The relevance of non-farmland habitats, uncropped areas and habitat diversity to the conservation of farmland birds

Authors


*Corresponding author. Email: rob.fuller@bto.org

Abstract

Use of non-farmland habitats by species generally perceived as ‘farmland birds’ is common, yet these habitats are not always considered in conservation strategies aimed at population recovery. At the national scale, many farmland species occur in landscapes not dominated by farmland. An analysis of distribution atlas data coupled with remotely sensed habitat data showed that for 16 out of 28 farmland species, less than half of the breeding range was associated with high cover of lowland farmland. However, with a few exceptions, populations breeding in non-farmland habitats are likely to depend on farmland at some time in the year. Within farmland landscapes, uncropped areas and patches of non-farmland habitat can provide nesting, foraging or roosting resources. Habitats that are scarce on farmland and that provide potential supplementary or complementary resources to those available within the productive areas of farmland include ruderal vegetation, rough grassland and scrub. Enhancing habitat diversity through provision of modest quantities of these habitats will benefit farmland birds. Complete knowledge of year-round habitat requirements and patterns of resource use at all scales is needed if robust national conservation plans are to be developed for farmland species. Similarly, interactions between the farmland and non-farmland sections of populations need to be determined.

Agricultural intensification has reduced habitat heterogeneity at all scales within farmed landscapes with detrimental effects on biodiversity (Benton et al. 2003). More specifically, there is substantial evidence that specific changes in farming practices have been principal drivers of declines in farmland birds through effects on food resources and habitat quality (Chamberlain et al. 2000, Fuller 2000). Uncropped areas provide important resources for birds and other wildlife within farmed landscapes yet, with the exception of hedgerows (Hinsley & Bellamy 2000) and field margins (Vickery et al. 2002), relatively little attention has been paid to the implications of reduction in their diversity, quantity and quality.

It has become conventional to regard ‘farmland birds’ as species that depend largely on lowland fields or field margins for their food (Gibbons et al. 1993, Fuller et al. 1995). This approach has been useful in helping to focus research on species most seriously affected by agricultural practices and in formulating the UK Government's ‘Farmland Bird Index’ (FBI), which measures the general health of farmland bird populations in Britain (http://www.sustainable-development.gov.uk, Gregory et al. 2004). However, classification of species by general habitat associations is not straightforward and can be potentially misleading. The cues by which birds select their habitat (Hildén 1965, Wiens 1989) are very different to the criteria typically used by ecologists to define spatial variation in the environment. Many species use more than one broad habitat (defined as farmland, woodland, moorland, scrub, etc.). Their distribution across such habitats may vary in time depending on population level (Wiens 1989). Even where species appear to occupy mainly one habitat in one season, they may depend on another in other seasons, or they may require a combination of habitats within a season.

The use of non-farmland habitats can be viewed on a continuum of scales. At one extreme is the situation in which populations of farmland birds occur in habitats that do not form an intimate part of the farmed landscape. In some cases, individuals within these populations may be functionally isolated from farmland for a part of the year, especially in the breeding season. At the other extreme, within many agricultural landscapes, populations of farmland birds may use a complex of landscape elements comprising both uncropped and cropped areas. This paper takes a dual-scale view of habitat use by farmland birds in an attempt to assess the ecological significance of habitats that do not form a part of productive agricultural systems. First, broad habitat associations are examined at the national (i.e. British) scale. We ask to what extent the national distributions of farmland birds actually coincide with farmland and we review the use made by farmland birds of non-farmland habitat in Britain. Secondly, at a finer scale, the use made of non-farmland and uncropped habitats within farmed landscapes is reviewed and we summarize the availability of such habitats at this scale.

DEFINITIONS AND METHODS

We treated ‘farmland species’ as the 28 species classified by Gibbons et al. (1993) as farmland birds (Table 1). In its broadest sense, farmland in Britain embraces extensively grazed unenclosed grassland and moorland in the uplands as well as more intensively managed land in the lowlands. However, our definition of farmland is essentially a lowland one because we believe the term ‘farmland bird’ has become synonymous with the lowlands. Farmland habitats were broadly taken as all tilled land and all grassland excluding moorland grass. Non-farmland included woodland, scrub, saltmarsh, dune and shingle systems, fens and swamps, unimproved upland margins, upland grass, dwarf shrub heath, moorland, gardens and urban areas. When considering fine-scale habitat use within farmland, we considered non-farmed habitats as all uncropped and infrequently managed areas, including rough grassland and the immediate vicinity of farmsteads. Cropped areas are defined as all productively managed arable land and grassland. We do not review the use of hedgerows or set-aside by birds because the resources they provide for farmland birds have been relatively well documented (Henderson & Evans 2000, Hinsley & Bellamy 2000).

Table 1.  Distribution of species perceived as farmland birds in relation to farmland landscapes in Britain. Proportion of occupied 10-km squares is shown for landscapes containing different proportions of farmland where (a) is all squares with the species and (b) are those squares containing the highest relative abundance of the species. Farmland habitats were defined using data from the CEH Land Cover Map 1990 and data on bird ranges was drawn from the BTO 1988–91 Breeding Bird Atlas (see Methods). For seven species (indicated by –), the data were insufficient to calculate an index of abundance; data were also unavailable for Woodpigeon (indicated by *). Species included in the Farmland Bird Index are shown in italics. n= number of 10-km squares occupied. Chi-square tests and P-values compare the observed distributions across the three landscape types with those evident for the whole of Britain.
 Proportion of farmland in mainland Britain according to the percentage of farmland per 10-km square
(a) n 30%31–70%> 70%χ2P(b) n 30%31–70%> 70%χ2P
0.330.220.450.330.220.45
Proportion of breeding rangeProportion of breeding range
  1. Scientific names: Stone Curlew Burhinus oedicnemus, Quail Coturnix coturnix, Turtle Dove Streptopelia turtur, Red-legged Partridge Alectoris rufa, Hobby Falco subbuteo, Corn Bunting Emberiza calandra, Yellow Wagtail Motacilla flava, Tree Sparrow Passer montanus, Little Owl Athene noctua, Yellowhammer Emberiza citrinella, Stock Dove Columba oenas, Grey Partridge Perdix perdix, Rook Corvus frugilegus, Barn Owl Tyto alba, Whitethroat Sylvia communis, Goldfinch Carduelis carduelis, Linnet Carduelis cannabina, Greenfinch Carduelis chloris, Jackdaw Corvus monedula, Montagu's Harrier Circus pygargus, Reed Bunting Emberiza schoeniclus, Kestrel Falco tinnunculus, Starling Sturnus vulgaris, Lapwing Vanellus vanellus, Skylark Alauda arvensis, Woodpigeon Columba palumbus, Cirl Bunting Emberiza cirlus, Corncrake Crex crex.

Stone Curlew440.000.110.89 22.8< 0.001
Quail2330.060.250.69 57.9< 0.001
Turtle Dove7000.040.280.67189.1< 0.0013380.020.220.75118.3< 0.001
Red-legged Partridge8940.080.280.64173.5< 0.0013850.030.200.77127.8< 0.001
Hobby2300.040.330.63 64.5< 0.001
Corn Bunting6860.110.270.62 96.4< 0.0011610.070.280.6535.0< 0.001
Yellow Wagtail7590.080.310.61146.4< 0.0012080.040.240.7261.3< 0.001
Tree Sparrow10400.070.340.59213.7< 0.0011880.010.240.7570.3< 0.001
Little Owl9420.080.360.56183.3< 0.001
Yellowhammer19620.120.350.53257.1< 0.00116630.050.360.59431.3< 0.001
Stock Dove14230.110.370.52214.2< 0.0016720.070.390.55152.4< 0.001
Grey Partridge12780.130.350.51151.6< 0.001540.040.390.5716.1< 0.001
Rook19540.170.380.46180.0< 0.00116410.050.400.55441.3< 0.001
Barn Owl7480.150.400.45 88.1< 0.001100.100.600.30
Whitethroat19340.210.350.45107.6< 0.00112150.080.380.55249.7< 0.001
Goldfinch18880.180.370.44153.3< 0.00110980.050.410.53295.7< 0.001
Linnet20020.210.360.43125.6< 0.0014290.080.400.5291.3< 0.001
Greenfinch20560.220.350.43110.9< 0.00115500.070.400.54355.1< 0.001
Jackdaw21490.230.360.41112.3< 0.00117160.070.440.48417.0< 0.001
Montagu's Harrier50.000.600.40
Reed Bunting18460.250.340.40 70.4< 0.0014730.130.420.4571.4< 0.001
Kestrel19920.270.340.39 69.6< 0.0011820.070.480.4547.7< 0.001
Starling24980.310.320.37 67.1< 0.00124310.100.420.48453.2< 0.001
Lapwing20910.310.330.36 69.9< 0.0013440.090.440.4772.4< 0.001
Skylark25710.350.300.35 71.0< 0.00121710.160.400.45241.0< 0.001
Woodpigeon23080.370.340.29149.5< 0.001******
Cirl Bunting180.390.330.28  1.71   0.557
Corncrake1050.790.140.07 50.8< 0.001

The Centre for Ecology and Hydrology's Land Cover Map 1990 (LCM) (Fuller et al. 1994) was used in conjunction with BTO data from the 1988–91 breeding bird atlas (Gibbons et al. 1993) to examine distributions of farmland birds at the national scale within mainland Britain. These two contemporary data sets allow comparisons that were applicable in 1990; it should be noted that several of the species have shown subsequent changes in range. The LCM allows the whole of Britain to be allocated to 25 land cover classes. Lowland farmland was defined as comprising the following five land cover classes: class 5 – lowland grass heath; class 6 – pasture/amenity turf; class 7 – meadow/verge/semi-natural swards; class 8 – rough grass/marsh grass; class 18 – tilled land. This definition of farmland includes some classes that are likely to include some unimproved rough grassland. The areas of each of these classes were summed to estimate the total area of farmland in each 10-km square of the Ordnance Survey National Grid.

For each species in turn, the proportion of the total number of squares in which each was recorded as breeding was calculated for three categories of cover of farmland (0–30%, 31–70%, > 70%) and compared with the proportions for mainland Britain as a whole. These ‘all-squares’ analyses were repeated for a subset of 10-km squares where each species was likely to be most abundant. In the atlas, ‘relative abundance’ within each 10-km square was estimated from the proportion of tetrads (2-km squares) in which each species was recorded. Those 10-km squares in which a minimum of eight tetrads had been surveyed and in which the proportion of tetrads with breeding confirmed for the particular species was  0.4 were selected and are termed ‘high abundance squares’. This analysis was conducted for 21 farmland species; there were insufficient data for the remaining seven species. To determine which non-farmland land cover classes were most frequent in squares occupied by each species, mean percentages of each of the 25 land classes were calculated for each occupied 10-km square and then grouped into broad habitat types. Types and quantities of non-farmland and uncropped habitats within farmland were summarized from LCM 1990 for six English counties (mainly arable – Cambridgeshire, Lincolnshire and Essex; mainly grassland – Devon, Somerset and Dorset).

RESULTS AND DISCUSSION

Distribution of farmland birds with respect to landscape type

In Table 1a, the species are listed in approximate decreasing order of apparent dependence on farmland based on the ‘all-squares’ analysis. Thus species such as Stone Curlew (scientific names are given in Table 1), Quail and Turtle Dove are highly associated with farmland landscapes, being present in a relatively low proportion of squares with < 30% farmland, but in a relatively high proportion of squares with >70% farmland. In contrast, species such as Starling, Lapwing and Skylark have distributions suggesting a wider use of habitat types. For 16 out of 28 species, less than half of the breeding range was associated with squares with high cover of lowland farmland. This included diverse species such as Barn Owl, Linnet, Reed Bunting, Lapwing and Skylark. The apparently high dependency of Corncrake and Cirl Bunting on non-farmland squares was an artefact due to the concentration of these localized species in areas where the squares tended to be classified in coastal land classes. Differences between the proportions of species’ ranges in each of the farmland categories and those of farmland in mainland Britain as a whole were significant (P < 0.001, χ2 tests) for all species except Cirl Bunting, but given the generally large sample sizes (except Montagu's Harrier, sample size too small to test) this was not unexpected.

Repeating the analysis for the ‘high abundance squares’ (Table 1b) produced a similar pattern to that for ‘all squares’. With the exception of Barn Owl, these ‘high abundance squares’ were more strongly dominated by lowland farmland than was the case for all occupied squares. Nonetheless, for seven out of 20 species less than half of the ‘high abundance squares’ contained > 70% farmland. The differences between bird and farmland distributions were all significant (P < 0.001, χ2 tests).

The mean habitat composition of all occupied squares is shown in Table 2 by broad habitat groupings. Species are listed in decreasing order of the percentage cover of farmland within the breeding range and they show a similar order to that evident using the approach in Table 1. Four non-farmland habitat types emerged as potentially important for farmland birds (Table 2). For many species, the occupied squares contained on average > 10% of coastal habitats. Woodland and suburban/built-up areas also featured regularly. Mean cover of moorland was relatively high (> 10%) in squares occupied by Jackdaw, Kestrel, Lapwing, Skylark, Woodpigeon and Corncrake.

Table 2.  Major habitat types ( 4% mean cover) within the breeding ranges of farmland birds. Figures are mean percentage cover per occupied 10-km square drawn from the BTO 1988–91 Breeding Bird Atlas calculated for broad habitat types defined from the CEH Land Cover Map 1990 (see Methods). Broad habitats are defined using the following amalgamations of LCM1990 land cover classes: Coastal = classes 1 and 3; Woodland = 15 and 16, Rural/suburban development and urban = 20 and 21; Moorland = 9 and 10. Values of < 4% are indicated by –.
SpeciesFarmlandCoastalWoodlandRural/suburban development and urbanMoorland
Stone Curlew777 5 0
Turtle Dove72 56 9
Hobby70811
Red-legged Partridge70 65 8
Quail70 75 6
Yellow Wagtail68 54 9
Tree Sparrow68 75 8
Little Owl67 66 8
Corn Bunting67114 8
Stock Dove64 76 7
Yellowhammer64106 7
Grey Partridge62 85 7
Montagu's Harrier61205 6 0
Barn Owl60106 5
Rook59105 6
Goldfinch58115 6
Whitethroat57135 6
Greenfinch56125 6 4
Linnet56135 6
Jackdaw55125 610
Reed Bunting54125 6 9
Kestrel52105 612
Lapwing50134 512
Starling50164 5 4
Skylark47164 513
Wood Pigeon44138 417
Cirl Bunting4244 4
Corncrake204616

Patterns of large-scale use of non-farmland habitats

Using data at a national scale from a large sample of randomly distributed transect plots, Gregory and Baillie (1998) estimated broad habitat distributions for eight breeding species, four of which are farmland birds. In all cases, a substantial proportion of birds was associated with non-farmland habitats: Skylark 42%, Starling 65%, Linnet 32%, Reed Bunting 49%. The functional significance of different habitats was unclear from these data, but in Table 3 we summarize the uses that farmland birds make of some non-farmland habitats. These results, derived independently, are consistent with the findings in Tables 1 and 2.

Table 3.  Examples of farmland species widely associated with non-farmland habitats.
 Breeding seasonOutside breeding season
  • N

    Nests,

  • F

    F feeds,

  • R

    roosts in the habitata. Note that the list of habitats and species is not comprehensive and that only FBI species are considered (see Table 1).

  • a

    General sources: Fuller (1982), Cramp et al. (1979–1994).

  • b

    Upland heath and moor are unenclosed rough grazing land in north and west Britain, typically dominated by Calluna vulgaris, Molinia caerulea, Nardus stricta and other grasses. Upland margins includes unimproved enclosed land just below the open moors often consisting of rough grazing with extensive Pteridium aquilinum in some regions, notably in Wales. Sources: Ratcliffe (1990), Fuller et al. (2002), Milsom et al. (2002).

  • c

    Typically dominated by Calluna vulgaris and Ulex spp. Most farmland birds are associated with scrub on heathland. Source: Bibby (1978).

  • d

    Vegetation in ruderal areas consists of weedy species (e.g. grasses, thistles, Chenopodium spp., Rumex spp., Urtica spp., Epilobium spp.) typically found on waste land or uncultivated, but episodically cleared or disturbed, areas such as around farm buildings and on rough ground alongside streams, in field corners, etc. If left indefinitely it may develop into scrub.

  • e

    Scrub occurs in many forms; here it refers to successional development of woody vegetation (both open- and closed-canopy) on abandoned or marginal land in the lowlands. Sources: Fuller (1995), Gough (1999).

  • f

    Includes recently established farm woods (Vanhinsbergh et al. 2002) and establishment and thicket-stage conifer plantations (Fuller & Browne 2003).

  • g

    Rural gardens can attract a wide range of farmland species in winter but the three listed are those that most widely occur in gardens. Source: Bland et al. (2004).

  • h

    Sources: Morgan (1978), Brown and Atkinson (1996), Ferns (1992).

  • i

    With the exception of Skylark, which reaches high densities in some dune systems, nesting farmland birds are largely confined to areas of scrub within dunes. Sources: Delius (1965), Morgan (1978), Ferns (1992).

  • j

    Sources: Milsom et al. (1998, 2000, 2001).

Upland heath & moorbKestrelNF, SkylarkNF 
Upland marginsbKestrelNF, Grey PartridgeNF, LapwingNF, SkylarkNF, JackdawF, RookF, StarlingF, LinnetNF, YellowhammerNFKestrelF, JackdawF, RookF, StarlingF
Lowland heathcKestrelNF, SkylarkNF, WhitethroatNF, LinnetNF, YellowhammerNF, Reed BuntingNF 
Ruderal areasdSkylarkF, WhitethroatNF, GreenfinchF, GoldfinchF, LinnetF, YellowhammerF, Reed BuntingF, Corn BuntingFTree SparrowF, GreenfinchF, GoldfinchF, LinnetF, YellowhammerF, Reed BuntingF, Corn BuntingF
ScrubeWoodpigeonNF, Turtle DoveN, WhitethroatNF, GreenfinchN, LinnetN, YellowhammerNFWoodpigeonR, StarlingR, Tree SparrowR, GreenfinchR, GoldfinchR, LinnetR, YellowhammerR, Reed BuntingR, Corn BuntingR
Young plantationsfKestrelF, WoodpigeonNF, Turtle DoveN, SkylarkNF, WhitethroatNF, LinnetN, YellowhammerNF(thicket stage may hold roosts similar to scrub)
GardensgWoodpigeonNF, JackdawNF, StarlingF, GreenfinchNF, GoldfinchNFStarlingF, GreenfinchF, GoldfinchF
SaltmarshhKestrelF, Grey PartridgeNF, LapwingNF, SkylarkNF, Yellow WagtailNF, LinnetNF, Reed BuntingNFKestrelF, LapwingF, SkylarkF, JackdawF, RookF, GreenfinchF, GoldfinchF, LinnetF, Reed BuntingF
Dune systemsiKestrelF, WoodpigeonNF, Turtle DoveNF, SkylarkNF, WhitethroatNF, LinnetNF, YellowhammerNF, Reed BuntingNFDune scrub probably used as roost by finches and buntings
Coastal grazing marshjKestrelF, Grey PartridgeNF, LapwingNF, SkylarkNF, Yellow WagtailNF, StarlingFLapwingF, StarlingF

Certain habitats can be singled out as especially significant at a national scale. The uplands hold > 30% of the British breeding population of Skylarks (Chamberlain & Gregory 1999, Browne et al. 2000) with the highest densities on semi-natural grass (Brown & Stillman 1993, Chamberlain & Gregory 1999). Some of the highest concentrations of breeding Lapwings in England and Wales occur on the upland fringes and > 25% of pairs in 1998 were estimated to be on rough grassland, much of it presumably in the upland margins (Wilson et al. 2001). Nearly 50% of the Starling population may be associated with urban areas (Gregory & Baillie 1998) while gardens are also especially important for Greenfinch and Goldfinch (Bland et al. 2004). Young conifer plantations provide breeding habitat for large numbers of Yellowhammers. Coastal populations of breeding Skylarks and Lapwings are likely to be considerable. Almost any habitats with scrub will support Whitethroats, Linnets and Yellowhammers – this applies to dune systems, lowland heaths, commonland, downland, marginal uplands and the upper slopes of coastal cliffs. Outside the breeding season, concentrations of farmland passerines are found in gardens and saltmarshes while all habitats containing scrub are potentially roosts for finches, buntings, sparrows and Starlings. In a national survey of wintering Skylarks, the highest densities were found on saltmarsh, although the bulk of the population was on farmland (Gillings & Fuller 2001).

Habitat use within farmed landscapes

Where farming is the dominant land-use, lowland landscapes are generally patchy mosaics consisting of fields, hedges, farmsteads, woods, patches of scrub, streams and other wetlands. In our six sample counties, woodland and suburban/built-up areas provided the most extensive non-farmland habitats, ranging from 1 to 11% cover and 3 to 13% cover, respectively, across the six counties. Three structural aspects of habitat composition seem especially important in shaping the make-up of breeding bird communities on farmland: the extent of woodland, the density of hedgerows and the density of wetlands (streams and ponds) (Fuller et al. 1997). Within farmland, Grey Partridge, Lapwing, Skylark and Yellow Wagtail are strongly associated with cropped areas, but all the other FBI species are likely to make some use of uncropped areas.

Woodland is used by several farmland birds in the breeding season both for nesting and for feeding. This is especially true of young plantations (Table 3). However, more mature woods can also be used by Woodpigeon, Stock Dove, Turtle Dove, Greenfinch, Goldfinch and Yellowhammer. The last four species tend to be associated with the edges of mature woodland, probably because they depend on habitats outside woodland for food and their occurrence in woodland can be influenced by the proximity of food sources (Hinsley et al. 1995). However, Whitethroat, Linnet, Goldfinch, Greenfinch and Yellowhammer all show stronger association with hedgerows than with woodland (Fuller et al. 2001). Similarly, scrub is probably more important for farmland birds than woodland. Scrub is strongly selected as a nesting habitat by Turtle Doves (Browne & Aebischer 2004) and as a roosting habitat, especially by finches and buntings (Table 3).

An increasing number of studies have examined in detail how farmland birds utilize the habitat features available within farmed landscapes. Some of these show that particular habitats may provide essential feeding areas: for example, patches of rough grassland in the case of Skylark (Wilson 2001), Yellowhammer (Bradbury et al. 2000 and references therein) and Corn Bunting (Watson & Rae 1997), rank vegetation for Reed Bunting (Brickle & Peach 2004) and wet or emergent areas for Tree Sparrow (Field & Anderson 2004) and Reed Bunting (Brickle & Peach 2004). Corn Buntings in the Outer Hebrides make much use of uncultivated areas both for nesting and for feeding (Hartley & Shepherd 1997). Although technically not a farmland bird in the context of this paper, it is also worth noting that Song Thrushes Turdus philomelos on farmland depend strongly on uncropped habitats, including woodland edges, scrub, field boundaries and gardens (Peach et al. 2004). Ruderal vegetation can provide important localized feeding sites for birds both in summer and in winter, for both insectivorous and granivorous species (Table 3). Constituents of ruderal vegetation providing food sources for granivorous birds are Polygonaceae, Chenopodiaceae and Asteraceae (Wilson et al. 1999). These resources can occur in a variety of situations such as neglected land in the vicinity of farm buildings and the vegetation along linear wetlands and around ponds.

A process-based perspective

Dunning et al. (1992) provided a useful framework for discussing how the cropped and non-cropped components of landscapes may interact with population dynamics. They identified two processes involving the movement of individuals between patches to exploit resources. ‘Landscape complementation’ occurs where a species requires at least two different resources (equivalent here to habitats) within the same season. This requires that the resources are available in reasonably close proximity (Eybert et al. 1995). Seed-eating birds provide striking examples of this process. Turtle Doves depend on scrub, hedgerows and woodland edges as nesting habitat, although they forage on spilt grain and on seed-rich patches in fields and ruderal areas (Browne & Aebischer 2004). In winter, several seed-eaters roost in scrub, but forage in other patch types. ‘Landscape supplementation’ applies to situations where different habitats offer substitutable resources. An example is the use of ruderal areas as seed sources to supplement those available in fields and field margins. It is likely that ruderal areas have increased in relative importance as seed supplies in arable fields have become depleted in recent decades. Similarly, ruderal and other non-crop patches may provide resources at critical periods, for example in late winter/early spring when food supplies in fields may be exhausted. In summer, such substitutable resources would need to be in relatively close proximity, but in winter birds could potentially exploit substitutable resources on a larger scale, possibly showing seasonal shifts in the types of habitats used.

Source-sink dynamics form a third process whereby individuals in a population occupy habitats differing in quality (Dunning et al. 1992, Watkinson & Sutherland 1995). The consequence may be that poor-quality habitats could be sinks where occupancy is only sustained by immigration from high-quality source habitats where breeding production generates a surplus. To our knowledge, such effects have not been demonstrated for any farmland species, although they may occur across a variety of scales including regions and broad farming systems that vary in habitat quality (Wilson & Griffiths 1997) and more locally where potentially productive habitats exist close to unproductive ones. Given the poor performance of some species on farmland, the possibility exists that farmland habitats could constitute sink habitat on a large scale. Thus the ‘true’ status of farmland populations could be worse than apparent if numbers are being buffered by immigration from non-farmland source populations.

It is very likely that breeding output of some species is now higher in non-farmland habitats than in intensively farmed areas. Lapwings and Skylarks, for instance, may achieve higher productivity in some upland and saltmarsh habitats than in many areas of intensively managed lowland farmland. Sustained high breeding output in non-crop habitats does not, however, guarantee that populations in these habitats are buffered against change. Such buffering would depend in part on the distribution of the surplus between source and sink type habitats. In addition, at other times of year, a high proportion of the population may depend on lowland farmland where food supplies have been reduced. This may account for declines in upland and dune-nesting Skylark populations (Chamberlain 2001, Wolfenden & Peach 2001).

The extent and type of non-crop habitat in farmland landscapes almost certainly affects predation risk, although knowledge of the impact of predation at the population level is lacking. Generalist predators can attain high numbers in farmland mosaics and may be particularly effective at exploiting prey restricted to edges and patches of fragmented habitat. Little is known about relationships between landscape structure and breeding success in open-country species. It is clear, however, that some farmland birds avoid even lightly wooded areas (e.g. Hancock & Wilson 2003). The balance of cropped and non-cropped areas within farmland landscapes has a major effect on composition of bird communities (Fuller et al. 1997). This raises the possibility of interspecific competition for food resources varying according to landscape structure and habitat composition. For example, landscapes with a combination of woodland and rape crops generally support high numbers of Woodpigeons, which may compete at certain times of year with other granivores. It is also possible that species principally associated with woodland or scrub may compete with farmland birds for food in ruderal habitats. Thus non-cropped land within farmland has both positive and negative aspects.

CONCLUSIONS

Non-farmland habitats appear to support substantial proportions of the British breeding populations of several farmland birds. However, few of these non-farmland populations are isolated from farmland because (a) individuals breeding in non-farmland habitats may feed on farmland in summer and (b) many birds may depend on farmland outside the breeding season. The main exceptions are populations of Whitethroats that nest and feed within scrub and young woodland, and populations of Goldfinches, Greenfinches and Starlings associated with urban areas. Knowledge of the large-scale population structure with respect to interaction between farmland and non-farmland habitats is completely lacking.

Within farmland landscapes many species use a variety of habitats for different purposes, but detailed knowledge of habitat use is incomplete for most species. There is a need to study year-round ranging behaviour of farmland birds to ascertain more exactly the spatial, temporal and diurnal use of both cropped and non-cropped habitats, and to investigate the characteristics of movements between farmland and non-farmland sections of their populations. Many studies of foraging habitat requirements have focused on the use of fields by birds (for an exception see Hancock & Wilson 2003), but these do not give comprehensive information on resource use. A better understanding of ‘whole landscape use’ is needed for most species.

Some of the uncropped habitats that appear to provide roosting and foraging habitats for some species are rare in farmed landscapes. Rough grassland, ruderal vegetation and scrub are examples. LCM 1990 data indicate that these three habitats seldom contribute more than 1% of the land cover within farmed counties and are absent from substantial tracts of farmland. Cropped areas undoubtedly provide the major food resources for many farmland birds, especially in winter (e.g. Hancock & Wilson 2003). However, uncropped areas and non-farmland habitats potentially offer supplementary food resources and, in some landscapes, these may even become critical at certain times of year and help to buffer populations against negative environmental change. We contend that enhancing the diversity of habitat within farmland landscapes through modest increases in the extent of these habitats could benefit many farmland birds, as well as bringing wider biodiversity gains. Examples of this type of integrated approach to conservation within farmed landscapes are to be found in Germany (Berger et al. 2003) but are currently sparse in Britain. The introduction of the Environmental Stewardship scheme in 2005 in England, along with similar schemes in Scotland and Wales, offers an opportunity to move further in this direction (Smallshire et al. 2004; Vickery et al. 2004).

Acknowledgments

R.J.F.'s work on this paper was funded through a partnership between the British Trust for Ornithology and the Joint Nature Conservation Committee (on behalf of Scottish Natural Heritage, English Nature, the Countryside Council for Wales and the Environment & Heritage Service of Northern Ireland). We acknowledge the huge efforts of those who contributed to the 1988–91 breeding bird atlas and we thank Nicki Read for help in preparing this paper.

Ancillary