Can arable field margins be managed to enhance their biodiversity, conservation and functional value for soil macrofauna?

Authors

  • J. Smith,

    1. Centre for Agri-Environmental Research, The University of Reading, Earley Gate, PO Box 237, Reading RG6 6AR, UK; and
    2. Soil Biodiversity Programme, Department of Entomology, The Natural History Museum, Cromwell Road, London SW7 5BD, UK
    Search for more papers by this author
  • S. G. Potts,

    1. Centre for Agri-Environmental Research, The University of Reading, Earley Gate, PO Box 237, Reading RG6 6AR, UK; and
    Search for more papers by this author
  • B. A. Woodcock,

    1. Centre for Agri-Environmental Research, The University of Reading, Earley Gate, PO Box 237, Reading RG6 6AR, UK; and
    Search for more papers by this author
  • P. Eggleton

    1. Soil Biodiversity Programme, Department of Entomology, The Natural History Museum, Cromwell Road, London SW7 5BD, UK
    Search for more papers by this author

Correspondence author. E-mail: joans2@nhm.ac.uk

Summary

  • 1The establishment of grassy strips at the margins of arable fields is an agri-environment scheme that aims to provide resources for native flora and fauna and thus increase farmland biodiversity. These margins can be managed to target certain groups, such as farmland birds and pollinators, but the impact of such management on the soil fauna has been poorly studied. This study assessed the effect of seed mix and management on the biodiversity, conservation and functional value of field margins for soil macrofauna.
  • 2Experimental margin plots were established in 2001 in a winter wheat field in Cambridgeshire, UK, using a factorial design of three seed mixes and three management practices [spring cut, herbicide application and soil disturbance (scarification)]. In spring and autumn 2005, soil cores taken from the margin plots and the crop were hand-sorted for soil macrofauna. The Lumbricidae, Isopoda, Chilopoda, Diplopoda, Carabidae and Staphylinidae were identified to species and classified according to feeding type.
  • 3Diversity in the field margins was generally higher than in the crop, with the Lumbricidae, Isopoda and Coleoptera having significantly more species and/or higher abundances in the margins. Within the margins, management had a significant effect on the soil macrofauna, with scarified plots containing lower abundances and fewer species of Isopods. The species composition of the scarified plots was similar to that of the crop.
  • 4Scarification also reduced soil- and litter-feeder abundances and predator species densities, although populations appeared to recover by the autumn, probably as a result of dispersal from neighbouring plots and boundary features. The implications of the responses of these feeding groups for ecosystem services are discussed.
  • 5Synthesis and applications. This study shows that the management of agri-environment schemes can significantly influence their value for soil macrofauna. In order to encourage the litter-dwelling invertebrates that tend to be missing from arable systems, agri-environment schemes should aim to minimize soil cultivation and develop a substantial surface litter layer. However, this may conflict with other aims of these schemes, such as enhancing floristic and pollinator diversity.

Introduction

The decline of farmland wildlife in Europe during the last 40 years has been attributed to the intensification of agricultural practices (Robinson & Sutherland 2002). Spatial, temporal and technological intensification have resulted in complex natural communities being replaced by less diverse systems with frequent disturbances (New 2005). This loss of biodiversity threatens the sustainability of the entire system, with a resulting reduction in ecosystem services such as pest control, nutrient cycling and maintenance of soil structure. The degradation of these services are often compensated for by costly external mechanical and chemical inputs, including pesticides, fertilizers and soil tillage, which can have further negative impacts on biota (Giller et al. 1997).

Since 1985, agri-environment schemes in the European Union (EU) have aimed to reverse the decline of farmland biodiversity by providing subsidies for farmers to adopt more environmentally friendly farming practices (Ovenden, Swash & Smallshire 1998). One such practice in western Europe is establishing grassy margins in arable fields. In general these margins are linear, uncultivated, semi-natural habitats, providing resources for native flora and fauna and a buffer against intensive crop production (Marshall & Moonen 2002).

Three key ecological functions of field margins are to increase species density in an agro-ecosystem (biodiversity value), provide habitats for rare or endangered species (conservation value) and enhance ecosystem services such as pest control and decomposition (functional value). Several studies have investigated the manipulation of the initial establishment of field margins and their subsequent management to optimize these values (Haughton et al. 1999; Thomas & Marshall 1999; Woodcock et al. 2005). Creating margins with a high floral diversity tends to enhance invertebrate communities, thereby rapidly increasing farmland biodiversity (Thomas & Marshall 1999; Meek et al. 2002). Management practices (e.g. grass cutting and herbicide application) may also influence invertebrate diversity by manipulating the vegetation and reducing highly competitive grass species.

Field margin management practices have generally been assessed with reference to the responses of above-ground biota, such as farmland birds, rare arable weeds and epigeal invertebrate taxa (Kirkham et al. 1999; Thomas & Marshall 1999; Meek et al. 2002). However, few studies have considered the impact on soil invertebrate communities (Brown 1999; Lagerlöf, Goffre & Vincent 2002). As soil communities are probably not regulated exactly as above-ground biodiversity (Bardgett, Yeates & Anderson 2005), below-ground taxa may not respond to habitat management like above-ground communities.

This study assessed the impact of field margin management on the soil macrofauna (> 2 mm diameter; Bardgett 2005). While less numerous than the smaller soil biota, macrofauna are a highly diverse and integral component of many ecosystem services, including nutrient cycling and biological control of pests. As soil macrofauna vary in mobility, habitat requirements and ecological function, they are likely to respond differentially to variations in habitat management, making them ideal candidates for assessing environmental transformations and impacts. Species of some groups (e.g. Staphylinidae) differ widely in their feeding ecology and this probably affects their response to the environment. Therefore using trophic groupings allows a synthetic functional approach to assessing the potential effects of treatments on soil assemblages and may provide information on ecosystem functioning (Clough, Kruess & Tscharntke 2007).

Litter-feeding invertebrates (e.g. woodlice, millipedes and epigeic and anecic earthworms) respond to the quantity and quality of litter inputs (Hopkin & Read 1992; Curry & Schmidt 2007) and may be affected by changes caused by establishing margins of varying vegetation diversity and management. These litter-feeders facilitate decomposition activities of bacteria and fungi by shredding residues, dispersing microbial propagules and mixing organic matter into the soil (Hopkin & Read 1992); they help regulate the early stages of decomposition and nutrient cycling.

In contrast, the soil-feeding endogeic earthworms are able to live in soils with as little as 1% organic matter (Lavelle 1997) and mediate the later stages of decomposition, modifying the soil structure for other soil organisms by creating a diverse array of soil structures (Lavelle 1997; Jouquet et al. 2006). Some species can survive the frequent disturbances of crop production (Ekschmitt & Griffiths 1998) so probably respond differently to seed mix and management treatments than the litter-feeders. Polyphagous predators (e.g. carabids and centipedes) are potentially important natural pest-control agents (Kromp 1999). Important determinants of predator assemblages include the availability of prey species and sites for shelter and overwintering (Dennis, Thomas & Sotherton 1994).

The objectives of this study were to assess the effect of field margin seed mix and management on the soil macrofauna, asking the following questions. (i) Do arable field margins have higher species abundances and densities than the crop? (ii) Which margin treatments maximize diversity? (iii) Are margins significant ‘refuges’ for species of conservation importance and, if so, which treatment has maximum conservation value? (iv) Which margin treatments have the potential for enhancing ecosystem services, such as pest control and organic matter decomposition, through their effects on soil macrofauna assemblages?

Materials and methods

study site

Experimental margin treatments were established in a 20·5-ha field on calcareous clay/clay loam, at ADAS Boxworth, Cambridgeshire, England (52:15:10N, 0:01:54W, 50 m a.s.l., annual rainfall 553 mm) as part of the Sustainable Arable Farming For an Improved Environment project (SAFFIE; http://www.saffie.info, accessed 18 April 2007).

treatments

Non-cropped field margins were established in 2001. Plots measuring 25 × 5 m, with a 5-m buffer between adjacent plots, were arranged in four replicated blocks of nine and sown with one of three seed mixes (Table 1). In 2002, three different management regimes were introduced (Table 1) that were combined with the seed mixes to produce a 3 × 3 factorial design, with a total of 36 plots (nine combinations × four replicates). The main field was cropped with winter wheat from 2001 to 2006, with a break crop of spring beans in 2003–04. The field boundary was mostly hedgerow (Crataegus monogyna Jacq.), except for a small patch of woodland adjacent to plots 1–3.

Table 1.  Field margin treatments at ADAS Boxworth (http://www.saffie.info)
TreatmentDescriptionRationale
  • *

    Tussock-forming grasses, e.g. Dactylis glomerata L.;

  • fine-stemmed grasses, e.g. Festuca rubra L. The remainder of the grass mix comprises a mixture of other grass species (see Table S1 in the supplementary material for grass and forb species included in the seed mixes).

Seed mixesSown 2001 
Countryside stewardship (CS)Grasses = 100%Represents the ‘cheap mix’ available to farmers under the CS
Tussock* 10% 
Fine 40% 
Sowing rate 20·0 kg ha−1 
Tussock grass and forbs (TG)Forbs 20%To increase ground-dwelling invertebrates by providing tussocky refugia
Grasses 80% 
Tussock 24% 
Fine 20% 
Sowing rate 35·1 kg ha−1 
Fine grass and forbs (FG)Forbs 20%To enhance insect diversity by increasing floral diversity
Grasses 80% 
Tussock 0% 
Fine 40% 
Sowing rate 35·9 kg ha−1 
ManagementImplemented from 2002 
Cut (CUT)Spring cutting of vegetation (March) to a height of 100–150 mm with a flail cutter. Cuttings left in situTo reduce abundance of dominant species (especially grasses). Provides gaps in sward for seedling recruitment
Scarification (SCAR)50–60% soil disturbance to depth of 5 cm (March) using a power harrowRemoves living and dead vegetation to create gaps in sward for seedling recruitment
Herbicide (GRAM)Application of Fluazifop-p-butyl (Fusilade Max 125EC; Syngenta, Guildford, UK) (March/April), 100 g active ingredient ha−1Selective graminicide to reduce competitive exclusion by grasses, therefore increasing species diversity

soil macrofauna extraction

As many soil invertebrate species peak in abundance in either spring or autumn, sampling was carried out in April and October 2005. Five soil cores measuring 25 × 25 cm and 10 cm deep were taken from each plot. Cores were located 3 m apart on a transect running parallel to the boundary, halfway between the boundary and crop edge. Twenty soil cores were also taken each time from the main field under winter wheat, arranged in four transects of five cores, parallel to each of the four blocks, 30 m from the boundary.

Each soil core was hand-sorted for 40 min and all macrofauna were extracted into 80% alcohol. Time-limited hand-sorting is a reliable method for estimating soil macrofaunal populations (Schmidt 2001), providing quantitative data from a known volume and resistant to variations in environmental conditions.

species identification

The Lumbricidae, Diplopoda, Chilopoda, Isopoda and adult Coleoptera were identified to species. Species identification of the Coleoptera was restricted to the Carabidae and Staphylinidae (95% of all beetle specimens). Identifications of the Aleocharinae subfamily were undertaken by P. Hammond at the Natural History Museum (NHM), London, UK. Experts at the NHM and the British Myriapod and Isopod Group (BMIG), UK, confirmed species identifications. Nomenclature followed Blakemore (2005), Barber (2006), Duff (2005, 2006), Gregory (2006) and Lee (2006b). To test the main effects of the treatments, overall species density (i.e. number of species per unit area; Magurran 2004) and abundance were calculated for each plot from the combined data (April and October).

conservation status

There is little information on the conservation status of soil invertebrates. Therefore we used information from a variety of sources, including field guides, conservation reviews and distribution atlases (Barber & Keay 1988; Hopkin 1991; Hyman & Keay 1992; Hyman & Parsons 1994; Sims & Gerard 1999; Lee 2006a).

feeding group classification

Although there may be a high degree of omnivory in soil communities (Gunn & Cherrett 1993), stable isotope analyses (McNabb, Halaj & Wise 2001; Schmidt et al. 2004) have recognized distinct trophic levels within the macrofauna, interpreted here as predators, herbivores, fungivores, litter-feeders and soil-feeders. Adult and juvenile earthworms were identified as either litter-feeders (epigeics and anecics) or soil-feeders (endogeics), as in Bouché (1972) and isotope work by Schmidt et al. (2004). Woodlice and millipedes were categorized as litter-feeders (Hopkin 1991; Hopkin & Read 1992) and centipedes as predators (Eason 1964). Beetle species were assigned to a particular feeding group on the basis of their species or genus description (Good & Giller 1991; Luff 1998; Clough, Kruess & Tscharntke 2007) (see Table S2 in the supplementary material). As feeding group density fluctuations may affect ecosystem functioning, abundances of each feeding group were summed for each plot per season to investigate seasonal variation in treatment effects.

other environmental variables

A soil core (127 cm3) was collected alongside each main invertebrate core, at a depth of 5 cm. The fresh weight was recorded, the core oven-dried overnight at 105 °C, and bulk density (g cm−3) and soil moisture (%) calculated (Elliot et al. 1999). A separate soil sample was taken for analysis of soil pH (water), carbon (C) and nitrogen (N). For C and N, the samples from each plot were combined into a composite sample, air-dried and finely ground. Total C and N and organic C (%) of three subsamples per composite sample were measured by dry combustion with gas chromatography in the Electron Microscopy and Mineralogy Analysis (EMMA) laboratory at the NHM.

To reduce the number of co-varying parameters entering analyses, Pearson's correlations were performed on the soil environmental data (bulk density, percentage moisture, pH, percentage total C, N, organic and inorganic C, and C : N ratio). Bulk density was significantly correlated with percentage moisture, percentage N and C : N ratio (r = –0·76, –0·67 and 0·57, respectively; P < 0·001) and percentage total C with percentage organic and inorganic C (r = 0·39, 0·85, respectively; P < 0·001), and so these two parameters, along with soil pH, were used in analyses.

data analysis

General linear mixed models were used to identify the effects of the treatments on different elements of soil macrofauna biodiversity using SAS 9·01 (SAS 2002). The treatments were included as fixed factors and replicate block as random effect. Abundances were log10(n + 1) transformed and non-significant parameters deleted to achieve model simplification. When a significant effect of the treatment was found, post-hoc pairwise comparisons, using a Tukey test, were performed in SAS 9·01. First, species densities and overall abundances of the Lumbricidae, Isopoda, Chilopoda, Diplopoda and Coleoptera in the crop and margin were compared. Densities in the margins were calculated as the average per plot within each replicate block. Secondly, an analysis tested the effects of seed mix and management treatments on species densities and abundances in the margins, and on abundances of individual species of conservation value.

Repeated-measures analyses identified the effect of the seed mix and management treatments, and interactions with season, on the responses of individual feeding group abundances and species densities. As the soil properties were potentially influenced by the treatments, a three-model approach was used. The first model (model I) analysed the fixed effects of seed mix, management and season, and the interactions between the three. Season was identified as the repeated measure, and replicate block as the random effect. The second model (model II) tested the fixed effects of the soil properties, and their interactions with season, on feeding group abundance, using the same repeated-measures mixed model. In all cases deletion was by least significant factors from a type 3 model (order of parameter inclusion was not important), where fixed-effect components of an interaction could not be deleted if the interaction was significant. A final model (model III) was constructed that combined the significant treatments and soil properties from models I and II. This tested whether this model accounted for significant additional variance to that of model I. Differences in the fit of model III relative to model I were assessed by comparing the Akaike's information criterion (AIC) fit statistic. This statistic adjusts for the number of parameters in a model, allowing models with different numbers of parameters to be compared (Whittingham et al. 2006). In SAS, the smaller the AIC, the better the relative fit. For graphical representation, abundances have been converted to values per m2.

Differences in soil invertebrate composition between treatments were assessed using direct ordination methods in canoco 4·5·1 (Ter Braak & Šmilauer 1997). Short gradient lengths (0·771) obtained through a preliminary DCA analysis indicated that a linear method (redundancy analysis; RDA) was most appropriate (Lepš & Šmilauer 2003). Abundances of each species were combined from the two seasons and log-transformed [log10(n + 1)], and species represented by a single individual were excluded from the analyses. The categorical environmental variables (seed mix, management and replicate block) were re-coded as dummy variables. Centring was by species only. Global Monte Carlo permutation tests of all canonical axes under the reduced model (999 permutations, restricted within blocks) tested the significance of the model. Soil cores taken from the crop were included as supplementary samples.

To identify the main effects of seed mix and management on species assemblages, two partial RDA were undertaken, the first with the three seed mixes as environmental variables and the managements and replicate blocks as covariables, and the second with the three managements as environmental variables and the seed mix and replicate blocks as covariables. As before, correlations between the soil macrofauna and the soil properties were tested independently from the treatment effects in a separate analysis. A third pRDA was performed with the soil properties as the environmental variables and the replicate blocks as covariables.

Results

A total of 11 591 individuals was sampled, comprising 115 species, 18 of which were represented by a single individual. The largest group was the Coleoptera, with 34 Carabid species (n = 371) and 51 Staphylinid species (n = 1348), while the Isopods (n = 1662) and Chilopods (n = 350) were the least diverse, with six species each. The Lumbricidae (n = 6949) and Diplopoda (n = 911) had nine species each.

Diversity in the field margins was higher than in the crop, with earthworms, woodlice and beetles having significantly more species and/or higher abundances in the margins (Table 2). Woodlice showed a clear preference for the margins, with no individuals being collected in the crop, and twice as many beetles were found in the margins as the crop. Millipede and centipede diversities were similar in crop and margin.

Table 2.  Results of general linear analysis with mixed models on species density and abundance in arable field margins and crop
 Margin, mean (SEM)Crop, mean (SEM)F1,6
  1. NS, not significant, *P < 0·05, **P < 0·01, ***P < 0·001.

Species density
Lumbricidae  5·11 (0·24)  3·75 (0·25)  15·16**
Isopoda  2·78 (0·21)  0·00 170·45***
Diplopoda  3·22 (0·24)  3·50 (0·29)   2·59NS
Chilopoda  2·08 (0·20)  1·75 (0·25)   7·71NS
Coleoptera 17·31 (1·03) 10·0 (1·58)  15·02**
Abundance (m−2)
Lumbricidae281·20 (39·40)244·40 (27·10)   1·12NS
Isopoda 73·87 (8·54)  0·001220·52***
Diplopoda 36·22 (2·95) 38·40 (12·20)   0·06NS
Chilopoda 13·47 (1·01) 18·40 (3·55)   2·65NS
Coleoptera 80·00 (10·10) 41·20 (6·41)  11·94*

Within the margins, seed mix and management treatments had no significant effect on the species abundances and densities for the Lumbricidae, Chilopoda, Diplopoda and Coleoptera. However, the species abundance and density of the Isopods showed a significant response to management, with scarified plots containing fewer species than the graminicide or cut treatments (abundance, F2,30 = 15·11, P < 0·001; density, F2,30 = 5·60, P= 0·009; Fig. 1).

Figure 1.

Isopod abundance (a) and species density (b) in margin management treatments. Cut, spring cut; Gram, graminicide; Scar, scarification. Different letters denote significant differences between means based on Tukey's post-hoc tests.

Three species were recognized as being of conservation value; the millipedes Polydesmus coriaceous Porat and Nanogona polydesmoides Leach feature in the UK Biodiversity Action Plan (UK BAP; http://www.ukbap.org.uk, accessed 24 April 2007) long list (Lee 2006a), and the carabid Ophonus ardosiacus Lutshnik is nationally scarce category B (Hyman & Keay 1992). Polydesmus coriaceous and Ophonus ardosiacus showed no significant preferences for any of the treatments. Only a single individual of Nanogona polydesmoides was recorded, and therefore this species was not included in the analyses.

Two species whose feeding preferences were unknown were excluded from the functional guild analyses, and low abundances of herbivores and fungivores also prevented inclusion of these groups. Within model I, significant effects of season were found in all cases, except for soil-feeder species density, and there was a significant interaction between season and management for soil-feeder and litter-feeder abundance and predator species density (Table 3). This was because of low numbers of individuals or species in the scarified plots in the spring that then increased to levels equal to, or greater than, the other management treatments in autumn (Fig. 2). Seed mix had no significant effect on feeding group abundance and species density (Table 3). Only soil-feeders responded significantly to the interaction between seed mix and management (Table 3), with differences between the management types in the tussock grass treatments only, where graminicide plots had significantly higher abundances than scarified plots (mean ± SEM, 289·2 ± 16·1 m−2 compared with 168·8 ± 38·8 m−2, n= 8, P= 0·02).

Table 3.  Results of repeated-measures analysis of feeding group abundances and species densities. Model I tested response to the treatments (seed mix and management), month and all interactions. Model II included only the soil properties and their interaction with season and model III established whether combining the significant parameters from models II and I increased the amount of explained variance significantly as judged by the AIC fit statistic (see text for further details)
 Soil-feeder speciesLitter-feeder speciesPredator species
AbundanceDensityAbundanceDensityAbundanceDensity
  1. NS, not significant, *P < 0·05, **P < 0·01, ***P < 0·001.

Model I
Seed mix (seed)NSNSNSNSNSNS
Management (mgmt)NSNSF2,33 = 7·87**F2,33 = 5·71**F2,30 = 10·43***F2,30 = 7·14**
SeasonF1,27 = 10·55**NSF1,33 = 13·01**F1,33 = 12·69**F1,35 = 43·88***F1,33 = 18·84***
Seed × mgmtF4,24 = 3·21*NSNSNSNSNS
Seed × seasonNSNSNSNSNSNS
Mgmt × seasonF2,27 = 10·16***NSF2,33 = 18·71***NSNSF2,33 = 3·46*
Seed × mgmt × seasonNSNSNSNSNSNS
Model II
SeasonNSNSNSNSNSF1,56·8 = 8·02**
Bulk densityNSNSNSNSNSNS
pHNSNSNSF1,38·1 = 8·73**NSNS
CarbonNSF1,61 = 11·7**NSNSNSNS
Bulk × seasonNSNSNSNSNSNS
pH × seasonNSNSNSNSNSNS
Carbon × seasonNSNSNSNSNSF1,57·2 = 5·16*
AICNANANA NA 
Model I   245·7 347·1
Model III   245·0 337·8
Figure 2.

The response of feeding groups to field margin management varies with season: (a) litter-feeder abundance; (b) soil-feeder abundance; (c) predator species density. Open columns, April; shaded columns, October; Cut, spring cut; Gram, graminicide; Scar, scarification. Different letters denote significant differences between means based on Tukey's post-hoc tests.

Within model II, there were very few significant responses to the soil properties and they explained only small amounts of additional variance when added to significant parameters from model I (Table 3). However, species density of the soil-feeders, which did not respond significantly to any of the margin treatments, was significantly negatively correlated with percentage soil carbon.

RDA analysis indicated that soil invertebrate assemblages were not significantly different between the three seed mixes, with the canonical axes (axes 1 and 2) accounting for only 5·6% of the variability in the species data (Table 4). However, the assemblage was significantly influenced by the management treatments, with axes 1 and 2 explaining 12·2% and 1·8% of the species variance, respectively (Table 4). The resulting ordination diagram (Fig. 3) indicated that the first axis divided the scarified treatment from the cut and graminicide treatments, while the second axis separated the cut and graminicide treatments. As the eigenvalue of the second axis was low, its statistical significance was tested by performing another partial RDA (Lepš & Šmilauer 2003). This was carried out as before but using the sample scores of the first axis as covariables in addition to seed mix and block. The second axis was found to be non-significant (F = 0·805, P= 0·771), indicating that only the first axis (i.e. the scarified vs. non-scarified treatments) was worth interpreting.

Table 4.  Redundancy analysis (pRDA) results for soil macrofaunal assemblage responses to field margin treatments. F-values from Monte Carlo permutation tests (999 permutations) of all canonical axes
Environmental variablesCovariablesEigenvaluesF-ratioP-value
Axis 1Axis 2Axis 3Axis 4
Seed mixManagement block0·0310·0250·090·0631·2130·101
Sum of all eigenvalues 0·698   
Sum of all canonical eigenvalues 0·056   
ManagementSeed mix block0·1220·0180·090·0633·069< 0·001
Sum of all eigenvalues 0·783   
Sum of all canonical eigenvalues 0·141   
Soil propertiesBlock0·0500·0320·0210·0171·1640·186
Sum of all eigenvalues 0·838   
Sum of all canonical eigenvalues 0·119   
Figure 3.

RDA triplot showing only species with a fit greater than 14%. The environmental variables are shown as filled triangles. CUT, spring Cut; GRAM, graminicide; SCAR, scarification. Samples plotted using ‘Samp’ scores (i.e. observed values) rather than ‘SamE’ scores (fitted values). Samples: cut plots, crosses; graminicide plots, squares; scarified plots, open triangles; crop plots, filled circles; crop samples are treated as supplementary and are plotted post hoc. Lumbricidae species (o): oAPOcal, Aporectodea caliginosa Savigny; oLUMcas, Lumbricus castaneus Savigny. Isopoda species (i): iPHImus, Philoscia muscorum Scopoli; iTRIpus, Trichoniscus pusillus Brandt. Diplopoda species (d): dBLAgut, Blaniulus guttulatus Fabricius; dBRApus, Brachyiulus pusillus Leach. Carabidae species: BEMlam, Bembidion lampros Herbst; BRAverb, Bradycellus verbasci Duftschmid; DEMatr, Demetrias atricapillus L. Staphylinidae species: ANOrug, Anotylus rugosus Fabricius; ATHspp, Atheta spp. Thomson; DRUcan, Drusilla canaliculatus Fabricius; MOCfun, Mocyta fungi Gravenhorst (sensu lato); TACsig, Tachinus signatus Gravenhorst; XANlon, Xantholinus longiventris Heer.

The first axis of the ordination diagram (Fig. 3) partitioned those species that had higher abundances in the scarified treatment (e.g. beetles Bembidion lampros Herbst and Anotylus rugosus Fabricius, and the millipedes Blaniulus guttulatus Fabricius and Brachyiulus pusillus Leach) from those with lower abundances (e.g. earthworms Lumbricus castaneus Savigny and Apporectodea caliginosa Savigny, the woodlice Philoscia muscorum Scopoli and Trichoniscus pusillus Brandt, and the beetles Tachinus signatus Gravenhorst and Demetrias atricapillus L.). The position of the crop samples, which were placed on the ordination diagram as supplementary samples, indicated a high degree of similarity in species composition between samples from scarified plots and those taken from the crop (Fig. 3). The species assemblage was not significantly correlated with the soil properties (Table 4).

Discussion

Diversity levels were generally higher in the field margins than in the crop, and the margins seemed to be particularly important refuges for woodlice and beetles. Establishing arable field margins using characteristic seed mixes is an effective way of increasing above-ground biodiversity (Thomas & Marshall 1999; Meek et al. 2002). Theoretically, increasing vegetation diversity could benefit the below-ground system by providing a more heterogeneous resource, both by food provisioning and providing shelter (Wardle & van der Putten 2002). However, studies of the relationship between above- and below-ground systems have produced inconsistent results (Wardle & van der Putten 2002) and, in this study, soil macrofaunal diversity and composition did not vary significantly between the three seed-mix treatments. This may indicate a time-lag in soil invertebrate response, as suggested by Hedlund et al. (2003), and, while epigeal invertebrates respond to new seed mixes in 2 or 3 years (Frank & Reichhart 2004), soil invertebrates may take longer to respond than the 4 years these particular margins have been in place.

In contrast, the physical disturbance of the soil environment has a profound effect on soil communities (Bardgett 2005) so, unsurprisingly, the soil macrofaunal community responded significantly to the field margin management treatments. Scarified plots had lower species densities of woodlice, a group that is sensitive to soil disturbance (Paoletti & Hassall 1999). Scarification also reduced the species densities and/or abundances of the soil- and litter-feeders and predators, although populations appeared to recover by the autumn.

Species densities of the soil-feeders did not respond significantly to any treatments, but there were significantly fewer species at higher levels of soil C. This is a surprising result, as soil communities may not be affected by competitive exclusion at high resource levels in the same way as above-ground communities (Bardgett, Yeates & Anderson 2005). However, Cole et al. (2006) suggested that, while competition is limited between the smaller soil fauna because of the heterogeneous nature of the soil environment (Bardgett, Yeates & Anderson 2005), this heterogeneity is at too fine a scale to constrain the more motile macrofauna.

It would appear that scarification, while less severe, has many of the negative effects caused by conventional tillage on soil communities (Curry 2002). These include direct mortality caused by machinery, as well as indirect effects, such as reduced surface residues and greater exposure. However, in the margins these negative effects seem short-term and populations are able to recover within a season, probably as a result of dispersal from neighbouring plots and boundary features. In addition, recovery of soil-feeding populations may be the result of the amount of organic matter mixed into the soil. Such a positive response has been found in populations of Lumbricids following conversion of grassland to arable (Edwards & Lofty 1977) and, whereas populations within arable fields decline with repeated cropping because of removal of organic matter through harvesting, field margins may represent a more ‘closed’ system in terms of nutrient cycling, suggesting that long-term reductions in Lumbricid populations may be less dramatic.

Nevertheless, scarification apparently influences species composition, with species more commonly associated with cropped or exposed habitats (e.g. the millipede Blaniulus guttulatus and the ground beetle Bembidion lampros) found in the scarified plots. Litter-dwelling species, such as the woodlice Philoscia muscorum and Trichoniscus pusillus and the epigeic earthworm Lumbricus castaneus, with their requirement for surface residues, had low densities in scarified plots. This suggests that scarification does not increase farmland biodiversity as it may maintain only those species already commonly found in cropped areas. In addition, by reducing populations of litter-feeders in the margins available to re-invade the field, scarification may impede the restoration of a more ‘natural’ system when a cropped field is set-aside, fallowed or converted to pasture.

The temporal response of feeding groups to the scarification treatment may have implications for ecosystem functioning. Grassy margins in arable fields are valuable overwintering sites for many polyphagous predators (Pfiffner & Luka 2000), from where they move into adjacent crops in the spring. Scarifying in early spring (March) may disrupt this process, through direct mortality of overwintering individuals, and by making the margin habitat unsuitable for immediate recolonization because of lack of suitable shelter, and may therefore impede biological control of crop pests.

The decomposition of organic matter in the margins is probably affected by the management treatments. In cut and graminicide plots, dead vegetation accumulate on the soil surface. Here, epigeic litter-feeders fragment the litter, thus increasing the surface area available for attack by micro-organisms, and incorporate it into the soil, where it may be used by soil microbes and fauna such as geophagous earthworms. In contrast, vegetation is incorporated directly into the soil through scarification and, while initial decomposition rates may be suppressed because of the mortality of macro-detritivore populations from mechanical damage, more favourable conditions in the soil can increase decay rates of surface litter (Beare et al. 1992). In no-till systems, the separation of surface carbon-rich litter and mineralized nitrogen in the soil results in fungal-based decomposition systems, with slower rates and reduced losses of soil nutrients (Holland & Coleman 1987). In comparison, in tilled systems, where litter carbon is in direct contact with soil nitrogen, decomposition rates are higher and bacteria-based, with greater nutrient loss (Holland & Coleman 1987).

However, while in cropping systems retention of soil fertility is essential for sustainable farming, the opposite may be true in field margins, where the aim to recreate more ‘natural’ systems requires a reduction in soil fertility followed by restoration of a fungal-based pathway of decomposition (Bardgett 2005). In restoration schemes, reduction in soil fertility is usually achieved by removing cut vegetation, but scarification may provide an alternative (Marrs 1993), although adverse effects on soil fauna, as seen in this study, may outweigh the benefits. In this study, re-invasion from neighbouring plots may have helped invertebrate populations to recover following soil disturbance, but this would not occur if the entire margin was scarified.

The conservation value of the margin treatments appears to be minimal, with only three notable species occurring within the margins. This finding agrees with the conclusions of Kleijn et al. (2006), who evaluated the success of five agri-environment schemes and found that, while most schemes encourage common species, their value for rare or uncommon species is questionable. However, assessment of such schemes for conservation of endangered species is a challenge that requires high sampling intensity, both spatially and temporally (Potts et al. 2006). The single year of sampling presented here may not provide a definitive evaluation. In addition, agri-environment schemes are targeted at wider biodiversity enhancement rather than benefiting rarer species (Potts et al. 2006). Other factors contributing to the lack of rare species may include a time lag in their response to habitat improvement, which is likely to be more significant for soil invertebrates with limited dispersal abilities, and the high intensity of land-use prevalent in the surrounding landscape of eastern England. This means that, despite being part of a presumably species-rich regional pool that covers a large part of lowland Britain and northern Europe, the low population densities supported by these agro-ecosystems prevent the development of highly diverse local species assemblages (Eggleton et al. 2005).

conclusion

This study shows that arable field margins supported a higher diversity of several soil macrofaunal groups than cropped areas. The seed mix used in arable field margins may have less influence on soil macrofaunal communities than subsequent management practices. In particular, if the aim is to increase farmland biodiversity and recreate a more ‘natural’ habitat for soil macrofauna, the use of scarification is not recommended. To enhance the below-ground macrofauna, development of future agri-environment schemes should consider the impact of management on the dynamics of the leaf litter, and minimize soil cultivation. A substantial surface litter layer will encourage the litter-dwelling invertebrates that tend to avoid conventional arable systems. However, such management may conflict with other aims of these schemes (e.g. high floristic and pollinator diversity) that would require a reduction in soil fertility. This emphasizes the problems faced when designing conservation schemes: development of management for one target group may not favour another (Olson & Wackers 2007).

Acknowledgements

We thank Roger Booth and Peter Hammond (NHM) and Paul Lee and Steve Gregory (BMIG) for their help with the identifications. Thanks also to Val Brown (CAER), Sarah Cooke (ADAS Boxworth) and Sarah James (EMMA lab, NHM) for their support. Jo Smith is funded by an RETF grant from the University of Reading, and a grant from the Natural History Museum. SAFFIE (LK0926): This project is sponsored by the Department for Environment, Food and Rural Affairs (Defra), the Scottish Executive Environment and Rural Affairs Department (SEERAD) and Natural England (formerly English Nature), through the Sustainable Arable LINK programme. The industrial funders are British Potato Council, Agricultural Industries Confederation (AIC), Crop Protection Association, Home-Grown Cereals Authority (HGCA), Jonathan Tipples, Linking Environment and Farming (LEAF), Royal Society for the Protection of Birds (RSPB), Sainsbury's Supermarkets Ltd, Syngenta, the National Trust and Wm Morrison Supermarkets PLC.

Ancillary