https://cpw.cvlcollections.org/files/original/4310952630a3335157d66d57dcf1cd85.pdf e6d117ea153618ee798ba24db1afa69a 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 �The Journal of Wildlife Management 82(1):130–137; 2018; DOI: 10.1002/jwmg.21334 Research Article Variation in Ungulate Body Fat: Individual Versus Temporal Effects ERIC J. BERGMAN,1 Colorado Parks and Wildlife, 317 West Prospect Avenue, Fort Collins, CO 80526, USA CHARLES R. ANDERSON, JR, Colorado Parks and Wildlife, 317 West Prospect Avenue, Fort Collins, CO 80526, USA CHAD J. BISHOP,2 Colorado Parks and Wildlife, 317 West Prospect Avenue, Fort Collins, CO 80526, USA A. ANDREW HOLLAND, Colorado Parks and Wildlife, 317 West Prospect Avenue, Fort Collins, CO 80526, USA JOSEPH M. NORTHRUP,3 Colorado State University, Fort Collins, CO 80523, USA ABSTRACT The use of ultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding ungulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals, even those taken from the same herd at that same time, are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in wildlife populations. We evaluated the components of variation in late-winter mule deer (Odocoileus hemionus) body fat estimates: sampling variation (i.e., variation induced by the particular set of individuals that were sampled) and process variation (i.e., variation stemming from biological processes) with a long-term data set (2002–2015) from Colorado, USA. We collected our data from across Colorado as part of historical research, ongoing research, and periodic population monitoring programs. Mean percent ingesta-free body fat (%IFBF) for sampled mule deer was 7.20 � 1.20% (SD). Covariates related to individual deer explained approximately 4% of the total variation in %IFBF and annual effects explained an additional 13% of the variation. Substantial residual variation in %IFBF (83%) remained unexplained. The source of the 83% of unexplained variation is partially linked to fine-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herdlevel inference that can be made from individual late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. However, an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics, with the goal of further refining data that can better allow biologists to incorporate herd health into population management recommendations. Ó 2017 The Wildlife Society. KEY WORDS body condition, Colorado, ingesta-free body fat, mule deer, Odocoileus hemionus, variance components. Wildlife researchers and biologists strive to inform and improve decision-making processes for large ungulate populations. In most cases, new opportunities for achieving these goals are linked to analytical or technological developments. However, on occasion these opportunities include new types of data. The use of body condition metrics to evaluate the status of ungulate populations has a long history in wildlife research (Riney 1960, Anderson et al. 1972, Kistner et al. 1980). During the past 10–15 years, the Received: 16 March 2017; Accepted: 25 July 2017 1 E-mail: eric.bergman@state.co.us Current Address: University of Montana, 32 Campus Drive, Missoula, Montana, 59812 USA. 3 Current Address: Ontario Ministry of Natural Resources and Forestry, Trent University, 2140 East Bank Drive, Peterborough, ON K9L 1Z8 2 130 use of ultrasonograhic measurements of muscle and body fat has become more attainable as a fine-scaled approach to estimating body condition. Historical body condition estimation procedures for ungulates often included either subjective, visual evaluation of live animals (Riney 1960), or body condition evaluation of dead animals (Anderson et al. 1972, Kistner et al. 1980). More recent methods have coupled standardized palpation scores with ultrasonographic measurements to generate live animal body fat estimates (Cook et al. 2001, 2010). The product of this process is an estimate of the percent ingesta-free body fat (%IFBF) for an animal, a parameter that has become more commonly used as a metric in studies of potential management actions (Bishop et al. 2009a, b; Bergman et al. 2014), as a tool to evaluate anthropogenic influences on wildlife (Lendrum et al. 2013, Northrup et al. 2016), and as a tool to better understand life history (Monteith et al. 2013, 2014; Searle et al. 2015). The Journal of Wildlife Management � 82(1) �Through analyzing this relatively new stream of data, we have learned that %IFBF, and presumably overall condition, among individual animals, even those taken from the same herd at that same time, is highly variable (Bergman et al. 2014). During nearly 2 decades of use in wildlife research, reporting of variation in body condition and other physiological parameters in wildlife populations has occurred, yet relatively little effort has been directed at quantifying and evaluating the sources of that variation. Inference from body fat and body condition estimates of freeranging ungulates has the potential to be used for multiple management purposes. For instance, with limited success, the effectiveness of habitat treatments meant to improve population performance has been assessed by measuring changes in body fat of individuals exposed to those treatments (Bergman et al. 2014). Similarly, the decision to provide supplemental feed during harsh winters could be based on the real-time condition of animals. Additionally, some states currently use condition of animals as one parameter, from a suite of additional parameters, to help set hunting license recommendations (J. M. Shannon, Utah Division of Wildlife Resources, personal communication, and A. A. Holland, Colorado Parks and Wildlife [CPW], personal communication). However, because such herd- or population-level management decisions are based on the scaling-up of individual-animal data, we need to better understand the relative contribution of sources of variation (e.g., individual vs. temporal variation) within these data. The total variation in wildlife population data can be broken into 2 distinct sources: sampling variation (i.e., variation induced by the particular set of individuals that were sampled) and process variation (i.e., variation stemming from phenomena such as annual weather patterns, habitat conditions, or predator-prey interactions, all of which influence biological processes). Sampling variation can cause bias in mean estimates of a parameter and is influenced by the specific subset of animals sampled. Process variation is tied to biological processes and thus, for management purposes, is typically of greater interest. Sampling variation may be viewed as a nuisance because it is purely an artifact of which individuals are sampled and not intrinsically linked to a biological process that can be incorporated into decisionmaking. Although the total variance (the sum of process and sampling variation) for parameters of interest is commonly estimated and reported, it is only the subset of variation stemming from biological processes that should influence management decisions. Thus, a robust decomposition of total variation into its 2 components has utility to wildlife biologists and researchers who wish to use population data, such as ungulate body fat, in decision-making processes. We evaluated the components of variation in late-winter mule deer (Odocoileus hemionus) %IFBF estimates using a long-term data set (2002–2015) from Colorado, USA. Our objectives were to explicitly quantify the magnitude of the different sources of variation within %IFBF data, for the more general purposes of determining how best to incorporate these data into harvest and habitat management Bergman et al. � Variance Components of Ungulate Body Fat decision-making. We hypothesized that process variation, and within this, the stochastic and random effects of annual variation would be the single largest source of variation. STUDY AREA Data for our analyses were collected in western Colorado. Mule deer were captured from 3 study units (Fig. 1): Piceance Basin, Uncompahgre Plateau, and HD Mountains. Within these study units, samples were partitioned among sampling sites. In all study units, mule deer were sampled from pinyon pine (Pinyon edulis)-Utah juniper (Juniperus osteosperma) forest winter range. These forests were typified by open understory with occasional sagebrush (Artemisia spp.), cliffrose (Purshia mexicana), antelope bitterbrush (Purshia tridentata), mountain mahogany (Cercocarpus spp.), Utah serviceberry (Amelanchier utahensis), or rabbitbrush (Ericameria spp.) plants. Grasses included western wheatgrass (Pascopyrum smithii), green needlegrass (Nassella viridula), needle and thread (Stipa comata), Indian ricegrass (Achnatherum hymenoides), and bluegrass (Poa spp.). Similarities among study areas also extended to large mammalian fauna communities and land use patterns. Elk (Cervus elaphus) were the primary other large herbivore in each study area, and black bears (Ursus americanus), coyotes (Canis latrans), and mountain lions (Puma concolor) were the primary predators. All study areas were primarily comprised of federally and state managed public lands, although each study area also contained privately managed land at lower elevations. Our northernmost study unit, Piceance Basin, was northeast of Grand Junction, Colorado (n ¼ 6 sampling sites, data collected 2009–2015). Piceance Basin was located in CPW’s northwest region. The climate of Piceance Basin was semi-arid, with average annual moisture of 29.4 cm, Figure 1. Three study areas in Colorado, USA from which adult female mule deer body condition data were collected between 2002 and 2015. Study areas, depicted as solid black polygons, are shown in relation to cities and towns across Colorado. Mule deer herd management boundaries across Colorado are depicted by light gray lines. The northern most study area was centered on the Piceance Basin. The central study area fell on the eastern and southeastern sides of the Uncompahgre Plateau. The southern study area was centered on the HD Mountains. 131 �average winter temperatures of 08C, and average summer temperatures of 15.48C. Topography of Piceance Basin ranged between 1,675 m and 2,285 m and was typified by many intermittent, ephemeral creek drainages. Lower elevations within Piceance Basin were primarily pinyon pine-Utah juniper forest, whereas upper portions gave way to mountain shrub transition range, primarily comprised of Gambel oak (Quercus gambelii) and aspen (Populus tremuloides) but also occasional Utah serviceberry and sagebrush. The Piceance Base mule deer population was largely migratory, exhibiting mean annual movement distances of 38–53 km between seasonal ranges (Lendrum et al. 2013). Land management within Piceance Basin was mixed use, with livestock grazing and energy extraction being the 2 primary factors. Our second study unit, Uncompahgre Plateau, was located near Montrose, Colorado in the west-central portion of the state. This unit overlapped the eastern and southeastern drainages of the Uncompahgre Plateau (n ¼ 5 sampling sites, data collected 2002–2004 and 2005–2009). The Uncompahgre Plateau study unit was located in CPW’s southwest region. It too was a semi-arid landscape with average annual moisture of 38.1 cm, average winter temperature of �18C, and average summer temperature of 13.38C. The general topography of our Uncompahgre Plateau study unit sloped downward in a northeastern direction from elevations of 2,380 m to 1,670 m. As was the case with Piceance Basin, higher elevation portions of the Uncompahgre Plateau study unit gave way to mule deer transition range, which was primarily comprised of Gambel oak and aspen. Land management across the Uncompahgre Plateau study unit was primarily focused on livestock grazing. Our final and southernmost study unit was centered on the HD Mountains to the southeast of Durango, Colorado (n ¼ 2 sampling sites, data collected during 2008). The HD Mountains study unit was also located in CPW’s southwest region. The HD Mountains study site was bordered on the west by the Los Pinos River, on the east by the Piedra River, on the south by Navajo Reservoir, and on the north by the San Juan Mountains. Elevations within the HD Mountains ranged from 1,890 to 2,490 m, average winter temperature was 08C, average summer was 14.18C, and average annual moisture was 34.2 cm. Mule deer sampled from the HD Mountains sampling sites typically migrated north to summer in the San Juan Mountains. Land management within the HD Mountains study unit was mixed use, with livestock grazing and energy extraction being the 2 primary factors. Energy extraction in this study unit was concentrated on private tribal lands at lower elevations. METHODS Field permits for this research were granted by Colorado Parks and Wildlife (Scientific Collection License no. CPW003). Capture and handling procedures for all aspects of this study were approved by the Institutional Animal Care and Use Committees at CPW (protocol no. 11-2000, 1-2002, 10-2005, and 15-2008) and Colorado State University (protocol no. 08-2006A and 10-2350A). Mule 132 deer included in this study were sampled as part of past research projects (Bishop et al. 2009a, b; Lendrum et al. 2013, Bergman et al. 2014, Northrup et al. 2016), ongoing research (Anderson 2015), and as part of periodic population monitoring for management purposes. We determined that a sample size of 30 adult female deer sampled per study site would provide the necessary power to detect a 15% difference in %IFBF between sites, if it existed. Sample size estimates were based on a ¼ 0.05, b ¼ 0.10, and preliminary data that suggested that mean mule deer %IFBF in Colorado likely ranged from 6.7% to 7.2% (SD ¼ 1.69). During the first 2 weeks of March of each winter, researchers captured adult female deer via helicopter netgunning (Webb et al. 2008, Jacques et al. 2009). The goal of capturing deer at this time of year was to estimate %IFBF at the end of winter but before deer moved up in elevation to transition and summer ranges. Prior to the onset of this research, we expected variation in %IFBF during early March to most meaningfully facilitate our understanding of the effects spatial and temporal variation on the overall condition of mule deer. In Colorado, early March is just prior to spring green-up and should represent the nadir of %IFBF for wintering mule deer and also be indicative of reproductive potential relative to late fetal stage maternal condition. Upon capture, researchers immediately blindfolded, hobbled, and ferried deer to a central processing site (�5 km). During the first 3 years of the study, researchers ferried up to 3 deer at a time but subsequently limited ferry events to 2 deer thereafter to reduce overall deer handling time. During the capture process when multiple deer were ferried at once, handlers blindfolded, hobbled, and moved the first captured deer to shade while the remaining capture crew pursued a second animal. To help reduce stress, beginning in 2012, researchers also gave all captured deer a sedative of 35 mg of Midazolam and 15 mg of Azaperone (Wildlife Pharmaceuticals Inc., Fort Collins, CO, USA) immediately at the capture site and prior to ferrying to a central processing site. At the field-processing site, researchers weighed deer, estimated age via tooth eruption and wear patterns (Severinghaus 1949, Robinette et al. 1957, Hamlin et al. 2000), and measured chest girth and hind foot length. We also measured the maximum subcutaneous fat thickness (cm) on the rump and the thickness of the longissimus dorsi muscle (Cook et al. 2001) using a portable ultrasound machine and a 5-MHz linear transducer (Universal Medical Systems, Bedford Hills, NY, USA). We determined a body condition score for each animal by palpating the rump (Cook et al. 2001). We combined body condition scores with ultrasound measurements to generate a scaled estimate of the %IFBF for each animal (Cook et al. 2010). For consistency, the same 4 individual observers made all body condition scoring and ultrasonographic measurements. At least 2 of these observers were present for each animal included in this study. Prior to collection of data in the field, all 4 observers became familiar with the techniques of Cook et al. (2001) in CPW’s wildlife health captive animal facility. The Journal of Wildlife Management � 82(1) �Analytical Methods At the onset, we recognized the limits to evaluating %IFBF using a post hoc approach. Whereas experimental manipulation of %IFBF, with simultaneous measures taken in multiple areas and repeated through time would offer a more robust design, such an opportunity would be cost prohibitive over the time-period and spatial scale we examined and our data provided a solid evaluation of relative variance contributions. To assess the different sources of variation influencing individual mule deer %IFBF, we fit a hierarchical Bayesian linear regression model to the log transformed %IFBF estimate of each deer. To address individual factors influencing %IFBF, we included covariates for hind foot length (HFL), chest girth (CG), and mass. To account for broad scale spatial variation in %IFBF, we included a covariate for the region in which the deer was captured (coded as 1 for the northwest region and 0 for the southwest region). Lastly, to account for variation across years, we allowed the intercept to vary by year (i.e., a random effect). We standardized the individual level covariates and assessed pairwise correlations among all independent covariates (no covariates were correlated at |r| > 0.7). Ultimately, we fit the model d ¼ b þ b ðHFLÞ þ b ðCGÞ þ b ðMassÞ þ b ðRegionÞ þ b ðYearÞ %IFBF 0 1 2 3 4 5 using the Stan programming language in the R statistical software (R Core Development Team 2015) using the package rstan (Stan Development Team 2014a, b). We obtained 10,000 Hamiltonian Monte Carlo (HMC) iterations for each model run, discarding the first 5,000 as burn-in. We ran the algorithm 4 times to obtain 4 HMC chains, each initiated with random starting values. We assessed convergence of parameters to their posterior distribution by examining all traceplots of the resulting chains and by calculating the Gelman-Rubin diagnostic (mean values close to 1 indicate convergence; Gelman and Rubin 1992). The above modeling procedure can be used to parse the variance in %IFBF into 3 components: variance explained by available information on individuals (i.e., foot length, chest girth, mass, and the region in which the animal was captured), annual variation, and residual variance. To assess these different sources of variance, we followed the procedure of Nakagawa and Schielzeth (2013) to determine the amount of variance explained by the fixed and random effects of a mixed (i.e., hierarchical) model. In this procedure, the marginal R2 indicates the amount of variance explained by the fixed (i.e., non-varying) terms and the conditional R2 indicates the variance explained by the full model structure. We calculated these values using our model results. Table 1. Spatial and temporal distribution of adult female mule deer body condition data that were collected in Colorado, USA. Data were collected as part of completed and ongoing research projects and management studies. We used data to evaluate temporal aspects of variation in body condition data. Year Study area n 2002 2003 2004 2005 2006 2007 2008 Uncompahgre Plateau Uncompahgre Plateau Uncompahgre Plateau No areas sampled Uncompahgre Plateau Uncompahgre Plateau Uncompahgre Plateau HD Mountains Uncompahgre Plateau Piceance Basin Piceance Basin Piceance Basin Piceance Basin Piceance Basin Piceance Basin Piceance Basin 18 28 30 0 60 60 60 40 60 150 103 79 120 120 118 117 1,163 2009 2010 2011 2012 2013 2014 2015 Total distribution of data collection among years was clustered. Between 2002 and 2008 data were collected from the Uncompahgre Plateau and HD Mountains study areas in southwest Colorado (Table 1). During 2009 our data were collected in both the Uncompahgre Plateau and Piceance Basin study areas. Between 2010 and 2015, data were collected in the Piceance Basin study area in northwest Colorado (Table 1). Mean %IFBF for mule deer we sampled was 7.20 � 1.20% (SD). Although data were collected from 3 distinct study areas across western Colorado over a 14-year period, mean estimates of %IFBF showed little variation among years (Fig. 2). Within this observation, an outlier appeared to occur in 2006 when mean %IFBF was higher. However, even during this year the estimated %IFBF was within 1 standard RESULTS As part of this research, we used %IFBF estimates from 1,163 unique mule deer (Table 1). Because of the post hoc nature of these analyses, sampling among study areas and years was unbalanced. Although data were collected on an annual basis (with the exception of 2005), the spatial Bergman et al. � Variance Components of Ungulate Body Fat Figure 2. Annual estimated percent ingesta-free body fat (%IFBF) values for adult female mule deer in Colorado, USA that were captured during early March from 2002 to 2015. Annual estimated mean values are bracketed by bars reflecting 1 standard deviation. The maximum and minimum observed value for each year is depicted by a black X. The overall mean estimated value (7.20), across years, is reflected by the horizontal dashed black line. 133 �deviation of the overall mean (Fig. 2). The greatest annual fluctuation within our data occurred within the maximum observed values. Maximum observed values for %IFBF ranged between 8.83% and 16.10%. Alternatively, annual variation within our minimum observed values was less because mean values were skewed low during late winter, and ranged between 1.99% and 5.90%. Our model converged for all parameters (Gelman-Rubin diagnostics all were estimated to be equal to 1). Our ability to correlate %IFBF data with spatial and individual metrics was limited by the spatiotemporal distribution of data collection and by the metrics that we collected at the time of capture. There was evidence that %IFBF estimates from southwest Colorado were lower than those from northwest Colorado (Table 2), although the observed effect size was small (b ¼ �0.042) and 6% of the posterior distribution was >0. However, this effect was also largely confounded with time because the 2 sites were only sampled simultaneously for a single year. Mass and chest girth were positively correlated with %IFBF (Table 2). Alternatively, hind foot length was negatively correlated with %IFBF (Table 2). Within these individual covariates, only mass had all of its posterior distribution >0. However, the effect size for predicting % IFBF from mass was small (b ¼ 0.034), thus predicting that each additional kilogram in body mass resulted in only a 0.03 increase in %IFBF. The marginal R2 was 0.038, and the conditional R2 was 0.166, indicating that our individual covariates explained approximately 4% of the total variation in %IFBF and annual effects explained an additional 13% of the variation. Thus, substantial residual variation in %IFBF (83%) remained unexplained (Fig. 3). In the context of process and sampling variation, our model estimated that only 13% of the variation was due to the effects of year. The source of the 83% of unexplained variation is linked to finer scale spatial dynamics and additional individual metrics we were unable to capture. DISCUSSION Mule deer have an adaptive ability to tolerate a negative energy balance over winter (Wallmo 1981). In the majority of their range, energetic demands on mule deer exceed what can be met via daily caloric intake and surviving winter becomes a slow race against malnourishment. Presumably, individuals who use the least amount of energetic reserves (i.e., %IFBF) while maximizing daily caloric intake will Table 2. Statistics of Bayesian hierarchical linear regression models, including covariates, median coefficient (b) estimates and the proportion (prop.) of posterior distributions falling below or above 0. We fit models to log transformed estimated percent ingesta-free body fat data from adult female mule deer in Colorado, USA, 2002–2015. Covariate Mean year Mass Foot length Chest girth Region 134 Median b Prop. <0 Prop. >0 1.914 0.034 �0.010 0.013 �0.042 0.00 0.92 0.06 0.94 1.00 0.08 0.94 0.06 Figure 3. Evaluation of the uncertainty and variation among years of predicted late-winter percent ingesta-free body fat estimates from adult female mule deer in Colorado, USA, 2002–2015. Black diamonds depict the annual posterior distributions of predicted body fat across years while holding all morphometric measurements constant at their respective means. Shaded bars reflect the predicted probability density of predictions including residual variance. Black dots depict the distribution of raw data collected each year. experience positive survival and fitness benefits. However, evidence from Monteith et al. (2013) and ongoing research in Colorado (C. R. Anderson, CPW, unpublished data) suggest that animals with higher %IFBF during early winter tend to lose more mass and thereby exit winter with %IFBF levels comparable to other animals. Anecdotally, it appears that mule deer with the greatest early winter %IFBF are afforded a greater opportunity for daily caloric mismatch. Thus, understanding the intermediate role of body condition in the population dynamics of deer serves as motivation for evaluating body condition metrics, such as %IFBF, of individual animals. However, careful dissection of body condition parameters is needed to make them useful at the population level. Under many management scenarios the role of sampling variation in parameter estimates is secondary to the role of process variation (Unsworth et al. 1999, Lukacs et al. 2009). This perspective is largely intuitive; the fact that population parameter estimates for herds can change based solely on which individual animals are sampled is a nuisance. However, in the case of %IFBF, and in stark contrast to our hypothesis, we found that sampling variation was substantial and process variation was nominal. Of the variation, 87% was due to sampling (individual) variation, whereas only 13% was due to annual effects. However, this somewhat contradictory result is also intuitive; the body condition of an individual mule deer in March is an artifact of that individual animal’s decision-making and life history over the preceding months and years. We were surprised by the relatively small contribution of year to the variance in our late-winter %IFBF estimates. Western Colorado is a diverse, mountainous landscape that experiences periodic, harsh winters. For instance, throughout The Journal of Wildlife Management � 82(1) �western Colorado harsh winters occurred during the winters of 2007–2008 and 2010–2011, but we did not observe herdwide drops in %IFBF during this period (Figs. 2 and 3). The stochastic effect of winter severity on mortality of deer, a commonly referenced biological phenomenon (Unsworth et al. 1999; Hurley et al. 2011, 2014), directly fits within our definition of process variation. Yet those effects on survival may not readily appear in estimated body condition parameters. Similarly, process variation may be less informative than expected because many adult female deer are near the minimum survivable body condition, and the deer in poorest condition would not have survived to be sampled during our late-winter sampling period. Preliminary survival data from Colorado indicate that adult females with lower %IFBF in November have lower survival to March (C. R. Anderson, unpublished data), which supports this speculation. Thus, the set of circumstances under which mule deer body condition and late-winter %IFBF estimates are most useful is obscured by the mortality that occurs as body condition declines. We speculate that the primary factors influencing late-winter mule deer %IFBF, and body condition in general, are individual in nature. Thus, utility of %IFBF data could be maximized if estimates were generated from animals with known histories. The animals included in our study had not been captured or observed prior to our work and we had no way of knowing which of those animals had raised offspring during the preceding 1–2 years. Similarly, we had no way of knowing which animals may have nursed twins and for how long, 2 factors that are influential to individual condition (i.e., as evidenced by observation of adult female deer with fawns during fall being highly influential for individual condition; Monteith et al. 2013). Previous authors have noted that pregnancy status at the time of capture can be correlated with body condition, although those correlations have tended to be weak (Bishop et al. 2009b, Monteith et al. 2013, Bergman et al. 2014). The correlation is weak likely because pregnancy rates in mule deer, especially in western Colorado, tend to be high and largely invariant (Bishop et al. 2009a, b; Bergman et al. 2014). Thus, we believe that pregnancy status, fetal rate, and fetal and neonatal survival of offspring during the preceding year would capture much of the variation in %IFBF we could not explain. Given the late-winter timing of our annual capture effort, and the high probability of having many false negatives, we do not expect that lactation status data at the time of capture would have led to a meaningful reduction in unexplained sampling variance. However, we do believe that if %IFBF sampling occurred earlier, for instance during December or January, when lactation status could be more accurately assessed, these additional data would indeed prove to be useful. Extension of this hypothesis highlights how spatial effects might also explain variation in body condition data. For instance, maternal deer living in areas with high predator densities may experience increased mortality of newborn neonates, especially during the first several weeks of life. The subsequent transfer of energy away from lactation and into maternal reserves would likely carry forward into the next winter. Bergman et al. � Variance Components of Ungulate Body Fat Despite these speculations, we recognize short-comings in our study design. For instance, we struggled to discern between the effects of space and time given the spatiotemporal clustering resulting from how our data were collected. However, there was little evidence that regional spatial dynamics had a major influence on deer body condition. Given that all data collected between 2002 and 2008 came from southwest Colorado, and data collected between 2009 and 2015 came from northwest Colorado, any regional effects should show up as an upwards or downwards shift in all of the data points for those respective time periods. We did not observe such shifts (Figs. 2 and 3), which largely serves as a visual confirmation of the weak statistical effect we detected (Table 2). In fact, the relatively consistent mean %IFBF values that we measured, despite these sampling issues, lends strong credence to the argument that factors acting at a finer scale (i.e., the individual) were the primary influences on %IFBF. We also recognize that our data came from a single state, and from a single type of winter range. Other regions of North America may experience more severe and a wider range of winter conditions among years, which could indeed have more dramatic herd-level impacts. However, although not explicitly evaluated, Monteith et al. (2014) report consistently high levels of individual variation in similar data for mule deer in California. This lends evidence toward the perspective that our observations likely are not unique to Colorado. Ultimately, these results lead us to speculate on the use of herd-level late-winter body fat estimates. Simple population models typically only account for intuitive parameters, such as births, deaths, immigration, and emigration. Yet assessment of the body condition of free ranging deer species has a long history. It is thus clear that wildlife managers would like to have a tool that is intermediate to the live or dead fate of an animal. Monteith et al. (2014) suggest that herd-level body condition can also be used to distinguish between the additive versus compensatory nature of predation. Alternatively, in specific instances, such as winter feeding, treatment effects can clearly be detected using %IFBF (Bishop et al. 2009a, b). But in the absence of such a dramatic treatment effect, our data would have had more utility if we had known more about the individual deer that were sampled and thus could better account for the obviously substantial individual variation in %IFBF. We also postulate that different subsets of body condition data may have more use in management scenarios. For instance, male deer and young-of-the year (fawns) do not face the lactation costs that maternal females face. This may indeed reduce individual variation in body condition and %IFBF data. Yet adult males face severe energetic costs during breeding seasons. Those costs can also be expected to be carried into winter. Similarly, body condition data from dead animals has been used in the past (Anderson et al. 1972). However, opportunistic sampling of road-killed and hunter-harvested animals are likely prone to sampling bias. In particular, hunter-harvested samples would not reflect the effects of winter because harvest typically occurs during autumn. The effect of fawn mass on overwinter survival of fawns has been reported in 135 �Colorado and elsewhere (Unsworth et al. 1999, Bishop et al. 2009b, Hurley et al. 2011). Estimates of short-yearling mass (i.e., 10 months old) of moose (Alces alces) has been used for evaluation of herd nutritional status (Boertje et al. 2007). However, until further investigated, we can only speculate on the utility of fawn body condition or fawn mass in evaluating mule deer herd population health. The question of how best to scale-up individual health and condition parameters to better understand and manage population health is far from intuitive and not unique to wildlife management. We feel that an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics, with the goal of further refining which data will best allow biologists to incorporate herd health into population management decision-making. Likewise, we caution against the inference of our results from mule deer to other large ungulates. For instance, the majority of mule deer in Colorado have 2 fetuses per year (Bishop et al. 2009b). Thus, the annual reproductive output of individual animals has the potential to be highly variable (i.e., mature deer may nurse 0, 1, or 2 fawns for different periods of time, depending on timing of fawn mortality). For ungulate species with lower expected individual reproductive output, such as elk or Shiras moose (Alces alces shirasi), the unexplained sampling variation in body condition or %IFBF could indeed be much less. Alternatively, for adult females of species that frequently carry up to 3 fetuses but also resorb fetuses when nutritionally stressed, such as pronghorn (Antilocapra americana), the observed and unexplained sampling variation may indeed be greater than what we quantified for mule deer. MANAGEMENT IMPLICATIONS Our results suggest that managers will better understand mule deer %IFBF data if they simultaneously have other relevant life-history data, such as reproductive success or failure, for each individual. In the absence of that additional information, we suggest that %IFBF data may be of greater use when collected earlier during the year, such as when accurate lactation status data are available. For instance, in Colorado, %IFBF data are now collected during early- and late-winter periods for each deer, and when repeated sampling is not an option, single sampling events occur during early-winter periods. Likewise, repeated samples of individual mule deer fawn weights (during early- and latewinter periods) are now collected to supplement the data and insights stemming from our estimation of the components of variation in %IFBF. ACKNOWLEDGMENTS We greatly appreciate the contributions of L. L. Wolfe and CPW’s Wildlife Health Lab to this research. Fixed wing pilots L. D. Felix, L. A. Gepfert, and S. Waters provided assistance with aerial observation and spotting during capture efforts. Helicopter pilots R. A. Swisher, B. R. Malo, and M. A. Shelton provided safe and efficient capture services. Thoughtful reviews on earlier drafts of this work were provided by A. D. Apa and A. C. Behney. The original 136 draft of this manuscript was greatly improved through comments provided by K. M. Proffitt, and 2 anonymous reviewers. Financial support for this research was provided by: Colorado Federal Aid Wildlife Restoration Project Funding, the CPW Habitat Partnership Program, the Mule Deer Foundation, the CPW Big Game Auction and Raffle Grant program, EnCana Corporation, Exxon Mobil Production Company, XTO Energy, Marathon Oil Corporation, Shell Petroleum, and WPX Energy. LITERATURE CITED Anderson, C. R., Jr. 2015. Population performance of Piceance Basin mule deer in response to natural gas resource extraction and mitigation efforts to address human activity and habitat degradation. Wildlife Research Report, Colorado Parks and Wildlife, Fort Collins, USA. Anderson, A. E., D. E. Medin, and D. C. Bowden. 1972. Indices of carcass fat in a Colorado mule deer population. Journal of Wildlife Management 36:579–594. Bergman, E. J., P. F. Doherty Jr., C. J. Bishop, L. L. Wolfe, and B. A. Banulis. 2014. Herbivore body condition response in altered environments: mule deer and habitat management. PLoS ONE 9:e106374. Bishop, C. J., B. E. Watkins, L. L. Wolfe, D. J. Freddy, and G. C. White. 2009a. Evaluating mule deer body condition using serum thyroid hormone concentrations. Journal of Wildlife Management 73:462–467. Bishop, C. J., G. C. White, D. J. Freddy, B. E. Watkins, and T. R. Stephenson. 2009b. Effect of enhanced nutrition on mule deer population rate of change. Wildlife Monographs 172:1–28. Boertje, R. D., K. A. Kellie, C. T. Seaton, M. A. Keech, D. D. Young, B. W. Dale, L. G. Adams, and A. R. Aderman. 2007. Ranking Alaska moose nutrition: signals to begin liberal antlerless harvests. Journal of Wildlife Management 71:1494–1506. 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., J. G. Cook, T. R. Stephenson, W. L. Myers, S. M. McCorquodale, D. J. Vales, L. L. Irwin, P. B. Hall, R. D. Spencer, S. L. Murphie, K. A. Schoenecker, and P. J. Miller. 2010. Revisions of rump fat and body scoring indices for deer, elk, and moose. Journal of Wildlife Management 74:880–896. Gelman, A, and D. B. Rubin. 1992. Inference from iterative simulation using multiple sequences. Statistical Science 7:457–511. Hamlin, K. L., D. F. Pac, C. A. Sime, R. M. DeSimone, and G. L. Dusek. 2000. Evaluating the accuracy of ages obtained by two methods for Montana ungulates. Journal of Wildlife Management 64:441–449. Hurley, M. A., M. Hebblewhite, J. M. Gaillard, S. Dray, K. A. Taylor, W. K. Smith, P. Zager, and C. Bonenfant. 2014. Functional analysis of normalized difference vegetation index curves reveal overwinter mule deer survival is driven by both spring and autumn phenology. Philosophical Transactions of the Royal Society B 369:20130196. Hurley, M. A., J. W. Unsworth, P. Zager, M. Hebblewhite, E. O. Garton, D. M. Montgomery, J. R. Skalski, and C. L. Maycock. 2011. Demographic response of mule deer to experimental reduction in coyotes and mountain lions in southeastern Idaho. Wildlife Monographs 178:1–33. Jacques, C. N., J. A. Jenks, C. S. Deperno, J. D. Sievers, T. W. Grovenburg, T. J. Brinkman, C. C. Swanson, and B. A. Stillings. 2009. Evaluating ungulate mortality associated with helicopter net-gun captures in the Northern Great Plains. Journal of Wildlife Management 73:1282–1291. Kistner, T. P., C. E. Trainer, and N. A. Hartmann. 1980. A field technique for evaluating physical condition of deer. Wildlife Society Bulletin 8:11–17. Lendrum, P. E., C. R. Anderson Jr., K. L. Monteith, J. A. Jenks, and R. T. Bowyer. 2013. Migrating mule deer: effects of anthropogenically altered landscapes. PLoS ONE 8:e64548. Lukacs, P. M., G. C. White, B. E. Watkins, R. H. Kahn, B. A. Banulis, D. J. Finley, A. A. Holland, J. A. Martens, and J. Vayhinger. 2009. Separating components of variation in survival of mule deer in Colorado. Journal of Wildlife Management 73:817–826. Monteith, K. L., V. C. Bleich, T. R. Stephenson, B. M. Pierce, M. M. Conner, J. G. Kie, and R. T. Bowyer. 2014. Life-history characteristics of The Journal of Wildlife Management � 82(1) �mule deer: effects of nutrition in a variable environment. Wildlife Monographs 186:1–62. Monteith, K. L., T. R. Stephenson, V. C. Bleich, M. M. Conner, B. M. Pierce, and R. T. Bowyer. 2013. Risk-sensitive allocation in seasonal dynamics of fat and protein reserves in a long-lived mammal. Journal of Animal Ecology 82:377–388. Nakagawa, S., and H. Schielzeth. 2013. A general and simple method for obtaining R2 from generalized linear mixed-effects models. Methods in Ecology and Evolution 4:133–142. 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. R Core Development Team. 2015. R: a language and environment for statistical computing. R Development Core Team, Vienna, Austria. Riney, T. 1960. A field technique for assessing physical condition of some ungulates. Journal of Wildlife Management 24:92–94. Robinette, W. L., D. A. Jones, G. Rogers, and J. S. Gashwiler. 1957. Notes of tooth development and wear for Rocky Mountain mule deer. Journal of Wildlife Management 21:134–153. Bergman et al. � Variance Components of Ungulate Body Fat Searle, K. R., M. B. Rice, C. R. Anderson Jr., C. Bishop, and N. T. Hobbs. 2015. Asynchornous vegetation phenology enhances winter body condition of a large mobile herbivore. Oecologia 179:377–391. Severinghaus, C. W. 1949. Tooth development and wear as criteria of age in white-tailed deer. Journal of Wildlife Management 13:195–216. Stan Development Team. 2014a. RStan: the R interface to Stan, Version 2.4. <http://mc-stan.org>. Accessed 16 May 2016. Stan Development Team. 2014b Stan: a C++ library for probability and sampling, Version 2.4. <http://mc-stan.org>. Accessed 16 May 2016. Unsworth, J. W., D. F. Pac, G. C. White, and R. M. Bartmann. 1999. Mule deer survival in Colorado, Idaho, and Montana. Journal of Wildlife Management 63:315–326. Wallmo, O. C. 1981. Mule and black-tailed deer of North America. University of Nebraska Press, Lincoln, USA. Webb, S. L., J. S. Lewis, D. G. Hewitt, M. W. Hellickson, and F. C. Bryant. 2008. Assessing the helicopter and net gun as a capture technique for white-tailed deer. Journal of Wildlife Management 72:310–314. Associate Editor: Kelly Proffitt. 137 � 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 Variation in ungulate body fat Individual versus temporal effects Rights Information about rights held in and over the resource <a href="http://rightsstatements.org/vocab/InC-NC/1.0/">In Copyright - Non-Commercial Use Permitted</a> Date Created Date of creation of the resource. 2017-08-30 Subject The topic of the resource Body condition Colorado Ingesta-free body fat Mule deer <em>Odocoileus hemionus</em> Variance components Description An account of the resource The use of ultrasonograhic measurements of muscle and body fat represent a relatively new data stream that can be used to address questions regarding ungulate condition. We have learned that measurements of body fat and presumably overall body condition among individual animals, even those taken from the same herd at that same time, are highly variable. Relatively little consideration has been given to the sources of variation in body fat and other physiological parameters in wildlife populations. We evaluated the components of variation in late-winter mule deer (<em>Odocoileus hemionus</em>) body fat estimates: sampling variation (i.e., variation induced by the particular set of individuals that were sampled) and process variation (i.e., variation stemming from biological processes) with a long-term data set (2002–2015) from Colorado, USA. We collected our data from across Colorado as part of historical research, ongoing research, and periodic population monitoring programs. Mean percent ingesta-free body fat (%IFBF) for sampled mule deer was 7.201.20% (SD). Covariates related to individual deer explained approximately 4% of the total variation in %IFBF and annual effects explained an additional 13% of the variation. Substantial residual variation in %IFBF (83%) remained unexplained. The source of the 83% of unexplained variation is partially linked to fine-scale spatial dynamics but also additional individual metrics we were unable to capture, primarily the presence or absence of dependent young. We speculate that the primary factors influencing late-winter mule deer body fat and overall condition are individual in nature. These results present a cautionary check on herdlevel inference that can bemade from individual late-winter body fat estimates and we postulate that for mule deer, alternative and additional body condition metrics may offer added utility in management scenarios. However, an important next step to better understand wildlife population health is to evaluate the sources and magnitude of variation within other body condition metrics, with the goal of further refining data that can better allow biologists to incorporate herd health into population management recommendations. Creator An entity primarily responsible for making the resource Bergman, Eric J. Anderson, Jr, Charles R. Bishop, Chad J. Holland, A. Andrew Northrup, Joseph M. Language A language of the resource English Is Part Of A related resource in which the described resource is physically or logically included. The Journal of Wildlife Management Extent The size or duration of the resource. 8 pages Bibliographic Citation A bibliographic reference for the resource. Recommended practice is to include sufficient bibliographic detail to identify the resource as unambiguously as possible. Bergman, E. J., C. R. Anderson Jr., C. J. Bishop, A. A. Holland, and J. M. Northrup. 2017. Variation in ungulate body fat: individual versus temporal effects. The Journal of Wildlife Management 82:130–137. <a href="https://doi.org/10.1002/jwmg.21334" target="_blank" rel="noreferrer noopener">https://doi.org/10.1002/jwmg.21334</a> Format The file format, physical medium, or dimensions of the resource application/pdf Type The nature or genre of the resource Article