Although the changes outlined above have been continuous since the 1940s, the period of greatest change was in the 1970s and early 1980s, when the widespread declines in farmland wildlife first received attention (Potts & Vickerman 1974; Chamberlain et al. 2000). Although there are many local studies of the biodiversity impacts of agriculture, discussion of the factors affecting the declines of specific taxa, however, is often hampered by the lack of data at a national (or even regional) scale. Local studies, usually at a plot or field scale, are usually difficult to interpret as their results may not be applicable at broader scales (Duffield & Aebischer 1994; Siriwardena et al. 2000a). Changes in agricultural practice have largely been monotonic and in the same direction, towards increasing yields. Distinguishing between the various factors and attributing causal relationships to a particular combination of factors is very difficult; simple temporal correlations are likely to be misleading (Chamberlain et al. 2000).
One reason often proposed for the wide range of declines in farmland taxa is the increased use of pesticides (Campbell et al. 1997). Although pesticides are undoubtedly effective at controlling their target species and often affect non-target organisms (Wilson et al. 1999), evidence of population changes from studies at scales larger than individual plots can be equivocal. Often this is done by comparing farms employing conventional or intensive management techniques with those employing lower intensity or organic practices. Farms with lower chemical inputs generally hold greater numbers, and higher diversity of species, of a broad range of taxa, although the reported statistical significance of these results can be weak and some find no differences (Braae, Nøhr & Petersen 1988; Holland et al. 1994; Moreby et al. 1994; BTO/IACR 1995; Campbell et al. 1997; Greenwood 2000). It is often difficult to attribute the greater numbers or diversity of species on organic farms solely to chemical use, because although farms are usually paired, the effects of other factors, such as hedgerow density and rotations, can be at least as important as any differences in chemical applications (Chamberlain, Wilson & Fuller 1999). For example, Yeates et al. (1997) recorded increased numbers of mites and nematodes and lower earthworm numbers on organic farms; however, this is likely to reflect differences in soil structure caused by differing tillage practices, rather than pesticide use.
A number of potential factors for the declines in arable plants can be identified: habitat drainage; changes in soil nutrient status; changes in the timing of cereal sowing; and the application of chemicals, either directly or indirectly (Hampicke 1978; Chancellor, Fryer & Cussans 1984; Wilson 1992). Detailed autecological work has elucidated the declines of some of the rarer species, for example, the corncockle Agrostemma githago L. declined due to improved seed-cleaning techniques, which are also likely to have reduced weed populations generally (Firbank 1988; Wilson 1992). The increase in autumn cereal sowing (Fig. 4b) has contributed to the declines of spring-germinating plants, such as the corn marigold, and earlier harvesting of winter crops will also have affected late autumn-germinating species, such as corn buttercup and shepherd’s-needle Scandix pecten-veneris L. (Wilson 1992; Hald 1999).
Many species that remain common on farmland are either resistant to, or difficult to target with, herbicides, or have prolific, persistent, seed banks. This tends to suggest herbicides may be responsible for the declines of the remaining species, as does the lack of a decline in frequency of weed occurrence between 1969 and 1987 reported by Whitehead & Wright (1989), who recorded plant densities before herbicide application. Long-term studies of individual fields show that timing of sowing is critical, as are past management practices (Chancellor 1985). Ewald & Aebischer (1999) also noted that timing of application was important, with spring herbicide spraying having the greatest impact on plant populations. Minimum tillage favours species that have an ephemeral seed bank, particularly grasses, limiting the uptake of reduced cultivation techniques (Bobbink 1991; Wilson et al. 1999).
In parallel with the increase in pesticide use, inorganic fertilizers are also applied much more frequently, particularly in grass leys. Although increased nutrient status is a major problem in grassland habitats (Marrs 1993), there seems to be little evidence that increased nitrogen use affects diversity or numbers of non-crop plants directly in current high-intensity arable systems (McCloskey et al. 1997; Kleijn et al. 2001). Spray drift into field margins reduces plant density and diversity there (Marshall 1988). Many plant species that are currently increasing are associated with elevated nutrient status, and fertilizer-intolerant species may have disappeared in the 1950s and 1960s (Wilson 1992; Smart et al. 2000). At least some of the effect of fertilizers comes from indirect effects, such as shading from species that can use the nutrients most efficiently (Bobbink 1991). Organic manures potentially have the same effects as inorganic fertilizers, but nutrients are usually much less concentrated and released more slowly, reducing their deleterious impacts (Simpson & Jefferson 1996). With decreased atmospheric sulphur dioxide levels, we might expect lichen numbers to be increasing, but there is some evidence that increased ammonia emissions from farmland may be slowing this process (Coppins, Hawksworth & Rose 2001).
In general, many insect populations, particularly butterflies and orthopterans, are expanding as the British climate is becoming milder (Menzel & Fabian 1999; Roy et al. 2001); however, changes in habitat management have had a large, mostly negative, impact (Thomas, Morris & Hambler 1994; Warren et al. 2001). For example, many butterfly populations have suffered from the loss of unimproved pasture (Rands & Sotherton 1986; Asher et al. 2001), sawfly numbers from a reduction in the number of longer-term grass leys (Barker, Brown & Reynolds 1999), and insect diversity generally is lower in more intensively managed fields (di Giulio, Edwards & Meister 2001). Local studies have shown that some species are likely to have been affected by the loss of non-cropped habitats such as field margins, which can act as a reservoir of individuals (Dennis, Thomas & Sotherton 1994; Lee, Menalled & Landis 2001), although Jones (1976) noted greater numbers and diversity of arthropods in winter wheat plots than adjacent fallow plots. Diversity did, however, decrease through the 8 years of this study with the simplified rotation.
Chemical control of insects can have long-term consequences, with effects evident several years after the initial application (Aebischer 1990), although some species can recover very rapidly (Brown, White & Everett 1988). A large number of studies has quantified the short-term effects of insecticides on populations of both target and non-target species (reviewed in Wilson et al. 1999). These have generally been at a small scale and yielded varying results: most show declines in insect numbers following insecticide applications, but others show little change, or even increases. At least some of these differences in population trends are likely to be due to varying susceptibility to insecticide control; the timing of spraying is important and particular life histories will determine which species are most susceptible (Burn 1989; Vickerman 1992; Marc, Canard & Ysnel 1999).
The soil fauna, particularly the microfauna, represents a major part of the biodiversity of agro-ecosystems, though their study presents formidable challenges (Paoletti 1999). In general, greater soil disturbance reduces their numbers. The amount of organic matter (i.e. food) is the most important determinant of earthworm abundance, and although light soil cultivation can be beneficial, deep ploughing can reduce numbers by up to 50% (Edwards & Bohlen 1996). Increased fertilizer use is generally beneficial to the soil fauna, although concentrated liquid forms (slurry) may not be (Paoletti 1999). Holland & Luff (2000) found different practices favoured different carabid species, although the presence of non-cropped habitats, minimum tillage and some fertilization seemed to offer the broadest benefits.
In their review of the population status of British mammals, Harris et al. (1995) noted that the most common causes of population change are likely to include habitat change (believed to affect 48% of the 65 species covered), use of chemicals (including pesticides, affecting 38%, mostly bats and rodents) and deliberate killing (28%, primarily carnivores and species perceived as pests), although few explicit data were available.
While a number of factors have been identified as potential causes of the decline in brown hare numbers, a reduction in landscape diversity has been suggested as being most important, at least in arable areas; in pastoral areas, increases in stocking densities and the increasing use of grass for silage may be more important (Hutchings & Harris 1996). Rabbit populations are recovering from a catastrophic decline caused by the myxoma virus in the early 1950s, and this has been linked to the decline and subsequent recovery of the stoat (for whom rabbits are an important prey species) and the population increase and current decline in the weasel (rabbit grazing reduces the amount of long grass for voles, which are a weasel’s primary prey) (Tapper 1992; Yalden 1999). Numbers of bats are likely to have suffered from destruction of roost sites, loss of foraging habitat and a general decline in their insect prey (Walsh & Harris 1996; Yalden 1999). For example, the greater horseshoe bat Rhinolophus ferrumequinum Schreber may have declined with the loss of hedgerows as feeding habitat (Mitchell-Jones 1998).
Population trends of small mammals on farmland are unclear, but are likely to be affected by food supply of both insects and seeds, which have declined (Tew, Macdonald & Rands 1992); they may also be particularly susceptible to direct mortality from rodenticides or eating seeds treated with molluscicides, although immigration from other habitats can occur rapidly (Shore et al. 1997; McDonald et al. 1998). The harvest mouse has been adversely affected by loss of grassland habitats and increased planting of winter cereals (Perrow & Jowitt 1995), and common vole Microtus arvalis Pallas numbers have declined in areas where their preferred grassland habitat has been converted to agriculture (Gorman & Reynolds 1993). There is, however, little quantitative evidence for the relative importance of changes in habitat structure, harvesting, chemical usage and food supply.
Changing agricultural practices have had a major impact on bird communities, and the reasons for the declines in some species can be pinpointed through population monitoring and autecological studies (Aebischer, Green & Evans 2000). For example, corncrake populations have fallen due to increasingly intensive grassland management, in particular a loss of hay meadows (Green & Stowe 1993), and the breeding populations of some waders, such as redshank Tringa totanus L. and lapwing Vanellus vanellus L., are declining with the loss of unimproved grassland and increased stocking densities (O’Brien & Smith 1992; Peach, Thompson & Coulson 1994). Agricultural changes have benefited some species, for example woodpigeon Columba palumbus L., because of increased oil-seed rape planting (Inglis et al. 1990). Declines in seed-eating passerines (larks, finches, buntings and sparrows) have received much attention recently and we review them in some detail, as a case study of the processes involved.
In general, British bird species are nesting earlier and with increased success because of changes in climatic variables (Crick & Sparks 1999), although amongst seed-eaters this may also be a density-dependent response to lower population levels (Siriwardena et al. 2000b). While productivity per nesting attempt has not declined (except in the case of the linnet Carduelis cannabina L.), there may have been a decline in the number of breeding attempts, particularly among open-field nesters. For example, skylarks Alauda arvensis L. generally prefer to nest in short vegetation, such as spring-sown cereal, particularly where food densities are high, but frequently abandon nests in winter cereal (the crop grows too fast), reducing opportunities for subsequent broods (Wilson et al. 1997; Wakeham-Dawson et al. 1998; Chamberlain et al. 1999). Corn buntings Miliaria calandra L. nest much later than other species and nest success may have decreased, because of both harvesting operations (Crick et al. 1994) and reduced insect abundance (Brickle et al. 2000). However, responses to cropping regime are likely to be complex; both corn buntings and skylarks nest at higher densities and with greater success in areas of high crop diversity (Ward & Aebischer 1994; Chamberlain & Gregory 1999).
For most passerines that eat seeds in winter, mortality appears to be relatively low in the early winter period and much higher in late winter when food resources have been depleted, with late winter mortality relatively more important since 1970 (Crick, Donald & Greenwood 1991). Adult and juvenile survival rates of most species decreased during their population declines, and for reed bunting Emberiza schoeniclus L., goldfinch Carduelis carduelis L. and house sparrow Passer domesticus L. these changes are sufficient to account for the declines (Siriwardena, Baillie & Wilson 1998, 1999; Peach, Siriwardena & Gregory 1999). In parallel with the declines in farmland birds, populations of predators, particularly sparrowhawks Accipiter nisus L. and magpies Pica pica L., have increased. This is very unlikely to have caused the population declines of farmland birds (Thomson et al. 1998), although predation may have increased because of habitat simplification (Donald & Vickery 2000), a subject that needs further study. Intriguingly, recent work has suggested that parasites transmitted from captive-reared pheasants Phasianus colchicus L. may be having an adverse impact on grey partridge Perdix perdix L. populations (Tompkins et al. 2000).
While the breeding season and winter cannot be considered in isolation, the decline of farmland passerine seed-eaters can be broadly related to their foraging on seeds in winter and changes in survival (Table 2), although the importance of changes in number of nesting attempts remains to be quantified. For example, most of the British populations of linnet and goldfinch winter in France and Iberia, where seed densities can be approximately five times those currently found in British soils (Díaz & Tellería 1994). Chaffinches Fringilla coelebs L. and greenfinches Carduelis chloris L. forage extensively in gardens, which are also becoming increasingly important for other species, particularly in the late winter period (Cannon 2000). Similarly, while set-aside has many potential benefits for wildlife (Firbank 1998), current management practices are likely to limit the extent to which weed seeds are present, which may explain why populations of farmland seed-eaters in Britain did not increase with the widespread introduction of set-aside in 1992.
Table 2. Population and dietary characteristics of British farmland seed-eating passerines. Population changes on farmland Common Birds Census plots between 1976 and 1995 from Siriwardena et al. (1998) (95% confidence interval in parentheses). Relative rates of survival (S), daily nest failure rates (NFR) and brood size (BS) are given for periods of increasing (I), stable (S) or declining (D) population trends, asterisks indicate where the change is sufficient to have caused the population change (Siriwardena, Baillie & Wilson 1998; Siriwardena et al. 2000b). Figures in bold indicate statistically significant results, and hyphens insufficient data. The use of gardens is given by: †recorded < 10% gardens; ‡recorded in 25% of gardens; §recorded in 80% of gardens during the late winter (Cannon 2000)
|Tree sparrow||−84 (−97, +0·02)||S > D(–)||I > D||D > I||†||Mostly sedentary|
|Corn bunting||−66 (−, −)||–||I > D||D > I||†||Sedentary|
|Skylark||−49 (−57, −39)||–||S > D > I||D > S > I||†||Within Britain|
|Reed bunting||−46 (−69, −0·08)||I = S > D(*)||I > D = S||S > I > D||†||Mostly sedentary|
|Linnet||−38 (−55, −14)||D > I||D > I(*)||I > D||†||Mostly France/Spain|
|Yellowhammer||−36 (−42, −10)||I > S > D(–)||I > S > D||D > S > I||†||Mostly sedentary|
|Greenfinch||−1 (−21, +34)||S > D > I||S > D = I||I > S > D||§||Britain (France, Belgium)|
|Goldfinch||+7 (−20, +52)||I > D(*)||D > I||D > I||‡||Mostly France/Spain|
|Chaffinch||+19 (+7, +31)||I > S||I > S||I = S||§||Within Britain|
The population decline of grey partridge in Britain has been linked to a decrease in the number of invertebrates available during the nesting season, and hence chick survival (Potts 1986), although the same may not be true in France (Bro et al. 2000). The linnet is the only bird where changes in breeding season have been shown to be sufficient to account for changes in population (Table 2). This suggests that declines in the summer insect fauna may not have affected granivorous passerines to the same extent (partridge chicks are nidifugous and forage independently). The potential importance of winter food for passerines is highlighted by the cirl bunting Emberiza cirlus L. (Evans 1997; Aebischer, Green & Evans 2000). The cirl bunting was once common over much of southern England, but by the 1980s had become restricted to the south-west with a population of just 118 pairs. When farmers were paid to leave weed-rich winter stubbles, the population increased to around 500 pairs in less than 10 years.
Bird populations elsewhere in north-west Europe are likely to be affected by agricultural intensification in a similar way; however, those in southern and eastern Europe are more often affected by loss of unimproved habitat (Schifferli 2000; Donald, Green & Heath 2001), and numbers of little bustard Tetrax tetrax L. have increased with low-intensity cultivation of steppe habitats (Wolff et al. 2001). In North America, many species are thought to be declining because of farmland abandonment, particularly in the east, farmland elsewhere being increasingly intensively managed (Askins 2000). The precipitous decline of the dickcissel Spiza americana Gmelin has been related to changes in food supply in their wintering grounds (Fretwell 1986).