Hypometabolism and basking: the strategies of Alpine ibex to endure harsh over-wintering conditions

Authors


Correspondence author. E-mail: claudio.signer@gmail.com

Summary

1. The extent to which free-ranging large north-temperate mammals seasonally adjust thermoregulation and their energy expenditure under fully natural conditions are unknown.

2. Therefore, using telemetry we measured the heart rate (as a proxy for metabolic rate), rumen temperature (Tr) and locomotor activity (LA) over 2 years for 20 free-ranging Alpine ibex (Capra ibex ibex) living at high altitudes in the Alps.

3. Ibex showed strong seasonal changes in mean daily heart rate with a winter nadir of about 60% below the summer peak. Only 40% of this variation could be attributed to the changes in daily mean Tr, LA, wind chill, body size and snowfall. The unexplained residual variation in heart rate still showed a strong seasonal pattern.

4. The amplitude of daily rhythms in Tr was twice as high during the winter when compared with summer. This was predominantly due to lower daily minimum Tr. Thus, the substantial down-regulation of endogenous heat production during winter – as indicated by heart rate – had surprisingly small effects on Tr, indicating decreased thermal conductance.

5. Rewarming from the daily Tr minimum during the morning hours was independent of heart rate throughout the year, and occurred phase-delayed to the increase in black bulb temperature (BBT). The effects of BBT and LA on the rate of rewarming were maximized within a small range of BBT around 0 °C. This suggests that the ibex moved at sunrise to the closest sunny spot to facilitate extensive basking.

6. The energetic benefits of basking can explain the strong residual seasonality of heart rate in Alpine ibex. This partially ectothermic strategy – together with metabolic depression – apparently enables a thrifty use of body fat reserves, the major metabolic fuel during winter, and thus survival of extremely harsh winter conditions despite the virtual absence of food. Therefore, hypometabolism and passive rewarming by basking may be of general importance as a strategy for non-hibernating mammals to survive winter in strongly seasonal habitats.

Introduction

Mammals in seasonal environments face pronounced changes in climatic conditions and food availability. These changes require modification of several physiological functions, and even morphological adjustments (Lovegrove 2005). A first avenue of seasonal acclimatization that is employed by both small and large mammals is a voluntary reduction of food intake and body size towards winter, which results in reduced overall energy expenditure (Worden & Pekins 1995; Mesteig, Tyler & Blix 2000). Secondly, endotherms may adjust to conditions of cold and scarcity of food by lowering the body temperature setpoint. The most profound type of voluntary hypothermia, i.e. deep hibernation which leads to a tremendous reduction of energy requirements, seems to be largely restricted to smaller mammals (<∼4 kg; Geiser & Ruf 1995). Amongst other reasons, this is likely because only small mammals can retreat to underground burrows and hence avoid the risks associated with immobility at low core body temperature. Large mammals, on the other hand, appear to lower the costs of thermoregulation mainly by decreasing the thermal conductance and hence heat loss to the environment (Lovegrove 2005). This is achieved not only by increasing fur thickness. Recent studies on red deer (Cervus elaphus; Arnold et al. 2004) and Przewalski horses (Equus ferus przewalskii; Arnold, Ruf & Kuntz 2006) revealed that the strongest reduction in energy expenditure occurs during profound cooling of the body periphery, which is accompanied by considerable hypometabolism. Together, reductions in body size and thermoregulatory adjustments, as well as reduced locomotor activity, can lead to significantly lowered energy expenditure during winter, even in large mammals (Nilssen, Sundsfjord & Blix 1984a; Renecker & Hudson 1985a; Arnold et al. 2004; Arnold, Ruf & Kuntz 2006).

However, there are much less data on seasonal acclimatization in large mammals than there are for small species (Lovegrove 2005). Moreover, all the studies on large ungulates mentioned above share the weakness in that they were conducted with animals in enclosures that were, at the very least, supplementary fed. Further, core body temperatures have never been continuously measured over several seasons, and previous statistical analyses have so far failed to adequately explain the overall seasonal variation of metabolic rate in ungulates. Therefore it remains unclear, for instance, to what extent the reduction in energy expenditure during winter can be attributed to thermoregulatory adjustments.

To better understand the extent of, and mechanisms involved in, seasonal acclimatization in free-ranging large mammals, we chose to study Alpine ibex (Capra ibex ibexFig. 1) in their natural habitat. Alpine ibex live above ground throughout the year and typically inhabit altitudes between 1900 and 3000 m a.s.l. (Abderhalden 2005), but as high as 3500 m a.s.l. (Onderscheka & Hartl 1990). Hence, this species faces the most extreme environmental conditions of all the alpine ungulates. Specific adaptations to this harsh, steep and strongly seasonal habitat at high altitudes are a high haematocrit, a compact body shape, a dark and well insulating winter fur, as well as the ability to accumulate large amounts of body fat during summer (Onderscheka & Hartl 1990; Deutz & Gressmann 2001; Meile, Giacometti & Ratti 2003). However, the physiological and behavioural strategies facilitating survival of this ungulate during the challenging alpine winter are unknown. A new telemetry system enabled us to explore these mechanisms for the first time by continuously measuring the heart rate, temperature in the rumen (Tr; a close correlate of core body temperature) and locomotor activity (LA) over a period of about 2 years (Signer et al. 2010). Although we were unable to calibrate heart rate against metabolic rate in the free-ranging animals studied here, we have used heart rate as a proxy for energy expenditure, as these parameters are known to correlate strongly, particularly within distinct activity levels (e.g. Butler et al. 2004).

Figure 1.

 Adult male Alpine ibex basking in the sun. The male in the foreground is equipped with the telemetry system used to measure the heart rate, rumen temperature and locomotor activity. Photo by C. Signer.

Materials and methods

Study area

The study was conducted from June 2007 to June 2009 in the Alpine ibex population of Albris, situated in the border region of south-eastern Switzerland and northern Italy (46°28′N, 9°56′E). The area used by this population covers approximately 1100 km2 with elevations ranging from 400 to 3500 m a.s.l. Altitudes of 1900 to 3000 m a.s.l. are most frequently used by Alpine ibex and animals show a pronounced vertical migration pattern between seasons (Abderhalden 2005). The tree line is located around 2100 m a.s.l. The prevailing climate is characterized by relatively low precipitation (long-term annual average 799 mm, recorded at MeteoSwiss station ‘Pontresina’, 1774 m a.s.l.), low cloud cover, low humidity, low temperatures during winter and relatively high temperatures during the short alpine summer (Gensler 1978). Long-term air temperatures (Ta) average −7·8 °C in January and 8·3 °C in July (recorded at MeteoSwiss station ‘Passo del Bernina’, 2307 m a.s.l.). The Alpine ibex population of Albris was founded, as with most other ibex populations, by re-introduction of individuals bred in enclosures. From 1920 to 1934, a total of 18 males and 27 females were released to establish a population that today consists of about 1100 individuals.

Study animals and telemetric measurements

Between June 2007 and November 2007, we equipped 10 adult males (aged 6–11 years) and 10 adult females (aged 3–14 years) with our telemetry system. Age was determined by counting annually formed horn rings (Ratti & Habermehl 1977). Two females were caught with box traps already located in the study area and handled without immobilisation. All other animals were caught by immobilisation with dart-application of anaesthetic drugs through the use of a gas-driven rifle [Hellabrunner mixture: 125 mg mL−1 Xylazine (Rompun®; Bayer HealthCare, Leverkusen, Germany) and 100 mg mL−1 Ketamine (Ketaminol® 10; Veterinaria AG, Zurich, Switzerland), dosage 0·8–1·5 mL; Abderhalden et al. 1998]. Legs of immobilised animals were tied and eyes masked. Hind leg length (cm) was measured from the peak of the hoof to the end of the heel bone.

The telemetry system used had an expected lifetime of 2 years and consisted of two units: a cylindrical ruminal unit (22 × 80 mm, 100 g), and a collar unit (450 g). The ruminal unit was administered per os after initiating anaesthetic reversal to allow recurrence of a pronounced swallowing reflex that is lacking during deep anaesthesia. The ruminal unit remained in the rumino-reticular tract and hence was located in close proximity to the heart. The unit detected heart beats with an integrated acceleration sensor that responded to the mechanical shock waves above a certain trigger level. Heart rates below 15 or above 120 beats per minute (bpm) were considered to be caused by false triggers and were discarded. These limits were sufficient to include both the lowest heart rate ever occurring and the highest heart rate that could be measured reliably with the system. The ruminal unit also measured Tr with a thermistor calibrated in a water bath at 5 °C-increments between 20 and 40 °C with a resolution of 0·1 °C. We were able to recalibrate thermistors in three devices recovered from animals that died during the study. Deviations from the first calibration were, in all cases, <0·1 °C over the entire measurement range after having been in situ for up to 19 months. A significant drift of measured temperatures due to low battery power could therefore be excluded. To extend battery life, heart rate was measured for a period of 3 min every 21 min only, whereas Tr was measured every 3 min. All heart rate and Tr measurements were transmitted via short-distance UHF link to a data logging system located in the collar unit. The collar unit decoded and processed signals received from the ruminal unit, measured LA at intervals of 3 min with two different activity sensors and stored all data. For a more detailed description of the telemetry system see Signer et al. (2010).

To collect information on spatial use and body condition we regularly observed the animals throughout the study. Body condition was estimated visually by classifying animals into five categories according to distinct anatomical characteristics: (1) pelvic bones sharply protruding from strongly immersed flanks, projecting spinal column forming an acute angle with torso, clear prominence of single ribs and muscles, very thin and wiry neck; (2) pelvic bones moderately protruding from moderately immersed flanks, projecting spinal column forming a right angle with torso, slight prominence of single ribs and muscles, rather thin and wiry neck; (3) pelvic bones slightly protruding from slightly immersed flanks, projecting spinal column forming a slightly obtuse angle with torso, single ribs and muscles only slightly visible, intermediate neck; (4) pelvic bones only slightly visible, no immersed flanks, spinal column only slightly projecting from torso and forming a clearly obtuse angle, no single ribs and muscles visible, strong and round neck; (5) pelvic bones not visible, convex flanks, spinal column does not project from torso, no single ribs and muscles visible, very strong and thick neck indicating considerable amounts of subcutaneous fat deposits. The reliability of our body condition estimate is supported by data on animals culled in the Albris population, which shows that body mass of these animals also peaked in late autumn and decreased to an annual minimum in May (Tataruch & Onderscheka 1996a; Giacometti et al. 1997; cf. Fig. 3).

Figure 3.

 Seasonal variation of residual stationary heart rate independent from extrinsic and intrinsic parameters, of snow height (reflecting accessibility of plant forage), of body condition and of fur insulation. Solid black line: residual stationary heart rate predicted from a spline fit to Julian day in the generalized additive mixed model (Table 1) ± 95% confidence intervals (CI) for this prediction (dashed lines). Solid grey line: snow height at 2450 m a.s.l. (2-year average over data shown in Fig. 2a). Dotted line: body condition predicted from a cosinor fit to body condition estimates (condition scores from 1 ‘very poor’ to 5 ‘very good’; significance of seasonal variation F2,171 = 523·0, P < 0·001). Bottom bar: periods when the animals were in summer and winter fur and periods of moult (grey; after Meile, Giacometti & Ratti 2003).

Two of our study animals died before the end of the study in avalanches, three due to winter starvation and one animal was subjected to hunting by the Hunting and Fishing Services of Grisons because of a technical failure in the telemetry system. The first death occurred in January 2008, the last in March 2009. The remaining 14 individuals were recaptured between April and June 2009 to retrieve the collar units.

Weather data

Weather data were obtained from two stations located within the study area. One was operated by the Swiss Institute for Snow and Avalanche Research and is situated at the midrange altitude of the habitat used by Alpine ibex throughout the year (SLF; ‘Bernina 2’ at 2450 m a.s.l.). This station recorded Ta (°C) and wind speed (m s−1) every 30 min. Total snow height (cm) and a modelled estimate of actual snowfall (mm) within every measurement interval were also recorded at this station (Lehning et al. 1999). The second station, located at 1800 m a.s.l. close to the bottom of the south-westerly oriented slope most frequently used by our study animals during winter (slope range 1750–3150 m a.s.l.), was set up by ourselves and recorded black bulb temperature (BBT) every 30 min from June 2008 to June 2009. Due to a technical failure, BBT from a midrange station (2100 m a.s.l.) were unavailable for most of the study period (but those BBT obtained were highly correlated with the station at 1800 m a.s.l.). BBT provides an integrated measurement of Ta and solar radiation and is a standard method for determining environmental temperature as effectively perceived by animals (Bakken 1976; Hill et al. 2004). BBT was recorded with a temperature logger (iButton DS1922L; Maxim Integrated Products, Sunnyvale, CA, USA) inserted into a black-painted copper bulb (ø 16 cm) mounted 1·2 m above ground.

Data analyses

Heart rate data measured by the acceleration sensor in the ruminal unit contained a considerable amount of noise caused by reticulum contractions and animal movements. We removed this noise with a filtering procedure described in detail by Signer et al. (2010). As a result, the remaining heart rate data originate from animals at rest or moving only slowly and are therefore referred to as ‘stationary heart rate’ (HRs) (Signer et al. 2010). To visualize daily and seasonal patterns in double plots, we used these almost continuous HRs data. However, before calculating individual hourly and daily means for more detailed analyses, we further improved data quality. All HRs values remaining after the first filter were higher than 25 bpm. In a second pass, we accepted only HRs measurements with at least 75 heart beat recordings during a 3-min measurement interval. This second filter pass removed all HRs data that were potentially unreliable due to a low proportion of accepted values.

LA estimates were improved by combining the outputs from both activity sensors in the collar with an algorithm classifying each 3-min interval as predominantly ‘active’ or ‘at rest’ with an accuracy of 87% (Signer et al. 2010).

Measurements of Tr were contaminated by non-physiological temporal declines due to water consumption and ingestion of cold food (Dale, Stewart & Brody 1954; Cunningham, Martz & Merilan 1964; Brod, Bolsen & Brent 1982; Crater & Barboza 2007). To remove these data points, we visually checked the Tr raw data and removed all instances of sudden changes in Tr that could not be explained by physiological processes. After this elimination, remaining Tr values were in the range from 35·6 to 41·8 °C.

We ignored data from the first 3 days after capture and from days when we unsuccessfully tried to recapture an animal. Physiological parameters measured in animals dying from starvation began to differ from those of surviving animals on approximately day 20 prior to death. Thereafter, the three animals reacted differently to the imminent death: one animal remarkably increased HRs, while HRs of the other two animals fluctuated around the values measured in surviving animals. However, to remove any possible influence of organ failure occurring close to death, we ignored the last 28 days of life in starving animals. The remaining data set comprised a total of 7642 individual days of LA data, 7014 days of Tr data and 5778 days of HRs data.

Statistical analyses

All statistical analyses were performed with the software package R (R Development Core Team 2010). We used linear mixed effect models (function ‘lme’) for most analyses and controlled for repeated measurements by entering ‘individual’ as a random factor. We accounted for the well-known relation between body size and resting heart rate (Heusner 1991; Noujaim et al. 2004) by including ‘hind leg length’ as a covariate in our statistical models. In Alpine ibex, sexes and age classes differ considerably in body size (Giacometti et al. 1997) and these differences correlate well with hind leg length (Giacometti et al. 1997; Buchli & Abderhalden 1998). We did not account for reproductive effects, as the reproductive state of females did not significantly contribute to our models (C. Signer, T. Ruf & W. Arnold, unpublished data). Furthermore, we did not test for interactive effects of linear predictors, except when such interaction was to be expected a priori. This was the case for the climatic factors Ta and ‘wind speed’. Their combined effect is termed ‘wind chill’ in the following. Preliminary analysis of seasonal changes in HRs revealed a strong non-random pattern in the residual variation remaining after modelling linear predictors and wind chill. Therefore, we analysed the seasonal changes in HRs with a generalized additive mixed model (function ‘gamm’ with family ‘gaussian’; Wood 2006) and modelled the remaining seasonal variation, after adjusting for all other predictors, with a spline fit to the day of the year (‘Julian day’). Like linear mixed effect models, generalized additive mixed models also account for varying samples sizes but still allow controlling for repeated measurements.

To analyse the relative increase in Tr during the morning rewarming phase, we arcsine-transformed these percentages to achieve a normal distribution of model residuals. The nonlinear effects of BBT on LA were analysed by generalized additive models (function ‘gam’). To differentiate between the main seasons we did not use meteorological definitions, but rather used the known timing of ibex’ fur coating (Meile, Giacometti & Ratti 2003) as a biologically more meaningful definition of summer (July–September) and winter (December–March). Inflection points in the daily course of hourly mean HRs, Tr, LA and BBT, and their confidence intervals (CI), were determined with segmented regression, i.e. by fitting piecewise linear regression lines with significantly different slopes to subsequent parts of a time series (package ‘segmented’; Muggeo 2008).

Significance of seasonal variation in body condition scores was tested by cosinor analysis, i.e. by entering a sine (t) and cosine (t) term in a linear mixed effect model, with t representing day of the year in radians. Sums of squares and d.f. of these terms were added to obtain a single F and P value for the periodic function.

Results

Seasonal changes of climate and physiology

During the study period, our study animals were exposed to enormous seasonal and inter-annual fluctuations in climatic conditions (Fig. 2a). Daily mean Ta dropped from a summer maximum of 13·7 °C to a winter minimum of −17·4 °C. The hourly means of BBT (which provides a better estimate of temperatures experienced by the animals) ranged from −23·4 to 41·7 °C during February, a typical winter month, and from −0·4 to 41·4 °C during August, a typical summer month. Due to continuously low Ta and higher snow fall, climatic conditions were harsher during the second winter of the study. Snow covered the ground continuously from late October to the end of June (first winter: mid November to mid June) and accumulated to a maximum height of 380 cm at 2450 m a.s.l. (first winter: 303 cm; Fig. 2a).

Figure 2.

 Seasonal variation of environmental conditions and physiological parameters. (a) Daily mean air temperature ± daily maximum and minimum and total snow height measured at 2450 m a.s.l.; (b) Daily mean stationary heart rate, rumen temperature and locomotor activity. Error bars (SE) represent variation between individuals. No error bars indicate SE smaller than symbol size or measurements of single individuals, indicated with white dots.

Seasonal and inter-annual changes of climatic conditions were reflected by physiological and behavioural reactions of the study animals. HRs peaked in June with values around 100 bpm and decreased by approximately 60% to a winter nadir (Fig. 2b). The annual course of Tr and LA were similar to HRs but had lower amplitudes. Analysis of potential determinants of HRs revealed a moderate influence of body size and the extrinsic factors snowfall and wind chill (Table 1). Unfavourable climatic conditions such as intensive snowfall and severe wind chill surprisingly caused a decrease in HRs. However, as indicated by F-values in Table 1, the intrinsic factors Tr and LA explained about nine times more of the variation in daily mean HRs than the extrinsic factors. Daily mean Tr had by far the strongest influence although its maximum fluctuation throughout the year reached only 1·2 °C, or 3% of the annual mean. In contrast, LA during winter was reduced to almost half the summer level (Fig. 2b), but compared with Tr, LA explained less than half the variation in daily mean HRs (Table 1).

Table 1.   Result of generalized additive mixed modelling of daily mean stationary heart rate
PredictorFPβ*
  1. β, Standardized regression coefficient.

  2. *Available for linear predictors only.

  3. †Termed as ‘wind chill’ in the text.

  4. ‡Spline fit independent of other predictors in the model (Fig. 3).

Rumen temperature290·78<0·0010·52
Locomotor activity119·41<0·0010·26
Hind leg length10·080·002−0·59
Snowfall5·730·017−0·04
Air temperature4·480·034−0·03
Wind speed17·20<0·001−0·01
Air temperature × Wind speed†19·77<0·001−0·08
Julian day‡751·80<0·001

Although their statistically highly significant contribution, both extrinsic and intrinsic factors together explained only 38% of the annual variation in HRs. Moreover, the residual variation in daily mean HRs, independent from other predictors in the model, still contained seasonality, as indicated by the significant nonlinear term in the model. In fact, this term explained most of the annual variation in HRs (Table 1). It revealed substantial changes during short spring and autumn periods, leading to an almost rectangular pattern with a considerably higher level during the early summer months. The rapid transition from winter levels to the summer plateau preceded the period of abundant food availability and fattening that took place when the animals were in their summer fur (Fig. 3).

Seasonal patterns of daily rhythms

To better understand the strong seasonal variation of HRs independent from the extrinsic and intrinsic factors measured, we analysed daily rhythms of physiological parameters and environmental conditions and their change with the season. Throughout the year, HRs and LA peaked during day light hours and were reduced during the night. Tr usually peaked between late afternoon and dusk and reached the daily nadir around sunrise (Fig. 4). Comparison of hourly means measured during the typical summer month of August with those from February, a typical winter month, showed higher daily amplitude of HRs during summer (13·0 bpm vs. 10·7 bpm or 14% vs. 23% of the monthly mean; F1,883 = 164·5, P < 0·001; Fig. 5), on top of the profound seasonal differences between levels of daily mean HRs. Interestingly, HRs gave no indication of any thermoregulatory costs, even as BBT decreased below −10 °C in February (Fig. 5). Instead, the daily variation in HRs closely reflected that of Tr, with HRs increasing significantly prior to the increase in Tr during both summer and winter. However, in striking contrast to HRs, daily peak Tr was only slightly lower during February than August (mean difference 0·32 °C; F1,1134 = 536·4, P < 0·001), although the minima differed more substantially (mean difference 0·87 °C; F1,1134 = 2473·4, P < 0·001; Fig. 5). As a consequence, the daily fluctuation in Tr during February (1·2 °C) was twice as high as that during August (0·6 °C; F1,1134 = 668·4, P < 0·001).

Figure 4.

 Double plots of daily rhythms of (a) stationary heart rate, (b) rumen temperature and (c) locomotor activity of a representative female Alpine ibex (ID23). Each horizontal line represents 24 h of data from a first day, followed by 24 h of data from the consecutive day. The latter is repeated on the left side of the next line. Grey scales indicate 10 equally spaced categories of a parameter’s value range with darker shades referring to higher values. Solid vertical lines indicate dawn and dusk according to civil twilight.

Figure 5.

  Hourly means of stationary heart rate, rumen temperature, locomotor activity and black bulb temperature (BBT) during February (white symbols) and August (black symbols). Data are double plots for better visualization of daily rhythms. Error bars (SE) represent variation between individuals, no error bars indicate SE smaller than symbol size. Triangles at the bottom of each graph indicate the time of morning inflection of each daily course with 95% CI (horizontal lines around symbol, not visible if smaller than symbol size), determined with segmented regressions. The dotted vertical line indicates the morning inflection time of BBT and reflects local sunrise in the ibex’s preferred habitat. To aid comparison of phase relations with BBT, we phase-shifted the time axis for February to the left (grey x-axis) to make local sunrise the common reference time. The solid grey line represents the time on February mornings when BBT was between −3·6 and 0·8 °C (cf. Fig. 7).

The most pronounced seasonal change in daily rhythms occurred in LA, which gradually changed from the summer pattern, with two distinct peaks at dawn and dusk, to a clearly unimodal winter pattern with a single afternoon peak (Figs 4c and 5). As expected, daily patterns of LA only poorly matched HRs because our telemetry system predominantly registered heart rate at rest. However, the even poorer match between daily patterns of LA and Tr was surprising and indicated considerably different reactions during summer and winter, presumably due to differences in the thermal environment.

The influence of the thermal environment on behaviour and physiology

We explored the influence of the thermal environment on behaviour and physiology with a detailed analysis of their relation with BBT. During both February and August, LA began to increase from its late night nadir before the daily rise of BBT began, i.e. before local sunrise (Fig. 5). However, during August the increase of LA was much steeper and reached the morning peak already around sunrise, while BBT was still low. Thereafter, LA decreased to a much lower level, while BBT climbed to its high noontime plateau. In the afternoon, LA increased again towards a second peak at dusk. In sharp contrast to August, the morning increase of LA during February was much slower than that of BBT but reached its single late afternoon peak at about the same ‘black bulb time’ during August. In line with these results, the morning increase in Tr began before sunrise in August but after sunrise in February, and, independent of season, daily peak Tr coincided with the dusk peak of LA.

Generalized additive modelling of the relation between LA and BBT confirmed the expected differences in summer and winter animals. LA of animals in summer peaked at BBT around 16 °C, but in winter at around 7 °C BBT (Fig. 6). Surprisingly, summer and winter animals reacted differently at identical low BBT. At BBT below 0 °C, LA of summer animals was reduced but did not drop below 40%, whereas LA of winter animals declined to levels as low as 10%.

Figure 6.

 The influence of thermal environment on locomotor activity (LA) in animals in winter (December–March, grey) and summer (July–September, black) fur (Meile, Giacometti & Ratti 2003). Solid lines are predicted values from spline fits of hourly means of LA to corresponding hourly means of black bulb temperature. Dashed lines indicate the 95% CI of these fits.

The phase-relation of the daily patterns in Tr and BBT suggested that the animals may have used basking, i.e. solar radiation to facilitate the daily rewarming phase (Fig. 5). To test this hypothesis, we investigated the speed of changes in Tr as a function of BBT. Indeed, the relative increase in Tr between consecutive morning hours was affected by BBT both during February and August (Fig. 7). However, in February this relation was clearly nonlinear. A segmented regression analysis identified two inflection points at −3·6 °C (95% CI ± 2·1 °C) and at 0·8 °C (CI ± 2·8 °C). Between these inflection points, the regression slope was about 17-fold higher than the slopes below or above (Fig. 7; difference between slopes of regression F2,3062 = 22·8, P < 0·001). Outside the inflection points, slopes of regression lines neither differed from each other nor from the corresponding regression line in August (F2,7503 = 0·3, P = 0·775).

Figure 7.

 Relative increase in mean rumen temperature from 1 h to the next during the morning rewarming phase in February (white dots, solid lines) and August (black dots, dashed lines) and the corresponding hourly means of black bulb temperature (BBT). Plotted symbols are means (±SE) calculated for 1 °C-categories of BBT.

Modelling the influence of all measured potentially thermogenic factors during the morning rewarming phase confirmed the influence of basking and its particular importance during winter (Table 2). HRs had no independent effect, during both February and August. Tr increased faster with increasing LA during August, but during February this occurred only in the BBT range between −3·6 and 0·8 °C. Interestingly, this range of BBT coincided exactly with the initial phase of rewarming after local sunrise (Fig. 5, solid grey line), and within this range both BBT and LA had by far the largest influence on the increase in Tr. Expectedly, animals with larger body mass rewarmed more slowly but this effect reached statistical significance only at BBT below −3·6 °C.

Table 2.   The independent effect of potentially thermogenic factors on the relative increase in mean rumen temperature from 1 h to the next during the morning rewarming phase in a typical winter month (February) and a typical summer month (August). The analysis for February was split according to the inflection points identified by segmented regression (Fig. 7)
PredictorFebruaryAugust
BBT < −3·6 °C−3·6 °C ≤ BBT ≤ 0·8 °C0·8 °C < BBT
d.f.FPβd.f.FPβd.f.FPβd.f.FPβ
  1. BBT, black bulb temperature.

Heart rate1,1431·330·2510·051,400·850·3620·001,330·380·542−0·471,9500·630·429−0·13
Locomotor activity1,1432·940·0890·541,4019·22<0·0011·321,333·720·0620·721,95034·10<0·0010·28
Black bulb temperature1,1430·980·3290·281,404·580·0393·091,3323·77<0·0010·991,95079·14<0·0010·33
Hind leg length1,117·180·021−0·231,91·140·313−0·191,50·810·408−0·221,140·050·821−0·01

Discussion

Down-regulation of metabolic rate during winter

This first detailed field study of physiological and behavioural responses that enable large ungulates to survive harsh winters has confirmed results from previous studies of captive animals (Arnold et al. 2004; Arnold, Ruf & Kuntz 2006). The energy expenditure of our study animals during winter was less than half that during summer. This conclusion rests on the significance of relative changes of HRs. Heart rate and metabolic rate are strongly correlated, as shown by a number of studies on ungulates (Nilssen et al. 1984b; Hudson & Christopherson 1985; Renecker & Hudson 1985b; Brosh et al. 1998), birds (Bevan et al. 1995; Butler et al. 2004; Green & Frappell 2007) and kangaroos (McCarron et al. 2001). These close correlations were found within distinct activity levels. Since our telemetry system provided a reliable measurement of heart rate only for animals at rest or with little activity (Signer et al. 2010), and because our statistical analyses adjusted for differences in mean heart-rate levels between individuals, the use of HRs as a proxy for resting metabolic rate seems justified.

Considerably lower energy requirements during winter, and thus a reduction of voluntary food intake, are widespread amongst northern ungulates (reviewed in Arnold et al. 2004). In red deer, a down-regulation of metabolism accompanied by massive peripheral cooling has been identified as a major mechanism explaining this phenomenon. The reduction in endogenous heat production during winter was the most important determinant of the seasonal decline in metabolic rate (Arnold et al. 2004). Our results demonstrate that free-ranging ibex use the very same mechanism to cope with the lack of adequate food supplies and with cold conditions lasting more than half the year in a high-altitude habitat. The reduction in daily mean HRs during winter was accompanied by lower levels of daily mean Tr and, in particular, by a strong reduction of Tr during the night (Fig. 5).

The influence of extrinsic and intrinsic factors on daily mean HRs was surprising. Unfavourable climatic conditions such as snowfall and severe wind chill caused a decrease in HRs, indicating a reduction in energy expenditure instead of an increase, which is expected if the lower critical threshold for thermoregulation had been surpassed. However, the extrinsic effects were rather weak, presumably because of low thermal conductance achieved by both peripheral vasoconstriction and the well insulating, thick, dark-coloured winter fur. In contrast, the effect of reduced LA during winter on HRs was comparably high, although still far below that of reduced Tr. While daily mean Tr during winter was only slightly lower than during summer, it explained a large amount of the annual variation in HRs. Thus, a considerable fraction of the seasonal decline in HRs– and hence energy expenditure – was certainly due to reduced endogenous heat production. Nevertheless, all extrinsic and intrinsic factors together explained only about 40% of the seasonal variation in HRs (Fig. 3, Table 1). This result is, again, in line with previous observations in red deer (Arnold et al. 2004) and requires explanation.

At first glance, the summer plateau of HRs coincided with the vegetation period (Fig. 3). Hence, heat increment of feeding, i.e. the elevation of metabolic rate and body temperature caused by food ingestion and digestion (Blaxter 1989; Hindle et al. 2003), seemed to be a prime candidate to explain the variation in HRs. In our study area, the rapid growth of fresh plant forage began immediately after snowmelt (April–June, earlier at lower altitudes) providing an abundant and protein-rich source of food. Previous analyses of rumen contents of Alpine ibex from the Albris population revealed that the proportions of fresh plant forage, crude protein and crude fat indeed peaked between mid June and mid September (Klansek, Vavra & Onderscheka 1995; Tataruch & Onderscheka 1996b), at about the same time as the nonlinear fit of HRs (Fig. 3). However, the available data do not allow us to discriminate whether the increase in HRs to high summer levels was due to the heat increment of feeding, or due to the growth and maintenance of the gut and visceral organs in preparation for processing large amounts of food. Such changes in the gut are common in animals inhabiting seasonal environments (Piersma & Lindström 2000; Kamler 2001; Piersma & Drent 2003; Ostrowski, Mesochina & Williams 2006). In another large mammal, the Przewalski horse, it was shown that the increase of heart rate in spring significantly precedes the increase in heat increment of feeding, thereby excluding it as the primary cause (Arnold, Ruf & Kuntz 2006). A similar scenario in Alpine ibex is suggested by the phase relations between the cyclic variations plotted in Fig. 3. The seasonal pattern of body condition scores clearly lagged that of HRs. In other words, the highest rate of fattening occurred at a time when the metabolic rate had already decreased substantially. Active seasonal up- and down-regulation of metabolic rate, independent from heat increment of feeding, is further supported by histological data on thyroid activity from ibex culled in the Albris population (Tataruch & Onderscheka 1996a).

Taken together, the available evidence suggests that HRs is influenced by variation in extrinsic and intrinsic factors, by heat increment of feeding (which may contribute to the seasonal cycle of Tr), as well as by energy expenditure for growth and maintenance of visceral organs necessary for fattening. However, these factors alone are unlikely to explain the full range of the residual annual change in HRs.

Energetic benefits of basking

A potentially important contribution to the ibex’s energy budget that has not been considered to date, is basking in the sun. The active use of exogenous heat for thermoregulation is a well-known strategy of many small heterothermic mammals and birds (e.g. Körtner, Brigham & Geiser 2000; Geiser, Goodship & Pavey 2002; Mzilikazi, Lovegrove & Ribble 2002; Dausmann et al. 2004; Geiser & Pavey 2007; Lovegrove & Genin 2008; Turbill & Geiser 2008). It significantly reduces the amount of endogenous heat production required for rewarming from torpor (Lovegrove, Körtner & Geiser 1999; Geiser & Drury 2003) or for maintaining constant body temperature in cold-exposed animals (e.g. Brown & Downs 2007). Our results reveal for the first time a central role of radiant heat-assisted rewarming in the overwintering strategy of a large mammal (Fig. 5, Table 2). During winter, rewarming the body from the nightly nadir of Tr was strongly associated with the simultaneous increase in BBT. Our study animals apparently took advantage of the thermal conditions that accompanied the ascending sun. As a result, the increase in Tr towards levels enabling high LA could be accomplished without increasing endogenous heat production, as indicated by the absence of any influence of HRs. This effect was amplified on winter mornings during the initial phase of rewarming when Tr was at its lowest level throughout the year. This suggests quick and substantial vasodilatation in the body’s periphery upon exposure to solar radiation. Simultaneously, the effect of LA on the rate of rewarming was also highest immediately after sunrise, i.e. at the inflection point of BBT (Table 2), although, in the range of BBT around 0 °C, the total LA was much lower during winter compared with summer (Fig. 6). The extraordinarily high thermogenic effect of comparably low LA indicates that at sunrise the animals moved only short distances towards the next emerging sunny spot.

Together, the daily patterns of LA and Tr suggest that Alpine ibex take full advantage of the heterogeneity of thermal properties in the highly structured geomorphology of alpine environments. Rocks, ridges and orientation of slopes create a rapidly changing pattern of shade and incident solar radiation. Thus, it seems likely that during the night the animals sought shelter from wind and cold – e.g. in crevices, caves or other lee-providing structures – and left these places in the morning to expose themselves to the first shafts of sunlight. Together with substantially increased insulation, this strategy allowed the animals to significantly down-regulate metabolic rate during winter while avoiding a drastic reduction in body temperature. For a ruminant, a large drop in body temperature would most likely be an unacceptable risk for two reasons. First, a severe impairment of manoeuvrability due to low body temperatures exposes a non-burrowing animal to a high risk of predation. However, this problem seems to be mitigated by the ibex’s strategy of retreating into steep rock walls that are inaccessible to terrestrial predators (Kohlmann, Müller & Alkon 1996; Toïgo 1999). Second, and presumably more important, experimental evidence from cattle and sheep shows that lowered Tr significantly impairs fermentation (Gengler et al. 1970), disturbs the ruminal micro-fauna (Eadie & Oxford 1955) and leads to increased food requirements (Bhattacharya & Warner 1968). Therefore, Tr must not fall drastically over long periods of time to guarantee the animal’s survival.

Conclusion

Many large, north-temperate mammals survive long, cold and snowy winters by fuelling metabolism, to a large degree, from body reserves accumulated during summer (Reimers & Ringberg 1983; Tyler & Blix 1990; Adamczewski, Flood & Gunn 1997; Parker, Barboza & Gillingham 2009). Small hibernating mammals use this strategy combined with a substantially lowered endogenous heat production, rendering them completely independent from food supplies during winter. Hibernators lose about one-third of their body mass during this period of fasting (Watts & Jonkel 1988; Arnold 1993; Hilderbrand et al. 2000). The body mass loss of species like Alpine ibex (Tataruch & Onderscheka 1996a; Giacometti et al. 1997), chamois (Rupicapra rupicapra; H. Schaschl, F. Suchentrunk, D. L. Morris, B. S. Hichem, S. Smith & W. Arnold, unpublished), reindeer (Rangifer tarandus; Reimers & Ringberg 1983) or muskoxen (Ovibos moschatus; Adamczewski, Flood & Gunn 1997) is of similar magnitude. However, it is extremely unlikely that food availability during winter in the habitat of these species is sufficient to compensate for the lack of hibernation. Our finding that ibex substantially down-regulate endogenous heat production without excessive cooling by ‘hitch-hiking’ on exogenous heat supplies solves this enigma. While basking is a well-known mechanism by which hibernators and daily heterotherms facilitate arousal from torpor, its importance for large mammals seems underappreciated. It may well be a common reptilian heritage of mammals in general, enabling a thrifty use of body reserves and survival during periods of negative energy budgets.

Acknowledgements

We thank D. Godli, T. Wehrli, G. Brosi, H. Jenny and others for assisting field work; G. Fluch, F. Schober and T. Paumann for technical support; R. Drury for proofreading; the Swiss Federal Office for the Environment for a stipend to C. Signer; C. Marty from the Swiss Institute for Snow and Avalanche Research for weather data and ‘Skywards’, Heli Bernina, the Swiss National Park, the community of Pontresina and the Bergbahnen Engadin St. Moritz AG for logistical support. All experiments were approved by the animal welfare and ethics act of the Canton of Grisons no. 24/2006 and followed the guiding Swiss principles for research involving animals and human beings.

Ancillary