SEARCH

SEARCH BY CITATION

Keywords:

  • acid grassland;
  • initial composition;
  • productivity;
  • sustainable management

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

1.Agrostis capillaris–Festuca ovina-dominated communities are widespread in the uplands of Great Britain. They are agriculturally productive but little is known about how to manage this community for specific goals. Vegetation trajectories were examined in this plant community under different sheep grazing management regimes at two sites in Scotland. One site had a substantial presence of moorland species, the other was characterized by a more productive vegetation. Management consisted of maintaining sward heights of 3, 4·5 or 6 cm during the growing season, or complete exclusion of grazing stock.

2. Changes in species composition were small over the 7 years of the experiment. Few species invaded or were lost during the course of the study. The observed changes were largely as a result of shifts in abundance of the dominant species.

3. Maintenance of sward height at low levels (3 or 4·5 cm) during the growing season resulted in the spread of Nardus stricta where present. Where N. stricta was absent, the sward developed a higher content of mosses, specifically Hypnum jutlandicum and Rhytidiadelphus squarrosus.

4. Removal of grazing resulted in an increase of cover of grazing-intolerant species, such as Deschampsia flexuosa and Molinia caerulea, and in the cover of dwarf shrub species where present.

5. The two sites differed in the treatment that resulted in the smallest change in species composition. At the more productive site, maintenance of the sward at 4·5 cm resulted in the smallest overall change in species composition. At the less productive site, grazing the sward to 6 cm resulted in the smallest shift in vegetation composition. Grazing at this height appeared to prevent the spread of both M. caerulea and N. stricta.

6. The study demonstrates that sustainable grazing regimes for upland Agrostis–Festuca grasslands need to take into account both the initial composition of the vegetation, specifically the presence of species capable of replacing A. capillaris and F. ovina and of achieving dominance, and the overall productivity of the site.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The acidic grasslands dominated by the grasses Agrostis capillaris and Festuca ovina[nomenclature follows Stace (1991) for higher plants and Smith (1978) for bryophytes] make up the most extensive type of pasture on well-drained soils throughout the uplands and upland margins of Great Britain (Rodwell 1992). They are plagioclimax communities derived from woodland and heathland communities by a long and complex history of grazing (Ratcliffe 1977), and from other wetter vegetation types by hill drainage and burning (Rodwell 1992). Recently the principal large herbivore has been breeding sheep, although previously cattle and wether sheep would have been the dominant grazing animals. The plant community, Festuca ovina–Agrostis capillaris–Galium saxatile grassland (U4 of the British National Vegetation Classification) as defined by Rodwell (1992), is typically species poor, with few other grass species or herbs but with a frequent and often high cover of mosses. These grasslands are threatened by inappropriate grazing regimes that result in degradation, abandonment, bracken encroachment or intensification (UK Biodiversity Steering Group 1995).

This grassland vegetation type is of considerable agricultural importance. This is in part because of the large area it covers and its wide distribution within Great Britain, and in part because of its productivity and nutritive value to grazing livestock, such that it can support a higher intensity of grazing than other vegetation in the uplands (Hunter 1962). However, under inappropriate grazing, sites can become dominated by the grass Nardus stricta, a change also attributed to the effects of changes in agricultural production systems. Between the 16th and 19th centuries there was a reduction in cattle numbers, a disappearance of wether sheep and an increase in breeding ewes producing store lambs. Ewes are more selective than either cattle or wether sheep, preferring other grasses to N. stricta (Grant et al. 1996b). It has also been proposed that the general increase in N. stricta is the long-term result of continuous high grazing pressures (Ratcliffe 1959), particularly by sheep. Under much reduced grazing pressures, succession results in the invasion of bracken, dwarf shrubs and, where seed is available, the development of scrub or woodland or, on wetter sites, the grass Molinia caerulea (Miles 1988; Rodwell 1992). As in many other grazing systems, long-term sustainability, both agricultural and ecological, is the result of maintaining an appropriate species balance and preventing the expansion of less desirable species.

Despite the agricultural importance of this grassland type, there is little information on the levels of grazing that are required to meet a specific management objective for this type of vegetation. This parallels the shortage of information for other British semi-natural grassland systems, although this has been dealt with to some extent by Grant et al. (1996a, b) for M. caerulea- and N. stricta-dominated grasslands, respectively, and by Bullock et al. (1994) for lowland grassland. The work reported here formed part of a study that investigated the effects of varying sheep grazing intensity on this community. Three treatments were employed. One treatment was set at a level thought to represent ‘normal’ grazing management, the other two were set to create conditions of heavy and light grazing, respectively. A subsequent paper will report on diet selection by sheep in the treatments employed. The intensity of grazing was manipulated by the maintenance of the sward at set heights by the addition or removal of animals. This method of grazing manipulation affects not only competitive interactions between plants, but also the ability of the grazing animal to forage selectively. It was chosen as it maintained a relatively constant treatment, as opposed to setting a stocking density, where competition between plants and foraging selectivity alters through the growing season. It was easily monitored and mimics the decisions made by land managers.

The experiment was duplicated at two sites to represent different parts of the spectrum covered by this community. One site (Kirkton) had a substantial initial cover of both M. caerulea and N. stricta, both species that may have a considerable influence on community dynamics, as well as other species typical of moorland habitats. The other site (Cleish) had a less diverse sward that included a higher contribution of more productive species such as Festuca rubra, less N. stricta and no M. caerulea. The variation in response between the two sites should improve the applicability of the results to the sustainable management of this community in different situations. It is increasingly important to understand how grazing management affects community composition as considerable areas of land are now entering agri-environment schemes (MAFF 1986) where the land has to support two objectives (Walker 1997). They must remain agriculturally sustainable to enable an economic return to be made, and ecologically sustainable to fulfil the conditions of the scheme. This paper addresses the sustainable management of an agriculturally important and widespread community for which there is little quantitative information available.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Sites

The Cleish site was located at 240–250 m on the Cleish Hills of Fife, Scotland (National Grid Reference NT080936, 56°08′N 3°29′W), on a freely drained brown earth soil. The site had previously been used in a study of grazing selection by sheep (Grant et al. 1985), during which the site had been managed by grazing cattle and sheep to maintain herbage utilization at moderate, characteristic levels (20–35%). The Kirkton site was located at 410–445 m on the Scottish Agricultural College's Kirkton Farm near Crianlarich, Scotland (NGR NN360310, 56°27′N 4°39′W), on a podsolic gley with a patchy, peaty top and frequent protruding boulders. Prior to 1989, the site had been used regularly by sheep for short periods around tupping, lambing and other gathering times, equivalent to a year-round stocking rate of 0·7 ewes ha–1 (J. Wyllie, personal communication).

At each site, six plots of 0·3 ha, in two blocks of three plots, were fenced. In order to describe the effects of no grazing, two 5 m × 5 m exclosure cages were sited in each plot to prevent livestock grazing. Vegetation measurements at both sites began in June 1989, with fences installed subsequent to this. The grazing treatments commenced at Cleish on 2 August 1989, and at Kirkton on 11 May 1990. Thus at both sites, differences between the vegetation recorded in 1989 and that in 1990 reflect only a part of a year's treatment. Herbage mass was also assessed. However, due to the difficulty of assessing where litter and shoot material graded into soil or root material, respectively, the estimates were highly variable between different years and observers. This measure was therefore not considered further in the analysis.

Treatments

Three treatments were imposed randomly within each block. Sward heights were maintained in the ranges 3·0–3·5 cm, 4·5–5·0 cm and 6·0–6·5 cm, respectively. These treatments are referred to subsequently as 3 cm, 4·5 cm and 6 cm. This was achieved by stocking plots with predominantly Scottish Blackface wether sheep during the growing season (May–October). Treatment heights were maintained by regular (approximately weekly) measurements of sward surface height and adjusting animal numbers as necessary. Forty random measurements of sward surface height were taken per plot using a HFRO sward stick (Barthram 1986).

Floristic measurements

Floristic composition was recorded annually in June from 1989 to 1995, except in 1994. The Cleish site was recorded during the first 2 weeks of June, the Kirkton site in the following 2 weeks. Floristic composition was estimated by recording contact with an inclined point quadrat (32·5° to the horizontal) (Grant et al. 1985). Twenty quadrat positions were sited in a grid pattern at regular intervals along four or five transects (depending on plot shape) and were relocated by measurement from the perimeter fences. All point contacts with vegetation were recorded to ground level by species (live and dead). A variable number of pin traverses (always completed to ground) were made at each quadrat location until a minimum of 25 contacts were recorded, to give comparable numbers of contacts across the height treatments (i.e. a minimum of 500 contacts per plot). A minimum of 100 contacts was recorded along transects in each exclosure. The data used in all analyses were percentage ground cover of each species.

Statistical analysis

Detrended correspondence analysis (Hill 1979) of the cover data was used to examine the association between time since treatment imposition and changes in species cover. All ordination was carried out using the version of decorana as amended by Oksanen & Minchin (1997).

Differences in individual species cover of the most abundant species at each site were analysed using a repeated measures analysis of variance using genstat (GENSTAT V Committee 1987) on square-root transformed data. All analyses were also carried out on relative frequency data, but these and the corresponding data are not shown as they yielded the same patterns of change and very similar statistical results.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Cleish

Treatment imposition

Sward heights were generally kept within the limits set out at the start of the experiment (Fig. 1a). The major problem with the technique of adjusting stocking density to achieve a set sward height is its reactive nature. Changes in stocking rate were based on the change in sward height since the previous measurement and an anticipation of the likely growth in the following week, and hence could not react to sudden changes in sward growth as a result of rapid changes in weather. Also, at times, stocking rates were such that the addition of an extra sheep meant a substantial increase in grazing pressure, so that flexibility in stocking small areas did not reflect that possible on large unenclosed pastures. In particular, problems were encountered in maintaining the 3-cm treatment as care had to be taken to prevent the sward getting too far below 3 cm such that the sheep present could maintain their intake without losing body condition. Problems were also encountered in maintaining the 6-cm treatment in some years when the growth of the sward was slow as a result of dry weather.

image

Figure 1. (a) Recorded sward heights and (b) sheep grazing days ha–1 (+ 1 SE) in the three grazing treatments (3 cm, 4·5 cm and 6 cm) at Cleish from 1989 to 1995.

Download figure to PowerPoint

Large numbers of annual sheep grazing days were necessary to achieve the target for the 3-cm treatment in the first 2 years of the experiment (Fig. 1b). In subsequent years, the differences between treatments were smaller. Little year-to-year variation was evident in sheep numbers needed to achieve the 6-cm treatment. The other two treatments appeared more variable in the sheep numbers needed to achieve the set heights.

Community dynamics

Detrended correspondence analysis of the data from Cleish showed species associated with shorter vegetation to the left of axis 1 (e.g. Hypnum jutlandicum and Trifolium repens) and taller growing species (e.g. the grasses Deschampsia flexuosa and Poa spp., mainly P. pratensis) arrayed to the right (Fig. 2a). Correlation between site scores and cover values showed that D. flexuosa, Potentilla erecta and Poa spp. were highly positively correlated (r = 0·86, 0·83 and 0·78, respectively, all n = 46 and P < 0·001), and Rhytidiadelphus squarrosus, T. repens and Luzula multiflora highly negatively correlated (r = –0·74, –0·67 and –0·63, respectively, all n = 46 and P < 0·001). In contrast, axis 2 separated species associated with more productive sites (e.g. Rumex acetosa and F. rubra), situated towards the bottom of axis 2, from species associated more with upland/moorland vegetation types towards the top. Species positively correlated with axis 2 included Pleurozium schreberi, Carex spp. and N. stricta (r = 0·62, 0·59 and 0·58, respectively, all n = 46 and P < 0·001), and the most significantly negatively correlated species was R. acetosa (r = –0·59, n = 46 and P < 0·001). However, all the more common species were concentrated towards the centre of the ordination. This and the relatively low eigenvalues (axis 1 = 0·10, axis 2 = 0·04) suggested that differences between plots and over time did not involve major shifts in species dominance or the loss or colonization of many species, i.e. a low β-diversity in the data set.

image

Figure 2. decorana ordination of vegetation data from Cleish. (a) Ordination of species; species codes for the common species only: Achimill, Achillea millefolium; Agrocapi, A. capillaris; Anthodor, Anthoxanthum odoratum; Conomaju, Conopodium majus; Descflex, Deschampsia flexuosa; Festovin, Festuca ovina; Festrubr, F. rubra; Galisaxa, Galium saxatile; Hypnjutl, Hypnum jutlandicum; Luzumult, Luzula multiflora; Poa spp., Poa species; Poteerec, Potentilla erecta; Rhytsqua, Rhytidiadelphus squarrosus; Rumeacet, Rumex acetosa; Trifrepe, Trifolium repens; Deadgras, grass litter. (b) Trends in species composition through time by each individual plot. Plots identified by treatment height code (3 = 3 cm, 4 = 4·5 cm, 6 = 6 cm or E = exclosure), block (a or b), and first and final year of recording (89 = 1989, 95 = 1995). No identifiers are present corresponding to E89a and b to indicate that these values are derived from mean values of the data from the other treatments.

Download figure to PowerPoint

The ordination of plots showed that there were initial differences between the vegetation in some of the plots (Fig. 2b). The plots assigned to the 3-cm and 6-cm treatments in block b had a higher initial cover of D. flexuosa and in consequence had a higher initial score on axis 1. All plots showed a movement up axis 2 through time, indicating that certain changes were occurring across all treatments. However, treatments varied considerably in the movement of plots relative to axis 1. Plots that received the 3-cm grazing treatment moved furthest left, and plots that received the 4·5-cm treatment also moved consistently but less dramatically left. As noted before, this was associated with an increase in the cover of species associated with short vegetation. The exclosure plots moved right along axis 1, indicating an increase in the cover of species associated with taller vegetation. The behaviour of plots in the 6-cm grazing treatment was less consistent, one moving further left than plots receiving the 4·5-cm treatment (although its initial vegetation had a higher compliment of less productive species) and the other plot moving slightly right. The overall movement shown by plot in the 4·5-cm treatment was smaller than that in the other treatments (Table 1).

Table 1.  Relative movement of plots in ordination space (axes 1 and 2) from 1989 to 1995 at Cleish. Units are standard deviations of species turnover
TreatmentTotal movement in 2-D space (± 1 SE)
3 cm0·52 ± 0·01
4·5 cm0·34 ± 0·01
6 cm0·46 ± (0·06
Exclosure0·7 ± 0·04

All samples were allocated to the U4 Festuca ovina–Agrostis capillaris–Galium saxatile grassland community of the National Vegetation Classification (Rodwell 1992) using the tablefit program (Hill 1993). Most had a close affinity to the typical subcommunity (U4a), which is recognized from Shetland to Cornwall, although later samples from the exlosures were closer to the Vaccinium myrtillus–Deschampia flexuosa subcommunity (U4e).

Individual species

The trajectories shown by the plots in ordination space are the results of changes in species composition. The dynamics of species that showed a significant response to treatment or a significant treatment–time interaction are displayed in Fig. 3, and the significance of this response, as well as that of all the common species, is shown in Table 2.

image

Figure 3. Trends in mean species cover (± 1 SE) at Cleish from 1989 to 1995. (a) Deschampsia flexuosa; (b) Festuca ovina; (c) Festuca rubra; (d) Luzula multiflora; (e) Poa spp.; (f) Potentilla erecta; (g) Trifolium repens; (h) total moss cover; (i) litter; and (j) total cover. Values for cover in the exclosures in 1989 are means derived from the data from the other treatments. Note different y-axes for different taxa.

Download figure to PowerPoint

Table 2.  Probabilities of treatment and time effects from repeated measures analysis of variance on the most common species (†) and selected other species and cover types at Cleish. NS equates to P > 0·05, *0·05 ≥ P > 0·01, ** 0·01 ≥ P > 0·001, ***0·001 ≥ P;ε (Greenhouse–Geissner epsilon) was used to multiply the degrees of freedom ( d.f.) for the Time and Treatment × Time comparisons
SpeciesTreatment (d.f. = 3,4)εTime (d.f. = 5,20)Treatment × Time (d.f. = 15,20)
Agrostis capillarisNS0·34*NS
Conopodium majusNS0·32*NS
Deschampsia flexuosaNS0·54**
Festuca ovinaNS0·45***
Festuca rubraNS0·51**
Galium saxatileNS0·40**NS
Hypnum jutlandicumNS0·32**NS
Luzula multiflora*0·65NS**
Poa spp. †*0·43NS*
Potentilla erecta**0·53NSNS
Rhytidiadelphus squarrosusNS0·36NSNS
Trifolium repens**0·50NS**
Dead grass**0·39****
Total moss coverNS0·51****

A number of general patterns were displayed. Some species showed a consistent behaviour across all treatments; A. capillaris declined in all treatments (76–58% over the course of the experiment), whereas Conopodium majus (1–5%), Galium saxatile (11–32%) and H. jutlandicum (3–23%) increased in all treatments. Other species were negatively affected by the elimination of grazing, and either increased in all grazing treatments, L. multiflora (Fig. 3d) and mosses taken as a single group (Fig. 3h), or only in the 3-cm treatment, T. repens (Fig. 3g).

Other species were favoured by an absence of or a low intensity of grazing. Poa spp., mainly P. pratensis (Fig. 3e), and P. erecta (Fig. 3f) increased only in the exclosures, and decreased where grazed, particularly P. erecta in the shorter swards. Deschampsia flexuosa (Fig. 3a) increased in the exclosures and to a lesser extent in the 6-cm treatment, whilst decreasing in the 3-cm and 4·5-cm treatments. Festuca ovina (Fig. 3b), in contrast, showed a higher cover in the 6-cm rather than the exclosure treatment, although not significantly so. Festuca rubra (Fig. 3c) showed a more complex behaviour. There was an initial increase in the exclosures, which was reversed latter, and a later increase in the 3-cm and 4·5-cm sward treatments that was maintained.

The accumulation of litter was highest in the exclosures, with cover positively correlated to sward height treatments (Fig. 3j). There was an overall increase in litter cover in all treatments, after an initial fall in cover in the 3-cm and 6-cm plots. The total cover of vegetation increased slightly between the beginning and the end of the experiment (Fig. 3k). However, there were initial falls in total vegetation cover in the grazed plots, which were partly a result of the decline in litter and A. capillaris.

Kirkton

Treatment imposition

As at Cleish, the effects of the treatments on sward height and number of sheep grazing days at Kirkton were well separated (Fig. 4a). However, sward heights in the 4·5-cm and 6-cm treatments were often below those sought. This partly reflected the short growing season at this site, which resulted in slow spring growth of the sward and an early reduction in growth in the autumn. The number of sheep grazing days was generally lower than at Cleish, as the community was less productive, so small proportionate adjustments in stock numbers were rarely possible.

image

Figure 4. (a) Recorded sward heights and (b) sheep grazing days ha–1 (+ 1 SE) in the three grazing treatments (3 cm, 4·5 cm and 6 cm) at Kirkton from 1989 to 1995.

Download figure to PowerPoint

The number of sheep grazing days was much lower than at Cleish to achieve the same sward height targets (Fig. 4b). Again, relatively high grazing pressures were necessary to achieve targets in the 3-cm and 4·5-cm treatments in the first year of treatment. Thereafter differences between treatments were less marked, although more grazing days were usually necessary to achieve the shorter sward heights.

Community dynamics

Ordination of the data from Kirkton showed species associated with moorland vegetation communities (M. caerulea and Narthecium ossifragum) at higher values along axis 1 (Fig. 5a) and species generally associated with grazed vegetation at lower values (A. capillaris and F. ovina). Species highly positively correlated with site scores on axis 1 included Erica cinerea, M. caerulea and E. tetralix (r = 0·79, 0·78 and 0·67, respectively, all n = 46 and P < 0·001), and species negatively correlated included Anthoxanthum odoratum, F. ovina, L. multiflora and A. capillaris (r = –0·79, –0·74, –0·73 and –0·72, respectively, all n = 46 and P < 0·001). The placement of species in ordination space along axis 2 may possibly have related to an association with vegetation height, taller plants or creeping plants at high values (M. caerulea, r = 0·65; P. erecta, r = 0·70, both with n = 46 and P < 0·001), or lower growing plants at low values (Sphagnum spp., r = –0·76, n = 46 and P < 0·001; N. stricta, r = –0·38, n = 46 and P < 0·01). As for the ordination of the Cleish data, the most common species were clumped tightly in the centre of the ordination, and the eigenvalues were relatively low (axis 1 = 0·10, axis 2 = 0·04). Again, differences over time and between plots did not reflect large shifts in species dominance.

image

Figure 5. decorana ordination of vegetation data from Kirkton. (a) Ordination of species; species codes for the common species only: Agrocani, Agrostis canina; Agrocapi, A. capillaris; Anthodor, Anthoxanthum odoratum; Carepani, Carex panicea; Carepilu, C. pilulifera; Dantdecu, Danthonia decumbens; Festovin, Festuca ovina; Festrubr, F. rubra; Galisaxa, Galium saxatile; Hylosple, Hylocomium splendens; Hypnjutl, Hypnum jutlandicum; Luzumult, Luzula multiflora; Molicaer, Molinia caerulea; Nardstri, Nardus stricta; Nartossi, Narthecium ossifragum; Pleuschr, Pluerozium schreberi; Poteerec, Potentilla erecta; Rhytsqua, Rhytidiadelphus squarrosus; Sphaspp., Sphagnum spp.; Triccesp, Trichophorum cespitosum; Deadgras, grass litter. (b) Trends in species composition through time by each individual plot. Plots identified by treatment height code (3 = 3 cm, 4 = 4·5 cm, 6 = 6 cm or E = exclosure), block (a or b), and year of recording (89 = 1989, etc.). No identifiers are present corresponding to E89a and b to indicate that these values are derived from mean values of the data from the other treatments.

Download figure to PowerPoint

As at Cleish, there were some differences between plots in the initial vegetation (Fig. 5b). One plot in one block (labelled 4a in Fig. 5b) had a much higher cover of M. caerulea. All the grazed plots showed little movement through time in relation to axis 1, but substantial movement along axis 2. This movement was highest in the plots grazed at 3 cm, and lowest in the plots grazed at 6 cm, and reflected an increase in dominance of N. stricta and a reduction in the cover of dead grass in particular. The exclosure plots showed a different pattern. They moved substantially along axis 1, but showed little movement along axis 2, reflecting an increased dominance of species associated with less heavily grazed moorland vegetation. Plots in the 6-cm treatment showed least change in position in ordination space over the course of the experiment (Table 3).

Table 3.  Relative movement of plots in ordination space (axes 1 and 2) from 1989 to 1995 at Kirkton. Units are standard deviations of species turnover
TreatmentTotal movement in 2-D space (± 1 SE)
3 cm0·44 ± 0·01
4·5 cm0·42 ± 0·13
6 cm0·21 ± 0·001
Exclosure0·63 ± 0·01

tablefit (Hill 1993) assessed most samples as borderline between the U4d (Festuca ovina–Agrostis capillaris–Galium saxatile, Luzula multiflora–Rhytidiadelphus loreus subcommunity) and the U5a (Nardus stricta–Galium saxatile species-poor subcommunity) grasslands of the National Vegetation Classification (Rodwell 1992). Both subcommunities have been recognized from Scotland, northern England and North Wales. However, plot 4a (4·5-cm treatment) was borderline between these two communities and M15d, Scirpus cespitosus–Erica tetralix wet heath, Vaccinium myrtillus subcommunity (Rodwell 1991). This plot had a higher initial cover of M. caerulea, N. ossifragum and Trichophorum cespitosum, as well as some Calluna vulgaris and E. tetralix, and hence started from a substantially different position in ordination space (shown as 4a in Fig. 5b).

Individual species

The dynamics of species that showed a significant response to treatment or a significant treatment–time interaction are displayed graphically (Fig. 6), and the significance of this response, as well as that of all the common species, is shown in Table 4.

image

Figure 6. Trends in mean species cover (± 1 SE) at Kirkton from 1989 to 1995. (a) Agrostis capillaris; (b) Anthoxanthum odoratum; (c) Carex pilulifera; (d) Molinia caerulea; (e) Nardus stricta; (f) Potentilla erecta; (g) Sphagnum spp.; (h) dwarf shrubs; (i) litter; (j) total cover. Values for cover in the exclosures in 1989 are means derived from the data from the other treatments. Note different y-axes for different taxa.

Download figure to PowerPoint

Table 4.  Probabilities of treatment and time effects from analysis of variance on the 12 most common species (†) and selected other species and cover types at Kirkton. NS equates to P > 0·05, *0·05 ≥ P > 0·01, ** 0·01 ≥ P > 0·001, *** 0·001 ≥ P;ε (Greenhouse–Geissner epsilon) was used to multiply the degrees of freedom ( d.f.) for the Time and Treatment × Time comparisons
SpeciesTreatment (d.f. = 3,4)εTime (d.f. = 5,20)Treatment × Time (d.f. = 15,20)
Agrostis caninaNS0·43*NS
Agrostis capillarisNS0·35***
Anthoxanthum odoratumNS0·36***
Carex piluliferaNS0·48***
Festuca ovinaNS0·51NSNS
Galium saxatileNS0·64**NS
Hylocomium splendensNS0·47*NS
Hypnum jutlandicumNS0·55***NS
Molinia caeruleaNS0·39NS**
Nardus strictaNS0·47**
Pleurozium schreberiNS0·43*NS
Potentilla erecta*0·39NSNS
Rhytidiadelphus squarrosus†NS0·48***NS
Sphagnum spp.*0·45NSNS
Trichophorum cespitosumNS0·39NSNS
Dead grass*0·47****
Dwarf shrubs*0·41*NS

A greater number of species than at Cleish showed a significant trend in time during the experiment without showing a significant treatment effect. Agrostis canina (5–12%), and the mosses Hylocomium splendens (2–6%), H. jutlandium (29–46%), Pleurozium schreberi (2–7%) and R. squarrosus (8–21%), all increased from 1989 to 1995. Total moss as a grouping behaved in the same way (45–92%). Galium saxatile initially increased in all treatments from 25% to 37%, but by the end of the experiment had declined again to values close to its initial level.

Two common species, F. ovina (24%) and T. cespitosum (12%), showed no response to treatment and did not change significantly over time. Two other species declined in all the treatments during the experiment, A. capillaris (Fig. 6a) and A. odoratum (Fig. 6b), but for both these species there were significant effects of treatment. Highest covers were maintained in the 6-cm treatment, and by the end of the experiment there had been a sizeable reduction (c. 85%) in A. capillaris and an apparent elimination of A. odoratum in the exclosures.

Species favoured by grazing included Carex pilulifera (Fig. 6c), N. stricta (Fig. 6e) and Spagnum spp. (Fig. 6f). Nardus stricta was favoured, and increased over time, by maintaining a short sward (3 cm or 4·5 cm), whereas both C. pilulifera and Sphagnum spp. were favoured by all the grazing treatments by the end of the experiment, despite both showing an initial increase in cover as a result of exclosure. A number of species were favoured when grazing was prevented: M. caerulea (Fig. 6d) and ericoid shrubs as a group (Fig. 6h). The latter group was dominated by E. tetralix, although on its own it did not show a significant response. Potentilla erecta was initially favoured by the absence of grazing, but later decreased in cover to similar levels to those shown in the grazed treatments (Fig. 6f).

As at Cleish, the cover of litter (Fig. 6i) was dramatically reduced in all treatments in the first year. Subsequently, it only partly recovered in the more heavily grazed treatments (3 cm and 4·5 cm), but substantially increased in the 6-cm treatment and the exclosures. Total cover increased over time in the 4·5-cm and 6-cm treatments, after an initial decline caused by the fall in litter cover (Fig. 6j). The total cover of plants in the 3-cm treatment was consistently lower than that of the other grazed treatments. The total cover of plants in the exclosures increased initially as many species responded to the removal of grazing. However, total cover then declined as many of the species declined in turn.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Changes in the grassland communities at these two sites have been relatively slow over the course of the experiments. This is in part due to the absence of any substantial invasion by new species, and the survival of almost all the original species in all treatments. The exceptions to this amongst the common species and species groups were few; A. odoratum was not detected in the Kirkton exclosures in 1995, dwarf shrubs were absent in the 3-cm plots at the same date, and at Cleish P. erecta was eliminated from plots grazed at 3 cm and 4·5 cm.

Despite the small changes in the species complement at both sites and the differences in species between sites, a number of patterns emerged from the data. However, a number of changes occurred across the treatments at each site, and such changes must be borne in mind when determining the actual response of the vegetation to the treatments. At Cleish there was a general decline in the importance of A. capillaris in all treatments accompanied by a general increase in G. saxatile. At Kirkton, there was an increase in a number of species and a decline in A. capillaris and A. odoratum, although the rate of decline differed between treatments.

Cleish

This site initially had a more agriculturally productive sward than that at Kirkton (cf. Figs 1b and 4b). The maintenance of a short sward height, 3 cm or 4·5 cm, resulted in an increase in dominance of F. rubra, L. multiflora, T. repens and total moss cover. Both the shorter sward treatments also resulted in a decline in cover of Poa spp. and D. flexuosa. Overall, these two treatments resulted in a more open, moss-rich sward, with a higher contribution of forb species. Less intense grazing, i.e. maintaining the sward at 6 cm, allowed the increase of D. flexuosa and F. ovina, but appeared to cause a reduction in the cover of F. rubra.

An absence of grazing again favoured taller growing species as well as more grazing-sensitive species such as D. flexuosa and Poa spp. The removal of grazing resulted in a decline in the cover of lower growing species and those species tolerant of high grazing intensities, including F. rubra, L. multiflora and total moss cover. Deschampsia flexuosa showed consistent increases in cover during long-term exclosure studies at five sites in the northern Pennines (Rawes 1981), while P. pratensis increased in exclosures set up on Rhum (Ball 1974).

Kirkton

This site had an initially higher content of moorland species than the vegetation at Cleish. The maintenance of a short sward, 3 cm or 4·5 cm, resulted in an increase in dominance of N. stricta and a decrease in representation in the sward of M. caerulea and dwarf shrubs. Thus the maintenance of a short sward resulted in the expansion of rejected species (N. stricta) and the reduction of species intolerant of grazing, especially summer grazing (e.g. M. caerulea) (Grant et al. 1996a, b).

The imposition of the 6-cm treatment resulted in relatively few changes in species composition. Even this intensity of grazing was sufficient to reduce the cover of M. caerulea[cf. the 33% utilization rate shown by Grant et al. (1996a) to cause a reduction in cover], but it was not sufficient to allow the increase of the less preferred species such as N. stricta. The total removal of sheep grazing allowed for the expansion of the grazing-intolerant M. caerulea and the dwarf shrub species, and a reduction in species associated with a short turf and heavy grazing, such as A. capillaris, A. odoratum, C. pilulifera and N. stricta.

The behaviour of the communities at these two sites was, at least in part, as expected. Vegetation left ungrazed became dominated by taller growing, often grazing-intolerant, species, although the species that became dominant varied considerably in how they ranked with Agrostis–Festuca mixes in terms of preference by sheep; D. flexuosa was similar, M. caerulea was less preferred (Hunter 1962). This increase in dominance by D. flexuosa and M. caerulea was the same as that described by Hill, Evans & Bell (1992) for exclosures in Agrostis–Festuca grasslands in North Wales, as well as in other situations (Watt 1976; Rawes 1981). The taller growing species exclude the shorter ones such as F. ovina, F. rubra, L. multiflora and N. stricta. Vegetation kept short by grazing was increasingly dominated by shorter growing species and species resistant to grazing or not preferred by sheep (Hunter 1962). However, the trends were complicated to some extent by site-wide changes in species composition; the overall decrease in A. capillaris seen at Cleish, and the increase in a number of mosses in all treatments at Kirkton. The general trends in species behaviour at the two sites are summarized in Table 5.

Table 5.  Overall trends in the cover of species that showed significant differences during the experiment at Cleish (C) and Kirkton (K) summarized by grazing treatment. ‘+’, an increase over the experiment, ‘.’, little change; ‘–’, a decrease. No entry means the species was not present
 3 cm4·5 cm6 cmExclosure
 CKCKCKCK
Cleish only
Conopodium majus+ + + + 
Deschampsia flexuosa . + + 
Festuca rubra+ + .  
Poa spp.  . + 
Trifolium repens+ . .  
Both sites
Agrostis capillaris.
Anthoxanthum odoratum... 
Festuca ovina..+.+.+.
Galium saxatile........
Hypnum jutlandicum++++++.+
Luzula multiflora+.+.+..
Potentilla erecta..++
Rhytidiadelphus squarrosus.+.+.+.+
Kirkton only
Carex pilulifera . + + 
Dwarf shrubs  . . +
Molinia caerulea   . +
Nardus stricta + + . 

However, the differences observed between the two sites are illuminating. The initial vegetation at Cleish had no M. caerulea and very little N. stricta; at Kirkton the former had an initial cover of 19%, the latter 29%. Thus the potential for the conversion of the vegetation to one that was relatively unproductive for grazing differed between the two sites. Both N. stricta and M. caerulea are low in nutrients and show lower digestibility values than mixtures of A. capillaris and F. ovina (Armstrong, Common & Smith, 1986). The treatments involving the maintenance of a short sward at Kirkton produced a community with an increasing cover of N. stricta. This was not possible at Cleish, and the resulting community was one with a high proportion of cover made up by mosses. Similarly, in the absence of grazing, there was an increase in the cover of M. caerulea at Kirkton. In its absence at Cleish, the exclosures became dominated by D. flexuosa and Poa spp.

It should be noted that at Kirkton N. stricta was present outside the experimental area, but it appeared not to be increasing despite the relatively short sward produced by sheep grazing. It is possible that N. stricta increased inside the plots because it was not subjected to winter grazing as the open hill was. This is an issue that needs investigation.

It would appear from comparison of movement in ordination space and analysis of the response of individual species, that to maintain the initial vegetation composition the most appropriate management at Cleish was to graze the sward at 4·5 cm (Tables 1 and 5). At Kirkton it appeared that grazing to 6 cm was most appropriate (Tables 3 and 5). There was sufficient grazing pressure in the summer to prevent the spread of M. caerulea, but it was not sufficient to encourage the spread of N. stricta at the expense of more preferred grasses, and changes in the cover of A. capillaris were minimal. The results from these two sites demonstrate that sustainable management of upland grazing, specifically British upland Agrostis–Festuca grassland, needs to take into account the presence of species capable of achieving dominance under different conditions, as well as the overall productivity of the vegetation and its constituents, a similar conclusion to that derived by Hill, Evans & Bell (1992).

In this grassland type, as in all others, the key to maintaining both agronomic and ecological sustainability is to manage the sward to maintain a substantial cover of palatable species, and to prevent the dominance of species that are unproductive or that alter the structure of the community. Changes in composition or structure, such as allowing the expansion of non-preferred species or allowing litter build-up, can make pastures more susceptible to change (Kemp, Dowling & Michalk 1996). Understanding the ecology of the species present is critical to predicting the effects of management on systems. This is true whether those systems are mediterranean-type rangelands threatened with the replacement of productive perennial grasses by annual grasses (Kemp, Dowling & Michalk 1996), or tussock grasslands with a balance to be maintained between productive intertussock grasses and tussock grasses (Allan, O’Connor & White 1992), or arid systems where dwarf shrubs expand at the expense of perennial grasses at high grazing intensities (Bisigato & Bertiller 1997). As in all systems, stocking rate (as set by sward height in this case) is the most important variable in grazing management. If this is not near the optimal level, then, regardless of other management practices, the community will undergo change (Walker 1997).

The overall small change in community composition over the 6 and 7 years of treatment at Kirkton and Cleish, respectively, has other implications. The relatively small number of species lost from these communities implies that moderating the severity of treatments should result in the reversal of trends in species composition. Vegetation change is considerably slower when species have to regenerate from seed or even re-invade the site (Mountford, Lakhani & Holland 1996). Thus these communities appear to have a certain resilience that allows the effects of a few years of inappropriate management to be rectified. It is not known how long this period lasts and how it is affected by the changes in sward potential for livestock production as species change in dominance. However, the increase in dominance of D. flexuosa (a species preferred by sheep, particularly in the spring), or the moss-dominated vegetation at Cleish in the more extreme treatments, may be more easily replaced by the desired A. capillaris and F. ovina than the less preferred M. caerulea and N. stricta, which are present at Kirkton, particularly as these latter two species are both less preferred species and can build up large quantities of unpalatable litter that hinder foraging.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We would like to thank John Wyllie and his staff from the Scottish Agricultural College for carrying out sward measurements and adjusting stock numbers at Kirkton. Thanks are due to David Suckling, Alison Smith and Titus Barthram for carrying out sward measurements and stock management at Cleish, and Andrew, William and Livvit Kemp for checking stock at weekends and out of season. This work was funded by the Scottish Office Agriculture, Environment and Fisheries Department.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Allan, B.E., O'Connor, K.F. & White, J.G.H. (1992) Grazing management of oversown tussock country. II. Effects on botanical composition. New Zealand Journal of Agricultural Research, 35, 719.
  • Armstrong, R.H., Common, T.G. & Smith, H.K. (1986) The voluntary intake and in vivo digestibility of herbage harvested from indigenous hill plant communities. Grass and Forage Science, 41, 5360.
  • Ball, M.E. (1974) Floristic changes on grasslands and heaths on the Isle of Rhum after a reduction or exclusion of grazing. Journal of Environmental Management, 2, 299318.
  • Barthram, G.T. (1986) Experimental techniques: the HFRO sward stick. Hill Farming Research Organisation Biennial Report 1984–85, pp. 2930. HFRO, Penicuik, UK.
  • Bisigato, A.J. & Bertiller, M.B. (1997) Grazing effects on patchy dryland vegetation in northern Patagonia. Journal of Arid Environments, 36, 639653.
  • Bullock, J.M., Clear Hill, B., Dale, M.P. & Silvertown, J. (1994) An experimental study of the effects of sheep grazing on vegetation change in a species-poor grassland and the role of seedling recruitment into gaps. Journal of Applied Ecology, 31, 493507.
  • GENSTAT V Committee (1987) GENSTAT V Reference Manual. Clarendon Press, Oxford, UK.
  • Grant, S.A., Suckling, D.E., Smith, H.K., Torvell, L., Forbes, T.D.A. & Hodgson, J. (1985) Comparative studies of diet selection by sheep and cattle: the hill grasslands. Journal of Ecology, 73, 9871004.
  • Grant, S.A., Torvell, L., Common, T.G., Sim, E.M. & Small, J.L. (1996a) Controlled grazing studies on Molinia grassland: effects of different seasonal patterns and levels of defoliation on Molinia growth and responses of swards to controlled grazing by cattle. Journal of Applied Ecology, 33, 12671280.
  • Grant, S.A., Torvell, L., Sim, E.M., Small, J.L. & Armstrong, R.H. (1996b) Controlled grazing studies on Nardus grassland: effects of between-tussock sward height and species of grazer on Nardus utilization and floristic composition in two fields in Scotland. Journal of Applied Ecology, 33, 10531064.
  • Hill, M.O. (1979) Decorana: A Fortran Program for Detrended Correspondence Analysis and Reciprocal Averaging. Cornell University, Ithaca, NY.
  • Hill, M.O. (1993) TABLEFIT version 0.0 for Identification of Vegetation Types. Institute of Terrestrial Ecology, Huntingdon, UK.
  • Hill, M.O., Evans, D.F. & Bell, S.A. (1992) Long-term effects of excluding sheep from hill pastures in North Wales. Journal of Ecology, 80, 114.
  • Hunter, R.F. (1962) Hill sheep and their pasture, a study of sheep-grazing in south-east Scotland. Journal of Ecology, 50, 651680.
  • Kemp, D.R., Dowling, P.M. & Michalk, D.L. (1996) Managing the composition of native and naturalised pastures with grazing. New Zealand Journal of Agricultural Research, 39, 569578.
  • MAFF (1986) Environmentally Sensitive Areas. First Report as Required in Section 18 of the Agriculture Act 1986. Ministry of Agriculture, Fisheries and Food, London, UK.
  • Miles, J. (1988) Vegetation and soil change in the uplands. Ecological Change in the Uplands (eds M.B.Usher & D.B.A.Thompson), pp. 5770. Blackwell Scientific Publications, Oxford, UK.
  • Mountford, J.O., Lakhani, K.H. & Holland, R.J. (1996) Reversion of grassland vegetation following the cessation of fertilizer application. Journal of Vegetation Science, 7, 219228.
  • Oksanen, J. & Minchin, P.R. (1997) Instability of ordination results under changes in input data order: explanations and remedies. Journal of Vegetation Science, 8, 447454.
  • Ratcliffe, D.A. (1959) The vegetation of the Carneddau, North Wales. I. Grasslands, heaths and bogs. Journal of Ecology, 47, 371413.
  • Ratcliffe, D.A. (1977) A Nature Conservation Review. Cambridge University Press, Cambridge, UK.
  • Rawes, M. (1981) Further results of excluding sheep from high-level grasslands in the North Pennines. Journal of Ecology, 69, 651669.
  • Rodwell, J. (1991) British Plant Communities, Vol. 2. Mires and Heaths.Cambridge University Press, Cambridge, UK.
  • Rodwell, J. (1992) British Plant Communities, Vol. 3. Grassland and Montane Communities.Cambridge University Press, Cambridge, UK.
  • Smith, A.J.E. (1978) The Moss Flora of Britain and Ireland. Cambridge University Press, Cambridge, UK.
  • Stace, C. (1991) New Flora of the British Isles. Cambridge University Press, Cambridge, UK.
  • UK Biodiversity Steering Group (1995) Biodiversity: the UK Steering Group Report, Vol. 2. Action Plans.HMSO, London, UK.
  • Walker, J.W. (1997) Viewpoint: grazing management and research now and in the next millennium. Journal of Range Management, 48, 350357.
  • Watt, A.S. (1976) The ecological status of bracken. Botanical Journal of the Linnean Society, 73, 217239.

Received 3 October 1998; revision received 21 June 1999