1. Abandonment and eutrophication are major threats to traditional pastoral landscapes and their wildlife in Europe. Social and economical developments have rendered traditional pastoralism impracticable. More knowledge is needed about the effects of grazing with free-ranging herbivores, which is increasingly used as a substitute for the traditional herding system.
2. We studied the effects of free-ranging cattle on the recovery of Calluna heather, tree encroachment and plant species richness in six habitats in a grass-rich Dutch heathland during a 10-year period. The habitats differed in soil fertility, initial heather–grass ratio and developmental stage of Calluna.
3. Despite preferential grass defoliation, cattle grazing neither reduced grass cover in grass heath nor prevented grass invasion in heather. Grass invasion failed only in a nitrogen-poor turf-stripped Calluna heath. Grazing induced a substantial Calluna recovery in grass heaths on podzolic soils, but its recovery failed in grass heath on a phosphorus-rich medieval arable field. As a consequence, the grass–heather mosaics generated by free-ranging cattle were restricted to habitats of intermediate soil fertility.
4. Grazing did not prevent encroachment by pine and birch. Removal by the site manager prevented conversion of 10–20% of the open heathland to forest.
5. During the first 5 years, grazing induced a significant increase in species richness in all habitats. During the second 5 years, species richness stabilized in grass heath and heather–grass mosaics and it declined in the pioneer Calluna heaths.
6. We found indications of various nutrient-mediated grazing effects on the competitive balance between grass and woody pioneers. These suggest that nutrient-mediated feedback might be an important explanatory mechanism for the described vegetation mosaic cycling in heathlands.
7. Free-ranging grazing did not remove the high atmospheric nutrient inputs of the whole area. Substantial amounts of nutrients were redistributed from the grass lawns to the forest.
8. Free-ranging grazing combined with tree cutting appeared to be a suitable management regime for the maintenance of species-rich open heathlands with dynamic grass–heather mosaics.
9. Without tree cutting, free-ranging grazing would have created dynamic tree–grass–heather mosaics in open heathland. Wood–pasture landscapes are fundamentally different from the open heather-dominated heathlands produced by the traditional sedentary farming system. Integrated grazing of heathland, nutrient-rich farmland and forests accelerates this change. Alternative grazing regimes are discussed.
Abandonment in conjunction with atmospheric eutrophication is a major threat to the high conservation values of European traditional pastoral landscapes (Gimingham 1972; Bignal, McCracken & Curtis 1994; Bignal & McCracken 1996; Bignal 1998; Ostermann 1998; Webb 1998). Increased labour costs have rendered traditional herding systems economically unfeasible. Herding is increasingly replaced by more profitable free-ranging systems, using livestock or wild herbivores grazing over extensive areas. Little is known about the long-term effects of this change in grazing patterns on the landscape structure and its conservation values. The design of effective grazing systems for conservation management requires predictive knowledge of the effects on biodiversity and the underlying ecological key processes and factors, e.g. vegetation succession and nutrient availability (McCracken & Bignal 1998). Long-term monitoring provides this information (Illius & Hodgson 1996; Hester & Baillie 1998; Piek 1998; Hester et al. 1999) and offers an opportunity for the validation of simulation models (Wallis de Vries & Van de Koppel 1998).This paper assesses the effects of free-ranging cattle on the vegetation dynamics and species richness of open heathland vegetation in the Netherlands, with a special focus on nutrient-mediated mechanisms.
Inland heaths in continental north-west Europe originated mainly by sedentary pastoralism and associated land use (De Smidt 1979; Pott & Hüppe 1991; Pott 1998; Webb 1998). The daily return of the herd to the stable maximized the harvest of manure, which was essential for sustained sedentary arable farming on the nutrient-poor Pleistocene sandy sediments. After the 11th century, the manure was enriched with organic matter from ‘plaggen’ (sods) from the rangeland. This practice created ‘plaggen soils’, with a typical thick, black–brown, humic, nutrient-rich topsoil on the arable fields (Pape 1970).
Traditional sedentary herding declined in north-west Europe at the beginning of the 20th century after the introduction of synthetic fertilizers and barbed wire fencing. The abandoned rangelands were transformed into fertilized farmland and forest. In the Netherlands, approximately 40 000 ha of lowland heath (5% of the original 800 000 ha) remained unchanged. The majority of this area is managed today as a nature reserve. An open landscape, low soil fertility and substantial heather cover are considered essential for the survival of many endangered species (Bal et al. 1995). Until 1970, management was mainly restricted to the removal of encroaching trees. After 1960, the dominant heather species Calluna vulgaris L. and Erica tetralix L. were increasingly replaced by the perennial grass species Deschampsia flexuosa (L.) Trin. and Molinia caerulea (L.) Moench (Diemont 1996). Similar changes were reported from surrounding countries (Bülow-Olsen 1980; Marrs, Hicks & Fuller 1986; De Blust & Schneiders 1989; Pott & Hüppe 1991).
The grass invasion and loss of biodiversity were attributed to increased nitrogen deposition and to abandonment by traditional livestock farming (including woodcutting, burning and turf-stripping; Diemont & Heil 1984; Anonymous 1988). Both changes were assumed to favour succession, eutrophication and acidification (Aerts 1993a,b; Bobbink, Hornung & Roelofs 1998; Bakker & Berendse 1999). Grass invasion was accelerated by a positive feedback between nitrogen availability and grass dominance (Berendse 1990, 1994). Turf stripping (removing grass, litter and nitrogen and exposing Calluna seeds) appeared to be an effective method for the restoration of heather dominance (Diemont & Lindhorst Homan 1989; Heil & Aerts 1993). However, turf stripping has serious financial and biodiversity drawbacks (Anonymous 1988). This generated new interest in free-ranging grazing as a cheap alternative for sedentary herding and turf stripping.
These different results of grazing in grass heaths might be attributed to differences in site (climate, management history, soil fertility, initial grass–heather ratio, heather and grass species, heather seed bank, developmental stage of heather), grazing regime (including wild herbivores) and spatio-temporal scale of observation. We hypothesized that free-ranging cattle might induce heather recovery, inhibit tree encroachment and increase vascular plant species richness in grass heath on nutrient-poor soils.
It was hypothesized that replacement of the traditional land use by free-ranging cattle would lead to nutrient redistribution rather than nutrient removal. The resultant soil fertility patterns and the free habitat choice might lead to greater feedback between vegetation development and use by the herbivores. This might, in turn, lead to the emergence of shifting mosaics.
Materials and methods
Study area and habitats
The Wolfhezerheide nature reserve is situated on the Veluwe in the centre of the Netherlands, about 12 km east of Wageningen (51°47′N; 5°41′E). Natuurmonumenten has owned this nature reserve since 1966. The grazed area (40 ha of open heathland and 20 ha of forest) covers podzolic soils in fluvio-glacial sand and cover sand, a small area of plaggen soils in cover sand, and some moist podzolic and peat soils on a valley bottom (Vrielink & Van den Hurk 1975). The plaggen soils belonged to the medieval settlement ‘Wolfheze’. Spanish soldiers destroyed it in 1585 and its arable fields were mainly used as pasture afterwards (Natuurmonumenten 1996). The annual atmospheric nutrient deposition (wet plus dry) in the open heathland accounted for about 40 kg nitrogen, 1·5 kg phosphorus and 5 kg potassium ha−1 (Erisman & Heij 1991). Until 1983, juvenile trees were removed from the open heathland and old heather was occasionally mown (Natuurmonumenten 1996). Deschampsia invasion started in about 1960 on the plaggen soil and spread between 1970 and 1980 to the dry podzolic soils (B.W. Jacobs, B. Jansen & W. Aandeweg, personal communication). Some parts of the remaining mature heather were turf-stripped in 1980, other parts were mown in 1982 (Table 1). In 1983, Deschampsia covered major parts of the open heathland. Minor parts were covered by mature Calluna, pioneer Calluna and Molinia (Table 1).
Table 1. Main characteristics of six open heathland habitats at the Wolfhezerheide in 1983
In January 1983, the area was stocked with 17 Dutch Friesian heifers, which from May 1983 were gradually replaced by suckling cows of various breeds. Annual stocking rates were maintained at an equivalent of approximately 0·2 animal units ha−1 (1 animal unit = 450 kg). Rates were approximately 30% higher in the summer than in the winter. During the whole study period one to three horses were allowed in the area. Late winter densities of roe deer Capreolus capreolus L. (2–4 km−2) and rabbits Oryctolagus cuniculus L. (100 km−2) were low. The latter value increased slightly during the study period (B.W. Jacobs, personal communication). From 1984 a minor quantity (1–2 kg dry matter animal−1 day−1) of supplementary food (concentrates, hay or corn silage) was provided during late winter.
After 1983, the manager occasionally removed juvenile trees (1–2 m), mainly Scots pine Pinus sylvestris L. and silver birch Betula pendula Roth. No fires occurred. The winters of 1984–85 and 1985–86 were extremely cold and winter browning of Calluna (MacDonald 1990) was observed. Heather beetles Lochmaea suturalis damaged Calluna locally in 1991 (personal observation).
Habitats and treatments
In 1983, we located 21 5 × 5 m plots randomly in six habitat types in the grazed open heathland (Table 1). The habitats differed in soil, land-use history, previous management and initial vegetation. In April 1984, seven grazed plots and four ungrazed plots were added (Table 1). The ungrazed exclosures (approximately 8 × 8 m; mesh size 5 × 5 cm) were constructed in the most heavily grass-dominated Deschampsia heath on plaggen soil and in Molinia heath.
Plant cover and Calluna density
Plant cover was recorded annually in August–September, using an extended point quadrat method (Goldsmith & Harrison 1976). Pins (diameter 2·5 mm) were lowered through 10 vertical holes at 10-cm intervals in a horizontal plastic tube (diameter 5 cm, length 100 cm, 40 cm above ground level). The tube was located randomly in each plot 10 times, providing 100 pins. Hits for vascular plants (living parts only) were noted. Mosses and lichens were recorded as a group. If a species was hit more than once by the same pin, it was recorded only once. Several species per pin indicated overlap. The number of hits per species by 100 pins reflected the species cover (0–100%). Second and third samples of 100 pins were added until the average cover remained constant (< 5% change to the average by adding a new series). The cover of species not hit by the pins (i.e. species with a low cover or with foliage above 40 cm height) was estimated visually. Cover measurements in the ungrazed Molinia heath started in 1987.
Calluna individuals were counted in the grass heaths to assess the re-establishment rate.
The soil fertility of the dry habitats was sampled during the winter 1987–88. Samples from Ao (the ectorganic layer, including litter, fermented horizon and humic horizon) were taken at 50 random locations within 5 m around each plot. At each location, thickness was measured on the four sides of a 1-dm2 sod. The Ao material was transported in plastic bags to the laboratory for chemical analysis. The thickness of the Am horizon (grey–blackish mineral horizon between Ao and the brownish B horizon) was measured using an Edelman auger (diameter 7 cm) at eight locations around each plot. Volume samples from A10 (the upper 10 cm of Am) and the total Am were collected at eight locations around each plot with a gouge auger (diameter 30 mm). Samples were mixed per plot. Laboratory treatments followed Houba et al. (1986). Subsamples of Ao were dried at 70 °C for 7 days, grounded and sieved. Mineral soil samples were only sieved (mesh size 2 mm). For all horizons, organic matter content was measured as loss-on-ignition. The pH was assessed after adding 1 m KCl solution to the soil. Total N and P concentrations (spectro-photometrically) and the total K concentration (flame-photometrically) in the soil were measured after digestion in H2SO4–salicylic acid–H2O2 and selenium. Extractable NH4+-N, NO3– -N, P and K in the topsoil (Ao + A10) were measured after digestion in 0·01 m CaCl2, using a Technicon Auto-analyser (Skalar, Breda, the Netherlands).
Calluna seed bank
To explain different rates of Calluna recovery, we assessed the Calluna seed bank in Ao and A5 (the top 5 cm of Am) in grass heath on podzolic soil (three plots) and plaggen soil (two plots) in June 1991. Two-hundred soil samples were taken with a gouge auger (diameter 30 mm) within a 5-m distance around each plot. The Ao (without unfermented litter) and A5 material were placed on perforated dishes. The dish area per plot and horizon was 1414 cm2 and the resulting thickness of the soil material on the dishes was 1 cm for the Ao material and 5 cm for the A5 material. The dishes were placed in full daylight in a non-heated, ventilated, greenhouse. They were kept moist and cool (< 35 °C) by daily sprinkling for 130 days (28 June−4 November 1991). Emerging seedlings were removed with minimal soil disturbance. Their number was expressed per m2 dish area, approximating the density of viable seeds per unit exposed area under field conditions. Our method did not yield an estimate of the seed bank per unit volume, as the material was not stirred, dried and remoistened during the germination period (Hester, Gimingham & Miles 1991; Bruggink 1993).
The defoliation pressure on grass and heather in the Deschampsia heaths on podzolic and plaggen soil was approximated by measuring the plant height of Deschampsia (20 random locations per plot) and Calluna (20 individuals or all available individuals). A perforated foam disc (diameter 10 cm; weight 8 g) was slowly lowered down a vertical measuring rod on to the plant canopy.
Calluna died in some plots in 1991–92 due to a heather beetle outbreak. This necessitated separation of affected and unaffected plots (Table 1). One of the plots in the grazed Deschampsia heath on plaggen soil was omitted because vegetation and soil had been severely affected by trampling and excretion during supplementary feeding at the site (Table 1).
The measured data were tested for normality using a one-sample Kolmogorov–Smirnov test (Norusis 1993). Viable seed numbers were log-transformed. The seed bank of the Ao as a percentage of the total seed bank was arcsine-transformed. Differences between habitats or treatments were tested using two-sample t-tests or anova followed by Student–Newman–Keuls multiple comparisons. Homogeneity of variance was checked using Levene's test. Differences between time series of grazed habitats were tested using anova with year as covariate. Grazing–year interactions were detected by comparing two test designs. When two-way or higher order interactions were absent, main effects were tested with a one-way anova with year as covariate.
All habitats were very acid (Table 2). The extreme soil fertility values occurred in the pioneer Calluna heath on podzolic soil and in the Deschampsia heath on plaggen soil (Table 2). The Ao thickness and organic matter pool were significantly lower (P < 0·05) in the two pioneer Calluna heaths than in the other habitats. The extractable nitrogen was also lower, but only significantly so (P < 0·05) in the turf-stripped Calluna heath. The grazed Deschampsia heath on plaggen soil differed significantly (P < 0·05) from the one on podzolic soil in its thicker Am layer, higher Ao phosphorus pool and higher extractable phosphorus. The latter value was also significantly higher in the grazed than in the ungrazed treatment (P < 0·05).
Table 2. Soil characteristics of five grazed and one ungrazed dry heathland site at the Wolfhezerheide (for site descriptions see Table 1 and text).Values in a row with a same letter are not significantly different (one-way anova and Student–Newman–Keuls test, P < 0·05). Ao= ectorganic horizon (L + F + H); Am= humus-rich mineral horizon; A10= upper 10 cm of the Am. Extractable nutrient values concern the topsoil (Ao + A10)
Grazed turf-stripped pioneer Calluna heath (2)
Grazed mown pioneer Calluna heath (2)
Grazed mature Calluna heath (5)
Grazed Deschampsia heath (7)
Grazed Deschampsia heath (2)
Ungrazed Deschampsia heath (2)
Thickness Ao (cm)
Thickness Am (cm)
Organic matter Ao (ton ha−1)
Organic matter A10 (ton ha−1)
Organic matter Am (ton ha−1)
pH KCl Ao
pH KCl A10
Nitrogen pool Ao (ton ha−1)
Nitrogen pool A10 (ton ha−1)
Nitrogen pool Am (ton ha−1)
Phosphorus pool Ao (ton ha−1)
Phosphorus pool A10 (ton ha−1)
Phosphorus pool Am (ton ha−1)
Potassium pool Ao (ton ha−1)
Potassium pool A10 (ton ha−1)
Potassium pool Am (ton ha−1)
Extractable nitrogen (kg ha−1)
Extractable phosphorus (kg ha−1)
Extractable potassium (kg ha−1)
Heather and grass cover
The cover of Calluna and Deschampsia displayed different trends in the four grazed habitats on podzolic soil (Fig. 1a–d; habitat × year effect, P < 0·001; Table 3). In the turf-stripped pioneer Calluna heath (Fig. 1a), Calluna cover increased during the study period to nearly 90%. This contrasted with the persistently low Deschampsia cover. In the mown pioneer Calluna heath, cover of both species increased (Fig. 1b). In the mature Calluna heath (Fig. 1c), Calluna declined initially, particularly between 1985 and 1986, but it recovered afterwards. Similar declines occurred between 1985 and 1986 in the other Calluna habitats. Calluna also recovered after the crash due to heather beetles in 1991–92 (Fig. 1b–d). Deschampsia cover increased or stabilized at a high level in all Calluna habitats, except in the turf-stripped pioneer Calluna heath.
Table 3. F-values and significance levels of main effects of habitat, year and habitat × year on the cover of Deschampsia and Calluna and on the species richness in four grazed Calluna habitats on podzolic soils during the period 1983–93 (anova, ***P < 0·001)
Source of variation
Habitat × year
The effect of grazing on the recovery of Calluna differed significantly between the Deschampsia heath on plaggen soil and the Molinia heath (Fig. 1e–h; habitat × grazing interaction, P < 0·01; Table 4). Calluna recovery failed in the grazed Deschampsia heath on plaggen soil (Fig. 1e). In the Molinia heath, cover reached 20% in 1993 (Fig. 1g). The Deschampsia cover in the ungrazed Deschampsia heath on plaggen soil declined after the tree invasion (Fig. 1f).
Table 4. F-values and significance levels of main effects of habitat, grazing and year and interactions on the cover of Deschampsia, Molinia, Calluna and on species richness in grazed and ungrazed Deschampsia heaths on plaggen soil and Molinia heaths on podzolic–peat soil 1983–93
anova * P < 0·05, ** P < 0·01, *** P < 0·001; NS = not significant.
No other herbaceous species became dominant at the habitat scale. On individual plots, the non-nitrophilous species Carex pilulifera L., Galium saxatile L. and Rumex acetosella L. occasionally reached a 10% cover level.
Calluna density in grass heaths
The measured density in the grass heaths reflected the net establishment success, as plots did not contain juveniles in 1983. Between 1983 and 1990 density increased in the Deschampsia heath on podzolic soils significantly more than on plaggen soils (Fig. 2; anova, habitat × year, F = 4·34, P < 0·05). After 1990 the rates were no longer significantly different. Calluna density in the grazed Molinia heath was also higher than in the Deschampsia heath on plaggen soil (anova, habitat × grazing effect, F = 10·5, P < 0·01). No significant difference occurred between the grazed and ungrazed Deschampsia heath on plaggen soil. In the ungrazed Deschampsia heath one seedling emerged, in the ungrazed Molinia heath none. Calluna density (Fig. 2) and Calluna cover (Fig. 1d,e,g) were significantly correlated in 1993 (R2 = 0·91).
Shrubs and trees
Young trees (see the Appendix) established in the grazed treatments, but their cover remained low (< 1%), partly because of browsing and partly as a result of removal by the site manager (see the Discussion). The ungrazed exclosures displayed different results. In the ungrazed Deschampsia heath on plaggen soil, trees started to invade after the winter of 1985–86 (Fig. 1f). In 1993 the mean cover had reached 20%. Tree seedlings were occasionally observed in the ungrazed Molinia, but they did not survive. Statistical tests were not carried out for tree cover, as the values in the grazed treatments were affected by tree removal.
The frequency of shrub and tree species remained rather constant; only birch and bramble Rubus fruticosus L. increased substantially (see the Appendix).
Calluna seed bank
The number of emerged seedlings from the Ao of the grass heath on podzolic soil (1324 ± 1039 m−2) and plaggen soil (96 ± 76 m−2) differed remarkably, but the difference was not statistically significant (P = 0·078; Fig. 3) due to the low number of replicates. The differences between the A5 seed banks were less distinct: 175 ± 145 m−2 for podzolic soil and 89 ± 86 m−2 for plaggen soil (Fig. 3).
Linear regression yielded a significant relationship between the Ao seed bank and Calluna density in 1990 (R2 = 0·97, n = 5, P < 0·01; Fig. 4) and between the Ao seed bank and Calluna cover in 1990 (R2 = 0·95, n = 5, P < 0·01).
Plant height of deschampsia and calluna
Between 1991 and 1993, the average plant heights of Deschampsia and Calluna in the Deschampsia lawns were significantly lower on the plaggen soil than on the podzolic soil (Fig. 5; Deschampsia: F = 8·0, P < 0·01; Calluna: F = 11·3, P < 0·01).
After an initial rise during the first 5 years in all grazed habitats, species richness declined during the last 5 years in the two pioneer Calluna habitats (Fig. 6a). Species richness more or less stabilized in the mature Calluna heath and in the other habitats (Fig. 6a,b). The effect of grazing was most clearly shown by comparing the grazed and ungrazed treatments of the Deschampsia heath on plaggen soils and the Molinia heath (Fig. 6b and Table 4; grazing × year effect, P < 0·001). Species richness increased in the grazed treatments from 3·0 to 8·0 (Deschampsia heath) and from 3·7 to 9·3 (Molinia heath).
It may be argued that the increase in species richness by grazing is mainly due to the introduction of dung and urine. The plant species in the plots (see the Appendix) were classified as nitrophilous species (IN > = 6) and non-nitrophilous heathland species (IN < 6) according to their Ellenberg's nitrogen indication value, IN (1 = < IN > = 9; see the Appendix; Ellenberg 1979). The group of non-nitrophilous herbs displayed the strongest increase in species richness, and nitrophilous herbs the smallest (Fig. 7a,b). None of the 11 nitrophilous species increased by more than 10% between 1983 and 1993 (see the Appendix). Among the 19 non-nitrophilous heathland species, 10 species (see the Appendix) increased more than 10%, e.g. bramble (0–48%), birch (29–71%) and buckthorn Frangula alnus Mill. (29–41%). No species disappeared from the plots (see the Appendix). Some rare species established after 1983 outside the plots (e.g. Lycopodium clavatum L. and Ophioglossum vulgatum L.).
Habitat use by cattle
The cattle used habitats and dominant species in an uneven fashion. Time–budget studies and records of numbers of bites over 24-h periods between 1983 and 1985 (Bokdam, Gleichman & Batterink 1986; J. Bokdam, unpublished data) showed that the cattle grazed grass preferentially during the growing season. Between November and March, Calluna was added to their grass diet, reaching a maximum of 50% of all bites in January. The annual foraging pressure, based on foraging time and expressed in animal units ha−1, was higher in Deschampsia heath (0·28) than in Calluna heath (0·23), Molinia heath (0·16) and forest (0·12). Forest was intensively used as a shelter during the night.
Soil fertility displayed a gradient across the four habitats on podzolic soils, and significant differences were found between the podzolic and plaggen soils. The increase in the amount of organic matter in the topsoil and extractable nitrogen across the podzolic soils was caused by mechanical removal before the start of grazing. The low amount of organic matter in the Ao of the mown pioneer Calluna heath may have been due to the interrupted litter production or to litter removal by wind or run-off.
The different thickness of the Am and extractable phosphorus level (grazed treatments only) between Deschampsia heaths on podzolic and plaggen soil were explained as long-lasting effects of differential use as common pasture and arable field, respectively, during the Middle Ages.
The significantly higher extractable phosphorus and (not significantly) higher extractable nitrogen in the grazed Deschampsia heath on plaggen soil compared with the ungrazed treatment indicates accelerated decomposition and mineralization by grazing. Similar effects were found by Kemmers et al. (1996) in Deschampsia swards in pine forest.
Transitions between grass and heather
Grazing did not stop the ongoing invasion of Deschampsia in the non-turf-stripped Calluna habitats (Fig. 1b,c). The failed grass invasion in the turf-stripped pioneer Calluna heath (Fig. 1a) was attributed to nitrogen deficiency. Ageing and accumulation of litter and nitrogen may allow Deschampsia to invade in the future in gaps in this habitat. Grazing may have accelerated grass invasion in the mature Calluna (Fig. 1c). The mature and degenerating stages of Calluna are less tolerant of defoliation and trampling than the pioneer and building stages (MacDonald 1990). The prostrate growth forms, developed by layering and grazing (MacDonald et al. 1995), also seemed to be tolerant to trampling.
In the grazed grass heaths, Deschampsia cover did not decline despite intensive defoliation (Fig. 1d,e,g). This indicates a high tolerance of defoliation. Trampling, urine scorching and digging by rabbits and dung beetles created small gaps in the grass sward. The only declines of Deschampsia cover by defoliation occurred in the Deschampsia heath on plaggen soil, before the first measurement. Grazing of the tussocks exposed the bare intertussock areas, which were colonized by the grass after 1983 (Fig. 1e,g). Competition by taller Molinia (Fig. 1h) and trees (Fig. 1f) caused all other declines of Deschampsia cover. Until 1993, Deschampsia was only locally suppressed by expanding Calluna (Bokdam 1996).
Grazing induced a significant recovery of Calluna in the grass heaths on podzolic and peat soils (Fig. 1d,g), but not on plaggen soils. Key mechanisms were a successful establishment from the seed bank (Fig. 5), seedling survival and expansion. Recovery included four phases. (i) During the lawn development phase, grazing changed the grass tussocks into a lawn by defoliation and trampling. (ii) In the heather establishment phase, the first seedlings emerged in litter gaps. They probably originated from the seed bank. During later years recruitment was also found on compacted litter and in open lawn patches. These seedlings may have derived from seed rain. Rabbits heavily browsed Calluna seedlings and juveniles in localized areas. (iii) During the building phase, Calluna plants expanded in height and width (Fig. 1a–d), despite substantial browsing by cattle and rabbits. Browsing kept the plants more or less in a prostrate or creeping growth form with adventitious roots (MacDonald et al. 1995). Browsing and trampling may have stimulated layering and the development of adventitious roots in the moss-rich grass turf. It may also have favoured a dense canopy and increased the ability of Calluna to ‘walk’ over short-cropped grass lawns. Unbrowsed Calluna plants displayed the well-known, classical, developmental stages (Watt 1955). Layering and vegetative regeneration may have had a positive effect on the longevity of Calluna. Rabbits created flat, shaved, Calluna cushions, described earlier by Welch & Kemp (1973). (iv) In the final mosaic phase, which became apparent after 5 years of grazing, Calluna shrubs coexisted with grass in a dynamic mosaic (Bokdam 1996).
The failure of Calluna to colonize Deschampsia lawns on plaggen soil (Figs 1e,f and 2) was attributed to inhibition of germination, caused by the thick grass-rich Ao (Table 2), and to intensive browsing of the young plants (Figs 1e, 2 and 5). The inhibitory effect of the tough grass litter on the plaggen soil became obvious after a simple additional turf-stripping experiment. In two plots, each of 20 m2, the turf was removed in 1986. The turf stripping triggered the germination of Calluna and resulted in a mean Calluna cover of 50% by 1993. The lower plant height of Deschampsia and Calluna (Fig. 5) and the cushion form of Calluna confirmed the more intensive defoliation by cattle and rabbits on the nutrient-rich plaggen soil. Our findings agree with the higher grazing pressure by sheep and red deer on heather plants near grass patches (Hester & Baillie 1998; Palmer & Hester 2000). Impoverishment of the seed bank was not judged to be of great importance, as the low germinable seed density (96 m−2 from the Ao) did not prevent a successful recovery in the turf-stripped plots.
Cattle browsed and suppressed all tree species (see the Appendix) except the unpalatable pine and silver birch. To maintain open heathland, the site manager had to remove several hundreds of young trees, mainly from heather stands. Without removal, 10–20% of the open area might have been colonized by trees (W. Aandeweg, personal communication). Similar removals are practised in other heathland areas with free-ranging cattle (Westerhoff 1992). The observed differential tree encroachment in grass and Calluna heaths, and the frequency of bite marks, suggested that pine and birch juveniles were more severely suppressed in grass lawns than in Calluna heaths, more on plaggen soils than on podzolic soils, and more by rabbits than by cattle.
Scots pine, buckthorn, silver birch, rum cherry Prunus serotina Ehrh. and goat willow Salix caprea L. invaded in the ungrazed Deschampsia heath on plaggen soil (Fig. 1f). Their establishment followed substantial Deschampsia die-off after a prolonged snow cover during the winter 1985–86. The establishment after die-off and the failure of tree invasion in the vigorous ungrazed Molinia (Fig. 1h) suggest a threshold for tree invasion in ungrazed tall grass swards. Causal factors may be shading, small rodents and fungi. Runways and bite marks revealed the presence of small rodents in our exclosures.
Despite the increased frequency of bramble (see the Appendix), its cover remained low (< 1%). Nutrient deficiencies and browsing by rabbits, roe deer and horses may have checked its expansion. Cattle avoided bramble (J. Bokdam, unpublished data).
During the study, more than 20 adult trees in the forest and open heathland died as a result of ageing and wind-throw. They were all replaced by grass because cattle had suppressed the shade-tolerant mid-successional pedunculate oak Quercus robur L. and rowan Sorbus aucuparia L.
The positive effect on species richness was explained by cattle-mediated seed dispersal, release from competitive exclusion by suppression of taller dominants and by increased spatial heterogeneity (Bullock & Pakeman 1997; Olff & Ritchie 1998). The nitrophilous herbs (see the Appendix; e.g. Poa annua L. and Stellaria media (L.) Villa) occurred mainly on (old) dung patches. The increased frequency of non-nitrophilous herbs (see the Appendix; e.g. Carex pilulifera L. and Potentilla erecta (L.) Räuschel) was explained by improved germination conditions for seeds from the seed bank and the seed rain. The stabilization of the species richness (Figs 6 and 7) on a relatively low level may have been the result of the low pH and a failing long-distance dispersal (Bakker & Berendse 1999).
Proulx & Mazumber (1998) suggested reversal of grazing impact on plant species richness in nutrient-poor vs. nutrient-rich ecosystems. We found positive effects of grazing in grass heaths on poor and rich soils (Fig. 6). The decline of the species richness during the second 5 years in the most nutrient-poor Calluna habitats may support Proulx & Mazumber's (1998) theory. The decline was explained by out-shading of the small light-demanding heathland species by the closing Calluna canopy.
Cattle grazing generated shifting grass–heather mosaics in more or less homogeneous Calluna heaths (Fig. 1b,c) and grass heaths (Fig. 1d,g) on soils of intermediate fertility. No mosaics developed in the nutrient-poor turf-stripped pioneer Calluna heath and in the nutrient-rich Deschampsia heath on plaggen soil (Fig. 1a,e). Diemont & Heil (1984) concluded that the sustained grass dominance in eutrophicated (non-grazed) Dutch heathlands conflicted with cyclic succession of the four classic Calluna stages (Watt 1955; Barclay-Estrup & Gimingham 1969) and with the Calluna–moss–grass cycle in fixed drifting sand (Stoutjesdijk 1959). In the Wolfhezerheide, free-ranging cattle could change the linear heather–grass transition into a heather–grass–heather cycle because invading trees were removed. Without tree removal, a heather–forest–grass–heather cycle would have been generated. Grazing acted as the driving force for the cycling in two ways: (i) by inducing the invasion of grass in degenerating forest and heather; and (ii) by inducing the invasion of Calluna and other unpalatable woody pioneers in the grass lawn after a period of exploitation and nutrient depletion.
Despite the obvious light-mediated effect of grazing on the grass invasion in ageing Calluna and the recovery of Calluna in grass heath, we may not conclude that grazing diminished the importance of nutrients for the heather–grass balance.
The nutrient requirements of Deschampsia and Calluna were apparently met in all open habitats of the study area, except in the turf-stripped Calluna heath. Here, nitrogen deficiency probably excluded Deschampsia (Table 2 and Fig. 1a). This result agrees with the relatively high nitrogen demand (Aerts 1993a,b; Alonso & Hartley 1998) and low phosphorus requirement (Bokdam & Wallis de Vries 1992) of this grass species. The shade-tolerant Deschampsia was only locally restricted by shading (Fig. 1f,h).
Calluna was indirectly limited by nutrients. The relatively high nutrient availability allowed the grasses to inhibit the germination and seedling survival of Calluna in the grass heaths. By short-cropped lawns and gaps, cattle eliminated the nutrient-based competitive advantage of the established Deschampsia over the establishing Calluna. The grass sward on the plaggen soil with grass litter may have been more resistant to trampling, gap creation and nutrient depletion than the grass sward on the podzolic soils with Calluna litter. The rich Calluna recruitment on the turf-stripped plots on plaggen soil emphasized the essential role of gaps. Ten years of nutrient depletion in a Deschampsia heath at Hoog Buurlo (the Netherlands) by hay making (‘defoliation without treading gaps’) did not result in any Calluna recovery, despite a rich seed bank (Diemont & Lindhorst Homan 1989).
There is growing evidence that soil fertility may also affect the disturbance risk and longevity of Calluna. The more intensive browsing of Calluna on the plaggen soil (compared with the podzolic soil; Fig. 5) might be attributed to three nutrient-mediated mechanisms: (i) a higher nutritional value of Calluna; (ii) a higher intrinsic palatability of Calluna; and (iii) a higher associational palatability, provided by a more productive and palatable grass matrix. Fertilization with a compound fertilizer including nitrogen and phosphorus induced higher nitrogen and lower lignin concentrations in Calluna shoots (Iason & Hester 1993). Brunsting & Heil (1985) found a positive effect of nitrogen availability on foliage nitrogen content and the performance of the heather beetles. Calluna plants growing near grass patches were more intensively used than individuals at larger distances (Clarke, Welch & Gordon 1995; Hester & Baillie 1998). High nitrogen availability may also increase the risk of damage by frost and drought for Calluna (Berdowski 1993). The findings emphasize that Deschampsia and Calluna are both affected by nutrients, but in different ways and directions. Deschampsia seems to be directly limited by nutrients as a resource for its performance. Calluna is affected indirectly and in the opposite direction by the same nutrients. High levels of nitrogen increased the competitive ability of competing grasses (Aerts 1993b) and its disturbance risk. The earlier grass invasion and failing Calluna recovery on the plaggen soil in our study area may ultimately be attributed to a higher soil fertility. It shows that the soil fertility pattern created by a medieval sedentary farming system may still control present day vegetation patterns.
The combination of a low phosphorus and high nitrogen availability attributes a limiting role to phosphorus for species with a relative high phosphorus requirement. Such species (e.g. bramble) might be expected to increase rapidly after phosphorus inputs (e.g. on dung patches).
A preliminary calculation revealed that the nutrient removal by defoliation exceeded the atmospheric inputs for potassium and phosphorus on intensively grazed grass lawn patches. Nitrogen inputs were nearly completely removed. As a consequence, lawn patches underwent phosphorus and potassium depletion until excreta replenished them. The calculated frequency of dung and urine deposition on these lawns was approximately once per 20–30 years, assuming no resting activities on the lawn (Bokdam 1989). This means that nutrient depletion may play a role in the establishment of Calluna in grass lawns.
At the scale of the whole grazed area, free-ranging cattle removed only a minor proportion of the atmospheric nutrient input. The animals excreted a major proportion of the ingested nutrients and redistributed a substantial amount from the grass lawns to the forest (J. Bokdam, unpublished data).
Continued free-ranging cattle grazing will lead to a further eutrophication of the open heathland because of ongoing atmospheric inputs and because nutrient-rich forest patches will be converted into grassland. Cessation of tree removal would lead to the restoration of a half-open wood–pasture mosaic (Pott & Hüppe 1991) This process is accelerated by integrated grazing of heathlands, arable fields, improved grasslands and forest. Eutrophication may increase in the long term the grass proportion in the open heathland. A higher soil fertility will induce a higher grazing pressure in grass lawns. It may progressively hamper the invasion of woody pioneers of nutrient-poor environments (e.g. Calluna, pine and birch) and change the successional pathway from a grass–heather–pine–birch–oak route to a grass–bramble–oak–beech route (Bokdam 1998).
Replacement of the actual year-round grazing by summer grazing might increase the grass consumption and favour the heather proportion in the grass–heather mosaic.
Replacement of free-ranging by traditional herding (with a daily return of the herd to the stable and a daily foraging time of about 8 h) would remove about 60% of the excreta, i.e. 40 kg nitrogen and 2·5 kg phosphorus per animal unit (or 8 kg N and 0·5 kg P ha−1). Removal of the total actual atmospheric nitrogen input of the grazed area would require the replacement of all forest and heather stands by intensively used grass lawns (J. Bokdam, unpublished data).
Multi-species grazing combining cattle with smaller herbivores (e.g. sheep, red deer and rabbits), might suppress shrubs and trees more effectively. Replacement of the large grazers by smaller ones may lead to dominance of coarse grasses (e.g. Molinia and Calamagrostis epigejos (L.) Roth) or other potential dominants that are unpalatable to smaller herbivores (J. Bokdam, personal observations)
We conclude that free-ranging grazing without additional management will generate dynamic tree–grass–heather mosaics in the long term. These wood–pasture mosaics are fundamentally different from the traditional heathlands in which trees and grasses were excluded by the land use. Free-ranging grazing combined with tree cutting appears to be a suitable management regime for the maintenance of open heathlands with dynamic grass–heather mosaics. Herding or a combination of free-ranging grazing with burning or turf stripping seems unavoidable if pure heather-dominated open heathlands are to be maintained under the present environmental conditions.
Natuurmonumenten granted admission to the study area and provided material and assistance in the field by W. Aandeweg. Historical information was received from W. Aandeweg, the late B. Janssen and B.W. Jacobs. We thank the students M. Augustijn, M. Batterink, A. van Boxtel, M. Hagg, F. Luttiken, M. Mans, N. Van der Ploeg, H. Smits, A.Vermaat, L.Vink and T. van Vlerken for their contributions to the vegetation assessments, the soil investigation and the seed bank analysis. J. Van Walsum and W. van Vark gave assistance during the chemical analysis. We thank H. Olff for his substantial help during the data analysis and R. Smit for his technical assistance. Comments from F. Berendse, H. Olff, H. van Oene, E.S. Bakker and two anonymous referees substantially improved an earlier draft.
Received 9 July 1998; revision received 19 January 2000
Table 5. Frequency (%) of vascular plant species in the 17 grazed plots established in 1983 in open heathland, in 1983, 1988 and 1993. Nomenclature follows Van der Meijden (1990). The herbaceous species were arranged as nitrophilous and non-nitrophilous species according to their nitrogen indication value (IN) of Ellenberg (1979) (1 = extreme nitrogen-poor environment; 9 = extreme nitrogen-rich environment; X = indifferent; ? = unknown).
Species (IN value) occurring in other years: Epilobium sp. (?), Holcus lanatus (4), Holcus mollis (3), Luzula campestris (2), Spergularia rubra (4), Chamerion angustifolium (8) Lolium perenne (7), Polygonum persicaria (7), Senecio sylvaticus (8), Solanum nigrum (8), Prunus sertina (?).