SEARCH

SEARCH BY CITATION

Keywords:

  • arable reversion;
  • biodiversity;
  • land-use change;
  • soil fertility

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information
  • 1
    Agricultural intensification has resulted in the reduction and fragmentation of species-rich grasslands across much of western Europe.
  • 2
    We examined the key ecological processes that limit the creation of diverse grassland communities on ex-arable land in a multi-site experiment over a wide variety of soil types and locations throughout lowland Britain.
  • 3
    The results showed it was possible to create and maintain these communities successfully under a hay-cutting and grazing management regime. Furthermore, there was a high degree of repeatability of the treatment effects across the sites.
  • 4
    Lack of seed of desirable species was the key factor limiting the assembly of diverse grassland communities. Sowing a species-rich seed mixture of ecologically adapted grassland plants was an effective means of overcoming this limitation. Community assembly by natural colonization from the seed bank and seed rain was a slow and unreliable process. However, there was no evidence to suggest that sowing a species-poor grass-dominated seed mixture made the vegetation any less susceptible to colonization by desirable species than allowing natural regeneration to take place.
  • 5
    Deep cultivation caused significant reductions in soil P and K concentrations across the sites. This had a significant beneficial effect on the establishment and persistence of sown forbs in all years. It also resulted in a significant reduction in the number of unsown weedy grasses. However, for both variables these differences were very small after 4 years.
  • 6
    Sowing a nurse crop significantly reduced the number of unsown grass species, but had no beneficial effect on the establishment of desirable species.
  • 7
    Treatments sown with the species-rich seed mixture following deep cultivation corresponded most closely to the specified target communities defined by the UK National Vegetation Classification. Natural regeneration and treatments sown with the species-poor seed mixture were much less similar to the target. The sites on circum-neutral soils achieved the greatest degree of similarity to the target. Those on calcareous and acid soils failed to achieve their targets and most closely resembled the target for neutral soils. This reflected the poor performance of the sown preferential species for these communities.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

Unimproved species-rich grasslands were once widespread over much of western Europe (van Dijk 1991), where they were mostly a product of low-intensity farming systems. Such grasslands are of high conservation importance because of the diverse floral and faunal assemblages they support (Duffey et al. 1974). Post-war intensification of agricultural management together with land-use change have resulted in the large-scale degradation and fragmentation of this habitat. It has been estimated that unimproved grassland declined in area by 97% in lowland England and Wales between 1930 and 1984 (Fuller 1986), so that no more than 865 km2 of this habitat now remain (Crofts & Jefferson 1999). As a result, the UK Biodiversity Action Plan (BAP) requires the sympathetic management of all remaining grassland fragments together with the recreation of 2000 ha of various types of species-rich grassland over the next decade (Anonymous 1998). Much of this will be achieved within the agri-environment schemes (Ovenden, Swash & Smallshire 1998) such as the Environmentally Sensitive Areas (ESAs).

Carefully targeted habitat creation can be a useful tool for the conservation of such biodiverse habitats. However, the results of previous restoration under the agri-environment schemes have been inconsistent (Carey et al. 2001). In order to understand the reasons for this it is important to consider the limitations on ecological restoration (Mortimer, Hollier & Brown 1998; Bakker & Berendse 1999). Furthermore, it is essential to test these hypotheses over a wide range of sites and conditions. It has been suggested that the assembly of diverse plant communities on intensively managed agricultural land is severely limited by a lack of propagules of desirable species, both in the seed bank (Hutchings & Booth 1996; Bekker et al. 1997; Pywell, Putwain & Webb 1997) and in the surrounding landscape (Poschlod et al. 1998). In order to accelerate the assembly of species-rich communities it may be necessary to introduce seed of desirable species deliberately, either into gaps created in the grassland swards or to a suitably prepared seed bed on ex-arable land (Wells, Cox & Frost 1989; Hopkins et al. 1999).

A further important determinant of plant community assembly is competition for space and resources (Bullock 1996). The rapid establishment of competitive and undesirable species in the vegetation might prevent the colonization of desirable species or out-compete established species for resources (Hansson & Fogelfors 1998). However, with few exceptions (van der Putten et al. 2000), little work has been carried out into the importance of these competitive interactions in influencing the outcome of succession. An alternative aspect of biotic interactions is facilitation. The bare soil of ex-arable land may be quite inhospitable to seedlings of species adapted to regenerate in grasslands. Cover by established plants may protect seedlings from abiotic extremes, such as low moisture or freezing temperatures (Callaway 1995). Such facilitation can outweigh negative competitive effects of other plants (Brooker & Callaghan 1998). Nurse plants might help the establishment of sown species by suppressing more competitive species and/or ameliorating a harsh abiotic environment. They have been used routinely in forestry (Choi & Wali 1995; Ray & Brown 1995) but their potential in grassland restoration is poorly studied.

A number of abiotic factors are also likely to have important effects on key ecological processes controlling restoration, both directly and indirectly through their interaction with biotic factors. High residual soil fertility resulting from intensive farming is likely to place severe constraints on the enhancement and long-term maintenance of plant species diversity (Marrs 1993; Pywell, Webb & Putwain 1994). In particular, low levels of soil P seem to be a pre-requisite for long-term species co-existence in grasslands (Janssens et al. 1998). Techniques to reduce soil fertility have been reviewed by Marrs (1993). These can be divided into techniques that seek to: (i) remove or dilute the pools of nutrients; (ii) manipulate the stores and fluxes of nutrients in the soil to limit nutrient supply; and (iii) increase off-take.

A review of past attempts at habitat restoration concluded that a major cause of failure was a lack of clearly stated objectives or targets at the start of the project (Pywell & Putwain 1996). Without such targets, it may be impossible to monitor the progress of restoration and take appropriate action to ensure success. Suitable ecological targets for restoration might be the reinstatement of a particular species, a specific plant community, or a fully functioning ecosystem (Bakker et al. 2000). The National Vegetation Classification (NVC) is a systematic phytosociological description of British plant communities (Rodwell 1991, 1991–2000). The NVC describes 860 communities and subcommunities that have been derived from the analysis of 35 000 quadrats interpreted using information on management and environmental variables. Such classification systems provide a useful template for selecting appropriate ecological targets for habitat restoration and for monitoring progress towards their attainment (Rodwell & Patterson 1994).

This study examined the effectiveness of practical treatments to overcome possible limitations on diverse grassland creation. These treatments were designed specifically to investigate the relative importance of the following hypotheses: (i) diverse plant communities described by the NVC are useful ecological targets for restoration on ex-arable land; (ii) the assembly of these plant communities is seed limited; (iii) reduction of soil fertility can promote assembly of diverse communities; and (iv) establishment of desirable species can be enhanced by the amelioration of abiotic and biotic conditions through sowing a nurse crop. In order to test the generality of the treatment effects, they were applied at five sites distributed across different regions of southern England, with differing climates, management histories, soil types and nutrient status. Therefore a further hypothesis was tested by this meta-analysis: (v) these treatment effects are reproducible over a range of sites.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

study sites

In September 1994 one cereal field was selected from five ESAs in lowland England (Table 1). The sites represented a range of soil types and fertility. Each site had grown cereals, with the usual fertilizer and pesticide additions, in the previous season.

Table 1.  Descriptions of the experimental sites
SiteLocationElevation (m a.s.l.)Annual rain (mm)Soil typeNVC target community (Rodwell 1992)European classification (Braun-Blanquet & Tüxen 1952; Willems 1978)
South Downs50°54′N 0°35′W195700–800Brown calcareous earthCG3b Bromus erectus grassland: Centaurea nigra subcommunityCirsio–Brometum
South Wessex Downs50°8′N 2°10′W230800–1000Brown calcareous earthCG3b Bromus erectus grassland: Centaurea nigra subcommunityCirsi–Brometum
Upper Thames Tributaries51°52′N 1°03′W 70600–700Alluvial gleyMG5a Cynosurus cristatus–Centaurea nigra grassland: Lathyrus pratensis subcommunityCentaureo–Cynosuretum cristati typicum
Norfolk Broads52°44′N 1°36′E< 5500–600Humic alluvial gleyMG5a Cynosurus cristatus–Centaurea nigra grassland: Lathyrus pratensis subcommunityCentaureo–Cynosuretum cristati typicum
Suffolk River Valleys52°1′N 1°20′E  10500–600Brown sandMG5c Cynosurus cristatus–Centaurea nigra grassland: Danthonia decumbens subcommunityCentaureo–Cynosuretum cristati typicum, Sieglingia decumbens

targets for restoration

The NVC (Rodwell 1992) was used to derive target diverse grassland communities that were appropriate to the location, soil, hydrology and proposed management of each site. These are given in Table 1, together with the equivalent European grassland communities. All these grassland communities are typically managed by hay cutting and aftermath grazing (Rodwell 1992; Walker & Pywell 2000).

restoration treatments

A randomized experiment with four replicate blocks, each with seven restoration treatments, was set up at each site. The treatments were:

  • 1
    natural regeneration from cereal stubble;
  • 2
    shallow cultivation + species-poor ESA seed mixture;
  • 3
    shallow cultivation + species-rich NVC seed mixture;
  • 4
    shallow cultivation + species-rich NVC seed mixture + nurse crop;
  • 5
    deep cultivation + species-poor ESA seed mixture;
  • 6
    deep cultivation  + species-rich NVC seed mixture;
  • 7
    deep cultivation + species-rich NVC seed mixture + nurse crop.

The core of each plot measured 6 × 4 m with a discard area at one end. The plots were separated by 1-m guard rows. The cereal stubble was left intact for the natural regeneration treatment (1). Shallow cultivation was achieved with harrows or discs depending on the soil type. This was contrasted with deep cultivation to a depth of 30–40 cm using a plough, a method that was intended to bury fertile topsoil and the seed bank of undesirable species. A seed mixture designed to create the composition and structure of each NVC target community was made up from commercially available seed of native species (Appendix 1). Seed of a small number of NVC community constant species was unavailable (e.g. Trifolium pratense in MG5a). Other constant species, such as Dactylis glomerata, were considered to be sufficiently common in agricultural landscapes to exclude them from the seed mixtures. All forb species and the majority of the grasses were of lowland British provenance. These seed mixtures typically comprised between 25 and 41 species, of which 80% by weight were grasses and 20% forbs. The seed rate was between 24 and 28 kg ha−1. The performance of this species-rich NVC seed mixture was contrasted with that of the more species-poor seed mixture, which was recommended in the Department of Environment, Food and Rural Affairs (DEFRA) guidelines for establishing moderately diverse grassland on arable land. This prescription varied slightly between each ESA. Typically it recommended sowing six to eight common grasses at between 20 and 31 kg ha−1 (Appendix 1). However, the seed mixture recommended for the South Wessex Downs ESA required the inclusion of nine forbs with the basic mix of eight grass species. Finally, the effect of an annual nurse crop on forb seedling recruitment in the NVC seed mixtures was investigated by sowing Westerwolds rye-grass Lolium multiflorum at a rate of 20 kg ha−1.

In the first year after sowing the vegetation was cut and removed in early June to prevent the nurse crop from persisting. In subsequent years cutting was carried out in July and the herbage was left to dry for several days before removal to allow some seed return. In all years the aftermath growth was grazed by sheep between October and December at an annual stocking rate of 0·4–0·7 livestock units ha−1 year−1 depending on the grass growth. This typically equated to between 25 and 40 sheep ha−1 grazing for 6–8 weeks.

ecological assessments

Soil nutrients

Soil samples were collected from each plot following cultivation in September 1994. Ten samples were taken at random from each plot using a 6-cm diameter and 20-cm deep auger. The samples were split into 0–5 cm and 6–20-cm depth fractions and bulked. A subsample of c. 500 cm3 was analysed for soil nutrients according to the methodology described by Allen et al. (1974). Soil sampling and analysis were repeated in September 1998.

Vegetation

In early June of each year three quadrats were placed at random within each plot, avoiding a 1-m strip around the edge. Each quadrat measured 40 × 40 cm and was subdivided into 16 10 × 10-cm cells. The presence of all vascular plants was recorded in each cell. Abundance of each species was expressed as the percentage occupancy of the 48 cells per treatment. Species were classified as sown and unsown grasses and forbs. These were summed to provide a measure of plant species richness. It was difficult to distinguish between the two sown cultivars of Festuca rubra and Festuca ovina. These were therefore treated as a single species aggregate (Festuca rubra/ovina). The nomenclature for the vascular plants follows Stace (1997).

statistical analysis

The effects of restoration treatment, year and treatment × year interactions on the vegetation and soil variables across all sites were examined using analysis of variance (anova) with repeated measures (SAS for Windows Version 8). Vegetation data included the species richness of sown and unsown forb and grass species and the total number of species. Because the restoration treatments were not fully factorial, their effects were analysed using three different anova structures (analyses A–C).

Analysis A

Effects of cultivation depth (deep vs. shallow) and seed mixture (ESA vs. NVC). Treatments 2, 3, 5 and 6 were compared, giving factorial combinations of the cultivation and seed-mixture treatments.

Analysis B

Effects of cultivation depth (deep vs. shallow) and nurse crop (nurse crop vs. no nurse crop) within the NVC seed-mixture treatment. Treatments 3, 4, 6 and 7 were compared, giving factorial combinations of the cultivation and nurse crop treatments. Only the NVC seed mixture was used in these four treatments (see above). Therefore, these results were only applicable to the NVC seed mixture. Furthermore, this comprised a second, partially independent, analysis of cultivation depth effects for comparison with analysis A.

Analysis C

Effects of seed addition: natural regeneration (no seed addition) vs. ESA seed mixture vs. NVC seed mixture, by comparison of treatments 1, 2 and 3. The two seed-addition treatments 2 and 3 had shallow cultivation, which we considered as the minimum seed bed treatment, and thus the most comparable to natural regeneration. The natural regeneration differed from the other two treatments not only in having no seed added, but also in having no cultivation.

In all anovas blocks were nested within sites and therefore the site mean square was tested against the block mean square while other factors were tested against the error mean square. Furthermore, each anova had repeated measures by including data from every measurement year (2 years for soil data and 4 years for vegetation data). Univariate and multivariate repeated-measures anovas (Maxwell & Delaney 1990) were carried out using PROC GLM in SAS (1990) and both gave identical qualitative results. Within the univariate analyses, ɛ-values for adjustment of degrees of freedom according to the amount by which the population covariance matrix departs from homogeneity (Maxwell & Delaney 1990) were calculated using the Geisser–Greenhouse method and the less conservative Huynh–Feldt method (Maxwell & Delaney 1990). Both gave the same qualitative results, and the results from the Geisser–Greenhouse adjustment are reported.

As well as analyses of the vegetation and soil data, analyses A–C were carried out on the arcsine-transformed mean shoot frequency of selected plant species. These were 24 sown species that had a constancy value of ≥ 3 in NVC target communities, i.e. they were common species in target communities and thus we were especially interested in their responses to the treatments. To these were added a further six sown and 10 unsown species that were commonly found in the restored vegetation. The latter species were found in 120 or more plots across the whole experiment in any given year. For the sown species, only sites where the species had been sown in the seed mixture were included in the analysis (Appendix 1).

Finally, the composition of the restored grassland communities resulting from the natural regeneration (1), and the ESA (5) and NVC (6) seed mixtures both following deep cultivation, were compared with that of the diverse grassland target communities, together with all other British vegetation types described by the NVC, using the Tablefit computer program (Hill 1996). Tablefit calculates percentage fit between the observed species composition and constancy, and that specified for each NVC subcommunity in turn (Wilson et al. 2000). Both the percentage fit to the target community and the identity of the best fitting community are reported.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

Analyses A–C on soils, vegetation and individual species showed a large number of site and site × treatment interactions. Neither effects are examined in detail here. Site effects were to be expected through environmental differences (see below) and differences in compositions of seed mixtures. Site × treatment effects were less important for our purposes than consistent main treatment effects, as our aim was to find generalizations about appropriate treatments. In this section we concentrate on main treatment effects, year effects and year × treatment interactions. Higher order treatment interactions are only discussed where they were consistent and biologically important.

soil nutrients – site differences

The five sites chosen for this study represented the reasonably typical range of soil pH and fertility for arable sites across lowland Britain (Allen et al. 1974). Overall there was considerable variation in pH and nutrient concentrations between sites (Table 2). There were relatively few differences in nutrient concentrations between the upper and lower depth fractions at each site. The calcareous soils of the South Downs had the highest pH (8·1) and the Suffolk River Valleys the lowest (5·6). The sandy soils of the latter site had significantly higher P than the other sites (71 mg l−1), and the clay soil of the Upper Thames Tributaries had the lowest P (13 mg l−1). The soil of the South Wessex Downs had a significantly greater K compared with the other sites (348 mg l−1), with the South Downs having the lowest K (169 mg l−1). Finally, the South Downs had significantly more N (%N) in both depth fractions compared with the other sites (0·6%), suggesting this site may have been grassland in more recent times.

Table 2.  Changes in mean soil nutrient concentrations at the five study sites between 1994 and 1998. Soil samples were taken at two depths: 0–5 cm and 6–20 cm
 Depth (cm)South DownsSouth Wessex DownsUpper Thames TributariesNorfolk BroadsSuffolk River Valleys
 1994199819941998199419981994199819941998
pH0–5  8·1 ± 0·01  7·8 ± 0·03  6·9 ± 0·04  6·4 ± 0·02  6·5 ± 0·1  6·4 ± 0·06  7·1 ± 0·04  6·4 ± 0·03  5·6 ± 0·1  5·7 ± 0·1
 6–20  8·1 ± 0·01  7·9 ± 0·03  7·0 ± 0·04  6·5 ± 0·03  6·4 ± 0·1  6·6 ± 0·06  7·1 ± 0·04  6·7 ± 0·02  5·7 ± 0·04  5·7 ± 0·1
P mg l−10–5 16·3 ± 1·1 15·5 ± 0·8 20·7 ± 0·6 18·0 ± 0·8 13·4 ± 1·0 10·4 ± 0·5 28·1 ± 1·3 28·0 ± 0·9 70·8 ± 2·6 63·6 ± 2·1
 6–20 13·0 ± 0·4  8·4 ± 0·5 20·5 ± 0·7 14·0 ± 0·5 10·9 ± 0·5  7·9 ± 0·4 28·2 ± 1·2 24·5 ± 0·8 71·0 ± 2·8 62·4 ± 2·0
K mg l−10–5168·7 ± 14·4183·6 ± 10·3347·9 ± 17·1351·2 ± 15·2241·3 ± 15·4  190 ± 10·7201·9 ± 10·1237·6 ± 10·2210·7 ± 22·9235·1 ± 13·7
 6–20113·1 ± 4·5 69·5 ± 2·3311·4 ± 9·3221·9 ± 11·4205·8 ± 10·0  157 ± 10·9200·5 ± 13·6160·6 ± 7·0148·6 ± 7·5163·0 ± 12·3
Mg mg l−10–5 48·9 ± 1·0 65·1 ± 2·4 48·4 ± 2·2 67·8 ± 2·5190·1 ± 5·3163·9 ± 4·4 66·4 ± 2·9 76·5 ± 2·8 72·0 ± 4·3 76·0 ± 4·2
 6–20 47·4 ± 0·8 32·2 ± 0·8 45·7 ± 1·9 47·7 ± 2·0190·6 ± 4·0173·2 ± 5·1 68·6 ± 3·6 66·4 ± 2·9 67·2 ± 1·7 62·5 ± 2·5
%N0–5  0·6 ± 0·02  0·5 ± 0·02  0·3 ± 0·01  0·2 ± 0·01  0·4 ± 0·01  0·3 ± 0·02  0·1 ± 0·01  0·1 ± 0·01  0·1 ± 0·01  0·1 ± 0·01
 6–20  0·6 ± 0·01  0·5 ± 0·01  0·3 ± 0·01  0·2 ± 0·01  0·4 ± 0·01  0·3 ± 0·02  0·1 ± 0·01  0·1 ± 0·01  0·1 ± 0·01  0·1 ± 0·01

After 4 years the differences in pH and soil nutrient concentrations between sites remained but were less pronounced (Table 2). In general, there were often small declines in pH and nutrient concentrations in the upper depth fraction of the soil, but there were often larger declines in the lower depth fraction.

effects of cultivation on soil fertility and species number

The nutrient status and pH of the uncultivated soil of treatment 1 (natural regeneration) were compared with the minimally cultivated soil of treatment 2 (ESA seed mix) using repeated-measures anova. Soil P, N, Mg, K and pH did not differ between these treatments in either 1994 or 1998 and there were no year × treatment interactions (results not shown). Further analyses therefore only compared the effects of shallow and deep cultivation on soil nutrients.

Analysis A on the soil data showed there were no significant effects of cultivation depth on soil pH at either 0–5 cm or 6–20 cm (Table 3). However, there was a highly significant year effect reflecting the decline in pH over 4 years. The decline in pH in the 0–5-cm depth fraction was significantly greater following deep cultivation. This treatment also caused a significant reduction in extractable P compared with shallow cultivation at both depths. The magnitude of this reduction was greater in the 0–5-cm fraction compared with the 6–20-cm fraction. However, the highly significant year effect reflected the decline in P with time in both treatments. This decline was relatively greater in the shallow cultivation treatment. Deep cultivation also caused a highly significant reduction in K in the 0–5-cm fraction, but not in the 6–20-cm fraction. K concentrations recovered in this treatment after 4 years, but declined slightly in the shallow cultivation treatment. There was no significant effect of cultivation on Mg at either depth. However, there was a significant decline in Mg in the 6–20-cm fraction after 4 years. Finally, there was no significant effect of cultivation on %N at either depth. However, %N did decline by a small amount at both depths over the period of the experiment. Examination of the effects of cultivation on soil fertility using analysis B showed a similar pattern to analysis A (Appendix 2).

Table 3.  Summary of analysis A repeated-measures anova showing the effects of cultivation on soil fertility at the two soil depths. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
 Shallow cultivation (2, 3)Deep cultivation (5, 6)anovaF-values
 1994199819941998Cultivation d.f. = 1, 45Year d.f. = 1, 45Year × cultivation d.f. = 1, 45
  1. Treatment 2 = shallow cultivation + species-poor ESA seed mix; treatment 3 = shallow cultivation + species-rich NVC seed mix; treatment 5 = deep cultivation + species-poor ESA seed mix; treatment 6 = deep cultivation + species-rich NVC seed mix.

pH (0–5 cm)  6·8  6·6  6·9  6·5 0·26 NS103·15***    6·55*
pH (6–20 cm)  6·8  6·7  6·9  6·7 0·01 NS 54·40***    1·68 NS
P mg l−1 (0–5 cm) 32·5 28·0 27·3 25·714·88*** 16·16***    3·35 NS
P mg l−1 (6–20 cm) 30·6 24·5 27·7 22·5 4·49*140·28***    0·90 NS
K mg l−1 (0–5 cm)273·7255·1179·7232·224·23***  4·02 NS   17·74***
K mg l−1 (6–20 cm)196·4157·3190·6141·4 2·16 NS 77·23***    1·00 NS
Mg mg l−1 (0–5 cm) 83·5 90·2 87·1 88·8 0·09 NS  3·05 NS    1·11 NS
Mg mg l−1 (6–20 cm) 83·3 74·8 82·0 73·4 1·00 NS 36·20***< 0·01 NS
%N (0–5 cm)  0·3  0·3  0·3  0·2 1·37 NS 41·23***    3·81 NS
%N (6–20 cm)  0·3  0·2  0·3  0·2 0·01 NS 98·65***    1·49 NS

The effects of cultivation depth on numbers of sown and unsown species examined by analysis A showed a weakly significant (P < 0·05) beneficial effect of deep cultivation on the establishment and persistence of sown forb species in all years (Table 4). Deep cultivation also resulted in a significant reduction in the number of unsown grass species recorded in all years. However, for both variables these differences were very small, especially by 1998. There were no significant effects of cultivation on numbers of sown grass, unsown forb or total species richness. Furthermore, there were no significant year × cultivation interactions. A more pronounced, but otherwise identical, pattern of response was found when the effects of cultivation were examined by analysis B (Appendix 3).

Table 4.  Summary of analysis A repeated-measures anova showing the effects of cultivation on vegetation variables across all sites. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
Species numberShallow cultivation (2, 3)Deep cultivation (5, 6)anovaF-values
19951996199719981995199619971998Cultivation d.f. = 1, 45Year d.f. = 3, 135Year × cultivation d.f. = 3, 135
  1. Treatment 2 = shallow cultivation + species-poor ESA seed mix; treatment 3 = shallow cultivation + species-rich NVC seed mix; treatment 5 = deep cultivation + species-poor ESA seed mix; treatment 6 = deep cultivation + species-rich NVC seed mix.

Sown forbs 4·3 4·5 4·8 5·7 4·9 4·8 5·1 5·8 4·1* 15·3***0·5 NS
Sown grasses 6·6 7·3 7·4 7·4 6·4 7·7 7·2 7·5 0·1 NS 16·1***1·6 NS
Unsown forbs 5·4 2·5 1·7 2·6 5·9 2·7 1·9 2·5 1·9 NS163·9***0·9 NS
Unsown grasses 5·0 2·2 2·8 2·3 4·0 2·0 2·2 1·911·4** 78·1***1·7 NS
Total species21·316·416·718·021·317·416·517·7 0·1 NS 61·0***1·1 NS

effects of seed addition and mixture type on species richness

Analysis C showed there were significantly higher numbers of sown forbs and grasses recorded in the NVC seed-mixture treatment (3) compared with the ESA seed mixture (2) and natural regeneration treatment (1) in all years (Table 5a). These variables showed highly significant treatment × year interactions that reflected the large increase in sown forbs in both the natural regeneration (Table 5a; analysis C) and ESA seed-mixture treatments (Table 5b; analysis A) with time compared with the NVC seed mixture. The number of sown grasses increased between 1995 and 1998 in the NVC treatments, and only between 1994 and 1995 for the natural regeneration treatment. After 4 years there were few differences in the number of sown species in the natural regeneration (7·9) and ESA seed-mixture treatments (8·9; Table 5a). The vegetation resulting from the NVC seed mixture had approximately twice the number of sown species (17·3; Table 5b).

Table 5.  Summary of repeated-measures anova showing the effects of (a) seed-addition treatments (analysis C) and (b) seed mixture type (analysis A) on vegetation variables across all sites. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
Species numberNatural regeneration (1)ESA seed mixture (2)NVC seed mixture (3)anovaF-values
199519961997199819951996199719981995199619971998Seed addition d.f. = 2, 30Year d.f. = 3, 90Year × seed addition d.f. = 6, 90
(a) Seed-addition treatments (analysis C)
Sown forbs 0·1 0·8 1·9 2·9 0·6 0·6 2·0 2·8 8·1 8·3 7·6 8·61120·8***36·4*** 8·5***
Sown grasses 1·0 2·4 3·3 4·9 5·5 6·0 5·5 6·1 7·7 8·6 9·2 8·6 283·8***30·4***12·5***
Unsown forbs 4·9 4·1 4·9 6·0 5·2 2·2 1·5 2·7 5·5 2·7 1·8 2·4  57·2***44·5***14·3***
Unsown grasses 4·4 3·9 4·9 4·8 4·8 2·0 2·9 2·3 5·1 2·3 2·7 2·3  37·0***28·7*** 8·0***
Total species10·411·315·118·716·210·812·014·026·422·021·422·0 250·4***24·5***29·2***
Species numberESA seed mixture (2, 5)NVC seed mixture (3, 6)anovaF-values
19951996199719981995199619971998Seed mixture d.f. = 1, 45Year d.f. = 3, 135Year × seed addition d.f. = 3, 135
  1. Treatment 1 = natural regeneration; treatment 2 = shallow cultivation + species-poor ESA seed mix; treatment 3 = shallow cultivation + species-rich NVC seed mix; treatment 5 = deep cultivation + species-poor ESA seed mix; treatment 6 = deep cultivation + species-rich NVC seed mix.

(b) Seed-mixture type (analysis A)
Sown forbs 0·6 0·7 2·0 2·5 8·6 8·6 7·8 9·03934·0*** 15·3***15·0***
Sown grasses 5·4 6·3 5·3 6·0 7·6 8·7 9·3 8·9 667·0*** 16·1***12·4***
Unsown forbs 5·8 2·5 1·8 2·7 5·4 2·7 1·8 2·3   0·6 NS163·9*** 1·1 NS
Unsown grasses 4·4 2·0 2·6 2·3 4·6 2·2 2·4 1·9   0·0 NS 78·1*** 0·9 NS
Total species16·311·511·813·526·322·321·422·2 820·9*** 61·0*** 2·4 NS

After the first year there were significantly more unsown forb and grass species in the natural regeneration treatment (Table 5a) compared with the ESA and NVC seed mixtures. There were no significant differences in the number of unsown forbs recorded in these latter treatments (Table 5b). However, the number of unsown species declined significantly in the ESA and NVC seed-mixture treatments with time compared with the natural regeneration treatment, where it remained relatively constant. Sowing the NVC seed mixture significantly increased total species richness more than any other treatment in any year. Richness declined in the NVC and ESA seed-mixture treatments after the first year (Table 5b) but increased steadily in the natural regeneration treatment (Table 5a). At the end of the experiment there was little difference in the total species richness of the vegetation resulting from natural regeneration or sowing an ESA seed mixture (Table 5a).

effects of nurse crop on species richness

Sowing the NVC seed mixture with a nurse crop resulted in a slight, but non-significant, reduction in the number of sown grass and forbs species (analysis B; Table 6). However, the nurse crop caused a highly significant reduction in the number of unsown grass species in the first year, but this effect did not persist. The nurse crop also significantly reduced total species richness, which persisted for the first 2 years after sowing and which accounted for the significant nurse crop treatment × year interaction. This reflected the combined effect of a reduction in the number of sown forbs and unsown grasses in the nurse crop treatments for the first 2 years after sowing.

Table 6.  Summary of analysis B repeated-measures anova showing the effects of sowing a nurse crop on vegetation variables across all sites. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
Species numberNurse crop (4, 7)No nurse crop (3, 6)anovaF-values
19951996199719981995199619971998Nurse crop d.f. = 1, 45Year d.f. = 3, 135Year × nurse crop d.f. = 3, 135
  1. Treatment 4 = shallow cultivation + species-rich NVC seed mix + nurse crop; treatment 7 = deep cultivation + species-rich NVC seed mix + nurse crop; treatment 3 = shallow cultivation + species-rich NVC seed mix; treatment 6 = deep cultivation + species-rich NVC seed mix.

Sown forbs 7·8 7·6 7·1 8·8 8·6 8·6 7·8 9·0 8·2 NS 14·9***1·1 NS
Sown grasses 7·2 9·0 9·9 8·8 7·6 8·7 9·3 8·9 0·3 NS 76·9***4·4**
Unsown forbs 5·5 2·6 1·7 2·3 5·4 2·7 1·8 2·3 0·2 NS139·2***0·1 NS
Unsown grasses 3·3 2·2 2·8 2·3 4·6 2·2 2·4 1·917·2*** 43·5***9·2***
Total species23·823·821·522·226·326·321·422·2 9·2** 46·2***6·8***

performance of sown and unsown species

Sowing a nurse crop had few significant effects on the mean shoot frequency of individual sown and unsown species (analysis B; Table 7). Significant reductions were recorded for the sown grasses Agrostiscapillaris, Festucarubra/ovina and Phleumpratense, the sown forbs Leucanthemumvulgare, Plantagolanceolata and Ranunculusbulbosus, and the unsown forb Trifoliumrepens. As expected, sowing a nurse crop caused a significant increase in the frequency of Loliummultiflorum, as well as Loliumperenne. The latter effect may have reflected difficulties in separating the two species using vegetative characteristics. The nurse crop effect declined over time for six of the 12 species and for both species of Lolium.

Table 7.  Summary of analysis B repeated-measures anova showing the effects of sowing a nurse crop and cultivation depth on the performance of the most abundant sown and unsown species across all sites. Significant effects and interactions are shown in bold. Trend indicates whether species declined (−), increased (+) or did not change in abundance (no value) over time. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
SpeciesMean frequency %No. sites with the speciesNurse cropDeep cultivationYear
F-valueTrendNurse crop × year interactionF-valueTrendCultivation × year interactionF-valueTrend
Sown grasses
Agrostis capillaris22·05 11·3**  2·0 NS 0·8 NS  2·9*114·1***+
Alopecurus pratensis 6·43  1·9 NS  3·9* 2·5 NS  0·7 NS 34·7***+
Anthoxanthum odoratum 9·03  0·0 NS  7·3*** 4·5*+ 0·9 NS 73·4***+
Briza media 4·44  0·0 NS   0·8 NS19·7 NS  0·4 NS100·5***+
Bromus erectus 1·52  3·1 NS   0·3 NS 1·2 NS  0·1 NS  4·8***+
Cynosurus cristatus29·85  3·4 NS 17·7*** 1·7 NS  2·8* 22·8***+
Festuca ovina/rubra62·35  5·1*  0·4 NS12·4***+ 0·3 NS 75·3***+
Koeleria macrantha10·33  2·8 NS   1·1 NS 8·8**+ 1·1 NS 48·5***+
Lolium multiflorum14·35613·5***+289·2*** 4·0 NS  0·05 NS230·8***
Phleum bertolonii26·02  0·03 NS   1·0 NS 0·0 NS  0·7 NS187·1***
Phleum pratense10·03 15·9***  5·3*** 0·5 NS  0·5 NS  7·3***+
Poa pratensis 5·65  0·0 NS   0·2 NS 3·0 NS  0·2 NS 11·6***+
Trisetum flavescens25·74  0·4 NS   3·5* 0·8 NS  1·7 NS147·1***+
Sown forbs
Achillea millefolium22·85  2·7 NS   4·0* 0·0 NS  0·5 NS136·5***+
Centaurea nigra 1·04  2·3 NS   1·5 NS 1·5 NS  0·3 NS  1·1NS 
Hypochoeris radicata 4·55  1·6 NS   0·3 NS 2·5 NS  0·5 NS 41·4***+
Lathyrus pratensis 0·62  0·1 NS   0·3 NS 0·1 NS  0·7 NS 14·3***+
Leontodon hispidus 0·85  1·2 NS   2·5 NS 0·0 NS  0·1 NS  4·7**+
Leucanthemum vulgare16·14  7·6**  1·2 NS 0·6 NS  0·2 NS108·1***+
Lotus corniculatus11·35  1·9 NS   0·7 NS 0·2 NS  0·4 NS131·8***+
Pimpinella saxifraga 0·14  0·8 NS   0·4 NS 0·3 NS  0·3 NS  2·6NS 
Plantago lanceolata 5·25 17·6***  0·9 NS 4·3*+ 3·7* 38·2***+
Prunella vulgaris 9·05  0·1 NS   0·8 NS 1·9 NS  0·5 NS 85·2***+
Ranunculus acris 0·73  0·1 NS   1·3 NS 3·6 NS  0·3 NS  2·3NS 
Ranunculus bulbosus 1·04  5·1*  0·2 NS 0·3 NS  1·7 NS  5·4**
Rumex acetosa 0·43  3·1 NS   1·3 NS 0·01 NS  1·6 NS  2·3 NS 
Sanguisorba minor 1·24  0·2 NS   1·5 NS 0·5 NS  0·3 NS  2·4 NS 
Scabiosa columbaria 0·12  3·1 NS   1·3 NS 0·04 NS  1·3 NS  1·8 NS 
Succisa pratensis 0·025  0·5 NS   0·3 NS 4·8* 1·0 NS  1·0 NS 
Thymus polytrichus 0·32  0·1 NS   1·2 NS 0·5 NS  2·3 NS  1·8 NS 
Unsown grasses
Agrostis stolonifera 2·95  0·4 NS   0·6 NS 4·5* 1·4 NS 16·5***
Elytrigia repens 3·25  0·3 NS   1·3 NS 5·5* 0·6 NS  2·0 NS 
Holcus lanatus 2·75  0·1 NS  1·3 NS 0·1 NS  0·2 NS 10·2***+
Lolium perenne 5·65 31·9***+ 26·6*** 1·3 NS  0·3 NS131·3***
Poa annua 9·35  1·0 NS   0·2 NS14·6***10·3***153·3***
Poa trivialis 9·05  0·1 NS   0·7 NS21·2*** 3·0* 28·8***
Unsown forbs
Cerastium fontanum 2·75  0·0 NS   0·5 NS 7·4***+ 0·8 NS  3·8*
Geranium dissectum 1·25  0·9 NS   0·4 NS 0·02 NS  0·3 NS  7·3***+
Trifolium repens 4·55  4·3*  3·0* 0·7 NS  0·6 NS 85·2***+
Veronica arvensis 6·25  1·9 NS   1·1 NS 2·3 NS  1·0 NS  5·1**

The cultivation treatments only had a significant effect on 10 species (analysis B; Table 7). Deep cultivation caused a significant increase in the sown grasses Anthoxanthumodouratum, Festucarubra/ovina and Koeleriamacrantha, the sown forbs Plantagolanceolata and Succissa pratensis and the unsown forb Cerastiumfontanum. Four unsown grasses were significantly less frequent following deep cultivation, namely Agrostisstolonifera, Elytrigiarepens, Poaannua and Poatrivialis. There were few significant cultivation × year interactions, suggesting that the cultivation effects did not change over time.

Only the species that were sown in both ESA and NVC seed mixtures at a given site, together with the more common unsown species, were included in the analysis of seed mixture and cultivation effects (analysis A; Table 8). This meant that for some species the effects were only tested at one site. Similarly, some of these effects corresponded to differences in the sowing rates of the species between the NVC and ESA treatments. A total of seven sown grasses was significantly more frequent in the vegetation resulting from sowing the ESA seed mixture compared with that of the NVC seed mixture. Conversely, four sown grasses and one unsown grass (Holcus lanatus) were significantly more abundant in the NVC seed-mixture treatments compared with the ESA treatments. Similarly, 12 of the 23 species considered in analysis A (Table 8) showed a significant increase in abundance over time. Of these, seven of the 11 sown grasses increased and one decreased. Both sown forbs and two of the four unsown forbs increased. Finally, four out of the seven unsown grasses decreased in frequency and one increased.

Table 8.  Summary of analysis A repeated-measures anova showing the effects of seed mixture type and cultivation depth on the performance of the most abundant sown and unsown species across all sites. Significant effects and interactions are shown in bold. Trend indicates whether species declined (−), increased (+) or did not change in abundance (no value) over time. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
SpeciesMean frequency %SitesNVC seed mixtureDeep cultivationYear
EffectTrendSeed mixture × year interactionEffectTrendCultivation × year interactionEffectTrend
Sown grasses
Agrostis capillaris22·05 44·4*** 8·6***    0·03 NS 1·9 NS120·5***+
Alopecurus pratensis 6·41 14·3*** 2·1NS    0·7 NS 0·3 NS  3·9*+
Briza media 4·41 52·0***+22·0***    0·1 NS 0·2 NS 40·6***+
Bromus erectus 1·51 15·2**+ 0·5NS    2·5 NS 0·7 NS  0·3 NS 
Cynosurus cristatus29·85138·4*** 1·7NS   12·4***2·6 NS 27·9*** 
Festuca ovina/rubra62·35  7·1* 3·1*   11·7**+1·1 NS 75·2***+
Koeleria macrantha10·31 38·8***+11·4***    0·1 NS 3·3*  7·7***+
Phleum bertolonii26·01153·5***75·8***    0·01 NS 0·6 NS 99·6***
Phleum pratense10·02 44·4*** 3·2*    0·01 NS 0·9 NS  2·3 NS 
Poa pratensis 5·64  5·6*10·4***    0·04 NS 0·9 NS 50·2***+
Trisetum flavescens25·71 23·4***+ 5·7**    1·3 NS 1·8 NS 14·9***+
Sown forbs
Leucanthemum vulgare16·21  1·2 NS  0·2 NS    7·4*+7·9*** 15·9***+
Lotus corniculatus11·31  2·1 NS  0·7 NS< 0·01 NS 0·2 NS 14·1***+
Unsown grasses
Agrostis stolonifera 2·95  2·7 NS  2·2 NS   20·1***5·5** 28·0***
Elytrigia repens 3·25  2·3 NS  0·1 NS    0·6 NS 0·5 NS  0·2 NS 
Holcus lanatus 2·75  7·1*+ 0·7 NS    0·1 NS 0·1 NS 10·4***+
Lolium perenne 5·65  0·02 NS  0·6 NS    0·8 NS 0·4 NS 49·9***
Poa annua 9·35  0·5 NS  1·1 NS   19·5***9·9***185·9***
Poa trivialis 9·05  0·1 NS  2·4 NS    9·0**1·5 NS 51·0***
Unsown forbs
Cerastium fontanum 2·75  2·2 NS  1·1 NS   16·1***+3·1*  2·4 NS 
Geranium dissectum 1·25  0·01 NS  2·1 NS    0·1 NS 0·7 NS  9·5***+
Trifolium repens 4·55 < 0·01 NS  0·03 NS    1·7 NS 1·6 NS 81·7***+
Veronica arvensis 6·25  0·1 NS  0·3 NS    7·4**+1·8 NS  6·3***

There was a high degree of correspondence between the species showing significant responses to cultivation (analysis B, Table 7; analysis A, Table 8). In analysis A deep cultivation resulted in a significant increase in the frequency of the sown grass Festuca ovina/rubra and the sown forb Leucanthemumvulgare. It also produced a significant increase in the frequency of the unsown forbs Cerastiumfontanum and Veronica arvensis. Deep cultivation caused a significant reduction in the frequency of the sown grass Cynosurus cristatus and the unsown grasses Agrostisstolonifera, Poaannua and Poatrivialis.

The performance of the most common unsown species was analysed by comparing frequencies in the natural regeneration, ESA and NVC treatments (analysis C). All, except Trifoliumrepens, had a greater abundance in the natural regeneration treatment compared with the ESA or NVC seed mixtures (Table 9). This analysis showed little difference in the performance of the unsown species between ESA and NVC treatments. With the exception of Elytrigiarepens, all of the common unsown species increased over time. This appeared to contradict the results of the previous analyses (Tables 7 and 8), but reflected the increase of these species in the natural regeneration treatment and their decline or stable status in the ESA and NVC treatments.

Table 9.  Summary of analysis C repeated-measures anova showing the effects of seed addition on the performance of the most abundant sown and unsown species across all sites. Significant effects and interactions are shown in bold. Trend indicates whether species declined (−), increased (+) or did not change in abundance (no value) over time. NS = not significant; *P < 0·05; **P < 0·01; ***P < 0·001
SpeciesMean frequency %SitesSeed additionYear
EffectTrendEffectTrend
Unsown grasses
Agrostis stolonifera2·9511·1*** 25·8***+
Elytrigia repens3·2520·4***  2·6 NS 
Holcus lanatus2·7547·9*** 14·9***+
Lolium perenne5·6514·8***   7·3***+
Poa annua9·3523·6***134·6***
Poa trivialis9·0582·9***  3·0*+
Unsown forbs
Cerastium fontanum2·7511·2***  9·5***+
Geranium dissectum1·25 7·2**  8·7***+
Trifolium repens4·55 0·2 NS  55·8***+
Veronica arvensis6·2512·3***  4·1***+

comparison to nvc target communities

After 1 year the vegetation regenerating naturally from the cereal stubble (treatment 1) most closely resembled the weedy ephemeral communities described by the NVC (Rodwell 2000; Table 10). Similarity to these communities was relatively high for most sites (mean similarity 61 ± 4%). After 4 years this treatment had a weak similarity to the MG5 grassland community (Rodwell 1992) on the circum-neutral soils of the Upper Thames Tributaries and Norfolk Broads. On the lighter soils of the South Wessex Downs and Suffolk River Valleys this treatment retained its affinities with the weedy communities throughout. Natural regeneration showed weak affinities to the Festuca ovina–Avenula pubescens grassland: Holcus lanatus–Trifolium repens subcommunity (CG2c) on the South Downs. In all but one case the treatment sown with the ESA seed mixture (5) showed a low similarity to the same weedy communities (52 ± 3%) as natural regeneration after 1 year (Table 10). After 4 years the resultant vegetation most closely resembled the MG5 grassland at four of the five sites, but the degree of similarity remained low (44 ± 3%). After 1 year the treatment sown with the NVC seed mixture (6) had a low similarity to the MG5 grassland (51 ± 4%) at all sites, with the exception of the South Downs which most closely resembled the CG2c subcommunity (48%). However, after 4 years this treatment achieved a high degree of similarity to the target on neutral soils (75% and 66% at the Upper Thames and Norfolk Broads, respectively). The sites on chalk most closely resembled the CG2c subcommunity (64 ± 3%). No treatment achieved a good correspondence to the specified target (CG3b). Similarly, this treatment on the acidic sandy soil most closely resembled the target for neutral soils (MG5a) but had poor correspondence the target vegetation (MG5c)

Table 10.  Similarity of the restored grassland communities to the National Vegetation Classification (NVC) target communities based on analysis using Tablefit (Hill 1996)
SiteTreatmentYearNVC Target% fit to targetSee legend for details of (sub)community
% best fit(Sub)community
  1. CG2c, Festuca ovina–Avenula pubescens grassland: Holcus lanatus–Trifolium repens subcommunity.

  2. MG5, Cynosurus cristatus–Centaurea nigra grassland;

  3. MG5b, Cynosurus cristatus–Centaurea nigra grassland: Galium verum subcommunity.

  4. MG6c, Lolium perenne–Cynosurus cristatus pasture: Trisetum flavescens subcommunity.

  5. OV 3, Papaver rhoeas–Viola arvensis weed community.

  6. OV10c, Poa annua–Senecio vulgaris weed community: Agrostis stolonifera–Rumex crispus subcommunity.

  7. OV12b, Poa trivialis–Myosotis arvensis weed community: Dicranella staphylina–Bryum sp. subcommunity.

  8. OV19, Poa annua–Tripleurospermum inodorum weed community.

  9. OV23b, Lolium perenne–Dactylis glomerata weed community: Crepis vesicaria–Rumex obtusifolius subcommunity.

South Downs1. Control1995CG3b Bromus erectus grassland: Centaurea nigra subcommunity< 557OV10c
  1998      3248CG2c
 5. Deep cultivate + ESA mix1995       654OV10c
  1998      2541MG5b
 6. Deep cultivate + NVC mix1995      4248CG2c
  1998      5767CG2c
South Wessex Downs1. Control1995CG3b Bromus erectus grassland: Centaurea nigra subcommunity< 568OV10c
  1998      1144OV10c
 5. Deep cultivate + ESA mix1995      752OV12b
  1998      1744MG5a
 6. Deep cultivate + NVC mix1995      3641MG5b
  1998      5061CG2c
Upper Thames Tributaries1. Control1995MG5a Cynosurus cristatus–Centaurea nigra grassland: Lathyrus pratensis subcommunity< 564OV19
  1998  4040MG5a
 5. Deep cultivate + ESA mix1995   4242MG5a
  1998      3352MG6c
 6. Deep cultivate + NVC mix1995      5959MG5a
  1998  7575MG5a
Norfolk Broads1. Control1995MG5a Cynosurus cristatus–Centaurea nigra grassland: Lathyrus pratensis subcommunity< 566OV12b
  1998      5256MG5
 5. Deep cultivate + ESA mix1995      2252OV3
  1998      4949MG5a
 6. Deep cultivate + NVC mix1995      5950MG5a
  1998  6666MG5a
Suffolk River Valleys1. Control1995MG5c Cynosurus cristatus–Centaurea nigra grassland: Danthonia decumbens subcommunity< 548OV10c
  1998      2255OV23d
 5. Deep cultivate + ESA mix1995 < 558OC10c
  1998      2036MG5
 6. Deep cultivate + NVC mix1995  3253MG5a
  1998  3552MG5a

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

is soil fertility a constraint on community assembly?

The soil of the five arable reversion sites contained moderate to large residual amounts of P and K, and in most cases low %N. This is typical of soils with a history of intensive arable agriculture supported by inorganic fertilizers (Gough & Marrs 1990). There is good evidence to suggest that high residual soil fertility places severe constraints on the restoration and long-term maintenance of plant species diversity (Marrs 1993; Janssens et al. 1998). Elevated nutrient levels are often associated with high productivity, which enables a small number of more competitive species to dominate the vegetation and exclude other species. However, this study suggests that the establishment of diverse grassland communities may not be constrained by high residual soil fertility on ex-arable land where N is initially in short supply. Monitoring is continuing in order to investigate the effects of elevated nutrient concentrations on the rate and direction of succession in the longer term.

Nevertheless, this study has shown that deep cultivation of ex-arable soil can cause a rapid and significant reduction in available P and K concentrations in the 0–5-cm depth fraction across a range of soil types. It is likely that this reflects the burial and dilution of fertile upper layers of the soil with less fertile soil from below (Marrs 1993). Analysis of individual sites showed that deep cultivation was most effective on shallow chalk soils, where up to 50% reductions in both nutrients were achieved (R. Pywell, unpublished data). However, it is important to note that the resultant nutrient concentrations achieved by this treatment (P 25·7 ± 0·3 mg l−1; K 179·7 ± 3·5 mg l−1) were still considerably higher than those typically found in the unimproved grassland target communities (P 11·1 ± 1·5 mg/ l−1; K 49·2 ± 3·2 mg l−1; R. Pywell, unpublished data). Furthermore, after 4 years P concentrations fell in the shallow cultivation treatments to those achieved by deep cultivation in the first year. Also, K concentrations recovered in the deep ploughed treatments so that after 4 years there were no significant differences between cultivation treatments. This might be explained by weathering and release of nutrients from the inorganic pool, as well as the mixing of the soil by macrofauna and the redistribution of nutrients by plant roots.

It is perhaps surprising that %N declined by a small amount at both depths over the period of the experiment. Several studies have shown N and associated mineralization rates increase during old-field succession (Marrs 1993). However, little is known about soil processes during the early stages of old-field succession. It is possible that the rate of assimilation of mineralized N into the rapidly growing perennial vegetation and multiplying soil microbial populations initially exceeds the amount returned by processes such as leaf fall and root excretion.

Deep cultivation caused a significant reduction in the number and abundance of unsown grass species. Many grasses have a light requirement for germination and therefore need to remain at or near the soil surface to establish (Froud-Williams, Drennan & Chancellor 1983). The beneficial effect of deep cultivation on sown forbs may therefore relate to decreased competition from unsown grasses, the short-term reduction in soil fertility, and the improved physical structure of the seed bed. Monitoring is continuing to see whether species richness will be maintained despite these comparatively high nutrient levels.

is community assembly seed-limited?

The vegetation communities resulting from natural regeneration and those sown with typical ESA seed mixtures had significantly fewer desirable species than the treatments sown with the NVC seed mixture throughout the experiment. This confirms that the creation of diverse vegetation on ex-arable land is highly seed limited and there is little potential for the colonization of later successional grassland species (Hutchings & Booth 1996; Mortimer, Hollier & Brown 1998). Sowing a seed mixture (ESA or NVC) led to a decrease in the number and individual abundance of unsown weedy species, but there was no evidence to suggest that sowing a species-rich NVC seed mixture was any more effective in suppressing weedy species than the more species-poor ESA seed mixture. This conflicts with the findings of another multi-site study where high- and low-diversity seed mixtures were sown on five ex-arable sites across north-west Europe (van der Putten et al. 2000). Plots sown with 15 species contained fewer ‘natural colonizer’ species than those sown with four species. In our experiment the species-poor and species-rich seed treatments were sown with an average of eight and 37 species, of which an average of 8·6 and 17·9 species established, respectively. The relative similarity in the number of sown species establishing between our species-poor and species-rich treatments may explain the lack of support for the results of the latter experiment. However, these differences were sufficient to cause large differences in productivity between species-rich and species-poor treatments after the first year (Bullock et al. 2001).

The plots sown with the NVC seed mixtures provided a source of desirable species to colonize the other plots. This allowed us to determine whether sowing a grass-rich, but forb-poor, mixture, as the ESA mix was, makes the vegetation less susceptible to colonization by desirable forb species compared with the vegetation produced by natural regeneration. This may be a function of species richness (Schläpfer & Schmid 1999) but also the fact that the ESA mix contains later-successional species than the ruderal-dominated natural colonizers. Thus the former may be less susceptible to colonization by other late-successional species (Hansson & Fogelfors 1998). However, there was no evidence that vegetation produced by sowing the low-diversity ESA seed mixture was any less susceptible to colonization by desirable later successional grassland forb species than the natural regenerating vegetation.

can bio-remediation of site conditions with a nurse crop facilitate species establishment?

Sowing of a nurse crop of Loliummultiflorum did not have any long-term significant effects on the number of sown species recorded. However, the nurse crop did cause a significant reduction in both total species richness and the number of unsown grass species present. The nurse crop also caused significant reductions in the mean frequency of six sown grass and forbs, and one unsown forb. These effects probably reflect increased competition for resources in the establishment phase caused by the rapidly growing nurse crop. Therefore, contrary to the aim of sowing a nurse crop, this treatment was of little benefit to the restoration process, and could potentially have negative effects. Similar negative effects were reported from a wet grassland restoration trial on ex-arable land (Manchester et al. 1997).

what traits are associated with good performance of sown species in habitat restoration?

The experiment demonstrated that sown grass species were both more successful and reliable in their establishment and persistence over a wide range of ex-arable sites compared with the sown forbs. All of the sown grasses established at all sites, whereas the establishment of the sown forbs ranged from between 54% and 84%. Sown forbs that established successfully (e.g. Achillea millefolium and Leucanthemum vulgare) were competitive perennials of wide ecological amplitude, with a tall growth habit and extensive root system (Ellenberg 1988; Grime, Hodgson & Hunt 1988; Hill et al. 1999). These species typically have comparatively large seeds with high seed viability and germination rates (R. Pywell, unpublished data). Sown species that increased in frequency over time and were able to colonize adjacent treatments where they were not sown were typically species that set seed before the July hay cut (e.g. Hypochoeris radicata,Lotus corniculatus and L. vulgare). Forbs that failed to establish or performed poorly (e.g. Thymus polytrichus), were stress-tolerators (Grime, Hodgson & Hunt 1988) with more extreme ecological indicator values (Ellenberg 1988; Hill et al. 1999). These species usually have smaller seeds with lower viability and rates of germination. Seed dormancy (e.g. hard-coat dormancy of Helianthemum nummularium) and high rates of seedling mortality (e.g. Viola hirta) are likely to have been common causes of the failure to establish. The decline in soil fertility and the effects of perturbations, such as drought and water-logging, suggest that the suitability of the sites for these species may increase in the longer term. Further studies are required into the establishment requirements of such species and the optimum stage for their introduction.

was the restoration of diverse grassland communities successful?

A simple measure of restoration success was applied to this experiment, namely a comparison of the composition of the restored vegetation to pre-defined desirable target communities described by the NVC. However, these communities were originally described from sample stands of high nature conservation value, so they represent an ambitious target for restoration on ex-arable land after 4 years.

The natural regeneration and ESA treatments corresponded poorly to the target vegetation types. This reflects the importance of the constraint of seed limitation on community development. The treatments sown with the NVC seed mixtures were reasonably similar in composition to the diverse target communities on neutral soils. Correspondence with the defined target subcommunities on the acidic and calcareous soils was much lower, reflecting the poor performance of the indicator species for these target vegetation types.

In order to resemble truly the target ecosystem, a restored system must be similar in terms of the following attributes, which can be considered as measures of restoration success (Ewel 1987; Pywell & Putwain 1996): (i) community structure and composition of all other biota, including non-vascular plants, invertebrates, soil microbiota; (ii) ecosystem function, including measures of nutrient utilization efficiency, retention and cycling, productivity of different trophic levels, community respiration, and water use; and (iii) the maintenance of communities and functions over time (stability, resilience and resistance). Future research and long-term monitoring are needed to address these measurements for a range of restored ecosystems (Mitchell et al. 2000).

In conclusion, this study has demonstrated that the most effective means of restoring diverse plant communities to ex-arable land was deep cultivation followed by the application of a diverse seed mixture comprising ecologically appropriate species. The most important constraint on achieving the assembly of these plant communities was seed limitation: high soil fertility appeared to be a less important constraint, the amelioration of which can have small beneficial effects on the assembly of diverse communities. The establishment of desirable species was not significantly enhanced by sowing a nurse crop for amelioration of abiotic and biotic conditions. Overall these treatment effects were found to be reproducible over a range of sites. The NVC proved to be a useful means of tracking the course of vegetation succession towards a pre-defined target. The success of restoration, as measured by these means, was greatest on neutral soils. Further research is required to enhance the establishment of species with a preference for calcareous and acidic soils. Finally, monitoring is continuing to ensure that these conclusion are upheld in the longer term.

Supplementary material

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

The following supplementary material is available from http://www.blackwell-science.com/products/journals/suppmat/JPE/JPE718/JPE718sm.htm, or from the authors.

Appendix 1. Percentage composition of the species-rich (NVC) and ESA seed mixtures sown at the arable reversion sites.

Appendix 2. Summary of analysis B repeated-measures anova showing the effects of cultivation on soil fertility at the two soil depths.

Appendix 3. Summary of analysis B repeated-measures anova showing the effects of cultivation on vegetation variables across all sites.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

This work was funded by a commission (BD1404) from the Department of Environment, Food and Rural Affairs. The authors are grateful to R. Lambourne, J. Blackburn (Norfolk Wildlife Trust), P. Bowden-Smith, M. Houghton-Brown and T. Tupper for hosting the trials. We are grateful to D. Myhill and E. Allchin for technical support. Finally, we would like to thank the two anonymous referees for their constructive comments.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information
  • Allen, S.E., Grimshaw, H.M., Parkinson, J.A. & Quarmby, C. (1974) Chemical Analysis of Ecological Materials. Blackwell Science, Oxford.
  • Anonymous (1998) UK Biodiversity. Tranche 2 Action Plans. Vol. II. Terrestrial and Freshwater Habitats. HMSO, London, UK.
  • Bakker, J.P. & Berendse, F. (1999) Constraints in the restoration of ecological diversity in grassland and heathland communities. Trends in Ecology and Evolution, 39, 6368.
  • Bakker, J.P., Grootjans, A.P., Hermy, M. & Poschlod, P. (2000) How to define targets for ecological restoration? Introduction. Applied Vegetation Science, 3, 36.
  • Bekker, R.M., Verweij, G.L., Smith, R.E.N., Reine, R., Bakker, J.P. & Schneider, S. (1997) Soil seed banks in European grasslands: does land use affect regeneration perspectives? Journal of Applied Ecology, 34, 12931310.
  • Braun-Blanquet, J. & Tüxen, R. (1952) Irische Pflanzengesellschaften. Veröffentlichungen Des Geobotanischen Institutes Rübel in Zürich, 25, 224415.
  • Brooker, R.W. & Callaghan, T.V. (1998) The balance between positive and negative plant interactions and its relationship to environmental gradients: a review. Oikos, 81, 196207.
  • Bullock, J.M. (1996) Plant competition and community dynamics. The Ecology and Management of Grazing Systems (eds J.Hodgson & A.W.Illius), pp. 69100. CAB International, Wallingford, UK.
  • Bullock, J.M., Pywell, R.F., Burke, M.J.W. & Walker, K.J. (2001) Restoration of biodiversity enhances agricultural production. Ecology Letters, 4, 185189.
  • Callaway, R.M. (1995) Positive interactions among plants. Botanical Review, 61, 306349.
  • Carey, P.D., Finch, C., Hunt, J., Morris, C., Parkin, A., Priscott, A. & Short, C. (2001) Monitoring and Evaluation of the Countryside Stewardship Scheme: Overview Report. Final Report. Ministry of Agriculture, Fisheries and Food, London, UK.
  • Choi, Y.D. & Wali, M.K. (1995) The role of Panicum virginatum (switch grass) in the revegetation of iron mine tailings in northern New York. Restoration Ecology, 3, 123121.
  • Van Dijk, G. (1991) The status of semi-natural grasslands in Europe. The Conservation of Lowland Dry Grassland Birds in Europe (eds P.D.Goriup, L.A.Batten & J.A.Norton), pp. 1536. Joint Nature Conservation Committee, Peterborough, UK.
  • Duffey, E., Morris, M.G., Sheail, J., Ward, L.K., Wells, D.A. & Wells, T.C.E. (1974) Grassland Ecology and Wildlife Management. Chapman & Hall, London, UK.
  • Crofts, A. & Jefferson, R.G. (1999) The Lowland Grassland Management Handbook, 2nd edn. English Nature and the Wildlife Trusts, Peterborough, UK.
  • Ellenberg, H. (1988) Vegetation Ecology of Central Europe, 4th edn. Cambridge University Press, Cambridge, UK.
  • Ewel, J.J. (1987) Restoration is the ultimate test of ecological theory. Restoration Ecology: A Synthetic Approach to Ecological Research (eds W.R.JordanIII, M.E.Gilpin & J.D.Aber), pp. 3133. Cambridge University Press, Cambridge, UK.
  • Froud-Williams, R.J., Drennan, D.S.H. & Chancellor, R.J. (1983) Influence of cultivation regime upon buried weed seeds in arable cropping systems. Journal of Applied Ecology, 20, 187197.
  • Fuller, R.M. (1986) The changing conservation interest of lowland grasslands in England and Wales: a review of grassland surveys 1930–1984. Biological Conservation, 40, 281300.
  • Gough, M.W. & Marrs, R.H. (1990) A comparison of soil fertility between semi-natural and agricultural plant communities: implications for the creation of floristically rich grassland on abandoned agricultural land. Biological Conservation, 51, 8396.
  • Grime, J.P., Hodgson, J.G. & Hunt, R. (1988) Comparative Plant Ecology: A Functional Approach to Common British Species. Unwin Hyman Ltd, London, UK.
  • Hansson, M. & Fogelfors, H. (1998) Management of permanent set-aside on arable land in Sweden. Journal of Applied Ecology, 35, 758771.
  • Hill, M.O. (1996) TABLEFIT, Version 1.0, For Identification of Vegetation Types. Institute of Terrestrial Ecology, Huntingdon, UK.
  • Hill, M.O., Mountford, J.O., Roy, D.B. & Bunce, R.G.H. (1999) Ellenberg’s Indicator Values for British Plants: ECOFACT. Vol. 2. Technical Annex. HMSO, London, UK.
  • Hopkins, A., Pywell, R.F., Peel, S., Johnson, R.H. & Bowling, P.J. (1999) Enhancement of botanical diversity of permanent grassland and impact on hay production in Environmentally Sensitive Areas in the UK. Grass and Forage Science, 54, 163173.
  • Hutchings, J. & Booth, K.D. (1996) Studies on the feasibility of recreating chalk grassland vegetation on ex-arable land. I. The potential roles of the seed bank and the seed rain. Journal of Applied Ecology, 33, 11711181.
  • Janssens, F., Peeters, A., Tallowin, J.R.B., Bakker, J.P., Bekker, R.M., Fillat, F. & Oomes, M.J.M. (1998) Relationship between soil chemical factors and grassland diversity. Plant and Soil, 202, 6978.
  • Manchester, S.J., Treweek, J.T., Mountford, O.M. & Pywell, R.F. (1997) The success of nurse crops in the establishment of species-rich lowland wet grassland. Grassland Management in the ESAs, Occasional Symposium of the British Grasslands Society (ed. R.D.Sheldrick), pp. 260261. British Grassland Society, Reading, UK.
  • Marrs, R.H. (1993) Soil fertility and nature conservation in Europe: theoretical considerations and practical management solutions. Advances in Ecological Research, 24, 242300.
  • Maxwell, S.E. & Delaney, H.D. (1990) Designing Experiments and Analysing Data. Wadsworth Publishing Company, Belmont, CA.
  • Mitchell, R.J., Auld, M.H.D., Le Duc, M.G. & Marrs, R.H. (2000) Ecosystem stability and resilience: a review of their relevance for the conservation management of lowland heaths. Perspectives in Plant Ecology, Evolution and Systematics, 3, 142160.
  • Mortimer, S.R., Hollier, J.A. & Brown, V.K. (1998) Interactions between plant and insect diversity in the restoration of lowland calcareous grasslands in southern Britain. Applied Vegetation Science, 1, 101114.
  • Ovenden, G.N., Swash, A.R.H. & Smallshire, D. (1998) Agri-environment schemes and their contribution to the conservation of biodiversity in England. Journal of Applied Ecology, 35, 955960.
  • Poschlod, P., Kiefer, S., Traenkle, U., Fischer, S. & Bonn, S. (1998) Species richness in calcareous grasslands is affected by disperability in space and time. Applied Vegetation Science, 1, 7590.
  • Van Der Putten, W.H., Mortimer, S.R., Hedlund, K., Van Dijk, C., Brown, V.K., Lepš, J., Rodriguez-Barrueco, C., Roy, J., Diaz Len, T.A., Gormsen, D., Korthals, G.W., Lavorel, S., Santa Regina, I. & Smilauer, P. (2000) Plant species diversity as a driver of early succession in abandoned fields: a multi-site approach. Oecologia, 124, 9199.
  • Pywell, R.F. & Putwain, P.D. (1996) Restoration for conservation gain. Conservation Biology (ed. I.F.Spellerberg), pp. 203221. Longman, London, UK.
  • Pywell, R.F., Putwain, P.D. & Webb, N.R. (1997) The decline of heathland seed populations following the conversion to agriculture. Journal of Applied Ecology, 34, 949960.
  • Pywell, R.F., Webb, N.R. & Putwain, P.D. (1994) Soil fertility and its implications for the restoration of heathland on farmland in southern Britain. Biological Conservation, 70, 169181.
  • Ray, G.J. & Brown, B.J. (1995) Restoring Caribbean dry forests – evaluation of tree propagation techniques. Restoration Ecology, 3, 8694.
  • Rodwell, J.S. (1991) The National Vegetation Classification. British Wildlife, 2, 266268.
  • Rodwell, J.S. (1992) British Plant Communities. Vol. 3. Grasslands and Montane Communities. Cambridge University Press, Cambridge, UK.
  • Rodwell, J.S. (2000) British Plant Communities. Vol. 5. Maritime Communities and Vegetation of Open Habitats. Cambridge University Press, Cambridge, UK.
  • Rodwell, J.S. (19912000) British Plant Communities. Volumes 1–5. Cambridge University Press, Cambridge, UK.
  • Rodwell, J.S. & Patterson, G. (1994) Creating New Native Woodlands. Forestry Commission Bulletin 112. HMSO, London, UK.
  • SAS (1990) SAS/STAT Users’ Manual. SAS Institute, Carey, NC.
  • Schläpfer, F. & Schmid, B. (1999) Ecosystem effects of biodiversity: a classification of hypotheses and exploration of empirical results. Ecological Applications, 9, 893912.
  • Stace, C. (1997) New Flora of the British Isles, 2nd edn. Cambridge University Press, Cambridge, UK.
  • Walker, K.J. & Pywell, R.F. (2000) Grassland communities on Salisbury Plain Training Area (SPTA): results of the ITE ecological survey. Wiltshire Botany, 3, 1527.
  • Wells, T.C.E., Cox, R. & Frost, A. (1989) The Establishment and Management of Wildflower Meadows. Focus on Nature Conservation No. 21. Nature Conservancy Council, Peterborough, UK.
  • Willems, J.H. (1978) Observations on north-west European limestone grassland communities: phytosociological and ecological notes on chalk grasslands of southern England. Vegetatio, 37, 141150.
  • Wilson, J.B., Steel, J.B., Dodd, M.E., Anderson, B.J., Ullmann, I. & Bannister, P. (2000) A test of community reassembly using the exotic communities of New Zealand roadsides in comparison to British roadsides. Journal of Ecology, 88, 757764.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Supplementary material
  8. Acknowledgements
  9. References
  10. Supporting Information

Appendix 1. Percentage composition of the species-rich (NVC) and ESA seed mixtures sown at the arable reversion sites. Appendix 2. Summary of analysis B repeated-measures ANOVA showing the effects of cultivation on soil fertility at the two soil depths. Appendix 3. Summary of analysis B repeated-measures ANOVA showing the effects of cultivation on vegetation variables across all sites.

FilenameFormatSizeDescription
JPE_718_sm_app.doc144KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.