Adaptive management and targeting of agri-environment schemes does benefit biodiversity: a case study of the corn bunting Emberiza calandra


Correspondence author. E-mail:


1. Agri-environment schemes (AES) are the main European policy response to biodiversity loss caused by agricultural intensification. Maximizing their effectiveness is a key policy challenge. Monitoring is essential to inform adaptation and improvement of schemes over time, and to understand how measures may need to vary across a species’ range.

2. We measured changes in breeding abundance of a severely declining bird, the corn bunting Emberiza calandra, in response to AES in Scotland over 7 years and 71 farms. Two AES were monitored, one with general management for farmland birds, and one with targeted, adaptive management for corn buntings. We use these data to estimate the proportion of the population that AES must influence to halt the overall decline.

3. Corn buntings increased by 5·6% per annum on farms in the targeted AES, showed no significant change on farms in the general AES, and declined by 14·5% per annum on farms outside AES.

4. In arable-dominated areas, AES management that increased food availability reversed population declines. However, where a high proportion of corn buntings nested in grasslands, an additional AES option that delayed mowing was essential to achieving population increase.

5. Results suggest that approximately 72% of the corn bunting population in mainland Scotland must receive targeted AES management to halt the current decline. In 2009, only 24% was targeted in this way.

6.Synthesis and applications. AES measures are capable of reversing corn bunting declines in Scotland, and the same measures are likely to benefit a wide range of other taxa too, but require geographical targeting and flexibility to adapt and improve management options, backed by expert advice. Targeted AES provision to the required level for corn buntings will cost approximately £120 000 per annum, with 500–600 ha under appropriate management. This is 0·02% of annual subsidies paid to Scottish farmers, and 0·5% of land in the remaining mainland range of the corn bunting. These outcomes illustrate the value of AES monitoring studies to assess scheme effectiveness, identify improvements, and determine the scale of implementation required for reversing species declines.


The development of agriculture over the past 50 years, driven by advances in plant and animal breeding and chemical crop protection, and supported by state subsidy has allowed global food production to outstrip human population growth (FAOSTAT 2009). Widespread biodiversity loss has followed on land farmed intensively (Reidsma et al. 2006). The ‘greening’ of European Union (EU) agricultural policy through vast financial expenditure on agri-environment management since the late 1980s has had limited success in reversing these biodiversity losses. Kleijn & Sutherland (2003) found little evidence of the effectiveness of agri-environment schemes (AES) for biodiversity conservation and, worse, many schemes lacked robust monitoring to make such assessments possible. Since then, studies have demonstrated benefits of AES interventions (e.g. Kleijn et al. 2006; Knop et al. 2006; Carvell et al. 2007; Maes, Musters & De Snoo 2008), and maximizing biodiversity conservation from AES is now a key policy challenge (Sutherland et al. 2006). However, studies which quantify large-scale, long-term population response to AES remain scarce (e.g. Peach et al. 2001; Wilson, Vickery & Pendlebury 2007; La Haye et al. 2010), and Memmott et al. (2010) note the continuing lack of applied ecological studies which implement and test effectiveness of management in an adaptive approach. Adaptive management that combines research with action on the ground, enabling practitioners to learn from successes and failures and adapt actions accordingly, is essential if we wish for better conservation (Salafsky et al. 2002). AES monitoring studies provide opportunities for such tests, to understand how the design of measures may need to vary across species’ ranges (Whittingham 2007), to adapt measures over time, and to estimate what proportion of a given species’ population must be targeted to halt its decline (Wilson, Evans & Grice 2010).

Here we use a 7-year monitoring study of corn bunting Emberiza calandra (L.) population response to agri-environment management in Scotland to test its effectiveness. We compared two different AES, one with general and one with targeted deployment of measures. The targeted scheme was adapted during the study by incorporating a novel measure in response to initial monitoring results from the same study populations. We also use the annual growth rates of corn bunting populations on and outside AES schemes to make initial estimates of the proportion of the total population that AES management will need to target to reverse overall population decline. The corn bunting is one of the most severely declining farmland birds in the UK (Wilson, Evans & Grice 2009) and across Europe (PECBMS 2010). In Scotland, fewer than 800 territorial males remain, and evidence of the effectiveness of agri-environment management for this species is needed urgently, as discussed in a Scottish Parliament debate in September 2009 (; accessed 6 August 2010).

Materials and methods

Agri-environment schemes used as the basis for the study

In Scotland, the national AES in 2001–2006 was the Rural Stewardship Scheme (RSS). It was voluntary but competitive (not all applicants succeeded). Each application (for a 5-year funding agreement) included a whole-farm plan incorporating several of 33 management options (SEERAD 2003), some of them designed to provide resources for farmland birds (Table 1). The RSS was our ‘general’ AES. Our ‘targeted’ AES was Farmland Bird Lifeline (FBL), an intervention scheme for corn buntings, operating since 2002. In FBL, farmers in areas with known breeding populations are invited to enter annually reviewed management agreements, backed by face-to-face advisory support. Management options are similar to those in the RSS, but have been adapted to the specific needs of corn buntings as knowledge has improved. For example, because corn buntings use ‘unharvested crop’ patches (Table 1) mostly in the first winter when they are rich in cereal grain (Perkins, Maggs & Wilson 2008), these patches are sown annually with a cereal-rich mixture (since 2006) to ensure grain supply throughout winter. In the RSS, re-sowing was required only every 2 years. Secondly, nest monitoring studies in 2004–2005 in Aberdeenshire revealed that, in contrast to elsewhere in the species’ UK range (e.g. Brickle et al. 2000), up to 30% of first nesting attempts were in grass fields for silage, with high nest-loss rates caused by grass harvesting (Wilson et al. 2007). Consequently, delayed mowing (late-cut grass) was added to FBL from 2005 (Table 1), which has increased fledging success in mown grasslands (mean probability ±1 SE of brood fledging in conventionally cut silage = 0·09 ± 0·11, n = 37 nests; in late-cut silage = 0·43 ± 0·15, n = 52 nests) (Wilson et al. 2007).

Table 1.   Management implemented on 30 Rural Stewardship Scheme (RSS) and 35 Farmland Bird Lifeline (FBL) study farms
Management optionMain resourcePayment rate (£ ha−1)No. farmsArea (ha)
  1. Those most likely to provide food (winter seed or summer insects) or safe nesting habitats for corn buntings are in italics. For each option, payment rate, frequency of uptake and area managed per farm (mean ± 1 SD) is shown.

  2. FBL, Farmland Bird Lifeline; RSS, Rural Stewardship Scheme.

  3. aExtensively managed spring cereal or rape followed by over-winter stubble that could not be sprayed or ploughed before 28 February (RSS) or 31 March (FBL), or extensively managed turnips.

  4. bAvailable in FBL from 2005.

  5. cUntil 15 September.

  6. dDependent on mowing date (permissible date of 1 July was too early to allow most corn bunting nests to fledge before mowing).

  7. eLower payment rate for crops grown on set-aside.

  8. fHigher payment rate for management applied to the same field for 3 years or more.

  9. gLower payment rate for extensively managed turnips.

  10. hHigher payment rate if nitrogenous fertilizer not applied.

  11. RSince 2008, available within the Corn Buntings package of the Rural Priorities scheme.

  12. LSince 2007, available as a Land Managers Option.

  13. Na, not available in this scheme.

Unharvested cropsL,RWinter seed600160e/45024301.8 ± 0.92.9 ± 1.8
Introduction or retention of extensive cropping followed by over-winter stubblea,L,RSummer insects/winter seed120/140f120g/15001705.1 ± 3.6
Management of conservation headlandsLSummer insects70/150h7014110.8 ± 0.31.1 ± 0.6
Management of grass margin or beetle-bank in arable fieldsL,RSummer insects73653328121.9 ± 1.51.3 ± 1.4
Provision of supplementary food in winterWinter seedNa50 13 <1
Late-cut grass for corn buntingsb,RSafe nestingNa260 14 7.6 ± 4.8
Delayed spraying/topping of set-asidecSafe nestingNa0 14 6.6 ± 5.5
Extensive management of mown grassland for birds(Safe nesting)d150Na6 16.4 ± 17.1 
Creation/management of species-rich grasslandRSummer insects250/100Na13 1.6 ± 1.7 
Management of water marginR 400Na24 0.9 ± 0.6 
Management of open grazed grassland for birds 100Na8 5.0 ± 4.4 
Creation/management of wetland 250/100Na15 3.0 ± 3.7 
Management of wet grassland for waders 100Na2 3.5 ± 2.2 
Management of flood plain 25Na2 16.5 ± 0.7 
Management of scrub (including tall herb communities) 55Na2 1.7 ± 1.5 
Creation/management of hedgerowsL 5000Na16 0.2 ± 0.2 
Management of extended hedges 500Na10 0.5 ± 0.5 
Management of native or semi-natural woodlandL 100Na1 5.1 
Management of a site of archaeological or historical interestL 80Na4 9.1 ± 16.9 
Pond creation 18000Na4 0.1 ± 0.1 

Farm selection

Between 2003 and 2009, surveys were carried out on 71 arable and mixed lowland farms covering 8845 ha in eastern Scotland (the region holding >85% of Scotland’s corn buntings), and split between RSS, FBL and ‘control’ groups (Fig. 1). Because we were studying real AESs and their uptake by farmers, the distribution of farms between treatment groups changed between years (Table 2).

Figure 1.

 Distribution of study sites and corn buntings (10-km squares occupied during 2002–2009) in mainland Scotland.

Table 2.   Farms monitored in each treatment group (RSS, FBL and Control)
YearAberdeenshire, Moray & Inverness-shireFife & Angus
  1. Numbers given in parentheses indicate farms with no corn buntings in any year during 2003–2009 and excluded from analyses.

  2. FBL, Farmland Bird Lifeline; RSS, Rural Stewardship Scheme.

200323 (4)15 (2)1707 (2)4
200425 (4)15 (2)1607 (2)4
200547507 (2)3
200630 (4)129010 (2)1
20074131011 (2)0
200823 (2)1216 (2)010 (2)1
200914 (2)1112010 (2)1

Rural Stewardship Scheme farms were selected from those in Aberdeenshire and Moray that joined the scheme in autumn 2002, and were within or adjacent to a 2-km square recently occupied by corn buntings during the breeding season (Francis & Cook in press). ‘Control’ farms with similar land use to the RSS farms, and within 10 km of them, were also selected. Seven ‘control’ farms later joined RSS (two each in autumns 2003–2004 and three in autumn 2005). Management options were established during the spring and summer following entry into the scheme. In autumns 2007 and 2008, management options were removed from 7 and 2 farms respectively, following expiry of 5-year RSS agreements, and these farms were controls in subsequent years.

Farmland Bird Lifeline farms were selected in areas known to hold breeding corn buntings. Management options on 14 farms in Aberdeenshire, one in Inverness-shire, four in Fife and three in Angus were first implemented in the springs of 2002 and 2003. This was extended to a further seven farms in Aberdeenshire, two in Inverness-shire, three in Angus, and one in Fife in springs 2006–2007 (Table 2). Two farms had prior RSS agreements, and joined the FBL treatment group from 2006.

Land use (determined from digitized maps) was a combination of autumn- and spring-sown barley and oats, autumn-sown wheat and oilseed rape, potatoes, turnips, grass mown for silage or grazed by cattle and sheep, and ‘set-aside’ (Table 3). In Fife and Angus, there was less grass, and more land was used for vegetables, including carrots, cabbages and broccoli.

Table 3.   Area (ha) surveyed and land use composition (%) per farm in each treatment group and region (mean ± 1 SD), 2003 and 2008–2009
RegionFarm typeYearNo. farmsaTotal area (ha)bSCACLEGVEGOSRGRAROU
  1. P values are for Kruskal–Wallis tests for differences in proportions of each crop type within each treatment group between 2003 and 2008–2009.

  2. SC, spring cereals; AC, autumn cereals; LEG, legumes (peas, beans) or legume/barley mixture mown for arable silage; VEG, root vegetables; OSR, oilseed rape; GRA, grazed pasture, grass mown for silage or hay, or newly sown; ROU, rough grass or set-aside (rotational and non-rotational); FBL, Farmland Bird Lifeline; RSS, Rural Stewardship Scheme.

  3. aSome farms switched treatment groups between years.

  4. bTotal area includes non-cropped land and minor habitat categories not presented in table.

AberdeenshireRSS200323264130 ± 1720 ± 160 ± 14 ± 106 ± 1324 ± 188 ± 8
Moray &RSS2008–200929345537 ± 2317 ± 2002 ± 55 ± 1029 ± 223 ± 4
FBL200315198227 ± 1613 ± 151 ± 35 ± 57 ± 1133 ± 206 ± 7
FBL200911139929 ± 1813 ± 182 ± 44 ± 45 ± 1041 ± 243 ± 3
Control200317161144 ± 259 ± 1903 ± 41 ± 432 ± 259 ± 11
Control2008–200919166737 ± 2318 ± 220 ± 14 ± 96 ± 2024 ± 204 ± 11
Fife & AngusFBL20037124342 ± 1917 ± 23020 ± 101 ± 311 ± 124 ± 4
FBL200910199328 ± 2332 ± 203 ± 418 ± 82 ± 59 ± 121 ± 2
Control2003478126 ± 3022 ± 31031 ± 77 ± 78 ± 90 ± 0


All FBL farms were monitored in each year that AES options were implemented, and eight were surveyed as controls in at least 1 year prior to them joining FBL. It was not possible to monitor every RSS and control farm in every year due to other data collection commitments. However, all of these farms were surveyed in summers 2003, 2004, 2006 and 2008, with additional partial surveys in 2005 (n = 7 farms), and 2007 (n = 5 farms) during detailed studies of breeding corn buntings. In 2009, we surveyed 17 of these farms, together with eight farms not monitored since their transfer from FBL to RSS in 2004–2005. Our study therefore had strong ‘control-intervention’ design over several years, but weaker ‘before-after’ design. However, preliminary studies of corn bunting response to FBL management had previously shown that population changes were consistent across control and intervention study areas before management began (Perkins et al. 2008).

Surveys took place between May and August, on mornings with no or light rain and calm or light winds, and usually involved three visits to each farm. However, in 3 years (2004, 2008 and 2009), some farms (47%, 23% and 59%, respectively) were visited only twice. Each survey route was selected to pass within 250 m of all points on the farm. During each visit, locations and activities of all corn buntings were recorded on a 1 : 10 000 map. Locations on land surrounding the farm, but within 250 m of the farm boundary, were also recorded. The number of territorial males was counted from clusters of map records (Marchant et al. 1990). Corn buntings sing frequently and conspicuously (Olinkiewicz & Osiejuk 2003), so detection rates are high. In this study, data from those farms on which more intensive fieldwork was carried out (so that territory count is known with certainty) show that across 98 farm-years and 300 recorded territories, 94% of territorial males were detected by the survey method, so it was not necessary to account for effects of imperfect detection (Gonzalo-Turpin et al. 2008) in population trend modelling.

Data analysis

We excluded eight farms that held no territorial corn buntings in any year (Table 2). Data analysis was therefore based on 63 farms with at least one territorial male corn bunting in one of the years 2003–2009. However, because 30 farms changed treatment group (in one case, twice) during the study due to entering or leaving AES agreements, we considered data after the treatment change as being a new time series of data from a ‘new’ farm.

To assess population change in response to AES, we modelled the density of territorial males on each farm (response variable = territory count; offset = farm area) as a function of three fixed effects:

  • (i) farm type (1 = RSS; 2 = FBL; 3 = Control),
  • (ii) years since a farm joined its ‘farm type’ group (covariate ‘duration’),
  • (iii) number of survey visits (‘visit’, two or three),

plus the farm type × duration interaction term, in a generalized linear mixed model (GLMM) framework. This was specified by the sas 9.1 glimmix procedure with a log-link function, Poisson error distribution, and standard errors adjusted for over-dispersion, fitting farm identity (n = 94 ‘farms’ after including treatment changes) as a random effect. Denominator degrees of freedom for tests of fixed effects were calculated using the Kenward-Roger method (Littell et al. 1996). First we fitted ‘visit’, ‘farm type’ and the ‘duration × farm type’ interaction term, from which back-transformation of regression coefficients to the scale of the response variable gave an estimate of the annual percentage rate of change in the density of territorial males for farms in each treatment group. We then added ‘duration’ to the model to assess differences in trend between each farm type. Finally, to check for confounding of treatment effects with calendar year, we repeated the analysis, replacing ‘duration’ with the covariate ‘calendar year’ (2003–2009). Differences in model output between the two approaches proved negligible.

Secondly, we repeated the analyses, but replaced ‘farm type’ as a descriptor of scheme identity with an alternative three-level fixed effect ‘option type’ that described the resources offered by these options. The three levels were food (standard RSS or FBL options), food plus safe nesting habitat (as above but including FBL ‘late-cut grass’), and no options (control farms). Because safe nesting habitat was implemented mainly in Aberdeenshire and Inverness-shire due to few fields of grass silage occurring in Fife and Angus (Table 3), this analysis was done separately on the two areas to investigate regional differences in the influence of the management options.

Finally, because differences between the three treatment groups other than AES management may have influenced population trends, we tested (Kruskal–Wallis test) whether agricultural land use (proportions of each major crop) differed between the three groups in 2003 and 2008–2009, and whether there was significant change in each farm group between these years (Table 3). Further, for 62 farms where cropping was recorded in both 2003 and 2008–2009, we tested for significant change in these proportions using a Wilcoxon signed rank test.


Population changes

Corn buntings were recorded on 63 of the 71 farms surveyed. The maximum number of males on an occupied farm varied from 1–46 and density from 0·21–14·71 km−2. Modelled population trends differed between FBL and control farms (t = 4·53, d.f. = 317, < 0·0001), and between RSS and controls (t = 2·85, d.f. = 317, = 0·0047). The density of territorial males on control farms declined at 14·5% per annum (t = −3·66, d.f. = 317, = 0·0003), and on FBL farms increased at 5·6% per annum (t = 2·73, d.f. = 317, = 0·0066). On RSS farms, the rate of decline of 2·0% per annum did not differ significantly from zero (t = −0·89, d.f. = 317, = 0·377), but the difference between RSS and FBL trends was significant (t = 2·52, d.f. = 317, = 0·0122) (Fig. 2a).

Figure 2.

 Corn bunting population trends on agri-environment scheme (AES) and control farms, plotting model estimates for mean density of territorial males per farm type (±1 SE) in each treatment year (defined as year of management for AES farms, and year of control within 2003–2009 for control farms). (a) All farms (n = 63) (b) Aberdeenshire, Moray & Inverness-shire (n = 52) (c) Fife & Angus (n = 11) Note that for clarity, SE not plotted for late-cut grass + food options (year 1: lower SE = 0.42, upper SE = 0.66; year 2: lower SE = 0.89, upper SE = 2.17).

In AES without delayed grass mowing in Aberdeenshire, Moray and Inverness-shire, populations showed no significant trend (2·3% pa decline: t = −0·89, d.f. = 255, = 0·376), but revealed weak evidence of increase (6·3% pa) on farms with late-cut grass (t = 1·73, d.f. = 255, = 0·0840). This difference was significant (t = 1·98, d.f. = 255, = 0·0491) (Fig. 2b). In Fife and Angus, populations increased (17·8% pa) on FBL farms with options providing food (t = 4·22, d.f. = 55, < 0·0001), but declined (33·6% pa) on farms without this (t = −2·38, d.f. = 55, = 0·0209) (Fig. 2c), and the difference between these two trends was significant (t = 3·25, d.f. = 55, = 0·0020).

Land use and farming systems

In both 2003 and 2009, farms in Fife and Angus had less grass (χ= 7·14, d.f. = 1, = 0·0075; χ= 10·29, d.f. = 1, = 0·0013) and more vegetables (χ= 20·31, d.f. = 1, < 0·0001; χ= 24·31, d.f. = 1, < 0·0001), and in 2009 more legumes (χ= 8·19, d.f. = 1, = 0·0042) and autumn-sown cereals (χ= 5·53, d.f. = 1, = 0·0187) than those in Aberdeenshire, Moray and Inverness-shire. Within each region, land use was similar across the three treatment groups, although in Aberdeenshire, Moray and Inverness-shire in 2003, the proportion of autumn-sown cereals was greater on RSS than control farms (χ= 8·02, d.f. = 1, = 0·0046), and of vegetables greater on FBL than on RSS farms (χ= 5·01, d.f. = 1, = 0·0253). In 2008–2009, FBL farms in this region had a larger proportion of legumes than RSS farms (χ= 8·32, d.f. = 1, = 0·0039) and a larger proportion of grass than controls (χ= 4·01, d.f. = 1, = 0·0453). In Fife and Angus, the only significant difference between treatment groups was in 2003 when FBL farms had a greater proportion of rough grass and set-aside than controls (χ= 3·85, d.f. = 1, = 0·0499). Across 62 farms surveyed in 2003 and 2008 or 2009, the only crop type whose proportion changed was set-aside and rough grass (< 0·0001), which declined in all treatment groups in both regions due to withdrawal of compulsory set-aside in late 2007 (Table 3).

Identifying a population target for agri-environment management for corn buntings

If a closed corn bunting population of size N is divided into a proportion, p, which benefits from AES and a proportion, 1 − p, which does not, and the annual population growth rates of these two proportions are a and b, respectively, then for inter-annual population change t to t + 1:


which by re-arrangement for the case where Nt + 1 = Nt (i.e. a stable population) gives:


Substituting in the best estimate values of a = 1·056 (from FBL farms) and b = 0·855 (from control farms), gives p = 0·72. In other words, given a current rate of decline of 14·5% per annum in the wider countryside outside AES, then at least 72% of the corn bunting population would need to benefit from agri-environment management to halt the overall decline, assuming all was at FBL standard. Given that a = 1 (or at least does not differ significantly from 1) for RSS farms, then the entire corn bunting population would need to benefit from corn bunting-relevant RSS management to halt the overall decline. These estimates assume that the annual rate of population change observed on control farms is representative of the population as a whole. Given that our study sites were spread throughout the remaining breeding range of the species in mainland Scotland, this seems a reasonable assumption. By 2009, farms under FBL-type management supported 167 corn bunting territories, 24% of the remaining Scottish mainland population of approximately 700 territories. The mean annual cost of AES options on these 16 farms over 4 years, covering 186 ha, was approximately £40 000 (see Table 1 for payment rates). Data were unavailable to allow us to assess the total extent of population coverage by RSS agreements.


In eastern Scotland, over 7 years, corn bunting declines halted on farms where AES management with options broadly designed to benefit farmland birds (RSS) were implemented, but increased (5·6% pa) where AES management was specifically targeted at corn buntings (FBL). Adaptive improvement of FBL by adding delayed mowing of grass grown for silage in fields where corn buntings were nesting may have been a critical addition to the scheme. Before this was introduced, preliminary monitoring of FBL farms revealed that populations were maintained, but did not increase (Perkins et al. 2008), and late cutting is known to increase nest success rates (Wilson et al. 2007). Outside AES management, corn buntings continued to decline at a rate (14·5% pa) now greater than that (10·3% pa) observed on a partially independent sample of study areas over a longer period (1989–2007) by Watson et al. (2009). Although non-AES land use did not vary greatly between farm types and years, some differences could have influenced corn bunting populations. Notably, the reduction in set-aside and rough grass over the years of study may have contributed to declines on control farms that did not benefit from the ameliorating effect of AES measures.

Overall, the AES available were capable of reversing corn bunting declines in eastern Scotland. However, success requires AES management with the biological and spatial targeting found in FBL to be made available to approximately three-quarters of the current population, a large increase on the current level of availability that targets only 24%. Adaptive improvement to management options has been possible in FBL because agreements are flexible and renewed annually. Participants also received regular advice, which may be critical for options that require frequent interventions to recreate or maintain a habitat in good condition (Morris 2004). In the RSS, however, agreements were fixed for 5 years, precluding annual adjustments to improve the effectiveness of options, and expert advice was usually lacking. These differences may have contributed to larger responses on FBL than on RSS farms.

Some FBL options, notably late-cut grassland, necessitate substantial payments for profit foregone (Table 1). The finer-scale breakdown of the trend analyses suggests that it may be possible to achieve successful outcomes more cost-effectively by targeting different combinations of management options in different areas. Thus, in arable-dominated Fife and Angus where few corn buntings nest in grassland, management to provide food may be sufficient, but in Aberdeenshire, Inverness-shire and Moray, provision of safer nesting habitat via late-cut grass silage fields is likely to be a crucial additional measure. Studies which demonstrate this need for geographically targeted variation in agri-environment management are rare, although Batáry et al. (2010) show that effectiveness of grassland extensification schemes for bees is high only in countries (e.g. Switzerland) with intermediate farming intensity.

The AES management implemented during this study is likely to have benefited other species of high conservation concern. For example, seed-eating passerines made great use of unharvested crops and stubbles on our study farms during winter (Perkins, Maggs & Wilson 2008), whilst delayed mowing of grass can provide safe nesting habitat for skylarks Alauda arvensis (L.), and refuges for a wide variety of invertebrates after conventional fields are mown (Wilson et al. 1997; Woodcock et al. 2009). Field margin management similar to that in our study can benefit arable flora, butterflies and small mammals (Askew, Searle & Moore 2007; Aviron et al. 2007; Walker et al. 2007), as well as arthropods and soil macrofauna, which may themselves enhance pollination, pest control and improved soil structure (Smith, Potts & Eggleton 2008; Albrecht et al. 2010).

Synthesis and applications

We have shown that current AES options have the potential to reverse losses of one of Scotland’s most rapidly declining farmland birds. However, geographical targeting and flexible, adaptive improvement of measures, backed by advice from experts with sound knowledge of the species are likely to be crucial. Fulfilling this potential depends upon increasing the proportion of the current population that is targeted, from approximately a quarter (in 2009) to around three-quarters. Based upon the scale of land management and financial cost of measures provided through FBL (but excluding costs associated with expert advisory input), meeting this target would require 500–600 ha of land managed appropriately, at a total cost of around £120 000 per year. This amounts to 0·02% of agricultural and agri-environment subsidies currently paid to Scottish farmers annually (Scottish Government 2009), and 0·5% of land in the current mainland range of corn buntings in Scotland. The potential for success certainly exists. In 2008, the RSS was replaced by Rural Development Contracts (RDCs), which fund relatively simple ‘Land Managers Options’ available to all farmers, and more demanding and expensive Rural Priorities (RPs), which like RSS, is competitive (Scottish Government 2010). However, unlike RSS, the RP scheme is structured to deliver national and regional agri-environment priorities, including biodiversity, through ‘packages’ of options tailored to achieve specific outcomes. One such ‘package’ targets corn buntings, offering management interventions implemented and tested in FBL (Table 1), and several of our FBL farms have now transferred into the RP scheme. Contrary to conclusions of a recent report on the future of agriculture funding in Scotland (Pack 2010), our results show such targeting of AES is essential to reverse species declines.

Our study illustrates the value of AES monitoring, to test scheme effectiveness, allow adaptive improvement of implementation, and to estimate the scale of provision needed. Since the review by Kleijn & Sutherland (2003), studies recommending improvements to AES management are more common (e.g. Douglas, Vickery & Benton 2009; Smith et al. 2009), but those estimating the scale of intervention needed to reverse large-scale population decline remain rare (Vickery et al. 2004). Perhaps the best example is Aebischer & Ewald’s (2004) estimate that recovery of British grey partridge Perdix perdix (L.) populations to 1990 levels would require management of 5% of arable land as insect-rich brood-rearing habitat through reduced use of agrochemicals, and 6·9 km km−2 of field boundary nesting habitat. Many agri-environment monitoring studies have compared biodiversity trends on AES and control farms over several years (e.g. Kleijn & van Zuijlen 2004; Swetnam et al. 2004; Stevens & Bradbury 2006; Wilson, Vickery & Pendlebury 2007; Roth et al. 2008; Davey et al. 2010). However, we are not aware of any that estimated the proportion of the population that must be targeted to halt population decline at the national scale. Such estimates have practical value because, accompanied by data on distribution and abundance of target species, they help AES administrators to assess the extent, cost and spatial targeting of AES implementation necessary to meet conservation targets for species of high conservation concern. This is particularly important in current times of financial pressure, when it is essential to ensure that public funds are used to best effect.


We thank farmers and landowners for allowing access and providing maps and information, and Stuart Benn, Amanda Biggins, Ken Bruce, Alan Bull, Steven Coyne, Karen Cunningham, Paul Doyle, Stephanie Ferguson, Richard Firmin, Ian Francis, Flora Grigor-Taylor, Clive McKay, John McMahon, Nicky Penford and Chris Smout for support and assistance with monitoring. Scottish Natural Heritage and the Scottish Government provided financial support for Farmland Bird Lifeline, and the Farming & Wildlife Advisory Group provided details of farm management plans. We are grateful to the Editors and reviewers for valuable comments.