SEARCH

SEARCH BY CITATION

Keywords:

  • agriculture;
  • agri-environment policy;
  • biodiversity;
  • conservation;
  • Skylark Alauda arvensis;
  • Yellowhammer Emberiza citrinella

Abstract

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Between 1988 and 2007, set-aside, a European Commission production control measure, took an average of 10% of arable farmland in the EU out of production each year. In 2007, the set-aside rate was set to 0% and the scheme was later abandoned altogether. By assessing associations of farmland birds with set-aside and quantifying the extent of set-aside loss, this study aims to assess the implications of set-aside loss for farmland bird conservation. During the lifespan of set-aside, a large number of studies assessed the biodiversity value of set-aside and other agricultural crops and habitats. Where possible we considered measurable benefits of set-aside. However, some studies did not specify the type of set-aside and in some cases set-aside fields were grouped with cereal stubble fields. In these cases, we took the pragmatic approach of assessing the value of generic stubble fields as a conservative minimum estimate of the value of set-aside fields. A re-analysis of data from 30 intensive studies demonstrates that farmland bird densities tended to be higher on set-aside than on either cereal or oilseed rape crops. Without mitigation, these are the two crops likely to replace most set-aside fields. We estimate that 26–52% of the farmland populations of key granivorous passerines were present on stubble fields, giving an indication of the proportion of birds likely to be present on set-aside fields within this broader category. An extensive survey of lowland farmland during winters 1999/2000, 2000/2001 and 2002/2003, repeated in February 2008, showed a doubling of the number of 1-km squares with no stubble and a halving of the number of squares with more than 10 ha of stubble. After set-aside abandonment, 72% of squares had no stubble in the important late winter period, confirming that many of the former stubble fields were retained as set-aside. A simple correlative model suggests that this could cause a small increase in the rate of decline of Skylark Alauda arvensis and Yellowhammer Emberiza citrinella populations, assuming causal links between stubble area and demography. However, even if this assumption cannot be supported, these results clearly indicate that a significant proportion of some farmland bird populations will need to find alternative breeding and foraging habitats.

Biodiversity loss has occurred across all terrestrial ecosystems but many of its drivers are associated with agricultural change (Green et al. 2005, Butler et al. 2007). Globally, agricultural land covers approximately 38% of the planet’s surface (Donald & Evans 2006) and, within the EU, farmers manage almost half the land area (http://www.eea.europa.eu/themes/agriculture). Over the last few decades, population declines and range contractions have been recorded for a broad suite of species associated with farmland throughout Europe and North America (Donald et al. 2001, Robinson & Sutherland 2002, Murphy 2003, Brennan & Kuvlesky 2005). Much attention has focused on farmland birds, whose populations have been the subject of a long history of monitoring in countries such as the UK, which has tracked agricultural change as a result of the impact these changes have had on food and nest-site resources (Chamberlain et al. 2000, Wretenberg et al. 2007). For many declining species, probable drivers of change have been determined (Newton 2004), and options for land management designed to stem and, ultimately, to reverse the declines have been developed and made available to farmers and landowners through agri-environment measures.

In contrast to these agri-environment measures, which have been specifically designed with biodiversity goals in mind, set-aside was introduced as an entirely agro-economic measure to reduce overproduction. Set-aside was introduced in 1998, initially as a voluntary measure, and then from 1992 became mandatory for the majority of arable farmers claiming EU subsidies. On average, growers were required to take around 10% (although the proportion varied among years) of their arable land out of production and leave it fallow for one (rotational set-aside) or more (non-rotational set-aside) cropping years. Farmers could allow the vegetation (‘green cover’) to establish by natural regeneration, by directly sowing a cover (such as rye grass, wild bird seed mixes), or by growing crops for non-food uses (‘industrial set-aside’). Set-aside fields not sown with industrial crops received considerably lower fertilizer and pesticide inputs than conventional crops (Wilson et al. 1995) and, as many rotational set-aside fields were created by the retention of the stubble of the previous crop, set-aside fields had the potential to be richer in weeds and their seeds than non-set-aside stubble fields.

Although it did not provide ideal management for biodiversity, the potential of this vast area of low-input fallow land was quickly recognized (Andrews 1992) and several studies soon showed set-aside to be valuable bird habitat in summer (Wilson et al. 1995, Chamberlain et al. 1999, Henderson et al. 2000) and in winter (Buckingham et al. 1999). Some species (e.g. Skylark Alauda arvensis: Eraud & Boutin 2000) achieved higher productivity on set-aside. This, combined with its high usage by birds in winter and evidence connecting declines with low overwinter survival of granivorous passerines (Siriwardena et al. 2000), led to a common perception that set-aside could stem or reverse farmland bird declines. However, the introduction and increase in set-aside was gradual and not associated with a clear recovery in farmland bird populations. The potential to detect such changes is complicated by a concurrent change in the national bird monitoring scheme method (Freeman et al. 2007) and spatially referenced data on set-aside area to relate to bird data are unavailable. There were several detectable upturns in farmland bird population trends around 1992–93 (fig. 8 in Siriwardena et al. 1998) but the evidence from more recent population trends (Risely et al. 2008) is that set-aside has not subsequently been sufficient to bring about population recoveries. What is less clear is whether set-aside slowed declines that would otherwise have been more severe than those observed.

Against a background of high commodity prices and falling production, in September 2007 the European Commission (EC) announced a 0% set-aside rate for the 2008 harvest season, and in the following year the Common Agricultural Policy Health Check brought about the abandonment of compulsory set-aside. In 2007, set-aside land accounted for 3.8 million hectares of farmland in Europe (Newell-Price et al. 2008) and 440 000 ha in the UK alone (http://statistics.defra.gov.uk). There was widespread concern that loss of uncropped set-aside would have a negative impact on a range of farmland taxa (Everett 2007), particularly birds. This loss of habitat was likely to be further compounded by higher commodity prices, which encouraged landowners to bring other uncropped land into production. In England the potential for biodiversity-friendly management through Environmental Stewardship (especially Entry Level Scheme, ELS) to offset the loss of set-aside has, so far, been limited by the low uptake of in-field options (Davey et al. 2010). Similar problems are likely to affect other ‘broad and shallow’ schemes such as the Rural Development Programme in Scotland.

This study aimed to assess the potential impact on farmland birds of this policy decision, specifically (1) evidence for associations of birds with set-aside as an indicator of possible benefits, (2) the magnitude of loss of uncropped land associated with the repeal of the set-aside regulation and (3) the extent to which these and other published results can be used predict the response of farmland bird populations to set-aside loss.

Methods

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Comparison of bird densities on set-aside relative to other crops, 1992–2007

To assess the degree of association of farmland birds with set-aside we collated existing data to compare the densities found on set-aside with those from various crops. Although set-aside was a Europe-wide policy, and had parallels in North America (Herkert 2009), we restricted our analysis to 30 British field studies carried out by the British Trust for Ornithology, the Centre for Ecology and Hydrology, Central Science Laboratory, the Game and Wildlife Conservation Trust, and the RSPB between 1992 and 2007 (Vickery et al. 2008a). Studies reporting absolute densities of individual species using set-aside (or other fallows) relative to oilseed rape and winter cereal were used. In the absence of mitigation measures, oilseed rape and winter cereals (hereafter ‘replacement crops’) were judged to be the two major crops most likely to replace set-aside. In summer, it was also possible to compare densities on set-aside with fields explicitly identified as stubbles. Unfortunately, due to differences in the range of studies undertaken and the habitats they covered, it was not possible to compare densities on set-aside and stubbles in winter. Additional, regionalized data within some of the 30 studies increased the number of samples to 37 (26 in summer and 25 in winter, Supporting Information Table S1). The studies spanned arable and mixed-arable farm systems in England and southern Scotland, from Cornwall to the Lothians (Table S1).

The synthesis concentrated on species in the Farmland Bird Index (FBI; Gregory et al. 2004) and particularly those with populations in long-term decline (Grey Partridge Perdix perdix, Kestrel Falco tinnunculus, Northern Lapwing Vanellus vanellus, Skylark, Yellow Wagtail Motacilla flava, Starling Sturnus vulgaris, Tree Sparrow Passer montanus, Linnet Carduelis cannabina, Yellowhammer Emberiza citrinella, Reed Bunting Emberiza schoeniclus and Corn Bunting Emberiza calandra). As far as possible, data were extracted from each study on: (1) region, (2) season (winter or summer), (3) a measure of density (e.g. birds per ha or pairs per ha) and (4) crop type: set-aside, rotational set-aside, non-rotational set-aside, wild bird cover crops, cereal stubble, winter wheat or oilseed rape. Inconsistencies among studies in the definition of habitats were treated in the following way:

  • Set-aside and stubble. Some studies explicitly defined ‘set-aside’ as either rotational or permanent/non-rotational set-aside, others had data for both categories and a third group simply used the generic term ‘set-aside’ without any further detail. Ideally, data from rotational and non-rotational set-aside would have been analysed separately, but in practice the data were too limited. Therefore, to maximize the data available for analysis, we used the generic term ‘set-aside/fallows’ to include all data provided for ‘average’ set-aside, taking mean densities across rotational and non-rotational set-aside where both appeared in a study. Industrial set-aside was excluded because it is an intensively managed crop and wild bird crops/game cover crops were treated separately. Using these criteria, almost all studies contributed relevant data. In winter, stubbles were only treated as set-aside if they had been explicitly defined as such. Thus the generic term ‘cereal stubble’ used in the present analysis may have included unidentified set-aside, but the winter ‘set-aside stubble’ category was distinct.

  • Crop types. Studies identified cereals as ‘winter wheat’ or ‘winter cereals’. To maximize the use of data from across studies, the generic term ‘winter cereal’ was used for the analysis, comprising mainly wheat but potentially other cereals, too. The other main arable crop type, oilseed rape, was always clearly identified.

For any given combination of bird species and season (summer or winter), the number of contributing studies was a sub-set of the 30 studies originally submitting information. Direct comparison of absolute mean densities across studies was not possible due to differing field methods among studies. Initially, three methods of calculating ratios were tested to measure the difference in the density of birds recorded on set-aside compared with crops (Vickery et al. 2008a). None was perfect because no single method made full use of the datasets without penalizing certain studies or observations due to the treatment of zeroes and large counts (overdispersion). Here we present results calculated using a model-based approach to compare within-study ratios of densities. Data were modelled using the formula:

  • image

where p = bird density on set-aside and q = bird density on alternative crop. This produces an index (bounded by zero and 2) that when back transformed [x/(2−x)] gives relative estimates of densities (and confidence intervals) between set-aside and the comparison crop. We opted for this method because it fully uses zero-counts and, as it ‘reins in’ outlying data (e.g. larger counts), is more conservative than alternatives. It should be noted that this method potentially risks undervaluing rarer, high-quality habitats supporting larger aggregations of birds.

Proportion of winter bird populations associated with stubbles, 1999–2003

Ideally, we would like to know what proportion of all farmland bird populations relied on set-aside fields and, consequently, could be affected by their loss. The analyses above are informative but are based on selected study sites and do not cover all farmland bird species nor all their habitats. For the next analysis, we used an independent dataset from an extensive volunteer survey, the BTO/Joint Nature Conservation Committee Winter Farmland Bird Survey (WFBS), to ensure representative coverage of lowland farmland. WFBS involved systematic bird and habitat surveys in three winters (1999/2000, 2000/2001 and 2002/2003) in 1078 1-km squares of a stratified random sample from lowland farmland throughout Britain (full details in Gillings et al. 2008). It was inappropriate to expect all volunteers to discuss management with farmers. Therefore, these data cannot determine which winter stubble fields were actually set-aside fields and which were retained for other reasons. We analysed these data to estimate the proportion of all farmland bird species that were associated with stubble fields as this was a minimum estimate of the proportion of birds using set-aside. There is reason to believe that the proportion of birds using set-aside was probably higher. Set-aside and non-set-aside stubbles tended to differ in their green cover: whereas most non-set-aside stubbles received post harvest herbicide applications, set-aside stubbles did not. On average, set-aside fields were therefore more likely to develop a weed and crop volunteer cover than true crop stubbles. As the measure of stubble use employed here was the average of the number of birds on all stubbles regardless of quality or management origin, this measure of bird association with stubble fields was a conservative indication of the value of rotational set-aside to birds in winter.

The procedure was to estimate bird densities in key habitat types and to multiply these densities by estimates of the total area of these habitats in lowland Britain (based on arable and pastoral landscape types: Bunce et al. 1996) to derive the proportion of birds associated with the following eight habitats: cereal stubble, other stubble, cereal crop, oilseed rape crop, other crop, bare till, grass and other agricultural habitat. We considered Skylarks and then the following species were grouped because some were too scarce to analyse as individual groups: (1) sparrows and finches (House Sparrow Passer domesticus, Tree Sparrow, Chaffinch Fringilla coelebs, Brambling Fringilla montifringilla, Greenfinch Cardeulis chloris, Goldfinch Cardeulis cardeulis, Linnet, Twite Cardeulis flavirostris and redpoll sp. (Lesser Redpoll Cardeulis cabaret and Common Redpoll Cardeulis flammea)), and (2) buntings (Yellowhammer, Reed Bunting and Corn Bunting).

For each visit, counts were summed within each species group for each habitat type and analysed with a repeated measures generalized linear model because bird counts were non-normal and squares were visited multiple times. Bird count was modelled as a function of habitat type using a log link function. The inclusion of log field area as an offset effectively makes this an analysis of bird densities. Initial models using a Poisson error structure were found to be over-dispersed and a binomial error structure gave better model fit. This gave habitat-specific estimates of the mean bird density with likelihood-ratio tests of differences among habitat types. Additional tests were performed of the difference in density between set-aside and winter cereal and oilseed rape because their predominance in the landscape and high agricultural value led us to believe these were the habitats most likely to replace set-aside. As densities and field areas varied regionally, models were run separately for five broad regions: East England, North England, West England, Scotland and Wales (Gillings et al. 2008).

Regional estimates of habitat area were calculated as follows. WFBS field visits were time-limited and on average surveyed 72% of the farmland present in each 1-km square (Gillings et al. 2008). Therefore, the actual area of farmland in each 1-km square was estimated by multiplying the total area of arable and grassland cover types from the Centre for Ecology and Hydrology’s Land Cover Map 2000 (Fuller et al. 2002) for each 1-km square by the proportional availability of habitats on the field visit to that square to give a corrected estimate of the amount of each habitat in each surveyed square (Gillings et al. 2008). For simplicity, areas were averaged across multiple visits to the same square to give the area of each habitat in mid-winter. These areas were averaged across all surveyed squares in each region, and then multiplied by the total number of 1-km squares in each region to give estimates of the total area of each habitat type in each region.

The product of habitat-specific bird density and habitat area estimates gave regional habitat-specific population estimates. These were summed across regions and as the field methods were not sufficiently intensive to provide absolute counts for certain species (Atkinson et al. 2006), all habitat-specific population estimates were divided by the total estimate to give the percentage of birds associated with each habitat type.

Estimating the change in stubble area post set-aside loss

WFBS data provide a baseline against which to quantify recent change in the availability of cereal stubble. During February 2008, a sample of the WFBS 1-km squares was revisited to map habitats in the same fields as those originally surveyed. To ensure representation of different farming systems, the squares were selected after stratifying lowland farmland into pastoral, mixed and arable landscapes as follows. From the Land Cover Map 2000 (Fuller et al. 2002) we derived the proportion of each 10-km square’s farm landcover types attributable to arable cover types (pArable) and classified all 1-km squares in the 10-km squares as pastoral (pArable < 0.33), mixed (0.34 < pArable < 0.66) or arable (pArable > 0.67). The sample was also stratified by the availability of cereal stubble as recorded by the original WFBS using the categories ‘nil’ (0 ha), ‘low’ (< 10 ha) and ‘high’ (> 10 ha) stubble availability (after Gillings et al. 2005). The aim was to re-survey, during February 2008, 15 squares in each region and former stubble availability class, giving a total sample of 135 squares. For analysis, the re-surveys were paired with an estimate of habitat availability from each WFBS square. As far as possible we used only WFBS data from February visits, but in a small number of squares no February visit was available, so we had to use data from January visits. Potentially this would give WFBS estimates that were too high, but the influence of this should be small because in most squares (78%) the estimate was based on more than one winter, in 75% of cases the underlying visits were made in February, and in only three squares was the estimate based on a single January visit. In each field we recorded habitat type and marked changes to field boundaries since the WFBS period onto maps to allow estimation of current field areas. All areas from WFBS visits and 2008 re-surveys were corrected for area surveyed to give estimates of habitat area per 1-km square. Differences in stubble area between periods were tested using Kruskal–Wallis and Sign tests.

Predicting population trends of Skylark and Yellowhammer post set-aside loss

Population trends in relation to stubble availability reported in Gillings et al. (2005) offer the potential to estimate how the breeding population trends of Skylark and Yellowhammer might change if stubble availability drives population trends. In reality, birds may redistribute to use alternative, less preferred, habitats, so these estimated trends should be considered worst case scenarios.

Both Skylark and Yellowhammer are widely distributed throughout lowland farmland (Gibbons et al. 1993) and we therefore assumed that estimates of the percentage of squares falling into the nil, low and high stubble categories are equivalent to the percentage of the range of these species experiencing different rates of population change. These values were used to re-weight previously published trend estimates for Skylark and Yellowhammer in nil, low and high stubble squares for 1994–2003 (Gillings et al. 2005) to derive an overall population trend under the new level of stubble availability.

All statistical tests were performed in sas (version 9.2; SAS Institute Inc., Cary, NC, USA).

Results

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Comparison of bird densities on set-aside relative to other crops, 1992–2007

The ratio of densities in set-aside vs. nearby crops (and stubbles in summer only) could be estimated for 17 species in summer and 16 in winter (Tables 1 and 2), including 11 and 10 species, respectively, that are currently in long-term decline. Irrespective of season, most species reached higher densities on set-aside than on either winter cereal or oilseed rape crops. The exceptions were Yellow Wagtail and Stock Dove Columba oenas, which showed higher summer densities on winter cereals, Woodpigeon Columba palumbus, Linnet and Reed Bunting, which showed higher summer densities on oilseed rape, and Northern Lapwing and Starling, which showed higher winter densities on winter cereals. On average, across all species, densities were at least seven times higher on set-aside than on cereals or oilseed rape in summer and at least 22 times higher in winter. Wild bird cover crops, some planted as set-aside wild bird mix, supported higher densities of birds than set-aside in winter (Table 1). Comparisons between set-aside and stubbles in summer were only possible for a small number of species and although densities were still generally higher on set-aside, the results were more equivocal (Table 2).

Table 1.   Model-based ratios of densities of Farmland Bird Index (FBI) species on set-aside relative to winter cereals, oilseed rape or wild bird crops in winter (October to March). Declining FBI = species in long-term decline (italics). n = the number of contributing studies to each species/crop comparison; missing values = insufficient data available.
SpeciesSet-aside relative to winter cereals± 1.96 × senSet-aside relative to winter oilseed rape± 1.96 × senSet-aside relative to wild bird crops± 1.96 × sen
Common Kestrel3.500.2231.471.343
Grey Partridge7.410.2068.600.1150.931.435
Northern Lapwing0.530.2455.000.485
Common Woodpigeon2.380.1851.110.435
Stock Dove2.820.4458.550.215
Skylark2.950.1792.510.1874.764.006
Western Jackdaw1.790.4654.220.605
Rook3.290.1654.000.645
Common Starling0.830.2052.670.285
Eurasian Tree Sparrow3.000.3945.000.4830.0412.502
Common Linnet30.330.04511.380.1751.564.005
European Goldfinch2.910.3754.000.805
European Greenfinch8.450.1352.750.645
Yellowhammer49.690.03581.420.0250.472.227
Common Reed Bunting230.210.015335.500.0250.396.256
Corn Bunting2.330.3252.330.3150.272.336
Declining FBI36.620.22 49.660.37 0.203.80 
All FBI22.020.23 30.030.42    
Table 2.   Model-based ratios of densities of Farmland Bird Index (FBI) species on set-aside relative to winter cereals, oilseed rape or cereal stubbles in summer (April to August). No data were available for Common Whitethroat or European Turtle Dove. Declining FBI = species in long-term decline (italics). n = the number of contributing studies to each species/crop comparison; missing values = insufficient data available.
SpeciesSet-aside relative to winter cereals± 1.96 × senSet-aside relative to oilseed rape± 1.96 × senSet-aside relative to cereal stubbles± 1.96 × sen
Common Kestrel2.640.4732.2610.701
Grey Partridge2.560.2071.780.6332.861.023
Northern Lapwing4.130.22847.370.0441.282.302
Common Woodpigeon4.220.1960.760.6130.810.793
Stock Dove0.980.2866.600.2830.770.913
Skylark4.100.07183.330.1454.320.234
Yellow Wagtail0.440.2255.370.2731.0519.012
Western Jackdaw7.920.12547.130.0432.281.483
Rook17.350.066151.310.0931.130.863
Common Starling10.920.1452.231.0331.451.003
Eurasian Tree Sparrow2.180.4944.190.4932.240.633
Common Linnet5.800.2150.950.8230.790.623
European Goldfinch5.550.1764.130.0341.890.233
European Greenfinch43.680.0552.510.6541.170.963
Yellowhammer1.740.19101.570.3651.880.083
Common Reed Bunting1.190.5150.761.2631.503.602
Corn Bunting2.730.5157.060.3221.0010.002
Declining FBI3.490.29 7.270.53 1.703.85 
All FBI6.940.24 17.010.44 1.592.71 

Proportion of winter bird populations associated with stubbles, 1999–2003

With the exception of buntings in Wales, there were significant differences in bird density among habitats for all species groups in all regions (Supporting Information Table S2). At the regional scale, habitat availability differed among regions following the east–west gradient in grass–arable farming. In Wales, arable habitat accounted for only 13% of field area. There was greater cover of stubble in Scotland (21%) than in East England (14%). The single most abundant agricultural habitat was pasture, accounting for 47% of lowland farmland, followed by cereal crops at 21% (Fig. 1). Cereal and other stubbles together accounted for 13% of lowland farmland, yet they held 52% of Skylarks, 26% of finches and sparrows, and 38% of buntings (Fig. 1). Exactly what proportion of these stubble fields were actually set-aside fields is unknown. However, given the management differences between crop stubbles and set-aside (see Methods) and the likely distribution of birds in true stubbles vs. set-aside fields, these figures can be taken as minimum estimates of association with set-aside. The only habitat that held a greater proportion of any bird group than stubbles was pasture, which held 32% of finches and sparrows. Cereal crops and oilseed rape crops each supported around 10% or less of populations (Fig. 1).

image

Figure 1.  Estimated percentage of the lowland populations of Skylark, finches and sparrows, and buntings found on different agricultural habitat types in winter. For comparison, the proportional area of each habitat type is shown (open bars). All crops are winter-sown.

Download figure to PowerPoint

Estimated change in stubble area post set-aside loss

Repeat WFBS surveys were completed for 129 1-km squares (six were no longer accessible). Each square was classified as having nil, low or high stubble availability in the later winter of each survey period (WFBS vs. 2008). In the early 2000s, 41% of the squares contained zero stubble; this increased to 72% of squares in 2008. In contrast, the occurrence of high stubble squares more than halved, from 24% of squares to 11% (Table 3). Overall, there was a significantly greater number of squares with a decrease in stubble area than an increase (Sign Test, M = −22, < 0.0001). In arable, mixed and pastoral landscapes, the area of stubble differed significantly between periods (Kruskal–Wallis tests: 7.2, = 0.0072; 8.2, = 0.0043; 7.5, = 0.0061, respectively). The median area of stubble per 1-km square changed from c. 2 ha in 2000 to zero in 2008 (Fig. 2).

Table 3.   Numbers of squares falling into different categories of stubble availability where nil = 0 ha, low = < 10 ha and high = > 10 ha of stubble per 1-km square.
Stubble availability2008 Re-survey
NilLowHighTotal
WFBS (1999/2000, 2000/2001, 2002/2003)
 Nil437353 (41%)
 Low329445 (35%)
 High186731 (24%)
 Total93 (72%)22 (17%)14 (11%)129
image

Figure 2.  The estimated late-winter area of stubble in a sample of 1-km squares in arable, mixed and pastoral landscapes of Britain in 1999–2003 (grey bars) and 2008 (open bars). Boxes show inter-quartile range, solid lines are medians, dashed lines are means, error bars show the 90th percentile and dots show individual outliers.

Download figure to PowerPoint

Predicted population trends of Skylark and Yellowhammer post set-aside loss

The WFBS data suggested that, in the early 2000s, 48% of lowland farmland had no stubble in late winter (at the 1-km square scale), 31% had low stubble availability and 21% had high stubble availability. The analysis of Gillings et al. (2005) estimated the breeding population decline (1994–2003) of Skylarks in each of these strata at −35%, −20% and −4%, respectively. The mean of these rates, weighted according to the percentage of the lowlands to which each could be applied in the early 2000s, gives an estimated overall trend in lowland farmland for the period 1994–2003 of −23%. This figure is consistent with the all-England trend over the same period of −19% (confidence limits −15% to −22%; Raven et al. 2004). If the 1994–2003 period had instead been characterized by the reduced availability of stubble we see post set-aside (i.e. the percentage of squares with nil stubble elevated from 48% to 72%), we estimate the overall trend of Skylarks would have been −28%. Applying the same logic to Yellowhammer trends suggests the trend could change from −23% pre set-aside to −26% post set-aside.

Discussion

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

The results presented here, spanning a wide range of arable and mixed farming systems, show that during both winter and the breeding season, set-aside held consistently higher densities of many farmland bird species than arable crops. This is particularly true in relation to winter cereals and oilseed rape, which represent the crops likely to replace set-aside in the short term in the absence of mitigation measures. The only habitat found to support higher bird densities on average than set-aside was wild bird crops (in winter), a habitat specifically designed to increase the seed resources for foraging birds, and available over very small areas compared with set-aside. Previous studies have demonstrated the value of these wild bird crops for granivorous and some omnivorous species (e.g. Dunnock Prunella modularis and Song Thrush Turdus philomelos; Henderson et al. 2004) suggesting that for some species in winter, set-aside loss may be compensated for by smaller areas of wild bird crops. However, timing and scale are critical (Siriwardena 2010). Most wild bird crops are depleted and collapsed by late winter. Those that are for game cover are typically ploughed in early February at the end of the shooting season and those in schemes do not have to be retained after 15 February or 1 March depending on the area of Britain (Siriwardena 2010). In contrast, set-aside had to be maintained until at least 15 July. Also, as most wild bird crops are planted as strips around fields and often comprise structurally dense habitat, they are unlikely to attract open habitat species, in particular Skylark (Henderson et al. 2004, R.H. Field, A.J. Morris, P.V. Grice & A. Cooke submitted).

The habitat resurveys were carried out in February to approach as closely as possible the late winter period, when food resources for birds are likely to be especially scarce (Siriwardena et al. 2008). By this time, most non-set-aside stubbles will have been ploughed, leaving a greater proportion of set-aside fields. The results showed a significant decrease in stubble availability compared with the early 2000s, with twice as many squares now lacking stubble in late winter and half the number of squares having more than 10 ha of stubble. We used this figure of 10 ha because in a correlative study of trends and habitat availability, areas with more than 10 ha of stubble were associated with stable populations of Skylarks (Gillings et al. 2008). Extrapolations of these trends using the observed low levels of stubble available post set-aside loss predict a small increase in the rate of decline of Skylark and Yellowhammer. It should be noted that 10 ha is really an arbitrary figure to indicate the scale of possible effect rather than a biological threshold. As mentioned earlier, set-aside tends to be higher in quality for birds than typical stubbles, meaning the loss of set-aside will probably result in a reduction both in habitat area and in the quality of that stubble which remains.

As our data came from two discrete periods, we cannot determine whether there was a step-change in stubble availability over 2007–2008 due to the zero set-aside rate or a gradual decline due to other factors. Defra’s June Survey of Agriculture indicated a 14% reduction in the area of non-productive set-aside and fallow between June 2006 and June 2007 (Langton 2008), followed by a further 55% decrease in the area of uncropped land that can be attributed to set-aside loss between the 2007 and 2008 June Surveys (Langton 2009). Surveys of 20, 4-km2 areas in East Anglia recorded a 66% decrease in stubble availability, measured as change in area in March from the 2005–2007 average to 2008. In that survey, data on cropping in summer made it possible to confirm that 82% of the stubble loss was attributable to the removal of set-aside fields (Vickery et al. 2008b). Together, these results provide strong evidence that the zero set-aside rate resulted in a large-scale reduction in stubble retention in 2007–2008 throughout lowland farmland in England.

The use of these results to assess the implications of set-aside loss for birds is hampered by uncertainty concerning habitat quality. From WFBS we estimated that during the early 2000s, 52% of Skylarks, 26% of finches and sparrows, and 38% of buntings were associated with stubbles. It is unfortunate that the WFBS data do not permit apportioning these birds to set-aside and non-set-aside stubbles, but our expectation based on our density results and knowledge of bird responses to stubble management (Robinson & Sutherland 2000, Moorcroft et al. 2002, Robinson 2003) leads us to expect that a high proportion of these birds will have been associated with set-aside. Whilst it is extremely unlikely that loss of set-aside will lead to immediate mortality among 52% of Skylarks, these analyses give an indication of the number of birds required to redistribute to alternative habitats. In the absence of demographic data, scientists have tended to assume that high density is a proxy for high quality. Although this is not always true (van Horne 1983), evidence from breeding (e.g. Donald 2004) and wintering (e.g. Green 1978) studies suggest that it applies at least to some species on set-aside/stubbles. Our results showing high density could therefore be interpreted as set-aside being a higher quality habitat than winter cereals and oilseed rape. If true, we must conclude that birds redistributing in the wake of set-aside loss will be forced to move to inferior habitats, whether that be in terms of food availability or predation risk. Also, it should be borne in mind that intensification has not ceased since the inception of set-aside and it is quite possible that the difference in quality between set-aside and managed habitats is greater than it was in 1988.

Ongoing research is assessing how farmland birds have redistributed in the absence of set-aside and is attempting to assess the quality of these habitats in order to identify the scale of mitigation measures that is actually needed. Mitigation of the effects of set-aside loss on other farm wildlife and resource protection, as well as on farmland birds, is currently being provided voluntarily by farmers under the industry-led Campaign for the Farmed Environment (CFE). It is important actively to monitor the extent to which the CFE is succeeding in replacing the benefits that set-aside provided to birds. This will identify the need for any changes to the management measures implemented and determine the success (or otherwise) of the voluntary approach. CFE measures do not just involve the re-establishment of set-aside management; this monitoring must consider real bird responses as well as simply recording areas of land managed. If CFE, alongside ELS, is to mitigate the loss of set-aside, there will need to be large-scale uptake by farmers of in-field management options, as these have the best chance of counteracting any negative effects of set-aside loss.

In conclusion, the answers to the original questions we posed are:

  • 1
     across a range of studies in both winter and the breeding season, higher densities of most farmland birds species were present on set-aside than in the dominant arable crops,
  • 2
     the zero set-aside rate brought about a rapid reduction in fallow land availability to such an extent that the proportion of our sample of 1-km squares that had no stubble in late winter increased from 41% in 1999–2003 to 72% in 2008,
  • 3
     we estimate that between a quarter and half of farmland bird populations might be affected by loss of set-aside, the need to redistribute to other habitats in winter and, at least in the case of Skylark and Yellowhammer, a small acceleration in the rate of decline could follow.

Whether such acceleration of declines will occur will depend on the change in overall habitat quality and, in practice, on how strongly the high densities on set-aside were correlated with demographic rates. The extent to which the CFE, or other changes in the farmed environment, succeed in replacing the benefits of set-aside will also have a critical impact in this context and represents a major priority for research.

This work was funded under two Defra contracts (BD1639 and BD1640). We would like to thank those administering the various studies that provided density figures for the comparisons made in this paper. We also thank the observers who collected the original WFBS data that made many of these analyses possible, and also Graeme Garner and Jeff Stenning, who conducted the repeat habitat surveys. The WFBS was funded under a partnership between BTO and the Joint Nature Conservation Committee. We thank the Centre for Ecology and Hydrology for permitting use of the Land Cover Map 2000 data. Comments from Gavin Siriwardena and referees improved this manuscript.

References

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Methods
  4. Results
  5. Discussion
  6. References
  7. Supporting Information

Table S1. A summary of the studies used in the comparison of bird densities on set-aside relative to other crops, 1992–2007.

Table S2. Tests of significant differences in density of species and species groups between habitat types in five regions of Britain.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
IBI_1058_sm_TableS1-2.doc69KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.