Effects of body size on the diurnal activity budgets of African browsing ruminants

Transcription

1 Oecologia (2005) 143: DOI /s BEHAVIOURAL ECOLOGY J. T. du Toit Æ C. A. Yetman Effects of body size on the diurnal activity budgets of African browsing ruminants Received: 8 September 2004 / Accepted: 17 November 2004 / Published online: 17 December 2004 Ó Springer-Verlag 2004 Abstract We compared the diurnal activity budgets of four syntopic species of African browsing ruminant that differ widely in body size. These were concurrently studied through all phases of the seasonal cycle, in the same area, using the same methods. We tested five predictions from the literature on how body size is expected to influence the behaviour of tropical ungulates: the smallest members of the browsing ruminant guild exhibit (1) the lowest allocation of diurnal time to activity; (2) the greatest hour-to-hour variation in activity and resting time; (3) the greatest reduction in activity time during the hottest days; (4) the least change between wet and the dry seasons in the ratio of feeding: ruminating time; and (5) the greatest time budget allocation to vigilance. Prediction 1 was supported in that the smaller species spent less time being active during the day. Prediction 2 was also supported in that the smaller species were more variable in their relative allocations of time to activity and resting through successive hours of the day. Contrary to Prediction 3, however, the greatest reduction in activity with increasing temperature was found for the largest guild member. The smaller species can achieve their daily food intake requirements by feeding at night and in the cool hours of the day, while the larger species have to feed during all hours of the day and are thus more susceptible to thermoregulatory constraints on foraging. Prediction 4 was partially upheld in that the largest species (giraffe) displayed the widest variation in feeding: ruminating time through the seasonal cycle. Prediction 5 was not supported, indicating that multiple factors interact with body size in determining vigilance behaviour. J. T. du Toit (&) Æ C. A. Yetman Mammal Research Institute, Department of Zoology and Entomology, University of Pretoria, Pretoria, 0002, South Africa jtdutoit@zoology.up.ac.za Tel.: Fax: Keywords Ungulate activity Æ Scaling Æ Foraging patterns Æ Savanna ungulates Æ Kruger National Park Introduction Many physiological and ecological attributes of animals scale significantly with body mass, and since these attributes influence behaviour, it is to be expected that behavioural patterns should also scale in some ways with animal size (Peters 1983). In this paper we use detailed records of field observations to contribute to the debate on how ruminant activity patterns vary with body size (Owen-Smith 1988, 1992; Mysterud 1998; Pe rez-barberı a and Gordon 1999). A basic premise is that if all species in a mammalian herbivore assemblage use the same quality food and ingest it at roughly the same rate, then daily time spent feeding and walking in search of food is expected to increase with body mass due to increased metabolic requirements (Peters 1983; Calder 1984; Schmidt-Nielsen 1984; Hudson 1985). Larger herbivores can, however, ingest food at faster rates, and due to the Jarman Bell Principle they have lower massspecific metabolic demands, enabling them to tolerate a wider range of diet qualities than smaller herbivores (Bell 1971; Geist 1974; Jarman 1974). This is because basal metabolic rate scales allometrically with body mass (a M 0.75 ; Kleiber 1961) while gut capacity scales isometrically (a M 1 ; Demment and van Soest 1985). As a result, larger herbivores may be able to reduce their foraging time, despite having greater absolute intake requirements, by increasing their intake rate and reducing their food quality. Indeed, for north-temperate grazing ruminants ranging from the 46 kg pronghorn (Antilocapra americana) to the 636 kg bison (Bison bison), our analysis of data from Belovsky and Slade (1986) indicates that diurnal feeding time is negatively correlated with body mass (r s = 0.97, P<0.001). The same trend has been inferred from the activity time of temperate ruminants, across feeding types (Mysterud 1998; Pe rez-

2 318 Barberı a and Gordon 1999). On the other hand, Owen- Smith (1988, 1992) found a positive relationship between daylight foraging time and body mass for African grazing and browsing ruminants. A complication in these meta-analytical reviews is that feeding (plucking and ingesting food), foraging (feeding and moving) and active (non-lying) time allocations are often conflated. In our paper we hope to bring some clarity by using detailed diurnal time budgets compiled from focal animal observations, thereby comparing time allocations across species for each independent category of behaviour. In addition to influencing feeding behaviour, body size also has implications for how animals respond to changing environmental conditions. Among ungulates, the smallest members of a trophic guild would be expected to make the greatest relative adjustments to their diurnal activity budgets in response to hour-to-hour variations in ambient temperature through the day, and day-to-day variations in maximum ambient temperature through the seasonal cycle. Smaller animals have greater surface : volume ratios and, therefore, higher rates of thermal conductance and lower thermal inertia compared to larger animals (Peters 1983; Calder 1984; Schmidt-Nielsen 1984; Hudson 1985; Haim and Skinner 1991). Under mild ambient temperatures they can dissipate heat relatively quickly, but when ambient temperatures are equal to or greater than normal body temperature (common in tropical ecosystems) they can quickly reach their limits of thermal tolerance. Small ungulates in hot environments are therefore expected to allocate most of their active time to the cooler hours of the day, and several studies support this (e.g. Tinley 1969; Tilson 1980; Norton 1981; Cloete and Kok 1986; du Toit 1993). Furthermore, despite textbook examples (e.g. Feldhammer et al. 1999), there is only limited evidence among ungulates for adaptive heterothermy (allowing core body temperature to rise while restricting evaporative cooling to afferent blood to the brain only) and it seems that ungulates primarily employ behavioural thermoregulation to avoid heat stress (reviewed by Mitchell et al. 2002). Indeed, both small and large ungulates have been found to significantly reduce their overall daytime activity during periods of high ambient temperature (e.g. Jarman and Jarman 1973; Leuthold and Leuthold 1978; Belovsky and Slade 1986; Klein and Fairall, 1986; Owen-Smith 1998), although the relative extent of this adjustment has not previously been examined across a range of syntopic species that differ widely in body size. Ruminants are able to offset reduced food quality to some extent by increased oral processing, to reduce particle size and thereby improve both digestion and passage rate. Due to the allometry of digestive efficiency (Demment and van Soest 1985), the smaller members of herbivore guilds should make the greatest investments in oral processing when diet quality declines (Gross et al. 1995). The larger members have wider dietary tolerances and can offset mastication to some extent by the extended retention time of ingesta, with thereby improved cellulolysis, in their more voluminous rumens (Hofmann 1989). In African savannas, however, the woody browse resource is highly spinescent and tree leaves in the dry season have to be plucked from lignified twigs, reducing bite sizes and extending feeding bouts for the larger browsers in particular (Cooper and Owen- Smith 1986). Hence, when the availability of high quality foods (forbs, soft new shoots and young leaves of woody plants, etc.) declines in the dry season, it is to be expected that the feeding: ruminating time ratio should be higher in the larger browsers than in the smaller ones. Finally, because small ungulates are obliged to seek out high quality foods, which are scarce and discretely distributed, intraspecific competition precludes gregariousness in such species (Jarman 1974), and therefore also precludes the benefits of group vigilance. Small ungulate species thus generally occur alone or in pairs and depend more on hiding than fleeing as their primary antipredator behaviour (Brashares et al. 2000). However, because the smallest ungulates are vulnerable to the widest range of predators (Cohen et al. 1993; Sinclair et al. 2003; Radloff and du Toit 2004) yet cannot share vigilance costs among group members, the smaller members of an ungulate guild could be expected to allocate a relatively larger portion of their daily time budget to vigilance. Most investigations into the scaling of ungulate activity patterns involve meta-analyses, which seek patterns in the combined results of several independent studies and therefore have inherent potential for inaccuracy (see Gates 2002). Biases and confounding heterogeneity may be incorporated due to differences in methodology among studies, and clustering of species by body size, phylogeny and/or feeding style (Mysterud 1998; Pe rez-barberı a and Gordon 1999). Here we report on an analysis aimed at testing predicted allometric effects on the diurnal activity budgets of African browsing ruminants that were observed within the same study area, using the same methods, over the same period. The four species we studied range widely in size (adult female body mass; Skinner and Smithers 1990) from steenbok, Raphicerus campestris, which is one of the smallest of all ruminants (11 kg), through impala, Aepyceros melampus (41 kg), and greater kudu, Tragelaphus strepsiceros (155 kg), to giraffe, Giraffa camelopardalis, which is the largest of all ruminants (828 kg). In terms of phylogenetic independence, we argue that within the Order Ruminantia these four species are sufficiently distinct from each other by virtue of separation of giraffe from the three antelopes at the family level (Giraffidae and Bovidae) and the separation of the three antelopes at the tribe level (Antilopini, Aepycerotini, and Bovini for steenbok, impala and kudu, respectively). We used diurnal time budget data obtained during behavioural observations on free-living animals to test the following five predictions: the smallest members of the browsing ruminant guild exhibit (1) the lowest allocation of diurnal time to activity; (2) the greatest hour-to-hour variation in activity and

3 319 resting time; (3) the greatest reduction in activity time during the hottest days; (4) the least change between wet and the dry seasons in the ratio of feeding: ruminating time; and (5) the greatest allocation of diurnal time to vigilance. Materials and methods Study area Our analysis was based on data collected during a 3-year study ( ) on resource utilization within the African browsing ruminant guild (du Toit 1988) in the vicinity of Tshokwane (24 47 S, E), in the central region of the Kruger National Park, South Africa. Mean annual rainfall at Tshokwane is 525 mm, of which >80% falls between October and March (Venter et al. 2003). Highest daily maximum temperatures are in November March (typically about 32 C, but up to 44 C), and lowest daily minimum temperatures occur in June and July (about 6 C). The study area is characterised by two major vegetation types: Sclerocarya birrea/acacia nigrescens savanna on basaltic clay plains, and Combretum-dominated scrub savanna on thin rocky soils of the rhyolitic Lebombo range (Venter et al. 2003). Animal observations Focal animal observations (Altmann 1974) were performed on individuals within each species from a 4-wheel drive vehicle, using 7 50 binoculars. Observation was facilitated by the relatively high degree of vehicle habituation within the study animal populations. Data were collected on adult females only and focal animals were either artificially marked (with radio-collars or marker-collars; see du Toit 1988), or individually identifiable from natural coat patterns or scars. Sessions of continuous observation varied in duration but were typically 3 5 h each. A known individual would not be used again for another observation session until at least four observation sessions had been performed on other individuals of the same species. By this system any one animal would not appear in the data set with a frequency of more than once per month and observations were distributed over at least 20 individuals per study species. Observation sessions were prescheduled for each species to spread data collection across all hours between 0600 and 1800 hours in every calendar month. Data capture involved the use of a portable microcomputer (Sharp PC-1500 with 8 K RAM) attached to the dashboard of the vehicle. The data-capture programme (written in BASIC) presented a menu of first-order (italics) and second-order (brackets) activity classes as follows: feeding; moving (walking/walkingruminating/running); standing (ruminating/resting/alert/ grooming/socializing); lying (ruminating/resting); obscured. Each observation session thus yielded a data series comprising time-stamped events recorded in realtime sequence, and each data series was uploaded into a database in which seconds elapsed during each event were computed (see du Toit 1993). Data from the 3 months of each seasonal quarter were pooled to produce four data sets: quarter 1= January March; quarter 2= April June; quarter 3= July September; quarter 4= October December. A total of 1,094 h of focalanimal observations were recorded across all species during the study (Table 1). Table 1 Breakdown of diurnal time ( hours) allocated (percent of the day) to each behaviour category by each study species during each quarter a of the seasonal cycle and on average over the year Behaviour category Steenbok Impala Kudu Giraffe 1/4 2/4 3/4 4/4 Year 1/4 2/4 3/4 4/4 Year 1/4 2/4 3/4 4/4 Year 1/4 2/4 3/4 4/4 Year Feeding Walking Walking ruminating Running Standing ruminating Standing resting Standing alert Standing other b Lying ruminating Lying resting Total foraging c Total ruminating Feeding: ruminating d Total lying Hours recorded e a Seasonal quarters: 1/4 January March (mid-late wet season); 2/4 April June (early-mid dry season); 3/4 July September (mid-late dry season); 4/4 October-December (early-mid wet season) b Grooming, socializing, urinating and defecating c Feeding and walking (without ruminating) d Feeding/total ruminating e Hours of focal-animal observations digitized per quarter, with the total per species (total for all four species = 1,094 h)

4 320 Diurnal activity budgets The four data sets per species (one for each seasonal quarter) were used as replicate samples for the Kruskal Wallis analysis of variance by ranks (Kruskal and Wallis 1952) to test whether the four ruminant species varied significantly in their mean proportional allocations of diurnal time to each activity. Spearman rank correlation coefficients were calculated for the relationships between body mass and the mean proportions of diurnal time allocated to each activity during each quarter for each species. Hourly adjustment of activity in response to changing ambient temperature Three statistics were calculated to quantify the adjustment of activity in response to changing ambient temperature across the day: (1) the standard deviation of the mean proportion of time allocated to each activity across the 12 h of the day; (2) the range in proportional allocation of time to each activity across all hours of the day, calculated from the maximum and minimum hourly proportions; (3) the difference between the mean hourly proportion of time allocated to each activity during the six cool hours ( hours and hours) and the six warm hours ( hours) of the day. Using the four seasonal quarters as replicate samples of behavioural data for each species, Kruskal Wallis tests were performed to determine if the four ruminant species varied significantly in terms of each of the three statistics calculated for every activity, and Spearman rank correlations were performed to test for relationships between body mass and the values of the three statistics calculated for each activity. Daily adjustment of activity in response to maximum air temperature Each observation session was labelled with the daily maximum air temperature on that date at Skukuza (25 00 S, E), the nearest weather station 36 km southeast of Tshokwane. Observation days were classified into four maximum temperature (T i ) classes: T 1 =20 24 C; T 2 =25 29 C; T 3 =30 34 C; T 4 =35 C+. For each species within each temperature class, the total time that a particular activity was observed was divided by the total time that all activities were observed. Because every effort was made to spread observations on each species as evenly as possible across all hours of the day during each month of the year, this method allowed the comparison of activity budgets across species within each of the four classes of maximum daily temperature. We used Friedman s analysis of variance test for ranks (see Zar 1984). Adjustment of activity in relation to daily maximum temperature was quantified for each species by calculating the difference between T 1 and T 4 in the proportion of total observation time that each activity was observed. Spearman s rank correlation coefficients were then calculated for each activity across the four species to assess the relationship between body mass and the degree of activity adjustment between cool (T 1 ) and hot (T 4 ) days. Seasonal adjustment of the diurnal feeding : ruminating ratio To derive activity budgets for the wet and the dry seasons for each species, the data from quarters 1 and 4 were pooled for the wet season, and quarters 2 and 3 for the dry season. Friedman s test was used to compare the wet and the dry seasons in terms of the ratios of feeding: ruminating time calculated for the four species. Nonparametric Tukey-type multiple comparisons (see Zar 1984) were performed to examine the difference in feeding: ruminating ratio between all possible pairs of species within the guild. Spearman s rank correlation was used to investigate the relationship between body mass and the difference in feeding: ruminating ratio between dry and wet seasons. Results Effects of body size on diurnal activity budgets Within the browsing ruminant guild we found that increasing body size was associated with increasing time budget allocations to feeding and moving (Fig. 1a,b), and decreasing allocations to resting (Fig. 1c). Giraffe was the only species to ruminate while walking (3.53% of total diurnal activity; 26.9% of total ruminating time; see Table 1). Minor time allocations to walking-ruminating recorded for some of the other species were the result of focal animals beginning to walk away from a rumination bout while still processing the last bolus. The highly skittish and social impala was responsible for significant variation among the four species in time allocations to running (H=9.02, P<0.05) and grooming (H=12.51, P<0.001). The four species did not vary significantly in their time allocations to ruminating, being alert, or non-grooming social interactions, although most correlations between body mass and the percentage of diurnal time allocated to specific activities were significant (Table 2). Hourly adjustment of activity in response to changing ambient temperature Using our three measures to compare diurnal activity budget adjustments in response to changing ambient temperature across the day (SD = standard deviation of the mean hourly proportion of time allocated to each activity; R = range in hourly proportions of time allocated to each activity; D = difference between the mean

5 321 Table 2 Results of Spearman s rank correlation analyses of the relationships between log 10 body mass (kg) and the following statistics calculated over the four quarters of the seasonal cycle for steenbok, impala, kudu and giraffe: mean% diurnal time allocated to each activity; standard deviation (SD) and range in daily% time allocated to each activity; difference between the mean hourly proportion of time allocated to each activity during the six cool hours ( and hours) and the six warm hours ( hours) of the day Behaviour category Spearman s rank correlation coefficient r s Mean SD Range Difference Feeding 0.74*** Moving 0.91*** 0.51* 0.67** 0.27 Active a 0.81*** 0.52* 0.59* 0.24 Lying 0.79*** 0.75*** 0.74*** 0.47 Ruminating Resting 0.86*** 0.82*** 0.82*** 0.56* *P < 0.05; ** P < 0.01; *** P < a Feeding and moving, combined hourly proportion of time allocated to an activity in the six coolest and the six warmest hours of the day), we found that the four browsing ruminant species differed significantly for all three measures with regard to lying, which includes lying resting and lying ruminating (SD: H=12.77, P<0.001; R : H=13.10, P<0.001; D : H= 7.49, P<0.05) and resting, which includes lying resting and standing resting (SD: H=10.68, P<0.01; R : H=10.74, P<0.01; D : H=8.56, P<0.01). The smallest species, steenbok, showed the highest variation in activity through the day (Fig. 2), with distinct bouts of activity in the morning and evening separated by a midday resting spell (see du Toit 1993). Giraffe showed the smallest difference in time allocation to feeding between the six coolest and six warmest hours of the day (H=8.54, P<0.05). Hourly variation in time spent resting (measured in terms of SD, R and D) was the most consistent factor to emerge as a body size effect, as determined from Spearman s rank correlation coefficients (Table 2). Daily adjustment of activity in response to maximum air temperature Fig. 1a c Percentages of diurnal time spent a feeding, b moving, and c resting by steenbok, impala, kudu and giraffe, plotted in order of increasing body size. The accompanying Kruskal Wallis H-statistic and P-values express the significance of the overall variation across species in each plot. Legend: central square, mean; box, ±1 standard error; bars, ±1 standard deviation All four species decreased their time spent feeding and moving on hot days, in favour of standing resting (Fig. 3). A clear body-size effect emerged with regard to the difference between cool and hot days in the proportions of time allocated to being active, and feeding in particular, with the larger species suffering a relatively greater loss of feeding time on hot days (Fig. 4). Seasonal adjustment of the diurnal feeding : ruminating ratio There was significant variation among species in terms of the difference between wet and dry seasons in the ratio of feeding : ruminating time (Friedman s test:

6 322 Fig. 3 Percentage of diurnal time spent standing resting, plotted against daily maximum ambient temperature for steenbok (circles), impala (squares), kudu (triangles) and giraffe (crosses). Temperature classes (T i ): 1=20 24 C; 2=25 29 C; 3=30 34 C; 4=35 C+ Fig. 2a,b SD of the mean hourly proportion (MHP) spent a lying and b resting during the day for steenbok, impala, kudu and giraffe, plotted in order of increasing body size. The accompanying Kruskal Wallis H-statistic and P-values express the significance of the overall variation across species in each plot. Legend: central square, mean; box, ±1 standard error; bars, ±1 standard deviation v r 2 =11.08, P<0.025). Multiple comparisons showed that the increase in this ratio from wet to dry seasons was significantly greater for giraffe and impala than for steenbok and kudu (Fig. 5). Discussion Results of our study conform to those of Owen-Smith (1988, 1992) and support our Prediction 1, that diurnal time spent feeding and moving in a savanna ruminant guild is strongly positively correlated with body mass (Table 2). This is clearly different to the pattern found for temperate species (Belovsky and Slade 1986; Mysterud 1998; Pérez-Barberı a and Gordon 1999), so there must be environmental factors that override physiological and morphological factors in determining how ruminant activity budgets scale with body mass. Fig. 4 Difference between daily maximum temperature classes T 1 (20 24 C) and T 4 (35 C+) in terms of percent diurnal time allocated to feeding [difference=(percent feeding time in T 1 ) (percent feeding time in T 4 )] plotted against body mass for steenbok, impala, kudu and giraffe respectively (r s =0.98; P<0.001) Mysterud (1998) suggested that in tropical ruminants the larger species might spend comparatively more time feeding during the day. We could not compare diurnal and nocturnal activity for all four of our study species, but using nocturnal data for steenbok (du Toit 1993) and kudu (J.T. du Toit, unpublished data), we found the ratio of diurnal : nocturnal activity (feeding and walking) in the dry season to be : 1 for steenbok (11 kg) and : 1 for kudu (155 kg), which is consistent with Mysterud s suggestion. Furthermore, among African ruminants the larger browsers seem to spend more time foraging than similar-sized grazers (Owen-Smith 1992), and this should steepen the positive slope of the overall activity body size relationship for any assemblage of grazing and browsing species. One likely reason is that the ratio of feeding : ruminating time differs between grazers (lower) and browsers (higher) due to differences

7 323 Fig. 5 Difference between dry and wet seasons in terms of the ratio of feeding : ruminating time [difference=(dry season ratio) (wet season ratio)] plotted against body mass for steenbok, impala, kudu and giraffe respectively (r s =0.80, P=0.20) in the fermentation rates (see Owen-Smith 1982) of grass (slower) and browse (faster). Future studies on the scaling of nutritional ecology in ruminants should consider feeding and rumination as linked processes, rather than separate them into active and inactive behaviour categories. An additional reason for increased feeding time among larger browsers in African savannas could lie in the physical properties of the foliage they browse on. We suggest that in African savannas large browsing ruminants cannot increase their instantaneous rate of food intake in direct proportion with their metabolic needs because they are constrained by the spinescence and branching architecture of their staple woody food plants (Leuthold and Leuthold 1978; Owen-Smith 1982; Pellew 1984a,b; Cooper and Owen-Smith 1986). This forces large ruminants like giraffe to increase feeding time to meet daily intake requirements. In contrast, large boreal forest browsers such as moose (Alces alces) may increase their instantaneous food intake rates by feeding on softstemmed aquatic forbs and nonspinescent browse (Belovsky 1981). This highlights the need to consider not only how differences between feeding guilds may influence activity patterns within a taxonomic grouping (e.g. Ruminantia), but also how differences between ecosystems may influence activity patterns within a feeding guild. Prediction 2, that hour-to-hour variation in activity and resting time is most pronounced in the smallest members of the guild, due to greater sensitivity to hourto-hour variation in ambient conditions, was partially supported by our results. The standard deviations and ranges in hourly proportions of time allocated to each activity revealed that hour-to-hour variation in time spent active and resting decreased significantly with increasing body mass (Table 2, Fig. 2). The larger species thus maintained more even levels of activity through the cool and hot phases of the day. Within that activity, however, they exhibited greater variation in the mean hourly proportion of time spent moving (Table 2). This reflects the tendency for the larger species, with their larger home ranges, to periodically engage in long bouts of walking between feeding patches and then long bouts of feeding. We also found that giraffes exhibited the smallest difference in mean hourly feeding time between the six coolest, and six warmest hours of the day (H=8.54, P<0.05). Interestingly, however, we found a significant positive correlation between body mass and the difference between hot and cool phases of the day in time allocations to resting (Table 2). To meet their intake requirements the larger species do not reduce feeding time when the sun is high (except on very hot days), but they reduce nonfeeding activities in favour of resting at these times. All four of the browsing ruminant species we studied in Kruger exhibited reduced activity (and feeding, when considered in isolation) and increased resting on days with high maximum temperatures (Fig. 3). This is consistent with expectations from other studies on behavioural thermoregulation in ungulates (e.g. Jarman and Jarman 1973; Leuthold and Leuthold 1978; Belovsky and Slade 1986; Klein and Fairall 1986; Owen-Smith 1998). However, Prediction 3, that adjustment of the diurnal activity budget in response to high ambient temperature (measured as daily maximum air temperature) is most pronounced in the smallest members of the guild, was not supported. The difference between cool (T 1 ) and hot (T 4 ) days in the proportion of time spent active was actually found to increase with increasing body mass, and this trend was particularly prominent for feeding time (Fig. 4). This implies that, in savanna environments, the foraging time budgets of the larger guild members are more susceptible to thermoregulatory constraints than those of the smaller ones. Although this may seem counter-intuitive, it is explained by the smaller species requiring less total daily foraging time, which they can allocate to the cooler parts of the day and to the night (e.g. steenbok; du Toit 1993). When high ambient temperatures render a portion of the day unsuitable for foraging, it has less effect on the smaller species that have a lower requirement to use that time for foraging anyway, and more effect on the larger species that need to forage throughout the day. Ruminants may be able to at least partially compensate for lower quality diets by increasing rumination time (Hofmann 1989), and experiments on Nubian ibex (Capra ibex nubiana) found that chewing investment (chews per gram ingested) while ruminating was 50% greater in the smaller-bodied females than in the males (Gross et al. 1995). In the Kruger browsing guild, Prediction 4 was partially upheld in that steenbok and kudu apparently compensated for the dry season decline in food quality because they maintained fairly constant feeding : ruminating time ratios through both seasons (Fig. 5). It appears that kudus (Owen-Smith and Cooper 1989) and steenbok (du Toit 1993) are able to maintain a stable nutritional plane if they have access to key foods and resource patches during the dry season. Furthermore,

8 324 habitat specialization in the smaller ruminants is probably an evolutionary response to the requirement for selective feeders to have access to high quality foods all year round (du Toit and Owen-Smith 1989). The dry season increase in feeding: ruminating ratio in impalas can be explained by the seasonal switch from grazing to browsing in this mixed feeder (Jarman and Sinclair 1979; du Toit 2003). The higher fermentation rate of browse compared to grass (Owen-Smith 1982) should allow a reduced investment in rumination. Of most interest, however, was our finding that during the dry and early wet seasons giraffes allocated disproportionately large amounts of diurnal time to feeding as compared with ruminating (Table 1). This reflects the increased feeding time required by giraffes to meet their high intake requirements while cropping foliage from strongly spinescent and lignified twigs (see Pellew 1984a,b), combined with the constraint that giraffes appear to have in extending their rumination time. Indeed, giraffes differ from all other sympatric browsers, and perhaps all other extant ruminants, in ruminating while walking. Giraffes are near the maximum size attainable by ruminants (Clauss et al. 2003) and so we assume that walkingruminating (>25% of total rumination time) is an adaptation for alleviating their passage rate constraint. Because the proportion of time an animal spends resting is inversely related to the proportion of time it spends active, it was not unexpected that the proportion of diurnal time spent resting by browsing ruminants in Kruger decreased with increasing body mass (Table 2). Steenbok, however, spent significantly more time resting while lying (H=13.58, P<0.001) when compared with the other three more gregarious browsing species that usually rested while standing. A large component of resting serves no physiological or ecological function other than energy conservation (Herbers 1981), although the resting behaviour of small and usually solitary ungulates may be part of an anti-predator strategy (Jarman 1974). For example, vigilance in steenbok requires the animal to stand (to see over the herb layer) and employ small but quick and distinctive movements of the ears and head. The associated risk of detection evidently outweighs the benefits because we found that steenbok allocated no more time to vigilance than the larger guild members did, and indeed they frequently abandoned vigilance in favour of lying cryptically immobile. This is the same strategy adopted by savanna lagomorphs and newborns of certain large antelope species ( hiders ; Estes 1991). Hence we found no support for Prediction 5 that the smallest ungulate, being the most solitary and the most vulnerable to predation, should have the highest time budget allocation for predator detection. This demonstrates how factors such as habitat specialization and gregariousness interact with body size to affect anti-predator behaviour (Geist 1974; Jarman 1974; Brashares et al. 2000). The number of species that can be meaningfully covered imposes a severe handicap on detailed concurrent studies of the activity patterns of syntopic species within a trophic guild. Covering both diurnal and nocturnal phases of the 24 h cycle with equal resolution is an additional challenge that should be addressed in future studies, using improved technology for observation and data capture. Despite the challenges, such studies are required to clarify outputs from metaanalyses and to provide explanations for trend differences between guilds and/or biomes. We hope the results of our study represent a step in that direction. Acknowledgements We thank Chris Smit for his statistical advice and Norman Owen-Smith for his stimulating discussions during the fieldwork. Managers and scientists in the Kruger National Park, especially Pat Wolff and Peter Retief, provided invaluable assistance, while funding was generously provided by SANParks. The only disturbance to animals during this study was chemical immobilisation (for radio-collaring) performed by SANParks veterinarians following the standard operating procedures in effect at that time. References Altmann J (1974) Observational study of behaviour: sampling methods. Behaviour 49: Bell RHV (1971) A grazing ecosystem in the Serengeti. Sci Am 225:86 93 Belovsky GE (1981) Food plant selection by a generalist herbivore: the moose. Ecology 62: Belovsky GE, Slade JB (1986) Time budgets of grassland herbivores: body size similarities. Oecologia 70:53 62 Brashares JS, Garland T, Arcese P (2000) Phylogenetic analysis of coadaptation in behaviour, diet, and body size in the African antelope. Behav Ecol 11: Calder WA (1984) Size, function, and life history. Harvard University Press, Cambridge Clauss M, Frey R, Kiefer B, Lechner-Doll M, Loehlein W, Polster C, Ro ssner GE, Streich WJ (2003) The maximum attainable body size of herbivorous mammals: morphophysiological constraints on foregut, and adaptations of hindgut fermenters. Oecologia 136:14 27 Cloete G, Kok OB (1986) Aspects of the water economy of Steenbok (Raphicerus campestris) in the Namib Desert. Madoqua 14: Cohen JE, Pimm SL, Yodzis P, Sadana J (1993) Body sizes of animal predators and animal prey in food webs. J Anim Ecol 62:67 78 Cooper SM, Owen-Smith N (1986) Effects of plant spinescence on large mammalian herbivores. Oecologia 68: Demment MW, Van Soest PJ (1985) A nutritional explanation for body-size patterns of ruminant and nonruminant herbivores. Am Nat 125: du Toit JT (1988) Patterns of resource use within the browsing ruminant guild in the central Kruger National Park. PhD dissertation, Wits University, Johannesburg du Toit JT (1993) The feeding ecology of a very small ruminant, the steenbok (Raphicerus campestris). Afr J Ecol 31:35 48 du Toit JT (2003) Large herbivores and savanna heterogeneity. In: du Toit JT, Rogers KH, Biggs HC (eds) The Kruger experience: ecology and management of savanna heterogeneity. Island, Washington D.C., pp du Toit JT, Owen-Smith N (1989) Body size, population metabolism, and habitat specialization among large African herbivores. Am Nat 133: Estes RD (1991) The behavior guide to African mammals. University of California Press, Berkeley Feldhammer GA, Drickhamer LC, Vessey SH, Merritt JF (1999) Mammalogy: adaptation, diversity, and ecology. McGraw-Hill, Boston

9 325 Gates S (2002) Review of methodology of quantitative reviews using meta-analysis in ecology. J Anim Ecol 71: Geist V (1974) On the relationship of social evolution and ecology in ungulates. Am Zool 14: Gross JE, Demment MW, Alkon PU, Kotzman M (1995) Feeding and chewing behaviours of Nubian ibex: compensation for sexrelated differences in body size. Funct Ecol 9: Haim A, Skinner JD (1991) A comparative study of metabolic rates and thermoregulation of two African antelopes, the steenbok Raphicerus campestris and the blue duiker Cephalophus monticola. J Therm Biol 16: Herbers JM (1981) Time resources and laziness in animals. Oecologia 49: Hofmann RR (1989) Evolutionary steps of ecophysiological adaptation and diversification of ruminants: a comparative view of their digestive system. Oecologia 78: Hudson RJ (1985) Body size, energetics, and adaptive radiation. In: Hudson RJ, White RG (eds) Bioenergetics of wild herbivores. CRC, Florida, pp 1 25 Jarman PJ (1974) The social organisation of antelope in relation to their ecology. Behaviour 48: Jarman MV, Jarman PJ (1973) Daily activity of impala. E Afr Wildl J 11:75 92 Jarman PJ, Sinclair ARE (1979) Feeding strategy and the pattern of resource partitioning in ungulates. In: Sinclair ARE, Norton- Griffiths M (eds) Serengeti: dynamics of an ecosystem. University of Chicago Press, Chicago, pp Kleiber M (1961) The fire of life. Wiley, New York Klein DR, Fairall N (1986) Comparative foraging behaviour and associated energetics of impala and blesbok. J Appl Ecol 23: Kruskal WH, Wallis WA (1952) Use of ranks in one-criterion analysis of variance. J Am Stat Assoc 47: Leuthold BM, Leuthold W (1978) Daytime activity patterns of gerenuk and giraffe in Tsavo National Park, Kenya. E Afr Wildl J 16: Mitchell D, Maloney SK, Jessen C, Laburn HP, Kamerman PR, Mitchell G, Fuller A (2002) Adaptive heterothermy and selective brain cooling in arid-zone mammals. Comp Biochem Physiol B 131: Mysterud A (1998) The relative roles of body size and feeding type on activity time of temperate ruminants. Oecologia 113: Norton PM (1981) Activity patterns of klipspringers in two areas of the Cape Province. S Afr J Wildl Res 11: Owen-Smith N (1982) Factors influencing the consumption of plant products by large herbivores. In: Huntley BJ, Walker BH (eds) The ecology of tropical savannas. Springer, Berlin Heidelberg New York, pp Owen-Smith RN (1988) Megaherbivores: the influence of very large body size on ecology. Cambridge University Press, Cambridge Owen-Smith N (1992) Grazers and browsers: ecological and social contrasts among African ruminants. In: Spitz F, Janeau G, Gonzalez G, Aulagnier S (eds) Ongulés/Ungulates 91. SFEPM- IRGM, Toulouse, pp Owen-Smith N (1998) How high ambient temperature affects the daily activity and foraging time of a subtropical ungulate, the greater kudu (Tragelaphus strepsiceros). J Zool London 246: Owen-Smith N, Cooper SM (1989) Nutritional ecology of a browsing ruminant, the kudu (Tragelaphus strepsiceros), through the seasonal cycle. J Zool London 219:29 43 Pellew RA (1984a) Food consumption and energy budgets of the giraffe. J Appl Ecol 21: Pellew RA (1984b) The feeding ecology of a selective browser, the giraffe (Giraffa camelopardalis tippelskirchi). J Zool London 202:57 81 Pe rez-barberı a FJ, Gordon IJ (1999) The relative roles of phylogeny, body size and feeding style on the activity time of temperate ruminants: a reanalysis. Oecologia 120: Peters RH (1983) The ecological implications of body size. Cambridge University Press, Cambridge Radloff FGT, du Toit JT (2004) Large predators and their prey in a southern African savanna: a predator s size determines its prey size range. J Anim Ecol 73: Schmidt-Nielsen K (1984) Scaling. Why is animal size so important? Cambridge University Press, New York Sinclair ARE, Mduma S, Brashares JS (2003) Patterns of predation in a diverse predator-prey system. Nature 425: Skinner JD, Smithers RHN (1990) The mammals of the southern African subregion, 2nd edn. University of Pretoria, Pretoria, South Africa Tilson RL (1980) Klipspringer (Oreotragus oreotragus) social structure and predator avoidance in a desert canyon. Madoqua 11: Tinley KL (1969) Dikdik Madoqua kirki in South West Africa: Notes on distribution, ecology and behaviour. Madoqua 1:7 73 Venter FJ, Scholes RJ, Eckhardt HC (2003) The abiotic template and its associated vegetation pattern. In: du Toit JT, Rogers KH, Biggs HC (eds) The Kruger experience: ecology and management of savanna heterogeneity. Island, Washington D.C., pp Zar JH (1984) Biostatistical analysis, 2nd edn. Prentice-Hall, Eaglewood Cliffs