FOOD AVAILABILITY

1.Foraging bird resource requirements and behaviour in conventional and alternative arable crop systems: a review of published research.

CONTENTS

1.1Background1

1.2Aims1

1.3Introduction1

1.4Food availability4

1.4.1Introduction4

1.4.2Effects of crop type5

1.4.3Effects of management of non-crop habitats6

1.4.3.1Stubble6

1.4.3.2Set-aside7

1.4.3.3Conservation headlands8

1.4.3.4Field margins9

1.4.4Effects of position within a field10

1.4.5Effects of agricultural practices11

1.4.5.1Invertebrates12

1.4.5.2Plants13

1.4.6Effects of pesticides15

1.4.6.1Insecticide: factors affecting their impact on non-target taxa………………...15

1.4.6.2.Insecticide: levels of impact on non-target taxa16

1.4.6.3.Herbicide: effects on weeds18

1.4.6.4Herbicide: indirect effects on invertebrates…………………………………...19

1.4.6.5.Fungicides20

1.4.7Effects of whole farm agricultural systems20

1.4.7.1Integrated farming systems20

1.4.7.2Systems using genetically-modified crops21

1.4.7.3Organic systems22

1.4.8Weather effects22

1.4.9Effects of food accessibility and predation risk23

1.4.10Discussion – food availability25

1.5Foraging preferences26

1.5.1Introduction26

1.5.2The preference of birds for farmland habitats26

1.5.2.1All year studies26

1.5.2.2Winter studies27

1.5.2.3Breeding season studies28

1.5.3Specific studies on the species being considered in this project29

1.5.4Foraging preferences and pesticide levels36

1.5.4Discussion – foraging preferences37

1.6Functional response37

1.6.1Introduction37

1.6.2Types of functional response37

1.6.3Factors affecting the functional response38

1.6.3.1Food availability38

1.6.3.2Relative profitability38

1.6.3.3Substrate39

1.6.3.4Interference39

1.6.3.5Search efficiency39

1.6.3.6Individual variation39

1.6.3.7Risk sensitivity40

1.6.4Discussion – functional response40

1.7Avian energetics40

1.7.1Introduction40

1.7.2Methods41

1.7.3Results41

1.7.3.1Daily energy expenditure41

1.7.3.2Moisture and energy contents of foods42

1.7.3.2Assimilation efficiency42

1.7.3.3Estimation of average daily food intake43

1.7.5Limitations of method45

1.7.5.1Appropriateness of data on food quality45

1.7.5.2Appropriateness of data on Daily Energy Expenditure45

1.7.5.3Plausibility of food intake predictions45

1.7.6Discussion – avian energetics45

1.8Discussion46

1.9References48

1.Foraging bird resource requirements and behaviour in conventional and alternative arable crop systems: a review of published research.

1.1 Background

In Britain, many species of farmland birds are undergoing long term population declines and range contractions (Fuller et al. 1995; Siriwardena, Baillie & Wilson 1998). A reduction in food availability at critical periods of the life cycle is likely to have been responsible for driving many of the observed population changes (Wilson et al. 1999; Marshall et al. 2001). Before models of the effects of pesticides on the food webs of farmland birds are considered, it is useful to review what is known about the relationship between farmland bird species and their food.

1.2 Aims

Using published literature this review aims to determine: (a) what is currently known about the availability of food for birds on arable farmland under different pesticide regimes, (b) what is known about the relative use made by farmland bird species of cropped areas receiving pesticide inputs and cropped and uncropped areas receiving no or reduced pesticide inputs, (c) the functional response of foraging birds to variation in their food supply and the effects of this variation on survival and fledging success of chicks and (d) how avian energetics may bridge the gap between our current understanding of behaviour and time budgets of foraging farmland bird species and their survival.

1.3 Introduction

This review addresses farmland bird species that feed primarily on plant material, seeds and on invertebrates dwelling on the ground or in crops and weeds. Twenty-one species fall into this category (Table 1.1). Invertebrates comprise a major component of the chick food of 16 out of 21 species listed. Of the remaining five species, four are pigeons or doves, which feed their young on crop milk and seeds. The final species, the linnet Carduelis cannabina, not only feeds its young on seeds but adults and juveniles rely on seeds throughout the year. Adults and juveniles of most of the remaining species take more than one food type, which varies seasonally. Eleven out of 17 resident species rely on invertebrates during the summer whereas only two out 17 do so during the winter. Seasonal variation in the incidence of seed taking is much less marked and fifteen out of the 17 residents feed on seeds in summer and winter. Invertebrates form a major component of the summer diet of three of the four summer migrants. The fourth summer migrant, the turtle dove Streptopelia turtur, feeds principally on seeds of weed species, fruits and cereal grain.

Table 1.1. Major food components of selected farmland bird species (chick food, summer & winter)(Cramp & Simmons (1977), Combreau, Fouillet & Guyomarc’h (1990), Evans (1997a), Buxton, Crocker & Pascual (1998)).

Bird species

Chick food

Summer food

Winter food

Chaffinch

Invertebrates

Invertebrates, seed

Seeds, plant material

Cirl Bunting

Invertebrates, seeds

Seeds, invertebrates

Seeds

Collared Dove

Crop milk, grain

Cereal grain, seeds

Cereal grain, seeds

Corn bunting

Invertebrates, seeds

Seeds, invertebrates

Seeds

Goldfinch

Invertebrates, seeds

Seeds, plant material, invertebrates

Seeds, plant material, invertebrates

Greenfinch

Seeds, invertebrates

Seeds, plant material

Seeds, plant material

Grey partridge

Invertebrates, plant material (seeds?)

Plant material, seeds, invertebrates

Plant material, seeds

House sparrow

Invertebrates

Seeds, invertebrates

Seeds

Linnet

Seeds

Seeds

Seeds

Quail*

Invertebrates

Seeds, invertebrates

N/a

Red-legged partridge

Invertebrates, seeds

Seeds, plant material

Plant material, seeds

Reed bunting

Invertebrates

Invertebrates, seeds

Seeds

Rook

Invertebrates, cereal

Invertebrates, cereals

Cereals, invertebrates

Skylark

Invertebrates

Invertebrates, plant material, seeds

Plant materials, seeds

Stock Dove

Seeds, crop milk

Seeds, plant material

Seeds, plant material

Stone Curlew*

Invertebrates

Invertebrates

N/a

Tree sparrow

Invertebrates

Seeds, invertebrates

Seeds

Turtle dove*

Seeds

Seeds, fruits of weeds, cereals

N/a

Wood Pigeon

Seeds, plant material, crop milk

Plant material, seeds, fruit

Plant material, seeds, fruit

Yellowhammer

Invertebrates, unripe grain

Invertebrates

Seeds

Yellow Wagtail*

Invertebrates

Invertebrates

N/a

* summer migrants.

The invertebrate taxa, which are known to comprise important components of farmland bird diets, are listed in Table 1.2. Important arable plants (weed species) in the diet of farmland birds are listed in Table 1.3.

Table 1.2Important invertebrate taxa in the diet of farmland birds, and those especially associated with declining farmland species (from: Wilson, Arroyo & Clark 1996; Wilson et al. 1999).

A taxon was rated as important if it comprised a quantitative mean of at least 5% in the diet.

Invertebrate taxa (order)

Invertebrate taxa (family or sub-order)

Found to be prevalent in diet of declining bird species

Arachnida

*

Araneae

*

Coleoptera

Carabidae

*

Chrysomelidae

*

Curculionidae

*

Orthoptera

*

Acrididae

*

Diptera

Tipulidae

Hemiptera

Aphididae

Hymenoptera

*

Formicidae

Symphyta

*

Lepidoptera

*

Table 1.3. Important plant families and genera in the diet of farmland birds (Marshall et al. 2001).

Very important

Important

Family

Poaceae

Compositae

Polygonaceae

Labiatae

Chenopodiaceae

Boraginaceae

Caryophyllaceae

Violaceae

Cruciferae

Genus

Stellaria

Cerastium

Chenopodium

Sinapis

Polygonum

Viola

Poa

Rumex

Senecio

This review will first consider food availability, and describe what is known about differences in food availability between different cropped and uncropped farmland habitats, especially in relation to pesticide management. Birds are often found not to use food resources in proportion to their availability (Olsson et al. 2001), i.e. they often exhibit preferences. The preference for different cropped and uncropped habitats will therefore be examined, in terms of the relative use made by farmland birds of these habitats. The functional response of birds to changes in food availability will then be considered, and examples given of the effects of these changes on survival and reproductive success. The amount of food available and the birds’ foraging behaviour (the functional response and any preferences) determine the amount of food eaten. The amount of food eaten in relation to the amount of food required may affect breeding success and survival. Therefore, in the final section of this review we will consider the nutritional value of bird food in relation to the amount of food needed – avian energetics.

Other sections of this report will also consider some of the subjects covered by this review. Objective 2 (an analysis of available models) and Objective 8 (further work on functional responses needed for the depletion model) will consider the functional response. The nutritional value of bird foods will be reviewed for Objective 3 (ranking weed species in terms of food value) and Objective 5 (propose groupings of bird food in terms of their value and availability).

1.4 Food availability

1.4.1 Introduction

Food availability in this review is defined as the quantity of potential food resources present in an area, which may be utilised by a bird searching for food (adapted from Wolda 1990).

The difference between “abundance” and “availability” in terms of food resources has only recently been appreciated in studies of bird utilisation of farmland habitats (Moorcroft et al., 2002; Hart et al., 2002). Indeed, the terms seem to be interchangeable in many studies. For instance, Aebischer (1998) refers to food availability in equivalent terms to food abundance for species such as grey partridge Perdix perdix, lapwing Vanellus vanellus, skylark Alauda arvensis and corn bunting Miliaria calandra. Aebischer & Ward (1997) and Brickle et al. (2000) related corn bunting distribution to crop type and invertebrate abundance, rather than availability. Benton et al. (2002) linked farmland bird declines to insect abundance. Campbell et al. (1997) discussed trends in the abundance of food of farmland birds. Chamberlain & Wilson (2000) discussed invertebrate abundance on organic and conventional farms in relation to birds. Moreby & Southway (2002) referred to the availability of invertebrate groups in relation to cropping and year effects, but actually equated abundance to availability.

As the vast majority of studies relevant to this review equated abundance to availability, these are included, but with the proviso that the abundance of a group or taxa may not necessarily reflect its availability to a particular bird species or guild.

Hutto (1990) argued that measurements of food abundance may not reflect food availability, because of scale-of-measurement problems and measures of availability being equated with standing crops, even though bird behaviour can depend on the renewability of the resource. A bird’s perception also differs from measures of availability because of a food resource’s crypticity, inaccessibility, difficulty of capture, and mechanical or chemical defences (Hutto 1990). The size, life stage, palatability, coloration, activity patterns and other characteristics of arthropods act as “translators”, which relate abundance to availability (Cooper & Whitmore 1990).

Scale-of-measurement factors affect availability measurements, as food is not uniformly distributed throughout a bird’s feeding territory, so extrapolations of abundance may not be accurate. Some food resources (e.g. fruit and seeds) are not continuously renewing, whereas others (e.g. some invertebrates, by colonising an area), can renew. Bird behaviour is affected by the depletion/renewal of a food resource (Hutto 1990). Hutto (1990) suggested that quantitative measures of behaviour that are correlated with food abundance might provide a “check” on the reliability of food availability measurements. However, extensive fieldwork is usually required to quantify these measures (Poulin & Lefebvre 1997).

A knowledge of food availability, how it differs between habitats and in response to management, and how it changes over time, is a necessary prerequisite for both the depletion and deterministic models being considered in this project.

Food availability will therefore be considered here in terms of both food abundance and accessibility. The factors that affect food abundance include vegetation cover (crops and other farmland cover), the effects of crop management practices, pesticide usage, environmental and temporal effects.

1.4.2Effects of crop type

It has been shown repeatedly that the abundance of arthropods taken by farmland birds during the breeding season varies between arable crop types (e.g. Reddersen 1994; De Cornulier et al. 1997). In some of the most recent studies, field data on the abundance of arthropod prey have been matched to the results of dietary studies for particular bird species, e.g. grey partridge and yellowhammer, to derive species-specific measures of food abundance.

Moreby & Southway (2002) showed that the combined abundance of key arthropod groups in the diet of grey partridge chicks varied between crops on a farm in Leicestershire. Densities of prey in wild bird food cover were much higher than in grazed grassland or any of four arable crops (winter wheat, winter barley, winter oilseed rape & winter beans). Within the arable crops, densities were similar in winter cereals and beans but they were much lower in winter oilseed rape. These differences were maintained between years.

Holland et al. (2002) estimated arthropod densities in a range of arable crops from two farms, in Hampshire and Lincolnshire. They derived measures of food abundance for grey partridge chicks and yellowhammer. The grey partridge Chick Food Index (CFI) varied seasonally but differed significantly between crops on both farms. It tended to be highest in peas and spring beans, and lowest in root crops (potatoes and sugar beet). Cereals had intermediate values. At both farms, the monthly CFIs for spring barley tended to be higher than that for winter wheat. Densities of arthropods taken by yellowhammers also varied between crops in Hampshire (field edge only) and in Lincolnshire (field centre and edge). The ranking of crops on the Hampshire farm, in descending order, by their mean monthly densities of yellowhammer prey (winter barley>winter oilseed rape>spring barley>winter wheat) differed markedly from that for the grey Partridge CFI (winter wheat>winter oilseed rape=winter barley>spring barley).

A third study examined differences between crops in the abundance of arthropod prey taken by corn buntings on arable farmland in Sussex (Ward & Aebischer 1994). Densities of key arthropod groups (caterpillars of Lepidoptera & Symphyta, spiders and harvestmen) were higher in spring than in autumn sown crops.

A finding common to all three studies was that spring-sown cereals tended to support higher densities of grey partridge chick food than winter cereals. However, the crop rankings were not consistent across farms. In particular, Moreby & Southway (2002) reported that the grey partridge CFI for winter oilseed rape was lower than those for cereals, whereas it was higher on the farms surveyed by Holland et al. (2002).

Comparative assessments of the abundances of individual arthropod taxa listed by Wilson et al. (1999) in different arable crops are numerous. Coleoptera: Several studies have reported that numbers of carabid and/or staphylinid beetles differ between cereals, brassicas and root crops (Booij & Noorlander 1992; Holland et al. 1994b; Thiele 1997). Spring-sown crops tend to have fewer carabids than winter-sown crops, and a different suite of species is present. Spring-breeding carabids (overwintering as adults) were more common in winter cereals than spring-sown crops (Luff & Sanderson 1992). Diptera: Reddersen (1994) found that winter cereals held more lauxaniid flies than spring cereals. Hymenoptera: Sawflies (Symphyta) are an important chick-food for some farmland species, including grey partridge, pheasant Phasianus colchicus, skylark, reed bunting Emberiza schoeniclus and corn bunting (Barker & Reynolds 2000). Sawflies differ in density between different crops. Densities of Dolerus and Pachynematus larvae were low in spring barley, winter wheat and heavily-grazed permanent pasture but much higher in perennial rye-grass Lolium perenne crops (Barker, Brown & Reynolds 1999). Arachnida: Frank & Nentwig (1995) found that Araneae were found in greater numbers on oilseed rape than on bean or wheat crops.

Differences between crops in the densities of arthropods may be due differences in crop structure, the presence or absence of particular non-crop host plants, the micro-climate in the crop, and to crop management, including pesticide inputs and timing of sowing. In most cases, the relative importance of these factors has not been assessed but Booij & Noorlander (1992) found that carabid diversity and density was greatest in crops with early and persistent ground cover, and with no or reduced cultivations. These relationships were not contingent upon pesticide regime.

A change to winter cereals from spring cereals is likely to result in a 25% reduction in weed density and species diversity (Hald, 1999) In addition, plants that are important food resources for arthropod herbivores occurred at greater densities in spring rather than winter cereals.Barley crops have been found to have greater number of weed seeds than wheat crops, possibly because they receive fewer herbicide treatments (Thomas, Garthwaite & Banham 1996; Don 1997).

Summary: the diversity and abundance of arthropods taken as prey by birds differ between crops. This may be partially due to differences in vegetative structures between crops, and partially to the effects of management regimes specific to particular crops. Variation between crops in arthropod densities differ between taxonomic groups. For grey partridge chick food, spring sown cereals held higher densities than winter sown crops, whereas opposite applied to the densities of carabid beetles. There is some evidence that weed densities are higher in spring than in winter cereals.

1.4.3Effects of management of non-crop habitats

1.4.3.1 Stubble

Stubble fields are often rich in spilt grain and weed seeds and comprise an important food resource for seed-eating farmland birds. Robinson & Sutherland (1997) showed that seed densities in the upper soil horizons on cereal (mean density = 1093 seeds m-2) and linseed (4558 seeds m-2) stubbles were much higher than those on other crops including winter cereals (410 seeds m-2). However, Hart et al. (2002) found that other crop types (tilled fields, autumn-sown crops) had similar or higher seed densities to stubble. Seed densities also vary between different types of cereal stubble and are dependent on stubble management. Barley stubble had significantly higher seed abundance than wheat stubble (Moorcroft et al. 2002). Barley stubble is sparser and weedier than wheat stubble, as barley has a short straw and is harvested earlier. Undersowing may reduce seed densities. Non-undersown stubbles had seed densities of 1566 m-2 in November, whereas the equivalent density on undersown stubbles was 293 m –2 (Robinson & Sutherland 1999). Seed density was found to be high on organic stubble fields by Moorcroft et al. (2002), but food accessibility may be lower as a result of previous undersowing and greater ground cover inhibiting weed establishment. Stubbles, which had been sprayed post-harvest with herbicide, had the lowest seed densities (Robinson 2001).

The switch from spring to autumn sowing of most arable crops has reduced the availability of winter stubbles, thereby greatly reducing the amount of weed seed available during the winter (Campbell et al. 1997). This is believed to have affected the populations of seed-eating passerines, particularly corn bunting and cirl bunting Emberiza cirlus (Baillie et al. 1997). The quantity of seed available on the remaining stubble fields has also decreased significantly (Donald 1998; Robinson & Sutherland 1999).

The differences in seed densities between crop types reported by Robinson & Sutherland (1999) appeared to be sufficient to affect crop type preferences by foraging birds. Winter densities of four seed-eating species (skylark, yellowhammer, grey partridge and red-legged partridge) were positively correlated with seed density in the upper soil horizons and were often, but not always, highest in stubble fields. However, Hart et al. (2002) found yellowhammer and skylark field preferences were affected by crop type rather than seed density. Stubbles did not contain the highest seed densities but they was favoured by foraging birds possibly because the habitat provided better cover from potential predators than other crop types or because seeds in stubbles were more accessible to foraging birds.

Though undersown stubbles contain less food for seed-eaters in winter, they are of high value for invertebrate-feeding birds in spring and summer. Many invertebrates benefit from the lack of cultivation, as sown grass leys follow the harvesting of the arable crop (Barker, Vinson & Boatman 1997). The retention of overwinter stubbles was shown to increase carabid larvae numbers but did not affect adult abundance in the following summer (Gardner et al. 2001).

Summary: stubbles are generally perceived as an important winter food source for farmland birds. The switch from spring-sown to autumn-sown crops has reduced stubble area and food availability for many birds. Seed abundance may be similar to other crops, but it may also be more accessible for various reasons.

1.4.3.2 Set-aside

Set-aside was introduced as a voluntary scheme in 1988 as a means of reducing agricultural surpluses (Evans et al. 1997b). Set-aside was widely taken up in 1992 when Arable Area Payments to farmers became conditional on farmers leaving part of their land fallow as set-aside. Set-aside can be rotational (for one year only), or non-rotational (for longer periods). Set-aside must have a green cover over winter. The rotational set-aside is managed as part of the crop rotation, with non-rotational becoming semi-permanent. The vegetation cover on set-aside can be either naturally regenerated, industrial crops, sown grass or wild bird cover. Herbicides (usually glyphosate) may be used after April 15th on naturally regenerated set-aside to suppress weeds. Set-aside left to regenerate naturally is effectively traditional stubble, but remains fallow all summer, unlike crop stubbles.

Seed rain from set-aside can be very high (Jones & Naylor 1992). Set-aside can thus be a valuable source of seeds for birds, and of greater value than adjacent cropped farmland. Donald et al. (2001a) found that set-aside occupied by wintering skylarks had significantly higher weed seed availability than set-aside that was unoccupied, suggesting that there may be marked differences in food availability between different set-aside fields, possibly related to previous and current management. Draycott et al. (1998) found considerable variation in seed densities on set-aside in late winter, with many having very low densities. Even so, set-aside in general had significantly higher numbers of weed seeds, compared to arable fields. Chenopodium spp. and Polygonum spp. were the most common weed species found on all field types (spring sown, autumn sown and set-aside). These are important diet components of many farmland birds (Marshall et al. 2001).

Set-aside often supports a greater abundance and diversity of invertebrates compared to other crop types (Kennedy 1992; Moreby & Aebischer 1992; Holland et al. 1994b; Poulsen et al. 1998; Henderson et al. 2000b; Henderson, Vickery & Fuller, 2000a) but this is not always the case. For example, Gates et al. (1997) found that numbers of invertebrates were generally lower on set-aside than in adjacent field boundaries.

Skylark broods reared on set-aside were found to be fed more on soft-bodied invertebrates than those reared on grassland or spring cereals (Donald 1999; Poulsen, Sotherton & Aebischer 1998). More Hemiptera, spiders, Araneae and sawfly larvae were provisioned, and fewer adult Lepidoptera, Coleoptera, Diptera and other Hymenoptera. This may reflect differences in prey availability between the crop types, as the abundance of skylark nestling food was also found to be much greater in set-aside than in grassland or spring cereals (Poulsen et al. 1998).

The invertebrate fauna of set-aside may depend on its cover, age and management (Moreby & Aebischer 1992; Moreby & Southway 2002; Moreby & Southway 2000). Cutting set-aside has been found to reduce densities of most arthropod groups (Moreby & Southway 2000). First-year set-aside has large numbers of small invertebrates that are preferred by skylarks whereas cirl buntings and yellowhammers prefer older set-aside, which has more large prey items such as orthopterans (Evans et al. 1997b). Rotational set-aside was preferred by birds to non-rotational set-aside, as the diversity and structure of the former type probably supported a greater abundance and availability of food (Henderson et al. 2000b).

Wild bird cover comprises a small but important proportion of set-aside cover. Boatman, Stoate & Watts (2000) showed that wild bird cover is preferred to farm crops by farmland seed-eaters. Seed abundance and variety is greater in wild bird cover than in other crops. A greater number of chick-food invertebrates were also found in wild bird cover. Moreby & Southway (2002) found that wild bird cover (either a cereal-based mixture, or a kale-based mixture) contained the highest numbers of invertebrates, followed by wheat and barley crops.

Summary: set-aside often supports higher abundances and diversities of arthropods compared to arable crops and grassland. Management regimes and enhancement (eg. wild bird cover) of set-aside affect the availability of food on set-aside, and some bird species may prefer one type to another.

1.4.3.3 Conservation headlands

Conservation Headlands are cereal crop edges (6m strips) that are not treated with insecticides in summer, and receive only selective herbicides (Potts 1997). They support more weed species, higher weed densities, and a greater weed biomass than fully-sprayed headlands. Sotherton (1992) found a 10-fold increase in broad-leaved weed diversity, with an attendant increase in weed numbers, in conservation headlands because of reduced spraying.

By not spraying 6m wide crop margins with herbicides and insecticides, insect abundance was found to increase by three times (de Snoo & de Leeuw 1996). Densities of partridge chick-food arthropods in conservation headlands may be two or three times as high as those of fully-sprayed headlands (Rands 1985; Sotherton, Rands & Moreby 1985; Sotherton, Boatman & Rands 1989; Chiverton & Sotherton 1991; Aebischer 1998). Aphid predator numbers also increased in relation to the number of aphids present (Chiverton & Sotherton 1991; de Snoo & de Leeuw 1996). Hassall et al. (1992) found carabid density and diversity greatest on Conservation Headlands, compared to control field margins. However, a more recent study found no effect on carabid activity when Conservation Headlands were present, compared to control field margins (Gardner et al. 2001), while Chiverton & Sotherton (1991) also found no significant differences in numbers of the two most common carabids between sprayed and unsprayed headlands. Butterfly numbers increased on Conservation Headlands compared to sprayed headlands (Dover, Sotherton & Gobbett 1990). Spring emerging butterflies also foraged more on field margins containing unsprayed headlands than on field margins lacking headlands (Dover 1997).

Summary: the reduced pesticide applications on conservation headlands appear to positively affect the abundance of non-crop plants and arthropods taken by farmland birds.

1.4.3.4 Field margins

Field margins can be categorised as “boundaries” (eg. hedges, windbreaks, verges, ditches etc; separating one field from the next), “boundary strips” (eg. grass margins, herbaceous strips, farm tracks; situated between the boundary and the crop) and “crop margins” (crop edges, the outer edge of the crop, sometimes cultivated separately as conservation, managed or unfertilised headlands) (Greaves & Marshall 1987; Thomas 1996). This section refers to the first two categories.

Field margins are important refuges for biodiversity on arable farmland (Boatman 1994; Hopkins 1997; Thomas & Marshall 1999; Pfiffner & Luka 2000). Several farmland bird species can be regarded as hedgerow specialists: eg. Whitethroat Sylvia communis, Linnet, Greenfinch Carduelis chloris and Yellowhammer (Fuller et al. 2001), indicating the importance of field margins to farmland birds.

The lack of suitable overwintering sites, such as field margins, may influence arthropod predator densities (Thomas, Wratten & Sotherton 1992). Sotherton (1984) found that densities of overwintering polyphagous predators were significantly higher on field boundaries (1.39/m-2), wintersown cereals (1.41/m-2) and established grass leys (1.36/m-2), than on woodland (0.78/m-2), cereal stubbles (0.60/m-2), stubble turnips (0.77/m-2) and first-year leys (0.61/m-2). Individual species differed in their habitat preferences.

Butterflies tend to be associated with field margins rather than crops, as food and shelter are provided in the former (Hopkins & Feber 1997). Higher numbers of Lepidoptera larvae were thought to occur in adjacent mature field boundaries and grassy headland strips than in the crops themselves (Boatman 1998).

Andersen (1997) found that carabids and staphylinids were significantly more abundant in the boundaries between spring cereals and grass fields, than in the fields themselves. In a review of carabid densities, a range of 14. 5 to 1113 beetles m-2 (average of 233 beetle m-2) was found in field boundaries (Lovei & Sunderland 1996). Field margins have been shown to be significant in benefiting carabid populations (Hawthorne & Hassall 1995; Hopkins 1997; Lys & Nentwig 1994). Factors such as boundary structure (vegetation and soil characteristics, which influence micro-climate), pre-winter crop husbandry, food supply and parasitism may affect the dispersal and abundance of arthropods within field margins, indicating the complex interactions that determine overall densities in these areas (Thiele 1964; Coombes & Sotherton 1986; Dennis, Thomas & Sotherton 1994).

There is considerable movement between field margins and crop fields (Wratten & Thomas 1990). Field margins provide refugia for many arthropods, so their presence can affect the abundance of certain groups within crops, as well as within field margins. Crops had significantly higher numbers of predatory arthropods when adjacent to field margins (Dennis & Fry 1992). Variation in densities of over-wintering arthropods may also be related to the “landscape” scale matrix of field margins, so that abundances may need to be related to a farm-scale, rather than a single-field scale (Dennis & Fry 1992).

Management techniques influenced the abundance of butterflies in field margins, with grass-cutting, herbicide use and sowing the most important factors (Feber, Smith & Macdonald 1996).

The provision of grass margins to arable fields can restore Orthoptera populations locally in intensive farmland (Hill et al. 1995). If wide field margins, seeded strips and good edge management are employed in a farm regime, arthropod diversity and abundance will be enhanced (Gardner & Brown 1998).

Beetle banks are raised strips planted with tussocky grasses, located across the centre of fields (Thomas et al. 1992; Thomas, Goulson & Holland 2001). They support large numbers of predatory insects, which move into the adjacent crop during the summer (Thomas et al. 1992). Carabids are the most important group of epigaeic arthropods found in agro-ecosystems (Tischler 1980). Experimental studies by Lys & Nentwig (1994), Nentwig (1989) and Thomas, Wratten & Sotherton (1991) showed that beetle banks supported greater numbers of carabids than the surrounding crop, with densities of up to 1500 beetles/m2. Fields with 1.5 m wide beetle bank strips had 12 times the arthropod activity level of non-strip fields (Lys & Nentwig 1994). Thomas et al. (2001) found that beetle banks had lower overall densities of chick-food invertebrates than conventional field margins, but are important in the absence of well-managed hedgerows. However, the numbers of many key invertebrate groups did not differ significantly between beetle banks and other types of field margins (Thomas et al. 2001).

Field margins are vulnerable to pesticide spray drift. The effects of herbicide on spiders in an arable field margin were studied by Haughton et al. (1999). An arable field margin was sprayed with three rates of glyphosate (90 g active ingredient/hectare (a.i/ha), 180 g a.i./ha & 360 g a.i./ha) and control plots left unsprayed. The highest rate of glyphosate consistently reduced the total number of spiders, and the numbers of web-spinners, but not numbers of wandering spiders. Changes in vegetation structure and microclimate caused by the glyphosate were implicated in the reduction of numbers of spiders in plots receiving the highest rate of glyphosate. De Snoo (1994) found that weed cover and butterfly abundance increased significantly in unsprayed margins, though soil invertebrates such as carabids were less affected. High rates of mortality in butterfly larvae have been attributed to spray drift of pesticides (Longley et al. 1997; Longley & Sotherton 1997).

Summary: non-crop field margins tend to have higher numbers of invertebrates than the adjacent crop. They are also an important source of colonising and re-colonising invertebrates into an adjacent crop. The presence or absence of field margins, and their structure, are important determinants of a crop’s invertebrate fauna. Field margins also have to be viewed in a landscape context, rather than at a single-field scale.

1.4.4 Effects of position within a field

The variability of invertebrate abundance within a field can often be shown to be greater than the variability between fields under different management systems (e.g. Green et al. 1995). Within a cropped area, arthropod densities may differ between micro-habitats. Arthropod density was found to be highest within spring barley crops, compared to the tramline and edge of tramline areas of the crop (Odderskaer et al. 1997). However, food availability was reduced in the cropped areas, because of vegetation cover, and foraging areas reduced to some 5% of the field area. The spatial heterogeneity of arthropod distribution has important implications for appraising pesticide effects (Holland, Winder & Perry 2000). Most groups of beneficial arthropods show such distributions, and this may be why some large-scale trials do not detect expected reductions after pesticide applications.

Reddersen (1997) found that arthropod density, abundance and biomass decreased with increasing distance from the crop margins. This edge effect disappeared after 9 m, in both organic and conventional crops. Gardner & Brown (1998) found the edge effect less distinct on conventional fields than on integrated fields, the reverse of the results found in other studies (Reddersen 1997). Panek (1997) found higher numbers of Heteroptera in small arable fields (<10 ha.) than in large arable fields (10-50 ha.) He related this to a diverse crop mosaic, as well as less intensive pesticide use. In small fields there was no difference in Heteropera numbers between the edge, 100m in the field, or the field centre. In large fields, bug numbers were higher at the edge, compared to 100m in, and the centre. Green (1984) found higher insect densities 5m into a cereal field, compared to 50m into a field. Heteropterans were at higher densities 3m from the edge, compared to further in (Moreby 1995). These higher densities are probably related to the presence of permanent cover at the field edge, acting as a reservoir for insects (Panek 1997). Insect densities, as exemplified by predatory arthropods, vary spatially within a crop, although a crop may appear to be a uniform habitat (Kromp 1999). Soil factors, prey densities, and heterogeneity of vegetation density and microclimate all influence arthropod densities (Powell et al., 1995).

Summary: arthropod densities vary spatially within a field because of soil factors, prey densities, vegetation structure and heterogeneity, microclimatic influences, crop management, and colonisation from edges. Because of this variability, it may be difficult to assess the impact of different management regimes and pesticide applications on arthropods.

1.4.5 Effects of agricultural practices

Wilson et al. (1999) reviewed current knowledge of the effects of modern agricultural practices on invertebrate and plant foods found in the diets of farmland granivorous birds. Information on the effects of agricultural practices was collated from 900 references (Table 1.4). Table 1.4 also indicates the most-researched topics relating to farmland practices, with the effects of pesticides being particularly dominant. Agricultural practices varied in their effects on invertebrate orders and plant families. Pesticide applications, specialisation of farmland, loss of uncultivated field margins and increasing frequency of tillage generally reduced invertebrate abundance. Gardner & Brown (1998) also summarised the effects of agricultural practices on plants and invertebrates. Invertebrate numbers were negatively affected by inversion ploughing and pest control (chemical and biological). Numbers increased comparatively with practices such as direct drilling, green manure/intercropping, set-aside and stubbles, rotation with grass leys and the provision of permanent pasture.

Synthetic fertiliser, non-organic slurry and weed control (mechanical and chemical) decreased plant and seed abundance. Weed seed abundance was enhanced by the use of post-cropping incorporation and the presence of set-aside and stubbles. The loss of invertebrate-rich pastures is thought to have affected several bird species (e.g. cirl bunting: Campbell et al. 1997). Food availability has been found to be lower in solely arable areas than in areas of mixed farming (Aebischer 1998). In heterogeneous farming landscapes, there is a higher biomass of most insect groups compared to simplified landscapes (Panek 1997). Simplified landscapes are characterised by large fields and a lack of permanent cover. Intensification reduces non-cropped areas, which are important as refugia and food sources for invertebrates (Dennis et al. 1994; Andersen 1997).

Table 1.4. Number of references to specific agricultural operations and practices (adapted from Wilson et al. 1999).

Agricultural operation

Number of references

Application of insecticides

173

Application of fertilisers

154

Application of herbicides

138

Application of unspecified agricultural sprays

20

Application of fungicides

17

Use of avermectins

4

Application of manures

17

Ploughing regime

89

Management of uncropped field margins

29

Crop rotation/monocultures

28

Sowing grasses

10

Sowing arable crops

29

Field drainage

6

Biological control

5

Mechanisation

5

Seed-cleaning techniques

3

Pollutants

1

1.4.5.1 Invertebrates

Benton et al. (2002) found that changes in arthropod abundance were linked to changes in agricultural practice in Scotland, and that their findings were probably applicable to Britain overall. Intensive farm practices such as fertiliser input, pesticide usage and increased winter wheat sowing reduced arthropod abundance. Ewald & Aebischer (1999) also found that intensification (increased outputs per unit area of crop) of farming resulted in a decrease in biodiversity and abundance of many arable weeds and invertebrate species, affecting food availability. The loss of traditional rotational cropping is thought to have decreased invertebrate numbers in both autumn and spring-sown cereals. Aebischer & Ward (1997) showed that numbers of large Coleoptera, Lepidoptera/Symphyta and Arenae/Opiliones were significantly lower in both types of crops in 1994, compared to 1970. The difference may be related to the increasing use of foliar fungicides and insecticides over time on cereal crops (Aebischer, 1998). Krooss & Schaefer (1998) studied the effects of varying farming practices on rove beetles. Reduced tillage and fewer pesticide applications enhanced abundance, whereas no fertilisation led to reduced population densities because of sparse crops with unfavourable microclimates. All these studies considered the effects of a range of agricultural practices and it is often difficult to isolate the effects of particular ones.

Crop cultivation may reduce soil invertebrate populations, by direct mechanical damage, loss of insulating vegetation and the use of pesticides (Wilson, Taylor & Muirhead 1996a). Tillage often reduces arthropod populations, as many insects spend a life stage (eg. pupa) in the soil (Norris & Kogan 2000). Ploughing was shown to reduce numbers of linyphiid spiders by 89% whereas, over the same period, numbers on unploughed fields increased by 43-105% (Thomas & Jepson 1997). Inversion ploughing, a common practice on both organic and conventional farms, reduced invertebrate, particularly earthworm, abundance (Fuller 1998; Edwards & Lofty 1982b). The recent change from spring ploughing to later summer ploughing of cereal stubbles has adversely affected invertebrate food availability for birds in spring, by removing the fresh spring till (Newton 1998). Autumn ploughing has an adverse effect on carabid beetles overwintering as adults (Holland et al., 1996). Minimal tillage, where the soil is not turned over, results in increased survival of polyphagous predators, such as carabids, which are important in bird diets (Reddersen 1994). Increased applications of farmyard manure increased earthworm densities on cultivated fields (Tucker 1992).

Under-sown leys once covered almost 25% of the arable area of Europe (Potts 1997). Increased land usage, the decline in under-sowing, and the reduction of fallow periods for land, has very probably resulted in decreases in the survival of insects such as sawflies that over-winter as eggs or larvae. These insects are important components of farmland bird diets (Barker et al. 1997; Wilson et al. 1999). Vickerman (1978) found that three times as many adult sawflies emerged from under-sown fields in the subsequent year, compared to cultivated fields. Potts (1986) attributed this effect to the damage caused by ploughing and cultivation. Where fields are under-sown, sawfly densities remain high, as over-wintering insect pupae are not affected by cultivation after harvest (Aebischer 1998). Under-sown crops are weedier and more attractive to invertebrates than conventional crops (Aebischer 1998). Under-sowing may increase natural enemy abundance (Burn 1989).

The adoption of pure arable systems instead of mixed farming may be associated with declines in Coleoptera, Lepidoptera, Orthoptera and Arachnida populations (Wilson et al., 1999). Conversely, some common tipulids (Diptera) may increase in improved grassland swards, as do many hemipterans in arable monocultures (Wilson et al. 1999). Aebischer (1998) suggested that food availability is lower in purely arable areas, compared to areas of mixed farming.

Bruce et al. (1999) assessed the effects of sewage sludge on euedaphic and hemiedaphic Collembola in grassland. Five different treatments: uncontaminated (i.e. low levels of heavy metals) digested sludge; uncontaminated undigested sludge; zinc-rich digested sludge; copper-rich undigested sludge and no sludge control did not influence the overall abundance of Collembola in the study area, but significant differences were found at the species level. Seasonal and successional effects were also found and, for most species, these were more pronounced than the effects of sludge treatment.

Summary: modern farmland practices relating to the increasing intensification of farmland, generally have an adverse effect on invertebrates. Practices such as increased pesticide usage, loss of uncultivated field margins, increased tillage, and the decline in under-sown crops have all been shown to decrease food availability.

1.4.5.2 Plants

Important plant items in the diet of granivorous farmland birds include: cereal grain, knotgrasses and persicarias (Polygonum), chickweeds (Stellaria) and goosefoots (Chenopodium). Asteraceae, Fabaceae and Brassicaceae are also widely taken (Wilson et al., 1999). The latter include crop components, such as clovers and oilseed rape. Late summer ploughing has removed much of the spilled grain and other seeds that were formerly available (Newton 1998). There have been widespread reductions and declines in the diversity and abundance of many food plants on farmland as a result of modern farm techniques, such as frequent tillage, grassland management, herbicides and competition from farmland crops (Sotherton & Self 2000). Wilson et al. (1999) suggested that the main factors affecting grass seed and cereal grain availability for birds have been more efficient harvesting and storage of cereals, more frequent harvesting of grass, grazing intensification, early cultivation of cereals, and the elimination of under-sowing.

In contrast, some food plants may have increased in availability: e.g. Poa spp., curled dock (Rumex crispus), chickweeds, mouse-ears (Cerastium), and some Asteraceae, as they respond well to practices such as increased nitrogen application, frequent cutting or grazing (Wilson et al. 1999). Black-grass (Alopecurus myosuroides) and Barren Brome (Bromus sterilis) have increased in abundance because of the failure to bury seed during non-inversion cultivation (Sotherton & Self 2000). Improved harvesting efficiency and birds being unable to access grain stores because of measures to improve hygiene have also reduced grain availability (Brickle & Harper 2000). The major change in the number of weed species in the most important arable crop (winter wheat) occurred in the late 1970s, with the introduction of ioxynil + bromoxynil, which controlled a broad spectrum of weeds (Marshall et al. 2001).

Crop sowing date may also affect some broad-leaved weed species. Species that set seed over autumn and early winter may not set seed when winter cereal is sown immediately, but will on stubble prior to spring-drilling (Sotherton 1991; Sotherton & Self 2000). Spring sowing of wheat significantly increased weed germination and performance, and thereby the availability of weed seeds and invertebrates (Cosser et al. 1997). The introduction of continuous cropping is likely to have resulted in an increase in some weed populations, as annual rotations were an efficient method of controlling weed infestations (Rademacher, Koch & Hurle 1970).

Sowing practices may influence the availability of exposed cereal seed on fields (Pascual et al. 1999). Sowing depth, seedbed condition, and sowing method influence seed densities on cereal fields. Headlands often have higher surface seed densities than the main-field, due to poor soil conditions caused by more traffic, and double sowing and unintentional broadcasting of seed.

Application of fertilisers, especially nitrogen, has a major effect on plant species composition, and reduces the abundance of some plant species (Wilson, Boatman & Edwards 1990; Marshall et al. 2001). Fertiliser drift is thought to have a greater impact than herbicide drift on margin flora (Marshall et al. 2001).

Intensive management leads to a rapid decrease in abundance of weed species in the seed-bank (Roberts 1962), but numbers soon increase again if management is relaxed. In the TALISMAN project, spring sowing of crops was re-introduced, and herbicide usage halved (Squire, Rodger & Wright 2000). Most weed species increased in abundance, and some species increased dramatically. These latter were economically important weed species (Alopecurus myosuroides, Galium aparine, Papaver sp., Anagallis arvensis and Chenopodium album), which increased to such massive abundances (>10,000 m-2) as to have a potentially detrimental effect on the crop.

The trend to minimal cultivation of cereal crops will encourage annual grass weeds, but annual broad-leaved weed abundance will decline (Froud-Williams, Chancellor & Drennan 1984). Herbicides and species-specific periodicity of germination were thought to be important factors influencing these trends. More recent studies have, however, found that there is a general increase in weed species with reduced tillage (Torresen & Skuterud 2002). Tillage has different effects on broad-leaved weeds, depending on the particular species’ ecological characteristics (Pollard & Cussans 1981). Separate studies have found contrasting results on the same weed species (eg. Froud-Williams et al. 1984; Torresen & Skuterud 2002).

Summary: as for invertebrates, modern farming practices have significantly reduced weed plant and seed (both crop and weed) availability. Frequent tillage, increased fertiliser usage, improved harvesting and storage of cereals, increased herbicide usage and the switch to autumn-sown crops all adversely affect broad-leaved weed populations. A few weed species may have increased as a response to some of these practices.

1.4.6Effects of pesticides

1.4.6.1. Insecticides: factors affecting their impact on non-target taxa

The effects of insecticides on arthropod diversity and abundance may vary depending upon weather effects, crop type and growth stage, and differences in arthropod species ecology (e.g. Vickerman 1992; Young et al. 2001). Spatial distribution, the scale of the trial and natural population fluctuations also influence test results (Holland et al. 2000). The impact of an insecticide is also mediated by a combination of chemical, toxicological, ecological and operational factors (Jepson 1989). Many studies have, however, indicated an adverse effect on non-target arthropods (Barrett 1968; Vickerman & Sunderland 1977; Potts 1986; Theiling & Croft 1989; Inglesfield 1989a; Everts et al. 1989; Davis, Lakhani & Yates 1991; Wilson et al. 1999).

Holland et al. (2002) found insecticide effects only in spring beans, and not in winter wheat, potatoes, peas, sugar beet, linseed and lucerne. Spring beans were sprayed twice (mid-May and mid-June); other crops were sprayed once in mid-May. Invertebrate numbers subsequently recovered as a result of reinvasion from surrounding fields. Although carabids are affected by pesticides in laboratory tests, other factors in the natural environment such as temperature, scale and re-immigration may be important in determining numbers (Heimbach & Baloch 1994; Heimbach & Abel 1994; Purvis 1992). Chiverton (1984) found that pitfall catches of carabids increased after pesticide treatment, even though prey species declined. The apparent increase was perhaps due to increased activity of hungry survivors.

Basedow (1991) found the main factors affecting carabids included crop rotations, field size and margins, as well as insecticides. Booij & Noorlander (1988) and Hance, Gregoire-Wilbo & Lebrun (1990) also considered that carabids were more affected by crop type than by pesticide applications. Luff, Clements & Bale (1990) found that grassland carabid populations were affected by insecticide use and other management, but that soil water and density were the main factors affecting numbers.

Impact of particular insecticides may vary depending upon the timing and frequency of application but the evidence is not consistent. For example, Brickle et al. (2000) showed that corn bunting invertebrate chick-food density was negatively correlated with the number of insecticide applications, both on cereal fields and on other crop types. However, Holland et al. (2002) found that raising the number of insecticide applications did not always result in a decrease in invertebrate abundance. Indeed, for winter oilseed rape, there was an increase in invertebrate abundance, and for other crops (winter wheat, spring barley, winter barley) no discernible effect was apparent. A significant decrease was noted only in spring beans when extra insecticides were applied.

Different types of pesticides differ in their impact depending upon the type of active ingredient and formulation. For example, Jepson, Efe & Wiles (1995) recorded that dimethoate (an organo-phosphate) and deltamethrin (a pyrethroid) had similar levels of toxicity in the laboratory. However, in the field, it was found that dimethoate had a severe impact on invertebrates, whereas the impact of deltamethrin was far less (Vickerman et al. 1987). Similarly, Cole et al. (1986) found that c.10% of non-target species showed significant reductions in numbers, which persisted for two months after autumn spraying of dimethoate, whereas a summer application of cypermethrin affected 4-6% of non-target species for a month. Pyrethroid sprays could have a greater effect on flying insects, because of their contact knock-down action. Pyrethroids also have repellent effects on some insects, particularly Hymenoptera. Pirimicarb is a selective insecticide, controlling aphids, but not affecting other non-target arthropods (Ewald & Aebischer 1999). Granular insecticides incorporated in the soil before crop planting can have a major impact on soil invertebrates, but less so on flying or foliage invertebrates (Tones et al. 2000).

Thieling & Croft (1988; 1989) reviewed the effects of pesticides on arthropods, from the analysis of their SELCTV database. Comparisons were made between the various pesticides, and their effects on arthropods. Sub-lethal effects were also reviewed, as these also influenced population densities. Burn (1989) showed that groups of arthropod predators, defined by their dispersal abilities and over-wintering habitats, differed in their susceptibility to long-term effects of pesticides. Poorly-dispersing predators were highly susceptible, whereas groups that were more dispersive were less affected. Indirect effects were also discussed. Reductions in prey availability, reproductive rate and alterations in micro-habitats may also indirectly reduce arthropod populations.

Vickerman (1992) found that dispersal ability was the most important factor affecting long-term susceptibility to pesticides. Recovery times on farmland may be longer than surmised from small-scale trials, if large contiguous areas are sprayed. Many trials are carried out on smaller areas, where recolonisation may occur more readily (Campbell et al. 1997). Trials on differing sizes of plots showed that the scale of treatment affects the recovery of the invertebrate population (Duffield & Aebischer, 1994). Groups of arthropods also differed in their patterns and rates of recovery. Staphylinidae recovered the fastest, followed by Linyphiidae then Carabidae, due to their relative mobility. These groups colonised from the outside in, whereas aphids and springtails recovered from the inside outwards, as predation pressure was less (Duffield & Aebischer 1994).

Summary: The effects of insecticides on arthropod diversity and abundance may vary depending upon weather effects, crop type and growth stage, differences in arthropod species ecology, the active ingredient in the insecticide, and spatial scale of application.

1.4.6.2. Insecticide: levels of impact on non-target taxa

Wilson et al. (1999) reviewed the effects of insecticides on invertebrates. The number of detrimental (d) and non-detrimental (n) effects was given by the notation (d: n), to indicate the overall effects of insecticides on each group. Coleoptera (48:14), Diptera (8:1), Lepidoptera (5:0), Hymenoptera (22:11), Hemiptera (7:1), Orthoptera (2:0), Arachnida (17:6), and Annelida (5:3) were all generally reduced in numbers by insecticides.

Three long-term projects funded by MAFF (now Defra) have studied the environmental effects of contrasting pesticide regimes. The Boxworth project indicated that some species were susceptible to intensive pesticide use in winter, whilst others appeared to benefit (Greig-Smith, Frampton & Hardy 1992). The SCARAB project found that polyphagous predators were not affected over the long-term, although chloropyrifos did cause short-term reductions in some groups (Young et al. 2001). The TALISMAN project found only limited and short-term effects on non-target arthropods by insecticide spraying (Young et al. 2001). It was suggested that spring and summer spray applications are intercepted by crop foliage, and non-target invertebrates are, therefore, protected. Also, in winter arthropods are less active, and less susceptible to treatments. Spring and summer applications of deltamethrin had no effect on autumn-breeding carabids, but winter applications reduced carabid numbers by 30% (Matcham & Hawkes 1985). Other studies have also shown either decreased invertebrate food availability, or little effect (Rands 1985; Cigli & Jepson 1995; Moreby et al. 1994).

De Snoo (1999) found that phytophage insect presence and abundance increased significantly on unsprayed margins of winter wheat, sugar beet and potato crops. Butterflies in hedgerows have been shown to be susceptible to spray drift from insecticides such as deltamethrin (Davis et al. 1991; Cilgi & Jepson 1995). Initial mortality in Coleoptera after spraying may be high, between 60-90%, but some taxa may recover their abundance after a month, while others may remain suppressed for several months (Brown, White & Everett 1988). Purvis, Carter & Powell (1988) recorded a 70% decrease in carabid numbers after pyrethroid spraying, but complete recolonisation occurred within 2 months. Pyrethroids are generally considered to have no long-term adverse effects on entomophagous arthropods (Inglesfield, 1989b). Some autumn-breeding carabids had summer populations reduced by 50%, but rapid dispersal recovered these populations. Barrett (1968) found that Sevin (a carbamate insecticide) reduced arthropod biomass and numbers by 95%, but after seven weeks biomass, though not total numbers, recovered.

Casteels & de Clercq (1990) studied the effects of insecticides on epigeal arthropods in winter wheat. Parathion and dimethoate reduced carabid numbers by 28-29%, whilst pirimicarb, fenvalerate and phosaline had a non-significant effect. Parathion and dimethoate also reduced staphylinids by 67% and 31% respectively. Fenvalerate was most toxic to spiders (30% reduction), and phosalone to springtails (23-47% reduction). Effects were most important in the first weeks after application.

Cole et al. (1986) found that winter pesticide applications reduced Carabidae and Linyphiidae populations only until the following spring. Recolonisation is an important factor in population recovery. A reduction in beetle larvae did not affect the numbers of adults trapped the following spring (Cole et al., 1986). Basedow (1987) found an 81% decline in trapping rates, and a 90% decrease in biomass in carabids between 1971-1974 and 1978-1983, caused by increases in insecticide applications, particularly parathion.

Aebischer (1990) found that the use of dimethoate, an aphicide, suppressed sawfly numbers to such an extent that it would take seven years for numbers to fully recover. Sawflies are an important component of partridge and pheasant chick diets. Sotherton (1990) also found that synthetic pyrethroids adversely affected sawfly caterpillar numbers. The impact of a single treatment of a broad-spectrum insecticide can last several years because of the sawfly’s slow reproductive rate (Aebischer, 1990).

Some pesticides may have favourable indirect effects on some species (Frampton et al. 1992). Their predators may be reduced in numbers, thereby enhancing the prey’s survival and numbers. Springtail abundance was reduced by chloropryifos, but not by cypermethrin and pirimicarb (Frampton 1999). Springtails actually increased after cypermethrin applications. Competitive interactions may also result in some groups or life-stages increasing in abundance after pesticide use. Vickerman (1992) and Duffield & Aebischer (1994) found increased numbers of carabid larvae after spraying, possibly because adult numbers, and hence competition or predation, were reduced.

For aphid control, compounds containing pirimicarb were found to be less harmful to non-target arthropods (Anon. 1997). Insecticides differed in their effect on Heteroptera (Moreby, Sotherton & Jepson 1997). Phosalone and pirimicarb had little effect on heteropteran numbers, whereas demeton-S-methyl and dimethoate had significant short-term effects (Sotherton, 1991; Moreby et al., 1997). Dimethoate was also found to reduce densities of most non-target arthropods in cereals (Vickerman & Sunderland 1977). Dimethoate, parathion, phorate and fonophos are extremely toxic to carabids, whereas pirimicarb, fenvalerate, phosalone, chlorofenvinphos and carbofenothion had less or no effect (Casteels & de Clercq 1990; Campbell et al. 1997). However, dimethoate drift was found to have no long-term effect on non-target invertebrates, either in a crop or its field margin (Tones et al. 2000).

Summary: measuring the effects of insecticides on food availability is difficult to assess, due to bias in sampling methods caused by trap effectiveness, trap type, spatial variability, insect activity, temperature, ground cover, vegetation structure, micro-habitat preferences, trial scale, natural population fluctuations and crop effects. Insecticide impact is also mediated by a combination of chemical, toxicological, ecological and operational factors. Many studies have found that insecticides reduce invertebrate numbers but this may be a short-term effect in some instances. Long-term (30 years) declines in selected arthropod taxa have been recorded from arable farmland.

1.4.6.3. Herbicide: effects on weeds

Recurrent use of herbicides can detrimentally affect the numbers and diversity of weeds over time (Fryer & Chancellor 1970), However, not all declines in weed numbers in fields can be attributed to changes in herbicide use, and some are due to changes in cropping patterns (Marshall et al. 2001). Wilson et al. (1990) considered that changes in sowing dates, seed cleaning improvements and increased fertilisation may be at least as important as herbicides in determining weed abundance. The effects of herbicides are mediated by herbicide characteristics, weather, soil, crop conditions (Boatman 1988).

Weeds are killed by herbicides before they can produce seeds, and this leads to progressive depletion of the seed bank. Linnets responded by switching to oil-seed rape as charlock Sinapsis arvensis and other species became rarer due to herbicide use (Campbell et al. 1997; Moorcroft & Bradbury 1998). Comparisons of organic v. conventional cereals (Brookes et al., 1995; Halberg 1997; Moreby et al. 1994), and sprayed v. unsprayed field headlands (Chiverton & Sotherton 1991; Sotherton et al. 1985), have demonstrated the scale of reduction of arable weed densities following herbicide application. Powell, Dean & Dewar (1985) found that autumn herbicide applications reduced weed numbers by over 90%, and that weed populations were still reduced the following April.

Ewald & Aebischer (1999) found that broad-leaved weeds were affected by dicotyledon-specific herbicide use, and grass weeds by broad-spectrum herbicide use. Contact and contact + residual herbicides reduced the abundance of both groups. No significant temporal trends were found in the 25 years of the Game Conservancy Trust (GCT) Sussex study, however. This may have been because herbicide use had already altered the weed flora at the beginning of the study. Herbicide use in spring and summer, rather than in autumn, reduced densities of Fallopia convolvus, Sinapsis arvenis, Viola arvensis, Chenopdium spp., mayweeds and Capsella bursa-pastoris, several of which are important in farmland bird diet (Table 1.3). However, Whitehead & Wright (1989) found that the weed species most widespread in 1967 had maintained their ranges since then, despite increased herbicide use.

Herbicides also have an effect on weed seeds in set-aside, as a result of their usage in the previous crop (Campbell et al. 1997). Robinson & Sutherland (2002) noted that there is evidence of declining seed banks in arable land in Britain, as a result of persistent herbicide use. A similar trend has been reported in Denmark (Jensen & Kjellsson 1995). Viable seed density declined by 50% in Danish arable fields between 1964 and 1989.

Summary: as with insecticide usage, herbicide effects are confounded by other variables such as crop management and natural variation. Generally, increased herbicide usage reduces overall weed and seed abundance, but some weed species may be more tolerant to herbicide effects.

1.4.6.4 Herbicides: indirect effects on invertebrates

Herbicides can reduce the availability of invertebrate food for birds (Moreby & Southway 1999) but the mechanisms by which this occurs, and the scale of the effects on bird populations are, in many cases, not fully understood (Campbell et al. 1997; Marshall et al. 2001). Herbicide application reduces the abundance and diversity of weeds, thereby affecting herbivorous insects, which are important chick-food diet components (Sotherton 1982; Sotherton 1991). Field experiments entailing the cessation of herbicide applications resulted in greater weed growth and insect densities in unsprayed plots than in those receiving herbicides (Sotherton 1991). Moreby & Southway (1999) found that herbicide-treated winter cereal headlands had significantly lower weed density and diversity, and lower numbers of invertebrates, particularly those important in farmland bird diet (Heteroptera, Auchenorrhyncha, Coleoptera), compared to non-herbicide treated headlands.

Southwood & Cross (1969) suggested that the introduction of herbicides in the 1950s destroyed many of the host plants of invertebrates, and probably halved their abundance within cereals. Four of the five most important partridge-chick food groups are dependent on broad-leaved weeds, and herbicide use thereby reduces their abundance (Potts 1997). Chiverton & Sotherton (1991) found large differences in gamebird chick-food arthropod abundance between herbicide-treated and untreated plots. Unsprayed plots also had higher densities and more species of weeds.

Weed species vary in the diversity of insect species supported. Three target weed taxa (Stellaria media, Poa annua and Polygonum spp. including Fallopia convolvulus) support a high diversity of insects. The Asterceae have a particularly rich fauna with Senecio vulgaris and Cirsium arvense having around 50 insect species associated with them. Sonchus oleraceus, Tripleurospermum inodorum and Sinapis arvense (latter in Brassicaceae) are also species rich (Marshall et al. 2001). Reducing or eliminating certain weed species thereby also reduces the abundance and availability of invertebrates in a crop. Arthropod density has been shown to be up to three times greater in weedy fields than weed-free fields (Potts & Vickerman 1974). The presence of weeds in cereal fields benefits many arthropods, including Carabidae and Staphylinidae (Vickerman 1974; Moreby et al. 1994).

Herbicide usage induces species replacement; as one weed species is eliminated, another replaces it, often from an entirely different family (Freemark & Boutin 1995). This can lead to marked effects on invertebrate abundance, many invertebrates being host-specific, and may influence food availability. Hald & Reddersen (1990) found that herbicides had an immediate negative effect on arthropod species diversity and abundance in conventional fields. Herbicides reduced the numbers of arthropods by 50%, and their biomass by 66%, in barley fields, compared to “weedy” barley fields (Southwood & Cross 1969).

Aebischer (1991) found that cereal invertebrate numbers halved between 1970 and 1990, and this was linked to an increasing use of herbicides. Sotherton (1982) found that knotgrass beetle Gastrophysa polygoni larvae had significantly higher mortality on host plants treated with 2-4D herbicide, than on untreated plants.

Effects of herbicide application on invertebrate prey of birds not always negative. For example, higher numbers of large beetles have been captured in cereal crops with few weeds, suggesting that clean (herbicide-treated) crops are easier for the beetles to colonise (Powell et al., 1985). Overall numbers of carabids caught were almost twice as high in herbicide-treated crops, as in untreated crops, and higher in crops treated twice. This was due to the larger species being more abundant, or possibly the capture methods used measuring increased activity rather than abundance. Movement is also easier in clean crops, so beetles may be easier to capture. Increases in numbers of springtails after herbicide applications have been attributed to increased rates of litter input into the soil (Conrady 1986).

Differences in the type of herbicide used also affect insect food availability. Vickerman (1974) found that metoxuron + simazine control of rough meadow-grass Poa trivalis reduced insect biomass by 43% compared to weed control by mecocrop.

Summary: herbicides may reduce the abundance of arthropod prey through the destruction of host plants but other factors may also be important. Differences in invertebrate species ecology, such as mobility and over-wintering strategy, and interactions between species (e.g. predator-prey relationships) mean that some species may be more abundant in herbicide-treated crops than untreated crops.

1.4.6.5 Fungicides

Many fungicides have no significant effects on Heteropteran populations (Moreby et al. 1997). However, some formulations do significantly reduce populations of beneficial arthropods because of their insecticidal properties e.g. pyrazophos (Sotherton, Moreby & Langley 1987; Frampton 1988; Sotherton & Moreby 1988). Carbamate fungicides (and molluscides) are toxic to soil invertebrates, and reduce their populations (Tucker 1992). Sotherton (1989) reviewed the mortality effects of 27 single active-ingredient foliar fungicides on hoverfly and leaf beetle larvae. Most fungicides had very low mortality impacts, except for pyrozophos. Springtails have been found to be susceptible to fungicides (Frampton 1988). Rove beetles Tachyporus spp. declined substantially in numbers between 1970 and 1990 in a Sussex study, possibly because they feed on fungi as well as live prey, and foliar fungicides may have thus indirectly affected them (Aebischer 1991).

1.4.7Effects of whole farm agricultural systems

1.4.7.1 Integrated farming systems

Integrated farming systems are defined as “farming systems that produce high quality food and other products by using natural resources and regulating mechanisms to replace polluting inputs and to secure sustainable farming” (IOBC 1993). Chemical and other inputs are minimised, and farming managed on an integrated structure that minimises environmental impacts. Several research projects studying integrated farming systems have been conducted in recent years (Holland et al. 1994a). The results of these projects in relation to food availability are outlined below.

The Boxworth project: Three pesticide regimes were applied to three separate blocks of fields. These were high input, “supervised” and “integrated” (see Greig-Smith & Hardy (1992) for details). Overall, densities of arthropods were 50% lower in high input fields than under the other two regimes. Effects were variable, some groups such as Lepidoptera, thrips, and some dipterans were substantially reduced, but overall effects on beetles, aphids and plant bugs and cereal leaf miners were smaller (Vickerman 1992). Ground beetles, springtails and spiders showed a very slow recovery after pesticide input was reduced, indicating that pesticide effects can last for years (Holland et al. 1994a). Variations in life-cycle strategies among Coloepteran species affected their susceptibility to pesticides (Vickerman 1992).

The SCARAB project: SCARAB was designed to follow Boxworth, over a wider range of conditions, and was concerned primarily with environmental effects (Holland et al., 1994a). Two pesticide regimes, current practice and reduced input were applied. Broad-spectrum insecticides caused reductions in some invertebrate groups. The severity of pesticide effects varied between years, species, crops and locations. No significant differences between weed cover in conventional and reduced-input fields were found (Green et al. 1995). Holland et al. (1996) found that there were no significant differences in carabid activity between conventional and integrated systems. Differences in activity were due to site and field characteristics, rather than treatments (insecticide input, crop type and rotation). However, a case study of integrated v. conventional fields found the greatest carabid diversity and densities on conventional fields (Gardner & Brown 1998).

The TALISMAN project: This was more concerned with the economic impacts of low input regimes (agrochemicals and nitrogen). Pesticides affected invertebrate numbers, but also, in the absence of pesticides, invertebrate catches showed considerable variation. This made it difficult to evaluate the importance of the pesticide effects (Hancock et al. 1995). In general, yearly variation and crop rotation affected invertebrate populations more than pesticide applications (Young & Ogilvy 2001). Differences in weed seeds were found between oilseed rape and linseed plots: higher densities of Brassica seeds were found in oilseed rape plots, and higher densities of Poa seeds in linseed plots.

The RISC project: RISC was similar to TALISMAN in design, but took into account the different farming practices utilised in Northern Ireland. Carabidae catches differed between treatments. Individual species differed in their responses to treatments, with some having higher densities, and some lower under the same treatment (Holland et al. 1994a).

The LIFE project: This project compared conventional and fully integrated farming systems. Catches of some beneficial invertebrates (Carabidae, Staphylinidae, Linyphiidae, Diplopoda) increased markedly under the lower input regimes. Rotation, chemical inputs, field and crop factors did not significantly affect overall invertebrate numbers (Holland et al. 1994a).

LINK (Integrated Farming Systems). Holland et al. (1996) studied the effects of integrated farming systems, as opposed to conventional systems, on Carabidae. The overall results were that non-target arthropods and earthworms had higher populations in lower-input and integrated regimes, compared to conventional regimes (Holland et al. 1994a).

Summary: integrated farming systems, with reduced chemical and nitrogen inputs, generally showed increased non-target invertebrate populations compared to conventional farms. These increases could also be attributed to other factors such as greater weed species diversity and abundance, improved field management and a change to non-inversion tillage practices. Site and field characteristics, as well as yearly variation and crop rotations, were also important modifiers of invertebrate abundance in some studies.

1.4.7.2 Systems using genetically-modified crops

Genetically-modified (GM) crops may influence food availability, by reducing insect and seed-food abundance within fields (Fuller, 2000). Watkinson et al. (2000) simulated the effects of the introduction of genetically modified herbicide-tolerant (GMHT) crops on weed populations and the consequences for seed-eating birds, using fat-hen Chenopodium album as the model weed. They predicted that weed populations might be reduced to low levels or practically eradicated, depending on the exact form of management. Buckelew et al. (2000) have shown that herbicide-resistant soybean crops tend to have lower insect population densities. The effect is mediated through the impact of weed management, rather than the direct effects of herbicides.

1.4.7.3 Organic systems

Organic farmland comprises only 3% of UK agricultural land. Nonetheless, it is believed to support a greater diversity and abundance of species [plants, invertebrates, birds] than conventional farmland (Azeez 2000). A comparison of organic vs. an integrated arable system in Germany indicated that the abundance and diversity of weed flora increased on the organic system (Gruber, Handel & Broschewitz 2000). No-plough tillage increased weed abundance, notably grass species. Food abundance is often higher than on conventional farms (Fuller, 1998). For example, Hald & Reddersen (1990) found that 1.4 - 1.8 times the number of bird-food arthropods were present on organic farms compared to conventional farms. Brookes et al. (1995) found total numbers of invertebrates to be similar on organic and conventional cereals, but different groups differed in abundance on each type. A study of the carabid beetle fauna in fields undergoing conversion to organic production in Europe, demonstrated that increased activity-density could occur (Andersen & Eltun 2000). The increase in the number of carabids could in part be explained by the increase in the number of weed species present. Staphylinid beetles tended to show the opposite effect, which may be a response to competition from Carabidae.

Azeez (2000) summarised the results of research on organic farms, and that there were three times as many non-pest butterflies in organic crop areas, and one to two times as many species of spiders. Feber et al. (1997) found that vegetation differed significantly between organic and conventional sites, and this had a significant impact on spider communities. Abundance and diversity of spiders increased with increasing understorey vegetation. Higher densities of earthworms have also been found on organic than conventional cereal fields (Brookes et al. 1995).Numbers of aphids on organic are also thought to be lower than those on conventional farms (Azeez 2000).

Azeez (2000) has suggested that there are five times as many wild plants in arable fields, and 57% more species. Brookes et al. (1995) found weed cover to be higher in organic than in conventional cereals. Also, broad-leaved weeds dominated in organic cereals, and grasses dominated in conventional cereals. However, they found no difference in overall seed abundance. Intercropping is a common practice on organic farms, and results in a greater diversity of polyphagous predators compared to conventional wheat monocultures (Altieri & Letourneau 1982). Feber et al. (1997) considered that the pattern of cropping, rather than crop management, was the important factor in maintaining butterfly populations on organic farms. Oilseed rape is rarely found on organic farms, whilst grass clover leys, which are more attractive to butterflies, are more frequent on organic farms. Organic fields are considered to be more beneficial to birds because of their greater seed and invertebrate food availability (Brookes et al. 1995; Fuller, 1998; Reddersen 1997). Fuller (1998) listed the mechanisms by which food availability is enhanced on organic farms compared to conventional farms: reduced pesticides, use of animal and green manures, and a higher diversity of crops including rotational grass.

Summary: organic farms often have greater food availability than conventional or integrated farms. This may be due to the prevalence of mixed farming, crop rotation, spring sown crops, the avoidance of agrochemical usage, the maintenance of non-crop habitats such as hedges and field margins, green manuring, and cereal crop undersowing.

1.4.8. WEATHER EFFECTS

Weather may affect the availability of arthropod prey. Temperature has a direct effect on invertebrate activity, and can account for much of the variation in sampling (Tones et al. 2000). Temperature affects insect abundance, with increases beginning in April (Bryant 1975). Weather was shown to affect arthropod abundance in Scotland by Benton et al. (2002). Springtail numbers collapsed after two dry summers in Sussex, before recovering in subsequent years to pre-drought levels (Aebischer 1991). Grasshoppers are more active on warm days, and hence more available as prey (Brickle & Harper 1999). Some arthropod groups (Heteroptera, Pscoptera and Lepidoptera) were perhaps more susceptible to winter climatic factors (Benton et al. 2002).

High winds can influence penetration of pesticides, by opening up the crop canopy (Tones et al. 2000). Pesticide penetration can differ greatly between crops because of different canopy structure. In broad-leaved crops, pesticide penetration is more variable compared to thin-leaved crops such as cereals. Aebischer (1991) found that trends in invertebrate numbers were so similar on farms with different management methods and intensities, that either climatic factors were of over-riding importance, or that agricultural changes have occurred on such a wide scale as to affect even the least intensive farms.

Rainfall also affects invertebrate activity. In rain, corn buntings switched from provisioning invertebrates to provisioning cereal grain, presumably because insect activity, and thereby availability, was reduced (Aebischer, Green & Evans 2000). Fewer arthropods and more grain seeds are fed to cirl bunting chicks on wet days, as invertebrates become harder to find (Sitters 1991). Spring weather (specifically drought conditions in May) may affect the availability of arthropod prey, reducing arthropod abundance (Frampton, van den Brink & Gould 2000). Rainfall can also be a limiting factor in sampling. Micro-habitat requirements may also affect susceptibility to exposure to pesticides. Increasing dryness in recent years has reduced food availability for insectivorous birds in upland areas, as wet moorland areas with concentrations of insects have dried out (Fuller et al. 2002).

McCracken, Foster & Kelly (1995) studied leatherjacket (Tipula spp.) populations in Scotland. Numbers in pastures were related to aspect, silage use, tendency to waterlogging in the pasture, organic fertiliser use, sward height, numbers in previous years and prevailing wind direction. Standard soil characteristics did not have any effect. Studies such as this indicate the complex interactions of factors that determine abundance in arthropod populations.

In addition to immediate effects on arthropod behaviour, weather may have long-term effects on arthropod populations in arable crops. For example, annual sawfly density on some West Sussex farms was related, with a one-year lag, to summer rainfall and temperature, in addition to the proportion of cereal fields that were undersown (Aebischer 1990).

Summary: weather factors, such as rainfall, temperature and wind, can affect invertebrate food availability, and can mask the effects of other factors, such as pesticides. Weather variables also interact with other factors such as pesticide effects to influence arthropod populations.

1.4.9. Effects of food accessibility and predation risk

Resource-independent factors such as food accessibility and predation risk can also affect food availability. Many species of ground-feeding birds on agricultural land have been shown to forage preferentially in sites that are relatively sparsely vegetated, irrespective of food abundance (skylark: Buckingham 2001; granivorous passerines: Moorcroft et al. 2002; starling: Whitehead, Wright & Cotton 1995; waders: Milsom et al. 1998). Morris, Bradbury & Wilson (2002b) suggest that both abundance and accessibility may determine patch selection by foraging birds. They investigated invertebrate communities and crop structure at locations in cereal fields used by yellowhammers collecting food for nestlings. When these were compared to random locations, they found that foraging locations had sparser, shorter vegetation and more invertebrates. Odderskaer et al. (1997) found that skylarks preferred tramlines and unsown plots within spring barley fields. They suggest that this was probably due to unhindered ground locomotion, facilitating detection of prey items, rather than prey density. Absolute density of arthropod food items was higher in the crop than in the more sparsely vegetated areas.

Vegetation may physically impede the movement of foraging birds (Brodmann, Reyer & Baer 1997) and may result in food items becoming inaccessible or more difficult to detect in dense swards. It has also been suggested that dense vegetation growth in modern cereal crops might impede movement and/or detectability of food items (Odderskaer et al. 1997). Similarly, crypticity may affect food availability; if a food is not easily located in a habitat, despite being abundant, intake rate may be low and the food effectively unavailable (Nystrand & Granström 1997).

Ausden, Sutherland & James (2001) found that the accessibility of soil macroinvertebrates to foraging waders was likely to increase following flooding of lowland wet grasslands, yet flooding also tended to reduce macro-invertebrate biomass. This probably explains why many of the highest densities of breeding wading birds are found on sites with low densities of soil macro-invertebrates.

Diaz & Telleria (1994) investigated the distribution and abundance of seed-eating over-wintering birds in relation to seed abundance in central Spanish croplands. They found no relationship between bird distribution and seed abundance in habitats, even when both were converted to a common energy currency (kJ 10ha-1 seed abundance in January and whole winter food requirement of the bird assemblage). Seed abundances in all the habitats they studied were found to be a magnitude greater than whole winter food requirements of the birds. They cite studies in which summer seed crops and winter seed-eating bird abundances were found to be related, and also apparently contradicting studies in which they were not. They suggest that this might be explained by considering resource-independent factors such as food accessibility and predation risk, both of which are mediated by vegetation structure. They suggest that food is less accessible and predat