The herbivores dilemma: trade-offs between nutrition and parasitism in foraging decisions

Transcription

1 Oecologia (2000) 124: Springer-Verlag 2000 Michael R. Hutchings Ilias Kyriazakis Thomas G. Papachristou Iain J. Gordon Frank Jackson The herbivores dilemma: trade-offs between nutrition and parasitism in foraging decisions Received: 30 November 1999 / Accepted: 14 February 2000 Abstract An experiment was carried out using a tradeoff framework to determine the rules of sward selection, in relation to gastrointestinal parasite dispersion, used by mammalian herbivores, and the effect of level of feeding motivation and parasitic status on these rules. Twentyfour sheep divided into four animal treatment groups resulting from two levels of feeding motivation (high and moderate) and two parasitic states ( with Ostertagia circumcincta and non-) were presented with pairs of experimental swards which varied in N content (high and low), sward height (tall and short) and level of contamination with faeces and thus parasites (contaminated and non-contaminated). The selection for tall swards outweighed both the selection for N-rich swards and the avoidance of faecal contaminated swards. The selection for N-rich swards could not completely overcome faecal avoidance. Parasitism in animals with a moderate feeding motivation reduced their bite rates and grazing depths, thereby probably reducing the rate of ingestion of parasitic larvae. In contrast, highly feedingmotivated animals (including those ) increased their bite rates and grazing depths, thereby increasing the rate of ingestion of parasites. The inclusion M.R. Hutchings I. Kyriazakis Animal Biology Division, Scottish Agricultural College, West Mains Road, Edinburgh EH9 3JG, UK M.R. Hutchings ( ) Animal Biology Division, Scottish Agricultural College, Bush Estate, Penicuik, Midlothian EH26 0PH, UK m.hutchings@ed.sac.ac.uk Tel.: , Fax: T.G. Papachristou Forest Research Institute, National Agricultural Research Foundation, Vassilika, Thessaloniki, Greece I.J. Gordon Macaulay Land Use Research Institute, Craigiebuckler; Aberdeen AB15 8QH, UK F. Jackson Moredun Research Institute, Pentland Science Park, Bush Loan, Penicuik, EH26 0PZ, UK of parasite distributions, both in the environment and within herbivore host populations, is likely to advance optimal foraging theory by enhancing its predictive power. Key words Grazing Diet selection Parasitism Sheep Feeding motivation Introduction Herbivores interact with parasites in their grazing environment and are often presented with, and forced to make, foraging decisions involving trade-offs between the benefits of nutrient intake and the risks of parasitism. These trade-offs are created by the deposition of faeces, containing both nutrients and parasites, on pasture. Herbivores do not defecate at random. Faeces tend to be concentrated in areas where animals graze, where they rest (Marsh and Campling 1970), or are concentrated in latrines which may have a function in territorial marking (Estes 1991). As a consequence of the concentrating effect of digestion and defecation there tend to be high levels of nutrients (N, P and K) in areas of faecal deposits. This has an important effect on nutrient cycling, and in low nutrient ecosystems, typical of many extensive rangelands, can lead to an increase in soil nutrient levels around defecations. Consequently, the plants associated with areas where faeces are concentrated tend to have higher concentrations of nutrients in their tissues than the average in the system (Edwards and Hollis 1982). The increased nutrient concentration has the potential to make these plants highly desirable food items for herbivores in nutrient-limited environments. Numerous studies have shown that animals selectively graze on patches with high levels of N and digestibility (Bazely 1990; Langvatn and Hanley 1993; de Vries and Schippers 1994). As well as expelling nutrients in their faeces herbivores also expel the eggs and larvae of macroparasites (primarily helminths) living in the alimentary tract. As a

2 consequence the areas around faeces are frequently highly contaminated with infective-stage larvae. Gastrointestinal parasites, which are acquired through the action of grazing, are a pervasive challenge to the survival and reproductive ability of their hosts (Anderson 1978; Gulland 1992). Mammalian herbivores cannot detect gastrointestinal parasitic larvae themselves, but use faeces as a cue to avoid ingestion of parasites (Cooper 1997; Hutchings et al. 1998). If, as suggested above, animals should forage on nutrient-enriched and tall herbage in areas of high faecal contamination, they then run the risk of ingesting infective helminth larvae. The foraging herbivore is, therefore, faced with a trade-off between nutrient and parasite ingestion, and foraging decisions should be expected in relation to the consequences of the benefits of increased nutrient ingestion versus the costs of increased parasitic infection. Mammalian herbivores may use sward height and sward tone as cues to determine the potential forage intake rate and N content of swards, respectively (Black and Kenney 1984; Arnold 1987; Bazely and Ensor 1989; Bazely 1990), and olfactory cues to determine faecal and thus parasite-contaminated swards (Dohi et al. 1991; Hutchings et al. 1998). These environmental cues may enable grazing mammals to assess the relative costs and benefits of a grazing trade-off, which in turn determines their grazing strategy (Lafferty 1992). However, the relative costs and benefits of a trade-off to an animal may also depend on its physiological state (Kyriazakis et al. 1999). Level of feeding motivation, and immune and parasitic status, may affect foraging strategies of herbivores grazing faecal contaminated and thus parasitecontaminated pasture (Hutchings et al. 1998, 1999). The increased risks of further intake of parasites by already herbivores may lead to increased faecal and thus parasite avoidance (Cooper 1997; Hutchings et al. 1998, 1999). In contrast, the nutritional benefits (increased N intake and increased forage-intake rate) associated with faecal contaminated pasture may lead to reduced faecal avoidance in undernourished animals (Hutchings et al. 1999). Herbivore grazing behaviour in relation to faecal contaminated pasture may have knock-on consequences for the distribution of nutrients, parasites and herbage structure in the environment, as well as for parasite population dynamics. Grazing trade-offs may, therefore, play a key role in determining herbivore optimal foraging strategy. Here we used an experimental approach to determine the foraging rules used by mammalian herbivores by comparing the relative impacts of sward nutrient, height and faeces distribution on herbivore grazing behaviour. We also determined the impact of the physiological state of the animals (level of feeding motivation and parasitic status) on these rules of foraging behaviour. We tested the hypotheses that: (1) the benefits of grazing nutrient-rich and/or tall swards outweigh faecal avoidance in sheep, (2) highly feeding-motivated sheep select nutrient-rich swards and tall swards to a greater extent than sheep with lower feeding motivation, and (3) sheep will show an increased avoidance of faecal contaminated swards as compared to non- animals. Materials and methods Animals A pool of 32 Texel X Scottish Greyface ewe lambs (Ovis aries) were weaned at 6 8 weeks of age and kept indoors in individual pens. These animals were expected to be naive with respect to gastrointestinal parasites since they were reared indoors and had experienced only a high-quality pelleted feed. The experiment was carried out in June Day +1 corresponded to the first day sheep were presented with sward choice tests. Three weeks prior to the start of the experiment (day 21) the 32 sheep were divided into two groups balanced for live weight. Half of the sheep (n=16) were dosed daily with 2500 third-stage larvae (L 3 ) of Ostertagia circumcincta ( treatment) (Hutchings et al. 1999). The remaining 16 animals were non controls. From day 21, animals were trained to graze from sward trays (see Hutchings et al. 1999) and the twenty-four sheep (12, 12 naive), from the original 32 animals, which adapted most rapidly to the procedure (Illius et al. 1992) were chosen for the experiment. To ensure that sheep grazed the swards presented to them, a moderate feeding motivation was established from day 21 by feeding them to appetite ; the 24 sheep were offered two, 1-h ad-libitum feeding sessions/day on the high quality pelleted feed; the first feeding session occurred at 0800 hours and the second occurred at 1700 hours (Hutchings et al. 1999). On day 10, the 12 and the 12 non- animals were randomly divided into two feeding-motivation groups. Half of the animals remained on the feeding regime to establish a moderate feeding motivation and the remaining half were fed 60% of their to appetite ration, creating a high feeding-motivation treatment. This created four animal treatment groups; and moderately feeding motivated (P+FM ), and highly feeding motivated (P+FM+), non- and moderately feeding motivated (P FM ) and non- and highly feeding motivated (P FM+). The mean live weight±se of the experimental animals aged 3 months was 25.3±3.86 kg at the start of the experiment and was not different between the four treatment groups (P>0.5). Sward establishment and choices 243 One thousand two hundred sward trays (36 21 cm) were sown in a peat-based compost with perennial ryegrass (Lolium perenne) at 14,000 seeds/m 2, 12 weeks prior to the start of the experiment. The swards were maintained in an unheated polytunnel from the time of sowing until their use in an experiment. Swards were watered daily and cut to a height of 4 cm above the soil surface at 3-week intervals during the growing period to enhance sward establishment. After the first cut, half of the trays (n=600) were regularly treated with a fertiliser (1 N:3.75 P:1 K) at a rate of 24.4 kg N/ha/week and the remaining trays were regularly over-watered to ensure maximum leaching out of nutrients from the compost, creating two sward N contents (i.e. high and low). Swards were cut to 6 cm or to 12 cm the day of their use in the experiments, thus creating the two sward-height treatments. Three weeks prior to their use, half of the sward trays in each N treatment group were contaminated with 20 g faeces/tray from donor lambs infected with Ostertagia circumcincta. Twenty grams of faeces per sward tray was chosen since it represented a level of contamination which was just above the minimum level to which sheep show significant avoidance (Hutchings et al. 1998). The use of 3-week-old faeces ensured that the parasite eggs contained in the faeces had developed into infective stages and had contaminated the sward, creating a risk of infection to grazing animals

3 244 (Hutchings et al. 1998). The combination of N, height (H) and faecal contamination (F) created a possible eight sward tray treatments: N+H+F+, N+H+F, N+H F, N+H F+, N H+F+, N H+F, N H F+ and N H F (where N+=N-rich, N =N-poor, H+=12 cm tall, H =6 cm tall, F+=faecal contamination and F =no faecal contamination), and a possible 28 two-way choices. All choices which effectively compared levels of faecal contamination (i.e. F+ vs. F ; four choices) or N sward content (i.e. N+ vs. N ; four choices) or sward height (i.e. H+ vs. H ; four choices) or N+height (i.e. N+H+ vs. N H ; two choices) were not considered in the experiment as it was known that, with other things being equal, sheep avoid 20 g faeces/tray (Hutchings et al. 1998), select for N-rich swards (Hutchings et al. 1999) and tall swards (Black and Kenney 1984). All choices representing an inverse trade-off (i.e. where non-contaminated N-rich and/or tall swards were paired with N-poor and/or short swards contaminated with faeces) were also not considered (five choices). This left the following nine choices: 1. N H+F vs. N+H F (height vs. N). 2. N H+F+ vs. N+H F (height vs. N). 3. N H+F vs. N+H F+ (height vs. N). 4. N H+F+ vs. N+H F+ (height vs. N). 5. N H F vs. N+H+F+ (natural trade-off). 6. N+H F vs. N+H+F+ (height trade-off). 7. N H+F vs. N+H+F+ (N trade-off). 8. N H F vs. N H+F+ (height trade-off). 9. N H F vs. N+H F+ (N trade-off). Choices 6 and 7 having N+ and H+ common to both swards in the choice test, respectively, were effectively replicated in choices 8 and 9 having N and H common to both swards in the choice test, respectively. The latter two choices were also removed from the experiment leaving a total of seven choices. Comparing the grazing behaviour of sheep across these seven choices enabled the roles of sward N content and sward height in determining sheep foraging behaviour in relation to faecal contaminated swards to be quantified. Choices 1 4 directly compared the level of selection for N-rich and tall swards and the impact of faecal contamination of swards on that selectivity. Choices 5 7 compared the role of sward N richness and sward height in determining sheep foraging behaviour when presented trade-offs between nutrient and parasite intake. Of the four height vs. N choices: choice 1 was the key choice to which choices 2, 3 and 4 could be compared. Choice 1 represented a straightforward choice between grazing either tall swards (N H+F ) or N-rich swards (N+H F ). Choice 4 repeated this choice with the addition of faeces on both swards to determine whether selectivity for a particular sward characteristic (i.e. H+ or N+) was altered by the presence of faeces and thus parasites. Choice 2 represented a choice between faecal contaminated tall swards and non-contaminated N-rich swards to determine if faecal contamination of tall swards affected their selection when compared with choice 1. Choice 3 represented the inverse of choice 2 in that it determined whether faecal contamination of N-rich swards affected their selection when compared to choice 1. Of the trade-off choices: choice 5 represented the trade-off often occurring in natural systems between the benefits of increased N intake and herbage intake rate and the costs of parasitism. With all else being equal, choice 6 represented a trade-off between herbage intake rate and parasitism and choice 7 represented a trade-off between N intake and parasitism. Common experimental protocol The animals were offered the above choices a total of 3 times (three replicates) and allowed to graze from the swards for short periods either until they achieved 60 bites or until 10 min had elapsed (Hutchings et al. 1998). The experiment consisted of 504 choice tests (24 animals 7 choices 3 replicates). Ten animals were presented three different choices each day for 17 days in the experiment, and testing was balanced for period of day (see Hutchings et al. 1999). The structure of the experimental protocol enabled the effects of time of day or choice sequence on grazing behaviour to be taken into account in the statistical analysis (Hutchings et al. 1998). Measurements A faecal sample was taken once weekly from the rectum of all sheep from the time of allocation of animals to parasitic treatment group until the end of the experiment (day 21) and sampled for nematode eggs per gram of faeces (epg) (Christie and Jackson 1982). Blood samples were taken once weekly from day 21 from the animals by jugular venepuncture. Blood plasma samples taken on day 21 and day +17 were analysed for pepsinogen concentrations using the Moredun modification of the method described by Mylrea and Hotson (1969), to determine the effects of the parasitic treatments on the extent of abomasal damage in sheep. Pepsinogen concentrations are expressed as units of tyrosine, where one unit equals 1 µm tyrosine released per litre of plasma per minute at 37 C. Prior to each choice being presented to an animal, each tray was weighed and the mean sward height calculated from 10 samples/tray using a sward stick (Barthram 1985). The sward trays, which made up the choice, were presented to the animals in a metal frame which held the swards 10 cm apart. The bite number, i.e. number of head pulls associated with severing of herbage (Newman et al. 1992), taken from individual trays, was recorded on a hand-held Atari Portfolio computer (Atari Corporation, Slough, UK) using KEYTIME behaviour recording software (copyright J. Deag 1990). Mean herbage mass consumed (and hence bite mass), time taken to consume (and hence bite rate of) each tray, and grazed height was measured for each choice test using the methods of Hutchings et al. (1999). An extra sward from each sward-tray treatment was used in each testing period to measure the mass loss from sward trays due to evaporation and/or transpiration during each choice test. A total of five sample trays from each treatment were used to determine any differences in sward characteristics in the experiment. Each of these sample trays were divided into four cm quadrats. In two of these quadrats the sward was cut to the mean bite depth (calculated for each tray treatment) and the total vegetation was taken from the remaining two quadrats. The two total and the two bite depth samples taken from each sward tray were bulked, creating five total sward samples and five bite-depth samples for each sward tray treatment. Bite-depth sward samples (i.e. cut to the mean bite depth) from individual trays were often too small for individual analysis and, therefore, were bulked for each sward treatment, creating a composite mean for each tray treatment. Bite-depth sward samples were large enough to be analysed individually for sward tone and dry matter content (see below). All sward samples were analysed for dry matter (DM) content, N content, using the method of Pella and Colombo (1973) and in vitro organic matter digestibility (OMD), using the method of Alexander and McGowan (1966). The tiller densities of the five sample trays from each treatment were determined by sampling one of the cm quadrats of each tray with the 5 10-cm quadrat described above. Sward tone is correlated significantly with N content in perennial rye grass (Bazely and Ensor 1989). To determine whether a visible difference was associated with the sward treatments, the tone of all swards used in the experiment was measured using a Minolta CR-300 tristimulus colorimeter (Minolta, Milton Keynes). The key tone variable (L*) is measured on a 0 (black) to 100 (white) scale ( 60 to +60 scale; Thorogood 1995). To compare the benefit of grazing tall swards with the benefit of grazing nutrient-rich swards, N intake rates were estimated for non-, to-appetite (control, P FM ) animals grazing each of the two swards of choice 1 (N H+F vs. N+H F ) using the following equation: I i =BR i BM i pdm i pn i (1)

4 where I i =mean N intake rate for sward type i (g N/min), BR i =mean bite rate for sward i (bites/min), BM i =mean bite mass for sward type i [g wet matter (WM)], pdm i =proportion of dry matter in the grazed portion of sward type i and pni=n content (as a proportion) of the grazed portion of sward type i. Statistical analyses Sward tray characteristics (N, OMD, tiller density, DM and tone) were compared using ANOVA. Orthogonal contrasts were used to determine main effects of sward treatment on sward characteristics. Residual maximum likelihood (REML) (Patterson and Thompson 1971) was used to estimate the mean values for parameters of ingestive behaviour (bite rate, grazed height and bite weight) and sward selection (proportion of bites taken from sward trays) of the animals. The GENSTAT REML (Lawes Agricultural Trust 1993) option was used, which approximates SEs and SEs of the differences (SEDs) for the grazing parameters. The REML model was used to estimate the mean proportion of bites taken from the N-rich swards in each choice test, the sheep grazing height above the soil surface (cm) and bite mass (g WM) of the animals for each tray treatment in each choice type (seven choices), and the overall bite rate (bites/s) per choice test. The model included animal, day, period of day and their interactions as random effects to account for their potential effects on the variable under test; animal feeding motivation and parasitic status were used as fixed effects. Z-tests were used to determine whether the estimated proportion of bites taken from swards differed from indifference (0.5). All proportions were arcsine-transformed prior to their use in parametric statistical analyses (Zar 1984). Multiple comparison of means were carried out using the SED predicted by the REML model (Lawes Agricultural Trust 1993). Arcsine-backtransformed proportion means are given in tables with upper and lower 95% confidence limits due to the restriction on backtransforming SEs (Zar 1984). In all statistical tests, df were based on the number of different animals in the test and not the number of replicates (Zar 1984). Results Parasitic status of the sheep At the time of allocation of sheep to treatment groups all faecal egg counts were zero and remained zero for all Sward treatment Treatment effects 245 non- sheep throughout the experiment. Faecal egg counts of sheep increased to (mean±se) 258±59 epg by the end of the experiment (day +17). Blood plasma analysis showed that similar pepsinogen concentrations were apparent in and non sheep on day 21 (mean pepsinogen concentration, mu tyrosine±se: =143±28, non-=186±23). By the end of the experiment (day +17) pepsinogen concentrations had risen significantly in sheep, whereas in non- sheep no rise in pepsinogen concentrations was apparent (mean pepsinogen concentration, mu tyrosine±se: =677±127, t=4.13, df=11, P<0.001; non-=175±19, t=0.48, df=11, P>0.5). The increase in faecal egg counts and blood pepsinogen concentrations in the animals during the parasite trickle-dosing period suggested that the animals had an established nematode population in their abomasum. No clinical signs of disease were observed in the sheep, which suggested that animals contained a sub-clinical parasitism when choices were presented to them. Treatment effects on swards Table 1 shows the effects of sward tray treatment on sward characteristics. The total samples from N-rich swards contained higher N concentrations, had lower DM content and increased OMD than N-poor swards. The total samples from the tall swards contained higher N concentrations and were darker than short swards. Faecal contamination had no effect on sward characteristics. Bite-depth samples from N-rich swards had lower DM contents and were darker than N-poor swards. However, sward height did not affect the darkness of bitedepth sward samples. Grazing animals achieved greater N intake rates from the tall swards compared to the N-rich swards of choice 1. Using Eq. 1, N intake rates by non- animals Table 1 Effects of tray treatment on sward characteristics. Values are means of five sward trays from which the total sward was taken from two quadrats per sward tray (15 10 cm) which were bulked creating the total samples. The remaining two quadrats per sward tray were cut to the mean bite depth and bulked across the five trays per treatment to create the composite bite samples. Tiller density (tillers/cm 2 ) was estimated from a quadrat (5 10 cm) taken from one of the larger quadrats. N+ N-rich, N N-poor, H+ 12 cm height, H 6 cm height, F+ faecal contaminated, F noncontaminated, SED SE of the difference, N N content [g/kg dry matter (DM)], L* sward tone value (scale: 0=black to 100=white), OMD organic matter digestibility (g/kg DM); DM (g/kg wet matter) N+H+F+ N H F N+H F N H+F N+H F+ N H+F+ SED N Height Faeces N total *** ** NS N bite L* total NS * NS L* bite ** NS NS OMD total *** NS NS OMD bite DM total ** NS NS DM bite * NS NS Tiller density NS NS NS ***P<0.001, **P<0.01, *P<0.05, NS=P>0.05

5 246 Fig. 1 Sward selection by sheep presented trade-offs involving sward N content, sward height and faecal contamination of swards. Bars represent arcsine backtransformed mean proportion of bites from sward 2 with upper 95% confidence interval. All bars except for choice 7 differ from random selection (i.e. 0.5; P<0.05). Different letter combinations denote significant differences (P<0.05) with a moderate feeding motivation grazing the N-poor, tall sward were markedly greater than from the N-rich, short sward of choice 1 (N H+F g N/min grazing 6.44 cm from the soil surface; N+H F g N/min grazing 3.96 cm from the soil surface). Sward selection Figure 1 gives a summary of the results of animal sward selection for each of the seven choices in the experiment. When comparing sward selection across the N vs. height choices (choices 1 4), sheep always avoided the N-rich sward in favour of the tall sward. The N-rich, short sward was significantly and strongly avoided when paired with N-poor, tall sward, in choice 1, when both swards were not contaminated with faeces, and in choice 4 when both swards were contaminated with faeces. When comparing sward selection in choice 1 with choice 2 the addition of faeces to the tall sward reduced (P<0.05) the avoidance of the N-rich sward compared to the N-rich sward of choice 1. However, the N-rich swards remained significantly avoided in choice 2. When comparing sward selection in choice 1 with choice 3 the addition of faeces to the N-rich sward had no effect on sheep selectivity (P>0.05). All the trade-off choices (choices 5, 6 and 7) contained a common N-rich, tall sward contaminated with faeces (N+H+F+). When presented with choice 5, which often represents the trade-off in natural nutrient limiting conditions, the combination of N richness with tall swards overcame any avoidance of faeces, with strong and significant selection for the N+H+F+ sward (95.8% of bites) over the N H F sward (4.2% of bites). When presented with choice 6, the height trade-off choice, the attraction to graze tall swards overcame the avoidance of faeces, with strong and significant selection for the N+H+F+ sward (93.6% of bites) over the N+H F sward (6.4% of bites). When presented with choice 7, the N trade-off choice, the attraction to graze N-rich swards did not completely overcome the avoidance of faeces, with neither sward type being selected or rejected (N+H+F+ vs. N H+F : 53.3 vs. 46.7% of bites respectively). Sward selection was not affected by animal treatment (feeding motivation and parasitic status) in any of the choices except choice 7. Parasitised animals showed increased faecal avoidance, selecting the non-contaminated sward of choice 7 to a greater extent than the non- animals: arcsine backtransformed mean proportion of bites (± upper 95% confidence limit) taken from the N+H+F+ sward of choice 7; 0.44(0.38 to 0.53) and 0.61(0.53 to 0.68) respectively, P<0.05. No feeding motivation effect nor an interaction between parasitic status and feeding motivation was apparent in choice 7. Ingestive behaviour The interaction between feeding motivation and parasitic status was the dominant finding compared to main treatment effects on ingestive behaviour. Across all choices there was a consistent interaction between feeding motivation and parasitic status in relation to sheep bite rates:, moderately feedingmotivated sheep had the lowest bite rates of all animal treatments whereas,, highly feeding-motivated sheep (P+FM+) generally had the highest bite rates (Table 2). Parasitism reduced bite rates of animals in choice 1 (N H+F vs. N+H F ; Table 2). Increased feeding motivation was associated with increased bite rates, and this effect was significant in choices 2, 3 and 4. There were consistent interactions between feeding motivation and parasitic status in relation to grazing depths of sheep:, moderately feeding motivated sheep (P+FM ) grazed furthest from the soil surface compared to all other animal treatment groups, and highly feeding-motivated sheep (including the animals) (P+FM+; P FM+) grazed closest to the soil surface (Tables 3, 4). Feeding motivation and parasitic status had limited effect on grazing depths of sheep. Highly feeding-motivated animals (P FM+; P+FM+) grazed closer to the soil surface than moderately motivated animals in choices 3 and 5, and sheep (P+FM ; P+FM+) grazed further from the soil surface than non- animals (P FM+; P FM ) in choices 1, 4 and 5 (Tables 3, 4). Tall swards were never grazed as close to the soil surface compared to short swards (Tables 3, 4). A high degree of variability was associated with sheep bite mass. No consistent relationship between bite mass and any animal treatment in any of the choices was apparent (grand mean±se=0.25±0.037 g WM/bite).

6 247 Table 2 Effect of feeding motivation and parasitic status on sheep bite rate (bites/s) whilst grazing swards varying in N content, height and level of faecal contamination. Values are means. SED for choice comparisions=0.149, parasite comparisions=0.070 and feeding motivation comparisions= Treatment effects are given under P (parasitic effect), FM (feeding motivation effect) and (P FM) for their interaction. For other abbreviations, see Table 1 Choice 60% To appetite To appetite Overall Treatment effects Sward 1 vs. sward 2 means Non Parasitised Non Parasitised P FM P*FM (1) N H+F vs. N+H F * NS NS (2) N H+F+ vs. N+H F NS * * (3) N H+F vs. N+H F NS * NS (4) N H+F+ vs. N+H F NS * * (5) N H F vs. N+H+F NS NS NS (6) N+H F vs. N+H+F NS NS NS (7) N H+F vs. N+H+F NS NS * Overall means *P<0.05, NS=P>0.05 Table 3 Effect of feeding motivation and parasitic status on sheep grazing height above the soil surface (cm) whilst grazing swards varying in N content, height and level of faecal contamination. Values are means for sward 1 from each choice. SED for choice comparisions=0.529, parasite comparisions=0.379 and feeding motivation comparisions= For abbreviations, see Tables 1and 2 Choice 60% To appetite To appetite Overall Treatment effects Sward 1 vs. sward 2 means Non Parasitised Non Parasitised P FM P*FM (1) N H+F vs. N+H F * NS NS (2) N H+F+ vs. N+H F NS NS * (3) N H+F vs. N+H F NS * * (4) N H+F+ vs. N+H F * NS * (5) N H F vs. N+H+F NS * * (6) N+H F vs. N+H+F NS NS * (7) N H+F vs. N+H+F NS NS * Overall means *P<0.05, NS=P>0.05 Table 4 Effect of feeding motivation and parasitic status on sheep grazing height above the soil surface (cm) whilst grazing swards varying in N content, height and level of faecal contamination. Values are means for sward 2 from each choice. SEDs for choice comparisions=0.516, parasite comparisions=0.375 and feeding motivation comparisions= For abbreviations, see Tables 1 and 2 Choice 60% To appetite To appetite Overall Treatment effects Sward 1 vs. Sward 2 means Non Parasitised Non Parasitised P FM P*FM (1) N H+F vs. N+H F NS NS * (2) N H+F+ vs. N+H F NS NS NS (3) N H+F vs. N+H F NS NS NS (4) N H+F+ vs. N+H F NS NS NS (5) N H F vs. N+H+F * * NS (6) N+H F vs. N+H+F NS NS * (7) N H+F vs. N+H+F NS NS * Overall means *P<0.05, NS=P>0.05 Discussion The aims of this experiment were to determine the rules of sward selection used by mammalian herbivores and the effect of level of feeding motivation and parasitic status on these rules. In order to achieve these aims it was necessary to ensure that the different sward treatments used in the experiment could be distinguished by the animals. Sward height and N content were confounded when comparing N contents of the total (i.e. cut to the soil surface) sward samples with tall swards also having higher N contents than short swards. This effect was cre-

7 248 ated through the uneven distribution of N in grass plants. Grass laminae contain greater N concentrations than stems, and there is a gradient of increasing concentrations of N from the basal to the distal part of the lamina (Tallowin and Brookman 1996). A consistent sward cutting height across all sward treatments created a similar pseudostem height across all treatments (Haynes and Williams 1993). The total sward samples from short swards, therefore, contained a greater proportion of stem compared to the total sward samples from the tall swards, thereby explaining the lower N content in short compared to tall sward samples. The potential cue used by herbivores to determine sward N content (i.e. sward tone; Bazely and Ensor 1989) was also confounded with sward height, with the total samples of the tall swards being darker than the total samples of the short swards. However, both the N content and the sward tone of the portion of swards consumed by the animals (i.e. bitedepth sward samples) were not confounded with sward height. Herbivores are likely to base their grazing behaviour on the visual cues associated with the sward surface (i.e. that part of the sward visible during grazing; Hutchings et al. 1999). The bite-depth sward samples are, by definition, more indicative of the surface of the sward compared to the total sward samples. This suggests that the cue used by the animals to determine the N content of swards (sward tone) was similar across both short and tall sward treatments, but differed between the N-rich and N-poor sward treatments. Sward N treatment also affected their OMD value and their DM content. However, a herbivore s perception of the nutritional value of food is unlikely to be instantaneous since it cannot recognise most nutrients on consumption (Arnold 1981) and nutrient absorption may be delayed for several hours. It is, therefore, unlikely that animals based their sward selection or their grazing strategy on the OMD values of the swards, although changes in tensile strength of the tillers through the relationship between neutral detergent fibre and digestibility cannot be ruled out (Wright and Illius 1995). The differences in sward-n and sward-height treatments were of a similar magnitude; a two-fold difference in height between the tall (12 cm) and the short (6 cm) sward treatments and approximately a two-fold difference in N content between the N-rich (24.4 g/kg DM) and the N-poor (13.6 g/kg DM) sward treatments. The commensurate difference between the two sward N treatments and the two sward heights suggested that their effects on sheep grazing behaviour could be directly compared (i.e. the effect of a doubling of sward N content versus the effect of a doubling of sward height). The relative roles of sward height, sward N content and level of faecal contamination of swards in determining sheep grazing behaviour could then be quantified. The grazing decisions of all sheep, irrespective of physiological state, were dominated by the attraction to grazing tall swards. In choices 1 6, a tall sward was paired with a short sward. The strong and significant selection for the tall sward trays in these six choices clearly shows that animals of all the treatments preferred the tall swards offering increased herbage intake rates over the short swards (Black and Kenney 1984; Arnold 1987; Bazely 1990). However, adding faeces to tall swards and/or fertilising short swards in choices 1 6 did significantly affect the degree of selection for tall swards; both acted to reduce selection. In this experiment a doubling of sward height had far more impact on herbivore foraging decisions in relation to faecal contaminated swards, than a doubling of sward N content. All else being equal, a doubling of sward height transformed the strong inherent avoidance of faecal contaminated swards (Hutchings et al. 1998) into strong selection for tall, faecal contaminated swards. In comparison, a doubling of sward N content could only transform the inherent avoidance of faecal contaminated swards into an indifference to N-rich, faecal contaminated swards. This strong selection for tall swards is consistent with existing literature (Black and Kenney 1984; Arnold 1987; Bazely 1990) and not surprising as sheep maintained greater N intake rates when grazing N-poor, tall swards than when grazing N-rich, short swards. The greater N intake from grazing tall swards was largely achieved through greater bite mass of sheep grazing the tall sward and the greater DM content of the tall sward compared to the N-rich sward of choice 1. In addition, tall swards were grazed further from the soil surface compared to short swards. The increased N intake rate from grazing tall swards was, therefore, maintained whilst avoiding the majority of infective parasitic larvae which concentrate in the lower portion of the sward (Sykes 1987). This suggests that the benefits of grazing tall, N-poor swards can outweigh the costs of parasitism from faecal contaminated swards, whereas the benefits of grazing N-rich, short swards cannot. In order to determine whether the trade-off choices, created through faecal contamination of swards (choices 5, 6 and 7), used in this experiment are consistent with those occurring in natural environments it is first necessary to determine whether the sward treatments used in this experiment could occur in natural systems. For this to be the case not only must the effects of faecal deposition on the N content, parasite burden and height of the surrounding pasture in natural systems be equivalent to the sward treatments used here, but they must also coincide in time to create the trade-off choices. The fertilising effect of the deposition of faeces on pasture may result in up to 1.7 times more N in contaminated swards, compared to non-contaminated swards, 4 weeks post faecal deposition (i.e. a similar difference in N content between the N-rich and N-poor swards in this experiment) (Jorgensen and Jensen 1996). O. circumcincta eggs in faeces deposited on pasture take approximately 3 weeks to develop into infective-stage larvae (Coop et al. 1982). It is likely, therefore, that the timing of the development of parasites on swards and the uptake of nutrients from faeces coincide in natural systems, thus creating the trade-off. The relative heights of faecal contaminated and noncontaminated swards in natural systems are a function of

8 the overall level of faecal avoidance, which is determined by the amount of faeces and the age of the faeces (Hutchings et al. 1998). Both the amount and the age of faeces affect the olfactory stimulus of sheep, with increased avoidance of swards contaminated with greater amounts of fresher faeces (Hutchings et al. 1998). Avoidance of 3-week-old faeces (infective-stage larvae present) on swards by herbivores is significantly less than that of fresh faeces (Hutchings et al. 1998). In ideal conditions perennial rye grass growth peaks at about 5 cm/day (Robson et al. 1989). Taking the average growth rate as 2.5 cm/day, in 3 weeks a sward height difference of 52.5 cm could exist between heavily grazed and completely avoided swards. This suggests that in nutrient-limiting systems, the sward height differences used in this experiment could occur and coincide with the presence of parasitic larvae in natural systems. However, due to the huge potential for sward growth, swards contaminated with faeces are likely to be grazed within the 3-week period of parasite development. As shown in this and previous studies (Hutchings and Harris 1997; Hutchings et al. 1998, 1999), herbivores selectively graze the upper portion of swards contaminated with faeces. Faecal contaminated swards then remain tall in relation to the surrounding non-contaminated swards, even after being grazed (Hutchings and Harris 1997). It is, therefore, likely that 3 4 weeks after the deposition of dung, herbivores in nutrient-limiting natural systems face the trade-off between the benefits of ingestion of N-rich and tall swards offering increased nutrient intake rates and the costs of parasite intake, similar to that used in this experiment (choice 5: N+H+F+ vs. N H F ). The findings of this experiment are, therefore, likely to hold true and hence be of relevance in natural systems. Dung may be a source of infective stage larvae on swards for up to 12 months (Pandey 1972). Dung patches affect the N content of the surrounding sward for up to 24 months post faecal deposition (Haynes and Williams 1993). Trade-offs between nutrient intake rates and parasite intake (simulated in choice 5), are, therefore, likely to occur in natural systems in the relatively short term (3 4 weeks post faecal deposition) and persist over the long term (up to 12 months post faecal deposition). However, the average length of time for dung to persist on pasture, before complete decomposition, is 4.4 months (Haynes and Williams 1993). The stimulus used by herbivores to avoid parasites would then effectively be removed by 4.4 months, whilst the sward would remain relatively N-rich and tall, and thus attractive for grazing after this time. The trade-off between nutrient intake rate and parasite intake would continue to exist from the time of the disappearance of the faeces until the end of the survival time of the parasitic larvae. However, the herbivores would not be able to determine the presence of the trade-off due to the lack of a cue to the presence of parasites (Cooper 1997). This may result in the ingestion of infective stage larvae remaining on pasture from 4 to 12 months post faecal deposition. 249 Whilst the N+H+F+ vs. N H F trade-off choice (choice 5) is likely to occur in nutrient-limiting natural systems, other trade-off choices (choices 6 and 7) may also occur in natural systems. For example, trade-offs between N and parasite intake may perhaps occur under conditions of high grazing pressure where sward heights are kept permanently low, thus creating the N+H F+ vs. N H F choice. However, it is likely that the choice simulated by choice 5 of this experiment (N+H+F+ vs. N H F ) is most likely to be created through faecal contamination of pasture in nutrient-limiting systems typical of many extensive rangelands. The grazing decisions of herbivores faced with this trade-off will be determined not only by sward characteristics, which are indicative of the costs and benefits of the trade-off, but also the physiological state of the herbivore (feeding motivation and parasitic status) which may affect their motivation to feed and/or the risks associated with parasite intake (Hutchings et al. 1998, 1999). The physiological state of the sheep did not affect their sward selection in any of the choices, except choice 7. When faced with choice 7, which effectively acted as a N trade-off where sward height was controlled (both swards being tall) leaving N F vs. N+F+, different animal treatments selected significantly different sward treatments. When presented with choice 7, animals adopted a reduced parasite-risk grazing strategy (i.e. increased selectivity for the N F sward) which suggests that the risks of grazing faecal contaminated pasture outweigh the advantages of increased N intake for this group of animals and that sub-clinical parasitism increases faecal avoidance in herbivores (Hutchings et al. 1998). Although sward selection was only affected by animal physiological state in choice 7, the grazing behaviour (bite rate and grazing depth) of sheep was affected by both feeding motivation and parasitic status across all choices. It was the interaction between feeding motivation and parasitic status which dominated the statistical analyses. Parasitism in animals with a moderate feeding motivation resulted in reduced bite rates and grazing depths, which would have reduced the rate of ingestion of parasitic larvae as the majority of parasites concentrate in the lower portion of the sward (Sykes 1987). In contrast, all highly feeding-motivated animals (including those which were ) had increased bite rates and grazing depths which would have increased the rate of ingestion of parasites. This suggests that the need to feed in previously feed-restricted sheep with an existing parasite burden outweighed the risks of further parasite ingestion. Since the upper portion of the sward contains the highest levels of N (Haynes and Williams 1993) the reduction in grazing depths by, moderately feeding-motivated sheep would also result in them taking the most nutritious portion of the sward. A fact borne out in the results of this experiment, where sward N content was markedly higher in the portion of the sward consumed by the sheep during grazing compared with the total overall sward N levels. Herbivores generally reduce their grazing depths when grazing faecal contami-

9 250 nated pasture (Hutchings and Harris 1997; Hutchings et al. 1998, 1999). Despite the increased avoidance of faeces in, moderately feeding-motivated sheep, they still took an average of 46% of their bites from the faecal contaminated sward (N+F+) of the N trade-off choice (choice 7). The fact that, moderately feeding-motivated sheep grazed the upper portion of the N+F+ (parasite contaminated) sward when N F (noncontaminated) swards were available suggests that they were attempting to minimise parasite intake whilst maximising nutrient intake. Level of feeding motivation had no effect on sward selection irrespective of choice type. Increased feeding motivation has been shown to be associated with reduced faecal avoidance in grazing herbivores (Hutchings et al. 1999). The reason for the lack of a feeding-motivation effect in this experiment is unclear, however, it may be a reflection of the high level of selection for tall swards. For example, in previous experiments (Hutchings et al. 1999) where N trade-offs were presented to herbivores (i.e. N+F+ vs. N F choices) swards of 6 cm were used when creating the trade-off between N and parasite intake, whereas in this experiment the comparable choice (choice 7) used swards of 12 cm (the tall-sward treatment). As discussed above, sheep have more scope to maximise the nutritional advantages whilst minimising the parasitic costs associated with grazing tall rather than short swards. Sheep with a high feeding motivation were expected to show reduced faecal avoidance compared to the moderately motivated animals. As the moderately feeding-motivated sheep already strongly selected the faecal contaminated tall swards (a reflection of the height of the sward) it was unlikely that highly feedingmotivated sheep could select them to a significantly greater extent. The advantages associated with grazing tall swards are, therefore, likely to account for the lack of a feeding-motivation effect on sward selection in this experiment. This experiment has shown that the inclusion of sward characteristics and animal physiological state in a trade-off framework has the potential to provide a better understanding of what drives the contact process between herbivores and their parasites. The traditional assumption in optimal foraging models has been that maximisation of energy or N intake is the primary goal of foragers (Stephens and Krebs 1986). Other factors, such as the risk of predation (Abrahams and Dill 1989; Lima and Dill 1990), have only recently been incorporated into optimal foraging models to improve their predictive powers. The possible benefits of inclusion of parasite factors in optimal foraging frameworks has recently been discussed (Norris and Johnstone 1998; Hutchings et al. 1999). This and other recent experiments have shown that both parasitism of individuals and parasitic contamination of pasture via faeces affects herbivore grazing behaviour and sward selection (Hutchings et al. 1998, 1999). The effects of faeces and parasites on optimal foraging have been largely ignored but are likely to be significant (Lozano 1991). For example, at the end of a grazing season 2 4% of pasture can be covered by cattle faeces (Phillips 1993) and as, on average, each faecal deposit causes an area of herbage 6 times its own area to be avoided (Phillips 1993), up to 24% of pasture may, therefore, be avoided by grazing animals due to its association with faeces. This gives an indication of the scale of impact that faeces and thus parasite dispersion have on herbage utilisation and thus optimal foraging strategies of herbivores. The quantification of the various effects of faecal deposition on sward characteristics within a trade-off framework may, therefore, enhance the predictive power of optimal foraging models. It is clear from this and previous work that animal state impacts on herbivore foraging decisions, and thus plays an important role in determining herbivore grazing strategies and thus rates of parasite intake and output in faeces (Hutchings et al. 1998, 1999). The foraging decisions of herbivores faced with trade-offs between nutrition and parasitism are central to many questions on both parasitic infection rates in herbivores, parasite contamination levels on pasture and the knock-on consequences on animal fitness. In conclusion, when comparing commensurate differences in sward height and sward N content, sward height is the dominant cue which determines sheep foraging decisions. Herbivores may face trade-offs between nutrition and parasitism, and their grazing decisions are determined by both sward characteristics and their physiological state. The use of trade-off frameworks and the inclusion of costs and benefits determined from animal and plant factors associated with grazing trade-offs between nutrient and parasite intake are likely to greatly enhance the predictive power of optimal foraging models. Acknowledgements The authors would like to thank the technical team led by Dave Anderson, including Terry McHale, Leslie Deans, Sheila Young and Joan Chirnside. We would like to thank the Greek Mafia for saving our swards and we would also like to acknowledge Elizabeth Jackson and Dave McBean for parasitological technical support, Elizabeth Austin of Biomathematics and Statistics Scotland for statistical support and Bob Coop for useful discussions and comments on previous drafts of this manuscript. TGP was funded by a Royal Society visiting fellowship. The project was in part funded by the Natural Environment Research Council. The Scottish Agricultural College, Macaulay Land Use Research Institute and Moredun Research Institute receive support from the Scottish Executive, Rural Affairs Department. References Abrahams MV, Dill L (1989) A determination of the energetic equivalence of the risk of predation. Ecology 70: Alexander RN, McGowan M (1966) The routine determination of in vitro digestibility of organic matter in forages an investigation of the problems associated with continuous large scale operations. J Br Grassl Soc 21: Anderson RM (1978) The regulation of host population growth by parasite species. Parasitology 76: Arnold GE (1981) Grazing behaviour. In: Morely FHW (ed) Grazing animals. Elsevier, Amsterdam, pp Arnold GE (1987) Influence of the biomass, botanical composition and sward height of annual pastures on foraging behaviour of sheep. J Appl Ecol 24: