Energy, water and space use by free-living red kangaroos Macropus rufus and domestic sheep Ovis aries in an Australian rangeland

Abstract

We used doubly labelled water to measure field metabolic rates (FMR) and water turnover rates (WTR) in one of Australia’s largest native herbivores, the red kangaroo (Macropus rufus) and one of Australia’s dominant livestock species, the wool-breed Merino sheep, under free-living conditions in a typical Australian rangeland. Also, we used GPS technology to examine animal space use, along with the comparisons of urine concentration, diet, diet digestibility, and subsequent grazing pressures. We found smaller space-use patterns than previously reported for kangaroos, which were between 14 and 25 % those of sheep. The FMR of a 25-kg kangaroo was 30 % that of a 45-kg sheep, while WTR was 15 % and both were associated with smaller travel distances, lower salt intakes, and higher urine concentration in kangaroos than sheep. After accounting for differences in dry matter digestibility of food eaten by kangaroos (51 %) and sheep (58 %), the relative grazing pressure of a standard (mature, non-reproductive) 25-kg kangaroo was 35 % that of a 45-kg sheep. Even for animals of the same body mass (35 kg), the relative grazing pressure of the kangaroo was estimated to be only 44 % that of the sheep. After accounting for the energetic costs of wool growth by sheep, the FMRs of our sheep and kangaroos were 2–3 times their expected BMRs, which is typical for mammalian FMR:BMRs generally. Notably, data collected from our free-living animals were practically identical to those from animals confined to a semi-natural enclosure (collected in an earlier study under comparable environmental conditions), supporting the idea that FMRs are relatively constrained within species.

Appendices: FMR, WTR and space use by free-living kangaroos and sheep

Methods for isotope pool sizes and DLW calculations of FMR and WTR

Pool sizes (N _H for hydrogen or N _O for oxygen; moles) were estimated as:

$$ N_{\text{int}} = { (}I_{\text{ds}} --I_{\text{dist}} )\times (M_{\text{dist}} /M_{\text{W}} )\times (M_{\text{inj}} /(I_{\text{int}} - I_{\text{b}} ) )/ 1 8.0 2; $$

(2)

where N _int = initial intercept pool size of the isotope in the animal, I _ds = concentration of isotope in isotopically labelled water solution, I _dist = concentration of isotope in distilled water, M _dist = mass of distilled water, M _W = mass of isotopically labelled water used in dilution, M _inj = mass of isotopically labelled water injected into the animal, I _int = the intercept or equilibration concentration of isotope (¹⁸O or ²H) and I _b = concentration of isotope (¹⁸O or ²H) in background blood sample (Table 10).

We estimated WTR (r _H2O; mol day⁻¹) according to:

$$ r_{\text{H2O}} = { (}k_{\text{H}} *N_{\text{H}} /{\text{ R}} - {\text{displace)/}}\left[ {(f_{ 1} *X) + ( 1--X)} \right]; $$

(3)

where, k _H = ²H flux (turnover rate) during the experiment (Eq. 4), N _H = the isotope pool size calculated from ²H dilution (Appendix S2), R-displace = mean group displacement ratio for body water pools, estimated from initial pool sizes for ²H and ¹⁸O (i.e. N _H:N _O; Midwood et al. 1994; Table 11), f ₁ = fractionation constant for ²H₂O vapour relative to ²H₂O liquid (assumed to be 0.93; Lifson and McClintock 1966; Speakman 1997), X = estimate of the proportion of total water loss that is fractionated (assumed = 0.25; Speakman 1997).

Table 11 Animal parameters and doubly labelled water isotope-kinetics for red kangaroos and sheep

Deuterium flux during the experiment (k _H) was estimated as:

$$ k_{\text{H}} = ({ \ln }H_{\text{int}} --{ \ln }H_{\text{final}} )/t; $$

(4)

where ln = natural log of initial (H _int) and final (H _final) concentrations (ppm) of ²H in body water after correction for background levels, and t = time (days).

Production of CO₂ ($r_{{{\text{co}}_{2} }}$; mol day⁻¹) was estimated as:

$$ r_{\text{CO2}} = ((k_{\text{O}} \times N_{\text{O}} ) - [(r_{\text{H2O}} \times X \times f_{ 2} ) + ( 1- X) \times r_{\text{H2O}} ])/ 2f_{ 3} ; $$

(5)

where k _O = ¹⁸O flux (turnover rate) during the experiment (see Eq. 4, substituting ²H for ¹⁸O), N _O = isotope pool size calculated from ¹⁸O dilution, $r_{{{\text{H}}_{2} {\text{o}}}}$ = as per Eq. 2, X = estimate of the proportion of total water loss that is fractionated (assumed = 0.25; Speakman 1997), f ₂ = fractionation constant for H ¹⁸₂ O vapour relative to H ¹⁸₂ O liquid (assumed to be 0.99; Speakman 1997), and f ₃ = the fractionation constant for H ¹⁸₂ O₂ gas relative to H ¹⁸₂ O liquid (assumed to be 1.039; Lifson and McClintock 1966; Speakman 1997).

Methods for diet analyses

Diet was assessed from these forestomach samples via micro-histological identification of plant fragments (Dawson and Ellis 1994). Fragments were identified as being either grass, flat chenopod, round chenopod, forb, malvaceous sub-shrub or trees. Sub-samples (ca. 50 g) of foregut material were dried at 60 °C to constant weight and then ground through a 1-mm mesh (Glen Creston c.580 micro hammer mill, Glen Creston, London). Ground sub-samples (10 ml) were washed through two sieves yielding particles between 500 and 125 μm and greater than 500 μm. Particles smaller than 125 μm were discarded because they are mostly dust and microhairs. The relative volumes of the two size classes were determined by centrifugation. Five sub-samples of each size class were spread out on separate microscope slides. Random horizontal transects were chosen and the first 20 particles on transects were identified. For each size class 100 particles were examined, i.e., 200 in total per animal. Identification of plant particles was made using an extensive reference collection (see Dawson and Ellis 1994). Total proportion of plant categories in the diet was determined according to the ratio of particle size classes in each sample.

Methods for calibrating GPS collars and for estimating location errors and data screening

Location error of nine of the 11 collars was quantified by placing stationary collars at one of two known GPS locations at our study site. Known GPS-point locations were established using a hand-held GPS personal navigator (Garmin Etrex Vista, HCx model; Garmin International, Kansas, USA), scheduled to average points every second over 24 h (two of the 11 collars were not characterised for location error due to collar failure during a subsequent related study). Collar location error was established at an open-canopy area typical of that ranged by the animals at our study site, and at a second, shaded site beneath trees typical of those used by kangaroos as resting sites during hot daylight hours. At each site (open and shaded) collars were placed 20 cm above the ground to mimic the average height of collars on foraging or resting animals, and were scheduled to log location data every 10 min over 24-h. After retrieving the collars from re-captured animals, the stored data were downloaded to personal computer and examined. We excluded location data that were calculated using ephemeris data from only three satellites (i.e. 2D points; see Lewis et al. 2007), and further screened our data according to altitude, which varied widely and included readings not possible in both positive and negative directions (e.g. collars apparently located tens of meters below ground). Altitude data recorded by each collar (raw data before screening 2D readings) were normally distributed, and therefore location data were further screened by excluding all position fixes greater than one standard deviation from the mean of each collar. The absolute distance-error (±SEM) for all collars after screening was 3.8 ± 0.6 m (open and shaded sites pooled; average SEM was 0.27 ± 0.08 m). Collar error measured as distance from the known point was not significantly different between the open and shaded sites (t _2,8 = −0.212, P = 0.84), likely due to the relatively flat, open terrain of the study site, so that ‘available sky’ (D’Eon and Delparte 2005) was similar at the two sites. However, absolute distance-error may not adequately reflect collar location error (Frair et al. 2010), and so we further defined collar location error as the circular error of probability (CEP; see Lewis et al. 2007; Frair et al. 2010) that included 95 % of all locations, relative to the known point (i.e., open or shaded site; CEP95 %). We calculated CEP95 % distances for each collar using DNRGarmin software (version 5.4.1, Minnesota Department of Natural Resources 2001; projection WGS84, UTM zone 54S). The average (±SEM) location error for collars after screening was 10.4 ± 3.9 m (open) and 10.8 ± 3.1 m (shaded).