Predicting energy expenditures for activities of caribou from heart rates

Highly significant (P<0.001) linear relationships between oxygen comsumption (VO2) and heart rate (HR) were found for six caribou (Rangifer tarandus grand) at several times during the year. The standard error of the estimate for predicting V O 2 from H R was within 10% of the mean V O 2 for 9 of 13 caribou/season combinations. Energy expenditures by caribou while feeding on grain at a trough, grazing, browsing and walking within a large enclosure were 12%, 17%), 18% and 46% higher than the cost of standing. H R ' s recorded during a given activity decreased sharply during September and October, and reached a minimum in January. A n abrupt increase in H R ' s of female caribou occurred 3 weeks prior to parturition. Heart rate telemetry can be used to determine the relative energy expenditures of free-ranging caribou with reasonable accuracy.


Introduction
Measurements of energy expenditures by large animals are frequently made using closelyconfined animals encumbered by equipment that restricts their behavior.Development of the CO2 entry rate technique (Young et al, 1969) and the doubly labeled water method (Lifson et al, 1955) provided the methodology for estimating energy expenditures by free-ranging animals, but these methods are complex and expensive.Heart rate is common used to estimate energy costs by humans because it is easily measured and the measurements do not interfere with the subject's normal acitivites (Bradfield et al., 1969;Acheson et al., 1980).Some investigators working with ruminants have found poor agreement between heart rate and energy expenditure; the relationship is known to be affected by individual differences, season, time since the last feeding, type of activity, ambient temperature, excitement or stress, and other factors (Webster, 1967;Brockway and McEwan, 1969;Johnson and Gessaman, 1973;Hotter et al, 1975Hotter et al, , 1976;;Robbins et al, 1979).However, other researchers, by controlling some of these factors, established relationships which predicted energy expenditure within 10% for individually-calibrated animals (Yamamoto et al, 1979;Pauls et al, 1981;Renecker and Hudson, 1983;Nilssen etal, 1984;Richards and Lawrence, 1984;Fancy and White, 1985a).We investigated the heart rate/energy expenditure relationship for six caribou (Rangifer tarandus grand) throughout the annual cycle and used heart rate to estimate energy expenditures for activities as the caribou ranged within a large enclosure.Energy expenditures of caribou while cratering in snow are reported elsewhere (Fancy and White, 1985a).

Methods
Hand-raised caribou were kept within a 19-ha enclosure at the University of Alaska Large Animal Research Station in Fairbanks, Alaska, where they had ad libitum access to natural forage and a commercial livestock ration (Quality Texture, Fisher Mills, Seattle, WA).At the time of the experiments, the caribou were 16 -35 months of age and weighed between 85 and 123 kg.A heart rate transmitter (J.Stuart Enterprises, Grass Valley, CA; see Follmann et al. (1982) for transmitter design) was implanted subcutaneously on either side of each caribou adjacent and approximately parallel to the sixth rib.The pulsed signals were received by a Telonics (Mesa, AZ) TR-2 receiver and were either plotted on a chart recorder or counted directly.
The relationship between O2 consumption (VO2, l-min" 1 ) and heart rate (HR, beats-mirr 1 ) was determined for each caribou at several times during the annual cycle using an open-circuit respirometry system (Fancy, 1986).VO2 and HR were measured concurrently once the caribou reached a steady state of O2 consumption (i.e., when the rate of O2 consumption was relatively constant) while walking or standing on a treadmdl, or while standing in a respiration chamber.Caribou were taken directly from the enclosure prior to measurements.
Data from trials on the same caribou conducted within a 4 -5 week period were combined to develop regression equations predicting VO2 from HR for that season.These equations were then used to estimate VO2 from HR of each caribou at the same time of year as 2.

1.0-
as- they ranged within the 19-ha enclosure.Energy costs of individual activities of caribou in the enclosure were expressed as a multiple of the predicted energy cost of standing, thereby adjusting for seasonal differences in metabolic rates, time of day and other factors.An energy equivalent of 20.6 kj-l O2 1 (Brody, 1945) was used to convert VO2 to energy expenditure.This value was based on a respiratory quotient of 0.90 determined from air samples taken from caribou while in the enclosure.Eight activity categories were recognized: lying with head on the ground, lying with the head up but not ruminating, lying and ruminating, standing quietly, feeding on grain at a trough, grazing (without walking), browsing, and walking.Mean HR's were calculated from 30 sec counts taken continuously while the caribou was engaged in each activity.Statistical comparisons between regression lines were made by analysis of covariance (Zar, 1973;Neter and Wasserman, 1974).The standard error of the estimate for predicting VO2 from one additional measurement of HR (at the mean HR) was calculated according to Zar (1973) as: (1), where n is the sample size and Sy x 2 is the residual mean square.

Results
Highly significant (P<0.001)linear relationships between VO2 and HR were found for all six caribou (Figs. 1 -3).In 4 of the 13 cases, correlation coefficients increased about 1% when exponential equations were used, but these small improvements did not justify the transformations.Significant differences were found between the slopes of the regression lines (O2 pulse, 1 Û2"beat ') between seasons for individual caribou, and among caribou within a season (Table 1).For example, the O2 pulses of four Significant differences were also found between the VO2/HR relationships calculated for the same individual during a season.Although the O2 pulses of caribou F3 and F4 in November were the same as those calculated 6 months later, the elevations of the two regression lines calculated for caribou F4 were significantly different (Table 1).In contrast, the O2 pulse of caribou F2 in March was significantly greater than that calculated 3 months earlier.Some of these differences can be attributed to differences in the time since feeding, as fasting increases the elevation of the regression and lowers the range of HR's observed for the same activities (Fig. 3).
Heart rates of caribou decreased sharply during September and October, and reached a minimum in January (Fig. 4).Mean HR's of all female caribou increased by approximately 15 beats-min" 1 during the 3 weeks precceding parturition, then decreased sharply within 3 days following parturition (Fig. 4).The calves of caribou Fl and F2 died within 3 days of birth; therefore, these females lactated for only a few days.
The lowest energy expenditures occurred when the caribou were lying with their heads on the ground (Table 2).The expenditure of feeding at a trough was 12% higher than that while Test for equahtv among regressions (Neter and Wasserman, 1974:160) : Test for equalitv of slopes and elevations among regressions (Zar, 1973)  HEART RATE (beats, m i n ' 1 ) Fig. 3. Regressions of O2 consumption on heart rate for caribou Fl in September when ted and following a 15 h fast.Measurements were made as the caribou stood or walked on a treadmill.
females averaged 69% higher in late winter (March -April) than in late summer (September).Four of the five caribou had the same O2 pulse in late summer, but the elevations of the regression lines were significantly different (Table 1).The high O2 pulse of caribou Fl in summer may be a result of pooling data collected throughout the month of September when HR's were decreasing rapidly (Fig. 4).
Tabic 2. Energy costs of activities relative to the cost of standing, as estimated from heart rates of caribou Estimates are bases on regressions developed for each caribou at the same time of year as its application.
2 Activity codes are: LHD = lying with the head down; LHU = lying with the head up; LR = lying and ruminating; FT = feeding at a trough; GRZ = grazing without walking; BRW = browsing; WAK = walking

Discussion
The physiological components involved in the relationships between VO2 and HR can be expressed by the following equation (Morhardt and Morhardt, 1971): where HR is in beats-min 1 , SV is the stroke volume (1 blood-beat" 1 ), and (A-V) O2 diff is the difference between the O2 concentrations of arterial and venous blood (1 Cb-l blood" 1 ).Equation 2 can be rearranged to show that the volume of O2 consumed per beat, or O2 pulse, is equal to the product of SV and (A-V) O2 diff.If the O2 pulse is constant, or varies systematically over a range of HR, errors in the prediction of VO2 from HR will be small.This was the case in summer, when only slight increases in O2 pulse with HR were observed for four of five caribou (Fig. 5); this indicated that variations in O2 requirements within the range standing.The highest energy expenditures involved locomotion; the energy cost of walking was 46% above the cost of standing (Table 2). of HR's observed were met primarily by changes in HR.Changes in stroke volume and (or) the arteriovenous O2 differences were more pronounced in winter (Fig. 5).Because of seasonal differences in O2 pulse, regression equations based on pooled data collected in summer and winter, such as those presented by Pauls et al (1981), may be misleading.Nilssen et al (1984) measured VO2 and HR of Norwegian reindeer (R. t. tarandus) and Svalbard reindeer (R. t. platyrhynch^:) standing or running on a treadmill.The regression equations given by Nilssen et al. (1984) were recalculated for comparison with our results by assuming mean body weights of 89.1 and 76.0 kg for Norwegian and Svalbard reindeer, respectively (K.Nilssen, pers. comm.).The slopes of the regression lines for Norwegian reindeer (0.033 1 Ch'bear 1 ) and Svalbard reindeer (0.030 1 02'bear 1 ) in summer were within the range of those calculated for caribou (Figs. 1 -3), but the intercepts calculated for Norwegian and Svalbard reindeer were lower than those for caribou.Consequently, the same HR would be associated with a lower VO2 for reindeer when compared to caribou.The slope of the winter equation for Norwegian reindeer (0.035 1 02-bear 1 ) was also within the range of those calculated for caribou, but the slope for Svalbard reindeer (0.025 1 Cb-bear 1 ) was lower.Some of the differences noted above may be explained by the higher treadmill speeds used by Nilssen et al. (1984).Results of studies with humans (Rusher, 1965;Davies, 1968) and small mammals (Morhardt and Morhardt, 1971) indicate that as HR approaches a maximum, further increases in VO2 are associated with only small changes in HR.Pauls et al (1981) and Renecker and Hudson (1983) used HR to predict energy costs of activités for wapiti (Cervns elaphus) and moose (Alces alces), respectively.The relationships between metabolic rates and HR they presented were curvilinear, probably because they included data for lying as well as active animals.Studies with dogs and humans have found that the mean 02 pulse of inactive subjects is lower than that for the same individuals during exercise (Warnold and Avidsson Lenner, 1977), and that abrupt decreases in stroke volume can occur as a subject changes from a lying to standing position (Wang et al, 1960;Rushmer, 1965).These findings suggest that when relationships between VO2 and HR are developed using only standing and active animals, as in this study, they will give unreliable estimates of lying costs.Indeed, the predicted 8.7% increase for caribou in the cost of standing relative to the cost of lying and ruminating, is only half of that determined from 22 paired VO2 measurements of caribou lying and standing in a respiration chamber (16%; Fancy, 1986), and is much lower than the incremental cost of standing reported for other wild ruminants (Fancy and White, 1985b).Consequently, the costs of lying were overestimated and the values in Table 2 should not be used to calculate energy budgets.
Because ruminants spend a large portion of the day grazing and browsing, the costs of these activities are a major component of daily energy budgets.The estimated 17% increase in the energy cost of grazing above standing for caribou (Table 2) agrees closely with the mean difference of 16% between VO2 of caribou standing and feeding in a respiration chamber or headstall (n = 4;Fancy, 1986).Movements of the neck and mouth apparently account for most of the incremental cost, as VO2 returned to standing values within 30 sec after the end of the feeding bout.The reported incremental costs of grazing by wapiti (6.0%; Pauls et al, 1981) and moose (4.6%; Renecker and Hudson, 1983) are lower than those for caribou, possibly because standing costs for those two species are higher than those for caribou (Fancy and White, 1985b).The mean energy cost of browsing costs for moose and wapiti were also similar (2 -4% higher) to grazing costs (Pauls et al, 1981;Renecker and Hudson, 1983).
The energy cost of walking, above that required for standing, increases linearly with walking speed for caribou and most other species (Fancy and White, 1985b).The energy cost of walking predicted from HR's of caribou Ml, Fl and F2 was 46% higher than the standing metabolic rate.From VO2 measurements for these three caribou while walking on a treadmill (n=407), we determined that the metabolic rate of these caribou would be 46% above the standing rate when they walked at 3.4 km-h We did not measure walking speeds of caribou in the enclosure, but the energetically-optimal walking speed (Fancy and White, 1985b) for caribou is approximately 3.5 km-h -1 , and a mean walking speed of 3.4 km-h 1 during HR measurements appears reasonable.
The close relationships between VO2 and HR suggest that HR telemetry could be extremely useful in describing the energetics of wild caribou, particularly in summer when variations in O2 requirements are met primarily by changes in HR.The pulsed signals can be received several kilometers from the transmitter; therefore, it would be possible to monitor HR's without influencing the behaviour of the caribou.Reasonably accurate predictions of VO2 from HR could be made using an equation combining data from several caribou.For example, the standard error of the estimate calculated from the combined summer data for caribou M2, F2, F3 and F4 (Y=-0.242+ 0.020 X; r = 0.79; n= 144) is within 10.6% of mean VO2.Changes in stroke volume and O2 extraction from the blood are of greater importance in winter than in summer for meeting varying O2 requirements.Because these components differ substantially between individuals in winter (Fig. 5) it is unlikely that a general regression equation for accurately estimating activity costs from HR in winter could be developed.

Fig. 2 .
Fig. 2. Regressions of O2 consumption on heart rate for caribou in winter.

Fig. 4 .
Fig. 4. Seasonal trends in heart rates of two female caribou feeding on grain at a trough.

Fig. 5 .
Fig. 5. Regressions of O2 pulse on heart rate in summer and winter; values were calculated from data presented in Figures 1-3.

Table 1 .
Summary of analvses of covariance comparing regression lines relating VO2 to heart rates of caribou.