AHW2Supplementary winter bird food

What it actually is

GOV.UK's stated aim is that over the winter until mid-spring there's a mix of seeds spread on the ground at multiple feeding areas.

You can only apply for AHW2 if CAHL2, winter bird food on arable and horticultural land, is in the same SFI26 application, and the tonnage you can claim is capped at one tonne for every two hectares of CAHL2 you are doing.

Source: GOV.UK, AHW2 action page, checked 02/08/2026.

What it pays, and what that comes to

The rate is in the panel at the top of this page, and it is per tonne rather than per hectare. What you can claim follows from your CAHL2 area:

Source: the 1 tonne per 2 hectares cap, GOV.UK AHW2 action page, checked 02/08/2026, with the payment rate from Tilth's copy of the GOV.UK action catalogue. The 4 ha and 10 ha figures are illustrations; the rest is arithmetic on the rate and the cap.

Is it worth it on your ground?

Most action guides on this site compare the payment against what a three-year rotation would have returned on the same hectares. This one does not work that way, and there are two reasons. The ground it happens on is a track or a piece of hard standing rather than cropping ground, and the hectares it depends on are CAHL2 hectares, which have already been taken out of production and already had that comparison made on the CAHL2 page. Nothing further comes out of the rotation to do this.

So the whole question is narrower: what does a tonne of compliant mix cost you, and what is left over.

Route one: buy it ready mixed.

Source: seed price, Boston Seeds' own listed price for the AB12 Boston Basic Mix, looked up 02/08/2026, read in the inclusive-of-VAT state. The payment rate is from Tilth's copy of the GOV.UK action catalogue. The left-over figure is subtraction.

Route two: buy the small seeds and put your own cereals to it. The requirement is that cereals are no more than 70 per cent of the mix by weight and small non-cereal seeds are at least 30 per cent. So one route is to buy only the 30 per cent and make the rest up from grain in your own store.

Source: the 70/30 requirement, GOV.UK AHW2 action page, checked 02/08/2026, with the payment rate from Tilth's copy of the GOV.UK action catalogue. Gold Mix price, Boston Seeds' own listed price, looked up 02/08/2026, read in the inclusive-of-VAT state; the product page states it is intended to be mixed with 70 per cent of your own cereals. Own wheat valued at £206.00/t, AHDB Arable Market Report, 27/07/2026, East Anglia delivered, November 2026 delivery.

What is left has to cover the driving. GOV.UK's advice section, which is explicitly not a requirement, gives an example of feeding once a week for 20 weeks. Divided across that many rounds:

Source: arithmetic on the two tables above. The 20-week figure comes from the advice section of the GOV.UK AHW2 action page, checked 02/08/2026, where it is offered as an example and is not part of the action's requirements. The requirement itself is to spread at least once a week from over the winter until mid-spring, which GOV.UK says is usually early December to late April.

Where that leaves it.

The paragraph that follows is Tilth's own reading of the arithmetic above. It is not scheme guidance and it is not sourced from Defra or the RPA.

On the ready-mixed route the payment covers the seed and leaves £42 a tonne, out of which come your own time and diesel across the feeding rounds, in winter weather. On the route where the cereal fraction comes out of your own store, the same job leaves £272.80, because you are being paid for a tonne that includes 700 kg of grain valued at the price you would have got for it anyway. Which of those two you are looking at depends on your quote, your VAT position, and whether you have suitable grain and a way to mix it.

Does it work, and what makes it work better

This section is what we have found in published sources about the practice this action pays for. It is not scheme guidance, it is not Defra's or the RPA's view, and it adds no requirement beyond what the GOV.UK action page sets out. We have named every source and given the date we checked it. What it means for your ground is your decision.

A word about fit before anything else, because it changes how the numbers below should be read. Conservation Evidence hold two assessments of supplementary feeding for songbirds, and both sit in their Bird Conservation synopsis rather than the Farmland Conservation one. That synopsis covers songbirds everywhere, and a large part of both assessments is garden and woodland feeder work from North America, Europe, Japan and Scandinavia: nest boxes, clutch size, egg mass, chick condition, feeder height, sunflower seed in a hanging feeder. None of that is this action.

Inside the survival assessment, though, there is a run of British farmland studies that is close to AHW2. Weekly seed put out on English arable land through the winter, in landscape-scale experiments funded by Defra in East Anglia, in a UK-wide feeding programme, and in broadcast grain trials on three English farms. That run is what this section is built on. Where a finding comes from a garden or a wood we have said so on the line. We have also drawn on Farm Wildlife, a partnership of wildlife organisations that publishes farm management advice in one place, because it is British, it is written for this job, and it puts figures on the practical side that the studies do not.

The graded assessment. Conservation Evidence, based at the Department of Zoology at the University of Cambridge, summarises the published studies behind individual conservation actions and has independent subject experts score each one. Feeding songbirds to increase adult survival comes out at Beneficial, on 34 studies, with 65 per cent for effectiveness, 75 per cent for certainty and 0 per cent for harms. Feeding songbirds to increase reproductive success comes out a category lower at Likely to be beneficial, on 37 studies, with 51 per cent for effectiveness, 85 per cent for certainty and 6 per cent for harms. Effectiveness scores only the size and direction of the effects reported, and Conservation Evidence say themselves that it does not consider the quantity or quality of the studies, so a single poorly designed study could generate a high score. Certainty is the separate score for how many studies there are, how well built they are and how much ground they cover. Both scores answer the whole songbird question, gardens included, so neither is a score for this action.

Source: Conservation Evidence, action 552, Provide supplementary food for songbirds to increase adult survival, and action 537, Provide supplementary food for songbirds to increase reproductive success, both Bird Conservation synopsis, checked 06/08/2026. Cite as Williams et al. (2020), Bird Conservation, pages 137 to 281, in What Works in Conservation 2020, Open Book Publishers.

What the studies found, both ways. On the survival assessment as a whole, seven studies from Europe and the USA found higher densities or larger populations of songbirds near supplementary food, six from Europe, Canada and Japan found no difference between fed and unfed areas, and an American study found populations followed the food, rising after feeders went up and falling after they came down. Four studies from Canada, Europe, Japan and the USA found higher survival with food, though in two of them only in females or in one of two species. A controlled American study found no sign of dependency: when the food was taken away, previously fed birds did not survive worse than controls. Against that, a replicated controlled American study found song sparrows had lower survival where there were feeding stations in their territories. Six studies found better condition or larger fat reserves, with caveats in four of them, including one species that was in worse condition when fed. Of thirteen studies on how much the food gets used, four found high use, with up to 21 per cent of birds' daily energy needs coming from feeders, while another British study found very low use, which the authors put down to the feeder not being near natural food sources.

The British farmland studies inside that set are the ones that speak to this action. Raine set up three feeding stations on farmland in the South Pennines for twite, nyjer spread in a line on a two metre square patch of bare earth and topped up every week. Two of them were used heavily outside the breeding season, up to 250 birds at once at one and up to 150 at the other, drawing birds from colonies up to 20 kilometres away, while use dropped away in the breeding season when the birds went back to wild seed. The third station, about a kilometre from a colony of six pairs, was used only occasionally and by one or two birds at a time, which the author attributed to the small size of the colony and to the station not being positioned close to natural feeding areas. Robinson and others, in a randomised replicated controlled study at three English farmland sites over three winters, broadcast 36 kilogrammes a hectare of mixed grain over fields and found chaffinch densities up by 80 to 200 per cent on three of six fed plots and yellowhammer densities up by 230 to 400 per cent on four of six, with no matching change on the control plots, no consistent skylark response at any site, and no longer term effect on bird densities. Their reading of the sites where nothing happened was that natural seed density in the soil was so low that even with the food added the total was too small to draw birds in.

The two landscape-scale experiments are the largest pieces of work here. Siriwardena and Stevens reported on both. The BirdAid programme ran from October to March across three UK winters on 25 kilogrammes of seed a week, and all three target species, tree sparrow, yellowhammer and corn bunting, used it. The Winter Food for Birds project ran across ten replicates of seven sites in eastern England on 5 kilogrammes each of millet and sunflower seed a week, and six of eight target species used it enough to analyse. Their finding was that the things affecting use were not consistent across species or regions: tree sparrows and yellowhammers tended to use stations more when they were closer to cover and in mixed landscapes, corn buntings the reverse, and only human habitats and woodland had a uniform effect, raising use for three species and lowering it for four. A Defra study across ten East Anglian sites, each with seven feeding stations set at fixed distances from each other, found yellowhammer and chaffinch used 93 to 100 per cent of all stations on 10 kilogrammes of sunflower hearts and millet a week. A second Defra study, across ten experimental and ten control blocks of arable farmland in East Anglia over three winters, found winter seed significantly raised body mass and breeding population sizes of seed-eating species.

The counterweight comes from the same research group. Siriwardena and others, combining the BirdAid data with control areas and adding a second landscape-scale experiment, did not find robust evidence that supplementary winter feeding raised the breeding abundance of farmland songbirds. There were no effects of BirdAid on the target species, though sites used more heavily had higher yellowhammer and corn bunting populations and lower tree sparrow populations. Four of five insect-eating and generalist species declined faster in the Winter Food for Birds areas than in the controls, with no such effect for six seed-eating species, and declines in dunnock, robin and yellowhammer were smaller at sites given more food and at sites placed centrally in the cluster. Stoate, on one three square kilometre farm in Leicestershire against four unmanaged arable farms over seven years, recorded a 102 per cent rise in nationally declining songbirds where grain was put out through the winter, but as part of a package that also included hedge management, wild bird cover and predator control, and with no overall difference in bird abundance, species richness or diversity against the control sites.

On the reproductive success side, the picture is mixed and mostly not about farmland. Twelve studies worldwide found higher breeding productivity where birds were fed, against eleven from Europe and the USA that found no higher or even lower productivity or chick survival. Six found larger clutches against fourteen that found no larger or even smaller ones. Fifteen found earlier nesting or laying against four that found none. Nearly all of it is nest box work in gardens and woods. The one study on English farmland, across twenty paired nest box groups in Rutland, found tree sparrows showed no preference for boxes supplied with food.

Conservation Evidence say themselves that studies are not directly comparable or of equal value, and that study size, design, reported metrics and relevance to your own situation matter more than "simply counting the number of studies that support a particular interpretation". The British farmland studies here run from 1992 to 2008, so all of them predate this scheme and this action.

Source: Conservation Evidence, actions 552 and 537, key messages and supporting evidence from individual studies, checked 06/08/2026. The individual studies named are Raine (2004), Conservation Evidence volume 1; Robinson, Hart, Holland and Parrott (2004), Ibis; Siriwardena and Stevens (2004), Ibis 146; Defra project BD1616 (2005); Defra project BD1628 (2007); Siriwardena, Calbrade, Vickery and Sutherland (2006), Journal of Applied Ecology 43; Siriwardena, Stevens, Anderson, Vickery, Calbrade and Dodd (2007), Journal of Applied Ecology 44; Siriwardena, Calbrade and Vickery (2008), Ibis 150; Stoate (2002), Biodiversity and Conservation 11; and Field and Anderson (2004).

What we have not found. Nothing measuring AHW2 itself. Nothing testing a mix built to this action's own composition rule, the 70 per cent cereal cap with at least 30 per cent small non-cereal seeds. Nothing comparing seed spread on the ground against seed in hoppers for what it does to survival, which is the comparison the 10 per cent hopper limit implies. And nothing costing any of it, so no study puts a figure on what a tonne of seed buys in birds.

What the sources say makes it work better. All of the following comes from the studies listed on those two Conservation Evidence pages and from Farm Wildlife's own guidance, and it is a description of what those sources say rather than advice from us.

Start with when in the winter, because it is the finding that fits this action most exactly. Siriwardena and others, looking at how food use varied month by month across two landscape experiments, found the birds split into groups. Blackbird, goldfinch, greenfinch, house sparrow and robin, the generalists and the ones that live around people, used the food most in or before January. Yellowhammer, reed bunting, chaffinch and dunnock, the group heavily dependent on farmland seed, used it most in February or later. Great tit and blue tit use fell away across the winter. The authors read that as demand: the first group taking food when temperatures are lowest and the days shortest, the second when the seed in the countryside has run out. The Defra East Anglia experiment found the same shape, with the food most used in early to mid winter by generalists and in late winter by specialists such as chaffinch and yellowhammer. They caution that this depends on the mix of farming across the landscape, eastern England being arable dominated.

Then spacing, which is the one point three separate British studies agree on. The first Defra study, using colour-ringed birds and radio tracking, found target granivores move only short distances between food resources, 500 metres to a kilometre, and its authors suggested placing food resources such as overwinter stubbles and wild bird cover at a minimum of a kilometre apart to reach the greatest number of birds for the cost. Siriwardena and others, across three winters of the Winter Food for Birds data, found stations more than 500 metres from their neighbours were used by more birds and more frequently for blue tit, chaffinch, great tit and robin, with the same non-significant pattern in blackbird and house sparrow, and found that local populations divided themselves between multiple stations when those were closer than 500 metres apart but used a single station when they were further apart. Their recommendation was again at least a kilometre. The second Defra study put the figure at 1.1 to 1.3 kilometres, and added that year-round provision could be achieved by placing breeding habitat 2.7 to 3.6 kilometres from winter food. The exception inside the same data is worth stating: yellowhammers and reed buntings used clustered sites more, not less.

On where the food goes, three findings point the same way. Raine's station that failed was the one that was not close to natural feeding areas. Robinson's sites that failed were the ones with almost no natural seed in the soil. Conservation Evidence report a British study of very low feeder use with the same explanation attached, that the feeder was not near natural food. Farm Wildlife's advice runs alongside that: put the feeding area next to an existing resource such as stubble or a seed mix, and pick somewhere you pass anyway on the farm so it actually gets visited. Set against that, Siriwardena and Stevens found habitat effects were not consistent between species, with tree sparrow and yellowhammer favouring stations near cover and corn bunting the opposite, so there is no single answer that suits every bird on the list.

On what is in the mix, the clearest work is a set of randomised replicated preference trials at two English sites over two winters. Five songbird species took the food and preferred wheat to oats and oats to barley. Tree sparrows and reed buntings also took maize, preferring it to every cereal except wheat, house sparrows preferred maize to all cereals, and corn buntings and yellowhammers preferred every cereal to maize. Tree sparrows took both cereals and oily seeds such as sunflower and oilseed rape but avoided rye grass seed. Every species preferred cereals to sunflower seeds, and none of them distinguished between wheat and a weed seed mix. Food was in tubular feeders at one site and heaped on the ground at the other, the two were not compared, and survival was not monitored. Sitting alongside that, the Defra station experiment recorded yellowhammer and chaffinch using 93 to 100 per cent of stations stocked with sunflower hearts and millet, so the small seed was plainly taken even where the preference trials put cereal first.

On how much and how frequently, Farm Wildlife put the figure at between 10 and 25 kilogrammes a week at a feeding site, with half a tonne enough for a winter, and say to feed at weekly intervals at most, and that if seed is left over the answer is to reduce the amount rather than to stretch the interval. The study rates bracket that: 25 kilogrammes a week on BirdAid, 10 kilogrammes a week on Winter Food for Birds and on the first Defra experiment, and 10 kilogrammes twice a week on the second. Robinson's broadcast trial worked to 36 kilogrammes a hectare across whole fields rather than to a per-site figure.

On the ground itself and on keeping it clean, Farm Wildlife say to spread on firm dry surfaces such as farm tracks or short grassland, both to stop weed seed being introduced and so the site drains, and that moving the feeding location around the field reduces the risk of disease spreading and of drawing in rodents. That first half is the same condition the action states about firm free-draining ground. Farm Wildlife also start the clock earlier than this action does, advising that feeding begins in October so birds learn the food is there and runs through March, where the action pays from over the winter to mid-spring, which the GOV.UK page puts as early December to late April. And Farm Wildlife list tailings among the things that can be used, where this action excludes them.

Source: Conservation Evidence, actions 552 and 537, supporting evidence from individual studies, checked 06/08/2026. Farm Wildlife, supplementary feeding page, checked 06/08/2026. GOV.UK, AHW2 action page, checked 02/08/2026.

What follows is Tilth's own reading of how those findings sit against this action, and it adds no requirement beyond what the GOV.UK action page sets out.

The most useful thing in this evidence is not that birds come to seed. It is when. Two independent British datasets found the farmland seed specialists, which are the birds this action exists for, took the food most from February onwards, while the garden birds had already peaked by January. That is the hungry gap arriving in a graph, and it says the hard end of this action, the late rounds in February, March and April when the novelty has gone and the weather has not, is the end that carries the finding. Our reading is that anyone deciding whether to do this should be pricing the late rounds rather than the early ones.

The honest gap sits between two things that sound alike. That birds turn up, feed and put on weight is well evidenced on British arable land. That there are more of them the following year is not. One Defra experiment found winter seed raised body mass and breeding population sizes. The same research group, pooling two large-scale experiments, did not find robust evidence for higher breeding abundance, and found four of five insect-eating and generalist species declining faster in the fed areas than in the controls. Both are large British arable studies from the same decade and they do not agree with each other. We would not present this action as a proven way to raise bird numbers on a farm, and nothing on this page should be read that way.

The spacing findings are the ones we would think hardest about, because they cut across how a tonne actually gets used. Three British studies land on a kilometre or more between food resources, and one puts it at 1.1 to 1.3 kilometres. This action asks for the weight to be divided as equally as reasonable between multiple feeding areas, and on most farms those will end up a good deal closer together than that, because the constraint is a track that is firm in February rather than a map. The studies were measuring the spacing of whole resources across a landscape, stubbles and cover crops as well as feed sites, and were answering a cost-effectiveness question about reaching the most birds. It is not the same question as where to put four drops on one holding. But the numbers exist, and it seems better to know they exist than to assume that clustering the areas near the yard is what the research supports.

There is a rough sum in the quantities that may help. A tonne over the twenty-week example on the GOV.UK page is 50 kilogrammes a week across all the feeding areas together. Set against Farm Wildlife's 10 to 25 kilogrammes a week at a site, that is two to five feeding areas. That is arithmetic on two published figures rather than a rule, and the action states no number, but it puts a shape on what multiple areas looks like at this tonnage.

The composition finding and this action's composition rule pull in slightly different directions. The preference trials found every species tested preferred cereals to sunflower seeds, and the action caps cereals at 70 per cent and requires at least 30 per cent small non-cereal seed. Whether that matters is unclear from the evidence, because the stations stocked with sunflower hearts and millet were used by 93 to 100 per cent of the birds recorded, so the small seed was not being refused. What it does mean is that the second route on this page, buying the 30 per cent and putting your own wheat to the rest, produces a mix whose cereal fraction is exactly what the preference work says the birds went for first. That is two separate constraints happening to agree rather than a finding, but it is a pleasant coincidence to be on the right side of.

What none of it settles is the money, and on this action the money is the whole question. Not one of these studies costed the seed against anything. The comparison earlier on this page, between £42 a tonne left over on the bought mix and £272.80 on the home-mixed route, is arithmetic that no piece of published evidence has anything to add to.

What you actually have to do

Spread a winter supplementary feeding mix at multiple feeding areas, dividing the total weight as equally as you reasonably can between them, at least once a week from over the winter until mid-spring in the following year, which GOV.UK puts as usually early December to late April.

The mix must contain both cereals, not maize, at no more than 70 per cent of the total by weight, and small non-cereal seeds at at least 30 per cent by weight, with no individual species making up more than 50 per cent of the non-cereal element. The small non-cereal part must contain at least three of: canary seed, linseed, oilseed rape, red millet, sunflowers, white millet.

You must keep a written feeding diary. GOV.UK states that as a requirement of the action, and sets out what it must contain; the detail is in the evidence section below.

If the land is registered as fully organic or in conversion, GOV.UK says to check with your organic control body whether a derogation is needed to use non-organic seed.

Do it in each year of the three-year duration. If the start date makes it too late in the first winter, you must start within 12 months of the action's start date. In the final year, carry on until mid-spring or the action's end date, whichever comes first.

Source: GOV.UK, AHW2 action page, checked 02/08/2026.

What you cannot do

You must not use feed hoppers to supply more than around 10 per cent of the seed mix during each feeding period, and you must not use tailings, meaning the small seeds and chaff taken off the harvested crop, in the mix.

Source: GOV.UK, AHW2 action page, checked 02/08/2026.

Where it can go, and the cap

Land that is firm and free draining, such as farm tracks or hard standing, and close to existing winter bird food, enhanced overwinter stubbles or game cover. The compatibility list follows whichever SFI26 base action you have selected. The only cap stated is the tonnage one: a maximum of one tonne of AHW2 for every two hectares of CAHL2.

Source: GOV.UK, AHW2 action page, checked 02/08/2026.

Things worth watching

This section is Tilth's own reading of where this action is easy to get wrong. It is not scheme guidance, it is not sourced from Defra or the RPA, and none of it adds a requirement beyond what the GOV.UK action page sets out.

The cap is tied to CAHL2, so it is fixed on the day you apply. One tonne per two hectares of winter bird food. If you want more AHW2, the lever is the CAHL2 area, and that is a decision made at application rather than in January.

Three of the six, and none of them over half. The mix rules have two separate arithmetic tests in them: at least three named non-cereal species, and no single species over 50 per cent of the non-cereal element. A bag can pass one and fail the other, and the only way to know is to read the composition.

The 10 per cent hopper limit is a limit on the seed, not on the hoppers. It restricts how much of the mix goes out through hoppers in each feeding period. Reading it as a limit on how many hoppers you have is a different rule, and not the one on the page.

The word is spread, and the places are plural. GOV.UK asks for the total weight divided as equally as possible between multiple feeding areas, on firm free-draining ground. Those two conditions have to hold together, in February, at more than one point on the farm. Convenience is not one of the tests.

Tailings are out. Worth flagging on its own, because tailings are a free source of small seed sitting in the shed and they are specifically excluded from the mix.

The merchant prices above were read inclusive of VAT. Merchant sites can display either basis, and a quote you are given may be on the other one. Worth settling before the figures go into a budget.

Keeping the evidence

Also Tilth's own reading, except where marked. GOV.UK sets the requirements; what follows is a practical suggestion about record keeping.

The diary is not optional and it is not a suggestion. GOV.UK states that you must keep a written feeding diary containing details of the mix used, including the weight of the cereals and the small non-cereal elements and the cost; the feeding dates; the method, spreading or hopper; the amount fed; and the location of the feeding areas. It says it can be kept on paper or electronically, and that you must supply it if asked.

Source: GOV.UK, AHW2 action page, checked 02/08/2026.

What follows is our reading of what that takes in practice. On the once-a-week requirement across an early December to late April period, that is roughly twenty entries a year, made on the day, in the weather, for three years running. A diary written up in April from memory is a different document from one written as you went, and only one of those is what the requirement describes. Whatever you use, the test is whether it is quicker to fill in on the tailgate than to put off.

The part only your own ground can answer

The money on this one turns less on your soil than on your store: whether you have grain of the right sort to make up the cereal fraction, and what you would actually have got for it. That figure sits in your own harvest and sales records rather than in a national average. Tilth keeps those alongside the feeding diary, so the sum and the evidence live in the same place. Request an invite.