- Top of page
- Materials and methods
- Supporting Information
Alternative management of field margins has been used extensively in arable cropping systems to improve the nature conservation value of farmland (Marshall & Moonen 2002; Meek et al. 2002; Woodcock et al. 2005b) as well as to increase densities of beneficial invertebrates (Thomas et al. 2001). However, to date such field margin management has rarely been used in grasslands (but see Haysom et al. 1999). In England and Wales, permanent (> 5 years) and temporary (< 5 years) grassland covers 40% of all agricultural land (Defra 2004), of which a large proportion has been modified by intensive agricultural management (Duffey et al. 1974; Blackstock et al. 1999). Intensive grassland management normally involves the use of inorganic fertilizers (NPK), improved drainage, reseeding with one or two grass species/varieties and the replacement of hay with silage cutting as the principal conserved forage (Frame 2000). Technological developments in silage production have enabled greater flexibility in the timing and frequency of cutting as higher grass water contents are tolerated (Vickery et al. 2001). Increased productivity has also allowed higher stocking densities to be supported. The increased levels of disturbance associated with these changes in management have had a major impact on the composition and structure of agricultural grassland (Duffey et al. 1974; Blackstock et al. 1999; Vickery et al. 2001), leading to species-poor and structurally uniform grasslands of low nature conservation value (Duffey et al. 1974). Declines in populations of higher plants (Blackstock et al. 1999), farmland birds (Vickery et al. 2001) and invertebrates (Duffey et al. 1974; Morris 1978; Asher et al. 2001) have all been attributed to this intensification of grassland management.
Invertebrates represent a key functional component of agricultural grassland systems (Woodcock et al. 2005a, 2006) and have importance as pests (Norris 1994) and food resources for farmland birds (Vickery et al. 2001) and their own innate conservation value (Asher et al. 2001). Invertebrates also contribute to key ecosystem functions, such as nutrient cycling, biocontrol and pollination (Norris 1994). The causes of declines in invertebrate populations in response to improved grassland management are primarily driven by changes in the plant community as they respond to intensification in the form of cutting, grazing, fertilizer application and reseeding regimes (Duffey et al. 1974; Vickery et al. 2001). The direction of invertebrate responses to cutting has been related to the interaction between the insects’ lifecycles and the phenological development of the vegetation in the period since it was last cut (Duffey et al. 1974; Morris 1978; Asher et al. 2001). Increased stocking densities have negative impacts resulting from the disturbance caused by destruction of the sward canopy (Duffey et al. 1974; Morris 1978). Inorganic fertilization (Fenner & Palmer 1998) and residues of anti-helminth drugs in dung (Hutton & Giller 2003) have also been shown to have negative impacts on invertebrate populations.
An increased complexity of the above-ground vegetation structure (often referred to as sward ‘architecture’) is of key importance to both the abundance and diversity of invertebrates (Gibson, Hambler & Brown 1992; Dennis, Young & Gordon 1998; Morris 2000). The presence and availability of structures, such as flowers, seed heads, stems and leaves, have been shown to be important for many phytophagous and predatory invertebrates, as well as insect parasitoids (Gibson, Hambler & Brown 1992; Dennis, Young & Gordon 1998; Finke & Denno 2002). The intensification of grassland management, in particular multiple silage cuts and grazing, has been associated with reduced sward architectural complexity (Gibson, Hambler & Brown 1992; Morris 2000). This reduction in sward architecture would normally result in a reduced availability of reproductive plant structures, many of which represent key resources for phytophagous invertebrates (Morris 2000; Woodcock et al. 2005a). The strong temporal component of sward architecture would also impact on grassland invertebrates (Morris 2000).
The present study aimed to establish whether the beetle assemblages of intensively managed lowland grassland can be enhanced by modifying field margin management. The impacts of combinations of key management practices, namely the height and timing of sward cuts, cattle grazing and inorganic fertilizer inputs, were investigated by considering changes in the beetle species diversity and composition along a time series. This was intended to provide information to underpin future developments of agri-environmental policy for improved grasslands. The study presented here focused on beetle assemblages only. Beetles represent an important component of the grassland fauna in terms of overall abundance, species richness and the variety of functional groups they represent (Thiele 1977; Bohac 1999; Woodcock et al. 2005a). In addition, they are directly and indirectly dependent on plant assemblages and provide a link between plants and higher trophic levels, for example birds (Duffey et al. 1974; Vickery et al. 2001). By assessing beetle assemblage responses to management it is believed that recommendations of best practice will improve the biodiversity value of intensively managed lowland grasslands.
- Top of page
- Materials and methods
- Supporting Information
A total of 33 102 beetles was identified to one of 225 species (see Appendix S1 in the supplementary material), of which 42 species were represented by singletons. Staphylinidae (excluding Aleocharinae) were the most abundant (n = 13 894) and species rich (73 species), followed by Carabidae (46 species, n = 7284). Other families identified were the Apionidae (16 species, n = 4535), Curculionidae (42 species, n = 3460), Chrysomelidae (37 species, n= 2419), Coccinellidae (7 species, n = 1293) and Elateridae (4 species, n= 270). Plant species richness, while increasing generally between 2003 and 2005 (F2,165 = 18·4, P < 0·001), neither differed between treatments (F6,86·3 = 1·50, P > 0·05) nor showed an interaction with year (F12,153 = 0·59, P > 0·05). Grass sward architecture, which represented the dominant component of sward architecture, showed a treatment (F6,89·5 = 79·9, P < 0·001) and overall year effect (F2,170 = 110·9, P < 0·01) only. The architecture of grasses, forbs, legumes and dead vegetation increased from treatment 1 to 7.
Model I tested the response of beetle abundance, species richness, diversity and evenness to the management treatments and their interaction with year (Table 2). Significant effects of sample year were found in all cases, although only beetle diversity did not show a significant year × treatment interaction. Only beetle abundance and species richness were characterized by an overall treatment effect (for pairwise comparisons see Appendixes S2 and S3 in the supplementary material). In general, by 2005 the extensively managed treatments (5–7) supported the highest abundances and species richness of beetles relative to the other treatments (Fig. 1a,b). This pattern was reversed in the case of beetle evenness (Fig. 1d). To test that the response of beetle species richness was not an artefact of increasing beetle abundance, this latter parameter was included as a covariate. While beetle species richness was positively correlated with abundance (F1,228 = 89·5, P < 0·001), significant responses to the effects of treatment (F6·78 = 9·14, P < 0·001), year (F2,159 = 60·6, P < 0·001) and treatment × year (F12,155 = 2·72, P < 0·01) remained. Patterns of species richness between the treatments remained similar to those of the original model uncorrected for beetle abundance (Fig. 1b). Therefore the response of beetle species richness to treatment was not an artefact of treatment differences in abundance.
Table 2. Model I (MI): results of repeated-measures analysis with mixed models used to test responses of beetle abundance, species richness (loge n+ 1), Shannon–Wiener diversity and evenness to margin management treatments (treat) and their interaction with year (year). Model II (MII): considers only continuous environmental measures of variation. Model III (MIII): determines whether the addition of the significant treatment effects of model I to those of model II results in a significant increase in the explained variance. Only significant effects have been shown for model II. See the Methods for environmental variable abbreviations. NA, analysis not applicable; NS, P > 0·05; *P < 0·05; **P < 0·01; ***P < 0·001. Positive or negative correlations are indicated by + or – in parentheses, where three consecutive symbols represents the direction of correlations for successive years (2003–05)
|Abundance (loge n+ 1)||Species richness (loge n+ 1)||Species diversity||Species evenness|
|Treat: F6,81·1 = 3·60**||Treat: F6,79·1 = 12·5***||Treat: NS||Treat: NS|
|Treat × year: F12,156 = 5·61**||Treat × year: F12,156 = 4·16**||Treat × year: NS||Treat × year: F12,158 = 2·51**|
|Year: F2,156 = 95·5***||Year: F2,156 = 15·9***||Year: F2,170 = 19·2**||Year: F2,158 = 9·45***|
|Year: F2,211 = 30·6***||Year: NS||Year: F2,208 = 15·4***||Year: F2,189 = 16·1***|
|Sward architecture||Sward architecture||Sward architecture||Sward architecture|
|GA × year: F2,209 = 18·1*** (+ + +)||GA × year: F2,206 = 6·03* (+ + +)||FA × year: F2,207 = 3·28* (+ + +)||FA × year: F2,190 = 3·45* (+ + –)|
|LA: F1,186 = 34·1*** (+)||FA: F1,208 = 3·96* (+)||LA: F1,185 = 5·37* (–)||LA × year: F2,181 = 5·35** (–––)|
| ||LA: F1,199 = 11·6*** (+)|| ||DA × year: F2,178 = 8·25** (– ––)|
|DA: F1,228 = 9·02** (+)|| || |
|Plant percentage cover||Plant percentage cover||Plant percentage cover||Plant percentage cover|
|%Ru × year: F2,205 = 4·17* (– ––)||%Ru: F1,172 = 7·54** (–)||%Ru × year: F2,185 = 3·76* (– ––)||%Ra × year: F2,183 = 8·60** (– ––)|
|%Ra: F1,177 = 5·96* (–)||%Tu × year: F2,205 = 8·48** (+ + –)||%Ra × year: F2,198 = 4·78** (–––)||%Tu × year: F2,180 = 4·10* (+ + –)|
|%Tu × year: F2,216 = 4·77** (+ +–)||%LP × year: F2,207 = 9·12** (+ +–)||%Tu × year: F2,195 = 10·1** (+ + –)||%Ci × year: F2,188 = 3·93* (+ + +)|
|%Ci: F1,158 = 5·34* (–)|| ||%Ci × year: F2,192 = 5·43** (+ + –)||%Ta × year: F2,194 = 3·00* (+ + –)|
| ||%LP × year: F2,207 = 9·76** (+ + –)||%Pl: F1,195 = 3·87* (–)|
| ||%AS × year: F2,208 = 6·80** (+ + –)||%LP × year: F2,192 = 7·54** (– ––)|
| || ||%AS × year: F2,198 = 5·03** (– ––)|
| ||PSR: F1,151 = 12·2*** (–)||PSR: F1,191 = 9·36** (–)|
| = 0·80, NS|| = 3·87, NS||NA|| = 7·32, NS|
Figure 1. Response of beetle abundance (± SE) (a), species richness (± SE) (b), and Shannon–Wiener evenness (± SE) (c) to the seven field margin management treatments and sample year. Significance values (P) indicate the significance of the interaction between treatment and year.
Download figure to PowerPoint
Model II tested the interaction between the continuous environmental measures and year on beetle abundance and species richness, diversity and evenness. Significant responses to sward architecture, the percentage cover of floral groups and their interaction with year were found for beetle abundance and species richness, diversity and evenness (for the full list of effects see Table 2). Both beetle diversity and evenness also showed significant negative correlations with plant species richness. The significant continuous environmental parameters of model II did not, however, explain any additional variance when added to the significant treatment and year effects of model I.
Responses to margin management treatments were found for the functional structure of the beetle assemblages (Table 3). The proportional abundance represented by the predatory, root, foliage and seed/flower feeding functional groups all showed significant responses to treatment (Fig. 2a–d) and, with the exception of the foliage feeders, significant treatment × year interaction. Both the predatory functional group and the foliage feeders also showed a significant year effect.
Table 3. Response to margin management treatments of the proportion of the total abundance made up of predatory species. This analysis was repeated to assess the proportion of the overall abundance of phytophagous beetles composed of root/stem-, foliage- and seed/flower-feeding functional groups. *P < 0·05; **P < 0·01; ***P < 0·001; NS, P > 0·05
| ||Treatment||Year||Year × treatment|
|Predatory||F6,76·4 = 5·17***||F2,148 = 4·73**||F12,154 = 2·65***|
|Root feeders||F6,90·7 = 4·60***||F2,156 = 0·61, NS||F12,161 = 4·35***|
|Foliage feeders||F6,95·3 = 11·6***||F2,162 = 12·3***||NS|
|Seed/flower feeder||F6,67·6 = 13·6***||F2,153 = 0·77, NS||F12,161 = 2·03*|
Figure 2. Response to field margin management treatment of the proportional representation of beetles in four functional groups, the predatory (a), root/stem (b), seed/flower (c) and foliage (d) -feeding species (± SE). The response variable represents the proportion each functional group makes up of the total beetle abundance (n). The exception is for the root/stem-, seed/flower- and foliage-feeding groups which were represented as proportions of the phytophagous beetles abundance only.
Download figure to PowerPoint
Beetle assemblage structure was analysed using pRDA to assess the significance of both margin management treatment and secondary environmental factor interaction with year. The interaction between margin treatment and year had a significant effect on the beetle assemblage, accounting for 16·7% of the unexplained model variance. When the individual treatment and year interactions were tested, significant effects on beetle assemblage structure were found for treatment 1 (F = 1·54, P < 0·05), treatment 2 (F = 2·44, P < 0·01), treatment 3 (F = 1·84, P < 0·05), treatment 4 (F = 1·79, P < 0·01) treatment 5 (F = 1·42, P < 0·05), treatment 6 (F = 2·88, P < 0·01) and treatment 7 (F = 3·34, P < 0·005). There were also clear differences in the successional trajectories of the treatments between 2003 and 2005 (Fig. 3). In particular, the successional trajectories of the more extensively managed treatments (6 and 7) were moving in the opposite direction to those of the control plots. Conversely, treatments 3 and 4, which received NPK fertilizer, were characterized by successional trajectories similar to those of the control (treatment 1). In addition to the treatment effects, significant environmental parameter interactions with year were also found for sward architectural components (grass, legumes and dead vegetation) as well as dung density, bare ground cover and plant species richness. The interaction between year and the percentage cover of the legumes, tussock grasses, Cirsium spp., Ranunculus spp. and Agrostis stolonifera also had a significant effect on beetle assemblage structure (Table 4). When all the significant interactions were included in a single model (treatment × year and environmental variable × year), the beetle assemblage was significantly correlated with the environment and explained 39·7% of the variation in the species data not accounted for by the covariables.
Figure 3. Ordination diagrams of the pRDA for years 2003–05 based on the beetle assemblages. (a) The temporal interaction between sample year and the control and management treatments (T.2–T.7). The change with time of the beetle assemblages is emphasized by the connection of the centroids of the year × treatment interaction with arrows, from the 2003 × treatment (start of first arrow) to the 2004 × treatment (end of first arrow) to the 2005 × treatment (end of second arrow). (b) The companion beetle species scatter plot to (a). Only the 22 species with the best fit to the first two axes of the ordination have been shown, with the first four letters of the generic and specific names (see Appendix S1 for abbreviations). Species functional group was predatory unless otherwise indicated by: S, seed/flower feeding; F, foliage feeding; R, root/stem feeding.
Download figure to PowerPoint
Table 4. Results for partial redundancy analysis of beetle assemblage responses to both management treatment and the floristic composition and architectural structure of the field margins. All significances were tested using Monte Carlo permutation tests (1000 permutations) of both canonical axes. See the Methods for environmental variable abbreviations. *P < 0·05; **P < 0·01; NS, P > 0·05
|Environmental variable||F||Explained variance (%)||Environmental variable||F||Explained variance (%)|
|Treatment × year||2·19**||16·7||Plant percentage cover|| || |
| || ||%Le × year ||1·63*||2·4|
|Sward architecture|| || ||%Tu × year||1·44**||2·0|
|GA × year||2·33**|| 3·1||%LP × year||NS||–|
|FA × year||NS||–||%AS × year||1·31*||1·8|
|TA × year||4·30**|| 5·6||%Ru × year||NS||–|
|DA × year||2·21**|| 2·9||%Ra × year||1·36**||1·8|
| || ||%Pl × year ||NS||–|
|Others|| || ||%Ci × year||1·66*||2·2|
|Dung × year||1·00*|| 1·5||%Ta × year||NS||–|
|BG × year||1·41**|| 2·0|| || || |
|PSR × year||1·69**|| 2·4|| || || |
- Top of page
- Materials and methods
- Supporting Information
By managing field margins, the availability of extensively managed habitats within conventionally managed improved grasslands can be increased. As margin management will influence both flora and fauna, assessing the relative benefits of alternative practices is important if they are to be implemented as part of agri-environment schemes (McCracken & Bignal 1998). One of the clear implications of extensification of field margin management (i.e. the absence of NPK fertilizer, grazing and multiple silage cuts) was a change in the successional trajectory of the beetle assemblages away from what was characterized by the control (Fig. 3). This change was most clearly seen for those treatments that were either unmanaged (treatment 7) or received a single sward cut in May (treatment 5) or July (treatment 6). By 2005, the assemblages of these treatments were characterized by greater proportions of seed/flower-feeding beetles. Interestingly, of those treatments receiving multiple silage cuts each year (treatments 1–4), treatment 2, which was unique in receiving no NPK fertilizer, differed in its successional trajectory from that of the control. The occurrence of what appeared to be a successional change in the structure of the beetle assemblages of treatment 1 (control) was, however, unexpected. This successional change is thought to have been driven by the drier conditions that characterized the 2004 and 2005 sample years. This could have caused an overall reduction in beetle abundance, particularly in 2005, as well as successional shifts in beetle species composition in the control, most probably in response to changes in the plant assemblages as the plots became drier.
The continuous measures of both sward architecture and plant percentage cover were often characterized by strong year interactions. For example, the effect of grass sward architecture showed consistent positive correlations with beetle abundance and species richness for 2003, 2004 and 2005. This reflects the importance of structurally complex tussock grasses in providing an increased diversity of niches for epigeal beetles (Dennis, Young & Gordon 1998; Morris 2000). Legume architectural complexity also had positive effects on both beetle abundance and species richness, an affect attributed to the increased availability of reproductive structures important for a number of phytophagous beetles, in particularly the Apionidae (Woodcock et al. 2005a; Hoffman 1950–58). Conversely, legume sward architecture was negatively correlated with beetle evenness, an effect also seen for all years. This reduction in evenness is attributed again to members of the Apionidae, principally the seed-feeding weevil Protapion dichroum (Bedel), which became dominant species in architecturally complex legume swards.
The responses of beetle abundance and species richness, diversity and evenness to plant percentage cover and sward architecture were not always consistent in their direction between years. This was most apparent for grass percentage cover, in particular that of the tussock grasses, and was attributed to successional shifts in beetle assemblages in response to margin management. As species composition changed throughout the succession, so did the characteristics of how the overall assemblage responded to plant percentage cover and sward architecture. In each case these changes would reflect the individual species’ traits of the assemblages within each treatment and year.
Beetle abundance and species richness and evenness also responded to treatment and year interactions. Between 2003 and 2005 this was characterized by increases in beetle abundance and species richness in the extensively managed treatments (5–7) relative to that observed in the more intensively managed treatments (1–4). Increased sward architectural complexity and the establishment of key floral species in the extensively managed treatments would have increased the relative importance of these treatments in terms of beetle abundance and species richness by 2005 (Duffey et al. 1974; Morris 2000; Woodcock et al. 2005a). Conversely the levels of beetle evenness in the intensively managed treatments were proportionally higher relative to treatments 5–7, a difference that became more pronounced from 2003 to 2005. This was potentially a result of the drier conditions in 2004 and 2005 impacting negatively on dominant beetle species associated with the improved grassland treatments (Frampton, van den Brink & Gould 2000).
For the seed/flower-feeding functional groups, and indeed for many phytophagous species, management extensification was seen to be beneficial. Increased abundance of plant reproductive structures as a result of the greater sward architectural complexity of the extensively managed treatments was beneficial in terms of many larval feeding resources (Morris 2000; Woodcock et al. 2005a). The proportional representation of the seed/flower-feeding beetles was lowest in those treatments receiving NPK fertilizer (treatments 1, 3 and 4). Increased availability of NPK fertilizer may have resulted in recruitment limitation, competitive exclusion or the loss or reduction in seed set of plants important for phytophagous beetles (Kirkham & Wilkins 1994; Tilman 1997). For example, the grass L. perenne, although found to decrease over the duration of the experiment, remained prevalent in these treatments and this may have excluded other plants important to phytophagous beetles (Mountford, Lakhani & Kirkham 1993). Treatment 2, while being grazed and receiving two silage cuts, supported higher proportional abundances of the seed/flower-feeding functional group. This is again attributed to the lack of NPK fertilizer in this treatment reducing the competitive displacement by dominant grasses of important floral species for the beetles.
Invertebrate responses to grazing are common in the literature (Thiele 1977; Gibson, Hambler & Brown 1992; Woodcock et al. 2005a). In this study, response to cattle grazing were subtle and only seen in an assemblage-level response to the percentage cover of dung. This relatively small effect may be explained by the late application of grazing as a management practice (late August to September). The assemblage-level responses to dung density were, however, characterized by a relatively small proportion of the overall fauna, for example dung-associated species such as Philonthus varians (Paykull) (Staphylinidae).
Management in the intensive grasslands was extremely variable on a field by field basis, with stocking densities, timing of cutting and the application rate of NPK fertilizer differing considerably between fields within a single farm (Frame 2000). For this reason the control used in this study was not an unfenced area subject to the same management of individual fields, but rather an approximation of what was considered intensive grassland management. This management used for the control treatment could then be repeated for each replicate across all years. For this reason, both floral and beetle assemblages in the control treatments could have differed from what was found in the fields within which replicates were situated. While this difference was small, the choice of an artificial control could be interpreted as introducing some bias into the results; however, such an effect was believed to be minimal.
Without the implementation of alternative management practices to reverse downward trends in populations of grassland flora and fauna, it is likely that the conservation status of many currently infrequent or rare grassland species of a variety of taxa will become critical (Duffey et al. 1974; Blackstock et al. 1999; Asher et al. 2001; Vickery et al. 2001). Agri-environment schemes are now a mandatory component of European Community Rural Development Regulations and, while they may not always be effective (Kleijn & Sutherland 2003), they represent an important instrument for improving biodiversity in agricultural systems (Ovenden, Swash & Smallshire 1998). Conclusions from the 3 years of this study indicate that managing field margins can serve as a method for diversifying the structure of beetle assemblages while at the same time retaining the majority of the improved grassland under conventional management practices. Although many other invertebrates also respond to plant species richness and architecture, it is possible that the responses observed for beetles in this study may differ from those found for other orders (Meek et al. 2002). It is also true that the application of these management practices to field margins of different widths may also result in different responses in the beetle assemblages, particularly for narrow margins (< 2 m). There was some evidence for the benefits to beetle assemblages of stopping NPK fertilizer application, even when other conventional management practices were maintained. Such a simple modification of conventional improved grassland management may be useful as a cheap and simple practice to adopt in an agri-environment scheme, and has already been adopted by the new Entry Level Stewardship Scheme in the UK (Option EK3 Permanent Grassland with very low inputs; Defra 2005). Its benefits, however, would be minimal relative to those associated with the more extensively managed treatments, which are either unmanaged or receive only a single sward cut each year. These extensively managed treatments, which were characterized by higher abundances and species richness of beetles, may have increased cost implications because they required livestock exclusion fences. Therefore, it may be more cost-effective to place whole fields under this form of extensive management, negating the need for additional fencing. If such a whole-field approach is to be undertaken, investigation into the benefits of creating additional spatial variation in sward structure, possibly by using patchworks of low-intensity cutting regimes, would be valuable. All these factors would need to be considered in terms of agri-environmental policy.