GRAZING CHOICE BY SHEEP:CONTROLLED BY PLANT SPECIES OR HEIGHT? - G.T. Barthram^[44], E.I. Duff^[45] and G.R. Bolton^[46]

ABSTRACT

The aim of this study was to test the hypothesis that within a feeding station height and not other species characteristics, determines the grazing selection by sheep of plants of Lolium perenne, Poa annua and Poa trivialis. Species determined little of the variation in frequency of defoliation. Plants shorter than the surrounding feeding station were grazed less frequently but target plants that were taller than the rest of the feeding station were not preferentially grazed. The sheep showed a tendency to select uniform feeding stations. Thus, the results show that the probability of defoliation of the target plants is more strongly influenced by their height relative to their surroundings than by their absolute height.

INTRODUCTION

The management control which is applied to extensively managed pastures should be based on an understanding of ecological processes (McIvor, 1993). Differences in stock density alter the competitive interactions among species of plants, through differences in defoliation intensity and through differences in selective defoliation. These effects are rapidly expressed in changes in tiller demography and population structure (Briske and Silvertown, 1993), so that the composition of a plant community nearly always reflects its history of herbivory (Edwards and Gilman, 1987). Selective defoliation of species can operate at the community level (Milne, 1991) but there is equivocal evidence (Newman et al., 1992) that sheep may select for species in an intimate mixture (Trifolium repens grown with Lolium perenne). Selection can also arise as a result of differences in the heights of sward components; sheep tend to select the taller tillers in a pasture (Barthram and Grant, 1984). The aim of this study was to test the hypothesis that within a feeding station height, and not other species characteristics, determines the grazing selection by sheep of plants of L. perenne, Poa annua and Poa trivialis.

MATERIALS AND METHODS

Measurements were made in artificial feeding stations, or patches, which were transplanted in July and September into swards maintained with surface heights of 4 and 8 cm, respectively.

Sward heights were maintained by the weekly adjustment of the numbers of dry yearling sheep. There were two replicates of each sward that were dominated by L. perenne and P. annua, respectively. Before transplanting, the patches were prepared in 17 x 20 cm² pots that contained one target plant of each of L. perenne (cv. Magella), P. annua (un-named) and P. trivialis (cv. Dasas), surrounded by plants of L. perenne and P. annua. There was one target plant at each apex of an equilateral triangle with 10 cm sides centred in the middle of each pot. All the target plants in each patch destined for the 4 cm swards were cut weekly to heights of either 2, 4 or 6 cm (shorter, level and taller height treatments) and for each of these treatments the surrounding plants were also cut weekly to 2, 4 or 6 cm to create nine different treatments. The target and surrounding plants in the patches destined for 8 cm swards were similarly cut, but at 6, 8 or 10 cm (also called shorter, level and taller) once every fortnight. When moved into the swards each cutting set of nine patches was located in a transect with 2 m between each patch. There were two replicates of each transect in each sward replicate.

Two tillers on each target plant were marked with fine wire and after three days exposure to the grazers the number of grazed leaves on each tiller was recorded. Selection was assessed by comparing the proportion of grazed leaves per marked tiller, the proportion of marked tillers grazed per plant and the proportion of grazed target plants per patch (assuming that the defoliation status of the marked tillers represents the defoliation status of the plant).

The data were analysed by generalized linear modelling with a binomial error distribution. The results from this model were then used to produce a hierarchical ANOVA-type table which estimated the significance of treatment effects. Standard errors were calculated from the original data by averaging to the appropriate level of replication according to the hierarchy of the analysis and then calculating standard errors from these averages.

RESULTS

Differences between species only explained a significant (P<0.05) amount of variation in the proportion of leaves grazed per tiller. The proportions of leaves grazed per P. trivialis, P. annua and L. perenne tiller were 0.527 (se 0.034), 0.458 (se 0.034) and 0.438 (se 0.032), respectively.

The height treatments explained most of the variation. Averaged across patch heights, shorter target plants were selected least (Table 1). A lower proportion of the taller target plants were grazed in the 8 cm rather than in the 4 cm sward. For the shorter and level target plants the proportion of leaves grazed was greatest when the target and patch heights were the same. There was no evidence from any of the varieties that target plants taller than the rest of the patch were selected.

DISCUSSION

There was no evidence that sheep simply selected the tallest plants. There was some evidence that the sheep selected target plants which occurred within patches of uniform height. Together with the low level of selection for plants shorter than their surroundings these results show that the probability of defoliation of the target plants is more strongly influenced by the height of their surroundings than by their absolute height. The simplest explanation for this behaviour is that the sheep were in some way averaging the height of the patch and grazing all plants similarly. Any apparent selection for or against plants of different height was then a passive effect.

The results show that differences in the probability of grazing of the three grass species are determined mainly by differences in height. Swards maintained at different heights can show differences in the proportions of these species (C.A. Marriott, personal communication). If these differences in proportion are the result of differences in defoliation frequency, then this difference in defoliation is a result mainly of differences in the height of the three species relative to their surroundings, rather than of any other species characteristic which the grazing sheep might respond to.

Table 1. Proportions of grazed leaves, tillers and plants. Averages of two measurement periods. Shorter, level and taller heights are relative to the sward heights. Standard errors of means are shown in brackets. “Significance” is the level of significance of treatment effects from the ANOVA-type table (ns=P>0.05, *=P<0.05, **=P<0.01, ***=P<0.001).

ACKNOWLEDGEMENTS

We thank David Elston for his help and advice. This work was funded by the Scottish Office Agriculture, Environment and Fisheries Department.

REFERENCES

Barthram, G.T. & Grant, S.A. 1984. Defoliation of ryegrass-dominated swards by sheep. Grass and Forage Science 39, 211-219.

Briske, D.D. & Silvertown, J.W. 1993. Plant demography and grassland community balance: the contribution of population regulation mechanisms. Proceedings of the XVII International Grasslands Congress, Palmerston North, New Zealand, New Zealand Grassland Association, Palmerston North, Vol. 1, pp. 291-298.

Edwards, P.J. & Gilman M.P. 1987. Herbivores and plant succession. In: Gray, A.J., Crawley, M.J. and Edwards, P.J. (Eds) Colonisation Succession and Stability. Blackwell. pp. 295-314.

Mcivor, J.G. 1993. Distribution and abundance of plant species in pastures and rangelands. Proceedings of the XVII International Grasslands Congress, Palmerston North, New Zealand, New Zealand Grassland Association, Palmerston North, Vol. 1, pp. 285-290.

Milne, J. 1991. Diet selection by grazing animals. Proceedings of the Nutrition Society 50, 7785.

Newman, J.A., Parsons, A.J. & Harvey, A. 1992. Not all sheep prefer clover: diet selection revisited. Journal of Agricultural Science, Cambridge 119, 275-283.