Winter availability of cereal stubbles attracts declining farmland birds and positively influences breeding population trends

Abstract

Many studies have demonstrated the selection of stubble fields by farmland birds in winter, but none have shown whether provisioning of this key habitat positively influences national population trends for widespread farmland birds. We use two complementary extensive bird surveys undertaken at the same localities in summer and winter and show that the area of stubble in winter attracts increased numbers of several bird species of conservation concern. Moreover, for several farmland specialists, the availability of stubble fields in winter positively influenced the 10 year breeding population trend (1994–2003) whereas hedgerow bird species were less affected. For skylarks and yellowhammers, initially negative trends showed recovery with 10–20 ha of stubble per 1 km square. Thus, agri-environment schemes that promote retention of over-winter stubbles will attract birds locally and are capable of reversing current population declines if stubbles are available in sufficient quantity.

1. Introduction

There are now well-documented, large-scale population declines and range contractions in many bird species inhabiting farmland across northwest Europe (Pain & Pienkowski 1997; Donald et al. 2001a). Agricultural intensification since the 1970s, principally through the Common Agricultural Policy, has fostered an increased use of chemical inputs, a switch from spring to autumn sowing of crops, the loss of non-cropped habitats and the loss of rotations and farm-scale mosaics owing to local specialization (Chamberlain et al. 2000; Vickery et al. 2001; Robinson & Sutherland 2002; Benton et al. 2003). A wealth of research has shown that one or more components of these changes have negatively affected either farmland bird productivity, survival or both (e.g. Siriwardena et al. 2000).

In 2000, the UK government adopted a Public Service Agreement (PSA) target to reverse the farmland bird declines by 2020. Achieving this PSA target is likely to require sympathetic land management across large areas of the UK (Vickery et al. 2004). Given that reduced over-winter survival explains patterns of decline for several granivorous species (Siriwardena et al. 2000) and that their preferred winter feeding habitat is stubble fields (e.g. Wilson et al. 1996; Moorcroft et al. 2002), the retention of over-winter stubbles is one of the most widely advocated conservation measures. Prior to the 1970s, most crops were planted in spring, allowing seed-rich stubble fields to remain undisturbed, thus providing an essential food resource in winter. With the subsequent switch to autumn sowing, many stubbles are now ploughed before winter (Evans et al. 2004) with a consequent loss of winter food for farmland birds. Moreover, many stubble fields now provide poor resources owing to more efficient harvesting and weed control, depleting seed banks and reducing the abundance of broad-leaved weeds important in the diet of many farmland passerines (e.g. Donald et al. 2001b). For widespread species such as skylarks Alauda arvensis, the provision of seed-rich stubbles will be required throughout Britain but the quantity (and quality) and spatial arrangement of stubble required to stem and reverse bird population declines is unknown (Bradbury et al. 2004). However, it is clear that agri-environment schemes (AES) may be the most cost effective means of deploying sympathetically managed habitats on a national scale.

To date, AES have been successful in reversing population declines in only four farmland species (Aebischer et al. 2000). Two of these, corncrake (Crex crex) and stone curlew (Burhinus oedicnemus) require highly specialized breeding habitat, and AES could encompass the entirety of their restricted breeding ranges. The other two species, grey partridge (Perdix perdix) and cirl bunting (Emberiza cirlus) require provision of winter and breeding habitat. The grey partridge is relatively widespread but management has only been effective at the local level. The cirl bunting recovery might be informative because the species is ecologically similar to several of the more widespread species of concern. However, population recovery followed ad libitum provision of over-winter stubbles and breeding habitats (Peach et al. 2001), hence, no relationships between stubble provision rate and population trends can be determined. Therefore, although these success stories show promise, they do not help in determining levels of resource provisions needed for birds of the wider countryside. Moreover, in most cases, management was intensive and expensive, and thus unlikely to be tenable nationally.

The populations of widespread species must be restored if the PSA target is to be achieved. Here, we investigate the effects of stubble availability in winter on local breeding populations to determine the levels of sympathetically managed farmland required to reverse farmland bird declines. We use data from two extensive surveys in which volunteer surveyors visited the same 1 km squares in summer and winter, providing a unique opportunity to link summer and winter abundance with winter habitat data and answer two key questions.

1. Relative to the number of birds present in summer, does the presence of stubble fields in winter attract further birds into areas?

2. Does the winter availability of stubble fields within squares positively affect recent breeding population trends of farmland birds?

The first question considers whether AES are capable of having effects over wider areas of surrounding countryside, and the second question attempts to ascertain what resource levels are required to reduce or reverse farmland bird declines.

2. Methods

(a) Survey methods

This study uses two extensive volunteer surveys, the British Trust for Ornithology/Royal Society for the Protection of Birds/Joint Nature Conservation Committee (BTO/RSPB/JNCC) Breeding Bird Survey (BBS) and the BTO/JNCC Winter Farmland Bird Survey (WFBS). Since 1994, the BBS has been the national monitoring scheme for breeding bird populations within the UK (Raven et al. 2004). The WFBS aimed to document winter abundance, distribution and habitat selection of farmland birds in the three winters 1999/2000, 2000/2001 and 2002/2003 (Gillings 1999). Both were undertaken on stratified random samples of 1 km squares: BBS squares (ca 2000 squares annually) were stratified regionally and by human population density to afford representative coverage of regions and habitats while making the most of available volunteer resources (Raven et al. 2004). WFBS square selection (1090 squares) was constrained to lowland farmland areas and avoided largely urbanized or wooded areas (Gillings 1999). Any BBS squares that met the latter constraints were included within the WFBS sample. This gave a sample of 601 squares surveyed in both winter and the breeding season. Both survey methods aimed to sample bird populations within the square rather than yield exact population figures. BBS essentially involved two visits to each square each summer (between May and July) with all bird species counted along two 1 km transects. WFBS involved three timed visits each winter in which a suite of 30 farmland bird species and habitats were surveyed from field perimeters.

Of the 30 species surveyed in winter, 12 were excluded from analysis because they were strictly winter visitors, for example, fieldfare (Turdus pilaris) or essentially absent from lowland farmland in summer, for example, stonechat (Saxicola torquata) and were thus not present in summer to make comparisons. The remaining 18 species were typical of lowland farmland in summer and winter (table 1). We used the maximum count across visits in each season as the measure of bird abundance. On each winter visit, the area of broad agricultural habitat type (e.g. grass, crop, stubble) per 1 km square was estimated from WFBS habitat data. For classifying sites for the trend analysis, these values were averaged across all visits to derive an overall description of the farmland present in each sample square during the late 1990s and early 2000s.

Table 1.

Results of a repeated measures generalized linear model predicting winter numbers as a function of summer numbers, then testing the additional effect of increasing availability of different habitats. (n, number of squares recording the species at least once; slope, the direction and magnitude of the relationship between winter and summer abundance (significance tested against zero); r, correlation coefficient between observed and predicted winter numbers. Columns for habitats show the direction and significance of the presence of each habitat type on winter bird abundance relative to summer abundance. *p<0.05, **p<0.01, ***p<0.001, +, positive; −, negative, where number of symbols indicates significance as for asterisks.)

species	n	slope	r	grass	crop	stubble	farmyard
grey partridge	219	−0.43***	0.24			+
lapwing	327	−0.32***	0.16	−
stock dove	345	−0.03	0.02
skylark	512	0.74***	0.40	−−−		+++
pied wagtail	521	−0.22***	0.12	++	−−
song thrush	557	0.20***	0.13	+++	−
mistle thrush	500	−0.05	0.04	+	−−−
starling	566	0.06	0.04	+++	−−−	−
house sparrow	481	0.28***	0.19	++	−	++	+++
tree sparrow	119	−0.36	0.23			+
chaffinch	586	0.69***	0.36	+++	−−−	+++	++
greenfinch	549	0.25***	0.16	−		++	+
goldfinch	513	−0.30***	0.16			+
linnet	476	−0.10	0.05	−−−		+++
bullfinch	350	−0.45***	0.29	++	−−
yellowhammer	478	0.51***	0.26	−−−		+++
reed bunting	248	−0.34*	0.20			++
corn bunting	97	−0.09	0.05

(b) Are birds drawn into squares with key habitats in winter?

If all individuals are sedentary, then the number of birds in a square in winter should be proportional (although not equal owing to recruitment, mortality and different survey methods) to the number in the preceding spring. However, differences in habitat composition between squares ought to be apparent. For instance, a square with an abundance of seed-rich habitats in winter might be expected to have higher winter bird populations than the ‘average’ square owing to reduced mortality and birds being attracted from surrounding squares.

Because visits to the same square across years cannot be considered statistically independent, tests were performed using a repeated measures generalized linear model which modelled winter abundance as a function of summer abundance, then tested the additive effect of winter habitat availability. Summer counts for 1999, 2000 and 2002 were merged with winter counts and habitat areas for 1999/2000, 2000/2001 and 2002/2003, respectively. Initial analysis used Poisson errors and a log link, but over-dispersion was extreme. Instead, both counts were ln(x+1) transformed, which gave acceptable model over-dispersion (values of deviance/degrees of freedom in the range 0.4–4.7) and the square root of the deviance/degrees of freedom was used for scaling. For the repeated measures, square grid reference identified subjects and year denoted the within-subject order. Habitat effects were tested by examining the drop in deviance (with a likelihood ratio test) when a variable describing the availability of a single habitat type was added to the winter–summer relationship. The effect of each habitat was tested in a separate model, thereby avoiding statistical problems associated with the unit sum constraint. The model was run separately for each species after excluding any squares that never recorded any individuals of the species in either summer or winter. All tests were performed using the GENMOD procedure in SAS (SAS Institute 2001).

(c) Does over-winter stubble availability influence breeding population trends?

If preferred winter habitats reduce over-winter mortality, then squares with those habitats should have higher populations in the subsequent breeding season than squares without those habitats. Over several seasons, there should be detectable differences in trends between squares with and without these supposedly beneficial winter habitats, with higher growth rates (or at least lessoned declines) in squares with preferred habitats.

This was tested by computing breeding population trends for squares with and without cereal stubbles. For each of the 601 BBS squares, we used WFBS habitat data to derive an estimate of the area of cereal stubble present during winter in the late 1990s and early 2000s, and then computed separate breeding population trends from BBS data (between 1994 and 2003) for squares with and without over-winter cereal stubbles. Two separate sets of trends were produced: national lowland trends using all 601 squares and eastern England trends using 248 squares in the East Midlands, East of England and South East Government Office regions. The reason for this dual approach was that the national trend benefited from greater between-square variation in stubble availability but suffered from the problem that declines may have different drivers in different parts of the country. The eastern England-only analysis reduced this problem by concentrating on the predominantly arable zone of Britain, thus reducing the likelihood of multiple drivers. The eastern region also provided the most squares and thus the best power to detect what could be quite small changes in trends over just a 10 year period.

Breeding population trends were estimated from BBS data using a log-linear model with Poisson error terms (Raven et al. 2004). A baseline trend was calculated using site effects and a linear year effect. Models were re-run with a year×winter stubble presence–absence interaction term to test for trends specific to the availability of over-winter stubble. Trends were derived for the 18 resident farmland species targeted by both surveys, plus for eight common species of farmland that were adequately monitored by BBS. Of this pool of 26 species, we expected to see beneficial effects of stubble presence on trends for some species and not for others. Specifically, based on current knowledge of habitat preferences (e.g. Wilson et al. 1996; Moorcroft et al. 2002), grey partridge, skylark, sparrows, finches and buntings (hereafter referred to as ‘stubble species’) should show positive effects of stubble presence because they will benefit from the associated seed resources. In contrast, wren (Troglodytes troglodytes), robin (Erithacus rubecula), dunnock (Prunella modularis), thrushes and tits (hereafter referred to as ‘hedge species’) rarely use field habitats and should show no consistent effect of stubble presence on population trends. It is not clear what effect might be expected for the remaining six ‘other species’.

Trends for skylark and yellowhammer (Emberiza citrinella) were investigated further. Both are widespread species of particular conservation concern, but whereas yellowhammers may benefit simply from the provision of stubble fields though over-winter effects (Bradbury et al. 2000), skylarks may benefit doubly from the additional presence of spring-sown crops (Chamberlain et al. 1999) associated with over-wintering of stubble fields. Thus, squares were classified as having no, low or high stubble availability and trends were computed and tested for significant differences. The threshold between low and high availability was varied from 5 to 20 ha (of a maximum of 100 ha of land in the square). For visual presentation, annual indices were determined using categorical year factors. Finally, trends were modelled with a continuous stubble covariate.

Models were corrected for over-dispersion using the square root of the deviance/degrees of freedom and were weighted to account for the original square stratification. Only squares that were surveyed in at least two summers, and thereby generating a measure of ‘change’, were included in the analysis. Trends were only computed for species for which the mean number of sites contributing counts in each year was 30 or more (Raven et al. 2004).

3. Results

(a) Are birds drawn into squares with key habitats in winter?

The abundance of individual species was significantly related across seasons for 12 of the 18 species; seven negative and six positive (table 1). Of those species showing negative relationships, grey partridge, pied wagtail (Motacilla alba), bullfinch (Pyrrhula pyrrhula) and reed bunting (Emberiza schoeniclus) relationships were caused by a large number of sites that were apparently unoccupied in summer but with an abundance of birds in winter. In addition to these patterns of seasonal abundance, winter habitat covariates explained significant variation (at p<0.01) in the summer–winter relationship, with the availability of grass and stubbles benefiting six and seven species, respectively, and crops negatively affecting five species. Thus, for a species such as skylark, squares with high densities of skylarks in summer had relatively higher densities in winter if stubbles to some degree replaced crops. Notably, the availability of farmyards positively influenced house sparrow (Passer domesticus) and chaffinch (Fringilla coelebs) winter abundance.

(b) Does over-winter stubble availability influence breeding population trends?

Nationally, significant baseline trends were detected for 20 of the 26 species considered (table 2). Significant declines were evident for 50% of the stubble species but none of the hedgerow species. Moreover, significant increases occurred in only 25% of stubble species compared with 75% of hedgerow species. Results were similar in the east where 19 of the 27 species analysed showed a significant trend: 58% of stubble species declined compared with 13% of hedgerow species, and 17% of stubble species increased compared with 50% of hedgerow species. Mixed patterns were evident for the other species.

Table 2.

Percentage change in breeding bird numbers on BBS squares between 1994 and 2003. (All includes 601 squares distributed throughout Britain. East includes 248 squares in east England. A baseline trend is given followed by separate trends for squares with (Y) and without (N) over-winter cereal stubble. For baseline trends, asterisks indicate a significant change. For stubble trends, asterisks indicate a significant difference between trends in the presence or absence of stubble. *p<0.05, **p<0.01, ***p<0.001.)

species	number squares		baseline		stubble (Y/N)
	all	east	all	east	all	east
stubble species
grey partridge	68	34	−43***	−42***	−40/−58	−38/−58
skylark	348	167	−18***	−18***	−13/−34***	−14/−31**
house sparrow	300	122	5	−14**	7/1	−12/−20
tree sparrow	40	12	139***	−69***	30/434***	−
chaffinch	414	175	2	15***	5/−3	19/4*
greenfinch	332	144	15***	15*	11/24	25/−10*
goldfinch	273	113	12*	−13	14/7	−8/−28
linnet (Carduelis cannabina)	272	122	−22***	−36***	−18/−32	−29/−57**
bullfinch	102	45	−25**	−22	−19/−32	−28/−4
yellowhammer	305	148	−20***	−15***	−16/−30*	9/0
reed bunting	84	42	−7	−1	−13/4	−3/1
corn bunting (Emberiza calandra)	50	30	−35***	−32**	−37/−22	−33/−28
hedge species
wren	396	169	21***	16***	21/22	22/−3**
dunnock	367	156	18***	12*	17/20	15/6
robin	388	163	17***	18***	14/21	14/28
blackbird (Turdus merula)	420	181	17***	10***	18/15	13/4
song thrush (Turdus philomelos)	315	132	23***	−3	23/21	27/−81
mistle thrush	211	99	−9	−34***	−3/−18	−25/−50*
blue tit (Parus caeruleus)	393	164	−5	−7	−3/−7	−6/−9
great tit (Parus major)	358	152	15***	5	15/14	7/3
other species
lapwing	142	57	16*	56***	44/−33***	84/−1*
stock dove	173	88	14	8	24/−8*	94/19**
woodpigeon (Columba palumbus)	423	183	19***	20***	18/22	−2/−1
pied wagtail	229	91	20***	54***	31/4*	50/66
rook	262	110	−13*	−7	−20/3*	−26/47***
starling	348	151	−36***	−46***	−41/−29	−18/−33

In total, 16 species showed a positive effect of stubble presence on national trends and 10 species showed a negative influence. For example, skylarks declined by 34% on squares with no stubble compared with only 13% on squares with stubble present (table 2). Only five positive effects were statistically significant: two for stubble species, none for hedgerow species and three for other species. Of the other species showing significant effects, lapwing (Vanellus vanellus), stock dove (Columba oenas) and pied wagtail responded positively to stubble presence whereas rook (Corvus frugilegus) responded negatively.

Whether squares had stubble or not differed markedly between regions of Britain. In east England, only 30% of squares had no stubble compared with 83% of Welsh squares. Constraining the analysis to east England removed much variation in stubble availability, but still 18 species (eight significantly) showed a benefit of stubbles and seven a negative influence (only rook significantly). Of the species showing significant positive effects of stubble, four were stubble species, two were hedgerow species and two were other species.

The effect of increasing the area of over-winter stubble on trends was evaluated further for skylark and yellowhammer. Of the 601 squares, 237 contained no stubble, leaving 364 squares among which there was appreciable variation in the extent of stubble (quartiles for non-zero squares were 9 and 69 ha). For different thresholds of stubble availability (5–20 ha), skylark declines were lessened with greater stubble availability (table 3a). Populations on less than 10 ha stubble squares declined by 20% compared with the 34% decline in the complete absence of stubbles. Moreover, populations on greater than 10 ha stubble squares declined by only 4% (figure 1). Crucially, only when stubble availability exceeded 20 ha per square was the 10 year linear trend stable/increasing (table 3a). Results were similar for yellowhammer, with a lessening of declines in the highest stubble availability category and approximate stability above 15 ha of stubble (table 3b). Tested more formally, when stubble area was included as a covariate, there was a significant interaction with year for skylark (likelihood ratio ${\chi }_1^2=18.3$, p<0.001) and yellowhammer (likelihood ratio ${\chi }_1^2=10.4$, p<0.01). For skylarks, for each addition of 5 ha of stubble, the 10 year decline was lessoned by four percentage points. Six other species showed a significant positive linear effect of stubble area (lapwing, stock dove, mistle thrush (Turdus viscivorus), starling (Sturnus vulgaris), goldfinch (Carduelis carduelis) and bullfinch); and two showed a significant negative effect (house sparrow and tree sparrow (Passir montanus)) on breeding population trends.

Table 3.

Breeding population changes for (a) skylark and (b) yellowhammer under different levels of over-winter cereal stubble availability. (Using different thresholds of stubble availability to denote the difference between low and high availability, trends are given for squares with no, low or high stubble availability. Sample sizes are given for each category. The likelihood ratio (LR) tests for difference in trend in relation to stubble availability. *p<0.05, **p<0.01, ***p<0.001.)

low–high threshold (ha)	trend by stubble area (%)			number of squares			LR test ${\chi }_1^2$
	none	low	high	none	low	high
(a) skylark
5	−34	−22	−9	237	116	248	24.2***
10	−34	−20	−4	237	210	154	27.9***
15	−34	−17	−5	237	266	98	23.5***
20	−34	−16	1	237	308	56	26.4***
(b) yellowhammer
5	−30	−9	−19	237	116	248	6.3*
10	−30	−23	−9	237	210	154	9.5**
15	−30	−23	0	237	266	98	14.8***
20	−30	−20	−1	237	308	56	9.7**

Figure 1.

Skylark breeding population trends in squares with 0, ≥10 or >10 ha of over-winter stubble. Dashed lines indicate linear trends from table 3. Solid lines with symbols indicate annual fitted values. The same symbols are used for the last point of each linear trend (dashed lines) to aid interpretation.

4. Discussion

A wealth of research has focused on diagnosing the causes of population declines in farmland birds in Europe. For many species, the underlying demographic and environmental causes are known and land management options have been developed to reverse these population declines. One of the key questions that remains is how much of a given resource is required to have an impact at the national population level? This is the first study to link summer and winter bird communities on farmland across broad geographical areas and to attempt to relate breeding season trends to the availability of such a resource—over-winter stubbles. The results show that the availability of over-winter stubble can explain some of the variation in population trends for several declining species.

For most species, abundance in winter relative to summer was significantly affected by winter habitat availability, indicating that certain habitat features in one locality were consistently capable of attracting birds from the surrounding countryside. No species increased in response to the presence of crops. Species responding positively to grass were primarily invertebrate feeders, whereas those responding positively to stubbles were primarily granivorous passerines. Many autecological studies have shown strong preferences for stubble fields in winter by the latter (e.g. Wilson et al. 1996; Donald et al. 2001b; Gillings & Fuller 2001; Moorcroft et al. 2002). Interestingly, house sparrow, chaffinch and greenfinch (Carduelis chloris), showed a positive association with the availability of farmyards.

As expected, the benefits of stubble differed between species in relation to their foraging ecology. More of the positive significant effects were for stubble species than for hedgerow species. Furthermore, the remaining non-significant benefits tended to be greater for stubble species than for hedgerow species. That some benefits of stubble (albeit small) were apparent for hedgerow species may be a result of landscape effects where farms that retain stubbles also tend to have greater availability or higher quality non-cropped habitats (e.g. hedgerows). This may be the case if stubbles are more common on lower intensity (or organic) farms that tend to retain more non-cropped habitat (Benton et al. 2003). The beneficial effects of the stubble are likely to be twofold. Over-winter stubbles provide essential seed-rich resources for a wide array of species, plus stubbles are associated with summer cropping which benefits productivity in skylarks (Chamberlain et al. 1999) and lapwings (Wilson et al. 2001). Perhaps this dual mechanism explains why the response is greater in skylark than yellowhammer.

The detailed examples for skylark and yellowhammer clearly show that increasing the quantity of stubble in squares can reduce, and even reverse, local population trends during the ten-year period considered here. The skylark results can be used to indicate how much stubble might be needed in the wider countryside to reverse population trends for this species. At the 10 ha threshold, all three categories of squares showed a decline until around 1997/1998. Since this time, populations in greater than 10 ha stubble squares have increased slightly. Though only greater than 20 ha stubble squares demonstrated an absolute skylark recovery to 1994 levels by 2003, trends in greater than 10 ha stubble squares might recover by around 2011 whereas those in 0 ha stubble squares might continue to decline down to −40% by 2011. Nationally, only 50% of squares contained stubble and of those, the median area was 12 ha (Gillings unpublished work). Therefore, only 25% of squares have the greater than 10 ha of stubble required for recovery whereas 75% of squares will probably show sustained skylark declines.

These increases in stubble availability may be achieved either through AES, or through changes in wider farm management practices, such as incorporating more fallow land or spring cropping in rotations, or the addition of pockets of arable land in grass dominated areas (Robinson et al. 2001). In Britain, the ‘national roll-out’ of the new entry-level AES has the potential to provide beneficial resources on the national scale necessary to aid population recovery. This work suggests that if stubble areas can be increased from the current average of 3 ha per 1 km (Gillings & Fuller 2001) to 10 ha or more, then this should be sufficient to stem breeding population declines. Gillings & Fuller (2001) estimated that only 46% of stubbles were weedy and hence valuable as foraging habitat for birds. If management options are developed to enhance weed abundance in stubbles, then the area required could be significantly reduced. It should be borne in mind that the area and spatial arrangement of food patches required may differ between species. Skylarks avoid small fields and hedgerows (Donald et al. 2001a,b; Gillings & Fuller 2001) so stubble patches must be carefully sited if they are to be fully exploited by this species.

These are crude extrapolations and three points should be considered. First, the 1994 ‘recovery’ value used here is simply the beginning of the BBS scheme. Skylarks have been in decline since the mid-1970s, so a full recovery will take considerably longer. Second, these extrapolations could be misleading because only the processes operating under the current conditions are taken into account (e.g. density dependent effects are ignored; Bradbury et al. 2001). Third, there is enormous variability in the quality of stubble fields (Robinson 2003) and these extrapolations refer to the ‘average’ stubble field. Ideally, behaviour-based models alongside empirical data on stubble field quality should be used to predict future population trajectories under different management scenarios, but data are currently lacking (Stephens et al. 2003). Nevertheless, these results indicate that the provision of sufficient high quality, suitably placed stubble could at least make a significant contribution to population recovery for many of these species.

5. Conclusions and recommendations

These results demonstrate that the availability of over-winter stubbles in the last three years explains some of the variation in population trends for several declining farmland birds. The thresholds of stubble availability, at which changes in population trends were apparent, suggests that significant changes in land management may be required to sustain population recovery. Changes of this magnitude and scale are only likely to be possible within AES, and monitoring the benefits of these schemes is essential (Kleijn & Sutherland 2003). In particular, we advise the collection of spatially referenced information on the take-up of schemes and resulting areas of beneficial habitats that can be linked to bird population trend information at different spatial scales. Although we show that stubbles benefit population trends in squares with stubbles, the fact that birds were also attracted into those squares from outside indicates a likely effect on trends in the countryside surrounding the squares in question. While the distance over which this effect extends remains unknown, further research is needed to answer this question in order to determine the optimum spatial arrangement of resource patches.