARH JRB KMF KRC LMA. Performed the experiments: CLB ARH. Analyzed the data: CLB ARH JRB KMF KRC LMA. Contributed reagents/materials/analysis tools: CLB ARH JRB KMF KRC LMA. Wrote the paper: CLB ARH JRB KMF KRC LMA.
- 1. Barber JR, Crooks KR, Fristrup KM (2010) The costs of chronic noise exposure for terrestrial organisms. Trends in Ecology and Evolution 25: 180–189.
- 2. Barber JR, Burdett CL, Reed SE, Warner KA, Formichella C, et al. (2011) Anthropogenic noise exposure in protected natural areas: estimating the scale of ecological consequences. Landscape Ecology 26: 1281–1295.
- 3. Lynch E, Joyce D, Fristrup K (2011) An assessment of noise audibility and sound levels in U.S. National Parks. Landscape Ecology 26: 1297–1309.
- 4. Henderson D, Hamernik RP, Dosanjh DS, Mills JH (1976) Effects of noise on hearing. New York: Raven Press. 565 p.
- 5. Weisenberger ME, Krausman PR, Wallace MC, De Young DW, Maughan OE (1996) Effects of simulated jet aircraft noise on heart rate and behavior of desert ungulates. Journal of Wildlife Management 60: 52–61.
- 6. Owen MA, Swaisgood RR, Czekala NM, Steinman K, Lindburg DG (2004) Monitoring stress in captive giant pandas (Ailuropoda melanoleuca): behavioral and hormonal responses to ambient noise. Zoo Biology 23: 147–164.
- 7. Warren PS, Katti M, Ermann M, Brazel A (2006) Urban bioacoustics: it's not just noise. Animal Behaviour 71: 491–502.
- 8. Bee MA, Swanson EM (2007) Auditory masking of anuran advertisement calls by road traffic noise. Animal Behaviour 74: 1765–1776.
- 9. Habib L, Bayne EM, Boutin S (2007) Chronic industrial noise affects pairing success and age structure of ovenbirds Seiurus aurocapilla. Journal of Applied Ecology 44: 176–184.
- 10. Lengagne T (2008) Traffic noise affects communication behaviour in a breeding anuran, Hyla arborea. Biological Conservation 141: 2023–2031.
- 11. Slabbekoorn H, Ripmeester EAP (2008) Birdsong and anthropogenic noise: implications and applications for conservation. Molecular Ecology 17: 72–83.
- 12. Halfwerk W, Holleman LJM, Lessells CM, Slabbekoorn H (2011) Negative impact of traffic noise on avian reproductive success. Journal of Applied Ecology 48: 210–219.
- 13. Bayne EM, Habib L, Boutin S (2008) Impacts of Chronic Anthropogenic Noise from energy-sector activity on abundance of songbirds in the boreal forest. Conservation Biology 22: 1186–1193.
- 14. Francis CD, Ortega CP, Cruz A (2009) Noise pollution changes avian communities and species interactions. Current Biology 19: 1415–1419.
- 15. Francis CD, Paritsis J, Ortega CP, Cruz A (2011) Landscape patterns of avian habitat use and nest success are affected by chronic gas well compressor noise. Landscape Ecology 26: 1269–1280.
- 16. Frid A, Dill LM (2002) Human-caused disturbance stimuli as a form of predation risk. Conservation Ecology 6: 11.
- 17. Quinn JL, Whittingham MJ, Butler SJ, Cresswell W (2006) Noise, predation risk compensation and vigilance in the chaffinch Fringilla coelebs. Journal of Avian Biology 37: 601–608.
- 18. Rabin LA, Coss RG, Owings DH (2006) The effects of wind turbines on antipredator behavior in California ground squirrels (Spermophilus beecheyi). Biological Conservation 131: 410–420.
- 19. Nowacek DP, Thorne LH, Johnston DW, Tyack PL (2007) Responses of cetaceans to anthropogenic noise. Mammal Review 37: 81–115.
- 20. Maier JAK, Murphy SM, White RG, Smith MD (1998) Responses of caribou to overflights by low-altitude jet aircraft. Journal of Wildlife Management 62: 752–766.
- 21. Childress MJ, Lung MA (2003) Predation risk, gender and the group size effect: does elk vigilance depend upon the behaviour of conspecifics? Animal Behaviour 66: 389–398.
- 22. Fortin D, Boyce MS, Merrill EH, Fryxell JM (2004) Foraging costs of vigilance in large mammalian herbivores. Oikos 107: 172–180.
- 23. Ydenberg RC, Dill LM (1986) The economics of fleeing from predators. Advances in the Study of Behavior 16: 229–249.
- 24. Cassirer EF, Freddy DJ, Ables ED (1992) Elk responses to disturbance by cross-country skiers in Yellowstone National Park. Wildlife Society Bulletin 20: 375–381.
- 25. Rumble MA, Benkobi L, Gamo RS (2005) Elk responses to humans in a densely roaded area. Intermountain Journal of Sciences 11: 10–24.
- 26. Stockwell CA, Bateman GC, Berger J (1991) Conflicts in national parks: a case study of helicopters and bighorn sheep time budgets at the Grand Canyon. Biological Conservation 56: 317–328.
- 27. Bradshaw CJA, Boutin S, Hebert DM (1998) Energetic implications of disturbance caused by petroleum exploration to woodland caribou. Canadian Journal of Zoology 76: 1319–1324.
- 28. Thorpe WH (1963) Learning and instinct in animals. London: Methuen. 558 p.
- 29. Harms CA, Fleming WJ, Stoskopf MK (1997) A technique for dorsal subcutaneous implantation of heart rate biotelemetry transmitters in black ducks: Application in an aircraft noise response study. The Condor 99: 231–237.
- 30. Delaney DK, Grubb TG, Beier P, Pater LL, Reiser MH (1999) Effects of helicopter noise on Mexican spotted owls. Journal of Wildlife Management 63: 60–76.
- 31. Waynert DF, Stookey JM, Schwartzkopf-Genswein KS, Watts JM, Waltz CS (1999) The response of beef cattle to noise during handling. Applied Animal Behviour Science 62: 27–42.
- 32. Berger J (2007) Fear, human shields and the redistribution of prey and predators in protected areas. Biology Letters 3: 620–623.
- 33. Fahrig L, Rytwinski T (2009) Effects of roads on animal abundance: an empirical review and synthesis. Ecology and Society 14: 21.
- 34. Leighton PA, Horrocks JA, Kramer DL (2010) Conservation and the scarecrow effect: Can human activity benefit threatened species by displacing predators? Biological Conservation 143: 2156–2163.
- 35. Muhly TB, Semeniuk C, Massolo A, Hickman L, Musiani M (2011) Human activity helps prey win the predator-prey space race. PLoS ONE 6: e17050.
- 36. Stankowich T (2008) Ungulate flight responses to human disturbance: A review and meta-analysis. Biological Conservation 141: 2159–2173.
- 37. Bolger TD, Newmark D, Morrison TA, Doak DF (2008) The need for integrative approaches to understand and conserve migratory ungulates. Ecology Letters 11: 63–77.
- 38. Duchesne M, Côté SD, Barrette C (2000) Responses of woodland caribou to winter ecotourism in the Charlevoix Biosphere Reserve, Canada. Biological Conservation 96: 311–317.
- 39. Gavin SD, Komers PE (2006) Do pronghorn (Antilocapra americana) perceive roads as a predation risk? Canadian Journal of Zoology 84: 1775–1780.
- 40. St. Clair CC, Forrest A (2009) Impacts of vehicle traffic on the distribution and behaviour of rutting elk, Cervus elaphus. Behaviour 146: 393–413.
- 41. Schultz RD, Bailey JA (1978) Responses of national park elk to human activity. Journal of Wildlife Management 42: 91–100.
- 42. Freddy DJ, Bronaugh WM, Fowler MC (1986) Responses of mule deer to disturbance by persons afoot and snowmobiles. Wildlife Society Bulletin 14: 63–68.
- 43. Knight RL, Cole DN (1995) Wildlife responses to recreationists. In: Knight RL, Gutzwiller K, editors. pp. 51–70. Washington, DC: Island Press.
- 44. Phillips GE, Alldredge AW (2000) Reproductive success of elk following disturbance by humans during calving season. Journal of Wildlife Management 64: 521–530.
- 45. Manor R, Saltz D (2003) Impact of human nuisance disturbance on vigilance and group size of a social ungulate. Ecological Applications 13: 1830–1834.
- 46. Taylor AR, Knight RL (2003) Wildlife responses to recreation and associated visitor perceptions. Ecological Applications 13: 951–963.
- 47. Borkowski JJ, White PJ, Garrott RA, Davis T, Hardy AR, et al. (2006) Behavioral responses of bison and elk in Yellowstone to snowmobiles and snow coaches. Ecological Applications 16: 1911–1925.
- 48. Setsaas TH, Holmern T, Mwakalebe G, Stokke S, Røskaft E (2007) How does human exploitation affect impala populations in protected and partially protected areas? – A case study from the Serengeti ecosystem, Tanzania. Biological Conservation 136: 563–570.
- 49. Rost GR, Bailey JA (1979) Distribution of mule deer and elk in relation to roads. Journal of Wildlife Management 43: 634–641.
- 50. Pedevillano C, Wright RG (1987) The influence of visitors on mountain goat activities in Glacier National Park, Montana. Biological Conservation 39: 1–11.
- 51. Czech B (1991) Elk behavior in response to human disturbance at Mount St. Helens National Volcanic Monument. Applied Animal Behaviour Science 29: 269–277.
- 52. Rowland MM, Wisdom MJ, Johnson BK, Kie JG (2000) Elk distribution and modeling in relation to roads. Journal of Wildlife Management 64: 672–684.
- 53. Rumble MA, Gamo RS (2011) Resource selection by elk at two spatial scales in the Black Hills, South Dakota. Prairie Naturalist 43: 3–13.
- 54. Summers PD, Cunnington GM, Fahrig L (2011) Are the negative effects of roads on breeding birds caused by traffic noise? Journal of Applied Ecology 48: 1527–1534.
- 55. Bejder L, Samuels A, Whitehead H, Finn H, Allen S (2009) Impact assessment research: use and misuse of habituation, sensitisation and tolerance in describing wildlife responses to anthropogenic stimuli. Marine Ecology Progress Series 395: 177–185.
- 56. Smith BL (2007) Migratory behavior of hunted elk. Northwest Science 81: 251–264.
- 57. Berger J (2003) Is it acceptable to let a species go extinct in a National Park? Conservation Biology 17: 1451–1454.
- 58. Liley S, Creel S (2008) What best explains vigilance in elk: characteristics of prey, predators, or the environment? Behavioral Ecology 19: 245–254.
- 59. Goldstein MI, Poe AJ, Cooper E, Youkey D, Brown BA, et al. (2005) Mountain goat response to helicopter overflights in Alaska. Wildlife Society Bulletin 33: 688–699.
- 60. Dorrance MJ, Savage PJ, Huff DE (1975) Effects of snowmobiles on white-tailed deer. Journal of Wildlife Management 39: 563–569.
- 61. Tyler NJC (1991) Short-term behavioural responses of Svalbard reindeer Rangiger tarandus platyrhynchus to direct provocation by a snowmobile. Biological Conservation 56: 179–194.
- 62. Papouchis CM, Singer FJ, Sloan WB (2001) Responses of desert bighorn sheep to increased human recreation. Journal of Wildlife Management 65: 573–582.
- 63. Fortin D, Andruskiw M (2003) Behavioral response of free-ranging bison to human disturbance. Wildlife Society Bulletin 31: 804–813.
- 64. Cleveland SM (2010) Human predation risk infuences elk behavior in heterogeneous landscapes. MS Thesis. University of Montana. Missoula. 92 p. Accessed 2012 Jun 18.
- 65. Roberts G (1996) Why individual vigilance declines as group size increases. Animal Behaviour 51: 1077–1086.
- 66. Frid A (1997) Vigilance by female Dall's sheep: Interactions between predation risk factors. Animal Behaviour 53: 799–808.
- 67. Burnham KP, Anderson DR (2002) Model selection and multimodel inference: a practical information-theoretic approach. New York: Springer-Verlag. 496 p.
- 68. Doherty PF, White GC, Burnham KP (2010) Comparison of model building and selection strategies. Journal of Ornithology 152: S317–S323.
- 69. Richards SA, Whittingham MJ, Stephens PA (2011) Model selection and model averaging in behavioural ecology: the utility of the IT-AIC framework. Behavioural Ecology and Sociobiology 65: 77–89.
- 70. Symonds MRE, Moussalli A (2011) A brief guide to model selection, multimodal inference and model averaging in behavioural ecology using Akaike's information criterion. Behavioural Ecology and Sociobiology 65: 13–21.
- 71. Lipetz VE, Bekoff M (1982) Group-size and vigilance in pronghorns. Journal of Comparative Ethology 58: 203–216.
- 72. Lima SL (1998) Stress and decision making under the risk of predation: Recent developments from behavioral, reproductive, and ecological perspectives. Advances in the Study of Behavior 27: 215–290.
- 73. D'Angelo GJ, De Chicchis AR, Osborn DA, Gallagher GR, Warren RJ, et al. (2007) Hearing range of white-tailed deer as determined by auditory brainstem response. Journal of Wildlife Management 71: 1238–1242.
- 74. Dooling RJ, Popper AN (2007) The effects of highway noise on birds. Environmental BioAcoustics Report. California Department of Transportation website. Accessed 2012 Jun 18.
- 75. Lyon LJ, Ward AL (1982) Elk and land management. In: Thomas JW, Toweill DE, editors. pp. 442–477. Harrisburg: Stackpole Books.
- 76. Morrison JR, De Vergie WJ, Alldredge AW, Byrne AE, Andree WW (1995) The effects of ski area expansion on elk. Wildlife Society Bulletin 23: 481–489.
- 77. Thompson MJ, Henderson RE (1998) Elk habituation as a credibility challenge for wildlife professionals. Wildlife Society Bulletin 26: 477–483.
- 78. Lima SL, Bednekoff PA (1999) Temporal variation in danger drives antipredator behavior: the predation risk allocation hypothesis. American Naturalist 153: 649–659.
- 79. Miller SG, Knight RL, Miller CK (2001) Wildlife responses to pedestrians and dogs. Wildlife Society Bulletin 29: 124–132.
- 80. Berger J, Daneke D, Johnson J, Berwick SH (1983) Pronghorn foraging economy and predator avoidance in a desert ecosystem: Implications for the conservation of large mammalian herbivores. Biological Conservation 25: 193–208.
- 81. Wolff JO, Van Horn T (2003) Vigilance and foraging patterns of American elk during the rut in habitats with and without predators. Canadian Journal of Zoology 81: 266–271.
- 82. Buchholz R (2007) Behavioural biology: an effective and relevant conservation tool. Trends in Ecology and Evolution 22: 401–407.
- 83. Caro T (2007) Behavior and conservation: a bridge too far? Trends in Ecology and Evolution 22: 394–400.
- 84. Angeloni L, Crooks KR, Blumstein DT (2010) Conservation and behavior: Introduction. In: Breed MD, Moore J, editors. pp. 377–381. Oxford: Academic Press.
- 85. Angeloni L, Schlaepfer MA, Lawler JJ, Crooks KR (2008) A reassessment of the interface between conservation and behaviour. Animal Behaviour 75: 731–737.
- 86. Blumstein DT, Fernández-Juricic E (2010) A primer of conservation behavior. Sunderland: Sinauer Associates. 224 p.
- 87. Conover MR (2002) Resolving human-wildlife conflicts: the science of wildlife damage management. Boca Raton: CRC Press. 440 p.
- 88. Ament R, Clevenger AP, Yu O, Hardy A (2008) An assessment of road impacts on wildlife populations in US National Parks. Environmental Management 42: 480–496.
- 89. Gill JA, Norris K, Sutherland WJ (2001) Why behavioural responses may not reflect the population consequences of human disturbance. Biological Conservation 97: 265–268.