Conservation Evidence strives to be as useful to conservationists as possible. Please take our survey to help the team improve our resource.

Providing evidence to improve practice

Action: Provide supplementary food for birds or mammals Farmland Conservation

Key messages

Read our guidance on Key messages before continuing

Supporting evidence from individual studies


A replicated and controlled study in mixed farmland in north-east Scotland between 1971 and 1973 (Yom-Tov 1974) found that carrion crow Corvus corone nestlings in nests provided with supplementary food had significantly higher hatching, survival and fledging rates than those in control (unfed) nests. With human ‘predation’ included: 79% of fed nests hatching at least one chick (n = 11), 71% having at least one chick surviving for ten days and 71% fledging at least one chick (n = 10) compared to 61% (n = 28), 54% (n = 15) and 43% (n = 12) for controls. With human ‘predation’ excluded: 92% of fed nests hatching at least one chick (n = 12), 83% having at least one chick surviving for ten days and 83% fledging at least one chick (n = 10) compared to 55% (n = 22), 55% (n = 12) and 45% (n = 10) for controls. Nestlings from fed nests, however, were no heavier than those from controls, when comparing first-hatched with first-hatched etc. Supplementary food consisted of one domestic hen’s egg and five dead hen chicks provided every day from when laying began and increasing to one egg and ten chicks from the seventh day after hatching until fledging. Other experiments in this study found that winter feeding (a hen’s egg and five chicks provided between January and April 1973) led to crows laying clutches earlier but did not affect clutch size (average laying date of April 13th and 4.4 eggs/clutch for fed territories, n = 10 vs April 18th and 4.3 eggs/clutch for controls, n = 21). Further experiments examined the effect of moving supplementary food further from nests, but the author argues that these results are confounded by supplementary food being taken by non-target birds. Finally, additional experiments in the study found that crow nesting density did not increase following the provision of supplementary food and additional nesting sites in breeding territories.



A small study of feeding as a management option for grey partridges Perdix perdix at an arable farm in France (Westerskov 1977), found that partridge density was higher in an area with ‘partridge cafeterias’, than the area without. In spring 1973, the population on the 424 ha farm was 71 pairs (1 pair/6 ha) and four single birds. In spring 1974, a total of 48 pairs (1 pair/4.7 ha) and four single birds were recorded in the southern section (224 ha), where 27 partridge cafeterias had been constructed. The northern section (200 ha), with no cafeterias, had 24 pairs (1 pair/8.3 ha). Cafeterias comprised a barrel with a feed mixture (grain and weed seeds), a mini-midden to provide maggots and insects and a sand-bath, sheltered by a leaning roof that collected rainwater in a drinking trough. Stoats Mustela erminea and mice Mus spp. were also controlled with traps at the ‘cafeterias’. Small shrubs were planted next to cafeterias to provide shelter. Where possible they were placed one/territory.



A replicated and controlled study in the breeding seasons of 1985-1987 in grasslands on Öland, southern Sweden (Moreno 1989), found that female northern wheatear Oenanthe oenanthe, but not males, that were provided with supplementary food were significantly heavier than unfed controls (average of 26.9 g for 53 fed females and 24.4 g for 42 fed males vs 24.3 g and 23.7 g for 48 and 32 unfed controls). However, there was no effect when females were feeding older chicks, which were able to regulate their body temperature. A few days after hatching, most food was delivered to chicks, not consumed by adults. Food consisted of 7 g of mealworms provided either during incubation, or for the entire breeding season.



A replicated and controlled study in grasslands in southern Sweden between 1982 and 1990 (Källander & Karlsson 1993) found that common starling Sturnus vulgaris supplied with supplementary food showed only occasional differences in egg weight, no differences in clutch size or nestling weights and fledging rates were actually lower in fed nests in 1990 (4.3 young/nest for fed nests vs 5.6 for controls). However common starlings supplied with food began laying significantly earlier than controls (first laying date 21st April-5th May for fed nests vs 22nd April-10th May for controls). There were no such differences between nests in the years when supplementary food was not supplied. Supplementary food consisted of approximately 100 g of mealworms placed in small feeders either on the outside or inside of nest boxes, supplied to different colonies in 1982 and 1985 and a subset of nests at a third colony in 1990. Feeding began approximately one month before laying started and stopped once all females began laying. Feeding represented more than the daily energetic needs of a pair of starlings.



A study of habitat use by yellowhammer Emberiza citrinella in 1997 on a mixed farm in Leicestershire UK (Stoate & Szczur 1997), found that in winter supplementary feeding sites were used more than cereal and rape crops and in late winter more than wild bird cover. In early winter, cereal-based wild bird cover was used significantly more than all other habitats and supplementary feeding sites, kale-based wild bird cover and field boundaries were used more than cereal and rape crops. In late winter, supplementary feeding sites were used significantly more than all other habitats including wild bird cover. A 15% area of the arable land was managed for game birds and in winter grain was distributed along some hedges and supplied in hoppers at permanent feeding sites. A 60 ha area of the farm was walked seven times in November-December and February-March 1997 and habitat use was recorded.



A randomized, replicated and controlled study in heathland on Orkney Mainland, Scotland, in 1999-2000 (Amar & Redpath 2002) found that male hen harriers Circus cyaneus provided with supplementary food (chicken Gallus domesticus chicks and quarter pieces of European rabbit Oryctolagus cuniculus or brown hare Lepus europaeus) bred with significantly more females than control (unfed) males (100% of 11 fed males mated and 36% mated with more than one female vs 80% of nine unfed males mated, 11% mated with more than one female). There was no effect of feeding on clutch size or hatching success (average of approximately 5.1 eggs/clutch for 13 fed clutches vs 4.7 eggs/clutch for four unfed clutches), but productivity still increased. Hooded crows Corvus cornix were also removed from all territories.


A replicated cross-over study in 1999-2000 in ten mixed agricultural and natural habitat sites in Norfolk and Suffolk, England (Browne & Aebischer 2002), found that European turtle dove Streptopelia turtur reproductive success, territory size or territory density did not differ between years when supplementary food was provided and control (unfed) years (24 nests studied, daily survival rates of 79-97% for fed nests vs 85-98% for unfed). However, doves were frequently observed eating the food. The authors argue that the experimental sites were too small (mostly 200-400 ha) to affect the wide-ranging doves.



A replicated study of four house sparrow Passer domesticus populations in mixed farmland in Oxfordshire, UK (Hole et al. 2002) found that supplementary feeding in winter increased overwinter survival of a declining population, but not three populations understood to be stable. Monthly overwinter survival rate varied between populations, with that of the declining population lower than the other three populations (0.8 vs 0.9). The apparent survival rate of the declining population over the November-March period increased from 0.39 in 1998-1999 to 0.65 in 1999-2000. There was no effect of supplementary feeding on the other three populations. Three populations were selected at random and the fourth was selected for the availability of historical records, which indicated an 80% decline over the last 30 years. Populations were 6-24 km apart. Supplementary seed food was freely provided to the populations during winter 1999-2000. Nest recording, mark re-sighting and microsatellite-based molecular genetics were used for sampling.



A small replicated controlled study from May-June 1992-1998 in Leicestershire, UK (Stoate 2002), found that the abundance of nationally declining songbirds and species of conservation concern significantly increased on a 3 km2 site where supplementary food was provided from hoppers and by hand (alongside several other interventions). However there was no overall difference in bird abundance, species richness or diversity between the experimental and three control sites. Numbers of nationally declining species rose by 102% (except for Eurasian skylark Alauda arvensis and yellowhammer Emberiza citrinella). Nationally stable species rose (insignificantly) by 47% (eight species increased, four decreased). The other interventions employed at the same site were managing hedges, beetle banks, wild bird seed cover strips, predator control and reducing chemical inputs generally.



A replicated, controlled paired site study from March-August in 2000-2003 in 20 paired nest box groups (10 placed along wetland edges and 10 in farmlands) in Rutland, England (Field & Anderson 2004) found that tree sparrow Passer montanus showed no preference for nest boxes supplied with supplementary food (four fed boxes colonized vs four unfed). There was no difference in the number of nesting attempts made by birds with or without supplementary food although the mean clutch size was significantly higher in nests closer to supplementary food (5.6 compared to 5.0 eggs/clutch). The authors point out that the small spatial scale of the study (1 km between pairs) may have confounded any effect of supplementary feeding. Nest box groups consisted of five nest boxes placed 2-20 apart. Sunflower seeds were randomly provided to one nest box group within each pair.


A study at a farmland site in northwest England between January 2003 and February 2004 (Raine 2004a) found that twite Carduelis flavirostris used a supplementary feeding station (established in spring 2002) frequently outside the breeding season, with up to 250 birds seen at once. However, twite used the station far less during the breeding season, when they relied more on wild seeds. Birds from another feeding station (see (Raine 2004b)) and other breeding colonies up to 20 km away used the feeding station, as well as individuals from a nearby colony of 20-30 birds. Supplementary food consisted of nyjer Guizotia abyssinca seed spread in a thick 2 m x 5 cm line on a 2 m x 2 m patch of bare earth and replenished every week. This study was part of the same experimental set-up as (Raine 2004b, Raine 2004c).



A study at a farmland site in northwest England between January 2003 and February 2004 (Raine 2004b) used an identical procedure to (Raine 2004a) at a site 12.6 km away and found that twite Carduelis flavirostris used the supplementary feeding station frequently outside the breeding season, with up to 150 birds being seen at once. However, twite used the station far less during the breeding season, when they relied more on wild seeds. A large number of birds from near the feeding station in (Raine 2004a) and colonies up to 20 km away used the feeding station, as well as birds from the two nearby colonies (each approximately 1.5 km away and 20-30 birds). This study was part of the same experimental set-up as (Raine 2004a, Raine 2004c).



A study at a farmland site in northwest England between January 2003 and February 2004 (Raine 2004c) used an identical procedure to (Raine 2004a) to establish a feeding station approximately 1 km from a colony of six pairs of twite Carduelis flavirostris. This station was only used occasionally and only by one or two birds at a time. The author suggests that the lack of use could have been due to the small size of the colony and the fact that it was not positioned close to natural feeding areas for twite. This study was part of the same experimental set-up as (Raine 2004a, Raine 2004b).



A randomized, replicated and controlled study at three farmland sites in England in the winters of 1999-2000 until 2001-2002 (Robinson et al. 2004) found that farmland birds showed mixed responses to supplementary food. Chaffinch Fringella coelebs, linnet Carduelis cannabina and yellowhammer Emberiza citrinella all showed significant short-term increases on at least one plot provided with food (chaffinch densities increased by 80-200% on three of six fed plots, yellowhammer densities increased by 230-400% on four of six fed plots, data for linnets not provided). There were no corresponding short-term changes on nearby control plots. Skylark Alauda arvensis did not show any consistent response to food at any of the sites and there was no longer term impact of feeding on bird densities. Supplementary food consisted of 36 kg/ha of mixed grains broadcast over fields. The authors suggest that the lack of effect of feeding at some sites may be due to a very low natural seed density in the soil, meaning that even with supplementary food, the level of food was too low to attract birds.



The results from two replicated studies from the UK found that the factors affecting the use of supplementary food by a range of farmland songbirds were not consistent across species or regions (Siriwardena & Stevens 2004). The ‘Bird Aid’ programme (run between October and March in the winters of 2000-2001 until 2002-2003 across the UK) found that all three target species (tree sparrow Passer montanus, yellowhammer Emberiza citrinella and corn bunting Miliaria calandra) used supplementary food, consisting of 25 kg of seeds supplied each week. Tree sparrow and yellowhammer tended to use feeding stations more if they were closer to cover and in mixed landscapes, the opposite was true for corn bunting. The Winter Food for Birds project, run from October 2002 to March 2003 at ten replicates of seven sites across eastern England, found that six of eight target species used supplementary food, consisting of 5 kg each of millet and sunflower seeds supplied each week, sufficiently often for analysis. At both the local and landscape scale, only human habitats and woodlands had uniform effects, increasing and decreasing the use for three and four species respectively. All other habitats had different impacts on different species. Results from the same experimental set-up are also presented in (Defra 2005, Siriwardena et al. 2006, Defra 2007, Siriwardena et al. 2007, Siriwardena et al. 2008).



A replicated, controlled study from November-July 2002-2004 in 10 sites each containing seven feeding stations (placed at the centre of a 2 x 2 km tetrad) separated at set distances from each other (100 m, 500 m, 1 km, 2 km, 5 km and 10 km) in East Anglia, UK (Defra 2005) found that supplementary provision of seeds increased local seed-eating bird abundance, especially species of conservation concern. Yellowhammer Emberiza citrinella and chaffinch Fringilla coelebs used the feeding stations most extensively (93–100% of all stations). Although genuine population trends were difficult to infer from the experimental setup, the authors argue that food provisioning increased the local abundance of several otherwise declining species (yellowhammer, reed bunting Emberiza schoeniclus, house sparrow Passer domesticus and chaffinch) over two winters. Colour-ring re-sighting and radio-tracking revealed that target seed-eaters move small distances between food resources (500m – 1km) and the authors suggest placing food resources (overwinter stubbles and wild bird cover crops) at a minimum of 1 km apart in order to be cost effective in reaching the largest number of populations. Supplementary seed (10 kg of equally distributed sunflower hearts and millet) was replenished weekly. Bird use of the feeding stations was monitored twice weekly (20 min observation sessions). Results from the same experimental set-up are also presented in (Siriwardena & Stevens 2004, Siriwardena et al. 2006, Defra 2007, Siriwardena et al. 2007, Siriwardena et al. 2008).



A replicated, controlled, randomized paired study of 15-16 nest-box colonies of starling Sturnus vulgaris in 1998-1999 in southern Sweden (Granbom & Smith 2006) found that food supplementation increased growth and survival of nestlings. Greater availability of pasture also increased survival, but tended to have a smaller effect. Fledging success increased with supplemental feeding and local availability of pasture (<10% pasture: fed 0.92, unfed 0.80; >10% pasture: fed 1.0, unfed 0.95). Nestling growth (tarsus length) was significantly higher following supplemental feeding, but was not affected by habitat (<10% pasture: fed 32.2 mm, unfed 32.8 mm; >10% pasture: fed 32.6 mm, unfed 32.9 mm). Feather growth rate showed the same pattern. There was no effect on chick condition and no interactions between effects of habitat and feeding. Colonies were over 1 km apart and each comprised eight boxes. Nest boxes were visited to determine clutch size, hatching date, fledging success and to band (day 1) and measure (day 10 and 14) nestlings. Agricultural land-use was classified in a radius of 500 m around each colony. On the day of hatching, two broods with similar clutch size were matched from different habitat classes and were randomly selected to receive supplemental food or not. From the fourth day after hatching, half of the breeding pairs were given a bowl containing 84 g of mealworm larvae twice daily (1998-1999).



The Winter Food for Birds project (see (Siriwardena & Stevens 2004)) was continued in the winter of 2003-2004 and this study discusses the data from both winters (Siriwardena et al. 2006). For four songbird species (blue tit Parus caeruleus, chaffinch Fringilla coelebs, great tit P. major and robin Erithacus rubecula), feeding stations were used more frequently and by more birds if they were more than 500 m from other stations, compared with stations less than 500 m from neighbours. The same pattern was seen (but not significant) in blackbird Turdus merula and house sparrow Passer domesticus. Yellowhammer Emberiza citrinella and reed bunting E. schoeniclus, however, used clustered sites more. There was no significant impact of distance on feeder use by greenfinch Carduelis chloris, goldfinch C. carduelis or dunnock Prunella modularis. All species used multiple stations if they were closer than 500 m apart, but used only single stations if they were more widely spaced. The authors use this information to recommend that stations are placed at least 1 km apart to maximize cost-effectiveness (i.e. to ensure the maximum number of birds have access to supplementary food). Results from the same experimental set-up are also presented in (Siriwardena & Stevens 2004, Defra 2005, Defra 2007, Siriwardena et al. 2007, Siriwardena et al. 2008).



A replicated, controlled study from November-March 2004-2007 in 10 experimental and 10 control tetrads (composed of four 1 km2 sites) of arable farmland in East Anglia, UK (Defra 2007) found that provision of seeds during winter significantly increased body mass and breeding population sizes of seed-eating bird species. Supplementary food was most used in early to mid-winter for generalist species and late winter for specialist species (such as chaffinch Fringilla coelebs and yellowhammer Emberiza citrinella). Radio tracking and mark-recapture techniques revealed that resource patches (such as wild bird cover crops and overwinter stubbles) should be separated by 1.1-1.3 km to be both cost and conservation effective for priority species (like yellowhammer). The authors suggest that year-round resource delivery could be achieved by placing breeding habitat 2.7-3.6 km from winter food patches. They caution that specific inter-patch distances may vary according to species and habitat but should be based on species of conservation concern. Experimental sites contained one central feeding station provided freely with seed (10 kg of equally distributed millet, rape, wheat and sunflower seeds; replenished twice a week) and were fenced (50 cm in height) using 50 mm mesh wire and bamboo canes to exclude gamebirds. This study was an extension of and used partly the same experimental set-up as (Siriwardena & Stevens 2004, Defra 2005, Siriwardena et al. 2006, Siriwardena et al. 2007, Siriwardena et al. 2008).



A series of randomized, replicated trials at two sites in England in the winters of 2000-2001 and 2001-2002 (Perkins et al. 2007) found that five songbird species took supplementary food when provided and preferentially took wheat over oats and oats over barley. Tree sparrow Passer montanus and reed bunting Emberiza schoeniclus also fed on maize, preferring it to all cereals except wheat, whilst house sparrow P. domesticus preferred maize to all cereals. Corn bunting E. calandra and yellowhammer E. citrinella preferred all cereals to maize. Tree sparrow selected both cereals and oily seeds (e.g. sunflower seeds, oilseed rape) but avoided rye grass seed. All species preferred cereals to sunflower seeds and none showed any distinction between wheat and a ‘weed seed mix’. At one site, food was provided in tubular feeders, in the other it was heaped on the ground. Survival rates of birds were not monitored.



A replicated study using the same data as (Siriwardena & Stevens 2004) and combining it with data from control areas between 2000 and 2003 (Siriwardena et al. 2007) did not find robust evidence for supplementary winter feeding increasing breeding abundances of farmland songbirds. There were no effects of the Bird Aid programme on target species, although sites used more frequently had increased populations of yellowhammer Emberiza citrinella and corn bunting Miliaria calandra, but decreased populations of tree sparrow Passer montanus. Four of five insect-eating/generalist species declined faster in Winter Food for Birds programme sites than in controls. There was no such effect for six seed-eating species. Declines in dunnock Prunella modularis, robin Erithacus rubecula and yellowhammer Emberiza citrinella were lower in Winter Food for Birds sites provided with more food and centrally-placed Winter Food for Birds sites, compared to those provided with less food or those around the periphery of Winter Food for Birds clusters. Results from the same experimental set-up are also presented in (Siriwardena & Stevens 2004, Defra 2005, Siriwardena et al. 2006, Defra 2007, Siriwardena et al. 2008).



A replicated controlled trial in 1999-2002 on arable fields on three farms in Hampshire, Lincolnshire and Yorkshire, UK (Holland et al. 2008) investigated whether addition of supplementary bird seed affected the abundance of weed seeds on the soil surface by attracting seed-eating birds. The study found no difference in mid-winter weed seed densities in sites with and without addition of bird seed, even though a separate study at the same site recorded more birds on areas with added seed. This may have been due to low winter seed predation by birds. Post-treatment winter seed densities varied from 151 to 398 seeds/m2 on control plots and 92 to 365 seeds/m2 on treatment blocks. There was some evidence that birds changed the species composition of weed seeds, as there were fewer larger-sized weed seeds in treatment blocks. Bird seed (36 kg/ha, 444 seeds/m2, including cracked maize and linseed/soya/barley or sorghum/millet) was applied three times from November-March on two 100 ha blocks, with a further two control blocks, on each farm. Seed densities were estimated from 10 soil scrapes (0.2 x 0.2 m), from at least 10 field edge and 10 mid-field locations in each block, before the first bird seed application and 2-3 weeks after each application.



A further study, using the same data as (Siriwardena & Stevens 2004) investigated how use of supplementary food by farmland songbirds varied over winter months (Siriwardena et al. 2008). Supplementary food-use peaked in or before January for five generalists and ‘human-associated’ seed-eating species (blackbird Turdus merula, goldfinch Carduelis carduelis, greenfinch C. chloris, house sparrow Passer domesticus and robin Erithacus rubecula), whilst yellowhammer Emberiza citrinella, reed bunting E. schoeniclus, chaffinch Fringilla coelebs and dunnock Prunella modularis all used supplementary food most in February or later. Use by great tits Parus major and blue tits P. caeruleus declined overwinter. The authors suggest the first group use food when temperatures are lowest and daylight hours shortest, whilst the second group (which are heavily dependent on farmland seed) use food when naturally-occurring food sources are at their lowest. They caution that these results are likely to be dependent on the mix of farming types across the landscape, with eastern England being dominated by arable fields. Results from the same experimental set-up are also presented in (Siriwardena & Stevens 2004, Defra 2005, Siriwardena et al. 2006, Defra 2007, Siriwardena et al. 2007).



A replicated and controlled habitat selection study in four riparian field margins in the municipality of Jokioinen, south-western Finland (Yletyinen & Norrdahl 2008) found that supplementary food appeared to attract field voles Microtus agrestis in uncut plots in both wide buffer zones and narrow filter strips but not in mowed plots. Mowed food plots in wide buffer zones were avoided by voles. In mid-June 2005, one 210 m-long section in each margin was divided into fourteen 15 m-long experimental plots, half of which were mown to <20 cm. Food and/or cover was added to mowed/unmowed plots to create eight treatments. The remaining plots were interspersed between experimental plots. Trapping and radio-tracking field voles started two weeks after habitat manipulation.



A controlled study in 2002-2009 on mixed farmland in Hertfordshire, England (Aebischer & Ewald 2010), found that the number of grey partridges Perdix perdix increased significantly on an experimental site where supplementary food was provided (along with several other interventions), but only slightly on a control site without supplementary food. This increase was apparent in spring (from fewer than three pairs/km2 in 2002 to 12 in 2009, with a high of 18 pairs/km2 on the experimental site, compared to approximately one pair/km2 on the control site in 2002, increasing to approximately four pairs/km2 in 2009) and autumn (from fewer than 10 birds/km2 in 2002 to approximately 65 in 2009, with a high of 85 birds/km2 compared to approximately four birds/km2 on the control site in 2002, increasing to approximately 15 in 2009). Food consisted of wheat from a hopper, provided from October to March. The experimental site also had predator control and habitat creation.


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. (2019) Farmland Conservation Pages 291-330 in: W.J. Sutherland, L.V. Dicks, N. Ockendon, S.O. Petrovan & R.K. Smith (eds) What Works in Conservation 2019. Open Book Publishers, Cambridge, UK.