Rachel C. Cook

Animal migration amid shifting patterns of phenology and predation: lessons from a Yellowstone elk herd

Migration is a striking behavioral strategy by which many animals enhance resource acquisition while reducing predation risk. Historically, the demographic benefits of such movements made migration common, but in many taxa the phenomenon is considered globally threatened. Here we describe a long-term decline in the productivity of elk (Cervus elaphus) that migrate through intact wilderness areas to protected summer ranges inside Yellowstone National Park, USA. We attribute this decline to a long-term reduction in the demographic benefits that ungulates typically gain from migration. Among migratory elk, we observed a 21-year, 70% reduction in recruitment and a 4-year, 19% depression in their pregnancy rate largely caused by infrequent reproduction of females that were young or lactating. In contrast, among resident elk, we have recently observed increasing recruitment and a high rate of pregnancy. Landscape-level changes in habitat quality and predation appear to be responsible for the declining productivity of Yellowstone migrants. From 1989 to 2009, migratory elk experienced an increasing rate and shorter duration of green-up coincident with warmer spring-summer temperatures and reduced spring precipitation, also consistent with observations of an unusually severe drought in the region. Migrants are also now exposed to four times as many grizzly bears (Ursus arctos) and wolves (Canis lupus) as resident elk. Both of these restored predators consume migratory elk calves at high rates in the Yellowstone wilderness but are maintained at low densities via lethal management and human disturbance in the year-round habitats of resident elk. Our findings suggest that large-carnivore recovery and drought, operating simultaneously along an elevation gradient, have disproportionately influenced the demography of migratory elk. Many migratory animals travel large geographic distances between their seasonal ranges. Changes in land use and climate that disparately influence such seasonal ranges may alter the ecological basis of migratory behavior, representing an important challenge for, and a powerful lens into, the ecology and conservation of migratory taxa.

Development of predictive models of nutritional condition for Rocky Mountain elk

Despite its preeminence as a game species in North America, little research exists to validate nutritional condition indices for Rocky Mountain elk (Cervus elaphus nelsonii). We developed and calibrated indices of nutritional condition for live and dead Rocky Mountain elk. Live-animal indices included 20 serum and 7 urine chemistry variables, a body-condition score (BCS), thickness of subcutaneous fat and selected muscles using ultrasonography, bioelectrical impedance analysis (BIA), and body mass. Dead-animal indices included femur and mandible marrow fat, 3 kidney fat indices, and 2 carcass-scoring methods. Forty-three captive-raised cows (1.5 to 7 years old) were randomly divided into 3 seasonal groups (Sep, Dec, and Mar). Within seasonal groups, elk were fed different diets to induce a wide range of condition; all were fed identical diets 7 days prior to sampling to eliminate short-term nutritional effects. Cows were euthanized and homogenized for chemical analysis of fat, protein, water, and ash content. Estimates of fat and gross energy (GE) were compared to each condition indicator using regression, with age and season as covariates. Relations between condition and thyroxine (T 4 ) and insulin-like growth factor (IGF-1) varied seasonally, and the relation between condition and mandible marrow fat varied among ages. Subcutaneous fat depth and BCS were most related to condition for live animals (r 2 ≥ 0.87, P < 0.001); carcass scores and kidney fat were most related to fat and GE for dead animals (r 2 ≥ 0.77, P < 0.001); and IGF-1 and T 4 were the only serum and urine indices at least moderately related to condition (r 2 ≥ 0.54, P < 0.001). Nearly all other serum and urine indices, bone marrow indices, and BIA were either poorly correlated with condition or exhibited highly nonlinear relations. These results identify several indices of condition useful for assessing nutritional condition of live or dead elk, and indicate a number of previously used techniques that correlate poorly with total body fat.

Nutritional Condition of Northern Yellowstone Elk

Abstract We estimated nutritional condition for 96 female northern Yellowstone elk (Cervus elaphus nelsoni) during mid- to late winter 2000, 2001, and 2002. Neither year nor capture location significantly influenced any measure of condition (body fat, body mass, and longissimus dorsi thickness; P ≥ 0.14). Overall, age = 8.9 years ± 0.4 SE, body fat = 9.5% ± 0.4, body mass = 235.1 kg ± 2.2, and longissimus dorsi muscle thickness = 5.6 cm ± 0.1. Despite an age segregation pattern across the winter range (P = 0.016), we found no evidence of bias in our estimates of nutritional condition due to this pattern because condition was unrelated to age. Yearly pregnancy and lactation rates of all cows ranged from 78 to 84% and 8 to 16%, respectively, at the time of capture. Lactational status significantly influenced body condition (P = 0.003), with lactating cows having 50% less body fat than nonlactating cows. Probability of pregnancy observed for elk that we captured followed a logistic curve as a function of body fat levels. Based on mid- to late winter body fat levels, we would predict low mortality of adult cows during mild to normal winters. We suggest the possibility of nutritional limitations acting on this herd through summer–autumn forage conditions, in association with limitations during harsh winters.

Linking anti-predator behaviour to prey demography reveals limited risk effects of an actively hunting large carnivore

Ecological theory predicts that the diffuse risk cues generated by wide-ranging, active predators should induce prey behavioural responses but not major, population- or community-level consequences. We evaluated the non-consumptive effects (NCEs) of an active predator, the grey wolf (Canis lupus), by simultaneously tracking wolves and the behaviour, body fat, and pregnancy of elk (Cervus elaphus), their primary prey in the Greater Yellowstone Ecosystem. When wolves approached within 1 km, elk increased their rates of movement, displacement and vigilance. Even in high-risk areas, however, these encounters occurred only once every 9 days. Ultimately, despite 20-fold variation in the frequency of encounters between wolves and individual elk, the risk of predation was not associated with elk body fat or pregnancy. Our findings suggest that the ecological consequences of actively hunting large carnivores, such as the wolf, are more likely transmitted by consumptive effects on prey survival than NCEs on prey behaviour.

Revisions of Rump Fat and Body Scoring Indices for Deer, Elk, and Moose

Abstract Because they do not require sacrificing animals, body condition scores (BCS), thickness of rump fat (MAXFAT), and other similar predictors of body fat have advanced estimating nutritional condition of ungulates and their use has proliferated in North America in the last decade. However, initial testing of these predictors was too limited to assess their reliability among diverse habitats, ecotypes, subspecies, and populations across the continent. With data collected from mule deer (Odocoileus hemionus), elk (Cervus elaphus), and moose (Alces alces) during initial model development and data collected subsequently from free-ranging mule deer and elk herds across much of the western United States, we evaluated reliability across a broader range of conditions than were initially available. First, to more rigorously test reliability of the MAXFAT index, we evaluated its robustness across the 3 species, using an allometric scaling function to adjust for differences in animal size. We then evaluated MAXFAT, rump body condition score (rBCS), rLIVINDEX (an arithmetic combination of MAXFAT and rBCS), and our new allometrically scaled rump-fat thickness index using data from 815 free-ranging female Roosevelt and Rocky Mountain elk (C. e. roosevelti and C. e. nelsoni) from 19 populations encompassing 4 geographic regions and 250 free-ranging female mule deer from 7 populations and 2 regions. We tested for effects of subspecies, geographic region, and captive versus free-ranging existence. Rump-fat thickness, when scaled allometrically with body mass, was related to ingesta-free body fat over a 38–522-kg range of body mass (r2  =  0.87; P < 0.001), indicating the technique is remarkably robust among at least the 3 cervid species of our analysis. However, we found an underscoring bias with the rBCS for elk that had >12% body fat. This bias translated into a difference between subspecies, because Rocky Mountain elk tended to be fatter than Roosevelt elk in our sample. Effects of observer error with the rBCS also existed for mule deer with moderate to high levels of body fat, and deer body size significantly affected accuracy of the MAXFAT predictor. Our analyses confirm robustness of the rump-fat index for these 3 species but highlight the potential for bias due to differences in body size and to observer error with BCS scoring. We present alternative LIVINDEX equations where potential bias from rBCS and bias due to body size are eliminated or reduced. These modifications improve the accuracy of estimating body fat for projects intended to monitor nutritional status of herds or to evaluate nutritions influence on population demographics.

Validating Predictive Models of Nutritional Condition for Mule Deer

Abstract We developed new, and validated existing, indices of nutritional condition for live and dead mule deer (Odocoileus hemionus). Live animal indices included a body condition score (BCS), thickness of subcutaneous fat and selected muscles using ultrasonography, and body mass. Dead animal indices included femur, metatarsal, and mandible marrow fat, 3 kidney fat indices, and 2 carcass scoring methods. We used 21 female deer and 4 castrates (1–11 yr old) varying widely in nutritional condition (2–28% ingesta-free body fat). Deer were euthanized and homogenized for chemical analysis of fat, protein, water, and ash content. Estimates of fat and gross energy (GE) were regressed against each condition indicator using regression. Subcutaneous fat thickness, a rump BCS, and rLIVINDEX (an arithmetic combination of subcutaneous fat thickness and the rump BCS) were most related to condition for live animals (r2 ≥ 0.87, P < 0.001) whereas the Kistner score and kidney fat were most related to fat and GE for dead animals (r2 ≥ 0.77, P < 0.001). We also evaluated range of usefulness and sensitivity to small changes in body condition for all models. In general, indices with moderate or highly curvilinear statistical relations to body fat or those based on only one fat depot or a small number of ranking scores will have limitations in their use. Our results identify robust tools for a variety of research and monitoring designs useful for evaluating nutritions effect on mule deer populations.

Nutritional condition models for elk : Which are the most sensitive, accurate, and precise?

Traditionally, biologists have developed indices that assess nutrition and condition of wild ungulates. However, many attempts to validate such indices have failed to indicate the range of conditions under which they work. Furthermore, such validation tests often fail to identify sensitivity to small but biologically meaningful differences, and emphasize statistical rather than biological relationships. We evaluated 20 models that were developed to assess nutritional condition of Rocky Mountain elk (Cerous elaphus nelsonii). We analyzed sensitivity, bias, accuracy, precision, applicability across a wide range of body conditions, and field practicality. Because models were derived using captive elk, we incorporated data from 6 wild cows to assess suitability of condition-index models for free-ranging elk. We found that most condition indicators available to biologists were weakly related to actual nutritional condition, were insensitive to small changes in condition, or often showed nonlinear relations that restricted their value to a narrow range of body condition. An arithmetic combination of a rump body-condition score and subcutaneous rump-fat thickness for live animals, and a modified carcass-evaluation score for dead animals, were the most sensitive and accurate indices of nutritional condition that we tested.

Body condition and pregnancy in northern Yellowstone elk: Evidence for predation risk effects?

S. Creel et al. reported a negative correlation between fecal progesterone concentrations and elk:wolf ratios in greater Yellowstone elk (Cervus elaphus) herds and interpreted this correlation as evidence that pregnancy rates of elk decreased substantially in the presence of wolves (Canis lupus). Apparently, the hypothesized mechanism is that decreased forage intake reduces body condition and either results in elk failing to conceive during the autumn rut or elk losing the fetus during winter. We tested this hypothesis by comparing age-specific body condition (percentage ingesta-free body fat) and pregnancy rates for northern Yellowstone elk, one of the herds sampled by Creel et al., before (1962-1968) and after (2000-2006) wolf restoration using indices developed and calibrated for Rocky Mountain elk. Mean age-adjusted percentage body fat of female elk was similarly high in both periods (9.0%-0.9% pre-wolf; 8.9%-0.8% post-wolf). Estimated pregnancy rates (proportion of females that were pregnant) were 0.91 pre-wolf and 0.87 post-wolf for 4-9 year-old elk (95% CI on difference = -0.15 to 0.03, P = 0.46) and 0.64 pre-wolf and 0.78 post-wolf for elk > 9 years old (95% CI on difference = -0.01 to 0.27, P = 0.06). Thus, there was little evidence in these data to support strong effects of wolf presence on elk pregnancy. We caution that multiple lines of evidence and/or strong validation should be brought to bear before relying on indirect measures of how predators affect pregnancy rates.

Relations between nutritional condition and survival of North American elk Cervus elaphus

Abstract We related annual and seasonal survival of four populations of elk Cervus elaphus in the Pacific Northwest, USA, to measures and indices of individual nutritional condition. Among populations, for all mortality (human and non-human causes) sources inclusive, annual survival of adult females was correlated with a rump body condition score (rs = 0.627, P = 0.071), and survival over spring-summer-autumn (SSA) was correlated with mean ingesta-free body fat (IFBF; rs = 0.567, P = 0.088) and rump body condition score (rBCS; rs = 0.615, P = 0.050). For non-human mortality sources only, survival through SSA was correlated with IFBF (rs = 0.567, P = 0.088) and rBCS (rs = 0.615, P = 0.050), and survival over winter was correlated with withers body condition score (rs = 0.677, P = 0.045). For human-caused mortality sources only, survival over SSA was correlated with rBCS (rs = 0.696, P = 0.036) and IFBF (rs = 0.696, P = 0.036). For individuals, logistic analysis found that individual likelihood of dying from all mortality sources inclusive was best predicted (χ2 = 8.3, P = 0.004, β = −1.24) by longissimus dorsi (loin) muscle thickness, a measure of protein catabolism. For only non-human mortality sources, a model (χ2=16.1, P = 0.0003) containing both loin muscle thickness (χ2 = 5.7, P = 0.017, β = −1.02) and percent ingesta-free body fat (χ2 = 4.9, P = 0.027, β = −0.35) best predicted individual susceptibility to mortality. Odds ratios indicated that odds of dying increased approximately 3X for each centimeter of loin muscle catabolized and 1.4X for each percent less body fat. No condition indices at the individual level were related to survival from human-caused mortality sources. Our study populations were characterized by low-marginal condition (i.e. mean ingesta-free body fat levels of 5.9-12.3% for lactating cows in late autumn); this likely increased the prominence of measures of muscle catabolism relative to fat accretion in influencing individual elk survival. Elk populations throughout the Pacific Northwest likely show similar condition levels, and consequently individuals are predisposed to mortality to a much greater degree than under optimal foraging conditions. Management strategies which assume that nutritional condition affects vulnerability only at or near condition levels associated with ecological carrying capacity (i.e. near starvation mortality) may overestimate the impact of proximate mortality factors on adult female elk.

EVALUATION OF VAGINAL IMPLANT TRANSMITTERS IN ELK (CERVUS ELAPHUS NELSONI)

Abstract The effects of vaginal implant transmitters for tissue damage after 11 wk in 13 captive adult elk (Cervus elaphus nelsoni) and subsequent reproductive performance in 38 free-ranging elk were evaluated. Vaginal implant transmitters are designed to be shed at parturition and are used to locate birth sites of wild ungulates; however, potential adverse effects of these transmitters on tissues associated with the vaginal walls or subsequent reproductive performance have not been assessed. Vaginal implant transmitters consist of a transmitter encased in inert acrylic with an antenna trailing out the distal end and wings at the proximal end to hold the transmitter in place. Using a laparoscope on sedated captive elk, necrosis or measurable differences in tissue trauma between designs with wing spans of 80 versus 150 mm over an 11-wk trial were not observed. After the captive elk trial, vaginal implant transmitters with 80-mm wings were placed into 38 pregnant wild elk, and 31 live births were documented. Fates of seven calves were not determined, because their transmitters were not shed at the birth site. We recaptured 36 of these cow elk again in fall 2003 or spring 2004, and 32 were pregnant. This study was unable to document any short- or long-term effects of vaginal implant transmitters on reproductive performance of cow elk in captive and free-range environments.