Action

Plant nectar flower mixture/wildflower strips

How is the evidence assessed?
  • Effectiveness
    100%
  • Certainty
    75%
  • Harms
    not assessed

Study locations

Key messages

  • A total of 80 individual studies have in some way investigated the effects of flowering strips on biodiversity. Sixty-four individual studies show some benefits to one or more wildlife groups.
  • Sixty-five individual studies reported the effects of flower strips on invertebrates. Of these, fifty reported positive effects. Forty-one studies from eight European countries (including five reviews and twenty-three replicated controlled studies, of which one randomized and two site comparisons) found evidence that flower strips had a positive influence on invertebrate numbers with increased abundance, species richness/diversity, or both. Ten studies (nine replicated of which two controlled) found invertebrates visited or foraged on flower strips but did not specify increases/decreases in numbers. Two studies found effects on ground beetles other than changes in numbers. One replicated controlled study showed that ground beetles were more active or had enhanced feeding/reproductive conditions in flower strips. A review found flower strips supported ground beetle species that were rarely found in crops. Fifteen studies reported mixed or negative effects of flower strips on invertebrates. Six studies found no significant effects.
  • All five studies investigating the effects of wildflower strips on small mammals (four replicated studies from Switzerland and one review of studies from north-western Europe) found evidence that small mammals benefit from strips sown with wildflowers or flowers rich in pollen and nectar, with increases in abundance, density and species richness. One replicated study from Switzerland reported that most common vole home ranges and core regions of their territories were found within a wildflower strip.

About key messages

Key messages provide a descriptive index to studies we have found that test this intervention.

Studies are not directly comparable or of equal value. When making decisions based on this evidence, you should consider factors such as study size, study design, reported metrics and relevance of the study to your situation, rather than simply counting the number of studies that support a particular interpretation.

Supporting evidence from individual studies

  1. A replicated, controlled study in June to October 1985-1986 and February to August 1987 in a winter wheat field in north Germany (Nentwig 1989) could not detect consistent differences in arthropod numbers between sown flower strips (1 m-wide), winter wheat strips (12 m-wide) and the winter wheat control field. The number of arthropod pest species, weed species cover and crop yield did not differ significantly between the strip types. Arthropod abundance varied greatly over the season, between years and between trapping methods. Five flower strips (1 m-wide, each separated by a 12 m wheat strip) were sown in 1.2 ha of a 1.8 ha arable field with a seed mix containing Crimson clover Trifolium incarnatum, red clover T. pratense, lupin Lupinus angustifolius and winter rape Brassica napa in May 1985 and 1986. The remaining area of the same arable field was used as a control. Arthropods were sampled using pitfall and yellow bowl traps in all three treatments throughout the season in all three years. Plant biomass, vegetation cover, weed species composition and frequency were monitored monthly.

    Study and other actions tested
  2. A replicated study in 1989 in Hertfordshire, UK (Williams & Christian 1991) found that seven species of bumblebee Bombus spp., including the long-tongued common carder bee B. pascuorum, and one cuckoo bumblebee B. [Psithyrus] vestalis foraged on plots sown with phacelia Phacelia tanacetifolia. Of observed worker bumblebee visits, 97% were for nectar, not pollen. The plots each flowered for six to eight weeks, with a maximum flower density of more than 4,000 flowers/m2 on the plot sown in late May. The plot sown in July flowered until early December. Three 9 m2 plots of phacelia were sown at Rothamsted Research experimental farm in May and July 1989. Bee and flower densities were recorded weekly. Flowers were counted in a 0.25 m2 area of each plot. Bees were counted at 09:00, 11.00, 13.00 and 15.00 in each plot, their behaviour, species and gender were recorded.

    Study and other actions tested
  3. A replicated, controlled study in the summers of 1990-1991 in a cereal field in Switzerland (Lys & Nentwig 1992) found higher recapture rates of three ground beetle species (Carabidae: Carabus granulatus, Poecilus cupreus, Pterostichus melanarius) in wildflower strips (57.7%, 41.8%, and 19.8% recaptured in a different trap to the trap of first capture) than in the cereal control area (20.0%, 26.7% and 8.8% recaptured) indicating these species were more active in the wildflower strips. The activity density of four ground beetle species (P. cupreus, Pterostichus anthracinus, Pt. melanarius, C. granulatus) was significantly higher in wildflower margins than in the crop. In 1991, two species moved significantly more from the cereal crop to the wildflower margins than vice versa (P. cupreus and Pt. melanarius). After harvest, only two species Harpalus rufipes and Pterostichus niger showed a strong association with wildflower strips, with most individuals being recaptured in wildflower strips, irrespective of the habitat they were initially caught in (crop or wildflower strip). Four wildflower strips (1.5 m-wide) were studied. Strips were sown in 1989 at 12, 24 and 36 m apart in one part of a winter cereal field. The remaining area of cereal field was used as a control. After establishment, strips were left untreated for three years. Ground beetles were sampled from May-September 1990 and April-July 1991 using a network of numbered pitfall traps (diameter 7 cm) placed in rows in the strips and the cereal field. Captured beetles were individually marked and released about 10 cm from the trap they were caught in. This study was performed within the same experimental site as (Nentwig 1989, Bürki & Hausammann 1993, Lys & Nentwig 1994, Lys et al. 1994, Zangger et al. 1994).

    Study and other actions tested
  4. A controlled study in winter 1990-1991 in one cereal field in central Switzerland (Bürki & Hausammann 1993) found generally more overwintering arthropods in wildflower strips than in the adjacent cereal crop. Five times more beetles (Coleoptera) were recorded in soil samples from the wildflower strips than from the crop (1,032 vs 209 individuals/m2 respectively). Similar patterns were found for samples from photo-eclector traps. Rove beetles (Staphylinidae) and ground beetles (Carabidae) were more abundant in wildflower strips than in the crop, although the greatest abundance of both groups was found in conventional field margins. Other arthropod groups such as spiders (Aranae) and mites (Acari) also had higher densities in wildflower strips than in the crop. More arthropods overwintered in wild plants than in cereal stubbles. Of all arthropods found in cereal stubbles, 48% were found in cereal taken from the wildflower strips, 41% in cereals from conventional field margins and 11% from samples within the crop. Five 1.5 m-wide wildflower margins were established around one cereal field in 1989. The margins were sown with a mixture of wildflower species including clover Trifolium spp. and species from the Brassicaceae family. Overwintering arthropods were sampled from soil cores and photo-eclectors. Vegetation samples of 22 plant species and the cereal stubbles were taken twice a month from November 1990 to April 1991. Arthropods overwintering in the plants were hatched in the laboratory. Note that no statistical analyses were performed on the data presented in this paper. This paper summarizes a large study which is partly published elsewhere, it was performed within the same experimental site as Lys & Nentwig 1992, Lys & Nentwig 1994, Lys et al. 1994, Zangger et al. 1994.

    Study and other actions tested
  5. A replicated study in 1992-1993 in one arable field in Baden-Württemberg, Germany (Engels et al. 1994) recorded 58 species of wild bee (Apidae) either nesting or foraging on wildflower plots (sown with ‘Tübingen’ nectar and pollen mixture), including 11 species of true bumblebee Bombus spp. and five species of cuckoo bumblebee Bombus [Psithyrus] spp. Thirty-five bee species foraged on flowers from the Tübingen wildflower mixture. In total, over 50 herbaceous plant species were recorded in the Tübingen wildflower plots in 1992, and over 60 in Tübingen plots grown over two years in 1993. Ladybirds (Coccinellidae), hoverflies (Syrphidae), green lacewings (Chrysopidae) and butterflies (Lepidoptera) were also observed on the sown strips, including the swallowtail butterfly Papilio machaon. Three strips of the commercially available ‘Tübingen nectar and pollen mixture’ (40% phacelia Phacelia tancetifolia, 25% buckwheat Fagopyron esculentum) were sown at the edge of an arable field. Two strips were sown only in the first year, one strip was sown in both years.

    Study and other actions tested
  6. A randomized, replicated trial from 1987 to 1991 in Oxfordshire, UK (Feber et al. 1994) found that field margins sown with wildflower seed mix had more adult meadow brown butterflies Maniola jurtina but not more larvae than unsown margins in two of the three study years. In 1990 and 1991, sown plots had 4-52 meadow browns/50 m, and unsown plots 4-10 meadow browns/50 m. In all three years, there were more meadow brown butterflies on uncut margins, or margins cut in spring or autumn than in margins cut in summer (sown margins: 4-22 meadow browns/50 m with summer cut, 14-52 meadow browns/50 m without summer cut). There was no difference in the abundance of meadow brown larvae (three larvae/plot on average) between treatments. Two-metre-wide field margins were established around arable fields in October 1987. In 1988 margins were either left to naturally regenerate or sown with a wildflower seed mix (17 wildflower species, six grass species, with a wildflower:grass weight ratio of 1:4). Both treatments were rotavated before sowing. Fifty-metre-long plots were managed in one of the following ways: uncut, cut once in June with hay collected, cut April and June with hay collected, cut in April and September with hay collected, cut April and June with hay left lying (unsown margins only) or sprayed once a year in summer (unsown margins only). There were six replicates of each treatment. Adult meadow brown butterflies were monitored weekly along walked transects in the experimental plots from June to September 1989 and from April to September 1990 and 1991. Meadow brown larvae were sampled in spring 1991, by sweep netting and visual searching. This study is part of the same experimental set-up as Feber et al. 1996, Baines et al. 1998, Bell et al. 1999, Haughton et al. 1999, Smith et al. 1999, Bell et al. 2002, Smith et al. 2010.

    Study and other actions tested
  7. A replicated, controlled, site-comparison study in 1990 in the Kraichgau region, Germany (Gathmann et al. 1994) found set-aside fields newly sown with phacelia Phacelia tanacetifolia attracted many honey bees Apis mellifera (foraging bees not quantified), but no cavity-nesting solitary bees (Apidae) made nests in bundles of reed stems Phragmites australis placed in the phacelia-sown fields. In contrast, 12 bee species nested in reed stems placed in 2-year-old naturally regenerated set-aside fields mown in late June in the same study. Four set-aside fields were sown with phacelia. Bundles of reed stems for cavity-nesting bees (and wasps Sphecidae, Eumenidae) were placed in the four newly sown phacelia set-aside fields in April 1990 and sampled in October 1990. This trial was part of a larger study (Greiler 1994).

    Study and other actions tested
  8. A replicated, controlled study in 1989-1991 in up to 65 arable sites in the Kraichgau region, Germany (Greiler 1994) (same study as Gathmann et al. 1994) found lower plant species richness and invertebrate abundance on phacelia Phacelia tanacetifolia sown set-aside fields than on naturally developed set-asides. Plant species richness was lowest in sown set-asides (10-15 species/49 m2) and cereal fields (10-17spp./49 m2) and highest in orchard meadows (50 spp./49 m2) and naturally developed set-asides (37-45 spp./49 m2). Invertebrate numbers from suction samplers were lowest in phacelia-sown set-asides (500 individuals/5 m2), intermediate in naturally developed set-asides and cereal fields (ca. 1,000 ind./5 m2) and highest in set-asides sown with clover-grass-mixes (1,500 ind./5 m2). Invertebrate numbers caught in Malaise traps were highest in rye fields and clover-grass mixes (around 3,000 ind.) and lowest in naturally developed set-asides (1,000 ind.). Further studies and single species comparisons showed that the effect of field type and set-aside age was strongly species/family-dependent. Up to 11 field types (four to five replicates each) were investigated: one, two and three-year-old naturally developed set-asides (mown and unmown), one-year-old set-asides sown with either phacelia or a clover-grass mix, conventionally managed cereal fields (rye and barley), and low-intensity orchard meadows (<30 years old). Plant surveys (three visits) were conducted in May to October 1990-1991 on one 49 m2 permanent quadrat (meadows and sown fields) or on 120 m2 (systematically changed in naturally developed fields). Insects were sampled on four to five visits in April to October using Malaise traps (20 fields) and suction samplers (61 fields, 3 minute suctions in five 0.25 m2 plots).

    Study and other actions tested
  9. A replicated, controlled study in June-July 1993 in four pairs of winter wheat plots in Hampshire, UK (Holland et al. 1994) found a higher proportion of hoverflies (Syrphidae) with phacelia Phacelia tanacetifolia pollen in their stomachs in plots with sown phacelia strips than in plots without strips. There was no difference in egg production between female hoverflies in plots with and without Phacelia strips. There were non-significant trends of more Aphidius spp. parasitoid wasps, other wasps (Braconidae and Proctotrupoidea) and hoverflies in phacelia strips than in the crop. Four pairs of winter wheat plots (minimum size 5 ha, minimum width 100 m) were either managed according to the Integrated Farming System (IFS) or conventionally. At IFS plots, strips of phacelia were sown along the longest edge (300-400 m) in April 1993. Conventional control plots did not have strips. Invertebrates were sampled either using fluorescent-yellow water traps (19 cm diameter) located at different distances from the phacelia strip/field edge or using a D-Vac. Traps were emptied weekly. Five D-Vac samples were taken once in two plots. Hoverflies were dissected and pollen content in the stomach as well as number of eggs in females recorded. This study system was extended and further studied by Holland & Thomas 1996.

    Study and other actions tested
  10. A replicated, controlled study in the winter of 1990-1991 in four within-field wildflower strips in a cereal field in the Bernese Seeland, Switzerland (Lys & Nentwig 1994) found more than four times higher densities and more than twice as many overwintering species of ground beetles (Carabidae), rove beetles (Staphylinidae) and spiders (Araneae) in sown wildflower strips than in the winter cereal areas between them. The proportion of ground beetle and rove beetle larvae was significantly higher in cereal fields than in the wildflower strips. Four wildflower strips (1.5 m-wide) were sown in 1989, 12, 24 and 36 m apart in one part of a winter cereal field. The remaining area of cereal field was used as a control. Strips were sown with a variety of wild flowering plants and left untreated for three years. Soil samples (diameter 10 cm, depth 20 cm) were taken eight times from December to March in the four strips and the three cereal areas between the wildflower strips. After hand-sorting all samples for arthropods, the samples were extracted in a Berlese apparatus for five days and then hand-sorted again. Beetles were determined to species level, spiders to family level. This study was performed within the same experimental site as Nentwig 1989, Lys & Nentwig 1992, Lys et al. 1994, Zangger et al. 1994.

    Study and other actions tested
  11. A replicated, controlled study in Switzerland in the summers of 1990 and 1991 in one cereal field (Lys et al. 1994) found more ground beetle (Carabidae) species in wildflower strips than in the cereal area between these strips. The number of ground beetles was also higher in the wildflower strips, but only during the first year. Both ground beetle abundance and diversity was higher in the cereal area between the wildflower strips than in the control area of the same field. Ground beetle numbers decreased with increasing distance from the wildflower strips. Many of the ground beetle species that were only found in wildflower strips in 1990 dispersed into the cereal areas in 1991. Four wildflower strips (1.5 m-wide) were studied. Strips were sown in 1989, 12, 24 and 36 m apart in one part of a winter cereal field. The remaining area of cereal field without strips was used as a control. Strips were sown with a variety of wild flowering plants and left untreated for three years. Ground beetles were sampled weekly throughout the summer using pitfall traps (diameter 7 cm) placed in rows in the strips and the cereal field. This study was performed within the same experimental site as Nentwig 1989, Lys & Nentwig 1992, Lys & Nentwig 1994, Zangger et al. 1994.

    Study and other actions tested
  12. A series of four studies from 1991 to 1993 in Hampshire, UK (MacLeod 1994) found the abundance of some hoverfly (Syrphidae) species was higher in areas with sown flower strips than in control sections and hoverflies preferred foraging on certain plant species. Two trials showed hoverflies foraged on sweet alyssum Lobularia maritima, buckwheat Fagopyrum esculentum, coriander Coriandrum sativum, borage Borago officinalis, sunflower Helianthus annuus and dwarf marigold Calendula officinalis compared to other plant species or field margins. A field-scale trial found no difference in the total number of hoverflies, but more marmalade flies Episyrphus balteatus (39 vs 4 individuals/field boundary) in a winter wheat field with a 2 m-wide strip of coriander than a control field. More aphids (Aphidoidea) were found on marked wheat ears in the control field. In 1993 an unreplicated, controlled study found more males and females of three hoverfly species/genera (E. balteatus, Metasyrphus corollae and Eristalis spp.) in a 2 m-wide strip (240 m-long, divided into 0.75 x 10 m plots) sown with 13 plant species (including amaranthus Amaranthus spp., safflower Carthamus tinctorius and quinoa Chenopodium quinoa) in a spring barley field than a control strip on the same field between 7 and 14 July. Plant species used for foraging had small (less than 4 mm diameter) white or yellow flowers and easily accessible anthers and pollen (buckwheat, coriander, gold-of-pleasure Camelina sativa and texsel Brassica carinata). Hoverflies were recorded using transect walks and fluorescent yellow water traps. Ten wheat plants around each yellow water trap were used to count the number of aphids.

    Study and other actions tested
  13. A replicated, controlled study from April to July 1991 in Switzerland in one winter rye field (Zangger et al. 1994) found enhanced feeding and reproductive conditions for the ground beetle Poecilus cupreus (Carabidae) in sown wildflower strips and cereal strips adjacent to wildflower strips, than in a cereal control area in the same field. Male P. cupreus (and females in early season only) had generally higher crop-fullness and satiation in the wildflower strip-managed area than in the control area, indicating higher food availability. Females in the wildflower and cereal strips were generally larger and heavier, and had more ripe eggs in their ovaries (except in May), than females in the control area. Ground beetles were sampled weekly using five pitfall traps (diameter 7 cm) in each of the three habitats (wildflower strip, cereal strip and cereal control area). Captured beetles were dissected to analyse different size and reproductive measures and gut contents. This study was performed within the same experimental site as Nentwig 1989, Lys & Nentwig 1992, Lys & Nentwig 1994, Lys et al. 1994.

    Study and other actions tested
  14. A replicated, controlled study from April to August 1993 in the Kraichgau region, Germany (Steffan-Dewenter & Tscharntke 1995) found that wild bee (Apidae) species richness was lower on set-asides sown with phacelia Phacelia tanacetifolia (around 10 species) than on two-year-old naturally developed set-asides (around 27 spp.) and orchard meadows (around 28 spp.). The number of red-listed bee species and specialist species were lower on phacelia-sown set-asides than on orchard meadows and naturally developed set-asides. Wild bee abundance in phacelia-sown set-asides (around 75 individuals) was lower than in orchard meadows (around 120 ind.) and two-year-old naturally developed set-asides (around 100 ind.), but higher than in naturally developed set-asides of different ages. Seven field types (four replicates each) were investigated in 21 farmland sites: one, two, three, four and five-year-old naturally developed set-asides, one-year-old set-asides sown with phacelia, and orchard meadows. Wild bees were monitored on six 30 minute visits at each site. Bees were caught using sweep nets (100 sweeps/transect) along one 100 m transect in the field centre. Plant species richness and the abundance of flowering plants was recorded at each visit. Additional plant surveys on a 49 m2 quadrat were performed in July and August.

    Study and other actions tested
  15. A randomized, replicated study from 1989 to 1991 on the Oxford University Farm, Oxfordshire, UK (Feber et al. 1996) found that butterfly (Lepidoptera) abundance and species richness were higher in sown wildflower margins (21-91 individuals, 7-10 species) than in unsown, naturally generated margins (14-39 ind., 6-9 spp.) from the second year after establishment. Cutting during summer reduced butterfly diversity and density in the margins, but there were no such effects of cutting in spring and autumn. Cutting in summer also led to an immediate decline in the number of flowering plants directly after the treatment. However, the number of flowers in cut margins had increased by September when it was higher than in uncut margins. Existing field margins (0.5 m-wide) were extended by 1.5 m in October 1987. The extended margins were rotavated and either left to naturally regenerate or sown with a wildflower seed mix in March 1988. Fifty-metre-long plots were managed in one of the following ways: uncut, cut once in summer hay collected, cut spring and summer hay collected, cut spring and autumn hay collected, cut spring and summer hay left lying (unsown margins only), sprayed once a year in summer (unsown margins only). There were eight replicates of each treatment. Butterflies were monitored weekly along transects from June to September 1989 and from April to September 1990 and 1991. Transects were divided into 50 m sections corresponding to the experimental plots. This study was part of the same experimental set-up as Feber et al. 1994, Baines et al. 1998, Bell et al. 1999, Haughton et al. 1999, Smith et al. 1999, Bell et al. 2002, Smith et al. 2010.

    Study and other actions tested
  16. A replicated, controlled study in April-June 1993 in one winter rape field near Bern, Switzerland (Hausammann 1996) found lower numbers of pest species (mainly pollen beetles Meligethes spp. and cabbage weevils Ceutorhynchus spp.) near a sown weed strip than near a field boundary, at least early in the season. There was no difference in the abundance of predators and parasitoids between transects near the weed strip and the boundary. A 1.5 m-wide weed strip was sown with a seed mixture containing 25 varieties of annual, biennial and perennial plant species in the middle of a 3.8 ha winter rape field in spring 1992. The composition of the seed mixture was designed to provide flowering plants over the whole growing season. The strip was not cut or treated for three years. Adult and larval arthropods in the rape field were sampled weekly along transects at 3, 10, 20 and 50 m from the weed strips and the opposite field boundary from April-June 1993 using several different methods (visual counting, sweep netting, dissecting of rape pods and using water traps).

    Study and other actions tested
  17. A replicated, controlled study in June-July 1993 and 1994 using four pairs of adjacent winter wheat plots in Hampshire, UK (Holland & Thomas 1996) found higher numbers of total cereal parasitoids (average 33 vs 5/0.5 m2), gamebird chick food insects (25 vs 2) and parasitic wasps Aphidius spp. (13 vs 12) in phacelia Phacelia tanacetifolia sown strips than in the adjacent crop in one of the study years, 1994. In the same year, the abundance of braconid waps (Braconidae) was higher in plots with a sown phacelia strip (but not in the phacelia strip itself) than in plots without a strip. No significant differences in numbers of any other arthropod group considered were found between phacelia strips and the other habitats. Four pairs of winter wheat plots (minimum size 5 ha, minimum width 100 m) were either managed according to the integrated farming system (IFS) or conventionally managed. At IFS plots, 1 m-wide strips of phacelia were sown (1 g/m2) along the longest edge of the plot (300-400 m) in April each year. Conventional control plots did not have phacelia strips. Invertebrates were sampled either using fluorescent yellow water traps (19 cm diameter) located at different distances into the phacelia strip/field edge or using a D-Vac. Traps were emptied weekly, D-Vac samples (two set-ups) were taken once a year in each plot. Tiller counts were made to assess aphid numbers, species, life-stage and aphid mummies five times yearly in each plot. This study used an extended version of the experimental set-up in Holland et al. 1994.

    Study and other actions tested
  18. A replicated, controlled, randomized study of a sown wildflower margin at a farm in Oxfordshire, UK (Marshall & Nowakowski 1996) found that margin management affected plant species richness. Seventy plant species were recorded in the sown wildflower margin, including 28 of 36 sown species and 42 unsown species. A single cut in June resulted in a significant reduction of sown (2 vs 4 species/quadrat) and unsown species diversity (5 vs 6-8). Grass-specific herbicide did not affect overall species diversity, however sown and unsown grass diversity was reduced and sown and unsown herbaceous plant diversity significantly increased in herbicide-sprayed plots. Unsprayed plots were dominated by black grass Alopecurus myosuroides, however the species was eliminated by a December application of herbicide. Sown crested dogs-tail Cynosurus cristatus was eliminated by a second treatment of herbicide in April, late mowing in June also decreased this species. The wildflower/grass seed mix was sown on 21 contiguous margin plots (3 m wide by 12 m). Plots were grouped into three blocks, within which they randomly received one of seven treatments: unmanaged, cut April, cut April and May, cut May, cut in May and June, cut in June or grass-specific herbicide (fluazifop-P-butyl) application in April. Cuttings were removed. Half of each plot received grass-specific herbicide application in December. Vegetation in sub-plots was sampled in five 0.1 m² quadrats in July 1995.

    Study and other actions tested
  19. A replicated, controlled, randomized study of four field margins in southern and eastern England (West & Marshall 1996) found that plant cover was higher in margins sown with grass or grass/wildflower mixtures than naturally regenerated margins, and diversity tended to be higher with more complex seed mixtures. Percentage plant cover was significantly higher on spring-sown and Breckland autumn-sown grass or grass/wildflower plots than naturally regenerating plots. Plant cover did not differ with seed mixture diversity or management treatment (unmanaged, cut, grass herbicide), although cover tended to be lower on cut plots in the first year. In 1994 plant diversity was higher in plots sown with more complex seed mixtures (32-37 species) than those sown with grass-only (22-27) or naturally regenerated (21-25). In 1995, grass-seed-only plots tended to be the least diverse (15-21 species), but naturally regenerated plots (18-28) were as diverse as some complex seed mixtures (23-31). Species diversity did not differ between management treatments. Margins were created in each field and divided into six plots (4 x 30 m). Each was (randomly) sown with a seed mixture: grass, low cost mix (3 grass: 7 wildflower species), alkaline soil mix (6: 16), neutral soil mix (5: 15), acid soil mix (6: 16) and one naturally regenerated treatment. Plots were divided into 10 m sub-plots, which were either unmanaged, cut once, or treated with grass-specific herbicide. Plants were sampled in each sub-plot in summer 1994-1995.

    Study and other actions tested
  20. A replicated study in 1994 and 1995 in Hertfordshire, UK (Carreck & Williams 1997) found that plots sown with two commercial nectar and pollen seed mixtures, Tübingen mixture (40% phacelia Phacelia tanacetifolia) and Ascot Linde mixture (25% phacelia) attracted 14 species of bees/wasp (Hymenoptera), including all six common UK bumblebee (Bombus spp.) species and three cuckoo bumblebee species Bombus [Psithyrus] spp. across two years. A small number of solitary bees (Andrenidae, Megachilidae) of three species (no more than two individuals on any plot) were recorded. The plots also attracted 14 hoverfly (Syrphidae) species and six butterfly (Lepidoptera) species. Phacelia attracted 87-99% of all bee visits and 31-98% of all hoverfly visits over the two years. Buckwheat Fagopyrum esculentum, a nectar source that comprised 20% of both seed mixtures by weight, attracted 1% or less of all bee visits, but up to 36% of hoverfly visits. Phacelia flowered for a long period on all plots. The main flowering period lasted four weeks, but some flowering continued for several months afterwards. The sown species successfully competed with previously existing weeds. In April 1994 Tübingen mixture was sown on a 25 x 25 m plot. In 1995 both mixtures were sown on 19 x 14 m plots; Tübingen mixture sown in April and May, Ascot Linde mixture sown in May and June. In each plot plants and flowers were counted in four 1 m2 quadrats. Insect density and diversity were recorded at least three times a week/plot.

    Study and other actions tested
  21. A replicated, controlled study in 1995-1996 in Cambridgeshire, UK (Clarke et al. 1997) found a set-aside strip sown with a mix of 11 wildflower species (‘Tübinger Mischung’ or ‘bee mixture’) attracted more birds (average 45-131 individuals) than strips sown with three different grass mixtures (18-121 individuals) or a grass and wildflower mixture (33-100 individuals). However the ‘bee mixture’ attracted the lowest number of bird species (8-15 species). Strips sown with a grass and wildflower mixture attracted more bird species (16-25 species) than the bee mixture, but fewer species than strips sown with a diverse grass mixture (23-33 species). Most of the yellowhammers Emberiza citrinella recorded in the study were found on the bee mixture strip. No statistical analyses were performed on these data. Five seed mixtures were sown on 15 set-aside areas (minimum 20 x 100 m) on one farm in autumn 1993 and 1994. Only one strip was sown with the bee mixture, three to four strips were sown for all other set-aside strips. Seed mixtures contained: only grass species (three mixes of three to six species), mix of grasses and wildflowers (six grass and eight wildflower species) or only wildflowers (11 species). Birds were recorded on ten 15 minute point counts between June and September 1995 and July and October 1996. Individual bird locations were recorded in three categories: field boundary, set-aside strip or crop. After each count, the strips were walked to flush any birds present but not visible during the count.

    Study and other actions tested
  22. A replicated, controlled study in summer 1995 near Göttingen, Germany (Denys 1997) found higher arthropod species richness on potted mugwort Artemisia vulgaris plants placed in sown wildflower strips compared to the cereal field, but not compared to other margin types. The predator-prey ratio in wildflower strips did not differ from the control winter wheat field but was significantly lower than in a 6-year-old uncultivated field margin. The effect of wildflower strips on numbers of individual arthropod species varied between species, with some species (e.g. the aphid Macrosiphoniella oblonga and the fruit fly Oxyna parietina), but not all found in higher numbers in wildflower strips than in the control. Two types of wildflower strip were sown with either a wildflower seed mixture or a phacelia Phacelia tanacetifolia mix. Other margin types were one-year-old and six-year-old uncultivated margins and cereal strips. There were four replicates of each margin type. Potted mugwort plants (four pots) were placed in all margin types and the control. All herbivores and their predators on the plants were recorded during six visits in June and July. In September, all mugwort plants were dissected to assess numbers of arthropods feeding inside the plants. Results from the same study are also presented in Denys et al. 1997, Denys & Tscharntke 2002.

    Study and other actions tested
  23. A replicated, controlled study in summer 1995 near Göttingen, Germany (Denys et al. 1997) found higher species richness of plants and arthropods colonizing potted mugwort Artemisia vulgaris and red clover Trifolium pratense plants in wildflower strips than in unsprayed cereal control edges. However, the number of arthropod species on mugwort did not differ between any of the other established margin types. The number of arthropod species colonizing red clover flower heads decreased significantly with increasing distance from wildflower strips into adjacent cereal fields, but no such decline was found for individual numbers. Two types of wildflower strips were sown either with a wildflower seed mixture (19 species) or phacelia mixture (Phacelia tanacetifolia plus three species). Other margin types were one-year-old and six-year-old naturally regenerated margins and cereal strips. Potted mugwort (four pots) and red clover (three pots) plants were placed in all margin types and the controls. Mugwort plants were visited six times in June and July to count all herbivores and their predators on the plants before being taken to the lab in September to assess all arthropods feeding inside the plants. Red clover flower heads were collected five times in June-July and dissected for arthropods living inside the plants. Vegetation of all margins was surveyed in June. Results from the same study are also presented in Denys 1997, Denys & Tscharntke 2002.

    Study and other actions tested
  24. A replicated study in 1994-1996 in Gloucestershire, UK (Feber & Hopkins 1997) found higher plant species richness (23 vs 19 species) as well as higher abundance and diversity of butterflies (Lepidoptera) in sown wildflower margins than in naturally regenerated margins. Cutting and subsequent grazing of the sown margins significantly decreased butterfly diversity (5.6 vs 6.8 species) but not abundance (14.6 vs 16.3 individuals). Margins were established around two organically-managed arable fields by either sowing a seed mix (containing five grasses, six wildflowers) or by natural regeneration in 1994. In 1996 part of the margins were cut in June and grazed in July. The rest was left untreated. Butterflies were monitored along transects weekly from May to September 1996. Plant species and flower abundance were recorded in May and September 1996.

    Study and other actions tested
  25. A small replicated, controlled study in Switzerland (Frank 1997) found that ground beetle (Carabidae) species richness was not significantly higher in sown weed strips than in adjacent crops, but ground beetle species richness decreased with distance from the strips. The oldest weed strip (two years-old) contained the highest number of ground beetle species (10 species/trap), followed by the adjacent rape field (9/trap) and one of the one-year-old weed strips (9-10/trap), although the differences were not significant. The other one-year-old weed strip had 8 species/trap and other crops 6-8/trap. In 1992, numbers of ground beetle species in rape and wheat plots decreased with distance from weed strips (15% and 35% decreases respectively). Weed strips contained similar numbers of species in their first and second year. Three to five species were found only in the strips. Strips were sown with 25 weed species and were one (two strips) or two years old (one strip), they were not mown. Ground beetles were sampled using four pitfall traps/site, emptied every 14 days from April-September 1992 and 1993.

    Study and other actions tested
  26. A replicated trial from 1994 to 1996 in central Germany (Gathmann & Tscharntke 1997) found that few solitary bee and wasp (Hymenoptera: Aculeata) species occupied reed Phragmites australis stem nest boxes in set-aside fields sown with a clover-grass mix relative to nest boxes placed in semi-natural grasslands (quantitative details are lacking from the report of this trial). Three replicates in each of five habitat types were studied: set-aside fields (sown with clover-grass mixture), sown field margin strips, extensively-managed grassland, chalk grassland, orchard meadows. Ten reed stem nest boxes were placed in each site. In autumn, nests were dissected and occupants identified. This study is part of the same study set-up as Gathmann & Tscharntke 2000.

    Study and other actions tested
  27. A replicated trial in 1995 near Wageningen in the Netherlands (Kleijn et al. 1997) found that 4 m-wide field margins planted with wildflowers had more plant species than margins left to naturally regenerate or sown with rye grass Lolium perenne two years after establishment. On average there were 13.7 plant species/0.25 m2 in wildflower margins, 8.6 in naturally regenerated margins and 5.9 in grass-sown margins. There were fewer plant species in the 1 m of wildflower margin closest to the arable field (11-12 species/0.25 m2) than in plots more than 1.5 m away (14-17 species): this pattern was not observed in other treatments. Two prominent arable weeds, creeping thistle Cirsium arvense and couch grass Elymus repens, both had lower biomass in wildflower-sown margins than in naturally regenerated margins (0.1 g/m2 and 6 g/m2 respectively in wildflower-sown margins, 33 g/m2 and 28 g/m2 in naturally regenerated plots). Wildflower-sown margins had similar couch grass biomass to the grass-sown plots, but much lower creeping thistle biomass (8 g/m2 of creeping thistle in grass margins). In 1993, 27 experimental plots (8 x 4 m) were established on the boundaries of three arable fields. Wildflower plots were sown with 30 broadleaved (non-grass) wildflower species. All plots were mown once a year, without removing cuttings. There were three replicates of each treatment on each field. Plant biomass and species richness were measured in eight 0.5 x 0.5 m plots along a single transect across each margin in August 1995.

    Study and other actions tested
  28. A replicated, controlled study in 1994-1996 near Hannover, Germany (Lemke & Poehling 1997) found higher numbers of spider (Araneae) species and individuals (peak 435 individuals/m­2) in sown wildflower strips than in cereal fields. Spider abundance varied throughout the year. Spider species richness increased from the first to the third year following margin establishment. Abundance and population dynamics of aphids (Aphididae) on wheat tillers differed between the years (peak 7.5 aphids/tiller), but abundance generally increased with increasing distance from the wildflower strips. Note that no statistical tests were presented in this study. Two wildflower strips (1.5 x 30 m) were sown in two different winter wheat fields in 1994, with a seed mixture containing 19 non-grass plant species. The strips were cut annually after harvest. Spiders were sampled with a D-Vac both in the strips and in the fields at defined distances from the strip. Aphid numbers were assessed using tiller counts at the same sample sites.

    Study and other actions tested
  29. A replicated study in summer 1996 in central Germany (Weiss & Buchs 1997) found that both species richness and abundance of spiders (Araneae) caught in sown wildflower strips depended greatly on the species composition of the seed mixtures used. Highest species richness was reported in plots containing phacelia Phacelia tanacetifolia and Egyptian clover Trifolium alexandrinum (40 spider species) and lowest diversity (30 species) in plots with phacelia, buckwheat Fagopyrum esculentum, common sunflower Helianthus annuus and common mallow Malva sylvestris. Spider abundance was highest in plots containing sundial lupin Lupinus perenne and common vetch Vicia sativa in both pitfall traps and photoeclectors (155/124 individuals), significantly higher than in both naturally regenerated plots (97/56 individuals) and plots with fodder radish Raphanus sativus oleiferus (104/49 individuals). Note that most results in this study are not statistically tested. Eight different types of strip with three replicates each were tested: six seed mixtures contained mainly flowering plants (1-12 species), one mixture contained mainly grass seeds (two species plus white clover T. repens) and one naturally regenerated treatment. Spiders were sampled using two pitfall traps and two photoeclectors in each plot.

    Study and other actions tested
  30. A randomized, replicated trial from 1987 to 1991 in Oxfordshire, UK (Baines et al. 1998) found that field margins sown with a wildflower seed mix had more spiders (Araneae), but not more spider species, than naturally regenerated margins on all dates. Cutting, especially summer cutting, significantly reduced the abundance of spiders. Two-metre-wide field margins were established around arable fields in October 1987. They were either left to naturally regenerate or sown with a wildflower seed mix (17 wildflower species, six grass species, with a wildflower:grass weight ratio of 1:4) in March 1988. Both treatments were rotavated before sowing. Fifty-metre-long plots were managed in one of the following ways: uncut, cut once in June with hay collected, cut April and June with hay collected, cut in April and September with hay collected. There were six replicates of each treatment. Spiders were sampled using a suction trap (D-Vac) in September 1987 and 1988, and in May, July and September in 1989, 1990 and 1991. This study was part of the same study set-up as Feber et al. 1994, Feber et al. 1996, Bell et al. 1999, Haughton et al. 1999, Smith et al. 1999, Bell et al. 2002, Smith et al. 2010.

    Study and other actions tested
  31. A randomized, replicated trial from 1987 to 1996 in Oxfordshire, UK (Bell et al. 1999), found no difference in the number of pseudoscorpions (Pseudoscorpionida) between naturally regenerated field margins and those sown with a wildflower mix. More pseudoscorpions (Chthonius ischnocheles and C. orthodactylus) were found in unmanaged field margin plots (95 pseudoscorpions in total on sown and unsown plots) than in cut treatments (19-53 pseudoscorpions). Plots cut in spring and summer had fewer pseudoscorpions than other margins (19 pseudoscorpions on sown and unsown plots). Plots cut just once in June or cut twice but not in June had intermediate numbers of pseudoscorpions (29 and 53 pseudoscorpions respectively). Pseudoscorpions were sampled from the litter layer (not the soil) using a suction trap (D-Vac) in May, July and September 1995 and 1996. This study was part of the same study set-up as Feber et al. 1994, Feber et al. 1996, Baines et al. 1998, Haughton et al. 1999, Smith et al. 1999, Bell et al. 2002, Smith et al. 2010.

    Study and other actions tested
  32. Two replicated trials from 1995 to 1998 in Hertfordshire and Hampshire, UK (Carreck et al. 1999) monitored flower-visiting insects on sown flower strips. One trial (Hampshire 1995-1998) found more flower-visiting insect species and plant species on strips sown with a wildflower mix than on a naturally regenerated margin or a margin sown with wild bird cover mix in 1998. One trial (Hertfordshire 1996-1997) found plots sown with six annual plant species were visited by 39 invertebrate species (including bees Apidae, flies Diptera and butterflies Lepidoptera) in the summers after sowing. Wildflower strips attracted 24 invertebrate species, compared to 14 and 19 species on the wild bird strip and naturally regenerated strip respectively. There were 24 flowering plant species on the wildflower strips, compared to 20 and 16 on the wild bird strip and naturally regenerated strip (Hampshire). Five plant species attracted many insects or species: wild carrot Daucus carota, black knapweed Centaurea nigra, oxeye daisy Leucanthemum vulgare, bird’s-foot trefoil Lotus corniculatus and black medick Medicago lupulina. Butterflies only visited phacelia Phacelia tanacetifolia, borage Borago officinalis and marigold Calendula officinalis out of six plant species sown in the Hertfordshire study. Short-tongued bumblebees, buff-tailed Bombus terrestris/lucorum and red-tailed bumblebees B. lapidarius/ruderarius, were the most abundant wild bee visitors, and bees were most numerous on phacelia, borage and (second year only) cornflower Centaurea cyanus. Five field margin strips were established in the Hampshire study in 1995, three sown with perennial grass and wildflower mix, one with wild bird mix, one naturally regenerated. In the Hertfordshire study, four plots were sown with six annual plant species in 1996 and 1997. In both studies, the number of flowers, flower-visiting bees, wasps (Hymenoptera), flies and butterflies were counted (monthly from May-August 1998 in Hampshire study, several times a week in Hertfordshire study). The Hertfordshire study was part of the same study as (Carreck & Williams 2002).

    Study and other actions tested
  33. A 1999 review of research into field margins in north-west Europe (de Snoo & Chaney 1999) found that numbers of invertebrates and small mammals increased with the establishment of wildflower margins. Three studies reported that 1-1.5 m-wide flower strips resulted in higher numbers of invertebrates within the strips and field as a whole (Klinger 1987, Lys & Nentwig 1992, Nentwig 1992). One study in Switzerland found that 3 m strips were used intensively by small mammals and resulted in a population increase of common shrew Sorex araneus in spring and summer (Baumann 1996).

    Additional references:

    Klinger K. (1987) Auswirkungen eingesäter Randstreifen an einem Winterweizen-Feld auf Raubarthropodenfauna und den Getreideblattlausbefall. Journal of Applied Entomology, 104, 47-58.

    Nentwig W. (1992) Die nühtzlingsfördernde Wirkung von Unkrautern in angesäten Unkrautstreifen. [Augmentation of beneficial arthropods by sown weed strips in agricultural areas] Z. Pflanzenkrankheiten and Pflanzenschuzt, Sonderheft, 33, 33-40.

    Baumann L. (1996) The influence of field margins on populations of small mammals – a study of the population ecology of the common vole (Microtus arvalis) in sown weed strips. Field Margin Newsletter, 6/7, 26-33.

    Study and other actions tested
  34. A replicated, randomized study in Oxfordshire, UK (Haughton et al. 1999) found that from 1995 to 1996 total numbers of invertebrates and leafhoppers (Auchenorrhyncha) were significantly higher in sown wildflower margins than in unsown, naturally regenerated margins. Cut plots (cut in summer alone, spring and summer or spring and autumn) had significantly lower numbers of all invertebrates, spiders (Araneae), true bugs (Heteroptera) and leafhoppers than uncut plots in all seasons, apart from spiders and true bugs in May. Numbers of all invertebrates were significantly higher in treatments cut twice a year than annually. Cutting in spring and autumn resulted in higher numbers of invertebrates. The abundance of spiders was significantly higher in plots cut bi-annually in spring and autumn than in spring and summer (in July and September samples). Existing field margins (0.5 m wide) were extended by 1.5 m in October 1987. These were rotavated and left to naturally regenerate or sown with a wildflower seed mix. Six management treatments were applied with six replicates in a randomized block design on fifty metre-long plots: uncut, cut once in summer, cut spring and summer, cut spring and autumn, cut spring and summer (hay left lying), sprayed once a year in summer. Invertebrates were sampled using a D-Vac suction sampler at 10 m intervals along each plot in May, July and September in 1995-1996. This study is part of the same study design as Feber et al. 1994, Feber et al. 1996, Baines et al. 1998, Bell et al. 1999, Smith et al. 1999, Bell et al. 2002, Smith et al. 2010.

    Study and other actions tested
  35. A 1999 review of literature (Kromp 1999) found four experimental studies (Bürki & Hausammann 1993, Lys & Nentwig 1994, Zangger et al. 1994, Frank 1997) have found higher numbers or species diversity of ground beetles (Carabidae) in sown wildflower strips in cereal fields

    Study and other actions tested
  36. A replicated, randomized study from 1987 to 1992 in Oxfordshire, UK (Smith et al. 1999) found that species richness and abundance of sown plant species were higher in 1.5 m-wide extensions to margins than the original margin sections (0.5 m-wide). Species richness of sown wildflowers was significantly higher in new compared to old sections (3.6-6.3 vs 0.1-0.9/species quadrat), frequencies of species showed the same pattern. After three years, the original margin sections only had 20% of the species found in the new margins. Cutting in spring and autumn increased the number of species (6-7/quadrat), whereas under other treatments numbers declined sharply in the first year after sowing (from 6 to 4 species) and remained significantly lower (uncut: 4, summer cut: 3-4, spring/summer cut, 4-5). There was no significant difference between numbers in margins cut once or uncut. Individual species showed a range of responses to cutting regimes. Plants were sampled in three permanent quadrats (50 x 100 m) at 10 m intervals in existing and new sections of margins. Relative frequencies were recorded as presence/absence in eight sub-sections of the quadrat four times/year from July-September. This study is part of the same study design as Feber et al. 1994, Feber et al. 1996, Baines et al. 1998, Bell et al. 1999, Haughton et al. 1999, Bell et al. 2002, Smith et al. 2010.

    Study and other actions tested
  37. A randomized, replicated controlled trial from 1993 to 1996 near Bristol, UK (Thomas & Marshall 1999) found that 4 m-wide field margins sown with a nectar flower mixture had more suction-sampled invertebrates, but not more ground beetles (Carabidae), than control cropped margins or margins sown with grass. There were around 200 invertebrates/sample on margins sown with a wildflower/grass mix and naturally regenerated margins, compared to 110-130 invertebrates/sample on control or grass-sown plots. Wolf spiders (Lycosidae) were more abundant on grass and wildflower-sown margins than on control or naturally regenerated margins (numbers not given). There was no difference in the number of ground beetle species (average 8 species/plot), nor in the numbers of the four most commonly caught ground beetle species, between margin types. In a 2 m-wide margin, there were more over-wintering invertebrates in the soil of the wildflower-sown half than the naturally regenerating half, but this difference was not found in 4 m-wide replicated experimental plots. Three field margins were established in spring 1993. Experimental plots 10 x 4 m were either sown with arable crop (control), rye grass Lolium perenne or a wildflower and grass seed mix, or left to naturally regenerate. There were three replicate plots in each margin. All plots were cut annually after harvest, and cuttings left in place. Another 100 x 2 m wide field margin, 50 m sown with a wildflower mix and 50 m unsown, was used to monitor wintering invertebrates. Ground beetles were sampled in eight pitfall traps in or near each margin, for one week in June for four years. Invertebrates were sampled using a vacuum sampler on plots in two of the three margins in June 1994. Arthropods were extracted from soil samples taken from plots in two margins in December 1993 and February 1994.

    Study and other actions tested
  38. A replicated study in 1994-1996 near Göttingen, Germany (Gathmann & Tscharntke 2000) found no significant differences in the body mass and sex ratio of red mason bees Osmia rufa in sown wildflower strips on set-aside land compared to field margins (mostly naturally regenerated) and three types of grassland. Overall, female body mass was correlated with flower availability. Sex ratio was correlated with female body mass, relatively more female larvae were found in habitats with large females. Ten artificial nesting aids for solitary bees were placed in five arable habitats (set-asides sown with wildflower seed mixes, mostly naturally regenerated field margins, extensive bio-dynamic grasslands, chalky grasslands and orchard meadows). There were three replicates in each habitat type. Unparasitized cocoons of Osmia rufa were weighed and sex determined in the lab. This study is part of the same study set-up as Gathmann & Tscharntke 1997.

    Study and other actions tested
  39. A 2000 literature review (Holland & Luff 2000) looked at which agricultural practices can be altered to benefit ground beetles (Carabidae). It found four studies (Nentwig 1989, Lys & Nentwig 1992, Lys & Nentwig 1994, Zangger et al. 1994), showing that wildflower strips increased ground beetle numbers in adjacent cereal fields.

    Study and other actions tested
  40. A replicated, controlled study in summer 1997 and 1998 in Switzerland (Jeanneret et al. 2000) found that the number of spider (Araneae) and butterfly (Lepidoptera) species in wildflower strips sown on set-aside areas did not differ significantly from winter wheat fields. The number of spider species in wildflower strips was among the lowest reported (average 20 species), significantly lower than in low-intensity meadows (more than 25 species) and forest edges (more than 35 species). Butterfly species richness in the wildflower strips (more than 6 species) differed significantly only from forest edges (less than 4 species). However, for both taxa, wildflower strips attracted some species that were never or only rarely found in other habitats. The investigated habitat types were forest edges, arable fields (winter wheat and intensively managed meadow) and ecological compensation areas including hedgerows, extensively managed and low-intensity meadows, wildflower strips on set-aside land and orchard meadows. There were 109 sites in two arable regions. Spiders were collected in pitfall traps in May and June 1997. Butterflies were observed during six visits (10 minutes, covering 0.25 ha) in each site in 1998. This was part of the same study as Jeanneret et al. 2003.

    Study and other actions tested
  41. A replicated, controlled study in winter 1995-1996 in northwest Switzerland (Pfiffner & Luka 2000) found significantly higher abundances of arthropods in sown wildflower strips than in adjacent arable habitats in two of three paired sites on two arable farms. Species numbers were generally higher in the wildflower strips but this was not statistically tested. Many of the most frequent arthropod species were pest predators, e.g. rove beetles (Staphylinidae), ground beetles (Carabidae) and spiders (Araneae). Two of the wildflower strips (5 years old, 4-5 m-wide) were paired to winter wheat fields on an integrated farm. Both strips were sown with grass-clover mixtures and an additional 14 wildflower species and cut 2-3 times a year. The third wildflower strip (2 years old, 3 m-wide) was on an organic farm and paired to a ploughed strip (formerly wildflower strip, 6 months old). In each habitat, 24 soil samples were taken three times during the study period using a soil borer (25 cm depth, 8 cm diameter). In arable habitats, soil samples were taken 30 m parallel to the field margins.

    Study and other actions tested
  42. The initial findings of a controlled replicated site comparison study of the Swiss Ecological Compensation Areas scheme in 1999-2005 in Switzerland (Herzog et al. 2001) found more ground beetles (Carabidae) and ground beetle species in wildflower strips than in adjacent arable crops. The same was true for ground beetle species with specific habitat requirements. Ground beetles were sampled using funnel pitfall traps on 11 wildflower strips and comparison crop strips. Plants, ground beetles, spiders (Araneae), butterflies (Lepidoptera), grasshoppers (Orthoptera) and breeding birds were monitored on grasslands in three case study areas of around 5 km2.

    Study and other actions tested
  43. A replicated, controlled study in April to July 1999 in Witzwil, Switzerland (Luka et al. 2001) found higher species richness of ground beetles (Carabidae) and spiders (Araneae) in three sown wildflower strips than in adjacent crop fields. However, the number of individuals was generally lower in strips than in the crop. Abundance and species richness of specialist species was clearly higher in strips than crop fields for both ground beetles and spiders. On some of the sites, species richness (especially for spiders) appeared higher in the edge samples of the crop fields than in the centre. Note that no statistical analyses were performed on the data in this study. Wildflower strips (two, five and six-years-old) were sown with a mixture of 38 native wild and cultivated plant species and were not managed during the sample period. The crop fields (winter wheat, summer wheat and rye) were treated with pesticides in autumn 1998 and spring 1999. Ground beetles and spiders were caught in pitfall traps. Four traps each were placed in the wildflower strip. In the crop field, traps were placed 15 and 70–100 m from the wildflower strip.

    Study and other actions tested
  44. A site comparison study in 1996 in Wiltshire, UK (Moonen & Marshall 2001) found that coppiced and gapped-up hedges had higher plant diversity than those with adjacent sown grass and grass/wildflower strips. Hedges with adjacent sown strips had lower abundances of pernicious weed species. Sixty hedgerows on two neighbouring arable farms were studied. All 23 sampled hedges on Noland’s Farm were trimmed annually and had the vegetation at the hedge base cut. The 37 sampled hedges on Manor Farm were trimmed in alternate years, and nine were coppiced and gapped-up. Hedge vegetation was assessed in 25 m-long plots in the middle of a field edge, on both sides of each hedge in June.

    Study and other actions tested
  45. A replicated study in 1993 in Germany (Steffan-Dewenter & Tscharntke 2001) found that one-year-old set-aside fields sown with phacelia Phacelia tanacetifolia had similar numbers of bees (Apidae), but fewer bee species (13 species/field on average), than one- to five-year-old naturally regenerated set-aside fields (15-29 species/field). Bees found on phacelia were mainly common species of bumblebee Bombus spp. and the solitary bee genus Lasioglossum spp., whereas several endangered and specialized bees were found foraging on naturally regenerated set-aside. The percentage cover of flowers did not differ between ages of set-aside, but was higher (more than 25% cover of flowers) in phacelia-sown set-aside fields and old meadows than on naturally regenerated set-aside (around 10%). The following field types were studied: one- to five-year-old naturally regenerated set-aside fields, one-year-old phacelia-sown set-aside fields, >30-year-old orchard meadows. There were four replicates of each field type, with a total of 28 sites. Fields were set-aside after harvest in autumn and mown in July. Orchard meadows were mown once or twice in June-August. Between May and August percent flower cover was estimated on five occasions. Bee species and the flowering plant species visited by the bees were surveyed six times between April and August on 30 minute transects in each field.

    Study and other actions tested
  46. A randomized, replicated trial from 1987 to 1996 in Oxfordshire, UK (Bell et al. 2002) found no difference in numbers of the predatory sheet web spider Lepthyphantes tenuis between field margins sown with a wildflower mix and naturally regenerated margins. In September, when most of the spiders were caught, there were significantly fewer L. tenuis individuals in margins (sown and unsown) that were cut in June (around 10 individuals/m2) compared to more than 15/m2 in plots cut in spring and autumn, or not cut. In May and July, plots with a recent cut (April- or June-cut treatments respectively) also had lower numbers of L. tenuis than other plots. L. tenuis individuals were counted in invertebrate samples collected using a suction trap (D-Vac) in May, July and September 1990, 1991, 1995 and 1996. This was part of the same study design as Feber et al. 1994, Feber et al. 1996, Baines et al. 1998, Bell et al. 1999, Haughton et al. 1999, Smith et al. 1999, Smith et al. 2010.

    Study and other actions tested
  47. A replicated study in 1996 and 1997 in Hertfordshire, UK (Carreck & Williams 2002) (same study as Carreck et al. 1999) found that plots sown sequentially from mid-April to mid-July with a mix of six annual flowering species (cornflower Centaurea cyanus, common mallow Malva sylvestris (both native), borage Borago officinalis, buckwheat Fagopyrum esculentum, marigold Calendula officinalis and phacelia Phacelia tanacetifolia) provided continuous forage for pollinators from mid-June to mid-November. The mix attracted 15 bee (Apidae), 17 fly (Diptera) and six butterfly (Lepidoptera) species and the common wasp Vespula vulgaris. The most numerous insects were the honey bee Apis mellifera and red-tailed bumblebee Bombus lapidarius/B. ruderarius (not distinguished in the study). Abundance of flies varied over the season while abundance of butterflies was low. Butterflies and bumblebees Bombus spp. preferred borage and phacelia, while solitary bees and flies preferred marigold. Mallow and buckwheat did not contribute much to flower density or pollinator diversity. Four plots (22 x 14 m or 20 x 13 m) were sown each year (91 or 22 kg/ha) at monthly intervals, then harrowed and irrigated as necessary. Flower density was recorded weekly in four random 1 m2 quadrats in each plot. Pollinators were recorded in the outer 3 m of each plot on 21-34 days from mid/end of-June to end of October/beginning of November.

    Study and other actions tested
  48. A replicated study from April to September 1995 near Göttingen, Germany (Denys & Tscharntke 2002) (same study as Denys 1997, Denys et al. 1997) found that sown wildflower strips had higher plant species richness and could suppress the abundance of aggressive arable weeds. However, arthropod species richness and abundance in wildflower strips did not differ from the other margin types. Both the abundance and species richness of arthropods found on red clover Trifolium pratense plants in wheat fields decreased with increasing distance from the margins, however the decrease in abundance was less pronounced in fields with sown wildflower strips where dispersal from the margin into the field was higher than for control margins. Five margin types (3 m wide, 100-150 m long) around four cereal fields were studied: sown with a mixture of 19 wild flower species, sown with a phacelia Phacelia tanacetifolia mixture, one-year-old naturally developed, six-year-old naturally developed, control strips sown with winter wheat or oats. Potted plants of mugwort Artemisia vulgaris (four pots/margin) and red clover (three pots/margin) were used to study plant-arthropod communities. Red clover pots were also arranged in cereal fields at 4, 8 and 12 m from wildflower strips to assess dispersal. Mugwort pots were set out in May and visited weekly to count all arthropods feeding inside the plants, leaf miners and galls. In September, the plants were dissected and all larvae and pupae found inside the plants were individually reared in the lab to estimate parasitization rates. Red clover pots were set out in April. At five visits in June and July, flower heads were sampled, dissected and larvae and pupae found inside the plants were reared in the lab for species determination.

    Study and other actions tested
  49. A 2002 review (Evans et al. 2002) of two reports (Wilson et al. 2000, ADAS 2001) evaluating the effects of the Pilot Arable Stewardship Scheme in two regions (East Anglia and the West Midlands) from 1998 to 2003 found that ‘wildlife seed mix’ benefited plants, bumblebees Bombus spp., bugs (Hemiptera) and sawflies (Symphyta), but not ground beetles (Coleoptera). The wildlife seed mix option could be nectar and pollen mix for pollinators or wild bird seed mix, and the review does not distinguish between these. The effects of the pilot scheme on plants, invertebrates (bumblebees, true bugs, ground beetles, sawflies) and birds were monitored over three years, relative to control areas, or control farms. Only plants and invertebrates were measured within individual options. Wildlife seed mix was the least widely implemented option, with total areas of 106 and 152 ha in East Anglia and the West Midlands respectively.

    Additional references:

    Wilson S., Baylis M., Sherrott A. & Howe G. (2000) Arable Stewardship Project Officer Review. F. a. R. C. Agency report.

    ADAS (2001) Ecological evaluation of the Arable Stewardship Pilot Scheme, 1998-2000. ADAS report.

    Study and other actions tested
  50. A replicated trial in Switzerland from 1996 to 1999 (Fluri & Frick 2002) found 26 honey bees Apis mellifera/m2 and 0.2 bumblebees Bombus spp./m2 foraging on a plot sown with phacelia Phacelia tanacetifolia in 1996. Two plots sown with 50% white clover Trifolium repens, one in each of 1998 and 1999, had 1.7 and 3.9 foraging honey bees/m2 respectively. All three plots were located on one trial farm, plots measured approximately 0.3 ha. Five to six honey bee colonies were established adjacent to the plots several days before surveying.

    Study and other actions tested
  51. A small-scale replicated, controlled trial in summer 2000 in North Yorkshire, UK (Meek et al. 2002) found significantly more bumblebees Bombus spp. and butterflies (Lepidoptera) on four 6 m-wide margins sown with a grass and wildflower seed mix than on four naturally regenerated, grass-sown or control cropped margins. Spring numbers of ground beetles (Carabidae) and ground-dwelling spiders (Araneae) were higher in all treatments compared with cropped margins. Margins sown with a grass and wildflower mix harboured more pollen beetles Meligethes spp. than naturally regenerated margins. Plant diversity was higher in margins sown with a grass and wildflower mix. Four margins of winter cereal fields (all adjacent to hedges) on two farms were split into 72 m-long plots and sown in September 1999 with either a grass mix, a grass and wildflower mix, cereal crop or left to regenerate naturally. Ground and canopy-dwelling invertebrates, bumblebees, butterflies and plants were surveyed from late April to late September 2000 using pitfall traps, sweep netting, transects and quadrats.

    Study and other actions tested
  52. A small-scale study in 1994 in Vestby, Norway (Dramstad et al. 2003) found the number of bumblebees Bombus spp. visiting a single 2 x 210 m strip sown with phacelia Phacelia tanacetifolia peaked at 237 individuals (0.6 bumblebees/m2) on 17 July, and gradually declined to 93 bumblebees (0.2/m2) on 28 July. Maximum numbers of honey bees Apis spp. foraging on the phacelia strip were recorded on 14 July with 3739 honey bees (9.0/m2), honey bee abundance declined steadily after 18 July with the lowest numbers recorded on 28 July (22 honey bees). The strip was sown in May 1994 along the boundary of a cereal field and a ‘habitat island’ (area of semi-natural habitat within the farmed landscape). Bees were surveyed over a three week period (5-28 July).

    Study and other actions tested
  53. A site comparison study in 1997 and 1998 in the region of Rafz, Switzerland (Jeanneret et al. 2003) (part of the same study as Jeanneret et al. 2000) found that butterfly (Lepidoptera) species richness was significantly higher in wildflower strips planted as Ecological Compensation Areas than in intensively managed wheat fields. Eleven wildflower strips and 20 wheat fields were sampled. Butterflies were observed for ten minute periods on 0.25 ha of each site, on five occasions from May to August 1998, between 10:00 and 17:30 h on sunny days with temperatures of at least 18 °C.

    Study and other actions tested
  54. A replicated trial in 2001-2002 at three sites in the UK (May & Nowakowski 2003) found that margins of sugar beet Beta vulgaris fields sown with wildflowers had more plant species, but not more invertebrates (individuals or species) than margins sown with grasses, crops, or margins left to naturally regenerate. Wildflower margins had 35 plant species/m, compared to around 17 spp./m in naturally regenerated margins, 15 spp./m in grass margins and 6-11 spp./m in barley Hordeum vulgare or beet margins. Fertilized wildflower-sown margins, tested at two sites, had fewer species than those without fertilizer, around 30 plant species/m. Naturally regenerated margins had more invertebrate individuals (>1,700 caught) and invertebrate groups (45 groups) than other margin types. However, the difference in invertebrate numbers between different treatments was fairly small (>900->1,700 individuals, 35-45 groups caught). In autumn 2001, 50 x 6 m margins at the edges of beet fields were planted in three beet growing regions with either sugar beet, spring barley, grasses (eight species), wildflowers (20 % of seeds by weight, from 20 species) or allowed to naturally regenerate. At two sites, a sixth margin of wildflowers with nitrogen fertilizer applied was established. There were two replicates of each treatment at each site. In summer 2002, plants (including crop plants) were counted in the margins, and invertebrates sampled using pitfall traps, set for two weeks.

    Study and other actions tested
  55. A replicated, controlled before-and-after study from 2000 to 2002 in winter cereal fields on four UK farms (Powell et al. 2003) found no significant effects of flower-rich margins or an aphid sex-pheromone treatment (nepetalactone, designed to enhance Aphidiinae parasitoid wasps that use aphids as their hosts) on aphid populations due to cold and wet weather at the beginning of summer 2001, resulting in small parasitoid populations. Invertebrates were sampled by in situ counting, suction-sampling, pan-traps and pitfall traps along four transects in three fields (with or without wildflower-rich/tussocky grass margins) on each farm. Sex-pheromone was released from formulated strips in the grassy field margin in autumn and in the crop in spring, starting autumn 2000. This study was part of the same experimental set-up as Powell et al. 2004.

    Study and other actions tested
  56. A replicated controlled study from 1988 to 1997 in south-central Sweden (Bokenstrand et al. 2004), found higher plant species richness in two plots sown with wildflowers (32 wildflower and grass species) than in plots planted with rose bushes Rosa canina, sown with a clover Trifolium pratense and grass seed mix, or in adjacent untreated field boundaries (control), nine years after establishment. At the third site (organic), plots sown with wildflowers and/or planted with rose bushes had lower weed and couch grass Elytrigia repens cover compared with untreated field boundaries, naturally regenerated plots and plots sown with a clover and grass mix, seven years after establishment. In two of the field boundaries, total weed cover decreased in all treatments except ‘clover and grass’ where it remained stable or increased. Couch grass cover increased in all treatments in two of the boundaries. Plant species richness tended to decline in most treatments over time, however in experimental plots sown with wildflowers and/or planted with rose bushes, 14-20 of the original 32 wildflower species were still present seven or nine years after establishment. In ‘clover and grass’ plots the clover component decreased or totally disappeared, while sown and unsown grasses and weeds increased. At the organic site, wildflower sown plots and naturally regenerated plots had similar species richness but different species compositions due to a high cover of annual weeds in the naturally regenerated plots. Four replicates of three-four treatments were established in experimental plots on each field boundary in 1988 or 1990, either by widening an existing boundary or re-establishing a previously removed dirt road (organic site). All plots were cut annually in late summer and the cuttings removed. Vegetation surveys were carried out twice in experimental plots (1991-1993 and 1997) and once in untreated field boundaries (1997) in three to five 0.25 m2 quadrats. It is not clear whether the results for clover and grass plots were a direct result of planting nectar flowers or grass.

    Study and other actions tested
  57. A replicated controlled trial in 2000-2002 in North Yorkshire, UK (Carvell et al. 2004) found that 6 m-wide field margin plots sown, or half-sown with a native ‘grass and wildflower’ seed mix supported significantly more bumblebees Bombus spp. than margins sown with a ‘tussocky grass’ mix, or control cropped field margins. Wildflower-sown margins supported significantly more bumblebees than naturally regenerated margins, but only in the first year of the three-year study, and this difference was not significant when data were averaged across all three years. Wildflower sown margins supported consistently high numbers of bumblebees, whereas naturally regenerated margins had one bumper year for bumblebees and were poor in the other two years. The three most popular forage plant species were cornflower Centaurea cyanus, bird’s-foot trefoil Lotus corniculatus and spear thistle Cirsium vulgare. The study was carried out on three arable field margins of one farm. Each margin was split into five 72 x 6 m plots and each plot subjected to one of five treatments: naturally regenerated, sown tussocky grass mix, sown grass and wildflower mix, split treatment of 3 m tussocky grass and 3 m grass and wildflower mix, or cropped to the edge. Bumblebee activity was surveyed using a standard ‘bee walk’ methodology.

    Study and other actions tested
  58. A replicated, controlled study in May to September 2000 and 2001 in the suburban area of Vienna, Austria (Kromp et al. 2004) found more ground beetle (Carabidae) species in sown wildflower strips on set-aside land than in arable fields and on unsown set-asides. Ground beetle community composition differed between the three habitat types. No statistical analyses were presented in this paper. There were six farmland sites. Wildflower strips (four sites) were sown on set-aside land with the “Voitsauer” seed mix containing 25 species of wildflowers and weeds between 1998 and 2000. The two unsown set-asides aged six and >50-years-old were cut regularly. Typical crops for the region were sown on five arable fields. One of the arable fields was under conservation contract, growing a wildflower seed mix undersown in rye. Ground beetles were sampled using four pitfall traps 10 m apart in each habitat and site. There were five sampling periods each year, each lasting two to three days (2001) or seven days (2000).

    Study and other actions tested
  59. A series of three replicated, controlled studies from 2000 to 2003 in the UK (Powell et al. 2004) monitored beneficial invertebrates and aphids (Aphidoidea) in crop fields with and without wildflower margins. All three studies found some beneficial invertebrates were more abundant in fields with flower-rich margins. One study (2000-2003, four farms) found more ground beetles (Carabidae) and Pterostichus spp. and smaller aphid populations in fields with flower-rich margins than those with tussocky grass margins, or no margins. Numbers of adult aphid-eating hoverflies were similar in winter wheat fields with and without flower margins. In the same study in 2003 (four new locations with different crop types, same study as (Powell et al. 2003)) more aphids were parasitized and there were more parasitoids in a broccoli field with a flower-rich margin than in one with control (no field margins) and areas treated with a pheromone to enhance numbers of aphid-specific parasitoid wasps (Aphidiinae). In a 2002 study on four fields, more predatory invertebrates were found next to flower-rich set-aside strips than conventional field margins in mid-July (margins 24 m wide, including phacelia Phacelia tanacetifolia, sunflowers Helianthus spp., yellow sweet clover Melilotus officinalis). More cereal aphids but fewer rove beetles (Staphylinidae) were found next to flower-rich margins in wheat or pea fields respectively. Cereal aphid numbers were unaffected by field margins. The study was continued in four winter wheat and four pea fields. It found abundance of ground beetles, Harpalus spp., field overwintering and predatory invertebrates and ground beetle species richness was higher in pea fields with flower-rich set-aside strips than fields with control margins. However numbers of field overwintering invertebrates, predatory invertebrates, rove beetles, Pterostichus spp. and rove beetle species richness were lower in wheat fields with wildflower strips. Pea aphid Acyrthosiphon pisum abundance did not differ between pea fields with or without set-aside strips. Numbers of some hoverflies or aphids were unaffected by the presence of sown margins in all studies. Invertebrates were sampled using a range of methods, including suction-sampling, pan traps and pitfall traps. The 2000-2003 study consisted of four transects in three winter wheat fields: one field with wildflower-rich margin, one with tussocky grass margin, one with neither.

    Study and other actions tested
  60. A replicated study in 2000 and 2001 south of Bern, Switzerland (Briner et al. 2005) found that most home ranges and core regions of common vole Microtus arvalis territories lay within a six-year-old wildflower strip (91% total home range, 100% core region found within wildflower strip). Thus, vole activity in the adjacent crop fields (maize and wheat in 2000, maize and sugar beet in 2001) was very low. Vole home ranges in the wildflower strip were small (median size 125 m2) compared to findings from other studies and habitats. The authors suggest that an abundance of food in the wildflower strip may account for the small range size. Daily home range sizes were stable between days (overlap of 61-99%). The wildflower strip (130 x 6 m) was dominated by tansy Tanacetum vulgare, Fuller’s teasel Dipsacus fullonum, wild parsnip Pastinaca sativa and grasses, and had not been mown. In total, 118 voles were captured using Longworth traps. Radio-transmitters (2 g) were attached with a nylon cable around the animals’ neck. Each vole was tracked every 60 seconds using automatic telemetry for one to five days. Data from 40 voles tracked for at least 24 h with >100 bearings per day were analysed. Individual home ranges were analysed in 2000 and 2001, in 2002 above ground activity patterns of 20 voles were analysed.

    Study and other actions tested
  61. A replicated study from 2001 to 2004 in Switzerland (Jacot et al. 2005) showed that both seed composition of wildflower mixtures and micro-climatic conditions influenced the emerging plant community in sown wildflower strips. The number of plant species established from sown mixtures was relatively high (around 25 species/25 m2) in dry to moist sites, whereas fewer species (around 15 species/25 m2) established on wet or shady sites. Seed mixtures containing high proportions (>75%) of grasses often resulted in grass-dominated margins. Problematic weed species established but did not dominate any strip community. No clear effects of cutting could be shown during the four study years. Eighty wildflower strips on 35 farms were studied. The strips were hand-sown in April and May 2001 with four types of seed mixture, each mixture adapted to the micro-climatic conditions at the site: mixture with legumes, mixture without legumes, grass component of 75%, grass component of 90%. Additional strips were established in 2003 using seed mixtures with 20% or 40% grasses. Strips were cut once, twice or not at all in their first year, and annually in August from their second year onwards. Cuttings were removed. Half the strip was left uncut. Vegetation was surveyed annually in June from the second year after establishment.

    Study and other actions tested
  62. A replicated, controlled, paired sites comparison in 2003 in central and eastern England (Pywell et al. 2005) found bumblebee Bombus spp. foraging activity and species richness were significantly enhanced at 28 uncropped field margins sown with a ‘wildlife seed mixture’, compared to paired control sites of conventionally managed cereal or 16 ‘conservation headlands’. Wildlife seed mixture margins contained significantly more grass, non-grass and perennial plant species than control sites, with over double the total number of species. Flowering herbaceous plants were more abundant and diverse in wildlife seed margins, and these margins provided the widest range of forage species. This result was dependent upon key forage species being included in the seed mixture, including red clover Trifolium pratense, bird’s-foot trefoil Lotus corniculatus and borage Borago officinalis, the latter being of particular importance to short-tongued bumblebee species such as Bombus terrestris and B. lucorum. The seed mixture contained grasses, and annual and perennial broadleaved herbs. Nineteen farms were surveyed in East Anglia, and 17 farms in the West Midlands. Three agri-environment scheme (Arable Stewardship Pilot Scheme) options were studied: field margins sown with a wildlife seed mixture (28 sites), conservation headlands with no fertilizer (16 sites), naturally regenerated field margins (18 sites). Fifty-eight conventional cereal field margins were used as a control, and paired with Arable Stewardship Pilot Scheme sites. Bumblebees were surveyed along 100 x 6 m or 50 x 6 m transects twice, in July and August. Vegetation was surveyed in twenty 0.5 x 0.5 m quadrats.

    Study and other actions tested
  63. A replicated study in June and September 2002 in Yorkshire, UK (Woodcock et al. 2005) found that beetle (Coleoptera) abundance and species richness were strongly influenced by the type of seed mixture sown on experimental plots. A mix containing mainly flowering plants but no tussock grass species (‘Fine grass and forbs mix’) had fewer beetles and beetle species than a mix containing wildflowers, fine and tussock grass species (‘Tussock grass and forbs mix’) and a mix with tussock and fine grass species but no flowering plants (‘Countryside Stewardship mix’). Beetle diversity (Shannon-Weiner index) did not differ between the different seed mixes. Plant communities in the grass-only and tussock grass and wildflower mixes were more similar to each other than to the fine grass and wildflower mix. The three seed mixtures were each randomly sown on three of nine experimental plots in each of five blocks on one farm in autumn 2001. Plots measured 25 x 5 m. Seed mixes contained 3-7 grass and 0-19 wildflower species. The strips were cut once in July with cuttings left in situ. Plant diversity and cover and vegetation structure were surveyed in June and September 2002 using 0.5 x 0.5 m quadrats and a ‘drop disk’. Beetles were sampled using a Vortis (Burkland Ltd., UK) suction sampler. Five samples (15 suctions for 10 seconds) were taken in each plot (total area sampled 1.32 m2) on each visit. This study was extended in Woodcock et al. 2008.

    Study and other actions tested
  64. A replicated study in 2005 in the province of South Holland, the Netherlands (Alebeek et al. 2006) found that natural enemies (parasitoids, hoverflies (Syrphidae) and predatory bugs (Hemiptera)) were generally more abundant near sown flower strips than further into the crop. There were no clear effects of sown field margins and flower strips on pest populations. No figures were presented. A total of 15 km of perennial field margins and flower strips were sown along field edges and across fields on five neighbouring farms in a 400 ha area. Flower availability, natural enemy and key pest densities were measured in 2005.

    Study and other actions tested
  65. A replicated study from 2000 to 2006 in England (Allcorn et al. 2006) found red clover Trifolium pratense and other legumes tended to establish well after hay, rich in red clover, was spread over former arable fields, however seeds of these species may already have been present in the seedbank. Leguminous species tended to increase in abundance in three fields between 2003 and 2006 (red clover present in 18-55% quadrats in 2003, 44-90% of quadrats in 2006). However, in the two other fields, both red and white clover T. repens decreased (red clover present in 25-34% quadrats in 2003 to 2-11% in 2006, white clover 25-97% in 2003 to 22-48% in 2006). Of the undesirable weeds, creeping thistle Cirsium arvense (87-98% in 2003 to 37-94% in 2006) and spear thistle C. vulgare (10-54% in 2003 to 0-38% in 2006) tended to decrease in abundance between 2003 and 2006. Common ragwort Senecio jacobaea increased in three fields (0-15% in 2003 to 6-36% in 2006) and declined in one (19% in 2003 to 0% in 2006). Hay and cuttings were obtained from nearby farms and the study site and were spread over fields once between April and August 2000-2003. Fields were summer grazed by livestock. The presence of species was recorded in 100 random nested quadrats (1 x 1 m and 2 x 2 m) in each field until 2006.

    Study and other actions tested
  66. A replicated trial in 2001-2003 on a farm in North Yorkshire, UK (Carvell et al. 2006) found highest bumblebee Bombus spp. abundance on plots sown annually with a cover crop mix of five herbaceous species (including borage Borago officinalis, fodder radish Raphanus sativus and common melilot Melilotus officinalis). Short-tongued bumblebees (B. terrestris, B. lucorum and B. pratorum) strongly preferred this annual seed mixture over two perennial grass and wildflower seed mixes. Long-tongued bumblebee species B. hortorum and B. pascuorum preferred the perennial grass and wildflower seed mixtures, but were not more abundant on the ‘diverse’ than the ‘basic’ mix. On average 70% of pollen collected by buff-tailed bumblebee workers B. terrestris was from borage, and 76% of pollen collected by common carder bee B. pascuorum workers came from red clover Trifolium pratense. Five 6 x 30 m plots of each seed mixture were established in April 2001 (the annual seed mixture plots re-sown each year after ploughing). Bumblebees were monitored May-August 2002-2003 in 4 x 30 m transects down the centre of each plot. Both perennial grass and wildflower mixes contained five grass species, the basic mix contained three herbaceous species (black knapweed Centaurea nigra, bird’s-foot trefoil Lotus corniculatus and red clover) and the diverse mix contained 18 herbaceous species.

    Study and other actions tested
  67. A replicated trial from 2001 to 2004 in Belgium (Cauwer et al. 2006), found that field margins sown with wildflower mix had similar numbers of plant species to naturally regenerated margins after three years. The number of plant species decreased over time in sown plots (from 22-23 species/plot in July 2002 to 13-16 species/plot in July 2004), and the proportion of legumes also decreased. The relative abundance of perennial plants increased and the relative abundance of annuals decreased over time on all the field margin plots, regardless of treatment. In September 2001, 10 m-lengths of two 10 x 180 m arable field margins were either sown with 77 commercially available wildflower/grass species (mix 1), sown with 63 native, locally sourced wildflower and grass species (mix 2) or left to naturally regenerate. One margin was in a sunny location, the other shaded by trees. The margins were mown twice each year in late June and September, from 2002 to 2004. Each combination of treatments was replicated three times. Plants were recorded in July and October from 2002 to 2004.

    Study and other actions tested
  68. A 2006 review on the effects of the Swiss Ecological Compensation Areas scheme in Switzerland (Herzog et al. 2006) found wildflower strips sown with native flowers on set-aside land attracted ground beetle species (Carabidae) that were never or only rarely found in wheat fields. No details on study design, monitoring techniques or other methods were given.

    Study and other actions tested
  69. A replicated, controlled study in September 2004 in four regions north of the Alps in Switzerland (Jacot et al. 2006) found slightly more montane water vole Arvicola terrestris scherman hills in sown wildflower field margins than in conventional field margins and sown wildflower strips on set-aside land. Significantly more vole hills and holes were recorded in the different types of field margin than in the crops or on the edge of crop fields. Montane water voles were the most commonly recorded species (98% of observations), common voles Microtus arvalis made up 2% of observations. The European mole Talpa europaea was not recorded at any site. Three types of field margin were compared: 17 sown field margins (5 x 120 m) established in 2001-2003 with seed mixtures containing native wildflowers, grasses and legumes, 11 conventional field margins generally species-poor and cut several times yearly (0.5-2 x 100-200 m), and seven wildflower strips sown on set-aside land (at least 5 m wide and 120 m long). On each site, nine plots (5 x 5 m) were investigated: three plots on the field margin, three on the edge of the crop field and three plots 10 m into the crop. Vole and mole hills and holes were counted on three visits. Shape and distribution of the hills and holes were used to distinguish between the three species.

    Study and other actions tested
  70. A replicated, controlled study in 2002 and 2004 (April-July) in central Switzerland (Luka et al. 2006) found more spiders (Araneae) and ground beetles (Carabidae) in wildflower strips and extended field margins than in permanent road margins. There were more ground beetle species in wildflower strips and extended field margins than in road margins, whereas there were fewer spider species in wildflower strips than in road margins. No statistical analyses were performed on the data. Four extended field margins (one-year-old in 2002), four sown wildflower strips (one to four years-old) and four permanent meadow strips (road margins, less than 10 years old) in two different regions on 12 farmland sites were compared. All sites were 100-250 m-long and 0.5-5.0 m-wide, except for one 50 m-wide wildflower strip. No information was provided about seed mixtures used for margin establishment, however the existing vegetation on the sites was either grass dominated, or a species-poor to species-rich flora dominated by flowering herbs. Arthropods were sampled using pitfall (funnel) traps placed in groups of four at least 10 m apart in each site. The traps were emptied weekly for three weeks in April-May and two weeks in June-July.

    Study and other actions tested
  71. A replicated controlled trial in 2004 in thirty-two 10 km grid squares across England (Pywell et al. 2006) found significantly more bumblebee species Bombus spp. in field margins sown with wildflower or ‘pollen and nectar’ seed mixes (more than 3 bumblebee species/transect) than in grassy margins (1.3-1.4 spp./transect) and control cropped margins (0.1 spp./transect). Pollen and nectar margins had more individuals (86 bees/transect) than any other treatment. Wildflower margins had more individuals (43/transect) than grassy (6-8/transect) and control cropped margins (0.2/transect). Field margins were 6 m-wide and part of agri-environment scheme agreements. Five field margin types were investigated: grass and wildflower mix (sown between 1999 and 2003), ‘pollen and nectar’-rich margin (sown between 2002 and 2003), grass mix (sown between 1993 and 2000), grass mix (sown between 2002 and 2003), control cropped margins. Wildflower mixes were variable in species composition but typically consisted of perennial wildflowers, fine-leaved and tussock-grass species, sown in an 80% grass:20% wildflower ratio by weight. ‘Pollen and nectar’ mixes typically consisted of at least four nectar-rich wildflower species (such as clover Trifolium spp. and bird’s-foot trefoil Lotus corniculatus) and four grass species sown in an 80% grass:20% wildflower ratio by weight. All five margin types were surveyed within each 10 km grid square (excluding the grass and wildflower mix which was not present in all squares), giving a total of 151 margins. Bumblebees were counted on a 100 x 6 m transect in each field margin, once in July and once in August.

    Study and other actions tested
  72. A replicated, controlled study in the summer of 2003 in central Switzerland (Aschwanden et al. 2007) found higher densities of small mammals (mainly common voles Microtus arvalis) in wildflower strips than in low-intensity meadows, conventionally farmed artificial grasslands and autumn-sown wheat fields. Small mammal species richness in wildflower strips was equal to that in conventionally farmed habitats and low-intensity meadows, but lower than in herbaceous strips. Over the summer, small mammal density increased most in the wildflower strips and herbaceous strips than in low-intensity meadows, conventionally farmed artificial grassland and autumn-sown wheat fields. Wildflower strips (fallows sown with seed mixtures of native plants) and herbaceous strips (consisting mainly of herbaceous plants such as thistles Cirsium spp., common teasel Dipsacus sylvestris, St John’s wort Hypericum perforatum, common mallow Malva sylvestris and mulleins Verbascum spp.) were not cut regularly during the growing season, whereas the other grassland habitats were cut at least twice. A capture-recapture method was used to estimate small mammal densities. Small mammals were trapped and individually marked during 60 hour trapping sessions in March, May and July.

    Study and other actions tested
  73. A replicated controlled trial from 2001 to 2004 at six sites across central and eastern England (Carvell et al. 2007a) found 6 m-wide margins of cereal fields sown with pollen and nectar flower mixture supported significantly more foraging bumblebee Bombus spp. species and individuals than cropped, grassy or naturally regenerated field margins. Bumblebees included the long-tongued species B. ruderatus and B. muscorum. Wildflower mixture supported significantly more foraging bumblebee species and individuals than cropped field margins, including conservation headlands, in all three years of monitoring, and grassy or naturally regenerated unsown field margins in years two and three. In the third study year (2004), wildflower and pollen and nectar mixtures supported similar numbers of bumblebee species and individuals. Wildflower margins had more flowers in May-June than July-August (approximately 4,200 vs 2,000 forage flowers/plot). Pollen and nectar margins had few flowers in May and June when bumblebee queens of late-emerging species are foraging but a large number of flowers in July-August (2,000 vs 9,000 forage flowers/plot). The number of flowers declined in the nectar and pollen mix in year three. Flower numbers remained similar between years two and three in the wildflower mix. Native varieties of red clover Trifolium pratense and bird’s-foot trefoil Lotus corniculatus flowered earlier than agricultural varieties. Two experimental plots (6 x 50 m) were established in each field along two margins. There were six treatments: wildflower mixture (21 native wildflower, four fine grass species), pollen and nectar mixture (four agricultural legume, four fine grass species), tussocky grass mixture, conservation headland, natural regeneration, crop (control). Foraging bumblebees were counted May-late August, on 6 m-wide transects 6-11 times/margin. Flower abundance was estimated along bumblebee transects in 2002, 2003 and 2004.

    Study and other actions tested
  74. A 2007 review of two studies in England (Carvell et al. 2007b) found that margins sown with a nectar and pollen mix consistently attracted more foraging bumblebees Bombus spp. than other field margin options. Two replicated controlled trials (Pywell et al. 2006, Carvell et al. 2007a) monitored the use of arable field margins sown with grass, wildflower and pollen and nectar seed mixes. One used six sites (Carvell et al. 2007a), the other 32 sites (Pywell et al. 2006). Both studies found higher numbers of bumblebees on margins sown with pollen and nectar mix, although the number of bumblebee individuals and species increased over time in the wildflower mix in one study (Carvell et al. 2007a), and supported higher numbers of some species in the other (Pywell et al. 2006). The review recommends pollen and nectar mix for a rapid and positive impact on the number of foraging bumblebees, but suggests that wildflower mix is important in catering for a wider range of bumblebee species across the whole season.

    Study and other actions tested
  75. A replicated, controlled trial from 2003 to 2005 in eastern and central England (Heard et al. 2007) found that forage patches sown with a 20% legume seed (clovers Trifolium spp. and bird’s-foot trefoil Lotus corniculatus): 80% grass mix attracted significantly higher densities of bumblebees Bombus spp. than control patches of non-crop vegetation typical of the site (average 26 bumblebees/200 m2 on sown forage patches compared to 2 bumblebees/200 m2 on control patches). Honey bees Apis mellifera and cuckoo bumblebees (Bombus [Psithyrus] spp.) were not found in greater densities on forage patches. The study also showed that bumblebee densities on sown forage patches were higher in areas with a greater proportion of arable land in a surrounding 1 km-radius than in landscapes with less arable and more grassland, woodland and urban habitats. Eight areas with varying proportions of arable, grassland, woodland and urban areas in the surrounding landscape were studied. Four treatments were established in each area from autumn 2003 to spring 2004: sown forage patches of 0.25, 0.5 and 1 ha and one control patch of non-crop vegetation typical of the area. Bumblebees and honey bees were surveyed monthly from May to September 2005 on two 2 x 100 m transects in each forage patch.

    Study and other actions tested
  76. A randomized, replicated, controlled trial from 2002 to 2006 in eastern England (Henderson et al. 2007) (same study as Pywell et al. 2007, Ramsay et al. 2007) found that the number of birds using sown wildflower margins in summer increased by 29% between 2003 and 2006. The management of sown wildflower field margins affected bird use more than the seed mix used. Bird densities were higher on disturbed and grass-specific herbicide-treated plots than on cut plots (no actual bird densities given, only model results). Bird densities were linked to densities of diurnal ground beetles (Carabidae), especially in disturbed and grass-specific herbicide-treated plots. Field margin plots (6 x 30 m) were established using one of three seed mixes: Countryside Stewardship grass mix, tussock grass mix and a mixture of grasses and wildflowers designed for pollinating insects. The margins were managed in spring from 2003 to 2005 with one of three treatments: cut to 15 cm, soil disturbed by scarification until 60% of the area was bare ground, treated with grass-specific herbicide at half the recommended rate. There were five replicates of each treatment combination at two farms. Birds were surveyed five to eight times between April and July from 2002 to 2006.

    Study and other actions tested
  77. A replicated trial from 2001 to 2005 across nine regions in Switzerland (Jacot et al. 2007) found that wildflower strips sown with a ‘locally adapted’ mix of grass and flower species (species local to the area) had between 10 and 27 target plant species/20 m2, and more butterflies (Lepidoptera), grasshoppers (Orthoptera) and ground beetles (Carabidae) than conventionally cropped margins. The sown ‘locally adapted’ field margins had more butterfly and grasshopper species and individuals than standard wildflower strips, and four to forty times more grasshopper and butterfly species and individuals than conventional cropped margins. There were, on average 3.0 unusual (not ubiquitous) butterfly species and 5.4 unusual grasshopper species in the sown margins. Wildflower strips and ‘locally adapted’ sown field margins consistently had more ground beetle species and individuals than conventional margins. Conventional margins tended to have more spider (Araneae) species (statistically significant only in one region in 2002). The total abundance of spiders and ground beetles was highest in wildflower strips (2,500-4,800 individuals/margin/year/ in total), followed by locally adapted sown margins (2,000-4,000) compared with cropped margins (1,000-2,000). Seventy field margins (5 x 120 m) were sown with seeds of up to 38 grass and wildflower species (‘locally adapted’ mix) in 2001 and 2003. Butterflies and grasshoppers were counted five times between May and August 2003 and 2005 on seven locally adapted sown margins, and compared with standard wildflower strips, conventional cropped margins, extensively managed sown hay meadows and biodiversity-rich meadows. Ground beetles and spiders were sampled using pitfall traps for five weeks from April to July in 2002 and 2004 on four locally adapted sown margins, four standard wildflower margins and four conventional margins. Plants were monitored on all locally adapted sown margins in June 2002-2005.

    Study and other actions tested
  78. A small replicated site comparison study in 2005 in Oxfordshire, UK (Marshall 2007) found that field margins sown with a wildflower mix had fewer grasshopper and cricket (Orthoptera) species and individuals than margins sown with a grass and flower mix (floristically enhanced grassy margins). Wildflower margins had less than four individual insects from one species/margin on average, compared to 10 individuals from four species on narrow grass and flower margins (2 m-wide). They did not have more grasshopper/crickets than sown grassy margins, or existing grassy tracks. The wildflower margins had the lowest grass cover (less than 60%), compared to 100% for sown grass and flower margins. Three replicates of five field margin types were monitored on a large mixed farm: grass and wildflower mix (2 m), grass and wildflower mix (6 m), grass only mix, wildflower mix, grassy track. Grasshoppers and crickets were surveyed using a sweep net over two 20 minute periods in a 50 m-section of each margin, in late July or August 2005.

    Study and other actions tested
  79. A randomized, replicated, controlled trial from 2003 to 2006 in southwest England (Pilgrim et al. 2007) found that plots of permanent pasture sown with a grass and legume seed mix attracted more birds and more bird species than control treatments, in both summer and winter. Three plots (50 x 10 m) were established on each of four farms in 2002, re-sown in new plots each year and monitored annually from 2003 to 2006. Sown legumes included white clover Trifolium repens, red clover T. pratense, common vetch Vicia sativa and bird’s-foot trefoil Lotus corniculatus. This study was part of the same experimental set up as Potts et al. 2009.

    Study and other actions tested
  80. A replicated controlled trial in 2005-2006 in Warwickshire, UK (Pywell & Nowakowski 2007) found that field corners or margins sown with a wildflower mix had more plant species, more bumblebees Bombus spp. (species and individuals) and more butterfly (Lepidoptera) species than control plots sown with winter oats. There were 17 plant species/m2, 7 bumblebees and 2 bumblebee species/plot, and 5 butterfly species/plot on average in wildflower plots, compared to 2 plant species/m2, no bumblebees and one butterfly species/plot in cereal crop plots. Two declining butterfly species, small copper Lycaena phlaeas and common blue Polyommatus icarus were only found in wildflower plots. Wildflower plots did not have more butterfly individuals, or more birds in winter (species or individuals) than control crop plots. The wildflower mix (25 broadleaved non-grass species, four grass species 10%:90% wildflowers to grasses by weight) was sown in August 2005 and treated with grass-specific herbicide in November 2005. Plots were cut three times in 2006, and cuttings removed. Each treatment was tested in one section of margin and one corner in each of four fields on one farm. Plants were monitored in three 1 m2 quadrats/plot in July 2006. Butterflies, bumblebees and flowering plants were recorded in a 6 m-wide transect five times between July and September. Farmland birds were counted on each plot on seven counts between December 2006 and March 2007. Results from the second year of monitoring are presented in (Pywell & Nowakowski 2008).

    Study and other actions tested
  81. A randomized, replicated, controlled trial from 2002 to 2006 in eastern England (Pywell et al. 2007) (same study as Henderson et al. 2007, Ramsay et al. 2007) found that field margins sown with a flower mix designed for pollinating insects did not support more butterflies (Lepidoptera) or bumblebees Bombus spp. than a floristically enhanced tussocky grass seed mixture. There were 35-47 bumblebees of four species and 18-20 butterflies of six species/125 m2 plot on average in margins sown with some non-grass species in the mix, compared to 10 bumblebees of two species and 12 butterflies of five species on grass-only margins. Different types of management did not affect the abundance of bees and butterflies or the number of butterfly species, but there were more bumblebee species on plots treated with grass-specific herbicide in spring (average 4 species/125 m2, compared to 3 species on cut or disturbed plots). Field margin plots (6 x 30 m) were established in 2000-2001 using one of three seed mixes: Countryside Stewardship mix (seven grass species, sown at 20 kg/ha), tussock grass mix (7 grass species, 11 wildflowers, sown at 35 kg/ha) and a mixture of grasses and wildflowers designed for pollinating insects (4 grass species, 16-20 wildflowers, sown at 35 kg/ha). Margins were managed in spring from 2003 to 2005 with one of three treatments: cut to 15 cm, soil disturbed by scarification until 60% of the area was bare ground, treated with grass-specific herbicide in spring at half the recommended rate. There were five replicates of each treatment combination on three farms.

    Study and other actions tested
  82. A randomized, replicated, controlled trial from 2002 to 2006 in eastern England (Ramsay et al. 2007) (same study as Henderson et al. 2007, Pywell et al. 2007) found that field margins sown with a flower mix designed for pollinating insects supported fewer planthoppers (Auchenorrhyncha) than those sown with a grass-only seed mixture. Flower-rich sown margins had 25-40 planthoppers/plot on average (depending on management), while grass-only margins had 30-70 planthoppers/plot. Field margin plots (6 x 30 m) were established in 2000-2001 using one of three seed mixes: Countryside Stewardship mix (7 grass species, sown at 20 kg/ha), tussock grass mix (7 grass, 11 wildflower species, sown at 35 kg/ha) and a mixture of grasses and wildflowers designed for pollinating insects (4 grass species, 16-20 wildflowers, sown at 35 kg/ha). The margins were managed in spring from 2003 to 2005 with one of three treatments: cut to 15 cm, soil disturbed by scarification until 60% of the area was bare ground, treated with grass-specific herbicide in spring at half the recommended rate. There were five replicates of each treatment combination on three farms.

    Study and other actions tested
  83. A replicated, controlled study from 2005 to 2008 in England and Scotland (Edwards 2008) found the average number of worker bumblebees Bombus spp. was greater on margins where legume-rich seed mix was established than on other field margins (grassy margins or track edges). There was an observed decline in the relative number of foraging worker bumblebees on legume-sown margins after they had been established for more than three years (data from five farms). No formal statistical analyses were performed on these data. On each of 41 farms, four 100 x 2 m field margins were surveyed for bumblebees. Two of the margins were sown with a legume-rich seed mix in either April 2005, 2007 or 2008. The other two margins were track edges or grassy margins. Bumblebee surveys were made on all four margins on each farm from 2005 to 2008, twice in June/early July and late July/early August.

    Study and other actions tested
  84. A replicated, controlled study from 2001 to 2004 in the UK (Gardiner et al. 2008) found arable margins sown with a legume-grass seed mix had more bumblebee Bombus spp. forage plant species (almost 100% cover of Alsike clover Trifolium hybridum and red clover T. pretense one year after establishment) over four years, compared to naturally regenerated margins. The cover of Alsike clover declined from a peak of approximately 33% in 2002 to 2.5% in 2004, whilst red clover cover peaked at around 85% in 2003 and declined to 20% in 2004. Clover-sown plots were invaded by perennial grasses including false oat grass Arrhenatherum elatius in the third and fourth years of the study, when clover cover decreased substantially. Bee visits were not reported in this study however the results of fixed-time transect walks in the clover margins are reported in (Edwards & Williams 2004) which found a 300-fold increase in bumblebee forager numbers in the margins planted with clover, however no control count was carried out for comparison. Two 6 m-wide margins were established on one farm, and subdivided into three plots. There were two margin types: naturally regenerated, or sown with a mixture of grasses and leguminous species including two species of clover. Three different management treatments were applied to the subplots in the first year (2001): cut three times with cuttings left, cut three times with cuttings removed, cut six times with cuttings left. From 2002 to 2004, all plots were cut in late summer and the cuttings removed. Forage plants were monitored in 0.25 m2 quadrats every 1 m along a 30 m transect in early August 2001, 2002, 2003 and 2004.

    Additional references:

    Edwards M. & Williams P. H. (2004) Where have all the bumblebees gone, and could they ever return? British Wildlife, 15, 305-312.

    Study and other actions tested
  85. A replicated, controlled study in 2000-2003 in Norway (Hovd 2008) found that plant species diversity was higher in strips sown with a grass/wildflower mixture than strips left to regenerate naturally or in the grass crop. The number of plant species was significantly higher in sown grass/wildflower strips (17-18 species/quadrat) than naturally regenerated strips (10-12 spp.) or the grass crop (7-9 spp.). The same was true for the number of meadow herbs (sown: 7-10 species, unsown: 1-3 spp., crop: 1 spp./quadrat). Plant diversity (Shannon diversity index) was also significantly higher in sown grass/wildflower strips than in either naturally regenerated margins or in the crop (sown: 1.8-1.9, unsown: 1.3-1.6, crop: 1.0-1.3). Four of the 22 sown meadow wildflower species did not establish. Naturally regenerated strips and the grass crop had many species in common by the fourth study year, and grasses and perennial weeds dominated in the crop and unsown strips. By the fourth year some sown species were recorded in the unsown strips or grass crop and woody species from an existing semi-natural margin were recorded in the sown strips. The total number of plant species did not vary with distance from the existing margin. Four strips (2 m-wide) were ploughed perpendicular to an existing semi-natural margin, in May 2000. One half of each was left to regenerate naturally, the other half was sown with a grass/meadow wildflower (22 species) seed mixture (5 g/m²). Wildflower seeds were local to the area, grass seeds were cultivated varieties. Sown strips did not receive fertilizer and were cut once (late September). Permanent quadrats (0.5 x 0.5 m) were sampled in the grass crop and strips in June 2000-2003.

    Study and other actions tested
  86. A replicated, controlled site-comparison study in 2005 in the Netherlands (Kohler et al. 2008) found the number of flowers was 10 times higher in plots sown (or planted) with 17 insect-pollinated plant species than outside the plots (approximately 4,650 vs 480 flowers/plot within/outside flower plots respectively). The number and diversity of bees (Apidae) and hoverflies (Syrphidae) was significantly higher (60-80% higher) in flower plots than on control transects. Outside the flower plots, hoverfly abundance was significantly higher 50 m away from the flower plots but not at any other distance. The lowest numbers of bees and bee species were recorded 50 m away from the flower plots. Seventeen species of annual and perennial plants were either transplanted or sown in fenced 10 x 10 m plots at five locations in intensive farmland. Hoverflies and bees were surveyed at 10 sampling locations along a 1,500 m transect running away from each flower plot, and along five 1,500 m control transects. All transects ran alongside ditches. Bees and hoverflies were sampled using window traps, yellow water pans and nets four times between June and September 2004.

    Study and other actions tested
  87. Two randomized replicated studies from 2005 to 2007 in Yorkshire and Warwickshire, UK (Pywell et al. 2008) studied different ‘pollen and nectar’ seed mixes. The Yorkshire study (2005-2007) looked at six different seed mixes and found two agricultural varieties of red clover Trifolium pratense had the highest cover. There were more flowers of sown plant species in an agricultural clover mix in 2005 and in wild Somerset red clover in 2006. There were more red clover flowers in an agricultural mix in all three years. Bird\'s-foot trefoil Lotus corniculatus flowers were more abundant in the first and third years in wild Somerset red clover mix than in agricultural mixes. Plots measured 48 x 6 m and were replicated twice. The Warwickshire study (2006-2007) (same study as (Pywell et al. 2010b)) found butterfly (Lepidoptera) abundance and species richness were highest in plots sown with lucerne Medicago sativa or red clover (3-6 butterflies/plot, 2-3.5 species vs 0-3 butterflies and 0-3 species for all other plots). Phacelia Phacelia tanacetifolia plots had the highest number of bumblebees Bombus spp. (39-134 bumblebees/plot), followed by borage Borago officinalis (32-100). Phacelia, crimson clover Trifolium incarnatum, borage, sunflower Helianthus annuus and red clover (2007 only) plots had significantly more bumblebee species than all other plots in 2006 (3-4 vs 0-1 species/plot). Short-tongued bees preferred phacelia and borage in both years. Long-tongued bees preferred crimson clover, borage, phacelia or red clover. In the first year there were more annual than perennial flowers. Flowering of many important bee forage species peaked in late July. Thirteen annual and perennial plant species were sown individually in 6 x 4 m plots, replicated four times in May 2006, annual species were re-sown May 2007. Butterflies and bumblebees were surveyed six times in each plot (July-September 2006 and May-September 2007). On each visit the percentage cover of all flowers was estimated.

    Study and other actions tested
  88. A replicated controlled trial from 2005 to 2007 in Warwickshire, UK (Pywell & Nowakowski 2008) (the second monitoring year of the same study as Pywell & Nowakowski 2007) found that wildflower plots had more plant species, bumblebees Bombus spp. and butterflies (Lepidoptera) (individuals and species) than naturally regenerated or control cereal plots, and more vacuum-sampled invertebrates than control plots. Wildflower plots did not have more birds in winter than control plots. On wildflower plots there were 10 plant species/m2, 63 bumblebees and 5 bumblebee species/plot, 18 butterflies and 6 butterfly species/plot, compared to 3 plant species/m2, 0 bumblebees, and 1 butterfly/plot on control cereal plots. Control cereal plots had 254 vacuum-sampled canopy-dwelling invertebrates/m2 on average, compared to 840-1,197/m2 on other treatments. Plants were monitored in three 1 m2 quadrats in each plot in June 2007. Butterflies, bumblebees and flowering plants were recorded in a 6 m-wide transect six times between July and September 2006 and 2007. Invertebrates in the vegetation were vacuum-sampled in early July 2007. Farmland birds were counted on each plot on four counts between December 2007 and March 2008. The crop control in year two was winter wheat.

    Study and other actions tested
  89. A replicated controlled trial from 2002-2004 in County Wexford, Ireland (Sheridan et al. 2008) found that 1.5-3.5 m-wide margins of permanent pasture fields fenced, rotavated and sown with a wildflower seed mix had more springtails (Collembola: Anthropleona), spiders (Araneae), flies (Diptera) and plant species than control margins. Wildflower margins had 18 plant species/plot in July 2002, decreasing to 11 plant species/plot in 2004, compared to less than 5 plant species/plot throughout the study in control plots and unrotavated fenced margins. Some undesirable weeds, such as broad-leaved dock Rumex obtusifolius, established in rotavated plots, but were less abundant in plots sown with wildflower mix (broad-leaved dock cover was 6-25% in wildflower sown plots and 26-50% in naturally regenerated plots in 2002 and 2004). Grazing on half of each plot from 2003 to 2004 (one year after margin establishment) did not affect the number of plant species the following year. Plots were 30 m long (1.5-3.5 m-wide), with three replicates of each treatment combination. The wildflower mix had 10 grass species and 31 non-grass species. Plants were monitored in permanent quadrats in July 2002, May and July 2003 and May 2004. Invertebrates were sampled in six emergence traps/plot, between May and September 2003. Ground areas under the emergence traps were sampled with a vacuum sampler.

    Study and other actions tested
  90. A replicated study in summer 2002-2004 and 2006 on three farms in England (Woodcock et al. 2008) (study extended from Woodcock et al. 2005) found a greater abundance (but not species richness) of herbivorous beetle species (Coleoptera) in seed mixtures including wildflowers than in grass-only mixtures. However there were more predatory beetle species and individuals in margins containing tussock grass species regardless of whether the mixture also included wildflowers. Margin management (i.e. soil scarification) also had a positive effect on species richness of predatory beetles. Three different seed mixtures were sown: grass only, tussock grass and wildflowers, fine grass and wildflowers. Each of the seed mixtures was randomly sown on three of nine experimental plots (25 x 5 m) in each of five blocks on three farms in autumn 2001. From 2003, three different management practices were applied in each replicate block in May each year: cutting the vegetation to 10–15 cm, application of grass-specific herbicide (Fuazifop-p-butyl) at 0.8 l/ha, and scarification of 60% of the soil surface. Plant diversity and cover and vegetation structure were surveyed yearly in June using 0.5 x 0.5 m quadrats and vertical drop pins. Beetles were sampled using a Vortis suction sampler (75 suctions of 10 seconds each) over a fixed area (equivalent to 1.45 m2) in each plot on each sampling date. Rove beetles (Staphylinidae), ground beetles (Carabidae), ladybirds (Coccinellidae), leaf-beetles (Chrysomelidae) and weevils (Curculionoidea) were determined to species level and categorized as herbivorous or predatory.

    Study and other actions tested
  91. A site comparison study between 1997 and 2004 in central Switzerland (Aviron et al. 2009) found wildflower strips sown with 20-40 species contained significantly more (8-60% more) plant, butterfly (Lepidoptera), ground beetle (Carabidae) and spider (Araneae) species than crop fields in the same region. Estimated total numbers of species were 149 (plant), 19 (butterfly), 85 (ground beetle) and 134 (spider) on Ecological Compensation Area wildflower strips and 50, 19, 78 and 104 species on conventional crop fields respectively. Rare or threatened species were not found more frequently on Ecological Compensation Area sites. The increased number of species was a response of common species. The study sampled 78 wildflower strips and 72 crop fields in a predominantly arable region.

    Study and other actions tested
  92. A replicated, controlled study from March-July 2006 in mixed farmland near Bern, Switzerland (Fischer et al. 2009) found that Eurasian skylarks Alauda arvensis with territories that included undrilled patches sown with six annual weed species, were significantly less likely to abandon the territory and more likely to use undrilled patches as nesting and foraging sites. Nests were significantly more likely to be built within or close to undrilled patches (60% of skylark nests were within 5 m of an undrilled patch). Skylarks preferentially foraged in undrilled patches over all other crop types; undrilled patches covered 0.17–0.63% of the foraging area but were accessed on 12.6% of observed foraging flights. Plant cover ranged from 35 to 50%, and plant height ranged from 5 to 80 cm in the plots. Undrilled patches were composed of either four 3 x 12 m patches/ha (in seven fields) or a single strip 2.5 x 80 m (in 14 fields). In autumn 2005 undrilled patches were sown with six annual weed species including common corncockle Agrostemma githago in winter wheat fields. Skylark territories were surveyed over one breeding season (2006) in 21 experimental sites and 16 control wheat fields.

    Study and other actions tested
  93. A replicated, controlled study in summer 2001 in intensively managed farmland around Bern, Switzerland (Frank et al. 2009) found that the number of species and individuals, biomass and individual weights of most sampled arthropod predators increased with the age of sown wildflower sites. Conversely the number of rove beetles (Staphylinidae) and rove beetle biomass was highest in newly created wildflower sites, but the weight of individual rove beetles increased with age of wildflower sites. Control wheat fields had among the lowest species richness, density and biomass of predators, but these values were only significantly lower than in the oldest wildflower strips for spider (Araneae) and ground beetle (Carabidae) biomass and true bug (Heteroptera) density. Vegetation cover had a significant influence on spider assemblages. Ground beetle species assemblages were strongly correlated with vegetation cover, field size and soil water content in wildflower sites. Five different habitats with four replicates were surveyed at 20 sites (average 0.8 ha). The four sown wildflower habitats had been established for one, two, three and four years (one-year-old sites sown in May 2001) and were sown with a seed mixture containing 25 native plant species, not treated with fertilizer, pesticides or cut. Winter wheat fields were used as controls. Spiders, ground beetles and rove beetles were sampled using three photo-eclectors/site for two consecutive months. True bugs were sampled four times along 80 m transects using sweep-nets (100 sweeps/transect). Vegetation cover, volume of soil pores, and sand content were determined.

    Study and other actions tested
  94. A replicated, controlled study in June and July 2006 in north Germany (Haenke et al. 2009) found more hoverflies (Syrphidae) and hoverfly species in broad (12-25m wide) and narrow (3-6 m) sown wildflower strips (7 sites each) than in grassy margins (3 m-wide, 7 sites), wheat-wheat boundaries (7 sites) and within wheat fields adjacent to the margins (7 sites). Hoverfly density and species richness (total hoverflies and aphid-eating hoverflies) also increased with increasing amount of arable land around the site at smaller scales (0.5 and 1 km) but not at larger scales (2 and 4 km). Margins were located along a gradient of habitat complexities in the surrounding landscape, ranging from 30% to 100% arable land. Hoverflies were sampled by sweep netting (one sweep per footstep) along 100 m transects.

    Study and other actions tested
  95. A replicated, controlled study in the summers of 2004-2005 in northwest Switzerland (Pfiffner et al. 2009) found wildflower strips had a variable effect on parasitism and predation of eggs and larvae of two common butterfly/moth (Lepidoptera) cabbage pests. Parasitization rates of cabbage moth Mamestra brassicae eggs and larvae as well as small white butterfly Pieris rapae larvae on one farm did not differ between plots with and without wildflower strips. However on a second farm, parasitization rates of small white butterfly larvae and predation rates of cabbage moth eggs were significantly higher in plots with adjacent wildflower strips. Wildflower strips did not affect the spatial pattern of parasitization in the fields. Six cabbage Brassica oleracea fields were studied on two organic farms. Two blocks (45 x 25 m) were studied on each field, one with and one without a wildflower strip (3 x 35 m). Wildflower strips were sown with seed mixtures containing 24 native plant species, and were not treated with pesticides or mown. Egg parasitization rates were assessed by placing laboratory eggs pinned to paper cards on the ground underneath labelled plants in a 3 x 3 m grid for three days. Eggs were incubated for four weeks at 22°C to rear any parasitoids. Missing and damaged eggs were counted to estimate the predation rate. Butterfly larvae were sampled on randomly selected plants and parasitization rates were determined using DNA-based techniques.

    Study and other actions tested
  96. A randomized, replicated, controlled trial from 2003 to 2006 in southwest England (Potts et al. 2009) (same experimental set up as Pilgrim et al. 2007) found plots on permanent pasture annually sown with a mix of legumes, or grass and legumes, supported more common bumblebees Bombus spp. (individuals and species) than seven grass management options. In the first two years, there were more common butterflies (Lepidoptera) and common butterfly species in plots sown with legumes than in five intensively managed grassland treatments. No more than 2.2 bumblebees/transect were recorded on average on any grass-only plot in any year, compared to over 15 bumblebees/transect in both sown treatments in 2003. Plots sown with legumes generally had fewer butterfly larvae than all grass-only treatments, including conventional silage and six different management treatments. Experimental plots 50 x 10 m were established on permanent pastures (more than five-years-old) on four farms. There were nine different management types, with three replicates/farm, monitored over four years. Seven management types involved different management options for grass-only plots, including mowing and fertilizer addition. The two legume-sown treatments comprised either barley Hordeum vulgare undersown with a grass and legume mix (including white clover Trifolium repens, red clover T. pratense, and common vetch Vicia sativa) cut once in July, or a mix of crops (including linseed Linum usitatissimum) and legumes, uncut. Bumblebees and butterflies were surveyed along a 50 m transect line in the centre of each experimental plot, once a month from June to September annually. Butterfly larvae were sampled on two 10 m transects using a sweep net in April and June-September annually.

    Study and other actions tested
  97. A replicated, controlled, paired sites study in summer 2005 in northwest Switzerland (Schmidt-Entling & Dobeli 2009) found densities of several spider (Araneae) families were higher in wheat fields with adjoining sown wildflower areas than in fields with grassy margins. Crab spiders (Thomisidae), ground spiders (Gnaphosidae) and wolf spiders (Lycosidae) as well as young orb weaver spiders (Araneidae) had higher densities in fields with adjacent sown wildflower areas. However spider diversity and the total number of spider species were not significantly different in wheat fields adjoined by sown wildflower areas than fields with grassy margins. Twenty winter wheat fields were studied (0.5-4.1 ha in size), 10 fields had adjoining sown wildflower areas, 10 were adjoined by grassy margins. Wheat fields were treated with herbicides, fungicides and mainly mineral fertilizers but no insecticides. Sown wildflower areas (a Swiss agri-environment scheme) were sown with a mixture of 25 wildflower species, and were not treated with pesticides, fertilizers or mown. Sown wildflower areas were 0.4-2.3 ha in size (minimum 25 m wide) and between two and six years-old. Grassy margins were about 0.7 m wide and mown several times/year. Spiders were sampled from May to June using pitfall traps (0.2 l, 6.5 cm diameter) and a suction sampler (0.1 m diameter).

    Study and other actions tested
  98. A 2009 literature review of European farmland conservation practices (Vickery et al. 2009) found that field margins sown with a wildflower mix had higher arthropod diversities than adjacent crops, or margins sown with grass seed only. Several bird species were also found to use wildflower strips more than margins sown with grass seed only.

    Study and other actions tested
  99. A replicated study in 2005 in western Switzerland (Arlettaz et al. 2010) found that small mammal density and species richness were higher in wildflower areas than crops, but wildflower areas were avoided by barn owls Tyto alba. Wildflower areas (two years old, 1 ha) had more small mammal species and individuals (6 species, 458-1285 individuals/ha) compared to crops or meadows (2-5 species, 0-680/ha). In May and July, small mammal densities were significantly higher in wildflower areas (458-1,030 individuals/ha) and winter wheat Triticum aestivum (90-680/ha) than in tobacco Nicotiana tabacum, permanent and intensive meadows and in May maize Zea mays (0-10/ha) (in July the density in maize was 200/ha). In September, density was significantly higher in wildflower areas (1285/ha) than in winter wheat (0), other habitats had intermediate densities (5-60/ha). Barn owls significantly preferred cereal crops relative to availability and avoided wildflower areas and all other crop types. The estimated index of habitat selection by barn owls in order of decreasing preference was wheat, meadows, other crops and lastly wildflower areas. Four arable sites were studied. Small mammal population size was estimated using capture-mark-recapture. Mammal traps were placed at 20 points along two parallel 45 m transects in each habitat and set over three nights and days in May, July and September 2005. Seven breeding male barn owls were radio-tagged from June to September 2005 and hunting or resting locations recorded.

    Study and other actions tested
  100. A replicated study in summer 2007 in south Sweden (Haaland & Gyllin 2010) found higher densities and species richness of butterflies (Lepidoptera) and bumblebees Bombus spp. in sown wildflower strips than in strips consisting mainly of grass species (greenways or ‘beträdor’). Eighty-six percent of the recorded butterflies and 83% of the bumblebees were found in wildflower strips. Butterfly density was nearly 20 times higher in wildflower strips than in the grass strips. The most common flowers visited were field scabious Knautia arvensis and knapweeds Centaurea spp. for butterflies, and knapweeds for bumblebees (72% of all recordings). The presence of bushes adjacent to the strip positively affected the number of butterfly species and individual numbers of both butterflies and bumblebees. Butterflies and bumblebees were recorded on one wildflower strip (six transects) and three grass strips (14 transects) on five occasions on four arable farms. Butterflies and bumblebees were counted within 2 m either side of the observer, and the flower species visited by the insects noted.

    Study and other actions tested
  101. A replicated, randomized study from 2005 to 2007 in Warwickshire, UK (Pywell et al. 2010a) found no difference in the number of bumblebees Bombus spp. or bumblebee species between plots sown with ten different flowering plant and grass seed mixtures, but recorded a significant increase in the number of bumblebee individuals and species in sub-plots treated with the grass-specific herbicide propyzamide in 2007. These sub-plots also showed a significant decrease in grass cover (from 45 to 2%) and an increase in the cover of sown wildflowers (from 24 to 56%), bare ground (from 4 to 16%) and undesirable weeds (from 4 to 14%). The number and cover of sown wildflowers decreased over the years in favour of competitive grass species. Ten different seed mixes (three replicates each) were sown in plots (6 x 10 m) in April 2005. The seed mixes contained four to six flowering plant species and one to four grass species sown in different proportions. Plots were cut three times in 2005 and twice in 2006 with cuttings left in place. The grass-specific herbicide fluazifop-P-butyl was sprayed in plots with rye grass nurse crop in April 2006. In November 2006, all plots were split into two sub-plots (3 x 10 m) of which one was sprayed with propyzamide. The percentage cover of vascular plants was recorded in two randomly placed 1 x 1 m quadrats in each plot (2005-2006) or sub-plot (2007) respectively. Bumblebee abundance and diversity were monitored twice each year in late summer 2006 and 2007. These results were also presented in Pywell et al. 2008 but are only reported here.

    Study and other actions tested
  102. A replicated, randomized study in 2006 and 2007 in Warwickshire, UK (Pywell et al. 2010b) (same study as Pywell et al. 2008) found bee (Apidae) and butterfly (Lepidoptera) abundance and species richness were higher in plots sown with specific wildflower species. Bumblebee Bombus spp. abundance and species richness were significantly higher on plots sown with phacelia Phacelia tanacetifolia and borage Borago officinalis (32-85 bees/plot) compared to other treatments (1-22/plot). Crimson clover Trifolium incarnatum (10-21/plot), sunflower Helianthus annuus (10-22/plot) and red clover T. pratense (20/plot) also tended to have high bumblebee abundances (other species: 1-11/plot). Short- and long-tongued bees had different preferences. In 2006, butterfly abundance and species richness were significantly higher in plots with lucerne Medicago sativa compared to borage, chicory Cichorium intybus and sainfoin Onobrychis viciifolia. In 2007 butterfly abundance was higher in red clover compared with chicory, but the number of species did not differ between treatments. Mobile and immobile butterfly species had different preferences. Flowers of buckwheat Fagopyrum esculentum were the most abundant followed by phacelia, borage and sunflower in 2006. In 2007 fodder radish Raphanus sativus, red clover and sweet clover Melilotus officinalis also had high flower abundance. Mustard Brassica juncea and linseed Linum usitatissimum had the fewest flowers in both years, along with other species each year. Thirteen species were sown in single species stands: four wildflower species typically sown in pollen and nectar seed mixes and nine small-seeded crop species typically sown in wild bird seed mixes. The species were sown in May each year in adjacent 6 x 4 m plots in a randomized block experiment with four replicates. Butterflies and bumblebees were sampled by walking transects through each plot on six occasions from May-September. Flower cover was estimated at the same time.

    Study and other actions tested
  103. A replicated controlled site comparison study in summer 2008 in northwest Scotland (Redpath et al. 2010) found that croft sections (an agricultural system specific to Scotland, consisting of small agricultural units with rotational cropping regimes and livestock production) sown with a brassica-rich ‘bird and bumblebee’ conservation seed mix had 47 times more foraging bumblebees Bombus spp. than sheep-grazed sections and 16 times more bumblebees than winter-grazed pastures in June. In July the ‘bird and bumblebee’ mix sections had 248 and 65 times more bumblebees than sections grazed by sheep or both sheep and cattle respectively. The number of bumblebees in July was also significantly higher (4-16 times) in ‘bird and bumblebee’ sections than in arable, fallow, silage, and winter-grazed pasture sections. The availability of bumblebee forage plant flowers was lower in ‘bird and bumblebee’ sections than in silage sections in June, but no other significant differences involving the conservation mix were detected. Foraging bumblebees most frequently visited plant species in the legume (Fabaceae) family. Tufted vetch Vicia cracca was one of a few plant species favoured by bumblebees and was predominantly found in ‘bird and bumblebee’ sections in July-August, although it was not part of the seed mixture. Thirty-one crofts located on Lewis, Harris, the Uists and at Durness were studied. Species sown in the bird and bumblebee mix included kale Brassica oleracea, mustard Brassica spp., phacelia Phacelia spp. and red clover Trifolium pratense. In addition to the seven management types mentioned, unmanaged pastures were surveyed for foraging bumblebees and bumblebee forage plants along zigzag or L-shaped transects in each croft section once in June, July and August 2008. Foraging bumblebees 2 m either side of transects were identified to species level and recorded together with the plant species on which they were foraging. Flowers of all plant species were counted in 0.25 m2 quadrats at 20 or 50 m intervals along the transects.

    Study and other actions tested
  104. A randomized, replicated trial from 1987 to 2000 in Oxfordshire, UK (Smith et al. 2010) found that the number of plant species on 2 m-wide margins sown with a wildflower seed mix in 1988 declined by about half over 13 years. There were 23-24 plant species/quadrat in 1988 and 9-12 plant species/quadrat in 2000. The most rapid decline was in the first two years, when many annual species were lost. Sown plots retained more perennial plant species than naturally regenerated plots throughout the 13 years (around 10 vs 8 perennial species/quadrat respectively, in 2000). After 13 years, sown plots tended to have more species than naturally regenerated plots (9-12 vs 7-9 species/plot respectively in 2000), but this difference was not statistically significant. There was no effect of different mowing regimes on the numbers of plant species, although in the early years mown plots had more plant species than uncut plots. Sown plots that were cut twice retained a greater proportion of sown species (50-60%) than plots cut once or uncut (<40%). Sowing reduced the colonization of margins by unsown perennial species at first, but by 2000 many perennial species, including couch grass Elymus repens, were similarly abundant in sown and unsown plots. Plant species were monitored three times a year from 1988 to 1990, and once in July 2000 in three 0.5 x 1 m quadrats/plot. This was part of the same study set-up as Feber et al. 1994, Feber et al. 1996, Baines et al. 1998, Bell et al. 1999, Haughton et al. 1999, Smith et al. 1999, Bell et al. 2002.

    Study and other actions tested
Please cite as:

Dicks, L.V., Ashpole, J.E., Dänhardt, J., James, K., Jönsson, A., Randall, N., Showler, D.A., Smith, R.K., Turpie, S., Williams, D.R. & Sutherland, W.J. (2020) Farmland Conservation. Pages 283-321 in: W.J. Sutherland, L.V. Dicks, S.O. Petrovan & R.K. Smith (eds) What Works in Conservation 2020. Open Book Publishers, Cambridge, UK.

 

Where has this evidence come from?

List of journals searched by synopsis

All the journals searched for all synopses

Farmland Conservation

This Action forms part of the Action Synopsis:

Farmland Conservation
What Works 2021 cover

What Works in Conservation

What Works in Conservation provides expert assessments of the effectiveness of actions, based on summarised evidence, in synopses. Subjects covered so far include amphibians, birds, mammals, forests, peatland and control of freshwater invasive species. More are in progress.

More about What Works in Conservation

Download free PDF or purchase
The Conservation Evidence Journal

The Conservation Evidence Journal

An online, free to publish in, open-access journal publishing results from research and projects that test the effectiveness of conservation actions.

Read the latest volume: Volume 21

Go to the CE Journal

Discover more on our blog

Our blog contains the latest news and updates from the Conservation Evidence team, the Conservation Evidence Journal, and our global partners in evidence-based conservation.


Who uses Conservation Evidence?

Meet some of the evidence champions

Endangered Landscape ProgrammeRed List Champion - Arc Kent Wildlife Trust The Rufford Foundation Save the Frogs - Ghana Mauritian Wildlife Supporting Conservation Leaders
Sustainability Dashboard National Biodiversity Network Frog Life The international journey of Conservation - Oryx Cool Farm Alliance UNEP AWFA Bat Conservation InternationalPeople trust for endangered species Vincet Wildlife Trust