1. Little is known about the interplay between tissue complexity and tissue selection and its effect on the ungulate functional response. Effects of sward height and of bison tissue selectivity on the functional response of bison Bison bison athabascae Rhoads were examined.
2. The height of stem present in a sward was varied. Grazing depth of bison was negatively related to height of stem, and this relationship was more pronounced in tall swards. Bison preferred leaf to stem tissue at all sward heights, though preference of leaf was positively related to sward height.
3. While controlling for selectivity, a factorial design was used effectively to uncouple sward height from sward biomass. Intake rates and bite sizes obtained on simple swards were used to quantify functional responses. Without tissue selection, sward height had no effect on the functional response. Sward biomass explained 90% of the variation in intake rates.
4. Rates of bison food intake on complex swards with both stem and leaf tissue were significantly depressed at low sward biomass, relative to those obtained on simpler leaf-only swards. Thus, sward tissue composition is a crucial covariate for characterizing intake rates of grazing ungulates.
Obtaining relevant estimates of resource availability is a long-standing problem in ecology owing to the disparity between human and forager perception. This is especially true of terrestrial herbivore foraging studies because the resource under exploitation (plants) is rarely presented in discrete packages. Crawley (1983) stresses this point, citing numerous examples in which seemingly identical patches of forage represent markedly different availabilities to the herbivores exploiting them. This presents difficulties for those who wish to measure the functional responses of herbivores, i.e. the relationships between forage intake and forage availability ( Solomon 1949). This issue is of central importance, because functional responses provide the crucial link between trophic levels in models of plant–herbivore dynamics.
To address these problems, we investigated the foraging patterns of captive Wood Bison grazing hand-constructed swards of natural forage, Carex atherodes Spreng. Our first objective was to test whether bison exhibit tissue selectivity, and, if so, to determine how selectivity affects the shape of the functional response. Previous research has shown that some grazing ungulates exhibit marked preferences for leaf tissue ( Poppi, Minson & Ternouth 1980; Akin 1989). Thus, only part of the total biomass may be perceived as food by the forager. This should be reflected in the shape of the functional response. For selective grazers, forage intake rate should increase more slowly with increasing total biomass than with biomass of preferred tissues.
Our second objective was to test whether sward height influences food intake rates independent of plant biomass. Even when biomass has little or no relationship with intake ( Spalinger et al. 1988 ), sward height may show a positive relationship to bite size and intake, independent of biomass. One mechanism suggested for this is that bite area increases with sward height because of a scooping action with the jaw or tongue that allows the grazer to prehend plants over a broader area ( Illius & Gordon 1987; Burlison, Hodgson & Illius 1991; but see Gordon, Illius & Milne 1996). Thus, increases in sward height should show concomitant increases in bite size and intake rate, independent of biomass. Alternatively, changes in sward height independent of biomass may not affect bite size or intake, if in fact bite area remains constant.
Our third objective was to test the effect of sward structure on bison intake rates by presenting bison with two types of experimental swards. The first type was a simple sward composed solely of preferred plant tissue, while the second type was a complex sward similar in structure to naturally occurring swards of C. atherodes. It was predicted that the functional response should increase more slowly with total biomass in the complex sward as animals strive to select preferred tissues.
The study site was located in the Slave River Lowlands, 20 km north of Fort Smith, Northwest Territories, Canada. Research was conducted from May to August in 1993 and 1994 at the government-operated Hanging Ice Bison Facility. The main 8-km2 enclosure contained 170 Wood Bison and abundant sedge and grass meadows predominated by C. atherodes, and several Calamagrostis species, respectively. Rowe (1972) and Reynolds (1987) describe the physiography and vegetation of the study area, and Reynolds, Hansen & Peden (1978) give detailed description of meadow communities, the primary habitats used by foraging bison. Reynolds & Hawley (1987) describe food habits of local free-ranging bison herds. For our experiments, six yearling bison were separated from the main herd and rotated among large holding corrals to provide them with a mix of natural free-standing forage and hay.
Procedure for constructing experimental swards
Experimental swards of C. atherodes were prepared following the methodology of Black & Kenney (1984) and Gross et al. (1993) . Carex atherodes is a primary constituent of Wood Bison diets in the Slave River Lowlands, at times composing up to 98% of natural diets ( Reynolds et al. 1978 ). Carex atherodes tillers exhibited an umbrella-like growth form, consisting of a central stem with leaves branching out from the apex. While the complex architecture of many gramineous species prohibits easy experimental manipulation of leaf to stem ratios, that of C. atherodes is manipulated with relative ease. Tillers were anchored to a 2-m × 1-m × 19-mm plywood platform in 13-mm holes spaced 100 mm apart, and held in place by rubber stoppers inserted into the holes from below. Plant spacing and total area covered by forage were constant in all trials.
Experimental swards were presented to bison isolated individually in a small corral for a short foraging interval (1–5 min). Trials were terminated when the sward had been completely grazed, or when an individual began to re-graze clumps of tillers, whichever came first. A training period identical in protocol to the experimental trials was conducted daily for 2 weeks prior to the experimental trials, to accommodate the animals to the foraging platforms and trial procedure. Animals fed readily from the platforms, and did not hesitate to walk on the foraging surface.
Tissue selectivity trials
Experiments were conducted to examine tissue selectivity by foraging bison. Using a factorial design, three sward heights (150, 300 and 450 mm) were crossed with four stem proportions (13%, 40%, 67% and 93% stem, calculated as a percentage of the total tiller height). The experiment was replicated six times, using each individual bison as a sample unit. Two tillers were inserted into each hole. To assess grazed sward height, vertical heights of 20 grazed tillers were recorded upon completion of the trial. The percentage stem consumed was calculated by subtracting final stem height from initial.
Relationships between grazing depth (the height at which plants were grazed, measured from the top of the sward) and percentage stem on offer were plotted for each of three initial sward heights. A two-way anova was used to test for homogeneity of slopes among the three groups. When this hypothesis was rejected, linear least squares regression techniques were used to analyse the relationship between percentage stem and grazing depth independently for each sward height ( systat 6·0; Systat 1994).
To assess tissue selectivity, Ivlev’s electivity index ( Krebs 1989, p. 394) was used. This index ranges from −1 to 1, with positive values indicating preference and negative values indicating avoidance; a score of zero indicates no selection. All results are shown as means ±1 SE, except where otherwise indicated.
An effect of sward height on bison bite size and functional response was tested for. The functional response was determined from instantaneous feeding rates, following Gross et al. (1993) . A factorial design replicated six times was used, crossing three sward heights (75, 150 and 300 mm) with five forage biomass densities ranging from 20 to 400 g m−1 (based on previous observations of bison foraging by Hudson & Frank (1987)).
Whereas biomass and sward height covary in most natural pastures, our experimental design permitted sward height to be effectively uncoupled from biomass. This allowed examination of the effect of sward height on intake, while controlling for changes in biomass. Biomass was varied by altering the number of tillers inserted in each hole, while sward height was kept constant. To minimize tissue selection, only leaf material was used in these trials, although 30–50 mm of stem was left at the base of each tiller to hold the leaves together. This stem tissue was not available to foraging animals because of its location below the foraging surface.
Two observers counted the total number of bites (identified as the distinctive jerk of an animal’s head and associated sounds produced when herbage was severed) prehended from the sward during each trial. In the rare event that counts differed, the mean of the two was used. Availability was measured as the sward mass before grazing, corrected for dry matter content and unavailable forage projecting beneath the foraging surface. Consumption was measured as the difference between initial sward mass and final (grazed) sward mass. Dry matter content of forage was estimated from a 50-g sample of fresh forage collected at the time of sward construction. Samples were dried at 70 °C to a constant mass at the time of collection.
Bite mass (g bite−1) was calculated as consumption divided by the number of bites prehended. Intake rate (g min−1) was calculated as consumption divided by foraging interval length. Following Real (1977), a Michaelis–Menten formulation of the functional response (intake = (a × biomass)/(b + biomass) where a is the maximum feeding rate and b is the half-saturation constant) was used. Parameters were estimated for each individual-sward height combination using a standard non-linear least squares algorithm ( systat 6·0). When data failed to satisfy the assumption of normality or exhibited severe heteroscedasticity, logarithmic or Box–Cox (a family of logistic transformations; Krebs 1989) data transformations were used. A Kruskal–Wallis one-way anova was used to examine the effect of sward height on the functional response parameters.
The relationships of bite mass and bite rate with biomass were analysed using least squares linear regression techniques and a standard non-linear least squares algorithm, where appropriate ( systat 6·0). Previous studies suggested that increased sward height allows animals to prehend larger bites, but also leads to slower bite rates ( Hodgson 1981; Penning 1986; Laca et al. 1992 ). Consequently, bite mass vs biomass was plotted for each of three sward heights, and one-tailed pairwise comparisons of slopes ( Zar 1984, p. 295) were performed to test whether bison obtained heavier bites when grazing tall swards. One-tailed comparisons are justified in this case because of the expectation of a positive relationship between bite mass and sward height observed in numerous previous studies ( Hodgson 1981; Penning 1986; Laca et al. 1992 ).
Complex sward functional response trials
Simultaneous with the fixed-height/leaf-only trials, a second set of trials was conducted to measure the functional response of bison grazing complex swards composed of both leaf and stem tissue. These swards exhibited concurrent changes in height and biomass similar to what one might expect in a natural sward, i.e. low biomass swards were typically short, whereas high biomass swards were tall. Sward height was varied from 50 to 450 mm by 50-mm intervals, with two tillers anchored in each hole in the foraging platform. Trial procedure was otherwise identical to those trials described previously. The functional response was calculated as before and compared with results from the leaf-only trials, using the 95% confidence intervals of the functional response parameters.
Tissue selectivity by bison
Grazing depth of bison decreased in response to increases in the length of stem present in the sward. Decreases in grazing depth were minor when the initial sward height was short, but more pronounced when the initial sward height was tall (F0·05,2,65 = 33·0; P < 0·001; Fig. 1a). A more detailed analysis of these relationships revealed that there was no significant relationship between grazing depth and percentage stem at the shortest height, whereas at intermediate and tall heights, there was a significant negative relationship ( Fig. 1a) that was significantly more pronounced at the tallest height (t0·05,2,43 = 2·9; P < 0·01). This suggests that bison were more sensitive to stem when grazing high biomass swards, attempting to decrease the amount of stem they ingested by decreasing grazing depth.
Comparisons of the amount of stem consumed in relation to the amount of stem present in the sward ( Fig. 1b) supports this idea. When grazing short swards, bison maintained a relatively constant grazing depth, resulting in large increases in the amount of ingested stem when the proportion of stem in the sward was high ( Fig. 1b). When grazing tall swards, bison markedly decreased their grazing depth, resulting in proportionally less stem consumption than in short sward trials ( Fig. 1b).
Avoidance of stem and preference for leaf were both positively associated with initial sward height (F0·05,2,595 = 34·6; P < 0·001 and F0·05,23,595 = 72·1; P < 0·001, respectively), suggesting that bison are more selective when grazing taller swards ( Fig. 1c). However, as the proportion of stem increased, preference for leaf became stronger (F0·05,3,595 = 171·2; P < 0·001), whereas avoidance of stem became weaker (F0·05,3,595 = 186·7; P < 0·001). Although bison actively attempted to select against stem by decreasing grazing depth when percentage stem was increased ( Fig. 1a), it appears that they failed to stabilize the proportion of stem ingested ( Fig. 1c).
Functional response of bison grazing simple swards (leaf-only)
Food intake was positively associated with leaf biomass (F0·05,1,18 > 7·89; P < 0·012 for all individuals), taking the form of a type 2 functional response for all individuals tested ( Fig. 2). Parameter values for each functional response by individual-sward height combination are presented in Table 1. To facilitate comparison with functional responses obtained in natural composition trials, intake rates obtained at biomass above 250 g m−1 in the leaf-only swards were excluded from the analysis, since this was the maximum value at which intake was measured in the complex (leaf + stem) sward trials. For trials at all sward heights, maximum feeding rates (a) ranged between 23·2 and 55·6 g min−1. Values of b ranged between 19·2 and 176·1 g m−1. Some of the variation among individuals in functional response parameters could be accounted for by differences in body size. Although it was not feasible to weigh bison directly, individual body size was visually ranked from smallest to largest. a was positively correlated with body size (Spearman r0·05(1),6 = 0·83; P < 0·05), whereas b showed no significant relation.
Table 1. Functional response parameters (Michaelis–Menten formulation where a is the maximum intake rate and b is the half-saturation constant) estimated with non-linear least squares methods. Estimates are for each individual-sward height combination in the simple sward (leaf-only) trials. There were no differences between functional responses obtained on swards of different heights (Kruskal–Wallis test; a: H0·05,2 = 0·503, P = 0·778; b: H0·05,2 = 0·713, P = 0·700). Individuals are ordered from smallest to largest, by body mass. Population means are shown with 95% confidence intervals for comparison
a (g min−1)
b (g m−1)
Mean (leaf only)
37·8 ± 6·5
35·4 ± 12·2
Mean (leaf + stem)
54·8 ± 10·8
99·2 ± 45·5
The effect of sward height on forage intake of bison
When sward height was uncoupled from biomass, sward height had little effect on the functional response of wood bison ( Fig. 3; Table 1). Neither the maximum feeding rate a nor the half-saturation constant b were related to sward height (Kruskal–Wallis test; H0·05,2,15 = 1·604; P = 0·587 and H0·05,2,15 = 0·573, P = 0·751, respectively). One-tailed pairwise comparison of slopes revealed that bite mass was unrelated to initial sward height when biomass was held constant (t0·05(1),73 < −1·46 and P > 0·05 in all cases); Fig. 4. In contrast, log(biomass) explained over 90% of the variation in Box–Cox transformed bite mass ( Fig. 4; r2 = 0·92; P < 0·001). Similarly, bite rate was unrelated to initial sward height when biomass was held constant (t0·05(1),73 > 48·79 and P < 0·01 in all cases); however, there was a highly significant negative association between bite rate and log(biomass) (Y = 116·5 − 15·7X;r2 = 0·76; P < 0·0005). Bite rate was negatively correlated with log(bite mass) (Y = 34·3 − 21·4(X); r2 = 0·76; P < 0·001).
To facilitate comparison of our functional response parameters (a and b) with those used in the Spalinger & Hobbs (1992) formulation (i.e. h , average time required to crop a bite in the absence of chewing; Rmax, the maximum rate of processing plant tissue), we provide the following transformation: h = b/(a × plant density). In our experiment, plant density was a constant of 126 plants/m2, and Rmax = a. Thus, for bison grazing preferred tissue, h = 7·43 × 10−3 min bite−1, and Rmax = 37·8 g min−1.
Functional response of bison grazing complex swards (stem plus leaf)
As was the case in the fixed-height trials using only leaf tissue, food intake rate was a monotonically decelerating function of the total biomass of leaf plus stem tissue, corresponding to a type 2 functional response. Maximum feeding rate was 54·8 g min−1, declining by 50% as forage biomass was reduced to 99·2 g m−1 ( Table 1). There was a significant effect of the sward structure on the shape of the functional response (P < 0·05; Table 1). Both the maximum intake rate (a) and the biomass at which the intake is half the maximum (b) were significantly larger in the stem-plus-leaf trials compared to the leaf-only trials. Hence, sward complexity had the effect of reducing consumption rates, as predicted.
In simple swards composed of preferred leaf tissue, sward height had no effect on intake rate and bite mass of bison, contrary to some studies on other ungulate species ( Hodgson 1981; Penning 1986; Laca et al. 1992 ). However, neither Hodgson (1981) nor Penning (1986) successfully uncoupled the effects of biomass and sward height, which often covary strongly in vegetative swards. Intuitively, one would expect sward height to have little effect on bite mass or intake rate if bite area and depth remain constant. Laca et al. (1992) found that bite area of cattle increased with height of Lucerne and Dallisgrass swards owing to a sweeping action of the tongue, although the interaction with bulk density was also significant.
Illius & Gordon (1987) stated that large Bovinae exhibit a complex feeding style in which they use their tongues to gather forage into their mouths. Although use of the tongue by bison has been reported in some instances (D. Fortin, personal communication), we did not observe this in our experimental trials, consistent with observations of free-ranging bison by Hudson & Frank (1987). This may explain why sward height had no detectable effect on intake rate of bison. Unlike other Bovinae, bison rarely use their tongues to increase bite area on tall swards, so bite area remains constant. This likely explains the low variation in bite size with sward height that was observed.
Although sward height had no effect on intake rates of bison, sward biomass explained over 90% of the variation in bison intake rate ( Fig. 2). The functional response was quite similar to that described by Hudson & Frank (1987). The maximum intake rate reported for subadult bull and cow bison (450 kg) by Hudson & Frank (1987) was 68 g min−1 compared with 38 g min−1 for yearling bison (180 kg) in this study. These differences can readily be explained using Shipley & Spalinger’s (1992) allometric relationship relating intake rate to body mass. A strong negative association of bite rate and bite mass was found, whereas Hudson & Frank (1987) described the relationship as weak. Their finding was perhaps a methodological artefact of the large variance associated with their bite mass estimates that were obtained indirectly with hand-grab samples intended to mimic bites cropped by bison.
It seems rather odd that some researchers have failed to find any relationship between forage intake rate and biomass. By definition, intake rate must be zero when biomass is zero, and some higher value when forage biomass is abundant. This implies some sort of positive relationship between the two variables. It is therefore instructive to revisit some of the widely cited examples of uncoupling of herbivore intake from biomass. Allden & Whittaker (1970) manipulated pasture biomass by ploughing under sod strips of various widths while leaving other sod strips undisturbed. Thus, although within-strip biomass was the same in all fields, total biomass summed over the entire field was dependent upon the spacing and width of sodded strips. After observing that sheep intake rates were identical on pastures of different total biomass, they concluded that biomass is an unreliable measure of availability. Clearly, the scale at which sheep were responding to changes in biomass was not at the level of the field, but rather at the level of the strip of vegetation within the field.
Spalinger et al. (1988) also found that the forage intake rate of Black-Tailed Deer was independent of forage biomass. They manipulated biomass by spacing plants at increasing distances, so that at the lowest biomass level, a small amount of forage (2·3 g) was distributed over a large space (23 m2), resulting in very low bite densities (one bite per m2). The distribution of forage was not continuous, but rather fragmented into many patches separated by empty space. Under these unusual circumstances, it is not surprising that the foraging process was also discontinuous, with eating bouts interspersed with bouts of travel to the next feeding station. Perhaps Spalinger et al. (1988) might have found a different result if biomass had been measured on a smaller spatial scale.
Trudell & White (1981) and Batzli, Jung & Guntenspergen (1981) found intake was independent of biomass for caribou and lemmings, respectively. In both cases, forage consisted of forbs protruding through a continuous mat of lichen. Here again, the measure of availability included the empty space between feeding stations. The appropriate scale over which to measure availability relevant to the foraging process would seem intuitively to be at the feeding station level, where vegetation is continuously distributed over the bite dimensions of the forager. Travel time between patches then becomes a separate issue, as do other complicating factors such as the overlap of search and handling time ( Farnsworth & Illius 1996).
These examples point to the critical need for biomass to be measured on a scale appropriate to the process in question. Under homogeneous circumstances, biomass is not scale dependent; unfortunately, heterogeneity is the norm in natural systems, and scale of measurement becomes a concern, as any botanist deciding on quadrat shape and size well knows. For the measurement of biomass to which herbivores respond during foraging, an appropriate spatial scale is maximum bite area ( Black & Kenney 1984), a variable showing a strong allometric relationship with incisor dimensions and body size ( Gordon et al. 1996 ). Empty space contained within bite area will reflect changes in bite size, the natural unit of intake. Increases (or decreases) in empty space beyond the bounds of the maximum bite area (i.e. a change in biomass at a large spatial scale, but not at a feeding station level), will not affect the time it takes to crop one bite. It would, however, affect travel time between bites. With regards to field sampling of vegetation biomass and the choice of quadrat size, it would be reasonable to expand the quadrat size in order to minimize sample variance, so long as the distribution of forage within the quadrat reflected that contained within maximum bite area.
In a more recent model of functional response, Spalinger & Hobbs (1992) acknowledged scale as an important consideration by using bite size to define the grain (sensuKotliar & Wiens 1990). However, conventional Michaelis–Menten models could also incorporate effects of spatial scale simply by changing the scale at which available biomass is measured. Of course, it would be desirable to quantify the travel time and the overlap that may occur with handling time ( Farnsworth & Illius 1996, 1998). Perhaps a multiscale approach to measuring biomass would be informative. At what spatial scale does the positive relationship between bite size and biomass degenerate? What effect might this have on the population dynamics of the plants? Different sized ungulates feeding on the same sward might well have different transition points, and potentially different effects on plant demography.
Once a scale has been chosen for measurement of availability, some thought should be given to how to sample the vegetation in a manner consistent with forager perception. Bison showed a decrease in grazing depth when stem tissue was artificially increased in the sward, particularly when sward height was tall. Bison preferred leaf tissue and avoided stem tissue at all sward heights, and this selectivity increased with sward height. The cue for selective feeding by bison is therefore a combination of the amount of stem in the sward and sward height. Avoidance of stem may thus indicate that bison do not consider the total biomass as available food, but rather only a portion consisting of primarily leaf tissue.
It may seem illogical that bison exhibit sensitivity to increased stem at tall sward heights, but not at short sward heights ( Fig. 1a). Such a pattern of behaviour is consistent, however, with changes in the quality of vegetation that naturally occur during the growing season. Both biomass and sward height are low when plants are immature, and the digestibilities of stem and leaf tissues tend to be similar, only diverging as the sward senesces. Assuming that bison use sward height as a cue for sward quality, this scenario would account for the relative insensitivity of bison to sward composition at short sward heights. Trade-offs between short-term forage intake and digesta passage rates indicate that bison should feed more selectively in taller swards. Our results are consistent with this hypothesis: bison showed strongest preference for leaf and strongest avoidance of stem when grazing tall swards. This implies that bison do not treat total available biomass as available food, and that availability cannot be described with a single variable. Thus, it is useful to know the tissue composition of a sward as well as patterns of tissue selection in order to predict short-term rates of intake accurately.
Many thanks to Rita Antoniak, Nadele Flynn, Indrani Ghosh, Derek Haley, Jonathan Kenny, Emma White, Karla Williamson and Anita Young for their countless volunteer hours in the field, and to Troy Ellsworth and the staff at the Department of Renewable Resources in Fort Smith, Northwest Territories for logistical support and guidance. We are grateful to Elizabeth Boulding, Micheline Manseau and Tom Nudds for comments on earlier drafts of this manuscript, and to John Wilmshurst for his insightful advice and criticisms. This study was supported by grants from the Natural Sciences and Engineering Research Council of Canada, the Department of Renewable Resources for the Northwest Territories, the Association of Canadian Universities for Northern Studies, and the Canadian Wildlife Service.
Received 11 February 1999; revised 20 June 1999;accepted 7 July 1999