Populations inhabiting the bioclimatic edges of a species’ geographic range face an increasing amount of stress from alterations to their environment associated with climate change. Moose Alces alces are large-bodied ungulates that are sensitive to heat stress and have exhibited population de- clines and range contractions along their southern geographic extent. Using a hidden Markov model to analyze movement and accelerometer data, we assigned behaviors (rest, forage, or travel) to all lo- cations of global positioning system-collared moose (n ¼ 13, moose-years ¼ 19) living near the south- ern edge of the species’ range in and around Voyageurs National Park, MN, USA. We assessed how moose behavior changed relative to weather, landscape, and the presence of predators. Moose sig- niﬁcantly reduced travel and increased resting behaviors at ambient temperatures as low as 15 C and 24 C during the spring and summer, respectively. In general, moose behavior changed season- ally in association with distance to lakes and ponds. Moose used wetlands for travel throughout the year, rested in conifer forests, and foraged in shrublands. The inﬂuence of wolves Canis lupus varied among individual moose and season, but the largest inﬂuence was a reduction in travel during spring when near a wolf home range core, primarily by pregnant females. Our analysis goes beyond habitat selection to capture how moose alter their activities based on their environment. Our ﬁnd- ings, along with climate change forecasts, suggest that moose in this area will be required to further alter their activity patterns and space use in order to ﬁnd sufﬁcient forage and avoid heat stress. Key words: Alces alces, climate change, heat stress, moose, temperature, wolves. Introduction potentially rapid and unpredictable ways, thus altering ecological Climate change is forecast to drive dramatic changes to the distri- communities (Parmesan and Yohe 2003). Ecosystem shifts can bution (Parmesan 2006; Chen et al. 2011) and abundance of spe- lead to greater abundance of competitors and predators (Huey cies worldwide (Dirzo et al. 2014; Ripple et al. 2016). Species will et al. 2009; Gilman et al. 2010), or changes in predator behavior face temperatures that more frequently exceed their thermal that modify competition or predator–prey dynamics (Post et al. thresholds (Po ¨ rtner 2001), and vegetative communities will shift in 1999). Animals inhabiting areas at the bioclimatic edges of their V C The Author (2017). Published by Oxford University Press. 1 This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited. For commercial re-use, please contact firstname.lastname@example.org Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 2 Current Zoology, 2017, Vol. 00, No. 00 range will encounter the most dramatic and earliest effects of cli- potential factors in nearly extirpating a moose population in north- mate change. Understanding how individual animals respond to western Minnesota, USA (Murray et al. 2006). Similarly, moose existing variation in those environmental conditions predicted to populations in the northeastern part of Minnesota have experienced sustain the greatest changes will provide the best indication of the population declines of more than 50% between 2005 and 2016 future persistence (Franco et al. 2006), abundance trends (DelGiudice 2016). Factors such as warmer year-round tempera- (Forchhammer et al. 2001), and range shifts (Hampe and Petit tures increased parasite loads and diseases from the northward ad- 2005; Hickling et al. 2006)ofa population. vance of white-tailed deer Odocoileus virginianus, and an increasing For mammals, it may take years for the effects of climate change wolf Canis lupus population are thought to all play an increasing to result in detectable population declines (Forchhammer et al. role in declining moose populations (Murray et al. 2006; Lenarz 2001; Parmesan 2006; Mason et al. 2014). Studies that attempt to et al. 2009, 2010, DelGiudice et al. 2011; Mech and Fieberg 2014). understand changes in mammal behavior using traditional In addition to the current threats facing moose, the boreal forests of approaches that lack direct observation, such as habitat selection northern Minnesota that moose inhabit are predicted to shift dra- studies, may not fully capture the altered behaviors of individuals matically to drier and more open cover types in the coming decades coping with novel climatic and ecological conditions. For example, (Galatowitsch et al. 2009), thus reducing the ability of moose to animals may select for multiple habitats, but this simple association find habitats typically used for foraging (Street et al. 2015) and ther- may not capture how an animal uses each (e.g., one habitat may mal refuges (McCann et al. 2016). provide thermal refuge and the other foraging opportunities; Street Here, we utilize HMMs to examine how moose living in a pro- et al. 2016). Likewise, these models may not identify whether an tected area along the southern extent of the species’ range alter their animal is only using a habitat at certain times to reduce its risk of behaviors when responding to variability in weather (including am- predation (Latombe et al. 2014). Fortunately, improved methodolo- bient temperature), and the presence of wolves, along with encoun- gies of analyzing animal movement can help us discern what types tering multiple habitats and landscape features. Our HMM of behaviors occur in different habitats, and how these behaviors characterized all moose global positioning system (GPS) locations as may change based on current ecological conditions (Edelhoff et al. one of three behavioral states: traveling, resting, or foraging. We 2016; Gurarie et al. 2016). incorporated activity levels from accelerometers in the GPS collars Hidden Markov models (HMMs) and similar approaches to better differentiate foraging versus resting behaviors that may be (e.g., state–space models) can classify animal behavior based on move- indistinguishable based on movement data alone. Our approach ment characteristics derived from the locations of individual animals enabled us to ask three important questions about moose behavior (Morales et al. 2004; Forester et al. 2007; Patterson et al. 2008; across seasons: 1) can we capture the thermal thresholds at which McClintock et al. 2012a; Beyer et al. 2013). These models assign an moose alter their behavior, 2) does moose behavior change in areas unobservable behavioral state, assumed to be the true behavior of the regularly used by wolves, and 3) how do moose change their behav- individual, to each location of the animal. Aside from movement ior in different primary habitats and relative to landscape features characteristics, HMMs can incorporate ancillary data, such as physi- such as lakes? Many studies have attempted to determine the ology measurements from biologgers and activity counts from acceler- thermal thresholds of moose, but they were often conducted in la- ometers (McClintock et al. 2012b; Fehlmann et al. 2017; Leos-Barajas boratory or captive settings. Here we develop a fuller understanding et al. 2017) to more accurately assign behavioral states or define add- of moose behavior by analyzing behavioral changes at a fine scale itional states (Nams 2014). Behaviors linked with each relocation can as moose encounter different stressors and habitats in the wild. We be associated with outside biotic and abiotic factors to gain inference expand on our findings in the context of long-term climate on how animals respond or activity levels change when encountering forecasts of the region and discuss how the southern extent of different conditions and habitats (e.g., Russell et al. 2015). the geographic distribution of moose is expected to change in the Moose Alces alces are keystone herbivores and are a particularly future. good candidate species to examine how a changing environment may affect the behavior of a mammal at the edge of its bioclimatic threshold. Moose are physiologically sensitive to heat (Renecker and Materials and Methods Hudson 1986, 1989; McCann et al. 2013) and dramatically alter their habitat selection when ambient temperatures rise (Schwab and Study area Pitt 1991; van Beest et al. 2012; Street et al. 2015, 2016) by selecting All GPS-collared moose inhabited the Kabetogama Peninsula of 0 0 for habitats that act as thermal refuges (Dussault et al. 2004; Voyageurs National Park (VNP; 48 30 N, 92 50 W; n ¼ 11; moose- McCann et al. 2016). Energetic costs are tied to higher ambient tem- years ¼ 17) and the adjacent Rat Root Lake area (n ¼ 2; moose- peratures primarily in two ways: 1) direct costs of dissipating heat years ¼ 2). The combined study area is located in north-central through increased metabolic rates, including in extreme cases, pant- Minnesota, USA along the southern edge of the geographic distribu- ing to reduce excess heat (Renecker and Hudson 1986; Renecker tion of moose. The Kabetogama Peninsula (305 km ) is a roadless and Schwartz 1998), and 2) indirect costs of forgoing foraging protected area, but maintains hiking trails in the summer and snow- opportunities while resting to avoid overheating (Street et al. 2016). mobile trails during the winter months. The Rat Root Lake area These costs may lead to a reduction in physical condition and thus (12 km ), which is located near the western edge of VNP, is pri- an increased risk of mortality (Renecker and Hudson 1992; Joly and marily state forest land with a low road density. Area summers, Messier 2004; van Beest and Milner 2013). while relatively short in duration, are humid and reach average daily Throughout the southern extent of their range, moose show signs July temperatures of 18.8 C (10-year average: 20032012). of poorer health (Ruprecht et al. 2016), reduced fecundity Winters are typically dry and cold with an average daily temperature (Monteith et al. 2015; Ruprecht et al. 2016), lower calf survival of 13.7 C (10-year average: 20032012) during January. The (Grøtan et al. 2009; Severud et al. 2015), or range contractions 10-year averages are based on the National Oceanic Atmosphere (Dou et al. 2013). Summer heat stress was found to be one of several Administration’s National Climatic Data Center [cited August, 10th Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 Ditmer et al. Factors influencing moose behavior at bioclimatic edge 3 2016 (https://www.ncdc.noaa.gov/)] Global Climate Station interpolated 6% of all locations for an average of 4.9% per Summary for International Falls, MN, USA. Lakes and ponds moose-year. The waddle package also calculates the movement rate throughout the park are generally ice-covered from late November and relative turning angle for all movement steps. We refer to this as until late April or early May (Kallemeyn et al. 2003). the full movement dataset. The landscape of the VNP ecosystem contains numerous lakes We associated activity level with each movement step by binning and ponds (23.3% open water, based on % areal coverage) with is- accelerometry data based on the beginning and end timestamps lands containing rocky outcrops and shoreline bluffs and a mosaic from each movement step in the full movement dataset and averag- of beaver-influenced wetlands (Johnston and Windels 2015a). ing the corresponding activity values. Vegetation is primarily forest (36.4%), dominated by aspen (Populus spp.), white birch Betula papyrifera, balsam fir Abies bal- Spatial covariates samea, spruce (Picea spp.), pine (Pinus spp.), and red maple (Acer For covariates relying on the locations of moose, we used the coord- rubrum; Faber-Langendoen et al. 2007). Shrublands (1.2%), typic- inates at the end of each step from the full movement dataset. We ally alder (Alnus spp.) and willow (Salix spp.), along with both determined habitat type by overlaying moose locations on a raster woody (30.5%; largely tamarack Larix laricina and black ash of the 2011 National Land Cover Database (Homer et al. 2015) and Fraxinus nigra) and herbaceous wetlands (7.8%) are pervasive extracting the habitat classification of each raster cell using the throughout the VNP ecosystem. These habitat classes represent sig- package raster (Hijmans 2015) in program R (R Core Team 2016). nificant variability in thermal properties (i.e., operative tempera- Lakes and beaver-influenced wetlands within VNP were provided as tures) at various times of day and year (Olson et al. 2014). ESRI Shapefiles of polygons (Johnston and Windels 2015b). We cre- Wolves, the main predator of moose, are abundant inside and ated a raster in ArcMap 10.3.1 (Redwoods, CA), which expanded outside of VNP with 16–22 individuals in two to three packs in- beyond the boundaries of all moose locations, where each habiting the centrally located Kabetogama Peninsula of the park 30 m 30 m cell contained the distance from the center of each cell alone (Olson and Windels 2014). However, wolves regularly make to the nearest edge of a lake or pond. We overlaid all moose loca- use of other prey (Gogan et al. 2004; Chenaux-Ibrahim 2015; Gable tions onto the distance-to-water raster in program R and extracted et al. 2016) such as white-tailed deer (3.8 deer/km ) and beaver the distance value. (5.0 beaver/km ) that are present at much higher densities relative to moose (0.13 moose/km ; Windels and Olson 2016), albeit with Weather covariates much smaller body mass per individual. We used weather conditions from a weather station located at the International Falls airport located 18 km from the eastern edge of Capture and handling of moose VNP (weather station: 48.561, 93.398). The data, collected ap- From 2010 to 2012, we captured adult moose (males: n ¼ 4, moose- proximately once per hour, were available through the National years ¼ 5; females: n ¼ 9, moose-years ¼ 14) during February and Oceanic Atmosphere Administration’s National Centers for March using helicopters to dart individuals (Quicksilver Air, Inc., Environmental Information [March 1, 2016 (https://www.ncdc. Fairbanks, AK, USA). We immobilized and anesthetized moose with noaa.gov/isd/data-access)]. For our weather variables of interest— 1.2 mL (4.0 mg/mL) carfentanil citrate and 1.2 mL (100 mg/mL) ambient temperature ( C), wind speed (km per hour), and liquid pre- xylazine HCl, and used 7.2 mL (50 mg/mL) naltrexone HCl. We cipitation (previous hour in centimeter)—we interpolated the values, used 3 mL (5 mg/mL) yohimbine HCl as antagonist. During each using package zoo (Zeileis and Grothendieck 2005) in program R moose capture, we extracted a blood sample for progesterone ana- (R Core Team 2016), to create a timestamp match with our moose lysis. We classified individuals with progesterone values larger than location data. 2 ng/mL as pregnant in a given year (see Schwartz et al. 1995; Schwartz 1998). We outfitted each moose with a GPS collar Temporal covariates (Sirtrack Limited, Hawkes Bay, New Zealand). Fix attempts were Using the timestamps associated with each moose location in the full scheduled at 15-min intervals during 2010 and 20-min intervals dur- dataset, we calculated the altitude of the sun in the sky at VNP using ing 2011–2012. We estimated the average GPS error of locations the package maptools (Bivand and Lewin-Koh 2017) in program R from stationary Sirtrack collars at 7 m for a 50% circular error (R Core Team 2016). Values less than zero represent times of day probable (McCann et al. 2016). Each GPS collar contained an accel- where the sun was below the horizon, and positive values represent erometer that provided an activity count on a 1-min average basis. times where the sun was above the horizon. All animal capture and handling protocols were approved by the The presence of snow cover, activities associated with calving, University of Minnesota and National Park Service Animal Care and the availability and quality of forage may strongly influence and Use committees. movement. Accordingly, our analyses considered three seasons that We removed any GPS locations associated with a mortality corresponded with declining snow cover and calving (spring, 1 event, any fix with a horizontal dilution of precision (HDOP)>15, April–30 June), the snow free period (summer/fall, 1 July–31 and checked that the final rate of movement was biologically feas- October; hereafter, “summer”), and full snow cover (winter, 1 ible (fastest rate ¼ 52.6 m per min). Additionally, we removed moose November–31 March). Within season variation was considered by that did not have an accelerometer in the collar, and GPS locations assigning a unique day number to each date within a season, starting collected after an accelerometer had become faulty or failed to re- with 1 on the first day of each season then increasing by one every cord activity entirely. day until the last day of the season. HMMs require temporal regularization of the data (i.e., no missed GPS fixes), so we used the package waddle (Gurarie and Bracis 2013) in program R (R Core Team 2016) to linearly interpol- Capture and handling of wolves ate missing locations in the moose data for both the 15- and 20-min We captured adult wolves in VNP using padded-foothold traps interval data independently. For our moose locations, we (Livestock Protection Company, Alpine, TX, USA) from 2012 to Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 4 Current Zoology, 2017, Vol. 00, No. 00 2014 during June–October. We fit each individual with either an The observer classified the behaviors of the focal moose from a Argos GPS (Telonics, Inc., Mesa, AZ, USA) or an Iridium GPS collar distance of 3–10 m. The time and a classification of behavior (walk- (Lotek Wireless, Inc., Newmarket, Ontario, Canada; Vectronic ing, feeding, drinking, standing, standing and ruminating, bedded, Aerospace GmbH, Berlin, Germany). The collars attempted a fix or bedded and ruminating) that occurred for a minimum of 3 s were once every 20 min–6 h, but most were for every 4–6 h, for up to recorded. The moose was observed for a total of 2,315 observation 2 years. We removed any location with a HDOP value >15. minutes over four 7–11-h observation periods (see Ness  for We collected GPS data from 26 individual wolves between 9 additional details). October 2012 and 17 November 2015. We removed any locations The mean, maximum, and standard deviation of activity levels that occurred more frequently than 4 h (3 h, 57 min) of the last from the accelerometer, and the mode of the behavior classification included location to reduce the bias in home range delineation were compared based on 10,000 sets of sub-sampled data. Activity among packs. We pooled data among individuals that we visually data were sub-sampled in both 15- and 20-min time spans to corres- determined to be in the same pack based on proximity in time and pond with the length of GPS fix intervals from wild moose in VNP space of GPS locations. We assumed the GPS locations of the collars using random starting points with replacement. All consecutive data represented the pack’s location for the duration of the study. We following the randomized starting point for the given time span (i.e., split all pack data into seasons (same seasonal delineations as the the next 15 or 20 min of data for both activity counts and direct ob- moose) and only included pack-seasons that had a minimum of 30 servational data) were then included in the analysis. We performed fixes that met all of our criteria. We retained 18,367 total fixes from an ANOVA with a post-hoc Tukey’s Honest Significant Difference 20 distinct packs or individuals inhabiting unique areas of study for (HSD) test to determine significant differences in the accelerometer at least one full season. Seasonal home ranges contained 6,093 loca- activity levels between the three behaviors resulting from the mode tions from 17 pack-seasons in winter, 1,704 locations from 6 pack- calculation of the direct observational data: feeding, ruminating, seasons in spring, and 7,087 locations from 17 pack-seasons during and moving. We considered these three mode behaviors to corres- summer. Areal coverage of combined annual wolf home ranges con- pond to our three behavioral states in our HMM model. We tested tained most of the VNP ecosystem and overlapped the home ranges which metric of activity levels (mean, median, max, or sd) provided of GPS-collared moose. the most significant differences among the three behaviors. We calculated seasonal home ranges for wolf packs with Activity levels from the accelerometer data of the captive moose Program R (R Core Team 2016) using package adehabitat (Calenge were significantly different among the four summary metrics (mean, 2006). We used the function kernelUD to create a bivariate normal median, max, and sd) for the observed mode behavioral state of ei- kernel for each seasonal home range and used the ad hoc method for ther feeding, ruminating, or moving (adjusted P< 0.001 and 95% the smoothing parameter. We created a utilization distribution ras- confidence intervals of the Tukey’s HSD did not overlap zero for all ter for each wolf pack’s seasonal home range where each raster cell metrics for both 15 and 20 minute intervals). Based on the contained the value of the smallest isopleth of the home range cover- ANOVAs, mean activity had the largest differences among the ing the cell. For each moose location in the full dataset, we extracted observed behaviors for both 15-min intervals (mean: F ¼ 4387, 2,9998 the values of all wolf pack utilization distributions calculated at that SD: F ¼ 2096, median: F ¼ 1676, max: F ¼ 552) and 2,9998 2,9998 2,9998 location and season. We binned the extracted values into 0–19, 20– 20-min intervals (mean: F ¼ 6390, SD: F ¼ 2894, median: 2,9998 2,9998 39, 40–59, 60–79, 80–99, and >99 categories of the isopleth values F ¼ 1978, max: F ¼ 765). Mean activity counts for the 2,9998 2,9998 of wolf home ranges (i.e., a value of 100 indicates that the moose lo- 15-min intervals were 13.9, 4.8, 0.67, and 13.1, 4.7, 0.75 for the cation was not overlapped by a wolf home range, a value between 0 20-min intervals corresponding to movement, ruminating, and rest, and 19 is within the very core of a pack’s home range). We assigned respectively. each bin a value of 0–5 with larger values corresponding with We used the mean activity counts from the captive moose behav- smaller utilization distribution values from the wolf pack home iors as informative priors for the true mean activity level in our ranges. The largest bin value for each moose location across all wolf HMM model for the 20-min interval data. For the data correspond- packs was retained for modeling. We refer to this value as “wolf ing to the 15-min location moose, we re-ran the same procedure on home range” going forward. the same captive moose activity data using a minimum activity level of 0.01, and adding a random variance of 0.001 for each observa- tion as was done with the VNP data to better fit the assumption of a normal distribution (there were a large number of low activity Calibration of collar activity counts from the VNP moose). The resulting values were used for the We combined direct human observations of moose behavior with informative priors in the HMM models for moose with 15-min GPS data from SirTrack GPS collars outfitted with accelerometers (the interval data. same make and model as the wild moose in VNP area) to determine what metric of activity count best corresponded to an observed moose’s behavior. From 30 July to 2 August 2009 an observer fol- Statistical analysis lowed a single human-habituated moose inside a 2.6 km pen at Our approach to understanding moose behavior involved two distinct the Kenai Moose Research Center (KMRC) near Soldotna, Alaska. steps. First, we assigned each moose location from the full dataset a The KMRC research facility is owned by the Alaska Department of behavioral state, using a Bayesian three-state switching HMM. This Game and Fish and houses three to six moose per 2.6 km holding HMM utilized step lengths and turning angles derived from sequential pen. Within each pen, the moose forage on naturally occurring vege- GPS locations, and activity counts from accelerometers within the tation and habitat primarily consists of a mix of mature white (Picea same GPS collars to classify each location into one of three behavioral glauca) and black (Picea mariana) spruce along with deciduous trees states. We did not use our HMM to test how weather conditions, spa- such as aspen Populus tremuloides, willow (Salix sp.), and cotton- tial attributes of the landscape and habitat, and wolf home range areas wood Populus trichocarpa. Moose also had access to ponds, small affected moose behaviors because the large increase in number of par- lakes, and sedge meadows. ameters would have made processing times and synthesis prohibitive. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 Ditmer et al. Factors influencing moose behavior at bioclimatic edge 5 Instead, we utilized a frequentist approach, generalized additive mod- We fit the models using a Bayesian MCMC framework with els (GAMs) with a multinomial distribution (categorical behavior clas- JAGs through the RJAGS (Plummer 2016) and R2jags (Yu-Sung and sification was the response), to model the influence of these covariates Yajima 2015) packages. We ran three chains of 5,000 iterations on moose behavior (e.g., Russell et al. 2015). with a burn in of 1,000 and we thinned the chains by eight resulting in 500 samples from each chain. We checked the Gelman–Rubin statistic, making sure it was close to a value of 1, and visually as- sessed the three chains of each parameter to judge convergence. If a HMM: state assignment model did not converge, we then ran additional iterations in inter- We expanded on code provided in the Supplementary Material in vals of 1,000. Typically, the initial 5,000 iterations were enough for Gurarie et al. (2016) for an HMM three-state switching model. Our convergence when proper initial values were provided for the initial aim was to classify each moose location as one of three categorical movement and activity values. We used the mode latent state associ- latent behavioral states: 1) traveling, 2) foraging, or 3) resting. ated with the samples as the behavioral state assigned for each Following the assumptions of Gurarie et al. (2016), we used a movement step. We report the mean and its associated variation for wrapped Cauchy distribution to describe turning angles and a our parameters by randomly sampling, with replacement, 1,000 pos- Weibull distribution to describe step lengths (Kareiva and Shigesada terior point estimates from each moose-year’s HMM posterior dis- 1983). We assumed that traveling would be associated with larger tribution, calculating the mean of these samples (i.e., 1 random travel distances and directional persistence (i.e., turn angles close to sample from each moose-year) and estimating the 95% credible 0, straight-line movements). We assumed resting would be associ- interval of the distribution of means by parameter, behavior, and fix ated with the smallest travel distances and turning angles that were rate (i.e., separate for 15 and 20 min fix data). more evenly distributed compared with those of the travel state, but To ensure the assigned behavioral states from the HMM- with a higher probability of being toward 180 (this is because all lo- provided reasonable classifications, we utilized a suite of diag- cations have some GPS error and stationary collars will yield “steps” nostic plots to visually inspect that behaviors were assigned with with a tendency to have a tight turning angle back toward the actual expectations based on movement distance (i.e., the largest dis- location of the collar). The foraging state should be characterized by tances for travel and the smallest for rest), turning angles (i.e., step-length distances commonly larger than the resting state, but most directional persistence for travel and the least for rest), and smaller than the traveling state, and a generally uniform distribution activity counts (i.e., largest activity counts for travel and the of turning angles as the moose meanders through its habitat search- least for rest). A moose may exhibit multiple behaviors in the 15 ing for and consuming forage or standing still while browsing during or 20 min time of our GPS fix intervals, but the HMM approach the 15- or 20-min period. To increase our ability to differentiate allows us to assign a single behavior associated with each GPS among the three latent behavioral states, we incorporated a third fix interval that is the most likely primary behavior based on the data source: activity data from the GPS accelerometer. We used a distributions for movement rate, turning angle, and activity log-normal distribution for the activity parameter, and assumed that level. We determined the mean proportions of moose locations activity levels would increase from resting to foraging to traveling. assigned each behavior by season and used a bootstrap to esti- We used vague priors for all parameters except those of move- mate the uncertainty among moose-years. The bootstrap esti- ment scale and activity. Specifying slightly informative priors (based mated 95% confidence intervals (using package boot; Canty and on an exploratory analysis of a few individual moose and data from Ripley 2016) based on the adjusted bootstrap percentile method captive animals) helped to keep the parameters aligned for all of our (Davison and Hinkley 1997) using 10,000 bootstrap samples of MCMC chains (without specifying the mean prior values, the order the proportions of behaviors assigned to individual moose-years of the behavior vectors would flip activity and travel). Other priors, by season. Additionally, we calculated the average bout length such as those for movement shape, turn angles, and transition (in minutes; a bout is defined by consecutive moose locations as- probabilities between states were unchanged from those of Gurarie signed the same behavior) per individual by behavior type, and et al. (2016). For the zero-truncated Gaussian priors of the move- generated bootstrap estimates of the predictive distribution. ment scale parameters (one for each behavioral state: travel, forage, and rest), we provided mean values of 100, 40, and 10 for the 20- min data and 75% of these values for the 15-min data (75, 30, and Modeling behavioral states and covariates 7.5); all of these priors had precisions set to 0.001. The informative Moose behavioral states were analyzed using a GAM with a multi- priors for the mean of activity were derived from the captive moose nomial distribution using package mgcv (Wood 2011) in program R data at both the 15- and 20-min intervals with a large variance term (R Core Team 2016). The multinomial distribution allowed us to es- (variance ¼ 1,000, JAGs precision value of 0.001). The prior for the timate how each covariate influenced the likelihood of moose being variance of activity was uniform from 0 to 30 for all three states. in either the resting or traveling states relative to the foraging state We added a maximum step length to be considered in the rest state (i.e., changes in the likelihood of foraging are contingent upon of 50 m. This value is larger than our GPS error estimates of 7m changes in non-foraging behavior [traveling or resting]; e.g., a for the 50% error, but given a time step of 15–20 min, we believed reduced likelihood of traveling with a constant likelihood of resting this would allow for a small amount of movement (e.g., moving to a must result in more foraging activity). This estimated value is called nearby bedding site) along with GPS error. All step lengths and ac- the log odds ratio and it measures the change in the log odds of the tivity measurements less than 0.001 were set to 0.001. Because large probability of being found in one state against the baseline. Because gaps in the GPS data can cause model fitting issues when using the the coefficient values are relative to the baseline state of foraging, a interpolated values to fill in missing locations (e.g., all turn angles negative coefficient value for the rest state and a positive for the are 0 and the step lengths are constant), we removed all data inter- travel state suggest that as the covariate increases, moose are more polated over large gaps. This removal was necessary for three likely to be found in the travel state and less likely to be found in the moose-years of data where the collar had stopped acquiring fixes rest state relative to foraging. Additionally, GAMs enabled us to in- and re-started later in the year. corporate smoothers for covariates we believed to have non-linear Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 6 Current Zoology, 2017, Vol. 00, No. 00 relationships within the behavioral state of the moose. Smoothers (males ¼ 5; females ¼ 14; parameter point estimates for HMM offer a greater flexibility relative to regression splines, allowing us to models in Online Appendix Table 1). Our spring (observa- estimate at what temperature moose reduce travel and increase rest, tions ¼ 120,603, n ¼ 11, moose-years ¼ 17), and summer HMM and estimate a general daily activity budget by including a smoother models (observations ¼ 87,847, n ¼ 8, moose-years ¼ 9) resulted for the altitude of the sun. in similar proportions of assigned behaviors; proportions of each We modeled each moose’s behavior on a seasonal level using behavior had overlapping confidence intervals, and within sea- GAMs. We standardized and centered covariates to compare the ef- sons, travel was assigned to significantly fewer moose locations fect size of each covariate to one another and interpret how strong than forage and rest (x [95% CI]; spring: forage ¼ 0.44 [0.39–0.47], of an effect each covariate has on the likelihood of being in either of rest ¼ 0.38 [0.33–0.42], travel ¼ 0.19 [0.13–0.27]; summer: for- the rest or travel states. We modeled the behavioral state of each age ¼ 0.40 [0.39–0.42], rest ¼ 0.42 [0.32–0.48], travel ¼ 0.18 moose location reported from the HMMs as a function of smoothers [0.12–0.29]). During winter (observations ¼ 139,756, n ¼ 13 indi- for (1) ambient temperature and (2) altitude of the sun, and linear viduals, moose-years ¼ 19), moose were less active (0.45 forage or predictors based on the (3) categorical habitat classification (decidu- travel) relative to spring and summer (0.62 and 0.58 forage or ous forest, mixed forest, conifer forest, shrub/scrub, emergent herb- travel, respectively) due to decreased travel (x ¼ 0.07, 95% aceous wetland, woody wetland, and open water), (4) distance to CI ¼ 0.05–0.11) and increased rest (x ¼ 0.55, 95% CI ¼ 0.51–0.57). lake or ponds, (5, 6) weather conditions (precipitation and wind Across seasons, length of continuous behaviors (minutes per bout) speed), (7) day number of the season, and (8) wolf home range. In a was least for traveling (x ¼ 66.8, 95% CI ¼ 58.6–75.0), intermediate few instances we removed the wolf home range covariate or a given for foraging (x ¼ 75.8, 95% CI ¼ 62.9–88.8), and greatest for rest- habitat class from a seasonal individual moose GAM model if an in- ing (x ¼ 98.8, 95% CI ¼ 79–118.6). dividual did not have enough variation in either overlap in a known wolf pack’s range (e.g., a moose could not be found exclusively Weather conditions and temporal effects within the 40–60% isopleth level for a pack or the model could not Nearly all moose altered their behavior based on the altitude of the run; locations were required within multiple isopleth values), or sun (i.e., amount of daylight; smoother P< 0.05) for both the travel habitat type (e.g., for moose-years with relatively few locations in a (96.9% of moose) and rest states (96.9% of moose) across all sea- season, a less common habitat type such as shrub, may have three sons (32 moose-seasons total). During spring, but especially during locations that were all classified the same) because of the multi- summer, moose were more likely to rest near-midday and the middle nomial distribution of the model. of the night, and were more likely to travel and forage during cre- To summarize the results of the individual moose GAMs, as a puscular times (Figure 1). Moose activity patterns were nearly re- way of providing a “population-level” result and to provide a meas- versed during winter when rest occurred much more frequently in ure of variability among individuals, we report a mean, based on the the dark, and moose foraged and traveled during daylight hours estimates of individual moose-year model coefficients and boot- (Figure 1). Within seasons, moose exhibited changes in behavior re- strapped 95% confidence interval using the same bootstrapping sulting in large effect sizes, by foraging more (less rest and travel) as method previously described. Relative effect sizes can be determined winter turned into spring, traveling and resting more (less foraging) by comparing coefficient values among categorical and continuous as spring progressed, and reducing travel between the start and end variables because they are scaled and centered. These effects high- of summer (see Day # of season; Table 1; summaries of coefficient light trends in the associations between covariates values and behav- direction  and significance in Online Appendix Table 2). iors from the HMM. We considered P-values at or below an alpha level of 0.05 to be significant for linear predictors and smoothers. To summarize the effects of the smoothed covariates (ambient tem- perature and sun altitude) across all individual moose models, we re- ported on the significance of the smoothed terms in the same manner as the linear predictors (i.e., based on the number of individ- ual moose models out of the total). The predicted mean response and 95% confidence interval were then plotted by behavioral state. We calculated the mean and confidence interval with the same boot- strapping method previously described using the predicted values of the response at a given grouped level of the independent variable (e.g., increments of 3 C). Because we hypothesized that wolf home range might have a larger influence on the behavior of pregnant moose during the spring, when moose in VNP give birth, we re-ran our GAMs based on moose-year instead of individual moose. This enabled us to link Figure 1. Predicted values of the relative-risk ratios from smoothers modeling the coefficient values for wolf home range of the resulting moose- the inﬂuence of solar elevation (i.e., amount of daylight; 2 ¼ dark/nightime year GAMs with classifications for: 1) pregnant moose, and 2) non- and 2 ¼ brightest/noon) on the resting behavior of GPS-collared moose in pregnant females and males. We performed an ANOVA with a post- VNP, Minnesota, USA. Each moose’s behavioral state, classiﬁed using a hoc Tukey’s HSD test to determine differences between the groups. HMM, was modeled individually on a seasonal basis using a GAM with a smoother for solar elevation for the behavior of rest relative to foraging. Here, the predicted log odds ratio represents the likelihood of choosing one outcome category (rest) relative to the baseline category (forage). Results Standardized and centered values of solar elevation were pooled by incre- We classified 348,226 observations of moose movement and activity ments of 0.2 to predict the mean and 95% conﬁdence interval for all pooled from 13 individuals (males ¼ 4; females ¼ 9) and 19 moose-years individual moose estimates using bootstrapping. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 Ditmer et al. Factors influencing moose behavior at bioclimatic edge 7 Table 1. Mean (695% bootstrapped conﬁdence intervals) of estimated coefﬁcients from generalized additive model (GAM) results of indi- vidual moose-year data by covariate and season Variable Season n Rest Travel Day number of season Spr 11 0.16 (0.06–0.39)* 0.25 (0.07–0.46)* Sum 7 0.03 (0.15–0.12) 0.24 (0.37–0.08)* Win 13 0.11 (0.31–0.02)* 0.39 (0.91–0.08)* Lake/pond distance Spr 11 0.03 (0.01–0.07) 0.14 (0.18–0.09)* Sum 7 0.09 (0.02–0.22)* 0.03 (0.09–0.16) Win 13 0.01 (0.07–0.03) 0.23 (0.81–0.06)* Precipitation Spr 11 0.02 (0.06–0.02) 0.05 (0.01–0.1)* Sum 7 0.01 (0.07–0.22) 0.06 (0.02–0.28) Win 13 0.03 (0.06–0.01)* 0.03 (0–0.06)* Wind speed Spr 10 0.07 (0.04–0.14)* 0.02 (0.07–0.07) Sum 7 0.03 (0.06–0.01)* 0.08 (0.17–0.01)* Win 13 0.02 (0.00–0.04)* 0.02 (0.15–0.05) Wolf home range Spr 8 0.04 (0.02–0.11) 0.12 (0.22–0.03)* Sum 7 0.03 (0.05–0.01)* 0.03 (0.1–0.03) Win 12 0.01 (0.04–0.02) 0.10 (0.68–0.11) Emergent herb. wetland Spr 11 0.09 (0.16–0.21) 0.35 (0.06–0.61)* Sum 7 0.32 (0.14–0.55)* 0.13 (0.45–0.38) Win 13 0.14 (0.62–0.07) 0.65 (0.19–1)* Open water Spr 9 0.25 (0.94–0.03) 0.74 (0.33–1.38)* Sum 7 0.24 (1.07–0.1) 0.12 (0.19–0.42) Win 7 0.2 (0.63–0.14) 1.61 (0.97–2.28)* Woody wetland Spr 11 0.12 (0.04–0.21)* 0.04 (0.22–0.22) Sum 7 0.25 (0.01–0.63) 0.07 (0.58–0.18) Win 13 0.13 (0.05–0.2)* 0.47 (0.3–0.67)* Shrub/scrub Spr 9 0.10 (0.06–0.3) 0.15 (0.14–0.59) Sum 6 0.18 (0.33–0.03)* 0.15 (0.33–0.11) Win 9 0.22 (0.30–0.08)* 0.36 (0.85–0.00)* Mixed forest Spr 11 0.03 (0.03–0.1) 0.09 (0.25–0.07) Sum 7 0.07 (0.05–0.19) 0.08 (0.1–0.36) Win 12 0.07 (0.02–0.13)* 0.12 (0.05–0.26) Evergreen forest Spr 11 0.11 (0.6–0.16) 0.14 (0.36–0.09) Sum 7 0.06 (0.49–0.49) 0.19 (0.04–0.48) Win 11 0.13 (0.06–0.21)* 0.00 (0.46–0.46) Notes: Behaviors (travel, forage, and rest) of GPS-collared moose in VNP, Minnesota, USA, were classiﬁed using a three-state HMM with switching. These behav- iors were analyzed using GAMs with a multinomial distribution by season. Habitat cover types (last six covariates listed) were categorical and relative to the cover type for deciduous forest. Results for the behavioral states of rest and travel are relative to the foraging state (log odds ratio). Sample size was based on unique moose-year., *95% bootstrapped conﬁdence interval did not overlap zero. Beyond the influence of the sun’s altitude, moose behavior 3). In contrast, moose were more likely to be active (i.e., traveling or changed in association with variation in ambient temperature foraging) during precipitation events especially during spring and (Figure 2). Nearly all moose altered their behavior in association summer (Table 1; Figure 3). with changing ambient temperature throughout the seasons (smoother based on temperature: % of moose for which P< 0.05; winter: travel ¼ 84.6%, rest ¼ 92.3%; spring: travel ¼ 100%, rest- Habitat and landscape effects ¼ 100%; summer: travel ¼ 85.7%, rest ¼ 85.7%). The one individ- Based on the effect sizes of the cover types, moose behavior was ual who did not significantly respond to temperature during summer strongly influenced by habitat cover types and landscape features had considerably less data relative to other moose (1,741 fixes or throughout the seasons, albeit with a large degree of variability 24.2 days of fixes; mean fixes per moose during summer for all among individuals (Figure 4). Moose used wetland habitats for moose¼ 10,980). In response to warm temperatures, moose reduced travel and rest more than for foraging throughout the year (Table 1; their travel in both spring and summer (Figure 2) and correspond- Figure 4). Moose were rarely located in open water (% of locations ingly increased resting behavior. The change from a positive associ- per moose; spring: x ¼ 0.1, range 0.1–2.6; summer: x ¼ 1.7, ation between travel and ambient temperature to a significantly range ¼ 0–5.5; winter: x 0.1, range ¼ 0–0.1). But when they did negative one occurred at a higher ambient temperature during sum- use these areas—which likely consist of open areas along the frozen mer (24–27 C) relative to spring (12–15 C; Figure 2). shoreline in winter and spring—it was consistently for travel (Table Precipitation and wind speed did not have strong effects on 1, Figure 4). Distance to the nearest lake or pond had relatively large moose behavior, but moose consistently responded to changes in effects on moose, but the behaviors associated with these areas both (Table 1; Figure 3). Moose generally increased resting behavior changed throughout the year (Figure 3). In the spring, most moose with increased wind speed, especially during spring, and increased reduced travel and reduced rest (Table 1, Figure 3A,D) in areas fur- foraging when wind speed increased during summer (Table 1; Figure ther from lakes and ponds, which suggests these areas are primarily Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 8 Current Zoology, 2017, Vol. 00, No. 00 used for foraging. The opposite was true for summer as moose sig- [71%; Figure 3B,E; Online Appendix Table 2]). During winter, nificantly increased resting and traveling further from shore (travel moose decreased travel when closer to shore (Table 1; Figure 3F). not significant [Table 1] but most moose had positive coefficients Moose used evergreen/conifer forests primarily for rest during winter, and while there was a large degree of variability among indi- vidual moose-years, 43% of moose-years had a significant positive association between rest and evergreen/conifer forests during sum- mer (Table 1, Figure 4; Online Appendix Table 2). Mixed forests were also primarily associated with resting behavior of moose dur- ing winter and summer and generally with reduced travel during spring (Table 1; Figure 4). However, while the influence of mixed forests was consistent across seasons, the influence relative to de- ciduous forests was not large (Figure 4). Moose used shrub/scrub habitats primarily for foraging during summer and winter (i.e., moose consistently reduced travel and rest in scrub/shrub; Table 1; Figure 4). Wolf home ranges and their effects on moose Wolves used the smallest area during the spring (n ¼ 6; x 50% 2 2 Isopleth ¼ 25.3 km , 95% CI: 11.9–38.7 km ; x 95% Figure 2. Predicted values of the relative-risk ratios of smoothers modeling 2 2 Isopleth ¼ 130.4 km , 95% CI: 87.3–173.5 km ) resulting in the the inﬂuence of ambient temperature ( C) on the travel behavior of GPS col- lared in VNP, Minnesota, USA, during spring and summer. Each moose’s be- lowest percentages of moose locations within known wolf home havioral state (rest, forage, or travel) was classiﬁed using a HMM. The ranges (25.3% of all locations, 72.7% of moose with any spatial behavioral states were modeled individually on a seasonal basis using GAM overlap). A high percentage of moose locations that did overlap with a smoother for ambient temperature. The log odds ratio represents the with wolf home ranges was primarily in the outermost area of wolf likelihood of choosing one outcome category (travel) relative to the probabil- home ranges (80–99% isopleth; 47.5% of locations with overlap) ity of choosing the baseline category (forage). The predicted log odds ratio and no moose was located within the 40% isopleth value of any predictions were pooled by increments of 3 C, and the mean and 95% conﬁ- wolf pack’s home range. A majority of moose (n ¼ 8, moose years; dence interval were calculated for all pooled individual moose estimates using bootstrapping. 63%) located in these outer areas of wolf pack home ranges Figure 3. Coefﬁcient values from GAMs showing the inﬂuence of variables on the behavior of GPS-collared moose in VNP, MN, USA. Moose behaviors, as classi- ﬁed by a HMMs, were modeled by individual moose and season using GAMs with multinomial distributions that provide the log odds ratio estimate. The log odds ratio estimates the likelihood of choosing one outcome category (rest or travel) relative to the probability of choosing the baseline category (forage). Each point represents an estimate from a single moose GAM for each given behavior. The model coefﬁcients for the resting behavioral state response during the A) spring, B) summer, and C) winter, and the traveling behavioral state response during the D) spring, E) summer, and F) winter are shown. Covariate coefﬁcients of log odds ratio estimates were classiﬁed as signiﬁcant if the P-value was 0.05. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 Ditmer et al. Factors influencing moose behavior at bioclimatic edge 9 Figure 4. Coefﬁcient values from GAMs showing the inﬂuence of habitat types on the behavior of GPS-collared moose in VNP, MN, USA. Moose behaviors, as classiﬁed by a HMMs, were modeled by individual moose and season using GAMs with multinomial distributions that provide the log odds ratio estimate. The log odds ratio estimates the likelihood of choosing one outcome category (rest or travel) relative to the probability of choosing the baseline category (forage). Each point represents an estimate from a single moose GAM for each given behavior. The model coefﬁcients for the resting behavioral state response during the A) spring, B) summer, and C) winter, and the traveling behavioral state response during the D) spring, E) summer, and F) winter are shown. Covariate coefﬁcients of log odds ratio estimates were classiﬁed as signiﬁcant if the P-value was 0.05. increased resting behavior closer to wolf pack cores (Online the wolf home ranges (<40% isopleth) during at least some portion Appendix Table 2) and the same percentage of moose reduced trav- of the winter. Most moose, with adequate variation in wolf home eling in these areas substantially (Table 1; Figure 3D). During the range, responded to the closer proximity to wolf home range cores spring, females that were pregnant at the time of capture (n ¼ 5 by altering their amount of travel (58% of individuals with signifi- moose-years) reduced their travel more than non-pregnant females cant coefficients; Online Appendix Table 2). While some moose had and males (n ¼ 7 moose years) in areas closer to wolf pack cores exhibited large effects in travel behavior from wolf home range (pregnant vs. other coefficient for wolf home range¼0.16). The cores (Figure 3C,F) the response was not consistent (50% increased largest reductions in travel were associated with three of the five travel and 50% decreased travel). moose-years for which the female was pregnant that year (b¼0.25, 0.26, 0.35). However, the small sample size did not Discussion result in a significant difference between the two groups (P ¼ 0.08, 95% CI pregnant vs. other¼0.35–0.02). Females Our approach, utilizing HMMs to classify the behavior of moose, adj. that tested pregnant were only slightly less likely to rest when provides additional insights about moose living along the southern located closer to wolf home range cores (difference pregnant vs. oth- extent of their bioclimatic range beyond that of traditional habitat er¼0.07, P ¼ 0.26, 95% CI pregnant vs. other¼0.19–0.06) selection studies. By incorporating movement characteristics and ac- adj. suggesting an increase in foraging or another behavior with similar tivity counts from accelerometers (validated by direct observations behavioral characteristics. on captive moose), we were able to distinguish between three behav- During summer, wolves expanded their home ranges (n¼ 17; x iors of moose remotely, at fine temporal and spatial scales, and asso- 2 2 50% Isopleth¼ 42.1 km , 95% CI: 28.2–56.0 km ; x 95% ciate the changes in behavior with different environmental variables. 2 2 Isopleth¼ 170.6 km ,95% CI:112.7–228.5 km ) resulting in all moose While individual moose exhibited variable behavior associated with locations overlapping at least one wolf pack’s home range, and nearly changes in weather, landscape, and the presence of predators, we half of locations falling in the inner <40% isopleth core (48.2% of lo- were able to identify several consistent behavioral responses, espe- cations). Moose in close proximity to the core of wolf home ranges cially regarding time of day and ambient temperature. Our findings reduced rest (Table 1), but the effect was small (Figure 3B). highlight the difficult situation facing thermally sensitive species Wolf home ranges were largest on average during winter months coping with climate change; rising temperatures will require individ- 2 2 (n ¼ 17; x 50% Isopleth ¼ 83.6 km , 95% CI: 83.6–145.6 km ; x uals to reduce heat production and heat exposure by resting more 2 2 95% Isopleth ¼ 345.1 km , 95% CI: 71.1–619.1 km ). Wolf packs frequently or shifting activity to nighttime when temperatures abate, overlapped all moose locations again, and the majority of moose thus reducing opportunities to seek out new food sources and for- (76.9% of individuals) were located in areas closest to the core of age, especially when predators are present. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 10 Current Zoology, 2017, Vol. 00, No. 00 Assessing thermal thresholds for moose can be difficult outside frequently and thereby forego travel and foraging to avoid heat of laboratory or captive settings, but our approach, which ac- stress. Galatowitsch et al. (2009) forecasts that the VNP area will counted for the influence of solar elevation, time of year, and habi- experience an increase in the average daily minimum and maximum tat, characterized moose behavior at a fine scale and captured the summer (June–August) temperatures of around 1.6 C by 2030 and ambient temperatures when moose became less active (i.e., travel) 3.3 C by the year 2060. Studies by Dussault et al. (2004) and Street and more likely to rest. The thermal thresholds of moose found by et al. (2015) found that moose coped with warm summer daytime Renecker and Hudson (1986), who studied captive moose living in temperatures by reducing daytime activity and switching to noctur- enclosed pens, were accepted as the best estimates of heat stress for nal activity. Our findings show that moose in VNP follow a similar many years. They found that moose increased respiration rates at seasonal pattern of resting during the heat of the mid-day in summer 14 C and started open-mouth panting at 20 C during the summer. months, and decreasing rest at night. Moose responded in a similar, Recently, McCann et al. (2013) found that heat stress thresholds but weaker pattern during the spring, and reversed this pattern dur- vary for moose housed in an outdoor enclosure and that shade and ing the winter by becoming active during daylight hours. This be- wind speed influenced the thresholds. With no wind, moose havioral plasticity may help moose cope with rising temperatures increased respiration rates for evaporative cooling at 17 C, while for years to come; however, it is clear that moose will be faced with with wind respiration rates did not increase until 24 C. Similarly, a difficult tradeoff between energy acquisition and resting to deal free-ranging moose in VNP began resting more frequently in sum- with higher ambient temperatures. Street et al. (2016) examined this mer when temperature reached 21 C and reduced travel when they inherent tradeoff and found that a more northerly population in reached 24 C. While we were able to capture strong behavioral Ontario, Canada strongly selected for areas of better forage while changes at these ambient temperatures, without the ability to dir- the more southerly population in Minnesota, USA selected for a bal- ectly observe the moose or take physiological measurements, we can ance between habitats associated with better thermal cooling prop- only state that we estimated behavioral-related changes in activity erties and those with better forage. Average wind speeds are also and not necessarily heat stress. A 24 C threshold was also corrobo- forecasted to be reduced under some climate change scenarios (e.g., rated by Broders et al. (2012), who found that moose in Nova Pryor et al. 2009), thus also impacting another primary source of Scotia sought thermal shelter when daytime temperatures reached cooling for moose (McCann et al. 2013). 24 C. However, it is also important to recognize that the thermal Climate change in the region is already shifting vegetative com- thresholds for moose are likely to vary from region to region, and in- munities in boreal ecosystems (Soja et al. 2007) and the pace of this dividuals with poorer health, as a result of disease or parasite loads, change is expected to accelerate (Galatowitsch et al. 2009). This may exhibit a lower threshold of thermal tolerance, as was indicated shift may eliminate the ability of moose to cope with heat by select- for populations near VNP (Murray et al. 2006; Lenarz et al. 2009) ing for habitats with the best thermal properties. Numerous studies and for captive moose in Minnesota (McCann et al. 2013). have found that moose select for habitats that provide better shade Spring thermal behavioral thresholds affecting moose were much and cooling when ambient temperatures increase (Renecker and lower than summer thresholds. Despite the rapid increases in spring Hudson 1989; Schwab and Pitt 1991; Demarchi and Bunnell 1995; temperatures observed globally, especially in the Midwestern United van Beest et al. 2012; Street et al. 2015, 2016; McCann et al. 2016). States (Schwartz et al. 2006), they often do not receive as much at- These studies generally agree that lowland forests with dense cano- tention as the more extreme values of ambient temperature fore- pies are the most preferred by moose during times of high ambient casted for summer months. We expected moose to exhibit a lower temperatures. spring behavioral thresholds because seasonally shifting thermal tol- Although some studies have found little difference in the abilities of erance is well established across many taxa of animals (Po ¨ rtner different habitats to provide thermal shelter (Lowe et al. 2010), Olson 2002). Our findings, which suggest a spring thermal behavioral et al. (2014) found large differences in thermal properties among habi- threshold around 15 C, may be a result of natural acclimation to tats in the VNP ecosystem that drive the operative temperatures experi- cold winter temperatures and the rapid transitions in weather and enced by moose in the various habitat types. Open habitat types, such phenology that occur during spring, exacerbated by remnant winter as scrub/shrub and wetlands had an average difference in afternoon coats. It could also reflect a lack of canopy and horizontal cover temperatures during the summer of 3.38 C (and a maximum of within deciduous vegetation at this time. Research by Lenarz et al. 8.10 C) relative to forested habitats. These findings support the pat- (2009) highlighted the importance of spring temperatures by associ- terns found in our results that moose forage in mixed forests and ating warmer spring ambient temperatures with lower adult survival shrublands, and travel through wetlands—both active behaviors—but in Minnesota (although see Mech and Fieberg  which ques- likely only when temperatures fall below the maximum thermal thresh- tions this association). However, earlier and warmer spring tem- olds of a given season. Moose select for better thermal cover in conifer peratures can also benefit moose by making spring forage more and deciduous forests with dense canopies when the thresholds are ex- abundant and available earlier in the year. Grøtan et al. (2009) ceeded—a finding that was supported by Street et al. (2016) using step- linked these potential benefits of warmer spring time temperatures selection functions to analyze fine-scale habitat selection elsewhere in experienced by moose in Norway to subsequent increases in calf sur- Minnesota, and an analysis of moose bed site selection by McCann et vival, while Monteith et al. (2015) found a negative effect on recruit- al (2016). We did not include an interaction between ambient tempera- ment associated with warmer spring temperatures due to a ture and habitat type in our models due to the large increase in the size mismatch in phenology of early season forage and nutrition. While and complexity of the results of our models, especially while incorpo- the overall impacts of a warmer and earlier spring season are still rating the effects of smoothers. debated, other environmental conditions, such as more frequent pre- The overall lack of a strong, consistent response of moose to cipitation events, may help counteract times of heat stress more so wolves in VNP is not surprising for both methodological and biolo- during spring than summer. gical reasons. However, while the overall effect size of wolves on the Temperatures are forecasted to increase further in the coming behavior of all moose was relatively small, we did find some evi- decades, which means that moose will likely need to rest more dence that pregnant females close to the expected calving season Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 Ditmer et al. Factors influencing moose behavior at bioclimatic edge 11 (i.e., before and after presumed parturition) were less likely to be Methods that enable researchers to more fully utilize the data traveling in areas more frequented by wolves relative to adult fe- from GPS units can offer new insights beyond habitat selection by males without calves and males. This response is likely a way to re- discerning how an animal responds behaviorally to their environ- duce the exposure of calves to predation risk, but our results also ment (Edelhoff et al. 2016; Gurarie et al. 2016). Activity data from suggest that pregnant females increased foraging behavior in these collar accelerometers are often unused or are analyzed in relative same areas. Increased foraging in areas with a higher likelihood of isolation from animal movement. Our HMM enabled us to combine attack seems counterintuitive, but our HMM classification scheme activity and movement data to more accurately separate moose be- was limited to three basic behaviors. By relying solely on movement havior among resting and foraging. Anecdotally, the HMM model and activity data, we potentially misclassified some vigilance behav- struggled to assign behavioral states to large numbers of moose loca- ior as foraging. These behaviors may appear relatively similar in tions in a way that made biological sense when utilizing only step terms of movement and activity characteristics. Our inability to lengths and turn angles. Additionally, by utilizing a captive study of identify vigilance behavior may be exacerbated because wolf GPS moose wearing identical GPS collars (including accelerometers) to data only temporally overlapped with the collection of moose GPS most of the moose in our study, we were able to provide biologically locations for 1 year. Even with better temporal and spatial overlap, appropriate priors that helped the model converge more quickly and it can be difficult to identify responses of prey species to predators give more sensible results. However, we caution that there are sev- when relying on GPS data alone (e.g., Eriksen et al. 2011). eral factors to consider when utilizing a three-state HMM model Biologically, studies in other parts of Minnesota have found the re- with activity data: 1) several accelerometers failed or provided unre- cently recovered wolf population to play a major role in moose liable and infrequent measurements later in the lifespan of the collar population abundance (Mech and Fieberg 2014). However, studies and thus reduced our sample size, 2) Moen et al. (1996) found the in VNP, where wolf and moose population levels have been fairly association between directly observed behaviors and activity data constant over the last decade (Olson and Windels 2014), have found were no longer reliable at GPS fix rates of >1 h due to multiple be- that moose comprise a relatively small percentage of wolf diet haviors occurring within that time frame (our fix attempts were 15 or (2–3%; Gogan et al. 2004; Chenaux-Ibrahim 2015; Gable 2016). 20 min), and 3) the relationship we observed in wild moose between Wolves in VNP can avoid the task of predating moose because less- movement rate and activity had a large degree of variability among risky prey alternatives such as white-tailed deer and beaver are pre- moose and even between moose-years of the same individual. Our be- sent at relatively higher densities than moose (Windels and Olson havioral classification might have been improved if were able to dir- 2016) and consistently make up large portions of wolf diets within ectly observe moose behavior in different habitats, and create unique VNP (Gogan et al. 2004; Chenaux-Ibrahim 2015; Gable 2016). priors for the mean of activity dependent on the habitat associated with While the direct influences of wolves may have been muted due the location of the moose. For instance, movements through thicker to methodological hurdles, moose patterns of forage and rest in rela- vegetation associated with certain habitats may result in higher activity tion to lakes and ponds may be an indirect function of predator counts than the same movement rate through a generally open habitat. avoidance and thermal cooling. Lakes in many regions of moose To further understand how moose along the southern edge of range are typically frozen well into spring, and thus do not provide their range will respond to climate change, additional research may forage opportunities until summer. Moose may utilize the cooling be needed to understand the effects of climate change on the quality properties of aquatic habitats during warmer months, and use the of browse available to moose. Moose make habitat selection deci- often forage-rich shorelines, wetlands, or shallow lakes and ponds in sions by balancing the tradeoff between thermal cooling properties the region, while quickly dissipating heat (Schwab and Pitt 1991). and forage (Street et al. 2016), but finding this equilibrium may be- However, these same areas along the shore may be the riskiest areas come extremely difficult if climate change also reduces nutrition of of the landscape during winter and early spring when wolves often available forage. Findings by Dearing (2012) and Kurnath et al. select for frozen surfaces and shorelines (Kuzyk et al. 2004) to travel (2016) show that plant secondary compounds in herbivore forage and hunt (Mech 1991). In a nearby study on Isle Royale, Michigan, may have temperature-dependent toxicity properties; as tempera- USA, Montgomery et al. (2014) found a strong negative association tures rise, these toxic properties may reduce the nutritional content between the distance to shore and successful predation events of of forage currently assumed to be important for moose diets. As an wolves on moose. Wolves in VNP forage near water bodies and wet- additional potential avenue of research, biologger technologies that land habitats to prey heavily on beavers in summer (Gable et al. allow for the remote measurement of physiological changes (e.g., ab- 2016). Our findings of VNP moose potentially support the idea of a dominal and vaginal implant transmitters and cardiac biologgers) tradeoff between forage intake, thermal cooling, and predator offer the next step in understanding how moose respond to stressors. avoidance whereby VNP moose may risk being around shorelines When biologger technologies are combined with traditional GPS where wolves frequent in order to access forage-rich habitat patches, units and accelerometers, HMMs could identify other unique behav- such as wetlands created by beavers (Johnston and Windels 2015a), ioral and physiological states that may not have been apparent with that also allows them to stay cool during the summer months, but GPS and activity data alone (e.g., Ditmer et al. 2015). Additional utilize areas further from lakes in the winter and spring when water states such as “resting and heat stressed” or “traveling and stressed” bodies are frozen. Morris (2014) reviewed studies that reported on potentially from a predator would greatly enhance our understand- how moose used aquatic habitats and found little support for the hy- ing of both animal behavior and physiology. pothesis of minimization of direct predation risk and heat stress Understanding how populations inhabiting areas along the geo- amelioration. Instead, the analysis found greater support for maxi- graphic edges of their distribution mechanistically respond to mizing nutrition and avoidance of biting insects. More research, changes in their environment can help predict future range shifts, possibly incorporating novel technologies and methodologies, may and inform conservation or management needs. Analytical help researchers better determine how and why moose use aquatic approaches that provide insights into how and why species alter habitats, beyond travel, forage, or rest, so that we may better under- their behavior in response to environmental change at fine spatial stand how or if moose can cope with future drier environments. and temporal scales are a critical first step to understanding the Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 12 Current Zoology, 2017, Vol. 00, No. 00 Dearing MD, 2012. Temperature-dependent toxicity in mammals with impli- potential for a population to persist. Our findings suggest that in the cations for herbivores: a review. J Comp Physiol B 183:43–50. coming 50100 years, the adaptability of moose inhabiting the DelGiudice GD, 2016. 2016 Aerial Moose Survey. Technical Report, VNP ecosystem will be tested in the face of a shifting bioclimatic Minnesota Department of Natural Resources, St Paul, Minnesota, USA. range (Galatowitsch et al. 2009), whereby moose will either need to DelGiudice GD, Sampson BA, Lenarz MS, Schrage MW, Edwards AJ, 2011. find more or higher quality forage in habitats used for thermal ref- Winter body condition of moose Alces alces in a declining population in uge, or risk the physiological impacts of heat stress. Our study high- northeastern Minnesota. J Wildl Dis 47:30–40. lights this predicament for moose in VNP and other southerly moose Demarchi MW, Bunnell FL, 1995. Forest cover selection and activity of cow populations by estimating the seasonal maximum behavioral ther- moose in summer. Acta Theriol 40:23–36. mal thresholds and determining how moose altered their behavior Dirzo R, Young HS, Galetti M, Ceballos G, Isaac NJB et al., 2014. based on habitat, landscape features, and risk of predation. Defaunation in the Anthropocene. Science 345:401–406. Ditmer MA, Vincent JB, Werden LK, Tanner JC, Laske TG et al., 2015. Bears However, moose are an adaptable species and have shown persist- show a physiological but limited behavioral response to unmanned aerial ve- ence in other southern extents of their range with even less vegeta- hicles. Curr Biol 25:2278–2283. tive cover (e.g., southern Manitoba, Canada and North Dakota, Dou H, Jiang G, Stott P, Piao R, 2013. Climate change impacts population dy- USA). While our study provides a closer look at the behavioral namics and distribution shift of moose Alces alces in Heilongjiang Province response to many aspects of their changing environment, future re- of China. Ecol Res 28:625–632. search should both expand upon new technologies and methodolo- Dussault C, Ouellet J-P, Courtois R, Huot J, Breton L et al., 2004. Behavioural gies to better discern animal behavioral responses while more responses of moose to thermal conditions in the boreal forest. Ecoscience holistically incorporating other critical factors that may influence 11:321–328. population persistence such as disease, parasite loads, and inter- Edelhoff H, Signer J, Balkenhol N, 2016. Path segmentation for beginners: an actions with new competitors and predators. overview of current methods for detecting changes in animal movement pat- terns. Mov Ecol 4:21. Eriksen A, Wabakken P, Zimmermann B, Andreassen HP, Arnemo JM et al., 2011. Activity patterns of predator and prey: a simultaneous study of Supplementary Material GPS-collared wolves and moose. Anim Behav 81:423–431. Supplementary material can be found at https://academic.oup.com/cz. Faber-Langendoen D, Aaseng N, Hop K, Lew-Smith M, Drake J, 2007. Vegetation classiﬁcation, mapping, and monitoring at Voyageurs National Park, Minnesota: an application of the U.S. National Vegetation Acknowledgments Classiﬁcation. Appl Veg Sci 10:361–374. Fehlmann G, O’riain MJ, Hopkins PW, O’sullivan J, Holton MD et al., 2017. We thank Bryce T. Olson and other VNP staff for assistance with data collec- Identiﬁcation of behaviours from accelerometer data in a wild social pri- tion and management, and Dr Brian Kot and Dr Seth Stapleton for their con- mate. Anim Biotelem 5:6. tributions to this work. Three anonymous reviewers provided insightful Forchhammer MC, Clutton-Brock TH, Lindstrom J, Albon SD, 2001. Climate feedback to reﬁne the initial draft of this manuscript. and population density induce long-term cohort variation in a northern un- gulate. J Anim Ecol 70:721–729. Forester JD, Ives AR, Turner MG, Anderson DP, Fortin D et al., 2007. Funding State–space models link elk movement patterns to landscape characteristics This work was supported by the US Geological Survey-National Park Service, in Yellowstone National Park. Ecol Monogr 77:285–299. Natural Resource Preservation Program, Voyageurs National Park, Franco AMA, Hill JK, Kitschke C, Collingham YC, Roy DB et al., 2006. Minnesota’s Environment and Natural Resources Trust Fund, and Impacts of climate warming and habitat loss on extinctions at species’ Minnesota’s Clean Water, Land, and Legacy Amendment. low-latitude range boundaries. Glob Change Biol 12:1545–1553. Gable TD, 2016. Wolf predation: where and how wolves hunt beavers, and confronting the biases in scat-based diet studies [Master’s thesis]. Northern References Michigan University. Gable TD, Windels SK, Bruggink JG, Homkes AT, 2016. Where and how Beyer HL, Morales JM, Murray D, Fortin M-J, 2013. The effectiveness of wolves Canis lupus kill beavers Castor canadensis. PLoS One 11:e0165537. Bayesian state-space models for estimating behavioural states from move- doi: 10.1371/journal.pone.0165537. ment paths. Methods Ecol Evol 4:433–441. Galatowitsch S, Frelich L, Phillips-Mao L, 2009. Regional climate change Bivand R, Lewin-Koh N, 2017. maptools: tools for reading and handling spa- adaptation strategies for biodiversity conservation in a midcontinental re- tial objects. R package version 0.8-41. Available from: https://cran.r-proj gion of North America. Biol Conserv 142:2012–2022. ect.org/web/packages/maptools/maptools.pdf. Gilman SE, Urban MC, Tewksbury J, Gilchrist GW, Holt RD, 2010. A frame- Broders HG, Coombs AB, McCarron JR, 2012. Ecothermic responses of work for community interactions under climate change. Trends Ecol Evol moose Alces alces to thermoregulatory stress on mainland Nova Scotia. 25:325–331. Alces 48:53–61. Gogan PJP, Route WT, Olexa EM, Thomas N, Kuehn D et al., 2004. Gray Calenge C, 2006. The package adehabitat for the R software: a tool for the wolves in and adjacent to Voyageurs National Park. Minnesota: Research analysis of space and habitat use by animals. Ecol Model 197:516–519. and Synthesis 1987–1991. Canty A, Ripley B, 2016. boot: bootstrap R (S-Plus) functions. R package version Grøtan V, Sæther B-E, Lillega ˚ rd M, Solberg EJ, Engen S, 2009. Geographical 1.3-18. Available from: https://cran.r-project.org/web/packages/boot/boot.pdf. variation in the inﬂuence of density dependence and climate on the recruit- Chen I-C, Hill JK, Ohlemu ¨ ller R, Roy DB, Thomas CD, 2011. Rapid range ment of Norwegian moose. Oecologia 161:685–695. shifts of species associated with high levels of climate warming. Science 333: Gurarie E, Bracis C, 2013. waddle: What is the animal doing? R package ver- 1024–1026. sion 0.11. Available from: http://onlinelibrary.wiley.com/doi/10.1111/ Chenaux-Ibrahim Y, 2015. Seasonal diet composition of gray wolves Canis 1365-2656.12379/suppinfo. lupus in no rtheastern Minnesota determined by scat analysis[Master’s the- sis]. Duluth: University of Minnesota. Gurarie E, Bracis C, Delgado M, Meckley TD, Kojola I et al., 2016. What is Davison AC, Hinkley DV, 1997. Bootstrap Methods and Their Application. the animal doing? Tools for exploring behavioural structure in animal Cambridge: Cambridge University Press. movements. J Anim Ecol 85:69–84. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 Ditmer et al. Factors influencing moose behavior at bioclimatic edge 13 Hampe A, Petit RJ, 2005. Conserving biodiversity under climate change: the Moen R, Pastor J, Cohen Y, 1996. Interpreting behavior from activity counters rear edge matters. Ecol Lett 8:461–467. in GPS collars on moose. Alces 32:101–108. Monteith KL, Klaver RW, Hersey KR, Holland AA, Thomas TP et al., 2015. Hickling R, Roy DB, Hill JK, Fox R, Thomas CD, 2006. The distributions of a Effects of climate and plant phenology on recruitment of moose at the south- wide range of taxonomic groups are expanding polewards. Glob Change ern extent of their range. Oecologia 178:1137–1148. Biol 12:450–455. Montgomery RA, Vucetich JA, Roloff GJ, Bump JK, Peterson RO, 2014. Hijmans RJ, 2015. raster: geographic data analysis and modeling. R package Where wolves kill moose: the inﬂuence of prey life history dynamics on the version 2.5-2. landscape ecology of predation. PLoS One 9:e91414. Homer CG, Dewitz JA, Yang L, Jin S, Danielson P et al., 2015. Completion of Morales JM, Haydon DT, Frair J, Holsinger KE, Fryxell JM, 2004. Extracting the 2011 National Land Cover Database for the conterminous United more out of relocation data: building movement models as mixtures of ran- States—representing a decade of land cover change information. dom walks. Ecology 85:2436–2445. Photogramm Eng Remote Sensing 81:345–354. Morris DM, 2014. Aquatic habitat use by North American moose Alces alces Huey RB, Deutsch CA, Tewksbury JJ, Vitt LJ, Hertz PE et al., 2009. Why trop- and associated richness and biomass of submersed and ﬂoating-leaved ical forest lizards are vulnerable to climate warming. Proc R Soc Lond B aquatic vegetation in north-central Minnesota [Master’s thesis]. Lakehead Biol Sci 276:1939–1948. University. Johnston CL, Windels SK, 2015a. Using beaver works to estimate colony ac- Murray DL, Cox EW, Ballard WB, Whitlaw HA, Lenarz MS et al., 2006. tivity in boreal landscapes. J Wildl Manage 79:1072–1080. Pathogens, nutritional deﬁciency, and climate inﬂuences on a declining Johnston CL, Windels SK, 2015b. Using beaver works to estimate colony ac- moose population. Wildl Monogr 166:1–30. tivity in boreal landscapes (online spatial data). CUAHSI Spatial Data Nams VO, 2014. Combining animal movements and behavioural data to de- Service, Consortium of Universities for the Advancement of Hydrologic tect behavioural states. Ecol Lett 17:1228–1237. Science Inc., Medford, Massachusetts, USA. Ness TE, 2010. Using activity counters to estimate behavior of free-ranging Joly DO, Messier F, 2004. The distribution of Echinococcus granulosus in moose [Master’s thesis]. Twin Cities: University of Minnesota. moose: evidence for parasite-induced vulnerability to predation by wolves? Olson BT, Windels SK, 2014. Status of gray wolves in and adjacent to Oecologia 140:586–590. Voyageurs National Park: 2012–2013 Summary Report. Voyageurs Kallemeyn LW, Holmberg KL, Perry JA, Odde BY, 2003. Aquatic Synthesis National Park technical document, 1–17. for Voyageurs National Park. USGS Information and Technology Report Olson B, Windels SK, Fulton M, Moen R, 2014. Fine-scale temperature pat- USGS/BRD/ITR-2003-0001. terns in the southern boreal forest: implications for the cold-adapted moose. Kareiva PM, Shigesada N, 1983. Analyzing insect movement as a correlated Alces 50:105–120. random walk. Oecologia 56:234–238. Parmesan C, 2006. Ecological and evolutionary responses to recent climate Kurnath P, Merz ND, Dearing MD, 2016. Ambient temperature inﬂuences change. Annu Rev Ecol Evol Syst 37:637–669. tolerance to plant secondary compounds in a mammalian herbivore. Proc Parmesan C, Yohe G, 2003. A globally coherent ﬁngerprint of climate change RSoc B 283:20152387. impacts across natural systems. Nature 421:37–42. Kuzyk GW, Kneteman J, Schmiegelow FKA, 2004. Winter habitat use by Patterson TA, Thomas L, Wilcox C, Ovaskainen O, Matthiopoulos J, 2008. wolves Canis lupus in relation to forest harvesting in west-central Alberta. State–space models of individual animal movement. Trends Ecol Evol 23: Can Field-Nat 118:368–375. 87–94. Latombe G, Fortin D, Parrott L, 2014. Spatio-temporal dynamics in the re- Plummer M, 2016. rjags: Bayesian graphical models using MCMC. R package sponse of woodland caribou and moose to the passage of grey wolf. J Anim version 4-5. Available from: https://cran.r-project.org/web/packages/rjags/ Ecol 83:185–198. rjags.pdf. Lenarz MS, Nelson ME, Schrage MW, Edwards AJ, 2009. Temperature medi- Po ¨ rtner H, 2001. Climate change and temperature-dependent biogeography: ated moose survival in northeastern Minnesota. J Wildl Manage 73: oxygen limitation of thermal tolerance in animals. Naturwissenschaften 88: 503–510. 137–146. Lenarz MS, Fieberg J, Schrage MW, Edwards AJ, 2010. Living on the edge: Po ¨ rtner H, 2002. Climate variations and the physiological basis of temperature viability of moose in northeastern Minnesota. J Wildl Manage 74: dependent biogeography: systemic to molecular hierarchy of thermal tolerance 1013–1023. in animals. Comp Biochem Physiol A Mol Integr Physiol 132:739–761. Leos-Barajas V, Photopoulou T, Langrock R, Patterson TA, Watanabe YY Post E, Peterson RO, Stenseth NC, McLaren BE, 1999. Ecosystem conse- et al., 2017. Analysis of animal accelerometer data using hidden Markov quences of wolf behavioural response to climate. Nature 401:905–907. models. Methods Ecol Evol 8:161–173. Pryor SC, Barthelmie RJ, Young DT, Takle ES, Arritt RW et al., 2009. Wind speed Lowe SJ, Patterson BR, Schaefer JA, 2010. Lack of behavioral responses of trends over the contiguous United States. J Geophys Res Atmos 114:D14105. moose Alces alces to high ambient temperatures near the southern periphery R Core Team, 2016. R: A Language and Environment for Statistical of their range. Can J Zool 88:1032–1041. Computing. Vienna, Austria: R Foundation for Statistical Computing. Mason TH, Apollonio M, Chirichella R, Willis SG, Stephens PA, 2014. Renecker LA, Hudson RJ, 1986. Seasonal energy expenditures and thermo- Environmental change and long-term body mass declines in an alpine mam- regulatory responses of moose. Can J Zool 64:322–327. mal. Front Zool 11:69. Renecker LA, Hudson RJ, 1989. Ecological metabolism of moose in McCann NP, Moen RA, Harris TR, 2013. Warm-season heat stress in moose aspen-dominated boreal forests, central Alberta. Can J Zool 67:1923–1928. Alces alces. Can J Zool 91:893–898. Renecker LA, Hudson RJ, 1992. Thermoregulatory and behavioral response McCann NP, Moen RA, Windels SK, Harris TR, 2016. Identifying thermal of moose: is large body size an adaptation or constraint. Alces 1:52–64. refugia for a cold-adapted mammal facing climate change. Wildl Biol 22: Renecker LA, Schwartz CC, 1998. Food habits and feeding behavior. In: 228–237. Franzmann AW, Schwartz CS, editors. Ecology and Management of the North McClintock BT, King R, Thomas L, Matthiopoulos J, McConnell BJ et al., American Moose. Washington (DC): Smithsonian Institution Press. 403–440. 2012a. A general discrete-time modeling framework for animal movement Ripple WJ, Chapron G, Lo ´ pez-Bao JV, Durant SM, Macdonald DW et al., using multistate random walks. Ecol Monogr 82:335–349. 2016. Saving the world’s terrestrial megafauna. BioScience biw092. McClintock BT, Russell DJF, Matthiopoulos J, King R, 2012b. Combining in- Ruprecht JS, Hersey KR, Hafen K, Monteith KL, DeCesare NJ et al., 2016. dividual animal movement and ancillary biotelemetry data to investigate Reproduction in moose at their southern range limit. J Mammal 97:1355–1365. population-level activity budgets. Ecology 94:838–849. Russell DJF, McClintock BT, Matthiopoulos J, Thompson PM, Thompson D Mech LD, 1991. The Way of the Wolf. Stillwater, MN: Voyageur Press. et al., 2015. Intrinsic and extrinsic drivers of activity budgets in sympatric Mech LD, Fieberg J, 2014. Re-evaluating the northeastern Minnesota moose grey and harbour seals. Oikos 124:1462–1472. decline and the role of wolves. J Wildl Manage 78:1143–1150. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018 14 Current Zoology, 2017, Vol. 00, No. 00 Schwab FE, Pitt MD, 1991. Moose selection of canopy cover types related Street GM, Fieberg J, Rodgers AR, Carstensen M, Moen R et al., 2016. to operative temperature, forage, and snow depth. Can J Zool 69: Habitat functional response mitigates reduced foraging opportunity: impli- 3071–3077. cations for animal ﬁtness and space use. Landsc Ecol 31:1939–1953. Schwartz CC, 1998. Reproduction, natality and growth. In: Franzmann AW, van Beest FM, Van Moorter B, Milner JM, 2012. Temperature-mediated habi- Schwartz CS, editors. Ecology and Management of the North American tat use and selection by a heat-sensitive northern ungulate. Anim Behav 84: Moose. Washington (DC): Smithsonian Institution Press. 141–171. 723–735. Schwartz CC, Monfort SL, Dennis PH, Hundertmark KJ, 1995. Fecal proges- van Beest FM, Milner JM, 2013. Behavioural responses to thermal conditions tagen concentration as an indicator of the estrous cycle and pregnancy in affect seasonal mass change in a heat-sensitive northern ungulate. PLoS moose. J Wildl Manage 59:580–583. One 8:e65972. Schwartz MD, Ahas R, Aasa A, 2006. Onset of spring starting earlier across Windels SK, Olson BT, 2016. Moose Population Survey at Voyageurs National the Northern Hemisphere. Glob Change Biol 12:343–351. Park: 2016. Natural Resource Data Series NPS/VOYA/NRDS–2016/1031, Severud WJ, Giudice GD, Obermoller TR, Enright TA, Wright RG et al., National Park Service, Fort Collins, Colorado. 2015. Using GPS collars to determine parturition and cause-speciﬁc mortal- Wood SN, 2011. Fast stable restricted maximum likelihood and marginal like- ity of moose calves. Wildl Soc Bull 39:616–625. lihood estimation of semiparametric generalized linear models. J R Stat Soc Soja AJ, Tchebakova NM, French NHF, Flannigan MD, Shugart HH et al., B 73:3–36. 2007. Climate-induced boreal forest change: predictions versus current ob- Yu-Sung S, Yajima M, 2015. R2jags: using R to run ‘JAGS’. R package version servations. Glob Planet Change 56:274–296. 0.5-7. Available from: https://cran.r-project.org/web/packages/R2jags/ Street GM, Rodgers AR, Fryxell JM, 2015. Mid-day temperature variation R2jags.pdf. inﬂuences seasonal habitat selection by moose. J Wildl Manage 79: Zeileis A, Grothendieck G, 2005. zoo: S3 infrastructure for regular and irregu- 505–512. lar time series. J Stat Softw 14:1–27. Downloaded from https://academic.oup.com/cz/advance-article-abstract/doi/10.1093/cz/zox047/4054583 by Ed 'DeepDyve' Gillespie user on 12 July 2018
Current Zoology – Oxford University Press
Published: Jul 28, 2017
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera