Providing evidence to improve practice

Action: Undersow spring cereals, with clover for example Farmland Conservation

Key messages


Supporting evidence from individual studies


A before-and-after study in Dumfries, southern Scotland (Owen 1977) found that the number of barnacle geese Branta leucopsis on a mixed agricultural site and nature reserve increased from 3,200 in 1970 to 6,000 in 1975 after all cereals sown on the site were undersown from 1970 onwards. The nature reserve consists of 220 ha of salt pasture, whilst the agricultural land is 50 ha arable fields. Most of the increased goose numbers feed on the arable land. In addition to undersowing, the proportion of cereals grown on the arable land decreased and no stock were allowed to graze on the arable land after November.



A replicated, controlled study of arable fields in West Sussex, England (Vickerman 1978) found that arthropod abundance, density and species richness was higher in undersown spring barley and undersown grass fields compared to mono-cropped fields. Arthropod abundance and diversity was greater in undersown barley fields (767-874 m², 19-23 species), compared to mono-cropped barley (677-714 m², 14-18 species) and grass (281-391 m², 12-15 species). Only the true bugs (Hemiptera) were found in greater numbers in barley than in undersown barley. On average, 70% more arthropods emerged from undersown grass (555-623 m²) than cultivated fields (280-391 m²). Species diversity was also higher in undersown grass (22-28 vs 11-16). Half of the cereal fields in the ley farming area contained over 200 parasitic Hymenoptera/m² compared to just 9% in the ‘modern arable area’ (no grass). Arthropods from the field and ground zone were sampled using a Dietrick vacuum (five sub-samples each 0.09 m²) at regular intervals across one field from each treatment in 1972 and 1973 and in one grass field in 1974. In addition one sample was taken from 150 fields in June 1972-1974. In 1970 and 1971, two adjacent fields, one undersown with grass and one cultivated, were sampled for emergent arthropods using two rows of five emergence traps (area enclosed 2.8 m²). Traps were emptied regularly between March and June.


A trial at an experimental farm in 1989 on the Swiss Plateau, Switzerland (Wyss & Glasstetter 1992) found that earthworm (Lumbricidae) abundance and biomass were higher in a maize Zea mays plot undersown with grass than in conventionally managed maize, although statistical analyses were not presented. Control and undersown plots had averages of 127 and 145 earthworms/m2 and 45 and 71 g earthworm biomass/m2, respectively. The proportion of deep-burrowing earthworms was similar with 14 and 12% of individuals in the control and undersown plots respectively. A test strip of maize 14 m-long was undersown with grass in summer and compared with a control strip of conventional maize. Earthworms were sampled by hand-sorting 0.1 m3 of soil from each test strip, to a depth of 40 cm, on six dates between April and October 1989. There was no replication.



A replicated, controlled, randomized study of an undersown cereal field from 1989 to 1991 in Helsinki, Finland (Helenius & Tolonen 1994) found that green manuring with undersown rye grass Lolium spp. resulted in a 50% increase in new generation adult ground beetles (Carabidae). A total of 33 species of ground beetles were caught. For three of the most common species, plots that had received rye grass green manuring (0.5 t/ha) in autumn 1990 resulted in a 50% increase in new generation adults in 1991 compared to mono-cropped plots (Trechus discus: 681 vs 442, Clivina fossor: 160 vs 137, Bembidion guttula: 108 vs 61). Rye grass provided more green manure than clover Trifolium spp. (49-412 vs 216-474 g/m²). A 1 ha block of the field was divided into 25 x 25 m plots with treatments in a 4 x 4 Latin square design. Clover or grass were sown straight after the cereal and were ploughed into the soil as a green manure in the autumn. Plant biomass was sampled within 0.25 x 0.5 m quadrats just before ploughing. Emergence rates of ground beetles were sampled using enclosures (0.5 x 0.5 m²) with four pitfall traps. Traps were emptied every seven days over a sequence of three (1989) or five (1990 and 1991) trapping periods of approximately one month (June-September).



A replicated, controlled, randomized study of ground beetles (Carabidae) in arable fields in Finland (Huusela-Veistola 1996) (same study as (Huusela-Veistola 1998)) found no significant difference in beetle abundance between conventional and integrated farming practices (including undersowing with grass/clover Trifolium spp.). Abundance was higher with reduced pesticide applications. There were six replicate blocks and treatments (in 0.7 ha plots) which were fully randomized within blocks (one treatment combination/plot). Treatments were conventional pesticide applications, reduced pesticides or no pesticides (control) and customary or integrated (including undersowing) cultivation. Beetles were sampled with pitfall traps at 12, 66 and 120 m into each crop 8-10 times (one week/sample) between sowing and harvest.



A replicated, controlled, randomized study of spiders (Araneae) in arable fields in Finland (Huusela-Veistola 1998) (same study as (Huusela-Veistola 1996)), found no significant difference in spider abundance between conventional and integrated farming practices (including undersowing with grass/clover Trifolium spp.). Abundance was higher with reduced pesticide applications. There were six replicate blocks and the treatments (in 0.7 ha plots) were fully randomized within blocks (one treatment combination/plot). Treatments were conventional pesticide applications or reduced pesticides and customary or integrated (including undersowing) cultivation. Spiders were sampled with pitfall traps at 12, 66 and 120 m into each crop 8-10 times (one week/sample) between sowing and harvest.



A replicated study in summer 1995 in southern England (Wakeham-Dawson et al. 1998) found that the density of singing Eurasian skylark Alauda arvensis was higher on undersown spring barley fields than on any other field type (approximately 22 birds/km2 on four spring barley fields vs 2-15 birds/km2 on 85 other fields). Other field types were arable fields reverted to species-rich or permanent grassland, downland turf (close-cropped, nutrient-poor grassland), permanent grassland, winter wheat, oilseed rape and set-aside. The number and location of singing skylarks were recorded in May-June 1995 on 89 fields.


A 2000 literature review (Holland & Luff 2000) looked at which agricultural practices can be altered to benefit ground beetles (Carabidae). It found just one study (Vickerman 1978) showing that some ground beetle species benefit from undersowing spring cereals, and that emergence the following spring is higher than in cereal fields.



A study of paired, intercropped and conventional wheat fields at four sites in the UK (Schmidt et al. 2001) found that intercropping resulted in higher earthworm (Lumbricidae) abundance, biomass and species diversity than conventional wheat management. Earthworm populations and biomass were greater in wheat-clover Trifolium spp. fields (individuals: 548/m², biomass: 137 g/m²) than conventional wheat fields (194/m²,36 g/m²) from autumn 1995-1997. Abundance varied more between conventional sites (55-408/m²) than between wheat-clover sites (337-733/m²). Population size ratios (wheat-clover:conventional wheat) ranged from approximately 2:1 to 9:1 and the overall mean ratio was 4:1. Species diversity was greater in wheat-clover fields (7-10 species) than conventional fields (5-9 species). White clover Trifolium repens was established in spring, and winter wheat was direct-drilled into the clover sward. Mono-cropped wheat was drilled at the same time. Intercropped fields received reduced applications. Earthworm communities were sampled in spring and autumn using the formalin method (10-12 quadrats of 0.25 m²/field) and an electrical sampling method (5-10 samples of 0.125 m²/field). Community biomass values refer to the live biomass.



A 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) in the UK from 1998 to 2001 found that undersown spring cereals did not benefit plants or invertebrates. The undersown cereals could be preceded by overwinter stubble or followed by a grass or grass/clover ley. There were 148 ha and 470 ha of this option in total in East Anglia and the West Midlands respectively. The effects of the pilot scheme on plants, invertebrates (bumblebees Bombus spp., true bugs (Hemiptera), ground beetles (Carabidae), sawflies (Symphyta)) were monitored over three years, relative to control areas.

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.


A replicated study in the winters of 1997-1998 and 1998-1999 on 122 stubble fields on 32 farms in central England (Moorcroft et al. 2002) found that of five bird species using stubble fields, only one species, woodpigeon Columba palumbus was found most frequently on undersown organic wheat Triticum spp. stubbles. Eurasian linnet Carduelis cannabina, Eurasian skylark Alauda arvensis, reed bunting Emberiza schoeniclus and corn bunting Miliaria calandria were found more frequently on intensively-farmed barley Hordeum spp. stubbles than intensive or undersown organic wheat. Weed seed densities in March were highest on undersown organic wheat stubble fields compared to intensive barley or wheat stubbles. Weed seed density decreased the least on undersown organic wheat stubbles between October and March compared to intensive barley or wheat stubbles (11% decline on undersown organic wheat stubbles, 23% decline on intensive wheat stubbles, 35% decline on intensive barley). Seventeen stubble fields contained organic wheat with the previous crop undersown with rye grass Lolium spp. and white clover Trifolium repens. Sixty-seven fields were managed for intensive wheat and 38 fields for intensive barley, both intensively-managed crops received inorganic fertilizer and pesticide applications. Each study field was either overwintering as stubble or entered into the first year of a set-aside scheme. Plants were surveyed in forty 20 x 20 cm quadrats in each field in October. Seed densities were recorded in 27 fields from 10 soil cores/field in October 1997 and March 1998. Birds were surveyed monthly on parallel transects.




A replicated study in the summers of 1999-2000 comparing ten different conservation measures on arable farms in the UK (Critchley et al. 2004) found that undersown spring cereals had more plant species than seven other conservation measures, but were not considered one of the best options for conservation of annual herbaceous plant communities. Average numbers of plant species in the different conservation habitats were undersown cereals 5.9, wildlife seed mixtures 6.7, uncropped cultivated margins 6.3, naturally regenerated grass margins 5.5, no-fertilizer conservation headlands 4.8, spring fallows 4.5, sown grass margins 4.4, overwinter stubbles 4.2, conservation headlands 3.5, grass leys 3.1. Plants were surveyed on a total of 294 conservation measure sites (each a single field, block of field or field margin strip), on 37 farms in East Anglia (dominated by arable farming) and 38 farms in the West Midlands (dominated by more mixed farming). The ten habitats were created according to agri-environment scheme guidelines. Vegetation was surveyed once in each site in June-August in 1999 or 2000, in thirty 0.25 m2 quadrats randomly placed in 50-100 m randomly located sampling zones in each habitat site. All vascular plant species rooted in each quadrat, bare ground or litter and plant cover were recorded.


A controlled study in May to September 2000 in the sub-urban area of Vienna, Austria (Kromp et al. 2004) found no difference in the number of ground beetle (Carabidae) species between one rye field undersown with a wildflower mix (approximately 20 species) and four conventional fields without wildflowers (approximately 12-26 spp.). The number of ground beetle species in the undersown rye field was lower than on two types of set-aside land (unsown or sown with a wildflower mix). No statistical analyses were presented in this paper. Typical crops for the region were sown on five arable fields. One of the fields was under conservation contract growing a wildflower seed mix undersown with 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).


A randomized, replicated, controlled trial in 2003 to 2006 on four farms in southwest England (Defra 2007) (same study as (Potts et al. 2009, Holt et al. 2010)) found that 50 x 10 m plots of permanent pasture sown with spring barley Hordeum vulgare and a grass and legume mix attracted more bumblebees Bombus spp. and adult butterflies (Lepidoptera) than control plots. However undersown barley plots had either similar numbers (for suction trapped beetles (Coleoptera), ground beetles (Carabidae), spiders (Araneae), grasshoppers and crickets (Orthoptera), flies (Diptera), butterfly larvae or sawfly larvae (Hymenoptera: Symphyta), slugs (Gastropoda)) or fewer numbers (true bugs (Hemiptera), planthoppers (Auchenorrhyncha)) of other invertebrate groups than control plots. Control plots were managed as silage, cut twice in May and July, and grazed in autumn/winter. Small insect-eating birds (dunnock Prunella modularis, wren Troglodytes troglodytes and European robin Erithacus rubecula) and seed-eating finches (Fringillidae) and buntings (Emberizidae) preferred undersown cereal plots to control plots for foraging. Dunnock, but not chaffinch Fringilla coelebs or blackbirds Turdus merula, nested in hedgerows next to the sown plots more than expected, with 2.5 nests/km, compared to less than 0.5 nests/km in hedges next to experimental grass plots. There were twelve replicates of each management type, monitored over four years. More information on the use of these plots by bumblebees and butterflies is described in (Potts et al. 2009).



A replicated, controlled, randomized study of undersown and conventional cereal systems in Denmark (Gravesen 2008) found that undersown crops had higher money spider (Linyphiidae) web density, adult Bathyphantes gracilis and Tenuiphantes tenuis (both money spiders), springtail (Collembola) and vegetation density compared to conventional crops. Web density was higher in undersown crops (unfertilized: peak 250-300/m², low fertilizer input: 200-250/m²) than conventional crops (low fertilizer input: 150-200/m², high-input: 100–150/m²). More adult Bathyphantes gracilis were found in undersown crops (5 individuals/m²) and Tenuiphantes tenuis in unfertilized undersown crops (4/m²) compared with the high-input conventional system (1/m²). Springtail density was significantly higher in the fertilized (2350 individuals/m²) than unfertilized undersown crops (1600/m²) and conventional crops (low-input: 1250/m², high-input: 300/m²). Sixteen experimental plots (12 x 50 m) were established in a randomized block design. Treatments were wheat with clover Trifolium spp. undersown, without or with nitrogen fertilization (50 kg/ha), or conventional wheat with low (50 kg/ha) or high nitrogen fertilization (160 kg/ha), only the latter received pesticide applications. Money spider web densities, vegetation density (lower layer only, i.e. clover and weed layer) were sampled between May-October 1995-1997. Money spiders and springtails were sampled in 1996.


A randomized, replicated, controlled trial from 2003 to 2006 in southwest England (Potts et al. 2009) (same study as (Defra 2007, Holt et al. 2010)) 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, numbers of common butterflies (Lepidoptera) and common butterfly species were higher 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. The two legume-sown treatments comprised either spring barley Hordeum vulgare undersown with a grass and legume mix (white clover Trifolium repens, red clover T. pratense, common vetch Vicia sativa, bird’s‐foot trefoil Lotus corniculatus and black medick Medicago lupulina) cut once in July, or a mix of crops including linseed Linum usitatissimum and legumes, uncut. Seven management types involved different management options for grass-only plots, including mowing and fertilizer addition. 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.



A replicated site comparison study from 2004 to 2008 in England (Ewald et al. 2010) found measures of grey partridge Perdix perdix populations were negatively related to the proportion of sites covered by undersown spring cereals (following overwinter stubbles). There were significant negative relationships with year-on-year density changes in 2006-2007 and with overwinter survival rates in conjunction with overwinter stubbles (across all years combined and significantly in 2004-2005 and 2005-2006). There were no relationships with brood size or the proportion of young birds to old. Spring and autumn counts of grey partridge were made at 1031 sites across England as part of the Partridge Count Scheme.



A replicated study from April-July in 2006 on four livestock farms in southwest England (Holt et al. 2010) (same study as (Defra 2007, Potts et al. 2009)) found that dunnock Prunella modularis, but not Eurasian blackbird Turdus merula or chaffinch Fringella coelebs, nested at higher densities in hedges alongside field margins sown with either barley undersown with grass and clover Trifolium spp. or wild bird seed crops, compared to those next to grassy field edges under various management options (dunnock: approximately 2.5 nests/km for seed crops vs 0.3/km for grass margins, blackbird: 1.0 vs 1.3, chaffinch: 1.5 vs 1.4). There were three replicates/farm. Margins were 10 x 50 m and located adjacent to existing hedgerows. Seed crop margins were sown with barley (undersown with grass/legumes) or a kale/quinoa mix. There were 12 replicates of each treatment, three replicates on each farm.


Referenced papers

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. (2017) Farmland Conservation Pages 245-284 in: W.J. Sutherland, L.V. Dicks, N. Ockendon & R.K. Smith (eds) What Works in Conservation 2017. Open Book Publishers, Cambridge, UK.