Factors Affecting Populations of Farmland Birds. a Study of Data Collected Between 1993 and 1998 At

MINISTRY OF AGRICULTURE, FISHERIES AND FOOD Date project completed: -. Research and Development 28/02/1999 Final Project Report BDml1 (Not to be used for LINK projects)

1. (a) MAFF Project Code IBD0911

(b) Project Title FACTORS AFFECTING POPULATIONS OF FARMLAND BIRDS - A STUDY OF DATA COLLECTED BETWEEN 1993 Ai'ill 1998 AT ADAS DRAYTON

(d) Name and address ADAS CONSULTING LTD of contractor OXFORD SPIRES BUSINESS PARK THE BOULEYARD KIDLINGTON, OXON Pastcade OX5 1NZ

(e) Contractor's Project Officer I_M_R_C_P_B_R_lT_T _

(f) Project start date 101/iO/l99R Project end date 13 1/01/1999

(g) Final year costs: approved expenditure I £9,663

actual expenditure

(h) Total project costs 1total staff input: approved project expenditure I £9,663

actual project expenditure

'approved staff input

'actual staff input

(i) Date report sent to MAFF 124/03/1999------

Ul Is there any Intellectual Property ari~ing from this project?

'staff years of direct science effort

CSG 13 (1/97) 1 Ilij!!!.:I~Jlll~ilwl!II~~ijlllI11~ll!llll~g,~!lill~11l§l!ll~~,el!!;!!!~~II.;~I:'!~~!,!tgll!lllill111111~l!lill~11~IIIII!li!Jl[IIIIIIIII'I~ 2. Please list the scientific objectives as set out in CSG 7 (ROAME B). If necessary these can be expressed in an abbreviated form. Indicate where amendments have been agreed with the MAFF Project Officer, giving the date of amendment.

01 To collate Common Birds Census and farm management data from the ADAS Drayton ECN site.

02 To statistically analyse bird population data, to detennine the habitat preferences of skylarks and yellowhammers and examine other possible causes of spatial and temporal variation.

03 To compare Drayton ECN birds' data with those from lACR Rotharnsted and other appropriate lowland ECN sites.

04 To evaluate the short-tenn effects ofagricultural practices on populations ofskylarks and yellowhammers and provide recommendations for future research.

3. List the primary milestones for the final year. It is the responsibility of the contractor to check fully that ALL primary milestones have been met and to provide a detailed explanation if this has not proved possible Milestones Target Milestones met? date Number Title in full on time 01/01 Complete collation ofADAS Dra)10n birds (CBC) 10/11/1998 YES YES data

01102 Complete collation of ADAS Drayton farm 10/11/1998 YES YES management data

02/01 Complcte analysis of ADAS Drayton data. 04/12/1998 YES NO

04/01 Complete project report for I\'lAFF 31/01/1999 YES NO

If any milestones have not been met in the final year, an explanation should be included in Section 5. 111111111~'I!llllil!;IIIIIIIII~III~llii!IIl:!!i~lil!!I'1111~1~1\liiljllllllil!;!':i!lli;lllg!I~liil\llllI11~i1111!llill1!1111~llill!I':I:lllliili!I;llll!II'ilillllil1iliiililllil~1 4. I declare that the information I have given in this report is correct to the best of my knowledge and belief. I understand that the information contained in this form may be held on a computer system.

Signature I 1 Date 1----

Name 1 Christopher P Britt I Position in Organistation ILS_e_n_io_r_R_e_se_a_rc_h_C_o_n_Sl_ll_lan_t _

CSG 13 (1/97) 2 , • This study examined population changes/activity levels in two farmland bird species (skylark and yellowhammer) over a six year period, at ADAS Drayton - a lowland mixed farm in south Warwickshire - using registrations data from annual Common Birds Censuses, undertaken as part of the UK Environmental Change Network (ECN) project. Results were also compared with data obtained from a second lowland agricultural ECN site, at the Institute of Arable Crops Research at Rothamsted, in Hertfordshire.

National populations of both species have declined in recent years and this study, as well as determining whether similar declines occurred at Drayton, also attempted to determine the effects of cropping and the management of non-cropped areas on bird activity. Carabid beetle activity was also monitored each year (within the ECN project), and the numbers of beetles captured annually were compared with skylark and yellowhammer registrations, to examine any possible correlations between these groups.

The number of breeding territories of skylarks declined steadily over the period of the study. The number of registrations per hectare, however, did not differ significantly between years but did indicate major differences in the attraction of various crops to skylarks. Rotational set-aside, where vegetation was allowed to naturally regenerate, was very attractive in comparison with other crops. The lowest numbers of registrations were recorded on intensively managed grass fields.

The results generally agree with fmdings from other studies, and suggest that the sparser vegetation present in rotational set-aside allows skylarks to forage and nest successfully, whereas the denser, more disturbed intensively managed grassland does not. The invertebrate diet of these birds is also likely to be present in greater amounts on set-aside land. The correlation between Carabid beetle numbers trapped in field boundaries and skylark registrations was poor. This probably reflects the fact that the skylark is reliant on the open field habitat and that invertebrate populations, as measured by Carabid beetle numbers in field boundaries, do not allow valid comparisons between years to be made.

Yellowhammer populations were affected by the management of cropped and uncropped areas of farmland. Comparisons of various cropping combinations occurring each side of a field boundary indicated a positive effect for rotational set-aside and a negative effect for intensively managed grassland.

Yellowhammer population declines occurred when Carabid beetle numbers in field margins fell, indicating that these beetles may be a good indicator of prey availability for yellowhammers, and that uncropped field margins are an important habitat for foraging yellowhammers.

The beneficial effects of field margin management for wildlife in arable situations has been examined in a number of studies. There is an urgent need to ascertain whether similar benefits would be obtained from alternative management of intensively managed grass field margins, especially in the western half of the country, where the intensity of grassland management has increased most, and is likely to have had the greatest affect on species such as the yellowhammer.

CSG 13 (1/97) 3 A study offarmland bird populations: an analysis ofskylark (Alauda arvensis) and yellowhammer (Emberiza citrinel/a) registrations data from common birds censuses at ADAS Drayton, 1993-98

INTRODUCTION

The populations of a number of fannland bird species have declined in recent years. These declines have coincided with a period of increased intensification of agricultural production.

TI,e Environmental Change Network (ECN) project is carried out on a number of terrestrial and freshwater sites throughout Britain. It is a long-tenn integrated monitoring progran>me designed to identify and improve the understanding of the causes of environmental change. The site at ADAS Drayton is sponsored by the Ministry of Agriculture, Fisheries and Food (MAFF).

A wide range of physical, chemical and biological variables are measured by the ECN. One of the variables which has been measured since 1993 at Drayton is the population of breeding birds. During this time changes in populations of particular species have been recorded and the changes agree, in general, with those detennined in the national Common Birds Census (CBC) carried out by the British Trust for Ornithology (BTO). Another variable being measured is the number and species of Carabid beetles present at the site. Changes in the population levels of these insects have also been noted during the period 1993 to 1998. The site is one on which agricultural research has been carried out for many years and, therefore, detailed infonnation about the agricultural management of the site is available, enabling populations changes to be measured against changes in management practices.

Skylarks (Alauda arvensis) and yellowhan>mers (Emberiza cilrine/la) are two of the species of farmland bird which have experienced large population declines in recent years and have been the subject of much recent research (e.g. Grynderup Poulson el al., 1998; Wilson el al., 1997; Kyrkos el al., 1998; Sparks el al., 1996; Green el al., 1994). Nationally, skylark populations have been recorded as declining by almost 60% since 1970 in BTO studies.

Yellowhammer populations were relatively stable during the late 1960's and 1970's, while other farmland birds declined; but populations started to decrease in the late 1980's (Bradbury el aI., in press), with a rapid increase in the rate of decline in recent years. Currently, yellowhan>mer populations are declining by around 10% per year (Bradbury el aI., in press).

The two species occupy different areas of farmland. Skylarks live in more open habitats, nesting on the ground amongst crops. Yellowhan>mers, on the other hand, nest along field boundaries, either on the ground, in ditches or grassy margins, or in shrubby growth such as hedges. It feeds mainly in nearby adjacent habitats.

Various possible causes of decline in British yellowhan>mer populations have been suggestcd by previous authors (Morris el al., in press). These include: I. The shift, in western Britain, toward grass monocultures, which harbour smaller invertebrate prey populations and invertebrate diversity (e.g. in the Orthoptera and Staphylinidae) 2. A resultant reduction in cereal grain availability. 3. Loss of over-winter fallow; with less spilt grain, weed seeds and invertebrates available for food in the winter and in thc following breeding season. 4. Loss of non-cropped boundary features, which are important as nest sites and foraging habitats. 5. Increased use of pesticides, which has adversely affected many important food groups.

This study examined the changes in populations of skylarks and yellowhan>mers during the six years in which a census has been carried out at ADAS Drayton. These data were analysed with respect to changes in agricultural management during the period to ascertain whether particular practices were linked with the recorded population changes. Changes in the bird populations were also compared with those of Carabid beetles in order to detennine any relationship between changing prey availability and population trends.

CSG 13 (1/97) 4 File reference: RSC

C'.es in bird populations (as measured by the STO territory analysis) and Carabid beetle numbers during the same period were also compared with those obtained at IACR Rothamsted, near Harpenden, Hertfordshire - another ECN site.

MATERIALS AND METHODS

Study site

AD AS Drayton is an agricultural research site of 192 hectares, owned by MAFF. It is located 5 Ion to the west of Stratford-upon Avon in Warwickshire.

During the period of the study approximately 80% of the farm grew arable crops or short term rye-grass (Lotium perenne) leys, the rest of the area being mainly permanent grass. The soils of the site are all heavy calcareous clays, which restrict arable cropping primarily to winter wheat with break crops offield beans and, occasionally, oilseed rape.

Land set aside from arable production (as required by the rules of the Arable Area Payments Scheme) was primarily in rotational (one year) set-aside; in which weed and crop volunteers were allowed to naturally regenerate until early summer, when the green cover was killed using a 'total' foliar-acting herbicide (glyphosate). There were also other, smaller, areas of three or five year set-aside - established for experimental purposes. In these cases the majority of the area had been sown with a grass cover crop.

Permanent grass fields and temporary leys were managed intensively - being either grazed by cattle or sheep, at relatively high stocking densities, or used for silage production.

Field boundaries were typical of the area, being mainly short, hawthorn (Cralaegus monogyna) dominant hedgerows. The majority of hedgerows were trimmed annually to a square or rectangular shape, approximately 1.5 m high and 1-1.5 m wide; but the density of woody vegetation present varied.

There were very few mature trees present on the site, mostly of ash (Fraxinus excelsior) or oak (Quercus robur), and no woodland within the CSC plot. However, one large field immediately adjacent to the CSC plot (and within the Drayton farm boundaries) contained large numbers of young woodland trees, planted in experimental plots throughout the period between 1988 and 1995, and there was an area of more mature woodland just beyond the northern boundary of Drayton. In another field immediately adjacent to the CSC plot, approximately one hectare of high density, short rotation willow coppice was established in 1996.

Methods

Birds census The area of Drayton used as the bird census 'plot' in this study comprised an area of 62 hectares in the centre of the farm. This CSC plot included most of the croplhabitat types present on the farm as a whole, but with a smaller proportion of temporary grassland.

Bird populations were monitored using the BTO's Common Birds Census methodology. This entailed ten visits (mostly in the early morning or, occasionally in the evening) per year, from mid March to the end of June to record bird species, numbers, activity and location within the census plot. The position of each bird seen or heard (a registration) was recorded on a 1:10,000 map of the site.

Beetle trapping Carabid beetle populations were monitored using pitfall traps. Ten traps were sited between the edge of a permanent grass field and a brook with a bank of coarse vegetation which was adjacent to an arable field. Twenty more traps were located in a coarse grass boundary to an arable field. Traps were operated fortnightly between May and November in each year of the study. The total number of individuals caught between May and the end of July, during the main bird breeding season when invertebrates are an important component of the diet, was calculated. Carabid beetle populations were monitored in order to provide a measure of change in invertebrate population density during the study period.

The ECN site at lACR Rothamsted lies 40 Ion NNW of London, near Harpenden, Hertfordshire. The soils here were well drained to moderately well drained flint silty loams. The site, used primarily for agricultural research, comprised mainly

5 File reference: RSC int.ely managed arable land, Protocols used for the Common Birds Census and monitoring of Carabid beetle populations at Rothamsted were exactly the same as those followed at Drayton,

Statistical analyses PopulatiDn density or, perhaps more accurately, bird activity at Drayton was measured as the number of registrations within the CBC plot area, No account was taken of the number of breeding territories which these registrations comprised, The study, therefore, assessed and compared the number of registrations for each species across the plot with respect to factors such as year, field crops present, boundary types and bDundary lengths, where applicable,

A comparison was carried Dut between the numbers of skylark registrations per hectare or yellowhammer registratiDns per 100 metres of field boundary in each year, in order that these data could be compared with the results of territory mapping analyses carried out by the BTO, This allowed a comparison to be made of pDpulatiDn trends to determine whether both methodologies led tD similar conclusions,

A number of statistical techniques were used as explDratDry means of examining the data, Although it is appreciated that the census data on skylarks and yellowhammers are obtained from a treatment structure which is unbalanced and not necessarily randDm, nevertheless statistical tools which are commonplace in the analysis of results from designed experiments were used to provide an insight into the relevant effects Df parameters measured, Consequently the numerical P values obtained are not necessarily rDbust, but should represent an approximate measure of statistical significance, The various techniques are further discussed under the appropriate headings, All computations were carried out using Minitab for Windows (Version II) software,

Skylarks The main method used to analyse skylark data with year and crop as treatment factors was based on a generalised linear approach to the analysis Df variance of unbalanced design, An important consideration as to whether the analysis is valid (i.e. whether the data and model are compatible) was the assessment of the randomness of the residual error abDut zero when residual error was plotted against fitted value,

The treatment effects were year and crop; the registrations being expressed as registrations per hectare and registrations per log hectare (there may be a non-linear relationship between the area of a field and the number of registrations observed).

Unless otherwise indicated the residual plDts were fairly random (some systematic trends exist for data containing zeros) and are adjudged to be indicative of a representative model for the data,

Other statistical techniques used in the analyses of the skylarks data included: Contingency Tables (cDnsidered unsuitable data nDt shDwn), Kruskal-Wallis Tests (a non-parametric test, akin to ANOVA, using scores based on ranking - not bIDcked), and Friedman Test (a nDn-parametric analysis of variance, using scores based on ranking - blocked).

Multiple range tests for the various treatment levels were nDt carried out because of the unbalanced nature of the design. Even in well balanced designs multiple range testing is not without controversy, Analysis of skylark registrations per hectare are shown in the tables without further description.

Yellowhammers The analysis of yellowhammers data followed the same principles used for the skylarks data i.e. a generalised modelling approach was used tD identifY significant sources Df variability.

The treatments used were year, hedge type, crop combination and ditch type and the response variable was registratiDns per 100 metres DfbDundary length and their cDrresponding logrithmically (IoglO) transformed data. To preserve the zero data a value of Dne was added to all values prior to transformation.

Residual plots were assessed as described for skylarks,

The yellowhammer dataset was reduced to a large subset of data in which two boundary types and two crop combinations were Dmitted as these did not substantially contribute to the data. Furthermore, residual plDt analysis revealed a gross Dutlier which was also removed from the dataset.

6 File reference: RSC

Tlabset was analysed and the log transfonned data (which gave a better residual fit) was used in the assessment of the different sources of variability (e.g. ditch or hedge type).

The analysis of variance was conducted on year, hedge type and crop combination all together; on year, hedge type and ditch type all together; and on a combination of all four factors taken together. In the latter case the independent nature of those factors is questionable (i.e. hedge and ditch type are inter-related and not mutually exclusive) and so the analysis of variance for these two factors will be blurred.

Additional statistical tests applied to yellowhammers' data included Principal Components Analysis, which was then considered to be unsuitable (so analyses are not presented in this report); and the Chi Square Test, which was used for an analysis of yellowhammer registrations along boundaries with or without ditches, by year.

RESULTS

Skylarks

Analysis, by the BTO, of the Drayton Common Birds Census data for the period 1993-98 showed a decline in the number of breeding birds using the site, from 21 in 1993 to 12 in 1998 (Table I). Analysis ofCBC data from Rothamsted, for the period 1993-97, also showed a decline, from 23 to 17 (Table 2).

Table 1. Estimated numbers ofskylark territories (i.e. breeding pairs) within eBe plot - Drayton, 1993-98.

1993 21 1994 18 1995 14 1996 14 1997 12 1998 12

Table 2. Estimated numbers ofskylark territories (i.e. breeding pairs) within eBe plot - Rothamsted, 1993-97'.

1994 25 1995 20 1996 19 1997 17

I Analysis of1998 CBC data from Rothamsted not yet completed.

Analysis of variance, using GLM as an exploratory tool, indicated that the total number of skylark registrations per hectare at Drayton did not differ significantly between years (P > 0.05) (Table 3).

7 File reference: RSC

erable 3. Mean numbers ojskylark registrations per ha - Drayton, 1993-1998.

1993 1.57 1994 1.62 1995 1.73 1996 1.55 1997 1.03 1998 1.34

Table 4. Analysis ojvariance table jar skylark registrations per ha (log lO+ J) - Drayton, 1993-98.

Year 5 3.156 3.794 0.759 1.53 0.197 Croo 9 122.610 122.610 13.623 24.47 < 0.001 Error 51 25.293 25.293 0.496 Total 65 151.059

Table 5. Mean numbers ojskylark registrations per ha on various crops at Drayton, 1993-98. Itl~lllllftIJII[i!~:i.::rwHlllriltlllII1111fi Permanent grass 0.12 Winter wheat 1.68 Permanent set-aside 2.37 Rotational set-aside 6.70 Winter oilseed rape 1.10 Soring oilseed raoe 0.73 Winter beans 1.75 Temporary grass 3.50 Maize 0.67 3 year set-aside 1.19

Highly significant differences between crops were observed over the six years of the study (P < 0.001), with numbers of skylark registrations being much higher on rotational set-aside than on other crops (Table 5). The lowest numbers of registrations were consistently recorded on permanent grass fields.

The 'permanent set-aside' in Table 5 was, in fact, a single field containing an atypical mixture of three and five year set aside plots (some drilled with grasses, others with a 'natural regeneration' green cover). The three year set-aside was mainly natural regeneration, but was situated in half ofa relatively small field, which had forage maize in the other half - in an area bounded on one side by a tree lined stream-bank and on two sides by hedges. There were also some cropped trial plots in this area in 1995, the first 'set-aside' year for this field. It is probably a mistake, therefore, to read too much into the relatively low numbers of registrations per ha in these 3 and 5 year set-aside areas.

Both Kruskal-Wallis and Friedman Tests indicated that there were also highly significant differences between the numbers of skylark registrations per ha in different fields over the six year period, irrespective of cropping (P < 0.00 I). Generally, however, this finding will have been heavily influenced by the fact that certain fields, in particular those in pennanent grassland, remained in the same 'crop' throughout the study period. Additional multiple comparisons using Friedman's sum of ranks, considering all six years together, showed significant differences in the number of skylark registrations per ha between certain specified pairs of fields. These results tended to confirm the importance of the cropping factor, discussed

6 File reference: RSC a. indicating that two large fields which were in rotational set-aside for one of the six years (Fields 12 and 13) were preferred to three permanent grass fields (Fields 8, 9 and 13A) (P < 0.05). This analysis also showed that the single 'pemlanent set-aside' field (Field 16) was preferred to two of the grass fields (Fields 8 and 13A) (P < 0.05). Fields 8, 9, 12 and 13 were all relatively large (6-11 hal, open fields; but Field 13A was very small (0.9 hal and adjacent to farm buildings.

Yellowhammers

Analysis, by the STO, of the yellowhanlffier ese registrations at Drayton showed a dramatic fall in breeding numbers in 1995, compared with the previous two years (Table 6). Numbers remained at relatively low levels in subsequent years.

Table 6. Estimated numbers ofyellowhammer territories (i.e. breeding pairs) within CBC plot- Drayton, 1993 98. l\ltlllllllllllllalll 1993 27 1994 32 1995 18 1996 18 1997 20 1998 14

Analysis, by the STO, of the yellowhanlffier ese registrations at Rothamsted showed a dramatic fall in breeding numbers in 1996 compared with the previous two years (Table 7). The population then remained stable at this low level in 1997. Analysed data for 1998 from the Rothamsted site was still not available at the time that this report was written.

Table 7. Estimated numbers ofyellowhammer territories (i.e. breeding pairs) within CBC plot - Rothamsted, 1993-9i. illli'tlllllilflllllbll 1993 II 1994 14 1995 15 1996 8 1997 9

I Analysis of 1998 cac data from Rothamsted not yet completed

Table 8. Mean number ofyellowhammer registrations per 100 m ofboundary - Drayton, 1993-98. 111,lIlllllllfltS'lfil 1993 5.17 1994 4.22 1995 2.86 1996 3.30 1997 3.83 1998 3.24

9 File reference: RSC o.ences in yellowhammer activity at Orayton, as measured by numbers of registrations per 100 m of boundary, between years were statistically significant (P < 0.001), when all four factors were analysed together (Table 8). Thc number of yellowhammer registrations was at its highest in 1993 (5.17 registrations/IOO 01), fell to its lowest level in 1995 and remained at relatively low levels thereafter.

Table 9. Analysis 0/ variance (ANOVA) table lor yellowhammer registrations per 100 m 0/ field boundary (Iog/ o+I) when year, hedge type and crop combinations are compared - Drayton, 1993-98.

Year 5 1.102 1.262 0.252 4.68 0.001 Hedl!e tvpe 4 1.1 58 0.430 0.107 1.99 0.098 Crop combination 8 2.174 2.174 0.272 5.04 <0001 Error 154 8.301 8.301 0.054 Total 171 12.735

Table 10. Analysis o/variance table/or yellowhammer registrations per 100 m o/boundary (log/o+l) when year, hedge type and ditch type are compared - Drayton, 1993-98.

Year 5 U93 0.886 0.177 2.66 0.024 Hedl!e tvpe 6 1.345 1.091 0.182 2.73 0.015 Ditch type 3 0.358 0.358 0.119 1.79 0.150 Error 174 11.587 11.587 0.067 Total 188 \4.484

Table 11. Analysis 0/ variance table lor yellowhammer registrations per 100 m 0/ boundary (Iog/ o+I) when year, hedge type. ditch type and crop combinations are compared - Drayton. 1993-98.

Year 5 1.102 1.320 0.264 5.13 <0.001 Hedl!e type 4 U58 0.129 0.032 0.63 0.644 Ditch tvpe 3 0.212 0.523 0.174 3.39 0.020 Crop combination 8 2.485 2.485 0.311 6.03 <0.00 I Error 151 7.778 7.778 0.052 Total 171 12.735

Analysis of variance showed that field boundary type also had a significant, though less strong, effect on the number of yellowhammer registrations per 100 01 length (P < 0.05) - with relatively good numbers alongside hedged boundaries and streamsides with mature coppiced willows and very few being recorded along fence-lines. However, re-analysis of the data after simplification of the boundary type categories (removing minor categories and the poorly favoured fence-lines), iadicated that hedge type did not significantly affect the number of yellowhammers per 100 metres of boundary (P > 0.05), when all other factors were analysed together (Table 12).

10 File reference: RSC

.Table 12. Mean number ofyellowhammer registrations per 100 m offield boundary, classified by different hedge types - Drayton, 1993-98.

Short dense rectangular hedge 3.15 Short, moderately dense rectangular hedge 4.10 Short, low density, rectangular hedge 3.69 Remnant hedge 4.69 Shrubby growth and trees 3.22

The number of yellowhammers was affected by the combination of crops adjacent to the field boundaries in which they were recorded (Table 13). Highly significant differences (P < 0.00 I) were found between the crop combinations present. The highest numbers of registrations were recorded where rotational set-aside was present in combination with a cereal crop. The lowest numbers were recorded where grass was present on both sides of the fIeld boundary.

Table 13. Mean number of yellowhammer registrations per 100 m of field boundary, classified by different combinations ofadjacent crops - Drayton, 1993-98. I~B_'I"1111111IlI11'lt'lllll' CereaVcereal 4.48 CereaVother arable croc 4.79 Cereal/rotational set-aside 5.20 CereaVother set-aside 4.45 CereaVgrass 2.5\ Other arable croc/grass 4.51 Rotational set-aside/grass 4.28 Other set-aside/grass 2.00 Grass/grass 1.72

The presence or absence of a ditch and a strip of unmanaged vegetation significantly affected the populations of yellowhammers recorded (P < 0.05) (Table 14). The highest number of yellowhammer registrations occurred where a ditch was present along with an unmanaged grass strip greater than \.5 metres in width.

Table \4. Yellowhammer registrations per 100 m offield boundary, classified by different ditch typesljield margin grass strip - Drayton, 1993-98.

No ditch or vegetation strip 3.29 Ditch with <1.5 m unmanaged grass strip 3.44 Ditch with >1.5 m unmanaged grass strip 4.60 Ditch with no adjacent unmanaged grass strip 3.75

Analysis, using the Chi Square Test, of the relative number of yellowhammer registrations per 100 m length, along boundaries (at Drayton) with or without ditches, in each of the years 1993-97, showed no significant differences (P = 0.621) - indicating that yellowhammers showed no stronger preference for boundaries with ditches in any particular year (e.g. in dry years).

11 File reference: RSC

Caid beetles

The total number of Carabid beetles trapped at Drayton each year, between the beginning of May and the end of July, varied markedly (Table 15). Numbers fell in 1994 (-34%) and 1995 (-56%) and remained low in 1996 before recovering dranmtically in 1997 (+175%). Numbers fell again in 1998 (-36%).

Table 15. Toral number ofCarabid beetles trapped and change in numbers from previous year - Drayton, 1993 98.

1993 1599 Not known 1994 1057 -542 -34% 1995 470 -587 -56% 1996 514 +44 +9% 1997 1413 +899 +175% 1998 907 -506 -36%

Table 16. Annual changes in total Carabid beetle numbers and ye//owhammer registrations per 100 m of boundaryfrom previous year, and subsequent annual changes in ye//owhammer registrations - Drayton, 1993-98. ••••

Table 17. Annual changes in total Carabid beetle numbers and ye//owhammer territories (i.e. breeding pairs) from previous year, and subsequent annual changes in ye//owhammer territories - Drayton, 1994-98.

-542 -34% +5 -14 -44% 1995 -587 -56% -14 -44% o 0% 1996 +44 +9% o 0% +2 +11% 1997 +899 +175% +2 +]\% -6 -30% 1998 -506 -36% -6 -30% Not known

12 File reference: RSC

.Table 18. Total number of Carabid beetles trapped and change in numbers from previous year - Rothamsted. 1993-97

1993 4446 Not known 1994 3074 -1372 -31% 1995 4218 +1144 +37% 1996 2206 -2012 -47% 1997 7649 +5443 +247%

As at Drayton, the number of Carabid beetles trapped at Rothamsted varied greatly from year to year, although numbers were generally much higher (Table 18). There were marked declines in 1994 (-31%) and 1996 (-47%), but big increases again in 1995 (+37%) and 1997 (+247%). As at Drayton, \993 and 1997 were the two best years for Carabid captures, and the greatest annual increase was recorded between 1996 and 1997.

Table 19. Annual changes in total Carabid beetle numbers and estimated yellowhammer territories (i.e. breeding pai") from previous year. and subsequent annual changes in yellowhammer territories - Rothamsted. 1994-97.

1994 -1372 -31% +3 +27% -I +7% 1995 +1144 +37% -I +7% -7 -47% 1996 -2012 -48% -7 -47% +1 +13% 1997 +5443 +247% +1 +13% Notknown

Agricultural management

The most relevant aspects of agricultural management with respect to potential impacts on skylark and yellowhammcr populations at Drayton are probably a) insecticide inputs on wheat crops and b) the date of cutting or glyphosate application on rotational set-aside fields. These details are summarised in Table 20 below.

Other 'routine operations' that may have a bearing on the bird registrations data described in the preceding tables, include thc following: o Grassland - livestock over-wintered indoors; significant proportion of fields cut and baled for silage between mid-May and early June. o Winter wheat - fields cultivated soon after August harvest, then new crop usually drilled in first two weeks of October (only slightly late drilling in study period was 1994/95 crop - drilled 19-22 October 1994); herbicides applied in autunm and again in spring; fungicides in late spring (May); no routine insecticides. o Spring rape - broad spectrum insecticide (e.g. deltamethrin) applied in June for control of pollen beetle.

13 File reference: RSC

.Table 20. Dates of insecticide applications to winter wheat and cutting/spraying off (glyphosate application) of rotational set-asidejields - Drayton, 1993-98.

Winter wheat Insecticide None Chlor- Delta- None None None pyrifos' methrin' 13 June 29 June

Rot. set-aside' Cut 10 May 6 May NoRSA Spray 16 June 20 May 27 April NoRSA 10May

,I Chlorpynfos

Meteorological conditions

Drayton experienced a particularly hot and dry summer ID 1995 (Table 21). 1996 was also very dry, but mean temperatures were lower.

Table 21. Mean air temperature and total rainfall at Drayton, I May to 31 July in each year ofthe bird study.

1993 14.4 187.2 1994 15.0 Ill.7 1995 15.2 70.0 1996 13.9 74.5 1997 14.4 215.2 1998 14.6 145.2

DISCUSSION

In this study analysis of bird census registrations data was used to determine the overall activity of two bird species, skylarks and yellowhammers, during the breeding season and how this changed within an intensively farmed area over a period of six years. In addition, the effects of changing cropping patterns and the characteristics of the field boundaries present were compared with the changes in the degree of activity recorded. An indication of changes in invertebrate numbers and, hence, bird food availability was obtained from the numbers of Carabid beetles trapped during the bird breeding season.

Skylarks

Although there was no statistically significant difference in the number of registrations of skylarks recorded over the six years of the present study, the analyses of the number of breeding territories by the BTO has shown a steady decline from 21 breeding territories in 1993 to 12 territories in 1997 and 1998 at ADAS Drayton. A similar decline was also recorded at the ECN site at IACR Rothamsted. These population trends are in general agreement with national trends for this species, as determined from the BTO's national Common Birds Census.

14 File reference: RSC reations per hectare, compared with all other crops grown in the study period. This effect is in agreement with the findings of a number of other studies. For example, Henderson el of. (1998), in a recent survey for MAFF, found that skylarks showed a very strong preference for set-aside, particularly rotational set-aside with a sparse vegetation structure (i.e. where there was a fairly high proportion of bare ground or straw/litter).

Wilson et 01. (1997) found higher skylark densities on set-aside and in organically cropped fields, and also recorded improved breeding success on set-aside, in comparison with intensively managed cereals.

GI)'nderup Poulsen et 01. (1998) also found that set-aside, and permanent grass, were preferred habitats for skylarks on predominantly arable farmland, with territory densities 2-3 times higher than on cereals. They also found that skylarks had relatively small territories (I.7 hal on set-aside and large ones on winter cereal crops (4.5 hal. Furthermore, Grynderup Poulsen et 01. (1998) found evidence that the density of successful nests was much higher on set-aside than on arable crops (more than double) and the average clutch size at hatching larger than in silage grass or spring barley. Overall, although fledgling suecess did not differ according to crop type, the other factors mentioned all combined to give a better brceding 'productivity' on set-aside (0.50 fledglingslha, compared to 0.13 in silage crops, 0.21 in spring barley and 0.0 in winter cereals).

The reasons for the apparent attractiveness of rotational set-aside to skylarks have been given as the development of a varied flora, which may result in increased numbers of invertebrates and the presence of a sparse vegetation layer allowing easy access to the ground or soil litter. Henderson et 01. (1998) have suggested that skylarks, and other bird species with relatively diverse dietary requirements that show a prererence for rotational set-aside, are probably attracted by both an increased availability of invertebrate food (and they had some evidence that set-aside fields in their survey had highcr dcnsities of ground-dwelling invertebrates than did adjacent arable crops) and enhanced amounts of weed seeds from indigenous plants.

Previous studies have also shown the importance of winter stubbles to skylarks. With winter-sown crops having largely superseded spring crops throughout England, and the consequent demise of spring cultivation, compulsory rotational set aside has ensured the large-scale return of winter stubbles.

The potential loss of most of the rotational set-aside area in Europe within three years, with a proposed 0% compulsory set aside rate in 200 I under a reformed Common Agricultural Policy, seems likely to seriously disadvantage bird species such as the skylark.

Skylark populations appeared to rapidly move to occupy areas of rotational set-aside, when present. The attractiveness of rotational set-aside compared with other crops can be gauged by the fact that, in the three years when rotational set-aside was not present in the plot area (1993, 1996 and 1997), the mean number of skylark registrations per ha on all 'crops' (including grass and non-rotational set-aside) was 1.38. This compares with only 0.93 registrations per ha on all other crops in the three years when rotational set-aside was present (1994, 1995 and 1998). The mean number of registrations per ha on the rotational set-aside in these three years was 6.70.

Other types of set-aside monitored during the period of this study were three-year and five-year set-aside. Much of this area was within a long-term 'permanent set-aside' experiment, with a large proportion sown with a grass covcr crop probably making it less attractive to skylarks than the areas of rotational set-aside, on which natural regeneration of vegctation took place. The density of the grass cover would not allow easy foraging for skylarks and would provide fcwer suitable nesting sites. As important here, however, is the fact, mentioned previously in the Results section, that the other area of 'three-year set-aside' at Drayton, between 1995 and 1997, was in a small area, enclosed by tall field boundaries and an adjacent maize crop. Skylarks are well known to prefer open habitats. For example, a previous study by Wilson et 01. (1997) found that skylark densities were "lowest on fields surrounded by tal! boundary structures or unsuitable habitat, and those with tal!, dense vegetation cover."

With this point in mind, it is worth noting that the rotational set-aside areas at Drayton in 1994 and 1995 were both in relatively large (6.87 and 8.63 ha respectively), open fields (rectangular with only low hedges or fence-lines surrounding them); whereas the rotational set-aside field in 1998 was smaller (4.78 hal and much more enclosed (with one very tall boundaI)'). The respective numbers of skylark registrations per ha in 1994, 1995 and 1998 were 7.13, 7.42 and 4.79. It would seem, therefore, that positioning of rotational set-aside may be important if the benefits to skylarks are to be maximised.

15 File reference: RSC

A small area of temporary grass occupied part of the field in which pennanent set-aside was present during the first two years• of the study. The adjacent pennanent set-aside is likely to have affected the results for this crop, resulting in a higher apparent number of registrations than might be expected if the crop was in a different location.

Little difference in the number of registrations on arable crops, excluding rotational set-aside, was recorded during the period of the study. However, higher numbers were recorded on winter beans and winter wheat than on other arable crops. Fewer were present on oilseed rape with little difference between autumn sowing and spring sowing of this crop. Maize, actually a fodder crop though grown in a similar manner to arable crops, had the lowest number of registrations associated with it.

Yellowhammers

Yellowhammers are largely granivorous during the autumn and winter (feeding on seeds of cultivated cereals, wild grasses and weeds), but invertebrates are a major component of the diet of the adult birds in the summer. The diet of nestlings is dominated by invertebrates (especially spiders, beetles, springtails, sawflies, larval Lepidoptera and grasshoppers), though semi-ripe grain is also consumed (Morris el al., in press). During the nesting season, therefore, adult birds need to be near habitats in which they can obtain invertebrates with which to feed young as well as themselves.

The number of breeding territories detennined by the BTO analysis of CBC data from Drayton declined from 32 in 1994 to 14 in 1998. A particularly large decline occurred in 1995. At IACR Rothamsted numbers also declined, from 14 in 1994 to 9 in 1997 (1998 data was not available). The largest decline occurred in 1996. When the yellowhammer populations were measured in tenns of the number of registrations per 100 metres of field boundary a severe decline was again noticeable at Drayton

Yellowhanlffiers nest most often in shrubby growth such as hedges, or on ditch banks (Bradbury el aI., in press). There is plentiful evidence in the literature that yellowhammers generally prefer relatively short hedges « 1.5 m), with few or no hedgerow trees. However, analysis of the number of Drayton yellowhammer registrations obtained in hedges with different densities and different hedge types indicated that there were no statistically significant differences between them. Additional data regarding the numbers of registrations on two other field boundary types were not included in the analysis as the number of observations was too small to allow comparisons with the hedge types. These were wire fences, where hardly any registrations occurred, and a limited area which contained pol1arded trees bordering a brook. The brook had a coarse grass bank which was adjacent to an arable field. In keeping with the results found during the analyses of ditch type data, the number of registrations on this boundary type was high.

The presence or absence of a ditch or unmanaged grass strip immediately adjacent to the field boundary significantly affected the number of yellowhammer registrations. The greatest number of registrations were recorded where a ditch and an unmanaged grass strip greater than 1.5 metres in width was present and the lowest number occurred where there was no ditch or vegetated strip. This result emphasises the importance of uncropped margins of fields for foraging in bird species such as the yellowhammer.

These results are very similar to those from a recent study by researchers from Oxford University and the RSPB (Bradbury el al. in press). They also found higher densities of yellowhammers along boundaries with ditches and reported that densities increased with the width of any uncultivated grassy margin. They concluded that filling or clearing ditches, and cropping or grazing right up to field boundaries were practices that would be adverse to yellowhammers.

The number of registrations per 100 metres of boundary, where various combinations of cropping occurred on each side of the field boundary, indicated a negative effect of grassland. This finding, i.e. the general avoidance of intensively managed grass, supports the evidence from several previous studies (e.g. Kyrkos el aI., 1998; Morris el al., in press; and Bradbury el al., in press). Bradbury el al. (in press) report that the lowest densities of yellowhammers in their study were found next to pasture and grass leys. Green el al. (1994) found that yellowharnmers (like greenfinches) showed a preference for hedgerows between two arable fields over those between two grass fields. They also preferred arable/road combinations over grass/road.

The lowest numbers were recorded where grass was present on both sides of the boundary, with numbers also being reduced when grass was present in combination with cereals or 'other' set-aside. 'Other' set-aside at this site was

16 File reference: RSC c.osed mainly of a sown grass cover crop and, therefore, although not so intensively managed as grazed grassland, would not be expected to be significantly more attractive to yellowhammers. Numbers of registrations were relatively high when grass was in combination with the category of 'other arable crops'. Most of the grassland on the site is intensively managed, either for grazing or for silage production. This degree of management is likely to reduce the food resource available to foraging yellowhammers due to the limited range of plant species present, high levels of fertiliser usage and consequently low populations of invertebrates. High densities of yellowhammers are known to use unimproved semi natural grassland where the invertebrate populations are mueh higher.

Rotational set-aside, when combined with a cereal crop, resulted in the highest numbers of registrations. The attraction of rotational set-aside to foraging birds has been shown in a number of studies where increased invertebrate densities have been shown to be of importance. When rotational set-aside occurred in combination with grass the number of registrations along the boundary was reduced to a similar level to that obtained when grass and 'other arable crops' were combined. Similar levels were also obtained when cereals were combined with cereals, 'other arable crops' or 'other' set-aside.

Carabid beetles

The numbers of Carabid beetles are given to provide a measure of invertebrate population change over the period of the study. Invertebrates are a major component of the diet of skylarks and yellowhammers during the period May to August. Carabid beetles are taken in large numbers by skylarks in summer. Beetles and other invertebrates also comprise a major component of the diet of adult yellowhammers during the summer as well as being the majority of the diet of nestlings.

During the period of this study Carabid beetle numbers declined in 1994 and 1995 at Drayton, remained at low levels in 1996, before recovering in 1997, and declined again in 1998. The largest decline occurred in 1995 and this was probably due to the very hot, dry summer in that year leading to poor survival of soil inhabiting beetle larvae in the heavy clay soils of the site, and consequently low numbers of adults. The lack of a recovery in populations in 1996 was also probably caused by a severe drought in that year. At Rothamsted Carabid beetle populations declined most severely in 1996.

Population changes in Carabid beetles are closely linked with the observed changes in yellowhammer populations during the same year, particularly when measured as registrations per 100 metres of field boundary. The lower beetle numbers recorded in 1995 coincide with the dramatic reduction in yellowhammer numbers. These results suggest that the adult yellowhammer population may have been affected by a lack of food. If nestling survival only had been affected then population declines would be experienced in the year following the reduction in food availability.

The close correlation between Carabid beetle and yellowhammer numbers does not appear to exist with Carabid beetles and skylarks. The breeding numbers of skylarks has declined at a more steady rate over the six years of the study than is the case with yellowhammers and the decline continued even when Carabid numbers increased. This might suggest that skylarks are more prone to effects not linked with food, supply such as land use change, whereas yellowhammers are protected from these effects to some extent by their choice of nesting sites in field boundaries and their greater aptitude for foraging in uncropped areas. Alternatively, the habitats in which pitfall trapping for Carabid beetles was carried out may not have reflected sufficiently closely the range of invertebrates available to skylarks in the more open field situation. Invertebrate groups less affected by the conditions which caused beetle populations to decline (e.g. Diptera) may have formed a greater proportion of their diet at this time.

The pitfall trapping sites for Carabid beetles were in uncropped field margins composed mainly of coarse grass vegetation. This habitat appears to be the preferred one for foraging yellowhammers and, therefore, the general agreement between population changes in yellowhammers and Carabid beetles indicates that Carabid beetle numbers were a good indicator of invertebrate prey abundance in this habitat.

CONCLUSIONS AND CONSERVATION IMPLICATIONS

The ability of skylarks to rapidly make use of rotational set-aside land within an otherwise intensively managed agricultural situation has been shown by this study. These results show the importance of the correct management of rotational set aside and similar schemes and that account continues to be taken of the fact that populations of some fauna are attracted to it at critical stages in their life-cycle. Poor management decisions on set-aside would, potentially, have a far more damaging effect on skylark populations than would be the case on an equivalent area of most other crops.

17 File reference: RSC

_ared with rotational set-aside, grassland at the Drayton site was very unsuitable for skylarks. The grassland was very intensively utilised either by grazing livestock or for silage making which makes it poor as a potential nesting site and also as a food resource. Land occupied by longer term set-aside sown with a grass cover crop is not as attractive to skylarks as rotational set-aside, though numbers present were higher than on most other crops. In order to make grassland more suitable for use by skylarks a significant reduction in the intensity of its management would be required.

Success in the maintenance of yellowhammer populations appears to be linked with cropping, field boundary management and invertebrate density. Intensively managed grassland is not favoured by yellowharnmers, probably due to the lack of easily available food. Managing field boundaries to include a relatively unmanaged coarse grass area, prefcrable of at least 1.5 metres in width, appears to increase the amount of yellowharnmer activity. This may be due to the availability of invertebrate food during the nesting season. Where the amount of invertebrate food in such field margins has declined, albeit probably due to climatic change in this study, this has had a rapid deleterious effect on yellowharnmer populations. Unlike the Carabid populations, the yellowhammers (like all bird species) do not have the capacity for very rapid population recovery following years of heavy losses.

The importance of invertebrate population levels apppears to have been shown at Drayton in 1995, when both Carabid bcetle and yellowhammer numbers were severely reduced. Although, in this instance, the declines were probably climatically related the intensification offarmland management and consequent reduction in suitable foraging areas is likely to have similar effects over a longer timescale and it may be the effects of these habitat related changes which are producing the declines in some farmland bird populations being recorded in this and other studies.

In this study the unmanaged grass strips were adjacent to fields containing arable cropping. The results show that intensively managed grassland is less suitable for yellowharnmers than arable cropping. Further examination of the effects of establishing 'unutilised' margins to intensively managed grass fields on bird and invertebrate populations is necessary. If similar effects are forthcoming using margins on grassland, as occur in arable situations then the beneficial effects for farmland birds are likely to be substantial. This could be of particular importance in the South-west of England, where the intensification of grassland management is thought to have had a particularly deleterious effect on birds such as the yellowhanuner.

This study has highlighted the need for the habitat requirements of individual species, or groups of species, to be taken into account when considering management options to encourage biodiversity or when attempting to increase the population of a particular species. In this exercise skylark population activity was seen to be increased by the provision of relatively large, field scale, areas of set-aside land. The presence of more limited areas of set-aside in, say, the field boundaries, is less likely to produce this effect. On the other hand, these methods, for example creating coarse grassy margins to arable fields, would greatly benefit populations of farmland birds similar in habit to yellowharnmers.

Long-term monitoring of a number of variables, such as that carried out by the Environmental Change Network, allows an increase in the possibility that the causal mechanisms of change in one variable can be determined by examination of the data gathered for another. When comparisons can be made between sites across such a network the understanding of such mechanisms and the certainty with which results can be expressed is greatly increased.

ACKNOWLEDGEMENTS

This project was funded by the Ministry of Agriculture, Fisheries and Food. Thanks are also due to the Defence Establishment Research Agency (DERA) for allowing Clive Bealey to participate in this work. The authors gratefully acknowledge the assistance of Tim Pepper, and other colleagues at ADAS Gleadthorpe, with the various statistical analyses.

REFERENCES

Bealey, C; Howells, 0 & Parr, T (1998). Environmental change and its effects on wildlife: the role of the Environmental Change Network. Brilish Wildlife n. 341-347. Bradbury, R B; Kyrkos, A; Morris, A J; Clark, S C; Perkins, A J & Wilson, J D (in press). Habitat selection, breeding success and demographic predictions for yellowhammers Emberiza cilrinella on lowland farmland. Gillings, S & Fuller, R J (1998). Changes in bird populations on sample lowland English farms in relation to loss of hedgerows and other non-crop habitats. Oecologia 116. 120-127.

18 File reference: RSC

6, R E; Osborne, P E & Sears, E J (1994). The distribution of passerine birds in hedgerows during the breeding season in relation to characteristics of the hedgerow and adjacent fannland. Journal ofApplied Ecology 31. 677-692. Grynderup Poulsen, J; Sotherton, N W & Aebischer, N J (1998). Comparative nesting and feeding ecology of skylarks Alauda arvensis on arable fannland in southern England with special reference to set-aside. Journal of Applied Ecology 35. 131-147. Henderson, I G; Cooper, J & Fuller, R J (1998). Spatial and temporal utilisation of set-aside by birds in summer on arable fanns in England. In Agronomic and Environmental Evaluation of Sel-aside under the Ee Arable Area Payments Scheme (Ed. L G Firbank), Vol. 4 Birds & Modelling. Centre for Ecology and Hydrology/AOAS/British Trust for Ornithology report to Ministry of Agriculture, Fisheries and Food. May 1998. ITE, Grange-over-Sands, Cumbria. Kyrkos, A; Wilson, J 0 & Fuller, R J (1998). Fannland habitat change and abundance of Yellowhammers Emberiza citrinella: an analysis of Common Birds Census data. Bird Study 45. 232-246. Moles, R T & Breen, J (1995). Long-tenn change within lowland fannland bird communities in relation to field boundary attributes. Biology and Environment: Proceedings ofthe Royal Irish Academy Vol 95B, No.3. 203-215. Morris, A J; Whittingham, M J; Bradbury, R B; Wilson, J 0; Kyrkos, A; Buckingham, 0 L & Evans, A 0 (in press). Habitat selection by foraging yeliowhammers on lowland fannland during the breeding season. Parish, T; Lakhani, K H & Sparks, T H (1994). Modelling the relationship between bird population variables and hedgerow and other field margin attributes. I. Species richness of winter, summer and breeding birds. .Jollrnal of Applied Ecology 31. 764-775. Sparks, T H; Parish, T & Hinsley, S A (1996). Breeding birds in field boundaries in an agricultural landscape. Agriculture, Ecosystems and Environment 60. 1-8. Wilson, J 0; Evans, J; Browne, S J & King, J R (1997). Territory distribution and breeding success of skylarks Alauda arvensis on organic and intensive fannland in southern England. Journal ojApplied Ecology 34. 1462-1478.

19 DRAFT A study of farmland bird populations: an analysis of skylark (A/auda • arvensis) and yellowhammer (Emberiza citrinella) registrations data from common birds censuses at ADAS Drayton, 1993-98

S J Corbett; D N Wilson; C P Britt; D Allen (all ADAS) and C Bealey (CBD/DERA, Porion Down)

Summary

This study examined population changes/activity levels in two farmland bird species (skylark and yellowhanuner) over a six year period, at ADAS Drayton - a lowland mixed farm in south Warwickshire - using registrations data from annual Common Birds Censuses, undertaken as part of the UK Environmental Change Network (ECN) project. Results were also compared with data obtained from a second lowland agricultural ECN site, at the Institute of Arable Crops Research at Rothamsted, in Hertfordshire.

National populations of both species have declined in recent years and this study, as well as determining whether similar declines occurred at Drayton, also attempted to determine the effects of cropping and the management of non-cropped areas on bird activity. Carabid beetle activity was also monitored each year (within the ECN project), and the numbers of beetles captured annually were compared with skylark and yellowhanuner registrations, to examine any possible correlations between these groups.

Yellowhanuner populations were affected by the management of cropped and uncropped areas of farmland. Comparisons of various cropping combinations occurring each side of a field boundary indicated a positive effect for rotational set-aside and a negative effect for intensively managed grassland.

Although the type of hedge forming the boundary, in terms of its density, did not affect the number of registrations significantly the presence of a ditch and an unmanaged grassy area more than 1.5 metres wide adjacent to the crop did result in significantly higher numbers of registrations. No grass fields contained a ditch with an unmanaged grass strip of greater than 1.5 metres wide and therefore no valid comparisons between effects of field margin type on the two major cropping regimes (arable vs. grassland) could be made. DRAFT Yellowhammer population declines occurred when Carabid beetle numbers in field margins fell, indicating that these beetles may be a good indicator of prey availability for • yellowhammers, and that uncropped field margins are an important habitat for foraging yellowhammers.

The beneficial effects of field margin management for wildlife in arable situations has been examined in a number ofstudies. There is an urgent need to ascertain whether similar benefits would be obtained from alternative management of intensively managed grass field margins, especially in the western half of the country, where the intensity of grassland management has increased most, and is likely to have had the greatest affect on species such as the yellowhammer.

INTRODUCTION

The populations of a number of farmland bird species have declined in recent years. These declines have coincided with a period of increased intensification of agricultural production.

The Environmental Change Network (ECN) project is carried out on a number of terrestrial and freshwater sites throughout Britain. It is a long-term integrated monitoring programme designed to identifY and improve the understanding of the causes of environmental change. The site at ADAS Drayton is sponsored by the Ministry of Agriculture, Fisheries and Food (MAFF).

A wide range of physical, chemical and biological variables are measured by the ECN. One of the variables which has been measured since 1993 at Drayton is the population of breeding birds. During this time changes in populations of particular species have been recorded and the changes agree, in general, with those determined in the national Common Birds Census (CBC) carried out by the British Trust for Ornithology (BTO). Another variable being measured is the number and species of Carabid beetles present at the site. Changes in the population levels of these insects have also been noted during the period 1993 to 1998. The site is one on which agricultural research has been carried out for many years and, therefore, detailed information about the agricultural management of the site is available, enabling populations changes to be measured against changes in management practices.

Skylarks (Alallda arven~is) and yellowhammers (Emberiza cilrinella) are two of the species of farmland bird which have experienced large population declines in recent years and have been the subject of much recent research (e.g. Grynderup Poulson el al., 1998; Wilson el aI., 1997; Kyrkos el aI., 1998; Sparks el al., 1996; Green el aI., 1994). Nationally, skylark populations have been recorded as declining by almost 60% since 1970 in BTO studies.

Yellowhammer populations were relatively stable during the late 1960's and 1970's, while other farmland birds declined; but populations started to decrease in the late 1980's (Bradbury el aI., in press), with a rapid increase in the rate of decline in recent years. Currently, yellowhammer populations are declining by around 10% per year (Bradbury el al., in press).

The two species occupy different areas of farmland. Skylarks live in more open habitats, nesting on the ground amongst crops. Yellowhammers, on the other hand, nest along field boundaries, either on the ground, in ditches or grassy margins, or in shrubby growth such as hedges. It feeds mainly in nearby adjacent habitats.

2 DRAFT Various possible causes of decline in British yellowhammer populations have been suggested by • previous authors (Morris el at., in press). These include: I. The shift, in western Britain, toward grass monocultures, which harbour smaller invertebrate prey populations and invertebrate diversity (e.g. in the Orthoptera and Staphylinidae) 2. A resultant reduction in cereal grain availability. 3. Loss of over-winter fallow; with less spilt grain, weed seeds and invertebrates available for food in the winter and in the following breeding season. 4. Loss of non-cropped boundary features, which are important as nest sites and foraging habitats. 5. Increased use of pesticides, which has adversely affected many important food groups.

This study examined the changes in populations of skylarks and yellowhammers during the six years in which a census has been carried out at ADAS Drayton. These data were analysed with respect to changes in agricultural management during the period to ascertain whether particular practices were linked with the recorded population changes. Changes in the bird populations were also compared with those of Carabid beetles in order to determine any relationship between changing prey availability and population trends.

Changes in bird populations (as measured by the BTO territory analysis) and Carabid beetle numbers during the same period were also compared with those obtained at IACR Rothamsted, near Harpenden, Hertfordshire - another ECN site.

MATERIALS AND METHODS

Study site

ADAS Drayton is an agricultural research site of 192 hectares, owned by MAFF. It is located 5 km to the west of Stratford-upon Avon in Warwickshire.

During the period of the study approximately 80% of the farm grew arable crops or short term rye-grass (Lo/ium perenne) leys, the rest of the area being mainly permanent grass. The soils of the site are all heavy calcareous clays, which restrict arable cropping primarily to winter wheat with break crops offield beans and, occasionally, oilseed rape.

Permanent grass fields and temporary leys were managed intensively - being either grazed by cattle or sheep, at relatively high stocking densities, or used for silage production.

Field boundaries were typical of the area, being mainly short, hawthorn (Cralaegus monogyna) dominant hedgerows. The majority of hedgerows were trimmed annually to a square or rectangular shape, approximately I. 5 m high and 1-1. 5 m wide; but the density of woody vegetation present varied.

3 DRAFT There were very few mature trees present on the site, mostly of ash (Fraxinus excelsior) or oak • (Quercus rubur), and no woodland within the esc plot. However, one large field immediately adjacent to the esc plot (and within the Drayton farm boundaries) contained large numbers of young woodland trees, planted in experimental plots throughout the period between 1988 and 1995, and there was an area of more mature woodland just beyond the northern boundary of Drayton. In another field immediately adjacent to the esc plot, approximately one hectare of high density, short rotation willow coppice was established in 1996.

4 DRAFT Methods • Birds census The area of Drayton used as the bird census 'plot' in this study comprised an area of 62 hectares in the centre of the farm. This CBC plot included most of the crop/habitat types present on the farm as a whole, but with a smaller proportion of temporary grassland.

Bird populations were monitored using the BTO's Common Birds Census methodology. This entailed ten visits (mostly in the early morning or, occasionally in the evening) per year, from mid March to the end of June to record bird species, numbers, activity and location within the census i . The position of each bird seen or heard (a registration) was recorded on a I: 10,000 map of the'/ site.

The ECN site at IACR Rothamsted lies 40 km NNW of London, near Harpenden, Hertfordshire. The soils here were well drained to moderately well drained flint silty loams. The site, used primarily for agricultural research, comprised mainly intensively managed arable land. Protocols used for the Common Birds Census and monitoring of Carabid beetle populations at Rothamsted were exactly the same as those followed at Drayton.

Statistical analyses Population density or, perhaps more accurately, bird activity at Drayton was measured as the number of registrations within the CBC plot area. No account was taken of the number of breeding territories which these registrations comprised. The study, therefore, assessed and compared the number of registrations for each species across the plot with respect to factors such as year, field crops present, boundary types and boundary lengths, where applicable.

A comparison was carried out between the numbers of skylark registrations per hectare or yellowhammer registrations per 100 metres offield boundary in each year, in order that these data could be compared with the results of territory mapping analyses carried out by the BTO. This allowed a comparison to be made of population trends to determine whether both methodologies led to similar conclusions.

A number of statistical techniques were used as exploratory means of examining the data. Although it is appreciated that the census data on skylarks and yellowhammers are obtained from a treatment structure which is unbalanced and not necessarily random, nevertheless statistical tools which are commonplace in the analysis of results from designed experiments were used to provide an insight into the relevant effects of parameters measured. Consequently the numerical P values obtained are not necessarily robust, but should represent an approximate measure of statistical signifIcance. The various techniques are further discussed under the appropriate headings. All computations were carried out using Minitab for Windows (Version 11) software.

5 DRAFT Skylarks The main method used to analyse skylark data with year and crop as treatment factors was based • on a generalised linear approach to the analysis of variance of unbalanced design. An imponant consideration as to whether the analysis is valid (i.e. whether the data and model are compatible) was the assessment of the randomness of the residual error about zero when residual error was plotted against fitted value.

Unless otherwise indicated the residual plots were fairly random (some systematic trends exist for data containing zeros) and are adjudged to be indicative of a representative model for the data.

Other statistical techniques used in the analyses of the skylarks data included: Contingency Tables (considered unsuitable - data not shown), Kruskal-Wallis Tests (a non-parametric test, akin to ANOVA, using scores based on ranking - not blocked), and Friedman Test (a non parametric analysis of variance, using scores based on ranking - blocked).

Multiple range tests for the various treatment levels were not carried out because of the unbalanced nature of the design. Even in well balanced designs multiple range testing is not without controversy. Analysis of skylark registrations per hectare are shown in the tables without funher description.

Yellowhammers The analysis of yellowhammers data followed the same principles used for the skylarks data i.e. a generalised modelling approach was used to identifY signifIcant sources of variability.

The treatments used were year, hedge type, crop combination and ditch type and the response variable was registrations per 100 metres of boundary length and their corresponding

logrithmically (log lO) transformed data. To preserve the zero data a value of one was added to all values prior to transformation.

Residual plots were assessed as described for skylarks.

The yellowhammer dataset was reduced to a large subset of data in which two boundary types and two crop combinations were omitted as these did not substantially contribute to the data. Funhermore, residual plot analysis revealed a gross outlier which was also removed from the dataset.

This subset was analysed and the log transformed data (which gave a better residual fit) was used in the assessment of the different sources of variability (e.g. ditch or hedge type).

The analysis ofvariance was conducted on year, hedge type and crop combination all together; on year, hedge type and ditch type all together; and on a combination of all four factors taken together. In the latter case the independent nature of those factors is questionable (i.e. hedge and ditch type are inter-related and not mutually exclusive) and so the analysis of variance for these two factors will be blurred.

Additional statistical tests applied to yellowhammers' data included Principal Components Analysis, which was then considered to be unsuitable (so analyses are not presented in this

6 DRAFT report); and the Chi Square Test, which was used for an analysis of yellowhammer registrations • along boundaries with or without ditches, by year.

RESULTS

Skylarks

Analysis, by the BTO, of the Drayton Common Birds Census data for the period 1993-98 showed a decline in the number of breeding birds using the site, from 21 in 1993 to 12 in 1998 (Table I). Analysis of CBC data from Rothamsted, for the period 1993-97, also showed a decline, from 23 to 17 (Table 2).

Table 1. Estimated numbers ofskylark territories (i.e. breeding pairs) within eBe plot Drayton, 1993-98.

1993 21 1994 18 1995 14 1996 14 1997 12 1998 12

Table 2. Estimated numbers ofskylark territories (i.e. breeding pairs) within eBe plot Rothamsted, 1993-971. l1illlltlllll{fl,"tIIIIIIIJ

1993 23 1994 25 1995 20 1996 19 1997 17

I Analysis of 1998 eBe data from Rothamsted not yet completed.

7 DRAFT Analysis ofvariance, using GLM as an exploratory tool, indicated that the total number of skylark • registrations per hectare at Drayton did not differ significantly between years (P > 0.05) (Table 3). Table 3. Mean nllmbers ofskylark registrations per hectare - Drayton, 1993-1998.

1993 157 1994 1.62 1995 1.73 1996 155 1997 1.03 1998 1.34

Table 4. Analysis of variance table for skylark registrations per hectare (logJo+ 1) Drayton, 1993-98. Jlll'IIII'llllllllllft~ll~tll: 1111~lllllll~IIIIIII~IIIIIIIII~lllllllllllllll

Year 5 3.156 3.794 0.759 153 0.197 Crop 9 122.610 122.610 13.623 24.47 < 0.00 I Error 51 25.293 25.293 0.496 Total 65 151.059

Table 5. Mean nllmbers ofskylark registrations per hectare on variolls crops at Drayton, 1993-98. 1'\ltla'i"la'lll~t~I~\'llilllllll Permanent grass 0.12 Winter wheat 1.68 Permanent set-aside 2.37 Rotational set-aside 6.70 Winter oilseed rape 1.10 Spring oilseed rape 0.73 Winter beans 1.75 Temporary grass 3.50 Maize 0.67 3 year set-aside 1.19

8 DRAFT Highly significant differences between crops were observed over the six years of the study (P < 0.00 I), with numbers of skylark registrations being much higher on rotational set-aside than on • other crops (Table 5). The lowest numbers of registrations were consistently recorded on permanent grass fields.

The' permanent set-aside' in Table 5 was, in fact, a single field containing an atypical mixture of three and five year set-aside plots (some drilled with grasses, others with a 'natural regeneration' green cover). The three year set-aside was mainly natural regeneration, but was situated in half of a relatively small field, which had forage maize in the other half - in an area bounded on one side by a tree lined stream-bank and on two sides by hedges. There were also some cropped trial plots in this area in 1995, the first 'set-aside' year for this field. It is probably a mistake, therefore, to read too much into the relatively low numbers of registrations per ha in these 3 and 5 year set aside areas.

Both Kruskal-Wallis and Friedman Tests indicated that there were also highly signifIcant differences between the numbers of skylark registrations per ha in different fields over the six year period, irrespective of cropping (P < 0.001). Generally, however, this finding will have been heavily influenced by the fact that certain fIelds, in particular those in permanent grassland, remained in the same 'crop' throughout the study period. Additional multiple comparisons using Friedman's sum of ranks, considering all six years together, showed significant differences in the number of skylark registrations per ha between certain specified pairs of fields. These results tended to confirm the importance of the cropping factor, discussed above - indicating that two large fields which were in rotational set-aside for one of the six years (Fields 12 and 13) were preferred to three permanent grass fields (Fields 8, 9 and I3A) (P < 0.05). This analysis also showed that the single 'permanent set-aside' field (Field 16) was preferred to two of the grass fields (Fields 8 and I3A) (P < 0.05). Fields 8, 9, 12 and 13 were all relatively large (6-11 ha), open fields; but Field I3A was very small (09 ha) and adjacent to farm buildings.

Yellowhammers

Analysis, by the BTO, of the yellowhammer CBC registrations at Drayton showed a dramatic fall in breeding numbers in 1995, compared with the previous two years (Table 6). Numbers remained at relatively low levels in subsequent years.

Table 6. Estimated numbers ofyellowhammer territories (i.e. breeding pairs) within CBC plot - Drayton, 1993-98. 1111[llllllllil411lQU

1993 27 1994 32 1995 18 1996 18 1997 20 1998 14

9 DRAFT Analysis, by the BTO, of the yellowhammer CBC registrations at Rothamsted showed a dramatic fall in breeding numbers in 1996 compared with the previous two years (Table 7). The population • then remained stable at this low level in 1997. Analysed data for 1998 from the Rothamsted site was still not available at the time that this report was written.

Table 7. Estimated numbers ofyellowhammer territories (i.e. breeding pairs) within CBC plot - Rothamsted, 1993-971

1993 11 1994 14 1995 15 1996 8 1997 9

I Analysis of1998 CBC data/rom Rothamsted not yet completed

Table 8. Mean number ofyellowhammer registrations per 100 m ofboundary - Drayton, 1993-98. lilltl,1111111 1993 5.17 1994 4.22 1995 2.86 1996 3.30 1997 3.83 1998 3.24

Differences in yellowhammer activity at Drayton, as measured by numbers of registrations per 100 m of boundary, between years were statistically significant (P < 0.001), when all four factors were analysed together (Table 8). The number of yellowhammer registrations was at its highest in 1993 (5.17 registrations/lOO m), fell to its lowest level in 1995 and remained at relatively low levels thereafter.

10 DRAFT Table 9. Analysis of variance (ANOVA) table for yellowhammer registrations per 100m of field boundary (loglO+ 1) when year, hedge type and crop combinations are compared • Drayton, 1993-98. IICllIllIIItill1ttllllllllf 1\'li~'llll~_~III~llllii1lilllllilil

Year 5 1.102 1.262 0.252 4.68 0.001 Hedl!e type 4 1.158 0.430 0.107 1.99 0.098 Crop combination 8 2.174 2.174 0.272 5.04 <0001 Error 154 8.301 8.30\ 0.054 Total 171 12.735

Table 10. Analysis of variance table for yellowhammer registrations per 100 m of boundary (loglO+ 1) when year, hedge type and ditch type are compared - Drayton, 1993 98.

Year 5 1.193 0.886 0.177 2.66 0.024 Hedee type 6 1.345 1.091 0.182 2.73 0.015 Ditch type 3 0.358 0.358 0.119 1.79 0.150 Error 174 11.587 11.587 0.067 Total 188 14.484

Table 11. Analysis of variance table for yellowhammer registrations per 100 m of boundary (logIO+ 1) when year, hedge type, ditch type and crop combinations are compared - Drayton, 1993-98.

Year 5 1.102 1.320 0.264 5.13 <0.001 Hedl!e type 4 1.158 0.129 0.032 0.63 0.644 Ditch type 3 0.212 0.523 0.174 3.39 0.020 Crop combination 8 2.485 2.485 0.311 603 <0.001 Error 151 7.778 7.778 0.052 Total 171 12.735

11 DRAFT Analysis of variance showed that field boundary type also had a significant, though less strong, effect on the number ofyellowhammer registrations per 100 m length (P < 0.05) - with relatively • good numbers alongside hedged boundaries and streamsides with mature coppiced willows and very few being recorded along fence-lines. However, re-analysis ofthe data after simplification of the boundary type categories (removing minor categories and the poorly favoured fence-lines), indicated that hedge type did not significantly affect the number ofyellowhammers per 100 metres of boundary (P > 0.05), when all other factors were analysed together (Table 12)

Table 12. Mean number of yellowhammer registrations per 100 m of field boundary, classified by different hedge types - Drayton, 1993-98.

Short dense, rectangular hedge 3.15 Short, moderately dense, rectangular hedge 4.10 Short low density, rectangular hedge 369 Remnant hedge 4.69 Shrubby growth and trees 3.22

The number of yellowharnrners was affected by the combination of crops adjacent to the field boundaries in which they were recorded (Table 13). Highly significant differences (P < 0.001) were found between the crop combinations present. The highest numbers of registrations were recorded where rotational set-aside was present in combination with a cereal crop. The lowest numbers were recorded where grass was present on both sides of the fteld boundary.

Table 13. Mean number of yellowhammer registrations per 100 m of field boundary, classified by different combinations ofadjacent crops - Drayton, 1993-98.

Cereal/cereal 4.48 Cereal/other arable crop 4.79 Cereal/rotational set-aside 5.20 Cereal/other set-aside 4.45 Cereal/grass 2.51 Other arable crop/grass 4.51 Rotational set-aside/grass 4.28 Other set-aside/grass 2.00 Grass/grass 1.72

The presence or absence of a ditch and a strip of unmanaged vegetation significantly affected the populations of yellowhammers recorded (P < 0.05) (Table 14). The highest number of

12 DRAFT yellowhammer registrations occurred where a ditch was present along with an unmanaged grass • strip greater than 1.5 metres in width. Table 14. Yellowhammer registrations per 100 m offield boundary, classified by different ditch typesljield margin grass strip - Drayton, 1993-98.

No ditch or vegetation strip 3.29 Ditch with <1.5 m unmanaged grass strip 3.44 Ditch with> 1. 5 m unmanaged grass strip 4.60 Ditch with no adjacent unmanaged grass strip 3.75

Carabid beetles

The total number of Carabid beetles trapped at Drayton each year, between the beginning of May and the end of July, varied markedly (Table 15). Numbers fell in 1994 (-34%) and 1995 (-56%) and remained low in 1996 before recovering dramatically in 1997 (+ 175%). Numbers fell again in 1998 (-36%).

Table 15. Total number ofCarabid beetles trapped and change in numbersfrom previous year - Drayton, 1993-98.

1993 1599 Not known 1994 1057 -542 -34% 1995 470 -587 -56% 1996 514 +44 +9% 1997 1413 +899 +175% 1998 907 -506 -36%

13 DRAFT Table 16. Annual changes in /o/al Carabid beelle numbers and yellowhammer regis/rations per 100 m of boundary from previous year, and subsequent annual changes • in yellowhammer regis/rations - Dray/on, 1993-98.

1993 Not known Not known -0.95 -18% 1994 -542 -34% -0.95 -18% -1.36 -32% 1995 -587 -56% -1.36 -32% +0.44 +15% 1996 +44 +9% +0.44 +15% +0.53 +16% 1997 +899 +175% +0.53 +16% -0.59 -15% 1998 -506 -36% -0.59 -15% Not known

Table 17. Annual changes in /o/al Carabid beetle numbers and yellowhammer territories (i.e. breeding pairs) from previous year, and subsequent annual changes in yellowhammer territories - Dray/on, 1994-98.

1994 -542 -34% +5 +19% -14 -44% 1995 -587 -56% -14 -44% 0 0% 1996 +44 +9% 0 0% +2 +11% 1997 +899 +175% +2 +11% -6 -30% 1998 -506 -36% -6 -30% NOl known

14 DRAFT Table 18. Total number ofCarabid beetles trapped and change in numbersfrom previous • year - Rothamsted, J993-97. IIJlllllllil.\'IIlllIllllIIIIIIII 1111"'illlll'~'11l'l 1993 4446 Not known 1994 3074 -1372 -31% 1995 4218 +1144 +37% 1996 2206 -2012 -47% 1997 7649 +5443 +247%

As at Drayton, the number of Carabid beetles trapped at Rothamsted varied greatly from year to year, although numbers were generally much higher (Table 18). There were marked declines in 1994 (-31 %) and 1996 (-47%), but big increases again in 1995 (+37%) and 1997 (+247%). As at Drayton, 1993 and 1997 were the two best years for Carabid captures, and the greatest annual increase was recorded between 1996 and 1997.

Table 19. Annual changes in total Carabid beetle numbers and estimated yellowhammer territories (i. e. breeding pairs) from previous year, and subsequent annual changes in yelIowhammer territories - Rothamsted, J994-97.

1994 -1372 -31% +3 +27% -I +7% 1995 +1144 +37% -I +7% -7 -47% 1996 -2012 -48% -7 -47% +1 +13% 1997 +5443 +247% +1 +13% Not known

Agricultural management

The most relevant aspects of agricultural management with respect to potential impacts on skylark and yellowhammer populations at Drayton are probably a) insecticide inputs on wheat crops and b) the date of cutting or Roundup application on rotational set-aside fields. These details are summarised in Table 20 below.

Other 'routine operations' that may have a bearing on the bird registrations data described in the preceding tables, include the following: • Grassland - livestock over-wintered indoors; significant proportion of fields cut and baled for silage between mid-May and early June.

15 DRAFT • Winter wheat - fields cultivated soon after August harvest, then new crop usually drilled in first two weeks of October (only slightly late drilling in study period was 1994/95 crop - drilled 19 • 22 October 1994); herbicides applied in autumn and again in spring; fungicides in late spring (May); no routine insecticides. • Spring rape - broad spectrum insecticide (e.g. deltamethrin) applied in June for control of pollen beetle.

Table 20. Dates of insecticide applications to winter wheat and CUlling/spraying off (glyphosate application) ofrotational set-aside fields - Drayton, 1993-98.

Winter wheat Insecticide None Chlor Delta None None None spray pyrifos' methrin'

13 June 29 June

Rot. set-aside' Cut 10 May 6 May NoRSA

Spray 16 June 20 May 27 April NoRSA 10 May

I Chlorpyrifos (as Dursban 4) applied to some wheat fields only, for control of wheat blossom midge. 2 Deltamethrin (as Decis) applied to all wheat fields, for control of aphids. , Rotational set-aside in 1993 and 1996 was not within the CBC plot.

Meteorological conditions

Drayton experienced a particularly hot and dry summer in 1995 (Table 2\). 1996 was also very dry, but mean temperatures were lower.

Table 21. Mean air temperature and total rainfall at Drayton, 1 May to 31 July in each year of the bird study.

1993 14.4 187.2 1994 15.0 111.7 1995 15.2 70.0 1996 13.9 74.5 1997 14.4 215.2 1998 14.6 145.2

16 DRAFT • DISCUSSION

In this study analysis of bird census registrations data was used to determine the overall activity of two bird species, skylarks and yellowhammers, during the breeding season and how this changed within an intensively farmed area over a period of six years. In addition, the effects of changing cropping patterns and the characteristics of the fIeld boundaries present were compared with the changes in the degree of activity recorded. An indication of changes in invertebrate numbers and, hence, bird food availability was obtained from the numbers of Carabid beetles trapped during the bird breeding season.

Skylarks

Although there was no statistically signifIcant difference in the number of registrations of skylarks recorded over the six years ofthe present study, the analyses of the number of breeding territories by the BTO has shown a steady decline from 2\ breeding territories in 1993 to 12 territories in 1997 and 1998 at ADAS Drayton. A similar decline was also recorded at the ECN site at IACR Rothamsted. These population trends are in general agreement with national trends for this species, as determined from the BTO's national Common Birds Census.

Analysis of the number of skylark registrations on different crops grown at Drayton showed that there were highly significant differences in the amount of activity recorded. Rotational set-aside had a much higher number of skylark registrations per hectare, compared with all other crops grown in the study period. This effect is in agreement with the findings of a number of other studies. For example, Henderson el a1. (I998), in a recent survey for MAFF, found that skylarks showed a very strong preference for set-aside, particularly rotational set-aside with a sparse vegetation structure (i.e. where there was a fairly high proportion of bare ground or straw/litter).

Wilson el at. (1997) found higher skylark densities on set-aside and in organically cropped fields, and also recorded improved breeding success on set-aside, in comparison with intensively managed cereals.

Grynderup Poulsen el al. (1998) also found that set-aside, and permanent grass, were preferred habitats for skylarks on predominantly arable farmland, with territory densities 2-3 times higher than on cereals. They also found that skylarks had relatively small territories (1.7 ha) on set-aside and large ones on winter cereal crops (4.5 ha). Furthermore, Grynderup Poulsen el at. (1998) found evidence that the density of successful nests was much higher on set-aside than on arable crops (more than double) and the average clutch size at hatching larger than in silage grass or spring barley. Overall, although fledgling success did not differ according to crop type, the other factors mentioned all combined to give a better breeding 'productivity' on set-aside (0.50 fledglings/ha, compared to 0.13 in silage crops, 0.21 in spring barley and 0.0 in winter cereals).

The reasons for the apparent attractiveness of rotational set-aside to skylarks have been given as the development of a varied flora, which may result in increased numbers of invertebrates and the presence of a sparse vegetation layer - allowing easy access to the ground or soil litter. Henderson el al. (1998) have suggested that skylarks, and other bird species with relatively diverse dietary requirements that show a preference for rotational set-aside, are probably attracted by both an increased availability of invertebrate food (and they had some evidence that set-aside fields in their survey had higher densities of ground-dwelling invertebrates than did adjacent arable crops) and enhanced amounts of weed seeds from indigenous plants.

17 DRAFT Previous studies have also shown the importance of winter stubbles to skylarks. With winter sown crops having largely superseded spring crops throughout England, and the consequent • demise of spring cultivation, compulsory rotational set-aside has ensured the large-scale return of winter stubbles.

The potential loss of most of the rotational set-aside area in Europe within three years, with a proposed 0% compulsory set-aside rate in 2001 under a reformed Common Agricultural Policy, seems likely to seriously disadvantage bird species such as the skylark.

Other types of set-aside monitored during the period of this study were three-year and five-year set-aside. Much of this area was within a long-term 'permanent set-aside' experiment, with a large proportion sown with a grass cover crop - probably making it less attractive to skylarks than the areas of rotational set-aside, on which natural regeneration of vegetation took place. The density of the grass cover would not allow easy foraging for skylarks and would provide fewer suitable nesting sites. As important here, however, is the fact, mentioned previously in the Results section, that the other area of 'three-year set-aside' at Drayton, between 1995 and 1997, was in a small area, enclosed by tall field boundaries and an adjacent maize crop. Skylarks are well known to prefer open habitats. For example, a previous study by Wilson et al. (1997) found that skylark densities were "lowest on fields surrounded by tall boundary structures or ullsuitable habitat, and those with tall, dense vegetation cover."

With this point in mind, it is worth noting that the rotational set-aside areas at Drayton in 1994 and 1995 were both in relatively large (6.87 and 8.63 ha respectively), open fields (rectangular with only low hedges or fence-lines surrounding them); whereas the rotational set-aside field in 1998 was smaller (4.78 ha) and much more enclosed (with one very tall boundary). The respective numbers of skylark registrations per ha in 1994, 1995 and 1998 were 7.13, 7.42 and 4.79. It would seem, therefore, that positioning of rotational set-aside may be important if the benefits to skylarks are to be maximised.

A small area of temporary grass occupied part of the field in which pennanent set-aside was present during the first two years of the study. The adjacent permanent set-aside is likely to have affected the results for this crop, resulting in a higher apparent number of registrations than might be expected if the crop was in a different location.

18 DRAFT • Yellowhammers

The number of breeding territories determined by the BTO analysis of CBC data from Drayton declined from 32 in 1994 to 14 in 1998. A particularly large decline occurred in 1995. At IACR Rothamsted numbers also declined, from 14 in 1994 to 9 in 1997 (\998 data was not available). The largest decline occurred in 1996. When the yellowhammer populations were measured in terms of the number of registrations per 100 metres of field boundary a severe decline was again noticeable at Drayton

Yellowhammers nest most often in shrubby growth such as hedges, or on ditch banks (Bradbury el aI., in press). There is plentiful evidence in the literature that yellowhammers generally prefer relatively short hedges « 1.5 m), with few or no hedgerow trees. However, analysis of the number of Drayton yellowhammer registrations obtained in hedges with different densities and different hedge types indicated that there were no statistically significant differences between them. Additional data regarding the numbers of registrations on two other field boundary types were not included in the analysis as the number of observations was too small to allow comparisons with the hedge types. These were wire fences, where hardly any registrations occurred, and a limited area which contained pollarded trees bordering a brook. The brook had a coarse grass bank which was adjacent to an arable field. In keeping with the results found during the analyses of ditch type data, the number of registrations on this boundary type was high.

The number of registrations per 100 metres of boundary, where various combinations of cropping occurred on each side of the field boundary, indicated a negative effect of grassland. This fmding, i.e. the general avoidance of intensively managed grass, supports the evidence from several previous studies (e.g. Kyrkos el al., 1998; Morris el aI., in press; and Bradbury el al., in press). Bradbury el al. (in press) report that the lowest densities of yellowhammers in their study were found next to pasture and grass leys. Green el al. (1994) found that yellowhammers (like greenfinches) showed a preference for hedgerows between two arable fields over those between two grass fields. They also preferred arable/road combinations over grass/road.

19 DRAFT The lowest numbers were recorded where grass was present on both sides of the boundary, with numbers also being reduced when grass was present in combination with cereals or 'other' set aside. 'Other' set-aside at this site was composed mainly of a sown grass cover crop and, • therefore, although not so intensively managed as grazed grassland, would not be expected to be significantly more attractive to yellowhammers. Numbers of registrations were relatively high when grass was in combination with the category of 'other arable crops'. Most of the grassland on the site is intensively managed, either for grazing or for silage production. This degree of management is likely to reduce the food resource available to foraging yellowhammers due to the limited range of plant species present, high levels of fertiliser usage and consequently low populations of invertebrates. High densities of yellowhammers are known to use unimproved semi-natural grassland where the invertebrate populations are much higher.

Carabid beetles

20 DRAFT pitfall trapping for Carabid beetles was carried out may not have reflected sufficiently closely the range of invertebrates available to skylarks in the more open field situation. Invertebrate groups • less affected by the conditions which caused beetle populations to decline (e.g. Diptera) may have formed a greater proportion of their diet at this time.

CONCLUSIONS AND CONSERVATION IMPLICATIONS

The ability of skylarks to rapidly make use of rotational set-aside land within an otherwise intensively managed agricultural situation has been shown by this study. These results show the importance of the correct management of rotational set-aside and similar schemes and that account continues to be taken of the fact that populations of some fauna are attracted to it at critical stages in their life-cycle. Poor management decisions on set-aside would, potentially, have a far more damaging effect on skylark populations than would be the case on an equivalent area of most other crops.

Compared with rotational set-aside, grassland at the Drayton site was very unsuitable for skylarks. The grassland was very intensively utilised either by grazing livestock or for silage making which makes it poor as a potential nesting site and also as a food resource. Land occupied by longer term set-aside sown with a grass cover crop is not as attractive to skylarks as rotational set-aside, though numbers present were higher than on most other crops. In order to make grassland more suitable for use by skylarks a significant reduction in the intensity of its management would be required.

Success in the maintenance of yellowhammer populations appears to be linked with cropping, field boundary management and invertebrate density. Intensively managed grassland is not favoured by yellowhammers, probably due to the lack of easily available food. Managing field boundaries to include a relatively unmanaged coarse grass area, preferable of at least 1.5 metres in width, appears to increase the amount of yellowhammer activity. This may be due to the availability of invertebrate food during the nesting season. Where the amount of invertebrate food in such field margins has declined, albeit probably due to climatic change in this study, this has had a rapid deleterious effect on yellowhammer populations. Unlike the Carabid populations, the yellowhammers (like all bird species) do not have the capacity for very rapid population recovery following years of heavy losses.

The importance of invertebrate population levels apppears to have been shown at Drayton in 1995, when both Carabid beetle and yellowhammer numbers were severely reduced. Although, in this instance, the declines were probably climatically related the intensification of farmland management and consequent reduction in suitable foraging areas is likely to have similar effects over a longer timescale and it may be the effects of these habitat related changes which are producing the declines in some farmland bird populations being recorded in this and other studies.

In this study the unmanaged grass strips were adjacent to fields containing arable cropping. The results show that intensively managed grassland is less suitable for yellowhammers than arable cropping. Further examination of the effects of establishing 'unutilised' margins to intensively

21 DRAFT managed grass fields on bird and invertebrate populations is necessary. If similar effects are forthcoming using margins on grassland, as occur in arable situations then the beneficial effects • for farmland birds are likely to be substantial. This could be of particular importance in the South-west of England, where the intensification of grassland management is thought to have had a particularly deleterious effect on birds such as the yellowhammer.

This study has highlighted the need for the habitat requirements of individual species, or groups of species, to be taken into account when considering management options to encourage biodiversity or when attempting to increase the population of a particular species. In this exercise skylark population activity was seen to be increased by the provision of relatively large, fIeld scale, areas of set-aside land. The presence of more limited areas of set-aside in, say, the field boundaries, is less likely to produce this effect. On the other hand, these methods, for example creating coarse grassy margins to arable fields, would greatly benefit populations of farmland birds similar in habit to yellowhammers.

ACKNOWLEDGEMENTS

REFERENCES

Bealey, C; Howells, 0 & Parr, T (1998). Environmental change and its effects on wildlife: the role of the Environmental Change Network. British Wildlife xx. 34 I-347.

Bradbury, R Bj Kyrkos, Aj Morris, A J; Clark, S Cj Perkins, A J & Wilson, J D (in press). Habitat selection, breeding success and demographic predictions for yellowhammers Emheriza citrinella on lowland farmland.

Gillings, S & Fuller, R J (1998). Changes in bird populations on sample lowland English farms in relation to loss of hedgerows and other non-crop habitats. Oecologia 116. 120-127.

Green, R Ej Osborne, P E & Sears, E J (1994). The distribution of passerine birds in hedgerows during the breeding season in relation to characteristics of the hedgerow and adjacent farmland. Journal ofApplied Ecology 31. 677-692.

Grynderup Poulsen, J; Sotherton, N W & Aebischer, N J (1998). Comparative nesting and feeding ecology of skylarks Alauda arvensis on arable farmland in southern England with special reference to set-aside. Journal ofApplied Ecology 35. 131-147.

22 ...

DRAFT Henderson, 1 G; Cooper, J & Fuller, R J (1998). Spatial and temporal utilisation of set-aside by birds in summer on arable farms in England. In Agronomic and Environmental Evaluation of Set-aside under the EC Arable Area Payments Scheme (Ed. L G Firbank), Vol. 4 Birds & " Modelling, Centre for Ecology and Hydrology!ADASlBritish Trust for Ornithology report to Ministry of Agriculture, Fisheries and Food. May 1998. ITE, Grange-over-Sands, Cumbria.

Kyrkos, A; Wilson, J D & Fuller, R J (1998). Fannland habitat change and abundance of Yellowhammers Emberiza citrine/la: an analysis of Common Birds Census data, Bird Study 45. 232-246.

Moles, R T & Breen, J (1995). Long-term change within lowland farmland bird communities in relation to field boundary attributes. Biology and Environment: Proceedings of the Royal Irish Academy Vol 95B, NO.3. 203-215.

Morris, A J; Whittingham, M J; Bradbury, R B; Wilson, J D; Kyrkos, A; Buckingham, D L & Evans, A D (in press). Habitat selection by foraging yellowhammers on lowland farmland during the breeding season.

Parish, T; Lakhani, K H & Sparks, T H (1994). Modelling the relationship between bird population variables and hedgerow and other field margin attributes. I. Species richness of winter, summer and breeding birds. Journa! ofApplied Ecology 31. 764-775.

Sparks, T Hj Parish, T & Hinsley, S A (1996). Breeding birds in field boundaries in an agricultural landscape, Agriculture, Ecosystems and Environment 60. 1-8.

Wilson, J D; Evans, J; Browne, S J & King, J R (1997). Territory distribution and breeding success of skylarks Alauda arvensis on organic and intensive farmland in southern England, Journal ofApplied Ecology 34. 1462-1478.