Disentangling the effects of fertilisers and pesticides on winter stubble use by farmland birds

Page 1

Db

A

a

b

c

d

R

A

htwswCNpayRf

Z

AdmVjv

p

1d

Basic and Applied Ecology 12 (2011) 80–88

isentangling the effects of fertilisers and pesticides on winter stubble usey farmland birds

ilsa J. McKenziea,∗, Juliet A. Vickeryb,c, Carlo Leifertd, Peter Shottond, Mark J. Whittinghama

School of Biology, Ridley Building, Newcastle University, Newcastle-Upon-Tyne NE1 7RU, UKBTO, The Nunnery, Thetford, Norfolk IP24 2PU, UKRSPB, The Lodge, Sandy, Bedfordshire SG19 2DL, UKNafferton Ecological Farming Group, Nafferton Farm, Stocksfield, Northumberland NE43 7XD, UK

eceived 1 February 2010; accepted 27 October 2010

bstract

Cereal stubbles are a preferred foraging habitat for overwintering granivorous farmland bird species. Levels of this habitatave declined in recent decades across much of western Europe with increasing agricultural intensification. Organic farmsypically hold more stubble fields than conventional farms and thus may provide important refuges for wintering birds. However,hile organic stubble fields often contain higher food densities than conventional stubble fields, the more complex vegetation

tructure associated with organic farming may decrease use by birds. Bird use, vegetation characteristics and seed densitiesere measured on stubble plots managed under four strategies (Organic [organic fertiliser only and no chemical pesticides],onventional [inorganic fertiliser and chemical pesticides], NOFERT [organic fertiliser only and chemical pesticides) andOPEST [inorganic fertiliser and no chemical pesticides]). Skylarks foraged most frequently on stubbles which received noesticide applications which also had the highest weed seed densities. Plots receiving either inorganic or organic fertiliserpplications did not differ in terms of use by skylarks, weed seed density or diversity, or vegetation structure. Plot use byellowhammers was not significantly related to pesticide or fertiliser applications. Possible reasons for this are discussed.esults suggest that the main benefit of organic stubble fields for birds is via reduced pesticide inputs. Use of inorganic

ertilisers is also beneficial for birds via increased weed seed densities, but to a lesser extent.

usammenfassung

Getreidestoppelfelder gehören zu den bevorzugten Nahrungshabitaten für überwinternde, körnerfressende Vogelarten der

grarlandschaft. Die Ausdehnung dieses Habitats hat in den letzten Jahrzehnten in großen Teilen des westlichen Europas miter zunehmenden Intensivierung der Landwirtschaft abgenommen. Organisch bewirtschaftete Betriebe bieten typischerweise

und könnten so einen wichtigen Rückzugsraum für überwinternde

ehr Stoppelfelder als konventionell bewirtschaftete Betriebe ögel bieten. Während organische Stoppelfelder häufig höhere Futterdichten als konventionelle Stoppelfelder aufweisen, könnte

edoch die komplexere Vegetationsstruktur, die mit der organischen Bewirtschaftung einhergeht, die Nutzung durch die Vögelermindern.

Die Nutzung durch die Vögel, die Charakteristik der Vegetation und die Samendichten wurden in Probeflächen in den Stop-elfeldern gemessen, die unter vier verschiedenen Bewirtschaftungsmethoden standen (Organisch [nur organische Dünger und

∗Corresponding author. Tel.: +44 191 2225952; fax: +44 191 2225229.E-mail address: [email protected] (A.J. McKenzie).

439-1791/$ – see front matter © 2010 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.oi:10.1016/j.baae.2010.10.007

Page 2

ku

suts

fd©

K

I

wbThtFAabmlad

trolga(

sgsmBbptr

gsbb(T

A.J. McKenzie et al. / Basic and Applied Ecology 12 (2011) 80–88 81

eine chemischen Pestizide], konventionell [künstlicher Dünger und chemische Pestizide], NOFERT [nur organischer Düngernd chemische Pestizide] und NOPEST [künstlicher Dünger und keine chemischen Pestizide]).

Feldlerchen suchten zumeist auf Stoppelfeldern nach Nahrung, die keine Pestizidanwendung erfuhren und die auch die höch-ten Unkrautdichten aufwiesen. Die Probeflächen, die entweder künstliche oder organische Düngeranwendungen erhielten,nterschieden sich nicht in der Nutzung durch die Feldlerchen, in der Samendichte oder -diversität oder in der Vegetationsstruk-ur. Die Nutzung der Probeflächen durch Goldammern stand mit den Anwendungen von Düngern oder Pestiziden in keinemignifikanten Zusammenhang.

Die möglichen Gründe dafür werden diskutiert. Die Ergebnisse lassen vermuten, dass der größte Nutzen der Stoppelfelderür die Vögel in dem reduzierten Pestizideintrag liegt. Die Vögel profitieren ebenso von der Nutzung organischer Dünger durchie erhöhten Dichten der Unkrautsamen, aber in einem geringeren Ausmaß.

2010 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.

ging; S

saiaWdmc

s(g(cdtttcTfdedhsetfgOit

M

eywords: Fertilisers; Fertilizers; Herbicides; Organic; Birds; Fora

ntroduction

Winter cereal stubble is an important food resource for aide range of granivorous farmland bird species, often har-ouring crop and weed seeds in large quantities (Wilson,aylor, & Muirhead 1996). Changes in agricultural practiceave, however, led to a significant decline in cereal stubbles inhe last 50 years (Robinson & Sutherland 1999; Chamberlain,uller, Bunce, Duckworth, & Shrubb 2000; Newton 2004).

large-scale shift from spring to autumn sowing of cropscross western Europe has meant that fields once left as stub-le until spring are now ploughed and replanted in autumn,aking them unavailable during the winter period for farm-

and birds. Chamberlain et al. (2000) reported that 99% ofll wheat in the UK is now sown during autumn rather thanuring spring.

Organic farms tend to possess more winter stubblehan those managed conventionally. Shepherd et al. (2003)eported a 27% increase in spring-sowing of cereals followingrganic conversion, which will likely equate to an equiva-ent increase in winter stubble area. This increase has beeniven as a contributing factor to the increased biodiversity andbundance of many farmland bird species on organic farmsMcKenzie & Whittingham 2009).

However, it remains unclear how organic management oftubble fields affects intake rates and food accessibility forranivorous birds. While organic stubble fields have beenhown to contain higher seed densities than conventionallyanaged stubble fields, a study by Moorcroft, Whittingham,radbury, & Wilson (2002) did not find the use made byirds of the two field types to differ consistently, some speciesreferring conventional fields (e.g. yellowhammer, grey par-ridge, skylark), others preferring organic fields (e.g. linnet,eed bunting).

Recent work has demonstrated that intake rates forranivores are reduced by both vegetation height and sub-trate complexity. For example, seed densities needed to

e approximately 2.5 times greater in patches of tall stub-le (13 cm) before equal use was made with short stubble3 cm) (Butler, Whittingham, Quinn, & Cresswell 2005).his means that while organic stubble may promote higher

S

(

eed diversity; Farmland

eed abundances, the increased vegetation structure associ-ted with the regime may, in turn, alter food availability byncreasing both predation risk (by increasing obstruction)nd lowering food accessibility (Butler & Gillings 2004;

hittingham, Devereux, Evans, & Bradbury 2006). Thus theifferences in vegetation structure on organic stubble fieldsay reduce food availability and ultimately influence patch

hoice (Stephens & Krebs 1986).Organic and conventional farming differ mainly in their

oil fertility management and crop protection protocolsLehesranta et al. 2007), the organic regime excluding inor-anic fertilisers and the majority of chemical pesticidese.g. insecticides, herbicides, fungicides, growth regulators)haracteristic of conventional farming. Several studies haveiscussed the likely influence exclusion of each of theseypes of input would have on seed abundance and vege-ation structure (e.g. Hyvönen & Salonen 2002). However,heir occurrence together in any one farming system has pre-luded absolute quantification of their relative importance.o understand fully how best to provide optimum conditionsor foraging farmland birds, there is a need to understand howifferent management techniques (e.g. pesticides and fertilis-rs) affect both food abundance and vegetation structure. Theesign of the current experiment allows us to do this for oneabitat type: barley stubble. We compared bird abundance ontubbles from barley crops which had received four differentxperimental treatments [(1) Conventional (inorganic fer-iliser and chemical pesticide inputs); (2) NOFERT (organicertiliser and chemical pesticide inputs); (3) NOPEST (inor-anic fertiliser and no chemical pesticide inputs) and (4)rganic (organic fertiliser only and no chemical pesticide

nputs)] on barley stubble plots, and related this to the vege-ation structure, seed density and seed diversity of the plots.

aterials and methods

tudy site

The study was carried out over three winters2006/2007–2008/2009) from November to March at

Page 3

82 A.J. McKenzie et al. / Basic and Applied Ecology 12 (2011) 80–88

Fig. 1. Diagram of the Nafferton Farm factorial systems comparison experiments (NFSC). Plots were located within a single 6 ha fieldsubdivided into series of field experiments. Four blocks of barley stubble were established per year (4 m × 56 m) – (a) positions in Years 1a ning blo se and( aded) a

N(peupifmafwwttwdfiwoeptr

B

otbwv

etoponbat

S

esfpta

ptTat

V

nd 3, (b) positions in Year 2. Other crops were grown in the remaine managed under each treatment. Hatching represents pesticide ushaded and hatched), NOFERT (white and hatched), NOPEST (sh

afferton Farm near Stocksfield, Northumberland, UKNZ066662) using the Nafferton factorial systems com-arison (NFSC) experiments (Fig. 1). These long-termxperiments were established in 2001 on a field with aniform sandy loam soil where, under a split–split–splitlot design, replicate blocks of a range of crops are grownn the context of two different rotational sequences andour production systems (subsequently referred to as “treat-ents”; for full description see Appendix A: Table 1). In

ll years four replicates of barley stubble under each of theour treatments were available and as such these 16 plotsere used in this study (note, location of these plots variedithin the field between years, but individual plots retained

he same treatments in each year – see Fig. 1). In two of thehree years of the study the stubble plots were undersownith grass. This factor was included in the statistical modelesign (see below). The landscape surrounding the trialeld was heterogeneous, containing a range of hedgerow,oodland and farmland habitats. The random assignmentf the trial plots within the field (Fig. 1), however, shouldliminate any biases associated with proximity of individuallots to other habitat types. A 10 m buffer also lay betweenhe edge of the trials plots and the overall field-edge, furthereducing the likelihood of landscape effects.

ird occurrence

Visits were carried out typically once weekly during eachf the three winters from mid-November to mid-February (a

otal of 43 visits over three winters). Visits were carried outetween 1 h after dawn and 1 h before dusk to avoid periodshen birds were arriving or leaving roost sites. Periods ofery wet or windy weather were avoided due to the negative

shor

ocks. Each block was subdivided into four 4 m × 12 m sub-blocks,shading represents inorganic fertiliser use, therefore, Conventionalnd Organic (white blocks).

ffect of such conditions on bird activity. A visit comprisedhe observer walking across all 16 barley plots to within 5 mf every part of each plot, ensuring all birds were flushed. Thelots were flushed in a random pattern, with all birds sightedn each plot being recorded to species level. Care was takenot to double-count birds by noting where previously flushedirds had landed. Birds typically left the experimental areafter being flushed from a trial plot, making it very unlikelyhat the same birds were recorded more than once.

oil seed densities

Soil samples were taken from each of the 16 plots twiceach winter, once in November and once in February (i.e. sixample periods in total). The top 3 mm of soil was removedrom four 20 cm × 20 cm randomly-placed quadrats in eachlot and bulked in resealable polythene bags. These werehen stored at 4 ◦C within 24 h to prevent germination untilnalysis.

Each sample was weighed and a random 25% subsam-le removed for analysis. Seeds were extracted by washinghrough sieves of decreasing mesh size (1 mm and 500 �m).he contents of the sieves were washed into sample trays andllowed to dry. Seeds were then sorted, counted and identifiedo family.

egetation sampling

Vegetation sampling was carried out concurrently with

eed sampling, twice per winter. Mean stubble and weedeights in each treatment were assessed by recording heightf the stems closest to the four corners of a 50 cm × 50 cmandomly placed quadrat. Percentage bare ground in each

Page 4

d Appl

qat

A

Mfsmfov“pitNApotm

swvtodt

Fidppf

Mdiw(wec

R

B

mtdabw

F[yYPp

A.J. McKenzie et al. / Basic an

uadrat was estimated to the nearest 5%. Four quadrats weressessed in each plot during each sample period and a meanaken.

nalysis

Bird data were analysed using Generalised Linear Mixedodels (GLMMs) in R (version 2.9.1), one model built

or skylark and another for yellowhammer (the only twopecies with sufficient data to allow analysis). Data wereodelled with a binomial error structure and a logit link

unction. Plot was declared as a random effect, presencer absence of a species in a plot on a visit as the responseariable and ‘year’ (a three level factor which encompassedundersowing”) and use of ‘pesticides’ and ‘fertilisers’ asredictors. Codes for the use of ‘pesticides’ and ‘fertilisers’ndependently in a treatment were used instead of four levelreatment codes, i.e. Organic plots were given the code 0/0,OFERT 1/0, NOPEST 0/1, Conventional 1/1 (see Appendix: Table 1). This allowed investigation of the impact ofesticide use versus the impact of inorganic fertiliser usen skylark and yellowhammer occurrence. The interac-ion term between ‘pesticides’ and ‘fertilisers’ was also

odelled.Data collection which involves multiple visits to a single

ite has the potential to become autocorrelated. While thisas controlled for to an extent in the current experiment asisits tended to be at least one week apart, steps were taken

o further reduce the risk. A resampling protocol was carriedut using the skylark dataset (this dataset only as insufficientata existed for other species) whereby years were split intohree time blocks (November/December, January/February,

Opw(

ig. 2. Skylark occurrence per visit on stubble plots receiving differentfertiliser no pesticide] and ORG). Year 1 at bottom of stacked bar, Yearear, therefore the graph shows the proportion of visits on which one orear 1, skylarks were observed in organic plots on 19 occasions, over a testicide applications significantly affected skylark presence, with significa< 0.005; Table 1). Fertiliser use had no effect on use of plots by skylarks

ied Ecology 12 (2011) 80–88 83

ebruary/March) and a visit taken at random from each blockn each year. The above model was then rerun with these nineata points. This was repeated 50 times and the mean χ2 and-value calculated. This was done to check that the sameatterns were present in the data whilst attempting to controlor repeat measures (see section ‘Discussion’).

For seed and vegetation measurements, GLMs were run ininitab (Version 15) with ‘pesticides’ and ‘fertilisers’ as pre-

ictors as before and each of the measurements as responsesn turn (seed density, mean percentage bare ground, meaneed height, mean stubble height). Further separate GLMs

or Kruskal–Wallis tests for non-normally distributed data)ere run on the seed data to assess the significance of differ-

nces in monocotyledon/dicotyledon ratio and seed familyomposition between the treatments.

esults

ird occurrence

204 records of birds of three species (skylark, yellowham-er and grey partridge) were made foraging in the plots over

hree winters (additional records of birds flying over wereiscarded). Grey partridge were removed from the analysiss they were only observed on one occasion. Skylarks madey far the largest contribution to the total birds observedith one or more birds flushed 112 times on 43 visits.

ne or more yellowhammers were flushed 15 times fromlots on 12 visits. The distribution of skylarks on the plotsas as follows – NOPEST (40) > Organic (30) > NOFERT

21) = Conventional (21) and for yellowhammers NOPEST

treatments (CONV, NOFERT [pesticide no fertiliser], NOPEST3 at top. Each field was visited a particular number of times eachmore birds was observed foraging in the plots. For example, in

otal of 27 visits – therefore 0.7 skylarks were recorded per visit).ntly more use of plots receiving no applications (z = −2.92, d.f. = 1,(z = 1.06, d.f. = 1, p = 0.29).

Page 5

84 A.J. McKenzie et al. / Basic and Applied Ecology 12 (2011) 80–88

Table 1. GLMM output from the models run in R.

Estimate SE z-Value p-Value

SkylarksIntercept −1.854 0.364 −5.099 3.41 × 10−07

Year 0.180 0.116 1.556 0.120Pesticide use −0.630 0.216 −2.919 0.004Fertiliser use 0.224 0.212 1.058 0.290Pesticide*fertiliser interaction −0.376 0.432 −0.869 0.385

YellowhammersIntercept −4.992 0.784 −6.365 1.96 × 10−10

Year 0.185 0.282 0.657 0.511Pesticide use 0.417 0.534 0.780 0.435

(A

pecpnsinysd

4mdttTp

or

S

dwfissnol(i1

Fu

Fertiliser use 1.034Pesticide × fertiliser interaction −17.014

6) > NOFERT (5) > Conventional (4) > Organic (0) (Fig. 2;ppendix A: Table 2).Pesticide use had a significant negative effect on skylark

resence, while inorganic fertiliser use had no significantffect (Fig. 2; Table 1). The interaction term between pesti-ide and fertiliser use was non-significant (z = −0.87, d.f. = 1,= 0.39). For yellowhammers, pesticide use did not sig-ificantly affect presence but there was a trend towards aignificant effect of inorganic fertiliser use (Table 1). Thenteraction term between pesticide and fertiliser use was againon-significant (z = −0.01 d.f. = 1, p = 0.994). The effect ofear (and, by proxy, the effect of undersowing) was non-ignificant in the models for either species (skylarks: z = 1.56,.f. = 1, p = 0.12; yellowhammers: z = 0.66, d.f. = 1, p = 0.51).

The result of the resampling protocol (dropping from3 visits to 9) was not significant (mean χ2 across 50odels = 2.57, d.f. = 1, p = 0.11). However, considering the

irection of the trend remained the same as that found by

he main analysis, with a sample size reduced by morehan three-quarters, our overall argument remains supported.he parameter estimates indicated that in all 50 models therobability of one or more skylarks occurring on plots with-

iTwC

ig. 3. Mean weed seed density in soil samples (±1 SE). Bar signatures asse (GLM: F1,21 = 19.67, p < 0.001) but not by fertiliser use (GLM: F1,21 =

0.589 1.754 0.0792224.935 −0.008 0.994

ut pesticide treatments was higher than on plots that hadeceived pesticides.

eed densities and diversity

Weed seed densities varied widely between samples fromifferent treatments. More weed seeds were found on plotsithout pesticide treatments (Fig. 3), densities being betweenve and seven times lower in Conventional and NOFERTamples than in Organic and NOPEST samples (Fig. 3). Weedeed densities were negatively affected by pesticide use butot affected by inorganic fertiliser use (Fig. 3). Depletionf weed seeds between the November and February col-ections varied between 2 and 26%. Broad-leaved speciesdicotyledons) made up the largest percentages of weed seedsn all samples from all years (75–95%; Appendix A: Fig.). The monocotyledon to dicotyledon ratio of weed seeds

n samples did not differ significantly between treatments.he four families of seeds found most frequently in samplesere Chenopodiaceae (fat-hen), Polygonaceae (knotweeds),aryophyllaceae (mainly chickweeds) and Poaceae (grasses)

in Fig. 1. Weed seed density was significantly affected by pesticide0.42, p = 0.53).

Page 6

A.J. McKenzie et al. / Basic and Applied Ecology 12 (2011) 80–88 85

F ood set ntly wC .

(saase

fsA

V

bsuci

D

B

fi(2l2wl

stldowhNentffs

baidctsfnArm(

r

ig. 4. Mean percentage composition (±1 SE) of important bird fures as in Fig. 1. Percentages of Chenopodiaceae varied significaaryophyllaceae with pesticide use (GLM: F1,21 = 17.21, p < 0.001)

Fig. 4). Abundance of Chenopodiaceae in samples wasignificantly positively affected by inorganic fertiliser use,nd abundance of Caryophyllaceae significantly negativelyffected by pesticide use (Fig. 4). The abundance of othereed families in samples was not significantly affected byither pesticide or inorganic fertiliser use (Fig. 4).

Very few crop seeds (mean = 25 ± 6.42 SE seeds/kg) wereound in the soil samples overall and their distribution was notignificantly affected by pesticide or fertiliser use (Appendix: Fig. 2).

egetation measurements

None of the vegetation measurements (mean percentageare ground, mean stubble height, mean weed height) wereignificantly affected by either inorganic fertiliser or pesticidese overall (i.e. when summed across the three years) or whenonsidered by season (i.e. November vs. February) (p > 0.10n all cases) (see Appendix A: Fig. 3).

iscussion

ird distribution and seed availability

Granivorous passerines have been shown to select stubbleelds during winter, often to the exclusion of other field typesRobinson & Sutherland 1999; Perkins, Maggs, & Wilson008). Overall use of stubble plots in the current study was

ow compared to that reported elsewhere (e.g. Moorcroft et al.002). Diversity of bird species using the plots was also low,ith only three species recorded over the entire study (sky-

ark, yellowhammer and grey partridge). However, the small

rs2i

ed families found in samples from the four treatments. Bar signa-ith fertiliser use (GLM: F1,21 = 9.07, p < 0.01) and percentages of

ize and the substantial availability of other non-crop habi-at in the surrounding area (e.g. large stubble fields, fallowand, crop edges) is likely to have led to reduced numbers andiversity of granivores using the stubble plots. Indeed, studiesf stubble use by birds in Spain have found usage to declineith increased presence of fallow land and other non-cropabitats in the surrounding countryside (Suarez et al. 2004).umerous other granivorous species were observed foraging

lsewhere on the farm (e.g. tree sparrow, reed bunting, lin-et). The stubble was also undersown with grass during yearswo and three of the experiment, a practice which has beenound to decrease use by birds (Moorcroft et al. 2002) (forull discussion of the effect of undersowing, see “Vegetationtructure” section below).

Skylark occurrence on the plots was significantly affectedy pesticide use, plots receiving no pesticide inputs (Organicnd NOPEST) being used more often than those receivingnputs (Conventional and NOFERT). Inorganic fertiliser useid not significantly affect use of plots by skylarks. Pesti-ide (specifically herbicide) applications have been showno reduce dramatically the availability of weed seeds in theoil, many of which are important winter food resourcesor granivorous bird species. While stubbles in the UK areot generally treated with herbicides (Wakeham-Dawson &ebischer 1998), applications in the previous crop tend to

educe weed and seed densities dramatically, making theajority of stubbles unsuitable foraging habitat for birds

Gillings, Newson, Noble, & Vickery 2005).The pattern of use by skylarks in the current experiment

eflects mean weed seed density in the plots. Significant

elationships have been reported previously between fieldeed mass and granivorous bird density (Moorcroft et al.002; Hotker, Jeromin, & Rahmann 2004), with birds tend-ng to avoid stubble fields with less than 300 seeds per

Page 7

8 d Appl

msafimNi(a

S

wfaBosdtobTcacwSgBlftdd2cGowtawi

V

awmTd

a(ap(h

fda&ftcftcph

bagshe

R

pvGdav2inymtfrssfhtbb

6 A.J. McKenzie et al. / Basic an

2 (Robinson, Hart, Holland, & Parrott 2004). Seed den-ities in plots receiving no pesticide inputs (i.e. NOPESTnd Organic plots) in the current experiment were betweenour and six times higher than those receiving pesticidenputs (Conventional and NOFERT plots). This is a com-on finding for organic and low input fields (e.g. Koocheki,assiri, Alimoradi, & Ghorbani 2009). Weed seed densities

n the Conventional and NOFERT plots were extremely low32–58 seeds/kg), likely as a result of the repeated herbicidepplications.

eed diversity

The species’ found most frequently in seed samplesere those shown previously to be key food resources

or granivores (e.g. Polygonaceae, Poaceae, Chenopodi-ceae, Caryophyllaceae) (Wilson, Morris, Arroyo, Clark, &radbury 1999). However, there appeared to be little impactf treatment on seed diversity, with densities of only twoeed families affected significantly by treatment (Chenopo-iaceae and Caryophyllaceae). Several authors have foundhe opposite result, with weed and seed bank diversity, notnly abundance, tending to be higher in the absence of her-icide spraying (Ball 1992; Robinson & Sutherland 1999).his increase tends to manifest itself in changes in the mono-otyledon to dicotyledon ratio of the seed bank. Herbicidepplications tend to promote growth of perennial mono-otyledons (i.e. grasses) rather than annual dicotyledons,hich repeated herbicide applications tend to eliminate (denoo 1997; Albrecht 2005). While some monocotyledonrass seeds are suitable food for granivores (e.g. ryegrass;uckingham & Peach 2006), such seeds are, on the whole,

ess adequate food for birds than dicotyledon seeds. Resultsrom the present study found no significant difference inhe monocotyledon to dicotyledon ratio of samples fromifferent treatments, or in diversity overall. This has beenocumented previously (e.g. Weibull, Ostman, & Granqvist003). The impact non-spraying has on weed diversity isomplex, and benefits may be site-specific. Roschewitz,abriel, Tscharntke, & Thies (2005), for example, foundrganic agriculture only to benefit weed diversity in areasith very little habitat heterogeneity overall. The field where

he stubble plots were located in the current study containedlarge amount of non-cropped habitat, as did the farm as ahole. This likely led to a higher degree of seed immigration

nto all plots than would occur in less complex habitats.

egetation structure

Vegetation structure is typically more complex in thebsence of herbicides, often with increased ground cover and

eed height. This was not the case in the current study, treat-ent having no significant effect on either of these variables.his may be explained by the stubble plots being undersownuring Years 2 and 3 of the experiment. Undersowing gener-

itm(

ied Ecology 12 (2011) 80–88

lly decreases weed abundance and increases ground coverMoorcroft et al. 2002). Indeed, when Year 1 is consideredlone, mean percentage bare ground is significantly higher inlots which received pesticides than in those which did notF1,5 = 30.80, p < 0.005). Therefore, undersowing is likely toave precluded any vegetation effects in subsequent years.

Increased vegetation complexity has been shown to impedeoraging by some granivorous bird species, with bird abun-ances tending to be inversely correlated with weed biomassnd ground cover (Whittingham & Markland 2001; Butler

Gillings 2004). While a dense weed cover may hold largeood resources, time taken to locate food items may reduceotal energy gain per unit time. Increased vegetation structurean also reduce vigilance in species which rely on fast escaperom habitats such as finches and buntings. These speciesypically prefer to forage in short vegetation from which theyan escape quickly. Crypsis-reliant birds such as skylarks andartridges prefer to forage in high cover in which they canide from predators (Whittingham et al. 2006).

While not significant in the present study (overall use madey skylarks of the plots did not differ between undersownnd non-undersown years), vegetation structure is likely, ineneral, to be greater in organic stubble fields. While thishould not be problematic for skylarks, opting to forage inigher vegetation, other species may respond differently, forxample, yellowhammers.

esampling issues

An assumption with our approach is that on each visit alllots are potentially available to birds and that on the nextisit the birds have the opportunity to make a “fresh” choice.iven that all visits were spread apart by on average sevenays, this assumption seems reasonable, and is no different tossumptions made about bird use at the field scale by repeatedisits to the same field in other studies (e.g. Perkins et al.000; Moorcroft et al. 2002). What is less clear is how manyndividual birds were involved in our study. The maximumumber of skylarks observed in a single visit was 11 and forellowhammers 4. The same skylarks and yellowhammersay have been encountered time and again on different visits

o the plots but there is no way to test this formally. However,or yellowhammers at least, a small number (n = 4) of colour-inged birds were sighted foraging on the plots (ringed at theite as part of another project) and on no occasion was theame bird re-sighted. Autocorrelation is a potential problemor all repeat field visits but is likely to be more of an issueere given the high number of repeat visits. As described inhe Methods, this problem was addressed (at least partially)y re-analysing the skylark data (largest available dataset)ased on splitting the data into six groups and randomly pick-

ng one visit from early, mid and late winter in each of thehree years and creating results based on this re-analysis. This

ethod does make the assumption that over the longer-termi.e. between early, mid and late winter periods) the like-

Page 8

d Appl

lwctbapVsii2bvnmrs

C

fivghfoicEMpsaaGiv

A

tBtWg

A

cd

R

A

B

B

B

B

B

C

d

G

H

H

H

K

L

M

A.J. McKenzie et al. / Basic an

ihood that the same birds will be visiting the same plotsill be reduced, thus also reducing potential temporal auto-

orrelation. We do not have any of our own data to supporthis assumption, however it is known that while seed-eatingirds such as yellowhammers stay within a fairly restrictedrea (several hundred metres) when monitored over shorteriods of time (e.g. days, weeks) (Siriwardena, Calbrade,ickery, & Sutherland 2006), these and other granivorous

pecies tend to move extensively across the arable landscapen the longer term (e.g. months) in order to exploit chang-ng food resources over the winter period (e.g. Butler et al.010). The mean result after 50 randomisations was found toe non- but near-significant (p = 0.11), likely as a result of theastly reduced samples size (n = 9 vs. n = 32). Plots receivingo pesticides remain as most-used after resampling in all 50odels (see Results). Given these results, temporal autocor-

elation is therefore unlikely to change the overall pattern ofkylark distribution in the current experiment.

onclusions and further study

The design of the current experiment has allowed for therst time the effect of pesticide use on bird food items andegetation structure to be disentangled from the effect of inor-anic fertiliser use. In this instance, pesticide inputs clearlyave the largest effect on the variables measured, inorganicertilisers not appearing to negatively affect food abundancer diversity (at least for skylarks). Agri-environment pol-cy ought, therefore, to focus on reducing pesticide use inrops preceding stubble, an option recently included in thentry Level Stewardship Scheme (J. Vickery, pers. comm.).oreover, as overwinter food shortages have been linked to

opulation declines in a number of bird species (e.g. houseparrows; Hole et al. 2002), the observed increase in foodbundance provided by the NOPEST treatment may also beuseful aid in the conservation of these and other species.ood stubble management will be of even greater importance

n the future with the loss of set-aside, which was once aaluable winter foraging habitat for many granivorous birds.

cknowledgements

AJM was supported by a NERC Case studentship withhe British Trust for Ornithology. MJW was supported by aBSRC David Phillips Fellowship. We would like to thank

he staff at Nafferton Farm for their help with the trial plots.e would also like to thank Katy Goforth and Alice Love-

rove for help with data collection.

ppendix A. Supplementary data

Supplementary data associated with this arti-le can be found, in the online version, atoi:10.1016/j.baae.2010.10.007.

M

ied Ecology 12 (2011) 80–88 87

eferences

lbrecht, H. (2005). Development of arable weed seedbanks dur-ing the six years after the change from conventional to organicfarming. Weed Research, 45, 339–350.

all, A. (1992). Weed seedbank response to tillage, herbicides, andcrop rotation sequence. Weed Science, 40, 654–659.

uckingham, D. L., & Peach, W. J. (2006). Leaving final-cut grasssilage in situ overwinter as a seed resource for declining farmlandbirds. Biodiversity and Conservation, 15, 3827–3845.

utler, S. J., & Gillings, S. (2004). Quantifying the effects of habi-tat structure on prey detectability and accessibility to farmlandbirds. Ibis, 146(Suppl. 2), 123–130.

utler, S. J., Mattison, E. H. A., Glithero, N. J., Robinson, L. J.,Atkinson, P. W., Gillings, S., et al. (2010). Resource availabilityand the persistence of seed-eating bird populations in agricul-tural landscapes: A mechanistic modelling approach. Journal ofApplied Ecology, 47, 67–75.

utler, S. J., Whittingham, M. J., Quinn, J. L., & Cresswell,W. (2005). Quantifying the interaction between food densityand habitat structure in determining patch selection. AnimalBehaviour, 69, 337–343.

hamberlain, D. E., Fuller, R. J., Bunce, R. G. H., Duckworth, J.C., & Shrubb, M. (2000). Changes in the abundance of farm-land birds in relation to the timing of agricultural intensificationin England and Wales. Journal of Applied Ecology, 37, 771–788.

e Snoo, G. R. (1997). Arable flora in sprayed and unsprayed field-edges. Agriculture, Ecosystems & Environment, 66, 223–230.

illings, S., Newson, S. E., Noble, D. G., & Vickery, J. A. (2005).Winter availability of cereal stubbles attracts declining farm-land birds and positively influences breeding population trends.Proceedings of the Royal Society B, 272, 733–739.

ole, D. G., Whittingham, M. J., Bradbury, R. B., Anderson, G.Q. A., Lee, P. L. M., Wilson, J. D., et al. (2002). Wide-spreadlocal house sparrow extinctions – Agricultural intensification inblamed for the plummeting populations of these birds. Nature,418, 931–932.

otker, H., Jeromin, K., & Rahmann, G. (2004). Importance of win-ter stubble fields and/or green vegetation for farmland relatedbirds – Results of a survey on organic and conventional cropland of heavy soils in northern Germany. LandbauforschungVolkenrode, 54, 251–260.

yvönen, T., & Salonen, J. (2002). Weed species diversity and com-munity composition in cropping practices at two intensity levels:A six-year experiment. Plant Ecology, 159, 73–81.

oocheki, A., Nassiri, M., Alimoradi, L., & Ghorbani, R. (2009).Effect of cropping systems and crop rotations on weeds. Agron-omy for Sustainable Development, 29, 401–408.

ehesranta, S. J., Koistinen, K. M., Massat, N., Davies, H. V., Shep-herd, L. V. T., McNicol, J. W., et al. (2007). Effects of agriculturalproduction systems and their components on protein profiles ofpotato tubers. Proteomics, 7, 597–604.

cKenzie, A. J., & Whittingham, M. J. (2009). Why are birdsmore abundant on organic farms? Journal of Food, Agriculture& Environment, 7, 804–814.

oorcroft, D., Whittingham, M. J., Bradbury, R. B., & Wilson, W. D.(2002). The selection of stubble fields by wintering granivorousbirds reflects vegetation cover and food abundance. Journal ofApplied Ecology, 39, 535–547.

Page 9

8 d Appl

N

P

P

R

R

R

S

S

S

S

W

W

W

W

W

8 A.J. McKenzie et al. / Basic an

ewton, I. (2004). The recent declines of farmland bird popula-tions in Britain: An appraisal of causal factors and conservationactions. Ibis, 146, 579–600.

erkins, A. J., Maggs, H. E., & Wilson, J. D. (2008). Winter bird useof seed-rich habitats in agri-environment schemes. Agriculture,Ecosystems & Environment, 126, 189–194.

erkins, A. J., Whittingham, M. J., Bradbury, R. B., Wilson, J.D., Morris, A. J., & Barnett, P. R. (2000). Habitat character-istics affecting use of lowland agricultural grasslands by birds inwinter. Biological Conservation, 95, 279–294.

obinson, A., & Sutherland, W. J. (1999). The winter distributionof seed-eating birds: Habitat structure, seed density and seasonaldepletion. Ecography, 22, 447–454.

obinson, R. A., Hart, J. D., Holland, J. M., & Parrott, D. (2004).Habitat use by seed-eating birds: A scale-dependent approach.Ibis, 146(Suppl. 2), 87–98.

oschewitz, I., Gabriel, D., Tscharntke, T., & Thies, C. (2005). Theeffects of landscape complexity on arable weed species diversityin organic and conventional farming. Journal of Applied Ecology,42, 873–882.

hepherd, M., Pearce, B., Cormack, B., Phillips, L., Cuttle, S., Bho-gal, A., et al. (2003). An assessment of the environmental impactsof organic farming. Report to DEFRA, London, UK.

iriwardena, G. M., Calbrade, N. A., Vickery, J. A., & Sutherland,

W. J. (2006). The effect of the spatial distribution of winter seedfood resources on their use by farmland birds. Journal of AppliedEcology, 43, 628–639.

tephens, D. W., & Krebs, J. R. (1986). Foraging theory. Princeton,NJ: Princeton University Press.

W

ied Ecology 12 (2011) 80–88

uarez, F., Garza, V., Onate, J. J., Garcıa de la Morena, E. L.,Ramirez, A., & Morales, M. B. (2004). Adequacy of winterstubble maintenance for steppe passerine conservation in cen-tral Spain. Agriculture, Ecosystems & Environment, 104, 667–671.

akeham-Dawson, A., & Aebischer, N. J. (1998). Factors determin-ing winter densities of birds on environmentally sensitive areaarable reversion grassland in southern England, with special ref-erence to Skylarks (Alauda arvensis). Agriculture, Ecosystems& Environment, 70, 189–201.

eibull, A.-C., Ostman, O., & Granqvist, A. (2003). Species rich-ness in agro-ecosystems: The effect of landscape, habitat andfarm management. Biodiversity & Conservation, 12, 1335–1355.

hittingham, M. J., Devereux, C. L., Evans, A. D., & Bradbury,R. B. (2006). Altering perceived predation risk and food avail-ability: Management prescriptions to benefit farmland birds onstubble fields. Journal of Applied Ecology, 43, 640–650.

hittingham, M. J., & Markland, H. M. (2001). The influence ofsubstrate on the functional response of an avian granivore andits implications for farmland bird conservation. Oecologia, 130,637–644.

ilson, J. D., Morris, A. J., Arroyo, B. E., Clark, S. C., & Bradbury,R. B. (1999). A review of the abundance and diversity of inver-tebrate and plant foods of granivorous birds in northern Europein relation to agricultural change. Agriculture, Ecosystems &

Environment, 73, 13–30.

ilson, J. D., Taylor, R., & Muirhead, L. B. (1996). Field use byfarmland birds in winter: An analysis of field type preferencesusing resampling methods. Bird Study, 43, 320–332.