Early season grazing effects on birch , grass , herbs and plant litter in coastal meadows used by reindeer : a short-term case study

The effects of short-term grazing by reindeer (Rangifer tarandus tarandus) on birch (Betula pubescens), grasses, herbs and plant litter in coastal meadows in spring were investigated in grazed and control plots in 1996 and 1997. The meadow contained 29 different plant species, all but one of which (Deschampsia caespitosa) were intensively grazed by reindeer. Young birch eaten by reindeer did not increase in mean height (9 cm), while birch protected from grazing grew from 9 to 22 cm (P<0.05) during the two years of the experiment. The ratio of grasses to herbs was higher (P<0.05) in the grazed plots than in the control plots, and the relative abundace of grasses increased during the summer in both years. The abundance of Rumex asetosa and Alchemilla subcrenata decreased (P<0.05) in response to grazing. From spring 1996, plant litter increased (P<0.05) on the control plots until the investigation came to an end in spring 1998, unlike grazed vegetation. Reindeer affects the coastal meadows in northern Norway in spring by browsing on birch and grazing on herbs and grass which in the long term might influence the cultural landscape in favour of the growth of grass species.


Introduction
more heterogeneous composition of grasses, herbs and small birch (Betula pubescens) (Losvik, 1993; In northernmost Norway, semi-domesticated rein- Alm, 1996).Such areas may subsequently be deer (Rangifer tarandus tarandus) migrate in late colonised by trees, and thus transformed into shrub winter from inland pastures to the coast, peninsu-lands (Gibson et al., 1987a;Harrison, 1976; las and islands (Paine, 1994) where they graze on Mitchell et al., 1997).During the spring and early grasses, herbs, trees and woody bushes during the summer, coastal meadows in early successional summer (Gaare & Skogland, 1975; Skogland, stages dominated by grass and herbs are charac-1980; Mathiesen et al., 1999;2000).When they terised by a high protein content (crude protein as reach the islands, they start grazing at sea level and high as 30% of dry matter (DM)) and easily follow the young vegetation close to the snow digestible water-soluble carbohydrates (26% DM) melt-line up to the mountain pastures.In North- (Eilertsen et al., 2000).Reindeer may consume as Norway (the counties of Nordland, Troms and much as 131 g DM/kg body mass 0 75 per day of Finnmark) 25% of former farmland has been aban-their dietary plants on such coastal meadows doned between 1959 and 1997 (Central Bureau of (Eilertsen et al., 1999) in order to increase their Statistics, 1960;1997), much of this land being in body protein (Eilertsen et al., 2001).Domestic the coastal region where meadows without herbi-sheep grazing on abandoned meadows in the north vores have reverted from sward dominated by tim-of England help to maintain a species-rich vegetaothy grass (Phleum pratense) to a vegetation with a tion (Smith & Rushton, 1994).Sheep grazing has increased the proportion of annual herbs (Gibson et al., 1987b) and thus influenced the future botanical composition of the cultural landscape.A number of studies suggest that a dense accumulated litter layer may limit the diversity of productive habitats by inhibiting species establishment through the effects of shading and mechanical impediment (e.g.Goldberg & Werner, 1983;Carson & Peterson, 1990;Tilman, 1993;Facelli, 1994).We investigated whether early spring grazing by reindeer affects the dynamics of birch, grass and herbs in coastal meadows.

Area and design
At Søreidet, on the island of Reinøya (147.3 km 2 ) in Troms County (69°53'N, 19°46'E), 0.8 ha of abandoned meadow near the shoreline was sur¬ rounded by an electric fence.The meadow had pre¬ viously been used for hay and silage production, but has not been used for traditional farming since 1990, having only sporadically been exposed to grazing by sheep and goats during the past few years.The sward was dominated by the grasses Festuca rubra, Poa alpigenea, Agrostis tenuis, Phleum commutatum and herbs such as Achillea millefolium, Rumex acetosa, Ranunculus repens and Alchemilla subcrenata.Inside the electric fence, five permanent enclosures (control plots) (1 m x 1 m) were estab¬ lished by dividing the area into a grid system (1 m x 1 m), and selecting plots at random.The enclo¬ sures were created by erecting steel fencing with a 10 cm mesh.The snow melted at the same time inside the enclosures as on the rest of the experi¬ mental vegetation.Adjacent to each enclosure an unfenced plot of the same size was established.In spring 1997 a further five plots were established in randomly selected squares and exposed to grazing.

Animals
Male yearling reindeer (n = 12, 15 reindeer ha -1 ) were allowed to graze in the fenced area from 7 June until 1 July in 1996, from 14 June until 7 July in 1997.Animals were released at equal phenological plant stages, immediately after the snow had disappeared from the fence and as soon as plant growth had started (Eilertsen et al., 1999;2000).
Occurrence of grass, herbs and plant litter Vegetation cover was observed in grazed plots and in control plots one week after grazing started in 1996 and at the start of grazing in 1997.The plots were recorded weekly during the grazing period in both years.All plant species present on the sample plots were registered, and the proportion of each species was assessed visually as percentage cover according to the Domin cover-abundance scale (e.g.Evans & Dahl, 1955;Økland, 1990).Furthermore, in spring (8 June) 1998, vegetation cover was recorded in the grazed and control plots.Previously grazed plots were registered again in summer  1998 (26 July).The same person made all the esti¬ mates of vegetation cover and all plots were pho¬ tographed before the records were made, thus per¬ mitting the estimates of cover to be checked against the photos.In all plots, the percentage of plant litter covering the ground was assessed at the same time.In order to check the cover estimates, in spring 1996 samples of vegetation (n=10) were harvested after the observations of cover had been made.The vegetation was cut by one-hand garden clippers and sorted, each plant species was weighed separately.The proportions of the total weight of grass and of each species of herb were calculated, and the estimates of cover were checked.
into the generic groups "herbs" and "grasses".The data sampling carried out in June is referred to as "spring", while the July sampling is referred to as "summer" (Tables 2-4).Significant differences (P<0.05) between birch on grazed and ungrazed meadow were calculated by Student's t-test (twotailed test assuming equal variances) (Bhattacharyya & Johnson, 1977).The assumption of equal vari¬ ance was tested by the Folded Form F Statistic (SAS Institute Inc., 1989).Analyses of differences in the number of plant species, plant cover, plant litter, and grass:herbs ratio were performed by ANOVA in the General Linear Models procedure described by SAS Institute Inc. (1989).

Birch
The height of randomly selected young birches (Betula pubescens) (<0.5 m high) were measured on 10 birches in the grazed area, and on 10 adjacent birches protected from reindeer grazing by the electric fence.The birches were individually marked and each height was measured by ruler to the nearest cm from ground level to top of the meristem at the beginning of the grazing period in 1996 (7 June).The measurements were repeated on 8 July, 1997 after the two years of spring grazing.
The leaves of the top meristem on the birches were inspected for any defoliation caused by the grazing reindeer.

Data analysis
The various herb and grass species were grouped

Plant species
Both grazed and control plots in the coastal mead¬ ow contained a total of 29 different plant species (Table 1), dominated by the grasses Festuca rubra, Poa pratensis, Agrostis capillaris, Deschampsia caespitosa, Phleum alpinum and the herbs Achillea millefolium, Rumex acetosa, Ranunculus repens and Alchemtlla subcrenata.No significant changes in the total number of plant species were detected during the two years of the experiment.All plant species were visibly and intensively grazed, except D. caespitosa.

Botanical composition and cover of individual plant species
The differences in plant cover between grazed and control plots after one week of spring grazing in  1996 is illustrated in Fig. 1.Further, the cover of grass and of certain selected herbs is shown in Table 2.The cover of grass increased (P<0.05) from spring to summer in both grazed and control plots in 1996 and 1997.In comparison with the control plots, the grass cover was higher (P<0.05) in grazed plots in spring 1998 and in summer 1997.
In contrast, cover of the herb R. acetosa decreased from spring to summer in grazed plots (Table 2).In summer 1997, the cover of R. acetosa was higher (P<0.05) in control plots than grazed plots.There were also significant differences between grazed and control plots for A. subcrenata in summer 1997 (Table 2).The dynamic changes in plant cover dur¬ ing the grazing period in 1997 is shown in Table 3.
Grass cover increased (P<0.05)rapidly from spring to summer in 1997.In contrast, there were no sig¬ nificant changes in the cover of the various herbs.
As a consequence of the increased cover of grass, the ratio of grass to herbs rose (P<0.05) during the grazing period in 1997 (Table 3).
In the summer observations in 1996 and 1997 and spring 1998, the grasses:herbs ratio in the grazed plots was higher (P<0.05)than the control plots (Table 4).Within the grazed plots, the ratio rose from 1996 to 1997 and 1998 in spring, and from 1996 to 1997 in summer (Table 4).Similar changes were observed in the control plots (Table 4).

Shrub development
The mean height of the birches in the grazed area remained unchanged between 1996 and 1997 (Table 5), while it more than doubled in the control area (P<0.05)(Table 5).All leaves were intensively browsed by reindeer, whereas on the control plots a large number of large green leaves were observed.

Litter accumulation
Plant litter cover increased (P<0.05) in the control plots during the three years of observation (1996, 1997 and 1998) (Table 6).There was significantly less plant litter in the grazed plots than in the con¬ trol plots in spring 1997 and 1998 (Table 6).

Dynamic change in the vegetation in response to grazing
In spring and summer, herbs make up a consider¬ able part of the diet for sheep and reindeer when they are available in the pasture (Gaare & Skogland, 1975;White et al., 1975;Skogland, 1980;Garmo & Skurdal, 1989;Mathiesen et al., 1999;2000).Herbs previously exposed to herbivores or mechan¬ ical cutting seem to develop smaller leaves than ungrazed plants.Further, intensive sheep grazing Rangifer, 22 (2), 2002 reduced the occurrence of herbs (Bowns & Bagley, 1986;Haggstrom, 1990), and rare species could be extinguished (Augustine et al. , 1998).By the end of the grazing experiment, a few grass species were dominant, indicating that herbs seem to be inten¬ sively grazed by reindeer on the coastal meadows in northern Norway in spring.However, none of the herbs growing in the meadow were exterminated during the experimental two years.The dominant grass species, such as P. pratensis, A. capillaris, D. caespitosa, are very common in old meadows in northern Norway (e.g.Nesheim, 1986;Sveistrup & Østgård, 1985).Traditional grazing by cattle, sheep and goats is common on these mead¬ ows and grasses tend to be more tolerant to grazing than herbs (Hester, 1996).Many grass species increased the number of tillers during recovery.Furthermore, in arctic grasses, species such as Dupontia fisheri can compensate for clipping by rapid regeneration (Wegener & Odaz, 1997).In grassland containing various proportions of herbs and grass species in Norway, Haugland (1999) reported that the biomass production of grasses was greater than that of herbs in the second harvest.Nesheim (1986b) and Lemieux et al. (1987) have also reported that second yield harvest decreased as the amount of herbs increased.
In late summer the grazed vegetation had a high¬ er protein content than the ungrazed vegetation (Eilertsen et al., 2000).In recovered leaves of previ¬ ously grazed vegetation the protein content and digestibility were both high in comparison with ungrazed plants at the flowering stage.Such graz¬ ing facilitation is discussed by e.g.McNorton (1976;1984).Likewise, experimental clipping of sedges increased concentrations of nitrogen and minerals, improving quality of forage eaten by Canadian caribou (Ouellett, et al. , 1994).Furthermore, early summer grazing by sheep was capable of improving the nutritive value of forage plants compared to locations with ungrazed plants (Alpe et al., 1999).Nesheim (1986a) and Haugland (1995a, b;1999) demonstrated that the ratio of herbs to grasses in plant communities in Norway did not affect the dietary quality (protein content and digestibility) of the pasture as long as the plant species involved were at the same developmental stage.
The nutritive quality of forage vegetation used by reindeer in the meadows was not affected by the grass:herbs ratio, while a predominantly grassy vegetation seems to increase biomass production late in the growing season, compared to vegetation dominated by herbs.

Colonization of birch
During the last decades birch forests in northern Norway seems to have changed in response to a fall in the number of grazing ruminants (43 000 fewer dairy cattle and 200 000 fewer sheep) from 1959 to 1997, a ride of 31 000 ha in the area of abandoned farmland (Central Bureau of Statistics, 1960, 1997).Climatic conditions have also favoured the growth of pioneer birch plants (e.g.Hanssen-Bauer & Frøland, 1998;Talkkari, 1998;Kullmann, 2001;Niemela et al., 2001).An invasion of shrubs and woody meadow vegetation would cause consider¬ able changes in the landscape vegetation (Bjor & Graffer, 1963;Harrison, 1976;Mitchell et al., 1997).In meadows at several places on Reinøya, where this experiment were carried out, pioneer birch plants seem to colonise what had previously been farmland.In our experiment, young birch leaves seemed to be highly selected by reindeer in early spring.Intensive browsing reduces the regen¬ eration of palatable trees (Bjor & Graffer, 1963;Risenhoover & Mass, 1987;McInnes et al., 1992) or might even eliminate all tree regeneration (Jordan, 1967;Marquis, 1974).Reindeer browsing on pioneer birches in the spring could prevent the coastal meadow from being colonised by bushes and subsequently being developing into shrub-land and woodland.

Accumulation of plant litter on abandoned meadows.
In spring 1997 and 1998, before grazing, accumu¬ lations of plant litter were lower in previously grazed vegetation than on control plots.First, rein¬ deer remove a large proportion of the biomass (as much as 16 kg per animal per day) during the summer leaving less for senescence and litter pro¬ duction (Eilertsen et al. , 1999).Secondly, a decrease in plant litter may also be due to trampling by reindeer.The hoof and foot areas of reindeer are large (Nieminen, 1990), and these animals crush a lot of litter at each step.Moderate trampling may increase the rate of decomposition of litter caused by soil micro-organisms (e.g.Stalfelt, 1960;Larcher, 1983), with the result that the availablility plant nutrients might increase, in turn stimulat¬ ing plant growth.Studies in South Georgia have demonstrated that intense reindeer trampling can transform the nature of the vegetation (e.g.Lindsay, 1973;Leader-Williams et al., 1987).Likewise, in summer, dry lichen mats are rapidly destroyed by trampling (Bayfield et al., 1981) in particular on migration routes (Boertje, 1990;Pegau, 1970).We did not detect any trampling damage on the vege¬ tation in the meadow grazed by the reindeer.The hoof load of sheep (approximately 850 g/cm 2 ; Spedding, 1971) is more than three times as high as the hoof load of reindeer (approximately 250 g/cm 2 ; Nieminen, 1990).The trampling of the reindeer should therefore have little effect on the coastal vegetation that we investigated.Reduced utilisation of coastal meadows in northern Norway by ruminants may therefore increase the accumula¬ tion of plant litter in these locations.
On the basis of two years of investigation, plant species growing on coastal meadows are assumed to be tolerant of grazing in the northern Norwegian farming system, but the period of study may have been too short to have a detrimental impact on the species involved.The reindeer seems to have increased the dominance of grass species in the abandoned meadows which may lead to higher bio¬ mass production late in the growing season.Furthermore, browsing on birch seems to prevent the area from being colonised by pioneer trees and may thus support the maintenance of an open land¬ scape in northern Norway.

Fig. 1 .
Fig. 1.Photo to the left, plots ungrazed for one week; photo to the right, plots grazed by reindeer for one week at Reinøya in mid June 1996.

Table 1 .
Plant species observed in the investigated plots at Reinøya.

Table 2
. Percentage plant cover of grass, Rumex acetosa, Alchemtlla subcrenata and Ranunculus repens in spring (June) and summer (late July) in meadow ungrazed and grazed by reindeer for 3.5 weeks in spring at Reinøya in 1996 and 1997.Significant differences (P<0.05) between grazed and ungrazed meadows each year and season is marked with '*', and different letters in superscript indicates differences (P<0.05) between years within grazed or ungrazed meadows.Standard deviation in parentheses.

Table 3 .
Percentage plant cover of grass, Rumex acetosa, Alchemtlla subcrenata, Ranunculus repens, Achillea millefolium Taraxacum sp. and Equisetum arvense in plots (n=10) grazed by reindeer for 24 days at Reinøya in 1997.Grass: herbs relationship is based on the cover estimates.Different letters in superscript indicates differences (P<0.05) in cover between dates.Standard deviation in parentheses.
CTable4.Gras: Herbs relationship in spring (June) and summer (late July) based on percentage plant cover in mead¬ ow ungrazed and grazed by reindeer for 3.5 weeks in spring at Reinøya in 1996 and 1997.Significant dif¬ ferences (P<0.05) between grazed and ungrazed meadows each year and season is marked with '*', and dif¬ ferent letters in superscript indicates differences (P<0.05) between years in grazed meadows.Standard devi¬ ation in parentheses.