https://cpw.cvlcollections.org/files/original/1e7ed7825852c86e2e3543dccde85d18.pdf 8fdb7a690eeddb6b80917318d01fd605 PDF Text Text The research in this publication was partially or fully funded by Colorado Parks and Wildlife. Dan Prenzlow, Director, Colorado Parks and Wildlife • Parks and Wildlife Commission: Marvin McDaniel, Chair • Carrie Besnette Hauser, Vice-Chair Marie Haskett, Secretary • Taishya Adams • Betsy Blecha • Charles Garcia • Dallas May • Duke Phillips, IV • Luke B. Schafer • James Jay Tutchton • Eden Vardy �Behavioral Ecology Advance Access published September 11, 2014 Behavioral Ecology The official journal of the ISBE International Society for Behavioral Ecology Behavioral Ecology (2014), 00(00), 1–8. doi:10.1093/beheco/aru158 Original Article Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Received 20 July 2014; revised 29 July 2014; accepted 11 August 2014. While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (Odocoileus hemionus) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a form of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Key words: acoustic ecology, odocoileus hemionus, vigilance, mule deer Introduction Vigilance is an important behavioral adaptation allowing early detection and evasion of predators, thereby increasing survival (Lind 2005). Vigilance manifests in several ways including neural mechanisms, behavioral strategies, and social strategies (Dimond and Lazarus 1974). Research on behavioral vigilance strategies has predominantly focused on the visual forms of vigilance behavior, such as scanning, alert behavior, or heightened awareness, which have been well documented across a wide range of animal taxa (Lima 1987; Quenette 1990; Frid 1997; Fortin et al. 2004). However, animals also employ other sensory cues such as smell (Muller-Schwarze 1994) or auditory vigilance by listening for the Address correspondence to E. Lynch. E-mail: emma.lynch@colostate.edu. Published by Oxford University Press on behalf of the International Society for Behavioral Ecology 2014. acoustic cues of predators (Barber et al. 2010), or the alarm calls of conspecifics (Randall and Rogovin 2002; Thompson and Hare 2010; Hare and Warkentin 2012). Such behavioral strategies may be particularly valuable when environmental conditions preclude the use of sight or for species where visual acuity is low. The time invested in vigilance is considered a tradeoff to time invested in foraging, as the act of feeding is often incompatible with predator detection (Lima and Dill 1990). Foraging is expected to detract from the visual detection of predators, and it creates incidental noise that masks acoustic cues necessary for the auditory detection of predators (Molinari-Jobin et al. 2004). Thus, prey species are expected to modulate their investment in vigilance (both visual and auditory) with varying levels of predation risk in order to service metabolic requirements (maintain energy intake) while also evading predation (Brown 1999; Brown et al. 1999). Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 Emma Lynch,a Joseph M. Northrup,b Megan F. McKenna,c Charles R. Anderson Jr,d Lisa Angeloni,a,e and George Wittemyera,b aGraduate Degree Program in Ecology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA, bDepartment of Fish, Wildlife and Conservation Biology, Colorado State University, 1474 Campus Delivery, Fort Collins, CO 80523, USA, cNatural Sounds and Night Skies Division, National Park Service, 1201 Oakridge Drive, Fort Collins, CO 80525, USA, dMammals Research Section, Colorado Parks and Wildlife, 317 W. Prospect Road, Fort Collins, CO 80526, USA and eDepartment of Biology, Colorado State University, 1878 Campus Delivery, Fort Collins, CO 80523, USA �Behavioral Ecology Page 2 of 8 autonomically engages the stapedius reflex, which reduces the sensitivity of the ear by damping the transmission of vibrations through the incus–malleus–stapes complex (Pang and Guinan 1997). This attenuation of acoustic signals inhibits an animal’s ability to survey the acoustic environment. However, in addition to serving a role in auditory surveillance, pauses during mastication also may reflect processes unrelated to listening, including physiological functions (e.g., the movement of ingesta, gut processing time, or jaw muscle relief), though pauses as defined in this study are bounded by and exclude regurgitation and swallowing. We assume physiological processes such as mastication and rumination would occur in a random pattern across the landscape, as the available forage species during the winter season in this pinyon juniper range are of universally poor quality (Bartmann et al. 1982). Therefore, if pauses are used for auditory vigilance, time invested in the behavior would be expected to vary with exposure to stimuli and changes in landscape properties, particularly those that influence perceived levels of risk and that impede visual vigilance. To investigate spatial and temporal structuring in pause behavior, we conducted this study in the Piceance Basin area of northwestern Colorado, a topographically diverse region with heterogeneous vegetative communities that was actively undergoing natural gas production and extraction. This type of development has been shown to affect behavior in a range of ungulates (Northrup and Wittemyer 2013) and elicit changes in mule deer behavior consistent with an anti-predator response (Sawyer et al. 2006) and thus may be expected to cause an increase in their auditory vigilance. However, the human landscape features related to energy development (i.e., roads, drilling well pads, producing well pads, and facilities such as compressors) produce substantial noise, potentially masking other acoustic signals, degrading the efficacy of auditory vigilance, and ultimately causing a reduction in auditory vigilance. Additionally, these disturbed areas may offer a certain level of shelter by deterring predatory species, thereby reducing perceived risk and the use of auditory vigilance. Therefore, we examined the influence of proximity to these features on the proportion of time spent paused, our metric of auditory vigilance. We also assessed investment in pausing under conditions known to increase the perceived predation risk of ungulates (Altendorf et al. 2001; Stankowich 2008; Laundré et al. 2010), such as the presence of visual barriers that inhibit visual detection of predators (Hopewell et al. 2005). Specifically, we tested the predictions that pausing increased inside forested areas (relative to open regions), in rugged terrain, during nighttime hours (relative to daytime), and at closer distances to the edge of forested cover. METHODS Study area The study took place in the Piceance Basin of Northwestern Colorado, in an area that serves as winter range for mule deer from October through May. The area consists of both mixed mountain shrub and pinyon-juniper woodlands at an elevation of approximately 2000 m. This landscape naturally provides topographical relief and a diverse range of habitats, from dense cover to open, exposed regions (Bartmann et al. 1992). The landscape is also shaped by human activities associated with energy development, largely in the form of road networks servicing natural gas wells that are in varied phases of production or development. Human activity levels are high in this area, with several wells and natural gas facilities running 24 h a day, 7 days a week. Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 Investment in antipredator behaviors such as vigilance is expected to vary with spatial changes in perceived predation risk, often referred to as the “landscape of fear” (Brown et al. 1999; Laundré et al. 2001; Laundré et al. 2010). Major landscape factors known to influence visual forms of vigilance behavior include food density (Beauchamp 2009), whether habitat is open or closed (Ebensperger and Hurtado 2005), distance to cover (Lima 1987), and level of human disturbance (Li et al. 2011), but the relationship between these factors is not always straightforward (Quenette 1990). Human disturbance, in particular, can have differing effects on risk perception and subsequently visual vigilance behavior, increasing perceived risk and scanning when it represents a form of predation risk (Frid and Dill 2002), or reducing perceived risk when it provides a spatial refuge from predators that avoid human activity (Berger 2007, Muhly et al. 2011; Rogala et al. 2011). However, little is known about the effect of landscape variables and human disturbance on auditory vigilance, or how this behavior may interact with visual vigilance to structure behavioral responses to predation risk. Employment of auditory vigilance may be coupled with visual vigilance, or these behaviors may trade off in relation to characteristics that make one or the other more effective (e.g., auditory vigilance may be prevalent where landscape characteristics preclude sight). In the case of a trade-off between auditory and visual vigilance, it is possible that the availability of visual cues themselves could impact the propensity to listen in addition to landscape features and acoustic stimuli. With respect to human activity, auditory vigilance may be affected in ways similar to visual vigilance, where animals potentially increase investment if an increase in risk is perceived; alternatively, human activity could reduce perceived risk by providing refuge from predators. Furthermore, anthropogenic noise could render auditory vigilance ineffective by masking sounds of interest, causing a decrease in its use. While there is extensive evidence that animals can hear and respond to the sounds of predators and conspecific alarm cues, studies that directly quantify investment in listening for predator cues are largely absent from the literature, perhaps because of the difficulty in observing listening behavior. We overcome this impediment through the use of recently developed acoustic recording collars (Lynch et al. 2013), applying this technology to evaluate the potential role of auditory vigilance for mule deer (Odocoileus hemionus). The mule deer is a prey species that is known to use vigilance as a form of antipredator behavior (Geist 1981; Altendorf et al. 2001; Lynch et al. 2013). The visual acuity of mule deer is well established, and potential dangers are often identified visually before they are close enough to be a concern (Geist 1981; MullerSchwarze 1994; VerCauteren and Pipas 2003). However, mule deer spend up to 60% of their time resting (often in cover; Kie et al. 1991), requiring the use of other keen senses such as hearing to detect approaching animals (Muller-Schwarze 1994) and other changes in the environment. In fact, both the morphology and behavior of mule deer, including their oversized pinnae that amplify sounds (Calford and Pettigrew 1984), their sensitivity to wide ranging signals (250 Hz to 30 kHz, (Geist 1981), and their ability to detect animals as far away as 600 m in any direction using a combination of hearing, olfaction, and sight (Geist 1981); suggest that acoustic signals play important roles in their sensory ecology (Lingle et al. 2007; Teichroeb et al. 2013). Mule deer periodically pause during rumination while masticating ingesta, and this behavior appears to be used for auditory vigilance (Lynch et al. 2013). In addition to increasing sound levels that could mask detection of auditory cues, the act of chewing �Lynch et al. • Auditory vigilance in mule deer Page 3 of 8 Acoustic data collection Detection of auditory vigilance behavior Frequency (Hz) Previous research revealed that mule deer frequently pause during mastication bouts, creating brief periods of relative quiet that would allow for auditory vigilance (Lynch et al. 2013). Mastication bouts are part of the rumination process during which deer are stationary and are regurgitating and breaking down ingesta. For the purpose of this study, bouts were defined as continuous periods of mastication, bounded by continuous periods of other behaviors such as browsing or walking. Pauses during bouts occurred periodically, typically bounded by swallowing and regurgitation, but it should be noted that we could not acoustically identify any other digestive 5000 2500 1250 630 315 160 80 40 20 :10 :05 -8 -2 4 :15 10 16 :20 22 :25 28 :30 Time (min) :35 34 40 46 Sound Pressure Level (dB) :40 52 :45 58 64 :50 70 :55 76 82 87 Figure 1 Hour-long spectrogram of mule deer behavior. High sound pressure levels are indicated by darker shades, low sound pressure levels are indicated by lighter shades. Two mastication bouts are distinguishable from 01 to 33 min, and again from 51 to 59 min as areas of higher sound levels (as bounded by arrows). Pauses are located within each bout, and are represented by lighter shaded vertical stripes of varying width. Between mastication bouts, the deer was stationary, intermittently inhaling, exhaling, and eructating. The dark horizontal stripe seen throughout the hour at 80 Hz corresponds to the sound of mechanized equipment (such as a generator or compressor station). Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 Ten wild adult female mule deer (aged 4.5–11.5 years) were randomly selected from a multi-year global positioning system (GPS) radio tracking study (Lendrum et al. 2012, 2013) and wore collarmounted acoustic recording devices for approximately 2 weeks during the winter to track audible behaviors and ambient environmental sounds (Lynch et al. 2013). The microphones contained within the acoustic recording collars were mounted such that they were positioned against the throats of the deer. This positioning, along with their sensitivity, allowed capture of the sounds of footfalls, vocalizations, foraging, rumination (regurgitation, mastication, and swallowing), and even respirations when the deer were otherwise inactive. Prior to collar deployment, the technique was validated on captive individuals. During this testing period, we verified, through comparison of direct observations and concurrent audio files, that the collars clearly documented sounds produced by deer. During the study period, deer also wore GPS radio collars to track their movements. Protocol and procedures employed for capture were reviewed and approved under the Colorado State University Institutional Animal Care and Use Committee (IACUC) protocol 10-2350A. Once recovered, the acoustic recording devices provided continuous date-time stamped MPEG-2 Audio Layer III recordings. Following protocols detailed in Mennitt and Fristrup (2012), the acoustic recording collar was calibrated using a Type-1 (American National Standards Institute [ANSI] S 1.4-1983 [R 2006]) sound level meter (Larson Davis 831, Larson Davis, Depew, NY). This calibration was necessary to acquire broadband background sound levels from the collar. activity during the pauses. The pauses analyzed in this study were completely distinct from the swallowing and regurgitation portions of the mastication process (see electronic Supplementary Material for a recording of the focal behavior). Periods of mastication during rumination and the pauses included within them were identified by examining recorded acoustic data, displayed as spectrograms with 1-s, one-third octave band resolution using the Sound Pressure Level Annotation Tool (U.S. National Park Service Natural Sounds and Night Skies Division, Fort Collins, CO). In the spectrograms, these periods were visually distinct from other behaviors (Figure 1), verified aurally, and then annotated manually. During these mastication bouts, deer were typically stationary (as indicated by audio and GPS data), and steadily processing ingesta (as indicated by acoustic data). Using custom software developed in Matlab (Mathworks v. 2012b, Natick, MA), the waveforms of manually selected mastication bouts, described above, were further processed to automatically detect the start and end of pauses. The pause detector software worked on the time series data by first performing a full-wave rectification of the acoustic signal, essentially transforming the amplitude values to positive numbers. Then, the detector stepped through the data at 45-s intervals marking pauses as instances where levels remained below a 14th percentile threshold for a minimum of 1.4 s (selection of optimal automatic pause detection parameters are described in electronic Supplementary Material). This dynamic threshold approach was selected because it promoted consistent detection across varying signal levels from individual collars. Figure 2 displays a detected pause, marked on the original signal (a) and the rectified signal (b). To identify all pauses, we applied a set of parameter values to all mastication bouts from all individual deer (see electronic Supplementary Material). Relevant metrics for each pause such as begin time, end time, and duration were logged, and a 1-s one-third octave wideband sound pressure level (SPL) extracted from the center of each pause event was used to represent the background ambient SPL during the pause. Anthropogenic noise tends to occur on the lower end of the frequency spectrum (<2 kHz) (Francis et al. 2009; Barber et al. 2010), so to better assess the potential impacts of anthropogenic sounds on pause length, we calculated a truncated wideband SPL (dBT) that focused on low frequency sound (20–1250 Hz) in addition to the dBW wideband SPL (25–6300 Hz). �Behavioral Ecology Page 4 of 8 (a) (b) 1 Signal Threshold Pause Begin Pause End 1400 0.8 0.6 1200 0.4 1000 Amplitude Amplitude 0.2 0 −0.2 800 600 −0.4 −0.6 400 −0.8 200 22 Time (seconds) 33 11 22 Time (seconds) 33 Figure 2 Example of automatic pause detection. Begin (*) and end (○) times of a single pause are marked in the original time series data (a) and on the rectified signal (b). Detection percentile threshold is marked by dotted line on the right panel. Detector was triggered to mark a pause when the signal dropped below this percentile for at least 1.4 s. The signal before and after the pause represents chewing during mastication. The detector code and detailed information on detector performance are provided in electronic Supplementary Material. Modeling natural and anthropogenic effects To examine factors influencing variation in pause characteristics, we fit models of pause behavior during rumination in a Bayesian hierarchical framework. Our dependent variable was the proportion of the mastication bout spent paused (calculated as the sum of the duration of all pauses within a bout divided by the duration of the bout to standardize for bout length). For this variable, we fit beta regression models with intercepts varying by individual to account for the nested structure of the data (multiple bouts for each individual). The model structure is provided in electronic Supplementary Material. To extract landscape covariates for each mastication bout, we matched the time of the midpoint of each mastication bout to the GPS location for that deer that was closest in time. Deer locations were taken every 30 min using a GPS collar (Model G30C, Advanced Telemetry Systems, Isanti, MN). Where GPS fix failure did not make this possible, the location closest in time to the start or end of the bout was assigned to the bout if the location was taken within 1 h of the bout midpoint. Bouts not associated with a successful GPS fix according to this definition were dropped from the analysis. Consecutive bouts that occurred in spatially overlapping and temporally adjacent locations were combined into a single bout, assuming the short period of activity by the deer (e.g., brief movements and foraging) separating the bouts did not merit independent treatment. All covariates, described below, were extracted using the “raster” package (Hijmans and van Etten 2013) in the R statistical software version 3.0.1 (R Core Team 2013). The natural covariates expected to influence auditory vigilance (Table 1) included the distance of the deer to the edge of forested land cover (Edge), a binary covariate for whether it was located in forested (0) or open (1) land cover (Open), a terrain ruggedness index measuring the change in slope between the cell of interest and those surrounding it (TR), and a binary covariate for whether the bout occurred during the day (0) or between sunset and sunrise (1) (Night). Each of the spatial data layers was displayed with 30 m resolution. These factors were selected because they were expected to influence perceived predation risk and the ability to detect predators visually by influencing the line of sight distance. The anthropogenic factors expected to influence auditory vigilance (Table 1) included distance to the center of the nearest well pad with wells that were being actively drilled (D drill), distance to the center of the nearest well pad with only wells that were producing gas (D prod), distance to the center of the nearest natural gas facility (D fac), distance to the nearest road (D rd), and median wideband sound level (dBW.med) during the pause. These anthropogenic factors were selected because they might increase perceived predation risk (by presenting a disturbance), or they might decrease perceived predation risk (by deterring predators); further, sound levels have the potential to influence the ability to detect predators aurally. Following definitions of the Colorado Oil and Gas Conservation Commission, a well pad was considered a drilling well pad between the time that drilling began until product began to be extracted; it was considered a producing well pad once product began to be extracted. For well pads on which multiple wells existed, a pad was considered a drilling pad as long as at least one well was being drilled. A natural gas facility was defined as either a gas plant or compressor station. After all covariates were extracted, we fit 4 separate models to the dependent variable (Table 2). All models contained all covariates described above, but the structure of each was organized to explore the functional form (linear vs. non-linear) of the anthropogenic covariates. Interaction effects (between night and landcover, and between night and sound level) were considered, but no evidence for interaction effects was found (i.e., coefficients for interaction terms were not different from 0). The 4 models without interaction terms were compared using deviance information criteria (DIC; Spiegelhalter et al. 2002; but as formulated by Plummer 2012), with the best fit model used to reveal which factors were significant in predicting proportion of the mastication bout spent paused. Though we calculated multiple measures of background Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 11 �Lynch et al. • Auditory vigilance in mule deer Page 5 of 8 Table 1 Names, descriptions, and predicted effect (+/− indicates possible influence in either direction; see explanation in Introduction section) on the proportion of bout paused for covariates used in Bayesian hierarchical models Natural covariates Edge Open TR Night Anthropogenic covariates D prod D drill D fac D rd dbW.med Description Predicted effect Distance to the edge of forested land covera Binary covariate for being in forested land cover (0) or not (1)a Terrain ruggedness index–measure of change in slope between the cell of interest and those surrounding itb Binary covariate for whether the bout was between sunsxet and sunrise (1) or not (0)c − − + padd Distance to center of producing well Distance to center of drilling well padd Distance to center of natural gas facilitye Distance to nearest roadf Median wideband sound pressure level (25–6300 Hz)g + +/− +/− +/− +/− +/− Table 2 Model structure and deviance information criteria (DIC) values for models predicting the proportion of a mastication bout during which a deer was silent (paused) Model Structure DIC 1 2 3 4 Edge + Open + TR + Night + log(D prod) + log(D drill) + log(D fac) + log(D rd) + dbW.med + dbW.med2 Edge + Open + TR + Night + D prod + D drill + D fac + D rd + dbW.med + dbW.med2 Edge + Open + TR + Night + D prod + D drill + D fac + D rd + dbW.med Edge + Open + TR + Night + D prod + D prod2 + D drill + D drill2 + D fac + D fac2 + D rd + D rd2 + dbW.med + dbW.med2 −1884 −1882 −1882 −1878 sound level, we only report on models fit with the median dBW values because they provided a better fit (lower DIC) than models with mean dBW, mean dBT, and median dBT values (assessed separately from the 4 presented models). All models were fit in the R statistical software (R Core Team 2013) using the “rjags” package (Plummer 2013). We ran 2 chains for 8 000 000 iterations, discarding the first 4 000 000 as burn-in and thinning the chains to every 10th sample. We used starting values for all parameters that were expected to be overdispersed relative to the posterior distributions and assessed convergence to the posterior distribution using the Gelman–Rubin diagnostic (Gelman and Rubin 1992) and by examining trace plots of the resulting chains. RESULTS The automatic acoustic detector yielded 53 856 pause detections (with a median of 86.0 pauses per bout per individual, interquartile range [IQR] lower and upper: 42.0, 133.3) during 500 mastication bouts. The median duration of a mastication bout was 1.38 h (IQR: 0.57, 2.21). The median pause duration was 2.29 s (IQR: 1.79, 2.90), though a number of long pauses (up to 6.3 s) were also noted in the dataset. A median of 3.8% (IQR: 3.4%, 4.4%) of the time spent masticating was allocated to pausing. All models converged to the posterior distribution (Gelman– Rubin diagnostic for all parameters < 1.1; Gelman and Rubin 1992). Aside from Model 4 (Table 2), which had quadratic terms on all distance and sound covariates, all models had similar DIC values, indicating little difference among models (Table 2). Because it provided the best fit for the data according to DIC, we report coefficients for each covariate from Model 1 (Table 3), which were similar to and representative of the other models. As predicted, after accounting for other covariates, deer paused for a greater proportion of bouts during the night than during the day (Figure 3), and for smaller proportions of bouts when in open areas (Figure 3). Contrary to predictions, deer paused for a greater proportion of bouts when they were further from natural gas facilities (Table 3). Lastly, there was a quadratic relationship between dBW and pause behavior, whereby deer paused for a greater proportion of bouts in areas of intermediate sound level (Figure 4). The other covariates analyzed, such as distance to a road, distance to producing and drilling well pads, terrain ruggedness index and distance to the edge of forested land cover, were not significant predictors in models (<95% of posterior distribution on one side of 0; Table 3). Discussion Although vigilance is an important antipredator strategy, the mechanisms driving the use of auditory vigilance behavior are not well studied. Here, we present one of the first detailed assessments of this potentially critical behavior, and demonstrate how acoustic technology can change what we understand about how the landscape influences behavioral decision making. Our results demonstrated that pauses during mastication bouts were structured by landscape properties that are expected to influence perceived Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 Sources for covariates are indicated in the following footnotes. aColorado Vegetation Classification Project (http://ndis.nrel.colostate.edu/coveg/) bCalculated from digital elevation model obtained from United States Geological Survey seamless data warehouse (http://nationalmap.gov/viewer.html). cThe United States Naval Observatory (http://aa.usno.navy.mil/data/docs/RS_OneYear.php). dColorado Oil and Gas Conservation Commission (http://cogcc.state.co.us/). eObtained via ground truthing. fUnited States Geological Survey seamless data warehouse (http://nationalmap.gov/viewer.html) and validated through ground truthing. gCalculated from on-deer recording devices. �Behavioral Ecology Table 3 Representative model (Model 1) for the proportion of bout paused with median coefficient values and 95% credible intervals (CI) for each covariate Covariate Median coefficient 95% CI Edge Open TR Night log(D drill) log(D fac) log(D rds) log(D prod) dbW.med dbW.med2 0.035 −0.119 0 0.377 −0.009 0.194 0.067 −0.03 0.079 −0.055 −0.022 −0.23 −0.002 0.272 −0.1 0.097 0.006 −0.1 −0.032 −0.104 0.09 −0.009 0.002 0.483 0.083 0.292 0.13 0.042 0.191 −0.006 50 60 Median dbW 70 80 0.06 0.08 0.10 Figure 4 Predicted proportion of bout paused by median dBW (25–6300 Hz) values (solid line) with 95% credible interval (dotted lines). Predicted proportion is standardized by bout length. 0.04 Predicted proportion of bout paused 40 Day Night Figure 3 The median and interquartile range (IQR) for predicted proportion of bout paused for day and night time periods (derived from fitted model), where predicted proportion is standardized by bout length. Dashed lines extend to 95% credible interval. Forested landcover is represented by white boxes and open areas are represented by gray boxes. risk and impede visual vigilance. Consistent with predictions, we found that deer paused more extensively where concealment cover abounded and thus where visual vigilance was likely to be less effective. In addition, deer allocated a larger proportion of time to pausing during mastication bouts at night, implying that auditory vigilance is an important defense mechanism when darkness reduces the effectiveness of visual scanning. Previous research on vigilance behavior in wild species (Lima 1987; Altendorf et al. 2001; Lind 2010) has overlooked the importance of auditory vigilance, potentially because this behavior is not easily seen during behavioral observations and may be obscured when it is predominantly employed. Our results suggest that exploring this behavior allows deeper understanding of an animal’s perception of risk and the costs associated with vigilance behavior. Furthermore, identification of landscape characteristics associated with increased auditory vigilance provides an additional way to behaviorally identify properties that can augment more typical assessments of perceived predation risk (e.g., giving up densities or measures of visual vigilance). It is important to note that our findings do not preclude additional explanations for pausing behavior; for example, pausing may allow deer to be quiet to reduce detection when predators are near or it may be involved in physiological processes (though we note that pauses in our analysis excluded any audible activity including swallowing and regurgitation). As such, pausing during mastication likely serves multiple, simultaneous purposes. It is notable that in our analysis, the role of auditory vigilance, as evident from the increase in pausing at night and in cover, was strong enough to overcome the confounding influence of other modalities of pausing. Our pause identification procedure excluded all pauses less than 1.4 s in length, which may have reduced the influence of strictly biophysical activity from the behaviorally oriented listening pauses. Anecdotally, we noted that extended pauses were influenced by ecological stimuli (e.g., coyote calls occurred in conjunction with longer pauses), suggesting that longer pauses play a greater role in listening. A more efficient procedure for separating biophysical activity from behaviorally motivated pause categories would provide stronger inference and merits further investigation. For instance, future studies could experimentally manipulate perceived predation risk or disturbance in a controlled setting to identify the extent to which pauses are adjusted for listening. One of the negative consequences of increasing vigilance due to perceived risk is time taken away from other fitness enhancing behaviors (Frid and Dill 2002; Lind 2005). Therefore, an increase in auditory vigilance could result in a reduction in time spent invested in other behaviors. It has been argued that vigilance can be costfree when anti-predator vigilance is conducted during spare time (Illius and Fitzgibbon 1994), such as the interval of time between cropping and bringing a mouthful of forage fully into the mouth (Blanchard and Fritz 2007). However, unlike visual scanning, auditory vigilance cannot be conducted during the act of chewing, both because the sound of chewing itself masks auditory cues of interest in certain frequency bands, and because chewing triggers the Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 0.12 0.03 Significance is indicated with bold font. Predicted proportion of bout paused 0.04 0.05 0.06 0.07 0.08 Page 6 of 8 �Lynch et al. • Auditory vigilance in mule deer Supplementary Material Supplementary material can be found at http://www.beheco. oxfordjournals.org/ Funding This research was supported by Colorado Parks and Wildlife (CPW), U.S. National Park Service Natural Sounds and Night Skies Division, U.S. Bureau of Land Management, ExxonMobil Production/XTO Energy, WPX Energy, EnCana Corporation, the Mule Deer Foundation, the Colorado Mule Deer Association, Safari Club International, Federal Aid in Wildlife Restoration, Marathon Oil Corporation, Shell Exploration and Production, the Colorado State Severance Tax Fund, the Colorado Oil and Gas Conservation Commission, and Piceance Basin land owners. The authors thank L. Wolfe, C. Bishop, D. Finley (CPW) and numerous field technicians for capture expertise and field assistance, and Quicksilver Air, Inc. and Larry Gepfert (CPW pilot) for assisting with deer captures. Finally, the authors thank the Editor, Behavioral Ecology, as well as 2 anonymous reviewers for their thoughtful comments. Ethical Standards: Protocol and procedures employed for capture were reviewed and approved under the Colorado State University Institutional Animal Care and Use Committee (IACUC) protocol 10-2350A. Handling editor: Alison Bell References Altendorf K, Laundré J, López González C, Brown J. 2001. Assessing effects of predation risk on foraging behavior of mule deer. J Mammal. 82:430–439. Barber JR, Crooks KR, Fristrup KM. 2010. The costs of chronic noise exposure for terrestrial organisms. Trends Ecol Evol. 25:180–189. Bartmann RM, Alldredge AW, Neil PH. 1982. Evaluation of winter food choices by tame mule deer. J Wildl Manage. 46:807–812. Bartmann RM, White GC, Carpenter LH. 1992. Compensatory mortality in a Colorado mule deer population. Wildl Monogr. 121:1–39. Beauchamp G. 2009. How does food density influence vigilance in birds and mammals? Anim Behav. 78:223–231. Berger J. 2007. Fear, human shields and the redistribution of prey and predators in protected areas. Biol Lett. 3:620–623. Blanchard P, Fritz H. 2007. Induced or routine vigilance while foraging. Oikos. 116:1603–1608. Brown J, Laundré J, Gurung M. 1999. The ecology of fear: optimal foraging, game theory, and trophic interactions. J Mammal. 80:385–399. Brown JS. 1999. Vigilance, patch use and habitat selection: foraging under predation risk. Evol Ecol Res. 1:49–71. Calford MB, Pettigrew JD. 1984. Frequency dependence of directional amplification at the cat’s pinna. Hear Res. 14:13–19. Dimond S, Lazarus J. 1974. The problem of vigilance in animal life. Brain Behav Evol. 9:60–79. Ebensperger LA, Hurtado MJ. 2005. On the relationship between herbaceous cover and vigilance activity of degus (Octodon degus). Ethology. 111:593–608. Fortin D, Boyce MS, Merrill EH, Fryxell JM. 2004. Foraging costs of vigilance in large mammalian herbivores. Oikos. 107:172–180. Francis CD, Ortega CP, Cruz A. 2009. Noise pollution changes avian communities and species interactions. Curr Biol. 19:1415–1419. Frid A. 1997. Vigilance by female Dall’s sheep: interactions between predation risk factors. Anim Behav. 53:799–808. Frid A, Dill L. 2002. Human-caused disturbance stimuli as a form of predation risk. Conserv Ecol. 6:11. Geist V. 1981. Behavior: adaptive strategies. Mule and black-tailed deer of North America. Lincoln: University of Nebraska Press. Gelman A, Rubin DB. 1992. Inference from iterative simulation using multiple sequences. Stat Sci. 7:457–511. Hare J, Warkentin K. 2012. The song remains the same: Juvenile Richardson’s ground squirrels do not respond differentially to mother’s or colony member’s alarm calls. Curr Zool. 58:773–780. Hijmans RJ, van Etten J. 2013. Raster: geographic data analysis and modeling. http://R-Forge.R-project.org/projects/raster/. Hopewell L, Rossiter R, Blower E, Leaver L, Goto K. 2005. Grazing and vigilance by Soay sheep on Lundy island: influence of group size, terrain and the distribution of vegetation. Behav Processes. 70:186–193. Illius AW, Fitzgibbon C. 1994. Costs of vigilance in foraging ungulates. Anim Behav. 47:481–484. Kie JG, Evans CJ, Loft ER, Menke JW. 1991. Foraging behavior by mule deer: the influence of cattle grazing. J Wildl Manage. 55:665–674. Laundré J, Hernández L, Altendorf K. 2001. Wolves, elk, and bison: reestablishing the “landscape of fear” in Yellowstone National Park, U.S.A. Can J Zool. 79:1401–1409. Laundré J, Hernández L, Ripple WJ. 2010. The landscape of fear: ecological implications of being afraid. Open Ecol J. 3:1–7. Lendrum PE, Anderson CR Jr, Monteith KL, Jenks JA, Bowyer RT. 2013. Migrating mule deer: effects of anthropogenically altered landscapes. PLoS ONE. 8:e64548. Lendrum PE, Anderson CRJ, Long RA, Kie JG, Bowyer RT. 2012. Habitat selection by mule deer during migration: effects of landscape structure and natural-gas development. Ecosphere. 3:1–19. Li C, Monclús R, Maul TL, Jiang Z, Blumstein DT. 2011. Quantifying human disturbance on antipredator behavior and flush initiation distance in yellow-bellied marmots. Appl Anim Behav Sci. 129:146–152. Lima S. 1987. Vigilance while feeding and its relation to the risk of predation. J Theor Biol. 124:303–316. Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 stapedius reflex, an involuntary muscular contraction which limits the transmission of acoustic signals (Pang and Guinan 1997). This trade off is emphasized by a study that reported 58% of roe deer were preyed upon whilst ruminating, perhaps because the act of chewing hindered the deer’s ability to hear approaching predators (Molinari-Jobin et al. 2004). As such, trade-offs between noiseproducing behaviors and auditory vigilance may be as pertinent to animal ecology and behavior as other, more commonly studied behaviors. In summary, the investigation of auditory vigilance provides novel insight to animal time budgets and perceptions of risk, augmenting the more frequently studied visual vigilance behaviors, and offers a new lens through which to view the landscape of fear. Mule deer allocated a substantial amount of time to pausing in the midst of mastication bouts during the study period, with both natural and anthropogenic landscape features differentially impacting the use of this behavior. However, further research is necessary to understand the relationship between vigilance modalities (visual, olfactory, and auditory), and how these different behaviors compliment, tradeoff, or supersede one another. Such studies might employ either simultaneous visual and auditory inspection of vigilance behavior, or accelerometry or magnetometry sensors (to gather fine scale information about head movements and orientations) to better understand interactions between these behaviors and their landscape context. Studies assessing different contexts where these behaviors are utilized and their relative roles across species also would further increase understanding. Additional research is also needed to gain a better understanding of the relationship between behavioral measures of vigilance, concomitant brain and sensory processing requirements, and the direct fitness consequences of investment in this activity, a critical component to determine the cost-benefit ratio of these behaviors. Finally, although the bout-level data acquired from this study did not allow us to investigate the immediate factors that influenced individual pauses, we identified a number of long pauses that closely followed significant acoustic events (such as coyote calls or vehicles passing nearby). The biologic import of these hyper-vigilant events is likely significant, and points to an exciting new approach for identifying specific predator-prey interactions. Page 7 of 8 �Page 8 of 8 Plummer M. 2012. JAGS Version 3.3.0 user manual. Lyon, France: International Agency for Research on Cancer. Plummer M. 2013. rjags: Bayesian graphical models using MCMC. R package version 3–10. http://CRAN.R-project.org/package=rjags. Quenette PY. 1990. Functions of vigilance behaviour in mammals: a review. Acta Oecol. 11:801–818. R Core Team. 2013. R: A language and environment for statistical computing. Vienna (Austria): R Foundation for Statistical Computing. [cited 2014 August 28]. http://www.R-project.org. Randall JA, Rogovin KA. 2002. Variation in and meaning of alarm calls in a social desert rodent Rhombomys opimus. Ethology. 108:513–527. Rogala JK, Hebblewhite M, Whittington J, White CA, Coleshill J, Musiani M. 2011. Human activity differentially redistributes large mammals in the Canadian Rockies national parks. Ecol Soc. 16:16. Sawyer H, Nielson RM, Lindzey F, McDonald LL. 2006. Winter habitat selection of mule deer before and during development of a natural gas field. J Wildl Manage. 70:396–403. Spiegelhalter DJ, Best NG, Carlin BR, van der Linde A. 2002. Bayesian measures of model complexity and fit. J Roy Stat Soc B. 64:583–616. Stankowich T. 2008. Ungulate flight responses to human disturbance: a review and meta-analysis. Biol Conserv. 141:2159–2173. Teichroeb LJ, Riede T, Kotrba R, Lingle S. 2013. Fundamental frequency is key to response of female deer to juvenile distress calls. Behav Processes. 92:15–23. Thompson AB, Hare JF. 2010. Neighbourhood watch: multiple alarm callers communicate directional predator movement in Richardson’s ground squirrels, Spermophilus richardsonii. Anim Behav. 80:269–275. VerCauteren KC, Pipas MJ. 2003. A review of color vision in white-tailed deer. Wildl Soc Bull. 31:684–691. Downloaded from http://beheco.oxfordjournals.org/ at colostateuniv on April 19, 2016 Lima S, Dill L. 1990. Behavioral decisions made under the risk of predation: a review and prospectus. Can J Zool. 68:619–640. Lind J. 2005. Determining the fitness consequences of antipredation behavior. Behav Ecol. 16:945–956. Lind J. 2010. Vigilance and models of behavior. In: Breed MD, Moore J, editors. Encyclopedia of animal behavior. Oxford: Academic Press. p. 506–510. Lingle S, Rendall D, Pellis SM. 2007. Altruism and recognition in the antipredator defence of deer: 1. Species and individual variation in fawn distress calls. Anim Behav. 73:897–905. Lingle S, Rendall D, Wilson WF, DeYoung RW, Pellis SM. 2007. Altruism and recognition in the antipredator defence of deer: 2. Why mule deer help nonoffspring fawns. Anim Behav. 73:907–916. Lynch E, Angeloni L, Fristrup K, Joyce D, Wittemyer G. 2013. The use of on-animal acoustical recording devices for studying animal behavior. Ecol Evol. 3:2030–2037. Mennitt DJ, Fristrup KM. 2012. Obtaining calibrated sound pressure levels from consumer digital audio recorders. Appl Acoust. 73:1138–1145. Molinari-Jobin A, Molinari P, Loison A, Gaillard JM, Breitenmoser U. 2004. Life cycle period and activity of prey influence their susceptibility to predators. Ecography. 27:323–329. 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. Muller-Schwarze D. 1994. The senses of deer. In: Gerlach D, Atwater S, Schnell J, editors. The wildlife series: deer. Mechanicsburg (PA): Stackpole Books. Northrup JM, Wittemyer G. 2013. Characterising the impacts of emerging energy development on wildlife, with an eye towards mitigation. Ecol Lett. 16:112–125. Pang XD, Guinan JJ Jr. 1997. Effects of stapedius-muscle contractions on the masking of auditory-nerve responses. J Acoust Soc Am. 102:3576–3586. Behavioral Ecology � https://cpw.cvlcollections.org/files/original/0e69734b13df84ef58380ae907a624ae.mp3 60d0b6e8822581cb908f7608fce5815b https://cpw.cvlcollections.org/files/original/07500104ba59b6f32b918ec00a1c660e.pdf 4e8f8734c8e06dd4bc822f85da11986f PDF Text Text 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 ELECTRONIC SUPPLEMENTARY MATERIAL A. Selection of parameters for pause detection To determine the automatic detection criteria for time interval, percentile threshold and minimum pause duration, we evaluated the performance of 5 different combinations of values for their ability to correctly identify pauses. After implementation of each combination, correctly assigned pauses, false positives and missed pauses were identified visually within the first and last mastication bout of the two-week study period for each individual. We selected the combination of values that resulted in the lowest percentage of false positives and a high percentage of correct detections. The selected parameters were minimum pause length = 1.4 seconds, time interval = 45 seconds, and 14th percentile threshold (Table 1), which resulted in a false positive rate of 2.025%. We applied these parameter values to all mastication bouts from all individual deer. Relevant metrics for each pause such as begin time, end time, and duration were logged (note that pauses shorter than 1.4 seconds were excluded). Moreover, a 1-second onethird octave wideband sound pressure level (SPL), extracted from the center of each pause event was used to represent the background ambient sound pressure level during the pause. Anthropogenic noise tends to occur on the lower end of the frequency spectrum (<2 kHz) (Francis et al. 2009; Barber et al. 2010). So, to better assess the potential impacts of anthropogenic sounds on pause length, we calculated a truncated wideband SPL (dBT) that focused on low frequency sound (20 Hz – 1250 Hz) in addition to the dBW wideband SPL (25 Hz – 6300 Hz). Table S1. Evaluation of detector performance, with parameter sets listed in the sequence tested. The bolded parameter set was utilized in the analysis because it minimized false detections. Percentile Minimum pause Time threshold duration (s) interval(s) % Correct % False % Missed 16 0.8 30 70.393 20.921 8.686 11 0.8 30 71.914 8.587 19.499 14 1.4 45 76.709 2.025 21.266 12 1.1 60 78.241 5.787 15.972 12 1.1 45 78.623 6.643 14.734 B. Code for beta regression model (JAGS) model{ #priors mu.b0~dnorm(0,0.00001) tau.b0~dgamma(0.001,001) beta[1:n.beta]~dmnorm(mu.beta[], omega.beta[,]) # tau.eps~dgamma(0.001,0.001) # sigma.eps<-1/sqrt(tau.eps) r~dunif(0,100) #individual intercepts �39 40 41 42 43 44 45 46 47 48 49 50 for(j in 1:n.indiv){ b0[j]~dnorm(mu.b0, tau.b0) } #model for(i in 1:length(y)){ log(lambda[i])<-b0[indiv[i]] + beta[1]*x1[i] + beta[2]*x2[i] + beta[3]*x3[i] + beta[4]*x4[i] + offset[i] p[i]<-r/(r+lambda[i]) y[i]~dnegbin(p[i],r) } } �51 52 C. Model outputs Pause duration Model 1 Covariate Edge Open D prod D prod^2 D drill D drill^2 D fac D fac^2 TR D rd D rd^2 dbW.med dbW.med^2 Night Median Coeff. 0.022243 -0.13082 -0.07797 0.045018 -0.00422 0.058304 0.251687 0.046768 0.000683 0.041298 -0.01139 0.034512 -0.0333 0.379152 Prob 0.257419 0.970566 0.92232 0.063904 0.521774 0.159119 0.000511 0.141828 0.28986 0.21755 0.680654 0.307808 0.85824 0 Prob + 0.742581 0.029434 0.07768 0.936096 0.478226 0.840881 0.999489 0.858173 0.71014 0.78245 0.319346 0.692193 0.14176 1 Model 2 Edge Open log(D prod) log(D drill) log(D fac) TR log(D rd) dbW.med dbW.med^2 Night Median Coeff. 0.034631 -0.11933 -0.02959 -0.00886 0.19435 0.000452 0.067082 0.078662 -0.05472 0.37745 Prob 0.151781 0.962809 0.75582 0.564614 0.000458 0.355923 0.035625 0.123956 0.967096 0 Prob + 0.848219 0.037191 0.24418 0.435386 0.999543 0.644078 0.964375 0.876044 0.032904 1 Model 3 Edge Open D prod D drill D fac TR D rd dbW.med Night Median Coeff. 0.025748 -0.10532 -0.02243 -9.6E-05 0.244615 0.000679 0.025445 0.00536 0.376781 Prob 0.220254 0.945793 0.688315 0.495956 0.000268 0.290155 0.27582 0.461314 0 Prob + 0.779746 0.054208 0.311685 0.504044 0.999733 0.709845 0.72418 0.538686 1 �Model 4 Edge Open D prod D drill D fac TR D rd dbW.med dbW.med^2 Night 53 54 55 56 57 Median Coeff. 0.024296 -0.12141 -0.01619 0.007259 0.24575 0.000719 0.023557 0.053858 -0.04827 0.383386 Prob 0.23062 0.966201 0.635313 0.459126 5.75E-05 0.278774 0.290991 0.215216 0.945503 0 Prob + 0.76938 0.033799 0.364688 0.540874 0.999943 0.721226 0.709009 0.784784 0.054498 1 � https://cpw.cvlcollections.org/files/original/279fba799f67034079580b9d15a6b823.pdf c1f875bdf89e1d6a2b51eb01bd5a64b1 PDF Text Text % Uses binary file (created with Compile_SRCID_wav.m) to detect pauses within % deer mastication bouts (pre-selected by E. Lynch) %NOTES: %(1) Must run Compile_SRCID_wav.m first to generate a binary file listing %the times of all the mastication events for an individual deer (e.g. %SRCID_RG15_events_221). Events file size exceeds that which can be uploaded (~32 GB), %but can be made available upon request. %(2)Works for a single deer and mastication bout, chosen by the user %(3)Allows user to visualize/listen to each pause detected to verify clear all; clc; close all %% INITIAL CODE SET-UP FOR PAUSE DETECTION Fs=44100; %sampling frequency of the audio data Percentile = 14; %sets the "trigger" threshold for pause detection winL = 1.4*Fs; %Upper duration bound for potential pauses(in samples) winU = 50*Fs; %Lower duration bound for potential pauses(in samples) ee = 3 ; %choose what binary file to analyze, %which corresponds to an individual deer (1-10) uu = 17; %choose event number you want to examine %(1-length of events for an individual deer) pltwin = 45; %sets analysis window when evaluating pause detections %(in seconds) test = 2; %if running the code in testing mode plt = 2; %if you want the graphics plotted as the code runs saveplot = 1; %if you want to save the graphics as the code runs %% PAUSE DETECTOR %% NAVIGATE to location of events files strdirs = uigetdir( 'C:\' , 'Pick folder with events files'); cd( strdirs ); elist = dir( '*events_*' ); %list of each event file in the directory filenameE = char( elist.name ); llist = dir( '*length_*' ); %list of each length file filenameL = char( llist.name ); display(['Processing ' filenameE(ee,:)]) %% GET the number of mastication events (This number is appended to the end of the file name) if filenameE(ee,10) == '_' %Datacheck for SRCID length or events file to allow for different deer name lengths (RG1 vs RG11) numE = str2num( filenameE(ee,18:end) ); %#ok<*ST2NM> elseif isnumeric( str2num( filenameE(ee,10) )) %EDIT THIS w/ conditional statement for 18 or 19 length filenames numE = str2num( filenameE(ee,19:end) ); else display('Events or length filename has error') end file:///C/Users/austerma/Desktop/PauseDetector.m.txt[7/2/2021 2:18:44 PM] �%% READ in correct mastication event based on the start time of the event % need to start at the correct location in the binary file, using the % corresponding length file len = dlmread(filenameL(ee,:)); %list of length of each event and starttime fid = fopen(filenameE(ee,:),'r'); if uu == 1 DATA = fread(fid, len(uu,1),'single'); else GoHere = sum(len(1:uu,1)); fseek(fid, GoHere, 'bof'); DATA= fread(fid, len(uu,1),'single'); end %% RECTIFY the audio signal numsamples = length(DATA); intwindowtime = 1; %seconds intwindowsize = intwindowtime*Fs; halfintwindowsize = intwindowsize/2; RDATA = zeros(1,numsamples); %sets up matrix to fill with rectified data %initialize integral for one window length... RDATA(1+halfintwindowsize) = sum(abs(DATA(1:1+intwindowsize))); %...then add 1 sample and subtract last sample's contribution % as move through waveform for i = 2:numsamples-intwindowsize RDATA(i+halfintwindowsize)= RDATA(i+halfintwindowsize - 1)-... abs(DATA(i-1))+ abs(DATA(i+intwindowsize)); end %% EXTRACT the lowest values (potential pauses) in the designated analysis window % calculated as a proportion of the data plotwin = 1:pltwin*Fs:length(DATA); % converts analysis window to samples output = zeros (length(plotwin),6); % set up empty matrix to fill output(:,1) = ee; output(:,2) = uu; for pp = 1: length(plotwin); % loop through each analysis window if pp < length(plotwin) %GRABS the correct data and calculates the value what would trigger % a pause detection based on previously set percentile VALpctile = prctile (RDATA (plotwin(pp):plotwin(pp+1)), Percentile); %record the percentile used for this analysis window output(pp,3) = VALpctile; %converts data to zeros and ones based threshold values % (VALpctile) file:///C/Users/austerma/Desktop/PauseDetector.m.txt[7/2/2021 2:18:44 PM] �TDATA = RDATA (plotwin(pp): plotwin(pp+1)); array = zeros( size(TDATA) );%set up empty matrix to fill for jj = 1 : length(TDATA) % if TDATA(jj) < VALpctile array(jj) = 0; %YES, above VALpctile! else array(jj) = 1 ; %NO, below VALpctile! end end %% PULL OUT "ISLANDS OF ZEROS" % FOUND METHOD here: % http://stackoverflow.com/questions/3274043/finding-islands-of-zeros-in-a-sequence % First, find the starting indices, ending indices, and duration of each % string of zeros using the functions DIFF and FIND: % essentially, makes the start of a sequence of zeros -1, the end of a sequence % 1, and all the rest zeros tsig = array; dsig = diff([1 tsig 1]); %% CALCULATE the druation of each "potential pause" startIndex = find( dsig < 0); endIndex = find( dsig > 0)-1; duration = endIndex - startIndex + 1; %ONLY select pauses within the defined duration stringIndex = (duration <= winU & duration >= winL); durationCut = duration(duration <= winU & duration >= winL); startIndex2 = startIndex(stringIndex); endIndex2 = endIndex(stringIndex); %% EVALUATE the pause(s) detected scrsz = get(0,'ScreenSize');% set the resolution of the figure if isempty(startIndex2) display('NO PAUSES!') fig1 = figure(1); subplot(1,2,1) %original signal % plots the signal plot( plotwin(pp):1:plotwin(pp+1), DATA(plotwin(pp):plotwin(pp+1)), 'color', [.7 .7 .7] ); % labeling axes title('Original Signal'); xlabel(['Samples (' num2str(pltwin) ' seconds)']) ; ylabel('Amplitude'); subplot(1,2,2) %rectified signal % plot the rectified signal plot( plotwin(pp):1:plotwin(pp+1), TDATA, 'color', [.7 .7 .7] ); hold on; % plot the threshold line for this analysis window plot(plotwin(pp):1:plotwin(pp+1)-1,repmat(VALpctile,(pltwin*Fs),1)','k-.'); % label axes, set legend file:///C/Users/austerma/Desktop/PauseDetector.m.txt[7/2/2021 2:18:44 PM] �axis([plotwin(pp) plotwin(pp+1) 0 max(TDATA)-50]); title('Rectified Signal'); xlabel(['Samples (' num2str(pltwin) ' seconds)']) ; ylabel('Amplitude'); set(fig1,'Position',[scrsz(1) scrsz(2) scrsz(3) scrsz(4)]) legend ('Rectified Signal','Threshold'); %% LISTEN to the data in the analysis window reply = input('Do you want to play sound Y/N? ', 's'); dn = reply; if dn == 'Y' % plot the data fig3 = figure(3); plot( plotwin(pp):1:plotwin(pp+1), DATA(plotwin(pp):plotwin(pp+1)), 'color', [.7 .7 .7] ); axis([plotwin(pp) plotwin(pp+1) -1 1]); xlabel(['Samples (' num2str(pltwin) ' seconds)']) ; %xlabel('samples (30 seconds of data)') ; ylabel('Amplitude'); % play the sound v = axis; y = [v(3),v(4)]; x1 = [min(plotwin(pp):1:plotwin(pp+1)),min(plotwin(pp):1:plotwin(pp+1))]; % create the scrolling line for animation hline1 = line(x1,y,'Color','r','LineWidth',2); player = audioplayer(DATA(plotwin(pp):plotwin(pp+1)),Fs); play(player); while isplaying(player) x = [min(plotwin(pp):1:plotwin(pp+1)),min(plotwin(pp):1:plotwin(pp+1))] + get(player,'CurrentSample'); set(hline1,'Xdata',x,'Ydata',y) drawnow end end else display('Pauses detected!') fig1 = figure(1); subplot(1,2,1) %original signal % plots the signal plot( plotwin(pp):1:plotwin(pp+1), DATA(plotwin(pp):plotwin(pp+1)), 'color', [.7 .7 .7] ); hold on; ind = zeros(1, length(startIndex2)); % plots the start and end of the pauses plot(startIndex2+plotwin(pp),ind, 'k*','MarkerSize',7); plot(endIndex2+plotwin(pp),ind, 'ko','MarkerSize',6); axis([plotwin(pp) plotwin(pp+1) -1 1]); % labeling axes title('Original Signal'); xlabel(['Samples (' num2str(pltwin) ' seconds)']) ; ylabel('Amplitude'); subplot(1,2,2) %rectified signal % plot the rectified signal plot( plotwin(pp):1:plotwin(pp+1), TDATA, 'color', [.7 .7 .7] ); % hold on; % plot the threshold line for this analysis window plot(plotwin(pp):1:plotwin(pp+1)-1,repmat(VALpctile,(pltwin*Fs),1)','k-.'); % plot the start and end of pauses file:///C/Users/austerma/Desktop/PauseDetector.m.txt[7/2/2021 2:18:44 PM] �ind2 = repmat(VALpctile, length(startIndex2),1); plot(startIndex2+plotwin(pp),ind2, 'k*','MarkerSize',7); plot(endIndex2+plotwin(pp),ind2, 'ko','MarkerSize',6); % label axes, set legend axis([plotwin(pp) plotwin(pp+1) 0 max(TDATA)-50]); title('Rectified Signal'); xlabel(['Samples (' num2str(pltwin) ' seconds)']) ; ylabel('Amplitude'); set(fig1,'Position',[scrsz(1) scrsz(2) scrsz(3) scrsz(4)]) legend ('Rectified Signal','Threshold','Pause Begin','Pause End'); %% LISTEN to the data in the analysis window reply = input('Do you want to play sound Y/N? ', 's'); dn = reply; if dn == 'Y' % plot the data fig3 = figure(3); plot( plotwin(pp):1:plotwin(pp+1), DATA(plotwin(pp):plotwin(pp+1)), 'color', [.7 .7 .7] ); hold on; ind = zeros(1, length(startIndex2)); plot(startIndex2+plotwin(pp),ind, 'k*'); plot(endIndex2+plotwin(pp),ind, 'r*'); axis([plotwin(pp) plotwin(pp+1) -1 1]); xlabel(['Samples (' num2str(pltwin) ' seconds)']) ; %xlabel('samples (30 seconds of data)') ; ylabel('Amplitude'); % play the sound v = axis; y = [v(3),v(4)]; x1 = [min(plotwin(pp):1:plotwin(pp+1)),min(plotwin(pp):1:plotwin(pp+1))]; % create the scrolling line for animation hline1 = line(x1,y,'Color','r','LineWidth',2); player = audioplayer(DATA(plotwin(pp):plotwin(pp+1)),Fs); play(player); while isplaying(player) x = [min(plotwin(pp):1:plotwin(pp+1)),min(plotwin(pp):1:plotwin(pp+1))] + get(player,'CurrentSample'); set(hline1,'Xdata',x,'Ydata',y) drawnow end end end %% CONTINUE with analyis reply4 = (input ('Do you want to advance to the next analysis window [Y/N]?', 's')); if reply4 == 'Y'; close all %CLOSE ALL FIGURE WINDOWS clear x y v player hline1 array startIndex2 endIndex2 continue else display('Done with analysis of pauses...') close all %CLOSE ALL FIGURE WINDOWS clear x y v player hline1 array startIndex2 endIndex2 break end end file:///C/Users/austerma/Desktop/PauseDetector.m.txt[7/2/2021 2:18:44 PM] �end file:///C/Users/austerma/Desktop/PauseDetector.m.txt[7/2/2021 2:18:44 PM] � Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Journal Articles Description An account of the resource CPW peer-reviewed journal publications Text A resource consisting primarily of words for reading. Examples include books, letters, dissertations, poems, newspapers, articles, archives of mailing lists. Note that facsimiles or images of texts are still of the genre Text. Dublin Core The Dublin Core metadata element set is common to all Omeka records, including items, files, and collections. For more information see, http://dublincore.org/documents/dces/. Title A name given to the resource Landscape and anthropogenic features influence the use of auditory vigilance by mule deer Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/InC-NC/1.0/" target="_blank" rel="noreferrer noopener">In Copyright - Non-Commercial Use Permitted</a> Date Created Date of creation of the resource. 2014-09-11 Subject The topic of the resource Acoustic ecology <em>Odocoileus hemionus</em> Vigilance Mule deer Description An account of the resource While visual forms of vigilance behavior and their relationship with predation risk have been broadly examined, animals also employ other vigilance modalities such as auditory vigilance by listening for the acoustic cues of predators. Similar to the tradeoffs associated with visual vigilance, auditory behavior potentially structures the energy budgets and behavior of animals. The cryptic nature of auditory vigilance makes it difficult to study, but on-animal acoustical monitoring has rapidly advanced our ability to investigate behaviors and conditions related to sound. We utilized this technique to investigate the ways external stimuli in an active natural gas development field affect periodic pausing by mule deer (<em>Odocoileus hemionus</em>) within bouts of rumination-based mastication. To better understand the ecological properties that structure this behavior, we investigate spatial and temporal factors related to these pauses to determine if results are consistent with our hypothesis that pausing is used for auditory vigilance. We found that deer paused more when in forested cover and at night, where visual vigilance was likely to be less effective. Additionally, deer paused more in areas of moderate background sound levels, though responses to anthropogenic features were less clear. Our results suggest that pauses during rumination represent a form of auditory vigilance that is responsive to landscape variables. Further exploration of this behavior can facilitate a more holistic understanding of risk perception and the costs associated with vigilance behavior. Creator An entity primarily responsible for making the resource Lynch, Emma Northrup, Joseph M. McKenna, Megan F. Anderson Jr, Charles R. Angeloni, Lisa Wittemyer, George Language A language of the resource English Is Part Of A related resource in which the described resource is physically or logically included. Behavioral Ecology Extent The size or duration of the resource. 8 pages Bibliographic Citation A bibliographic reference for the resource. Recommended practice is to include sufficient bibliographic detail to identify the resource as unambiguously as possible. Lynch, E., J. M. Northrup, M. F. McKenna, C. R. Anderson Jr, L. Angeloni, and G. Wittemyer. 2014. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology 26:75–82. <a href="https://doi.org/10.1093/beheco/aru158" target="_blank" rel="noreferrer noopener">https://doi.org/10.1093/beheco/aru158</a> Format The file format, physical medium, or dimensions of the resource application/pdf Type The nature or genre of the resource Article