https://cpw.cvlcollections.org/files/original/845526345b99ba6398a39e000e284655.pdf f8bea669e25b5e17026a50e8bf222ae5 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 �Journal of Mammalogy, 100(5):1479–1489, 2019 DOI:10.1093/jmammal/gyz124 On-animal acoustic monitoring provides insight to ungulate foraging behavior Joseph M. Northrup,*,† Alexandra Avrin,† Charles R. Anderson, Jr., Emma Brown, and George Wittemyer * Correspondent: joseph.northrup@ontario.ca † These authors contributed equally to this work. Foraging behavior underpins many ecological processes; however, robust assessments of this behavior for freeranging animals are rare due to limitations to direct observations. We leveraged acoustic monitoring and GPS tracking to assess the factors influencing foraging behavior of mule deer (Odocoileus hemionus). We deployed custom-built acoustic collars with GPS radiocollars on mule deer to measure location-specific foraging. We quantified individual bites and steps taken by deer, and quantified two metrics of foraging behavior: the number of bites taken per step and the number of bites taken per unit time, which relate to foraging intensity and efficiency. We fit statistical models to these metrics to examine the individual, environmental, and anthropogenic factors influencing foraging. Deer in poorer body condition took more bites per step and per minute and foraged for longer irrespective of landscape properties. Other patterns varied seasonally with major changes in deer condition. In December, when deer were in better condition, they took fewer bites per step and more bites per minute. Deer also foraged more intensely and efficiently in areas of greater forage availability and greater movement costs. During March, when deer were in poorer condition, foraging was not influenced by landscape features. Anthropogenic factors weakly structured foraging behavior in December with no relationship in March. Most research on animal foraging is interpreted under the framework of optimal foraging theory. Departures from predictions developed under this framework provide insight to unrecognized factors influencing the evolution of foraging. Our results only conformed to our predictions when deer were in better condition and ecological conditions were declining, suggesting foraging strategies were state-dependent. These results advance our understanding of foraging patterns in wild animals and highlight novel observational approaches for studying animal behavior. Key words: acoustic monitoring, Bayesian hierarchical model, Colorado, foraging behavior, herbivore foraging, mule deer, Odocoileus hemionus, spatial ecology Foraging is a fundamental animal behavior that influences an array of ecological processes. This behavior dictates patterns of animal movement and patch use (Charnov 1976; Wajnberg et al. 2006), determines the dynamics of predator–prey interactions (Brown 1999; Brown et al. 1999), and is a driver of community structuring (Petchey et al. 2008). Furthermore, assessments of the influence of human presence and activities on foraging can provide insight to behaviorally mediated population processes and potentially fitness (Frid and Dill 2002). Understanding © 2019 American Society of Mammalogists, www.mammalogy.org environmental and anthropogenic factors influencing foraging behavior in animals is thus an important topic in both basic and applied animal ecology. The foraging behavior of animals is typically examined in the context of optimal foraging theory. Optimal foraging theory assumes that natural selection has shaped foraging decisions to maximize the fitness payoffs of this behavior (Pyke et al. 1977). Early research on this topic focused on the costs of movement and predicted that animals will maximize food intake by 1479 Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Department of Fish, Wildlife and Conservation Biology, Colorado State University, Fort Collins, CO 80523, USA (JMN, AA, GW) Ontario Ministry of Natural Resources and Forestry, Wildlife Research and Monitoring Section, Peterborough, Ontario K9L 1Z8, Canada (JMN) Mammals Research Section, Colorado Parks and Wildlife, Fort Collins, CO 80526, USA (CRA) National Park Service Natural Sounds and Night Skies Division, Fort Collins, CO 80525, USA (EB) �1480 JOURNAL OF MAMMALOGY We present a novel method of quantifying the components of foraging by using on-animal acoustic monitoring. This technique allowed us to measure the relationships between bites taken, time, and movement that are fundamental components of foraging. We used the framework of optimal foraging theory to generate three sets of predictions about how deer foraging behavior is influenced by a suite of environmental, anthropogenic, and individual factors. We note that these predictions rely in part on assumptions about whether these factors are adequately represented by our measured variables (see methods below). 1. Optimal foraging theory predicts that animals will forage more intensely and efficiently (i.e., take more bites of forage per unit time and per step) in areas with more available forage and greater costs of movement (Charnov 1976; Wajnberg et al. 2006). We assumed that areas of higher vegetation biomass (measured via the normalized difference vegetation index, NDVI), closer to edges, and in open or shrub land cover relative to treed land cover, were related to higher forage availability, and predicted that deer would take more bites per step and per unit time in these areas. We assumed that deer would incur greater costs of movement in the areas of deepest snow and more rugged terrain, and thus predicted that deer would take more bites per step in these areas. 2. Optimal foraging theory predicts that within a foraging patch, animals will forage more intensely and efficiently when predation risk is highest, because they should only use high-risk areas when the return is high (Brown and Kotler 2004). Empirical research on animal foraging suggests that anthropogenic disturbance can be perceived as akin to predation risk for herbivore species (Frid and Dill 2002; Ciuti et al. 2012). Thus, we tested whether mule deer perceived anthropogenic disturbance as akin to predation risk by examining the influence of a suite of features related to natural gas development on foraging behavior. We predicted that deer would forage more intensely and efficiently in areas closest to development. 3. Lastly, optimal foraging theory predicts that animals in better condition will prioritize risk reduction and minimization of opportunity costs over forage acquisition (Brown 1999; Brown and Kotler 2004). Thus, we predicted that deer in better condition would forage less intensely and efficiently than deer in poorer condition. Materials and Methods Study area and data collection.—This study took place on mule deer winter range in the Piceance Basin in northwestern Colorado (39.92 N, 108.35 W). The Piceance Basin is semiarid, with hot dry summers and cold winters, where winter snowfall accounts for the majority of precipitation. The vegetation on mule deer winter range is predominantly two-needle pinyon pine (Pinus edulis) and Utah juniper (Juniperus osteosperma). For detailed information on vegetation in the area, see Bartmann (1983). The elevation ranges between 1,700 and 2,300 m. This Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 balancing the costs and benefits of local forage with those of more distant forage (MacArthur and Pianka 1966; Charnov 1976; Pyke et al. 1977). This theory has been expanded to incorporate predation risk into the trade-off between foraging and movement, including investment in vigilance behaviors (Brown 1999), and provides an important theoretical framework for interpreting observed foraging patterns and related behaviors in the wild. Indeed, foraging theory forms one of the foundations for much of the contemporary research on spatial ecology and movement of animals (Owen-Smith et al. 2010; Humphries et al. 2012; Wilson et al. 2012; Wittemyer et al., In press). The study of foraging in the wild is often limited by difficulty directly observing the foraging decisions of animals. Within a foraging bout, these foraging decisions can be divided into two components: food intake and movement (Novellie 1978). At the finest scale, for ungulate species, food intake consists of bites and movement consists of steps, the ratio of which can be influenced by forage abundance, quality, predation risk, and environmental factors that inhibit movement (e.g., snow). Direct measures of foraging behavior at this scale are difficult to obtain, particularly in the wild. While some studies have relied on captive-reared, tame animals (e.g., Bartmann 1983; Gross et al. 1993) to observe foraging at this scale, studies of wild animals depend on visual observations (e.g., Novellie 1978). Visual observations are time-intensive and prone to bias related to limited circumstances that allow animal viewing (OwenSmith 1979). As such, studies usually include observations occurring on the timescale of hours or minutes (Owen-Smith et al. 2010) and rarely at night, limiting inference. Further, topography and vegetation often influence observability, raising the potential for habitat-induced bias. Therefore, visual observations tend to only produce surrogate measures of foraging investment, such as departure time from feeding stations, given the number of actual bites and steps taken can be obscured. Lastly, many species, including large herbivores, travel long distances, which can make behavioral observations of the same individual difficult. Acoustic monitoring has the potential to address many of the common problems in studying foraging behavior of animals. Passive acoustic monitoring has been used in numerous behavioral studies to investigate communication (Kroodsma 2015), assess responses to human-caused noise (Shannon et al. 2016), and examine a suite of detailed behaviors (Kalan et al. 2016; Higashisaka et al. 2018). On-animal recording devices allow constant monitoring, despite potentially poor visibility due to weather or thick vegetation, time of day, or distance traveled by the animal (Lynch et al. 2013). These benefits allow for a more extensive study than traditional behavioral observations can supply and remove the potential biases caused by behavioral or physiological responses to human presence (MacArthur et al. 1982; Steen et al. 1988). On-animal acoustic recording devices have been deployed on marine mammals (Johnson et al. 2009) but have been less frequently applied in terrestrial systems. However, acoustic monitoring devices show great promise for use on large terrestrial herbivores that can be equipped simultaneously with tracking devices (e.g., Lynch et al. 2015). �NORTHRUP ET AL.—ACOUSTIC MONITORING OF UNGULATE FORAGING Service Natural Sounds and Night Skies Division, Fort Collins, Colorado) that allows users to view spectrograms and play back corresponding audio (Fig. 1; Supplementary Data SD1). Only times when behaviors of interest (see below) were audible were used in further analyses. Lynch et al. (2013, 2015) previously identified a number of behaviors distinguishable from these acoustic data, including foraging, walking, and rumination, and quantified activity bouts for the same animals in this study, determining that deer spent most of their time resting or ruminating and foraging (activity budgets that matched those reported in other studies using telemetry or observational approaches—Lynch et al. 2013). These authors further validated that the recorded behaviors corresponded to observed behaviors by observing captive deer equipped with audio collars and listening to corresponding audio files. The authors found 100% agreement between the behaviors observed and those identified through visual and auditory inspection of the resulting spectrograms and audio files. We were interested specifically in foraging behavior and thus first reviewed the data to identify foraging bouts, classified by the sound of cropping and chewing vegetation (see Supplementary Data SD1). A foraging bout was considered to begin with the first bite following a period of rumination, walking (with no bites), or resting (when little was audible as the deer was relatively inactive), and ended when rumination started (regurgitating, masticating, and swallowing—Lynch et al. 2015). We randomly selected start times within the acoustic data and moved forward in time, first identifying potential bouts visually (Fig. 1), and then by listening to them. If the randomly selected start time fell within a foraging bout, we moved forward in time until the next foraging bout began. When rumination did not occur after a foraging bout, the bout was considered to be over when no bites could be heard for > 5 min. Two to four foraging bouts were identified using this approach for each deer and day. For each foraging bout, the length of the bout was first recorded. We then quantified the number of bites (leaves or stems being cropped off a plant) and the number of steps (each time one of the deer’s feet hit the ground audibly). Chewing associated with foraging was considered part of the foraging bout but was not counted toward the bite total, which only consisted of leaves or stems being cropped off a plant. Any mastication associated with rumination was considered to be part of rumination in the above classifications. See Supplemental Data SD1 for examples of these different behaviors. Because bouts did not occur at the exact time that GPS locations were taken, we matched each bout to the GPS location closest in time to the bout. In some cases, the GPS failed to obtain a fix and thus there was a greater amount of time between when the bout occurred and when we had location data. In these cases, we randomly selected a new foraging bout until we found a bout for which there was a successful fix within 15 min (the longest potential time between a successful fix and a foraging bout). Occasionally, interference from the deer’s head rubbing against vegetation or the GPS collar knocking against the audio collar was too high for an accurate count of bites or steps. In these cases (11 bouts in total), a new foraging bout was selected. Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 area is subject to extensive oil and gas development as well as hunting of mule deer and elk (Cervus canadensis) in the fall. We captured adult female mule deer (> 1 year old) using helicopter net-gunning as part of a long-term mule deer monitoring project in the Piceance Basin (Anderson 2015). In December 2011, we captured 10 deer, nine of which had been previously fit with store-on-board Global Positioning System (GPS; G2110D, Advanced Telemetry Systems, Isanti, Minnesota) radiocollars (see Northrup et al. 2014a for a detailed description of capture procedures). In March 2013, we recaptured 10 deer including two that were part of the December 2011 sample. Upon capture, deer were hobbled, blindfolded, and transferred to a central processing site, typically within 1 km of the capture location (always ≤ 5 km). At the processing site, deer were weighed and a set of morphometric measurements were taken. In addition, we measured the thickness of subcutaneous rump fat and the depth of the longissimus dorsi muscle using ultrasound (Stephenson et al. 1998), and calculated a body condition score following Cook et al. (2001, 2007, 2010). During December 2011, we fit deer with a new store-on-board GPS radiocollar set to attempt a relocation once every 30 min. Deer captured in March maintained their previously fit GPS collars. During both December 2011 and March 2013, we fit deer with an additional custom-designed audio recording collar positioned in front of the GPS collar (i.e., closer to the mouth) with the microphone facing the mouth of the deer (Lynch et al. 2013, 2015). Recorders were set to begin recording at midnight on the day of the capture, as mule deer can display altered movement behavior for a short time following helicopter capture (Northrup et al. 2014a). As we were uncertain of how long the acoustic collars would remain functional, we attempted to balance behavioral effects of capture with the ability to collect sufficient data for analyses. Recorders were set to record for approximately 2 weeks, at which point a timed drop-off mechanism (Lotek Wireless, Inc., Newmarket, Ontario, Canada) would engage. Collars were spliced and reattached with a 6.35-mm diameter latex tubing that degrades over time in case the drop-off mechanism failed. Collars also were equipped with a very high frequency (VHF) ear tag transmitter (series M3600, Advanced Telemetry Systems) with a mortality sensor to facilitate collar recovery. For detailed information on the collar specifications, see Lynch et al. (2013). All capture and collaring procedures followed ASM guidelines (Sikes et al. 2016) and were approved by the Colorado State University (protocol ID: 10-2350A) or Colorado Parks and Wildlife (protocol ID: 15-2008) animal care and use committees. Quantifying foraging behavior.—We monitored deer daily from the ground and biweekly from a fixed-wing aircraft using VHF radio telemetry to determine if they had suffered a mortality or if the acoustic collar had fallen off. Once collars were collected, the continuous acoustic data were converted from MP3 (44.1 kHz sample rate, 128 kbps bit rate) to hourly sound pressure level files using a protocol described in Mennitt and Fristrup (2012). The resulting data were used to produce spectrograms ranging in frequency from 20 to 6,300 Hz (Lynch et al. 2013). We reviewed the acoustic data using the Acoustic Monitoring Toolbox software (U.S. National Park 1481 �1482 JOURNAL OF MAMMALOGY Statistical analysis.—The process described above provided us with information on the duration of bouts, and the number of steps and the number of bites taken within those bouts. The ratio of the number of bites taken per step is a measure of foraging behavior, with a greater number of bites per step indicative of higher forage payoff per unit of movement (see Novellie 1978). However, the time over which foraging occurs also is important, as time spent foraging in a location has opportunity costs for foraging in other, possibly more productive, areas or engaging in other activities. We first compared data collected from different time periods (e.g., March versus December and night versus day) using Wilcoxon rank sum tests. Next, we fit a set of regression models (see below) to the ratios of bites per step and bites per minute, allowing us to examine our predictions. First, we examined the influence of a suite of environmental variables representing either forage availability or cost of movement on mule deer foraging. These variables included vegetation type with three categories: shrub, treed, and open land cover types. Treed land cover was included as the reference category. Other environmental variables included in the models were the distance to treed edges, snow depth (estimated daily at a 30-m resolution using a distributed snow evolution model—Liston and Elder 2006; Northrup et al. 2016), a terrain ruggedness index (calculated based on the square difference between each cell of a digital elevation model obtained from the United States Geological Survey National Elevation Dataset; http://nationalmap.gov/elevation.html), and the normalized difference vegetation index (NDVI; a biweekly measure of primary productivity downloaded from http://www.vgt.vito.be/). In addition, we included a binary covariate indicating if foraging occurred during the day or night (defined as the time between sunset and sunrise using times for Meeker, Colorado from http://aa.usno.navy. mil/data/docs/RS_OneYear.php). We next fit models with only anthropogenic disturbance covariates. These covariates included the distance to pipelines (obtained from the Bureau of Land Management White River Field Office and refined by digitizing missing pipelines using National Agricultural Imagery Program [NAIP] imagery), and the distance to natural gas facilities including compressor stations, natural gas plants, and other industrial facilities obtained by digitizing and ground-truthing the NAIP imagery. We also examined the influence of the distance to natural gas well pads that were either being actively drilled or were in some other state (primarily actively producing natural gas, but also some undergoing hydraulic fracturing). Detailed description of how these classifications were developed can be found in Northrup et al. (2016). In brief, we obtained daily classifications of well pads using publicly available data from the Colorado Oil and Gas Conservation Commission (http://cogcc.state.co.us/), which provides daily updated status of all oil and gas wells in the state. Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Fig. 1.—Example spectrograms from acoustic recording collars deployed on female mule deer (Odocoileus hemionus) in December 2011 and March 2013 in the Piceance Basin of Colorado, United States. Spectrograms show (A) initiation of a foraging bout, (B) termination of a foraging bout with rumination, and (C) termination of a foraging bout with no bites or steps for > 5 min. Brightness represents sound intensity; brighter = louder. Arrow indicates beginning (A) or end (B and C) of foraging. �NORTHRUP ET AL.—ACOUSTIC MONITORING OF UNGULATE FORAGING yij ∼ NB(µij , φ) log(µij ) = αj + xi β + log(offset) � αj ∼ Normal µα , σα2 β ∼ Normal(0, 5I) µα ∼ Normal(0, 5) � σα ∼ Half − normal(0, 100) φ ∼ Half − cauchy(0, 10), where yij is the number of bites taken by individual j during bout i, αj is an individually varying intercept, xi is a matrix of covariates with corresponding coefficients β (note that bold indicates a matrix or vector; see Edwards and Auger‐Méthé 2019). We fit models using STAN (STAN Development Team 2014b) in the R statistical software (R Core Development Team 2015) using the “RStan” package (STAN Development Team 2014a). For each model, we ran four chains for 1,500 iterations, removing the first 750 as burn-in. We initiated each chain with random values and assessed convergence to the posterior distribution using the Gelman–Rubin diagnostic, with values below 1.1 indicative of convergence (Gelman and Rubin 1992). We ran models with data from March and December separately. We fit three models each for March and December, the first including all environmental covariates, the second including all anthropogenic covariates, and the last including only body fat. In all models, we standardized all covariates (x−x̄ σ ) and used the magnitude of the median coefficient and proportion of posterior distributions for coefficients that fell on each side of 0 to infer the degree of evidence for and strength of an effect of a covariate. Results We recovered the acoustic collars for all 10 deer collared in December 2011, and eight of 10 deer collared in March 2013. Each collar recorded 11–16 days of acoustic data, with a mode of 14 days. Of the collars from December 2011, the audio data were of sufficient quality for distinguishing bites and steps for all collars. However, the GPS collar on one deer malfunctioned for the first several days and thus we excluded this individual from analyses. Of the collars from March 2013, the acoustic data were masked by interference for three collars. These issues resulted in viable data for counting bites and steps, and subsequently modeling the ratio between them, from nine deer during December 2011 and five deer during March 2013, equating to 185 bouts in December 2011 and 140 bouts in March 2013. Overall, the number of bites taken per step was higher in March ( x̄ = 4.38, SD = 5.72) than December ( x̄ = 2.47, SD = 4.02; Wilcoxon rank sum test W = 17,922, P < 0.0001; Fig. 2A), while the number of bites taken per minute was Fig. 2.—Boxplots showing population-level summaries of (A) bites taken per step and (B) foraging bout length for adult female mule deer (Odocoileus hemionus) in the Piceance Basin, Colorado, United States in March 2013 and December 2011. Solid black lines in the middle of boxes represent medians, box edges extend to the interquartile range, and the whiskers extend to the data extremes. Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Lastly, we qualitatively compared models fit to the data collected in December to those collected in March. During December, deer in this area have high energetic reserves, having recently returned from summer range, while these stores are largely depleted in March (Anderson 2015). Thus, we expected deer to prioritize foraging over other activities in March relative to December. To further test this prediction, we fit a model to deer foraging behavior in each season with a single covariate for estimated percent body fat. We fit a negative binomial regression in a Bayesian hierarchical framework to the different data sets. We fit models to the number of bites per bout, with the number of steps or bout length (in minutes) included as an offset in the model. We allowed the intercept for the model to vary by individual to account for multiple bouts per individual. The model took the following form: 1483 �1484 JOURNAL OF MAMMALOGY happened during the day with 58% at night (at this time it is light for approximately 40% of a day). Mean body fat was higher in December ( x̄ = 15.0%, range = 9.4–20.2%) than March ( x̄ = 6.6%, range = 5.9–7.3%). There were strong differences in the regression model results between seasons with models of bites per step and bites per minute being largely congruent (Table 1). For the models assessing the influence of environmental covariates during December, there was strong support for increased foraging effort where costs of movements were high, which in our case was captured by increased bites per step and bites per minute in deeper snow (Table 1). We also found evidence for increased bites per step in areas with presumed higher vegetative biomass, identified as areas with higher NDVI (Fig. 3), but no effect on bites per minute. Bites per step were lower during the night but bites per minute did not follow the same pattern (Table 1). Both bites per step and bites per minute were lower when deer were in shrub land cover relative to treed land cover (Table 1). In contrast, there was no evidence for a snow effect on either foraging metric in March (Table 1), and deer took more bites per step and bites per minute at night. There was moderate evidence for fewer bites per step in areas with higher NDVI (Fig. 3), as well as fewer bites per step and bites per minute in open and shrub land cover relative to treed land cover. For the models assessing the influence of anthropogenic variables in December, the results for bites per step and bites per minute were less congruent. There was evidence that facilities affected deer foraging in that they took more bites per step and more bites per minute further from facilities (Table 2). Table 1.—Results of hierarchical negative binomial regressions on the number of bites of forage taken by mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado in either December 2011 or March 2013. Shown are the median posterior coefficient estimates as well as the proportion (prop.) of posteriors that were less than or greater than 0, which indicates the posterior probability of a coefficient being negative or positive, respectively. The results presented below are from models used to assess the influence of environmental factors on deer foraging. The number of steps a deer took or the number of minutes were included as offsets in the models to assess the rate of bites taken per step or minute, respectively. Bites per step model Month December March Median Prop. < 0 Prop. > 0 Median Overall intercept Snow depth NDVIa TRIb Openc Shrubd Edgee Night Overall intercept Snow depth NDVIa TRIb Openc Shrubd Edgee Night 0.94 0.24 0.13 0.01 −0.01 −0.24 0.11 −0.27 1.31 0.06 −0.12 −0.04 −0.75 −0.42 0.01 0.44 0.01 0.06 0.17 0.46 0.52 0.95 0.13 0.92 0.01 0.22 0.86 0.67 1 0.97 0.44 0 0.99 0.94 0.83 0.54 0.48 0.05 0.87 0.08 0.99 0.78 0.14 0.33 0 0.03 0.56 1 3 0.19 0.06 0.01 −0.03 −0.15 0.00 0.03 2.91 0.03 −0.01 0.03 −0.47 −0.28 0.01 0.09 Normalized difference vegetation index. Terrain ruggedness index. c Open land cover. Reference category is treed land cover. d Shrub land cover. Reference category is treed land cover. e Distance to treed edges. a b Bites per minute model Covariate Prop. < 0 0 0.03 0.26 0.44 0.57 0.95 0.51 0.35 0 0.22 0.59 0.31 1.00 0.99 0.40 0.16 Prop. > 0 1 0.97 0.75 0.56 0.44 0.06 0.49 0.65 1 0.78 0.41 0.69 0.00 0.01 0.61 0.84 Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 higher in December ( x̄ = 20.05, SD = 11.81) than March ( x̄ = 16.00, SD = 8.10; Wilcoxon rank sum test W = 15,621, P < 0.01). During March, the number of bites taken per step was not statistically significantly different between night and day, despite being higher at night (night x̄ = 5.29, SD = 7.12; day x̄ = 3.44, SD = 3.57; Wilcoxon rank sum test W = 2,141, P = 0.25). Similarly, in December there was no difference in the number of bites taken per step between night and day, though the magnitude of values showed the opposite pattern as in March (day x̄ = 3.25, SD = 5.66; night x̄ = 1.92, SD =2.10; Wilcoxon rank sum test W = 4,448, P = 0.87). The number of bites taken per minute did not differ between night and day during December (night x̄ = 19.84, SD =11.1; day x̄ = 20.34, SD =12.81; Wilcoxon rank sum test W = 4,334, P = 0.89) or March (night x̄ = 16.42, SD = 8.44; day x̄ = 15.55, SD = 7.75; Wilcoxon rank sum test W = 2,315, P = 0.68). Bouts were longer in March ( x̄ = 9.77, SD = 8.84) compared to December ( x̄ = 7.17, SD = 4.31; Fig. 2B), with the differences being nearly significant (Wilcoxon rank sum test W = 14,864, P = 0.052). Mean foraging bout lengths were not different between night and day in either season, though mean bout lengths were longer during the night (March: night x̄ = 10.71, SD = 9.97; day x̄ = 8.79, SD = 7.43; Wilcoxon rank sum test W = 2,191.5, P = 0.35; December: night x̄ = 7.27, SD = 4.47; day x̄ = 6.97, SD = 4.46; Wilcoxon rank sum test W = 4,102.5, P = 0.45). The temporal patterns of foraging varied between seasons as well, though these differences followed seasonal differences in daylight. In March, 55% of foraging bouts happened during the day with 45% at night (at this time it is light for approximately 47% of a day). In December, 42% of foraging bouts �NORTHRUP ET AL.—ACOUSTIC MONITORING OF UNGULATE FORAGING 1485 Table 2.—Results of hierarchical negative binomial regressions on the number of bites of forage taken by mule deer (Odocoileus hemionus) in the Piceance Basin of Colorado in either December 2011 or March 2013. Shown are the median posterior coefficient estimates as well as the proportion (prop.) of posteriors that were less than or greater than 0, which indicates the posterior probability of a coefficient being negative or positive, respectively. The results presented below are from models used to assess the influence of anthropogenic factors on deer foraging. The number of steps a deer took or the number of minutes were included as offsets in the models to assess the rate of bites taken per step or minute, respectively. Bites per step model Month December March Covariate Median Overall intercept Dist. drilling pada Dist. other padb Dist. facilitiesc Dist. pipelinesd Night Overall intercept Dist. drilling pada Dist. other padb Dist. facilitiesc Dist. pipelinesd Night 0.83 −0.07 0.13 0.11 −0.05 −0.24 1.21 0.09 −0.23 0.08 0.05 0.31 Prop. < 0 0 0.78 0.07 0.07 0.72 0.96 0.02 0.38 0.97 0.31 0.31 0.02 Bites per minute model Prop. > 0 Median Prop. < 0 Prop. > 0 1 0.22 0.93 0.93 0.28 0.04 0.98 0.62 0.03 0.69 0.69 0.98 2.94 −0.09 0.04 0.09 0.03 0.02 2.73 −0.04 0.00 0.01 −0.02 0.04 0 0.94 0.25 0.03 0.31 0.43 0 0.62 0.52 0.45 0.61 0.33 1 0.06 0.75 0.97 0.69 0.57 1 0.38 0.48 0.55 0.39 0.67 Distance to natural gas well pads with active drilling ongoing. Distance to natural gas well pads that were in a status other than drilling. c Distance to natural gas facilities, including compressor stations and gas plants. d Distance to natural gas pipelines. a b Moreover, there was evidence of an effect of producing well pads and nighttime on bites per step, and evidence of an effect of drilling pads on bites per minute (Table 2). Deer took more bites per step further from producing pads and facilities, fewer bites per step during the night, and more bites per minute closer to drilling well pads (Table 2). For the March model, contrasting results were found for bites per step with evidence for an effect of producing well pads and nighttime, with deer taking more bites per step close to pads and during the night (Table 2). No covariates had strong influence on bites per minute in March (Table 2). Lastly, regression models indicated that fatter deer took fewer bites per step (β = −0.57, prob. β < 0 = 0.97; Fig. 4A) and fewer bites per minute (β = −0.22, prob. β < 0 = 0.88) in December, while there was no apparent effect of body fat on bites per step (β = 0.50, prob. β > 0 = 0.58; Fig. 4B), or bites per minute (β = 0.23, prob. β > 0 = 0.77) in March when body fat was less variable. Discussion Optimal foraging theory provides the basis for understanding an array of processes in animal behavior and ecology. With increased ability to collect spatial data on animals through GPS radiocollars, this theory is being invoked increasingly to explain complex spatial behavioral patterns (Owen-Smith et al. 2010). However, the difficulties involved with observing foraging decisions of animals have limited the number of robust empirical examinations of foraging in the wild. Our combined location and acoustic data allowed us to examine characteristics influencing foraging effort within patches in a manner that has previously been unattainable. Although we quantified direct measures of foraging (i.e., bites, steps, and time investment), our metrics are not complete measures of energetic costs and gains, but rather represent aspects of the intensity and efficiency of foraging. Our results provide insight Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Fig. 3.—Predicted effect of normalized difference vegetation index (NDVI) on bites per step in March 2013 (left panel) and December 2011 (right panel) for adult female mule deer (Odocoileus hemionus) equipped with acoustic recording collars in the Piceance Basin, Colorado, United States. �1486 JOURNAL OF MAMMALOGY to the individual, environmental, and anthropogenic factors that shape these aspects of foraging. We predicted that deer in better condition would forage less intensely and efficiently due to the opportunity cost and potential predation risk of foraging. Our results largely supported this prediction. When deer were in poorer condition and ecological conditions were improving, they tended to take more bites per step and more bites per minute. In December, when vegetation is senescent, deer in the study system have recently returned from summer range and are in relatively good condition, while in March, when the spring green-up begins, they have depleted their fat stores (Northrup et al. 2014b; Anderson 2015). As a result, there is greater variance in residual body fat stores in December (SD = 3.46) than March (SD = 0.48), which may partly reflect reproductive effort over the summer (Bergman et al. 2018). Given this context, we would expect foraging behavior in December to be influenced to a greater degree by fat stores than in March, which mirrored our results. In December, foraging was strongly correlated with body fat; deer in better condition took fewer bites per step and per minute, potentially reflecting a trade-off between predation risk and resource access. Interestingly, we found somewhat contrasting results seasonally between the two metrics of foraging. Although the model results relating body fat to foraging were similar, on average, deer took more bites per step in March when they were in poor condition, but took fewer bites per minute due to longer overall foraging bouts. This finding indicates that during March, processing time per bite must be higher, or that the costs of movement are greater, thus requiring greater depletion of resources before it is worthwhile for deer to move. Deer in March also might have higher metabolic costs due to the increasing costs of pregnancy. We did not test for an effect of pregnancy, and based on the consistently low body fat across individuals in March, do not expect it to influence foraging (i.e., our results indicate all individuals should be focused on forage intake at this time). We also predicted that deer would forage more intensely and efficiently in areas that we assumed were related to higher costs of movement and higher forage availability. Our results provided seasonally differentiated support for our predictions. During December, deer took more bites per step and per minute in areas of deeper snow, and there was moderate evidence that deer took more bites per step in areas of higher NDVI. Results were the converse in March, when the number of bites per step was negatively related to NDVI and both metrics were equivocal relative to snow depth. These findings suggest that deep snow represents an energetic cost to deer in December, eliciting lower movement rates (Parker et al. 1984), but is less important during March. Likewise, NDVI appears to be related to forage availability in December but not March. We caution that we lack detailed information on the true relationship between NDVI and the nutritional value of plants eaten by mule deer in our study area, so these results should be interpreted with caution. During December, vegetation has ceased growing in the study system, reducing quality, and thus biomass might be the best predictor of forage availability. Further, the likely lower per-bite value of forage in December, and thus lower opportunity costs, could increase the relative cost of movement through snow at this time. In March, initial green-up occurs, correlated with a recorded increase in deer movements (Northrup 2015). At this time, NDVI appears to be a poor indicator of forage availability and the costs of movement through snow appear to be less influential. These factors seem to drive deer to switch foraging strategies between seasons. During March, when body condition was at the yearly low and the onset of spring primary production begins, deer generally employed a consistent strategy, with a greater than 2-fold increase in the number of bites per step relative to that in December. Body fat provided no additional explanatory power in March when all deer had used fat stores and were in similar condition (i.e., variation in body fat scores was much lower). In developing our predictions, we assumed anthropogenic features were associated with greater human activity and perceived as risky areas by deer, leading to adjustment in foraging hypothesized by the risk-disturbance hypothesis (Frid and Dill 2002). As such, we predicted deer would forage more intensely in areas close to human development because these areas should only be accessed if they are rich in forage resources (Brown and Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Fig. 4.—Predicted effect of percent body fat on bites per step in March 2013 (left panel) and December 2011 (right panel) for adult female mule deer (Odocoileus hemionus) equipped with acoustic recording collars in the Piceance Basin, Colorado, United States. �NORTHRUP ET AL.—ACOUSTIC MONITORING OF UNGULATE FORAGING the interaction between physiology and environmental factors that structure foraging. We found that individual condition was the overriding factor influencing foraging tactics, and that our predictions bore out during periods more conducive to differential forage returns from alternate foraging tactics. The differences between the March and December models in regards to snow and NDVI highlight this contrast. Similarly, the trade-offs between risk and forage access impacted foraging strategies in ways that were difficult to predict. For example, we expected edges to relate positively to forage availability, but there is evidence that deer perceive edges as more risky than other habitats, and thus the finding of fewer bites per step close to edges could be related to such perceptions (Altendorf et al. 2001). Future work on the influence of predatory risk and forage quality, leveraging improved remote sensing data to quantify forage availability and quality, offer opportunities for assessing the factors driving foraging decisions of large mammals. Leveraging acoustic monitoring for wildlife research necessitates recognizing and adjusting for its limitations. Acoustic data are information rich, but extracting detailed data on specific behaviors is time-intensive. Proxies for behaviors of interest can provide meaningful alternatives to detailed data extractions. In this study, measuring the length of each bout provided a direct metric of foraging effort that was easier to extract, though it lacked the nuance needed to test our specific hypotheses because longer bouts may or may not equate to more actual foraging. Our metrics were themselves limited because we do not know the true energetic value of each bite taken. Further, there is likely to be an inverse relationship between bite size and the number of bites taken, which could bias results. Distortion and interference from branches brushing against the recorder and high levels of ambient noise, such as sounds of human activities, obscured behaviors or sounds of interest in our data. Generally, we faced power limitations with our recorders, and batteries on some collars did not last as long as was planned, cutting short data collection. This was in part because we ran our recordings continuously. Specific sampling regimes can extend the life of acoustic monitors. Despite these potential challenges, on-animal acoustic monitoring is a unique and innovative method for collecting behavioral data that provides novel opportunities that we did not leverage. For example, extraction of ambient anthropogenic noise levels from collars can provide information on the acoustic environment that animals encounter. In this study, we focused on deer-produced sounds, but a different placement of the recorder in relation to the deer (i.e., back versus front of the neck) would allow a focus on environmental sounds (Lynch et al. 2013). We urge further development of this technology to optimize it for future research. Acknowledgments 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, Williams Exploration & Production, Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Kotler 2004). Our models assessing the effect of anthropogenic covariates indicated mixed support for these predictions. In contrast to our predictions, deer in December took more bites per step and minute in areas further from facilities and more bites per step in areas further from producing well pads. In accordance with predictions, deer took more bites per minute in areas closer to drilling pads in December. Deer did not respond to any other covariates in the expected manner and showed no congruent responses in March. The partial support for the risk-disturbance hypothesis in December, but not March, is in accordance with our other findings that deer showed more variability in their foraging decisions during December relative to March, presumably related to differences in deer condition. The opposite response to development in March could indicate that the need to forage on newly available, high-quality forage, overrides any potential perceived risk from human activities. Alternatively, these seasonal differences could be related to unknown differences in the types of forage around human development and that their availability might differ between seasons. The weak effect of development covariates may also be due to habituation and interference from other factors. While oil and gas development does seem to affect mule deer habitat selection (Northrup et al. 2015) and vigilance to a degree (Lynch et al. 2015), its effects on fine-scale foraging behavior are more complex. These fine-scale decisions in relation to habitat are directly dependent on large-scale choices (Northrup et al. 2016). Therefore, if deer are already modifying their behavior to avoid development, they do not need to further alter fine-scale foraging decisions. Previous work on deer in our study system determined that spatial patterns of predation were more strongly structured by anthropogenic features than natural features (Lendrum et al. 2018). Some features were associated with high risk (e.g., proximity to roads) and others with less risk (e.g., proximity to pipelines). However, these relationships changed over time and in relation to the general activity level, indicating predation was dynamic in time and space relative to development. Deer foraging was related to edges and shrub land cover, areas we assumed to be important for foraging but also associated with greater risk of predation (Lendrum et al. 2018). Further, deer did not respond to terrain ruggedness, which we hypothesized to be related to a greater cost of movement and potentially reduced predation risk (Lingle 2002). However, terrain ruggedness also was not predictive of the spatial patterns of predation in the system (Lendrum et al. 2018). Thus, it is likely that the documented patterns of foraging intensity are driven by spatial patterns of predation risk (as predicted by optimal foraging theory), but that these patterns are more complex than could be fully captured in our assessment. New technologies such as on-animal acoustic monitors provide exciting opportunities to enhance understanding of animal behavior. While foraging is fundamental to an understanding of ecology, direct assessments have been difficult to conduct in the wild. Our application of acoustic collars allowed detailed inspection of factors influencing foraging. The departure of our results from predictions provides valuable information regarding 1487 �1488 JOURNAL OF MAMMALOGY Supplementary Data Supplementary data are available at Journal of Mammalogy online. Supplementary Data SD1.—The supplementary.zip file contains five audio files and corresponding spectrograms for examples of behaviors that were examined in the paper. These include cropping of vegetation (two files), chewing vegetation (one file), walking (one file), and walking to moving water (one file). Literature Cited Altendorf, K. B., J. W.Laundré, C. A. López González, and J. S. Brown. 2001. Assessing effects of predation risk on foraging behavior of mule deer. Journal of Mammalogy 82:430–439. Anderson, C. R., Jr. 2015. Population performance of the Piceance basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Job Progress Report, Colorado Parks and Wildlife, Fort Collins, Colorado. Bartmann, R. M. 1983. Composition and quality of mule deer diets on pinyon-juniper winter range, Colorado. Journal of Range Management 36:534–541. Bergman, E. J., C. R. Anderson, Jr., C. R. Bishop, Jr., A. A. Holland, and J. M. Northrup. 2018. Variation in ungulate body fat: individual versus temporal effects. Journal of Wildlife Management 82:130–137. Brown, J. S. 1999. Vigilance, patch use and habitat selection: foraging under predation risk. Evolutionary Ecology Research 1:49–71. Brown, J. S., and B. P. Kotler. 2004. Hazardous duty pay and the foraging cost of predation. Ecology Letters 7:999–1014. Brown, J. S., J. W. Laundre, and M. Gurung. 1999. The ecology of fear: optimal foraging, game theory, and trophic interactions. Journal of Mammalogy 80:385–399. Charnov, E. L. 1976. Optimal foraging, the marginal value theorem. Theoretical Population Biology 9:129–136. Ciuti, S., et al. 2012. Effects of humans on behaviour of wildlife exceed those of natural predators in a landscape of fear. PLoS One 7:e50611. Cook, R. C., J. G. Cook, D. L. Murray, P. Zager, B. K. Johnson, and M. W. Gratson. 2001. Development of predictive models of nutritional condition for Rocky Mountain elk. Journal of Wildlife Management 65:973–987. Cook, R. C., et al. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Cook, R. C., T. R. Stephenson, W. L. Myers, J. G. Cook, and L. A. Shipley. 2007. Validating predictive models of nutritional condition for mule deer. Journal of Wildlife Management 71:1934–1943. Edwards, A. M., and M. Auger-Méthé. 2019. Some guidance on using mathematical notation in ecology. Methods in Ecology and Evolution 10:92–99. Frid, A., and L. M. Dill. 2002. Human-caused disturbance stimuli as a form of predation risk. Ecology and Society 6:11. Gelman, A., and D. B. Rubin. 1992. Inference from iterative simulation using multiple sequences. Statistical Science 7:457–511. Gross, J. E., L. A. Shipley, N. T. Hobbs, D. E. Spalinger, and B. A. Wunder. 1993. Functional response of herbivores in foodconcentrated patches: tests of a mechanistic model. Ecology 74:778–791. Higashisaka, H., T. Matsuishi, and T. Akamatsu. 2018. Presence and behavior of harbor porpoises (Phocoena phocoena) around set nets revealed using passive acoustic monitoring. Fisheries Research 204:269–274. Humphries, N. E., H. Weimerskirch, N. Queiroz, E. J. Southall, and D. W. Sims. 2012. Foraging success of biological Lévy flights recorded in situ. Proceedings of the National Academy of Sciences 109:7169–7174. Johnson, M., N. A. De Soto, and P. T. Madsen. 2009. Studying the behaviour and sensory ecology of marine mammals using acoustic recording tags: a review. Marine Ecology Progress Series 395:55–73. Kalan, A. K., A. K. Piel, R. Mundry, R. M. Wittig, C. Boesch, and H. S. Kühl. 2016. Passive acoustic monitoring reveals group ranging and territory use: a case study of wild chimpanzees (Pan troglodytes). Frontiers in Zoology 13:34. Kroodsma, D. 2015. The singing life of birds: the art and science of listening to birdsong. Houghton Mifflin Harcourt, New York, New York. Lendrum, P. E., et al. 2018. Predation risk across a dynamic landscape: effects of anthropogenic land use, natural landscape features, and prey distribution. Landscape Ecology 33:157–170. Lingle, S. 2002. Coyote predation and habitat segregation of whitetailed deer and mule deer. Ecology 83:2037–2048. Liston, G. E., and K. Elder. 2006. A distributed snow-evolution modeling system (SnowModel). Journal of Hydrometeorology 7:1259–1276. Lynch, E., L. Angeloni, K. Fristrup, D. Joyce, and G. Wittemyer. 2013. The use of on-animal acoustical recording devices for studying animal behavior. Ecology and Evolution 3:2030–2037. Lynch, E., J. M. Northrup, M. F. Mckenna, C. R. Anderson, Jr., L. Angeloni, and G. Wittemyer. 2015. Landscape and anthropogenic features influence the use of auditory vigilance by mule deer. Behavioral Ecology 26:75–82. MacArthur, R. A., V. Geist, and R. H. Johnston. 1982. Cardiac and behavioral responses of mountain sheep to human disturbance. Journal of Wildlife Management 46:351–358. MacArthur, R. H., and E. R. Pianka. 1966. On optimal use of a patchy environment. The American Naturalist 100:603–609. Mennitt, D., and K. Fristrup. 2012. Obtaining calibrated sound pressure levels from consumer digital audio recorders. Applied Acoustics 73:1138–1145. Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 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 L. Gepfert (CPW pilot) for assisting with deer captures. We thank E. Bergman, M. Alldredge, two anonymous reviewers, and the associate editor for comments on earlier versions of this manuscript and C. White, S. Brandenburg, and J. Job for assistance with processing acoustic data. �NORTHRUP ET AL.—ACOUSTIC MONITORING OF UNGULATE FORAGING R Core Development Team. 2015. R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/. Shannon, G., et al. 2016. A synthesis of two decades of research documenting the effects of noise on wildlife. Biological Reviews of the Cambridge Philosophical Society 91:982–1005. Sikes, R. S., and The Animal Care and Use Committee of the American Society of Mammalogists. 2016. 2016 Guidelines of the American Society of Mammalogists for the use of wild mammals in research and education. Journal of Mammalogy 97:663–688. STAN Development Team. 2014a. RStan: the Rinterface to Stan, version 2.4. http://mc-stan.org/rstan.html. STAN Development Team. 2014b. Stan: a C++ library for probability and sampling, version 2.4. http://mc-stan.org Steen, J. B., G. W. Gabrielsen, and J. W. Kanwisher. 1988. Physiological aspects of freezing behaviour in willow ptarmigan hens. Acta Physiologica Scandinavica 134:299–304. Stephenson, T. R., K. J. Hundertmark, C. C. Schwartz, and V. Van Ballenberghe. 1998. Predicting body fat and body mass in moose with ultrasonography. Canadian Journal of Zoology 76:717–722. Wajnberg, E., P. Bernhard, F. Hamelin, and G. Boivin. 2006. Optimal patch time allocation for time-limited foragers. Behavioral Ecology and Sociobiology 60:1–10. Wilson, R. P., F. Quintana, and V. J. Hobson. 2012. Construction of energy landscapes can clarify the movement and distribution of foraging animals. Proceedings of the Royal Society of London, B. Biological Sciences 279:975–980. Wittemyer, G, J. M. Northrup, and G. Bastille-Rousseau. In press. Behavioural valuation of landscapes using movement data. Philosophical Transactions of the Royal Society of London, B. Biological Sciences. doi:10.1098/rstb.2018.0046 Submitted 10 July 2018. Accepted 8 July 2019. Associate Editor was Robert Schooley. Downloaded from https://academic.oup.com/jmammal/article/100/5/1479/5555688 by guest on 22 July 2021 Northrup, J. M. 2015. Behavioral response of mule deer to natural gas development in the Piceance Basin. Doctoral dissertation, Colorado State University, Fort Collins, Colorado. Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2014a. Effects of helicopter capture and handling on movement behavior of mule deer. Journal of Wildlife Management 78:731–738. Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2015. Quantifying spatial habitat loss from hydrocarbon development through assessing habitat selection patterns of mule deer. Global Change Biology 21:3961–3970. Northrup, J. M., C. R. Anderson, Jr., and G. Wittemyer. 2016. Environmental dynamics and anthropogenic development alter philopatry and space-use in a North American cervid. Diversity and Distributions 22:547–557. Northrup, J. M., A. B. Shafer, C. R. Anderson, Jr., D. W. Coltman, and G. Wittemyer. 2014b. Fine-scale genetic correlates to condition and migration in a wild cervid. Evolutionary Applications 7:937–948. Novellie, P. 1978. Comparison between the foraging strategies of blesbok and springbok on the Transvaal highveld. South African Journal of Wildlife Research 8:137–144. Owen-Smith, N. 1979. Assessing the foraging efficiency of a large herbivore, the kudu. South African Journal of Wildlife Research 9:137–144. Owen-Smith, N., J. M. Fryxell, and E. H. Merrill. 2010. Foraging theory upscaled: the behavioural ecology of herbivore movement. Philosophical Transactions of the Royal Society of London, B. Biological Sciences 365:2267–2278. Parker, K. L., C. T. Robbins, and T. A. Hanley. 1984. Energy expenditures for locomotion by mule deer and elk. Journal of Wildlife Management 48:474–488. Petchey, O. L., A. P. Beckerman, J. O. Riede, and P. H. Warren. 2008. Size, foraging, and food web structure. Proceedings of the National Academy of Sciences 105:4191–4196. Pyke, G. H., H. R. Pulliam, and E. L. Charnov. 1977. Optimal foraging - selective review of theory and tests. Quarterly Review of Biology 52:137–154. 1489 � https://cpw.cvlcollections.org/files/original/5f5f0ef089c27480d60667d992748575.zip 975f37f25bb2736e8059aeb186b02b33 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 On-animal acoustic monitoring provides insight to ungulate foraging behavior Description An account of the resource <span>Foraging behavior underpins many ecological processes; however, robust assessments of this behavior for free-ranging animals are rare due to limitations to direct observations. We leveraged acoustic monitoring and GPS tracking to assess the factors influencing foraging behavior of mule deer (</span><em>Odocoileus hemionus</em><span>). We deployed custom-built acoustic collars with GPS radiocollars on mule deer to measure location-specific foraging. We quantified individual bites and steps taken by deer, and quantified two metrics of foraging behavior: the number of bites taken per step and the number of bites taken per unit time, which relate to foraging intensity and efficiency. We fit statistical models to these metrics to examine the individual, environmental, and anthropogenic factors influencing foraging. Deer in poorer body condition took more bites per step and per minute and foraged for longer irrespective of landscape properties. Other patterns varied seasonally with major changes in deer condition. In December, when deer were in better condition, they took fewer bites per step and more bites per minute. Deer also foraged more intensely and efficiently in areas of greater forage availability and greater movement costs. During March, when deer were in poorer condition, foraging was not influenced by landscape features. Anthropogenic factors weakly structured foraging behavior in December with no relationship in March. Most research on animal foraging is interpreted under the framework of optimal foraging theory. Departures from predictions developed under this framework provide insight to unrecognized factors influencing the evolution of foraging. Our results only conformed to our predictions when deer were in better condition and ecological conditions were declining, suggesting foraging strategies were state-dependent. These results advance our understanding of foraging patterns in wild animals and hig</span> Bibliographic Citation A bibliographic reference for the resource. Recommended practice is to include sufficient bibliographic detail to identify the resource as unambiguously as possible. <p>Northrup, J. M., A. Avrin, C. R. Anderson Jr, E. Brown, and G. Wittemyer. 2019. On-animal acoustic monitoring provides insight to ungulate foraging behavior. Journal of Mammalogy 100:1479–1489. <a href="https://doi.org/10.1093/jmammal/gyz124" target="_blank" rel="noreferrer noopener">https://doi.org/10.1093/jmammal/gyz124</a> </p> Creator An entity primarily responsible for making the resource Northrup, Joseph M. Avrin, Alexandra Anderson Jr, Charles R. Brown, Emma Wittemyer, George Subject The topic of the resource Acoustic monitoring Bayesian hierarchical model Colorado Foraging behavior Herbivore foraging Mule deer <em>Odocoileus hemionus</em> Spatial ecology Extent The size or duration of the resource. 11 pages Date Created Date of creation of the resource. 2019-08-28 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> Format The file format, physical medium, or dimensions of the resource application/pdf Language A language of the resource English Is Part Of A related resource in which the described resource is physically or logically included. Journal of Mammalogy Type The nature or genre of the resource Article