www.elsevier.com/locate/agee

Agriculture, Ecosystems and Environment 109 (2005) 59–74

Effects of field management and landscape context on

grassland wintering birds in Southern Portugal

Francisco Moreira a,*, Pedro Beja b, Rui Morgado a, Luıs Reino b,Luıs Gordinho b, Ana Delgado a, Rui Borralho b

a Centro de Ecologia Aplicada ‘‘Prof. Baeta Neves’’, Instituto Superior de Agronomia,

Tapada da Ajuda, 1349-017 Lisboa, Portugalb ERENA – Ordenamento e Gestao de Recursos Naturais, Av. Visconde Valmor,

11-38, 1000-289 Lisboa, Portugal

Received 4 June 2004; received in revised form 20 January 2005; accepted 15 February 2005

Abstract

This study examined the multi-scale effects of farmland management on the assemblages of grassland wintering birds in

Southern Portugal, evaluating the potential influences on bird conservation of ongoing transformations in agricultural

landscapes. The bird assemblages of 42 grassland fields were characterised and related to three sets of variables reflecting

field management, landscape context and large-scale spatial trends. Birds showed responses to the three sets of variables, with

variation partitioning indicating that species richness was primarily influenced by landscape context, whereas abundances were

mostly determined by field management. In general, high species richness was associated with small patches of arable land,

diverse landscapes, high stream density and forest and shrub cover, which probably acted as sources of non-agricultural species

to grassland fields. In contrast, fields located in homogenous arable landscapes tended to be species poor, though they were the

most favourable for open farmland species of conservation concern (e.g. skylark, calandra lark and little bustard). Old grazed

fallows were generally associated with the highest abundances of both invertebrate and seed-eating birds, particularly that of

winter visitors. Seed-eating winter visitors were also associated with stubble fields. The results indicated that conservation of

Iberian grasslands for wintering birds requires management at both the landscape and field scales. At the landscape level, large

blocks of open farmland habitat need to be maintained. At the field level, conservation management should strive to maintain

both stubble fields and old grazed fallows. The application of such prescriptions would require a combination of agri-

environmental programs and land planning policies, whereby the financial support to individual farmers should be conditional

on the maintenance of adequate landscape contexts at larger spatial scales.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Grasslands; Habitat management; Multi-scale; Pseudosteppes; Portugal; Wintering birds

* Corresponding author. Tel.: +351 21 3616083; fax: +351 21 3623493.

E-mail address: fmoreira@isa.utl.pt (F. Moreira).

0167-8809/$ – see front matter # 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.agee.2005.02.011

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7460

1. Introduction

In recent decades, major declines and range

contractions in European farmland birds have been

associated with changes in agricultural practices,

prompting a quest for sympathetic farming practices

and land-use policies (Bignal and McCracken, 1996;

Pain and Pienkowski, 1997; Stoate et al., 2001).

Although most research and conservation prescrip-

tions have initially been targeted at improving

breeding conditions (e.g. Potts, 1986), there is

growing evidence that changes in winter habitats also

affect farmland bird populations (Evans, 1997; Peach

et al., 1999). For instance, the reduction of winter

stubbles has contributed to the decreased abundance of

seed-eating passerines of conservation concern, which

are strongly dependent on the grain and arable weed

seeds provided by these habitats (Wilson et al., 1996;

Robinson and Sutherland, 2000; Moorcroft et al.,

2002). Likewise, invertebrate-feeding birds respond to

factors such as grassland age and manure application,

which influence food abundance and, consequently,

winter survival and population trends (Tucker, 1992;

Wilson et al., 1996; Perkins et al., 2000). Following

these studies, recommendations to improve winter

habitats have been incorporated into conservation

practice (e.g. Wakeham-Dawson and Aebischer,

1998). It is unclear, however, whether the current

prescriptions have general application, as the available

information derives almost exclusively from Northern

Europe. It therefore remains important to evaluate

elsewhere the processes by which wintering birds are

affected by agricultural change.

The Iberian Peninsula is a major winter quarter for

European migratory birds, receiving each year the

influx of huge numbers of both seed and invertebrate-

feeding farmland birds (Tellerıa, 1988). Although the

distribution and habitat preferences of these birds are

insufficiently known, there is evidence for their

association with low-intensity farming systems,

including extensive cereal cropland and agricultural

grasslands (Tellerıa, 1988; Suarez et al., 2003). In

particular, high densities of wintering birds have been

found in pseudosteppes, which are extensively

farmed, mixed rotational systems of winter cereals,

fodder crops and grazed fallow land, covering over

four million hectares in dry areas with low forest

cover (Suarez et al., 1997). Currently, the area of

pseudosteppes is declining, due to intensification in

the most productive land, compounded by agricul-

tural abandonment and afforestation in marginal areas

(Suarez et al., 1997; Stoate et al., 2001). These trends

are likely to reduce the quantity and quality of winter

habitats for both resident and migratory species, with

potential negative consequences for bird conservation

at the continental scale (Dıaz and Telleria, 1994;

Suarez et al., 1997, 2003). At present, however, little

information is available to predict the consequences

of ongoing agricultural changes and to devise

adequate agri-environmental management prescrip-

tions favouring birds wintering in Southern Europe.

This study addressed these issues by examining, at

multiple scales, the effects of agriculture on birds

wintering in a pseudosteppe landscape of Southern

Portugal. Specifically, the study focused on agricul-

tural grasslands (fallows and pastures), given their

spatial extent, vulnerability to changing farming

practices and value for both migratory and resident

species of conservation concern (Suarez et al., 1997;

Delgado and Moreira, 2000). In pseudosteppes,

typically 30–80% of the land is left fallow each year

and fallow fields remain uncultivated for 2–5 years (or

more), after which they are ploughed and sown to re-

initiate the rotation cycle (Suarez et al., 1997).

Intensification generally results in shorter rotations

and declines in the amount of land left fallow each

year, whereas abandonment is associated with longer

rotations and scrub encroachment (Suarez et al.,

1997). The study adopted a multi-scale approach,

recognising the need to focus on both the management

of fields and their context at larger spatial scales for

adequately assessing the effects of agriculture on

wildlife (Freemark, 1995). Therefore, the study

assessed the relative contributions of habitat variables

reflecting grassland management and landscape

context to variation in: (i) species richness and overall

bird abundance; (ii) richness and abundance of birds

grouped according to migratory status (residents

versus winter visitors) and feeding habits (seed eaters

versus invertebrate eaters); and (iii) relative abun-

dances of the most abundant species. Results were

then used to evaluate the impacts of expected

agricultural changes and to formulate agro-environ-

mental management recommendations for compar-

able grassland landscapes across the Iberian

Peninsula.

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 61

Fig. 1. Map of the study area, showing the main land uses and the location of sample sites (circles).

2. Methods

2.1. Study area

The study was conducted in a typical pseudosteppe

in Southern Portugal, within a 20 km � 30 km

rectangle encompassing a North–South gradient of

agricultural intensification–abandonment (Fig. 1).

This is a plain (100–300 m a.s.l.) with Mediterranean

climate, including hot summers (30–35 8C on average

in July), fairly cold winters (averaging 5 8C in

January) and >75% of annual rainfall (500–

600 mm) concentrated in October–March. An agri-

cultural landscape prevailed in the north, where the

terrain is flatter and the soils more productive, so the

proportion of land cultivated each year is high and

fallows are short-term, mostly consisting of stubble

fields. This landscape was in marked contrast with the

southern portion, a mosaic of shrubland interspersed

with old fallows resulting from agricultural abandon-

ment and scrub encroachment. In-between there was a

pastoral landscape, where the cultivation of cereals is

regular but with medium to long rotations, and so

grazed fallows predominantly occupy the arable land.

Throughout the region there are holm oak Quercus

rotundifolia Lam. woodlands of variable tree cover,

frequently with a grassy understorey grazed by

livestock. The forested areas are increasing, due to

afforestation of abandoned arable land with eucalyp-

tus Eucalyptus spp., umbrella pines Pinus pinea L. and

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7462

holm and cork oak Quercus suber L. The region was

designated as a Special Protection Area under the

European Union Birds Directive due to its importance

for steppe birds. In part of the area there is an agro-

environmental subsidy scheme, whereby farmers are

compensated for maintaining agricultural practices

favouring bird conservation.

2.2. Sampling design

To sample the whole variety of landscape types

along the North–South gradient, rough boundaries of

the three main types (agricultural, pastoral and

shrubland landscapes) were delineated using satellite

imagery and field surveys, and 14 grassland fields

were randomly selected within each type. To reduce

eventual dependencies due to similarities in farm

management and spatial autocorrelation, the stratified

random sampling was constrained so that the selected

fallow fields were on different farms and at least

500 m apart from each other. Fields were also required

to be over 5 ha, with shrub cover <20% and tree cover

<5%. Sampling in large fields (>10 ha) was restricted

to 10 ha quadrats defined at their geographical centres.

From the 42 fields thus selected, 30 were sampled in

the Winter of 1999–2000, and 12 in the winter of

2000–2001.

2.3. Bird censuses

Birds were counted five times in each field, at about

2 week intervals, between December and mid

February. In each case, the entire field was walked

so that the observer approached to within 25 m or less

of every point. All individual birds and flocks of each

species were identified and recorded on maps and,

when flushed, care was taken to check if they landed

further ahead to minimize double counting. To aid in

the recording of bird observations, detailed maps

(1:2000) of each field were prepared before sampling,

depicting the position of marking poles and note-

worthy topographical features (e.g. dirt tracks, stone

piles and shrub patches). Due to difficulties in reliably

identifying all birds of some species in the field,

categorisation to the genus level was made for the

crested and thekla larks (Galerida cristata L. and

Galerida theklae Brehm), house and Spanish sparrows

(Passer domesticus L. and Passer hispaniolensis

Temm.) and common and spotless starlings (Sturnus

vulgaris L. and Sturnus unicolor Temm.). Birds flying

over but not landing were not counted; swallows and

birds of prey were not considered in this study. The

mean count over the five visits (birds per 10 ha) was

used to index the relative abundance of each species at

each field. This measure had potential shortcomings,

due to the winter flocking behaviour of many bird

species (Donald et al., 2001). However, approaches

based on field occupancy or frequency of occurrence

over visits were considered to be less adequate in the

present case, given the need to differentiate between

sites where a species occurs frequently but in low

numbers, from sites where it is both frequent and

abundant. For most species, however, there was strong

positive correlation between mean counts (as log

[x + 1]) and frequencies of occurrence, suggesting that

eventual effects of aggregation were unlikely to affect

the results seriously. For subsequent analysis, birds

were grouped according to both migratory status

(residents versus winter visitors) and feeding habits

(seed versus invertebrate eaters), as these character-

istics may influence the use of winter habitats (e.g.

Wilson et al., 1996; Perkins et al., 2000). Winter

visitors were regarded as species absent from grass-

land habitats during the breeding season (Delgado and

Moreira, 2000).

2.4. Explanatory variables

Factors influencing wintering birds were analysed

using three sets of explanatory variables reflecting

field management, landscape context and spatial

(geographical) location. The spatial component was

included to account for the possibility of unmeasured

environmental factors underlying large-scale trends in

assemblage variation (Legendre and Legendre, 1998).

Fields were described from seven variables

characterising grazing intensity, grassland age and

vegetation structure (Table 1). Estimates of grazing

intensity were obtained by recording at each field the

frequency of occurrence of livestock (mostly sheep)

over the five bird counts. The time in years since the

last cultivation or ploughing of each grassland field

was assessed from inquiries to landowners, assigning

an age of zero to recently ploughed or stubble fields.

Mean vegetation height was estimated in January,

from 60 evenly spaced measurements taken along a

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 63

Table 1

Description and summary statistics of field management and landscape context variables used to characterise fallow fields sampled in Southern

Portugal (see text for details)

Variable Description Mean S.D. Range

Field management variables

Stubbles (Stub) Dummy variable for the occurrence of stubbles 0.31 0.47 0–1

Shrubs (Shrub) Dummy variable for the occurrence of shrubs 0.43 0.50 0–1

Trees (Tree) Dummy variable for the occurrence of trees 0.43 0.50 0–1

Grazing intensity (Graz) Livestock grazing frequency 0.95 1.23 0–5

Vegetation height (Vegh) Mean vegetation height (cm) 5.13 2.75 1.3–15.0

Bare ground (Bare) (%) Proportion bare ground 0.07 0.07 0–0.27

Fallow age (Age) Time since last ploughing or cultivation (years) 1.88 2.53 0–10

Landscape context variables

Pseudosteppe (Steppe) (%) Proportion of area within 1-km radius with pseudosteppe 0.58 0.27 0.07–1.00

Shrubsteppe (Shrstp) (%) Proportion of area within 1-km radius with shrubsteppe 0.13 0.17 0–0.56

Open pastoral woodland (Wood) (%) Proportion of area within 1-km radius with pastoral woodland 0.05 0.11 0–0.37

Forest and scrub (Forscr) (%) Proportion of area within 1-km radius with forest and scrub 0.21 0.17 0–0.65

Other (Other) (%) Proportion of area within 1-km radius with other uses 0.01 0.01 0–0.08

Land use diversity (Divers) Number of different land use types within 1-km radius 3.31 0.81 1–5

Arable patch size (Patch) Size (ha) of pseudosteppe patch where field plot is included 95.2 69.8 6.7–234.1

Drainage density (River) Length (km) of drainage lines within 1-km radius 12.1 2.2 6.6–17.9

Game (Game) Dummy variable for the occurrence of game management 0.67 0.48 0–1

Agro-environmental (Agroenv) Dummy variable for the occurrence of

agro-environmental management

0.38 0.49 0–1

transect crossing the longest axis of each field. Height

was estimated to the nearest centimetre with a ruler

and was defined as the highest point of vegetation

projection, for plants within 3 cm, on the ruler

(Moreira, 1999). Cover by bare ground was estimated

as the proportion of the 60 measurements with no

vegetation.

Landscape context was described from 10 variables

reflecting landscape composition, configuration and

management regime (Table 1). Proportions of five land

use categories were determined in a geographic

information system (GIS), within a 1 km-radius circle

centred on each sampling fallow, from maps prepared

using aerial photography and field surveys. This radius

was set large enough to encompass different land uses

despite the relative homogeneity of the landscape,

while remaining within the range of scales to which

farmland birds are responsive (e.g. Soderstrom and

Part, 2000; Best et al., 2001). The following land use

categories were considered: (a) pseudosteppes were

continuous expanses of open arable land (including

fallows, pastures, cereal and other dry crops and

ploughed fields) with virtually no trees and shrubs; (b)

shrubsteppes were largely similar, but with many

small patches of scrub (typically 0.5–1 ha) inter-

spersed in the arable land; (c) pastoral woodland was a

kind of open savanna (‘‘montado’’), with scattered oak

trees (<10% tree cover) and a grassy understorey

grazed by livestock; (d) areas of scrub and forest

habitats were clumped in a single category, as they

were considered largely unsuitable for farmland birds;

(e) rare habitats (e.g. irrigated crops, rivers and dams,

urban) were grouped in an ‘‘other’’ category. Land

cover maps were also used to determine the size (ha)

of the continuous patch of arable land (pseudosteppe)

where each sampled fallow was inserted. The length of

drainage lines (km) was estimated from topographic

maps (scale 1:25,000). Game management and the

inclusion of the farm in the agro-environmental

subsidy scheme were assessed from inquiries to

landowners and coded as dummy variables.

Spatial trends were quantified following Legendre

and Legendre (1998), using terms of a full third-order

polynomial equation of the x and y Cartesian

coordinates of field centres. The x and y terms were

centred to zero mean before computing the matrix of

spatial descriptors, to reduce collinearity between

successive terms when fitting the polynomial. Quad-

ratic, cubic and interaction terms were explored to allow

for complex spatial features like patches and gaps.

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7464

2.5. Data analyses

Prior to analysis, log-transformations were used to

reduce the influence of extreme values in bird species

richness and abundance data, and in explanatory

variables showing skewed distributions (VEGH, AGE,

PATCH, RIVER). For proportional data, the angular

transformation was used likewise thus reducing

skewness and overcoming the unity sum constraint.

Because environmental data were strongly intercorre-

lated, the main gradients of variation in field

management and landscape context were described

from separate principal components analyses (PCA;

Legendre and Legendre, 1998).

Linear regression was used to estimate the field,

landscape and spatial variables influencing bird

species richness and abundance, with partial regres-

sion partitioning explained variation among the three

sets of explanatory variables (Legendre and Legendre,

1998). The procedure included separate stepwise

regression analyses for each group of factors, to select

reduced subsets of variables that maximally accounted

for variation in each response variable. Before

stepwise selection, inter-annual variation was con-

trolled by forcing a covariate coding the sampling year

into the model. In common with similar studies (e.g.

Estades and Temple, 1999; Freemark and Kirk, 2001;

Gutzwiller and Barrow, 2001), stepwise models

retained both significant variables (P < 0.05) and

variables approaching significance (0.05 � P < 0.10),

so as to reduce the incidence of Type II errors and

thereby to avoid missing eventual effects with

conservation relevance (e.g. Buhl, 1996; Underwood,

1997). Variables included in the stepwise models were

used in partial regressions following Anderson and

Gribble (1998), thus estimating the amount of

variation (R2) of the response variable that can be

attributed to one set of factors, once the effect of other

set of factors is taken into account. This method

isolated seven components of explained variation: (i)

pure field variation; (ii) pure landscape variation; (iii)

pure spatial variation; (iv) combined field/landscape

variation; (v) spatial component of field variation; (vi)

spatial component of landscape variation; (vii) spatial

component of combined field/landscape variation.

Factors underlying variation in the matrix of

species abundances were examined taking a similar

approach, using redundancy analysis (RDA) and

partial-RDA (Legendre and Legendre, 1998). A linear

model of ordination was used because preliminary

detrended correspondence analysis (not presented

here) showed turnovers along the first axes <2 S.D.,

which is the recommended criterion for choosing

linear versus uni-modal models (ter Braak, 1995).

RDAwere carried out on covariance matrices to weigh

each species by the variance of its log-transformed

abundance, thus preventing the undue influence of rare

species (ter Braak, 1995). Species occurring in less

than five fields were excluded from the RDA.

Statistical significance of variables and RDA models

were assessed from Monte Carlo permutation tests

(103 permutations).

3. Results

3.1. Overall patterns in field management and

landscape context

Descriptive statistics of field and landscape

variables are shown in Table 1. The PCA on field

variables revealed a dominant gradient related to

grassland age, ranging from stubble fields to old

fallows, which accounted for 27.1% in local habitat

variability (Fig. 2a). The second PC axis (21.0%)

represented a gradient of grazing intensity and

vegetation structure, ranging from frequently grazed

fields with shorter vegetation, higher proportion of

bare ground and absence of shrubs, to ungrazed or

rarely grazed fields, with taller vegetation, shrub

patches and less bare ground cover.

The landscape PCA underlined a dominant

gradient of increasing habitat diversity, with the

first axis accounting for 37.9% of the overall

landscape variability and contrasting fields within

landscapes composed almost exclusively by arable

land, with fields located in landscapes with high

proportions of forest and shrubland (Fig. 2b). There

was also a trend for farms involved in the agri-

environmental subsidy scheme to be associated with

homogeneous arable landscapes, whereas game

management tended to occur in diverse landscapes.

The second PC (15.6%) contrasted landscapes with

higher proportion of open oak woodlands, with

landscapes with higher amounts of arable land

interspersed with shrub patches.

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 65

Fig. 2. Ordination diagrams of principal component analysis (PCA)

of habitat variables characterising 42 grassland fields sampled in

Southern Portugal for wintering bird assemblages. Separate PCAs

were carried out for variables reflecting field management (a) and

landscape context (b).

3.2. Bird assemblage composition

Altogether, 47 species were recorded during the

study (Appendix A), with an average of 13.5 � 3.6

species per field and 85.2 � 54.6 birds per 10 ha. Most

species were resident invertebrate-eaters (39.6%) and

seed-eaters (22.9%). In terms of abundance, the

assemblage was clearly dominated by seed-eaters,

including both winter visitors (41.1%) and resident

(23.9%) species. There was a core group of seven

widespread (>50% occurrence) and abundant

(>4 birds per 10 ha) species, which accounted for

81.7% of the birds counted. This group was dominated

by winter visitors, including both seed-eaters (skylark

Alauda arvensis L., linnet Carduelis cannabina L. and

goldfinch Carduelis carduelis L.) and invertebrate-

eaters (lapwing Vanellus vanellus L., golden plover

Pluvialis apricaria L. and meadow pipit Anthus

pratensis L.), but there was also one abundant resident

seed-eater (corn bunting Miliaria calandra L.).

3.3. Multi-scale effects on species richness

and abundance

The field models underlined a strong tendency for

the overall species richness and that of invertebrate-

eating residents and winter visitors to be highest in

fallows with shrubs, with tree presence also affecting

positively the species richness of resident invertebrate-

eaters (Table 2). The most consistent field effects on

abundance were fallow age, occurrence of stubbles and

grazing intensity. Overall abundance and that of seed-

eating winter visitors peaked in both stubble fields and

old fallows, with the former also responding positively

to grazing intensity. Invertebrate-eating winter visitors

were most abundant in old grazed fallows.

Species richness was generally lower in landscape

contexts dominated by arable habitats, with more

species occurring in fallows located in smaller arable

patches, with higher habitat diversity, higher coverage

by shrubsteppe, forests and scrub and increased

drainage density (Table 2). Landscape effects on

abundance were more varied. For example, resident

invertebrate-eaters were most abundant in diverse

landscapes, with small arable patches and reduced

coverage by pseudosteppe. Conversely, invertebrate-

eating winter visitors favoured landscapes with a high

proportion of shrubsteppe, low habitat diversity and

little coverage by pastoral woodland. Agro-environ-

mental management showed positive effects on the

abundance of resident species. Game management

favoured the abundance of wintering invertebrate-

eaters, but affected negatively the resident species.

Mapping of each polynomial selected in spatial

models (not shown herein) indicated simpler spatial

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7466T

able

2

Imp

ort

ance

of

exp

lan

ato

ryvar

iab

les

inli

nea

rre

gre

ssio

n(r

ich

nes

san

dab

un

dan

ce)

and

RD

A(a

bu

nd

ance

sm

atri

x)

mod

els

for

gra

ssla

nd

win

teri

ng

bir

ds

inS

outh

ern

Po

rtu

gal

Fie

ldm

od

els

Lan

dsc

ape

Mo

del

sS

pat

ial

Mo

del

s

R2

Sh

rub

Tre

eA

ge

Stu

bG

raz

R2

Ste

pp

eS

hrs

tpF

ors

crW

oo

dO

ther

Div

ers

Pat

chR

iver

Ag

roen

vG

ame

R2

Po

lyn

om

ial

Ric

hn

ess

Over

all

29.8

<0.

001

(+)

40

.00

.00

7(�

)0

.07

7(+

)0

.01

2(+

)0

.02

6(+

)<

0.00

1(�

)0.

046

(+)

33

.5y3

�y

Mig

rato

ryin

ver

t9

.10.

053

(+)

8.1

0.06

9(+

)0

.07

9(�

)3

0.9

�y

Res

iden

tin

ver

t2

8.0

0.00

2(+

)0.

057

(+)

33

.60

.00

9(�

)0

.01

1(+

)<

0.00

1(�

)0

.04

9(+

)0.

062

(�)

10

.5�

y

Mig

rato

ryse

ed–

18

.10

.02

7(+

)0.

057

(�)

0.07

3(+

)9

.5�

x2y

Res

iden

tse

ed–

12

.20.

058

(�)

0.08

4(�

)1

3.4

�y2

Abu

nd

ance

Over

all

28.9

0.00

2(+

)0.

014

(+)

0.07

1(+

)7

.20.

088

(�)

22

.0y�

y2�

y3+

xy2

Mig

rato

ryin

ver

t2

2.0

0.00

6(+

)0.

071

(+)

23

.30.

005

(+)

0.0

19

(�)

0.02

2(�

)0.

098

(+)

16

.9�

y3�

xy+

x2y

Res

iden

tin

ver

t3

1.7

0.00

1(+

)0.

043

(+)

38

.80

.03

6(�

)0

.09

4(+

)<

0.00

1(�

)0.

063

(+)

0.01

5(�

)1

2.2

�y

Mig

rato

ryse

ed2

0.8

0.00

3(+

)0.

014

(+)

10

.10.

040

(�)

0.0

89

(+)

30

.3y�

y2�

y3+

xy2

Res

iden

tse

ed–

19

.10.

009

(+)

22

.5�

x2�

y2

Mat

rix

of

bir

dre

lati

ve

abun

dan

ces

Over

all

21.1

0.01

10.

002

<0.

001

0.00

91

0.6

0.0

28

0.0

44

0.03

10.

006

0.0

96

12

.5x

+y

+xy

Dir

ecti

ons

of

asso

ciat

ion

(�o

r+

)an

dP

-lev

els

are

giv

enfo

rvar

iab

les

sig

nifi

can

tly

(P<

0.0

5)

or

nea

rsi

gn

ifica

ntl

y(0

.05<

P<

0.1

0)

rela

ted

toth

ere

sponse

var

iable

s.V

aria

ble

sin

bold

wer

ein

cluded

inth

ebes

tm

ult

ivar

iate

model

s;it

alic

s

ind

icat

evar

iab

les

show

ing

on

lyu

niv

aria

tere

lati

onsh

ips.

Th

ep

oly

no

mia

lte

rms

are

tho

sein

clu

ded

inth

eb

est

spat

ial

mo

del

s.T

he

amo

un

to

fex

pla

ined

var

iati

on

(R2)

isg

iven

for

each

mo

del

.S

eeT

able

1fo

rd

efinit

ion

of

var

iable

s.

trends in richness than in abundance models (Table 2).

The main tendency for species richness was an

increase from north to south, apparent for both the

overall assemblage and each of the functional groups

except resident seed-eaters. The latter peaked at the

central region, declining to both the north and the

south. The spatial abundance models often reflected

the presence of patches of high abundances, differing

in location among functional groups.

Partitioning of explained variation among field,

landscape and spatial sets of variables showed marked

differences between species richness and abundance

models, and among functional groups (Table 3). In

richness models, the landscape context accounted for

more of the explained variation than field character-

istics. The importance of field characteristics

increased in abundance models, contributing more

than landscape variables to explained variation in total

abundance and abundance of winter visitors. Spatial

trends showed unique significant contributions to

explained variation in richness and abundance of seed-

eaters and richness of invertebrate-eating winter

visitors. In general, a relatively large proportion of

variation was explained by spatially structured effects

(shared components), precluding a more exact

isolation of either field or landscape effects. The

negative values obtained for some shared components

indicate that one set of variables showed an inverse

effect of that of the other set.

3.4. Multi-scale effects on assemblage structure

The redundancy analysis showed several associa-

tions between assemblage structure and field, land-

scape and spatial sets of variables (Table 2; Fig. 3).

The first RDA axis (RDA1) of the field model reflected

a gradient from recent fallows, with scattered trees and

little grazing, with higher relative abundances of

woodlark Lululla arborea L., greenfinch Carduelis

chloris L., chaffinch Fringilla coelebs L. and serin

Serinus serinus L., to old grazed fallows with no trees,

where the abundance of golden plover, lapwing,

skylark, calandra lark Melanocorypha calandra L.,

starlings and goldfinch increased. The RDA2 primar-

ily contrasted a large group of species favouring

fallows with shrubs or trees (e.g. crested/thekla larks,

chiffchaff Phylloscopus collybita L., stonechat Sax-

icola torquata L., Sylvia warblers, greenfinch and

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 67

Fig. 3. Biplots from redundancy analysis (RDA1/RDA2) of the responses of wintering bird assemblages to sets of variables describing field

management, landscape context and large-scale spatial trends, in 42 grassland fields sampled in Southern Portugal. Diagrams show the overall

effects of each set of variables (RDA models), and the effects of each set of variables after partialling out the effects of the remaining two sets of

variables (partial RDA models). Dotted arrows are the environmental variables (see codes in Table 1) and full-line arrows are the bird species (see

codes in the Appendix A). Only species with scores >0.2 are shown.

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7468

Table 3

Partitioning among the pure and combined effects of field (F), landscape (L) and spatial (S) sets of variables, of explained variation in linear

regression (richness and abundance) and abundances matrix (RDA) models for grassland wintering birds in Southern Portugal

Pure components Shared components Unexplained

Field Landscape Spatial F + L F + S L + S F + L+ S

Species richness

Overall 10.8 (0.005) 15.7 (0.004) 1.1 ns �2.6 3.9 9.2 17.1 44.8

Migratory invert 1.6 ns 0.4 ns 14.4 (0.060) �0.1 7.3 6.6 2.3 67.5

Resident invert 10.5 (0.045) 14.6 (0.015) 0.00 ns 9.5 0.9 2.4 7.3 54.8

Migratory seed – 14.8 (0.031) 3.8 ns – – 7.6 – 73.7

Resident seed – 11.4 (0.075) 13.1 (0.016) – – �0.2 – 75.7

Abundance

Overall 15.6 (0.040) 2.2 ns 10.6 ns 0.3 6.3 0.0 4.8 60.2

Migratory invert 10.7 (0.049) 6.2 ns 2.8 ns �2.0 �3.6 6.6 9.5 69.8

Resident invert 7.0 ns 17.7 (0.013) 2.2 ns 14.9 4.4 0.4 5.0 48.4

Migratory seed 8.1 (0.082) 5.4 (0.067) 17.3 (0.038) 2.9 12.1 3.9 �2.2 52.5

Resident seed – 8.1 (0.041) 14.2 (0.028) – – 7.5 – 70.2

Matrix of bird relative abundances

Overall 20.0 (0.002) 6.6 ns 12.1 (0.062) 0.9 �1.6 1.4 1.2 59.4

P-levels for pure components are given in brackets.

chaffinch), with a few species associated with open

fields without either shrubs or trees (little bustard

Tetrax tetrax L., calandra lark, linnet). After control-

ling for landscape context and spatial trends, the

significant effects of fallow age and grazing intensity

remained largely comparable (Fig. 2). In contrast, the

presence of shrubs and trees lost significance in the

partial RDA, indicating that their influence could not

be disentangled from that of other sets of variables.

The landscape model included the significant effects

of arable patch size and cover by open pastoral

woodland. Along the gradient defined by RDA1, there

was increasing abundance of species favouring large,

continuous expanses of open arable land (little bustard,

skylark, calandra lark and corn bunting), and decreas-

ing abundance of species benefiting from the contact

with other land uses (e.g. red-legged partridge Alectoris

rufa L., crested/thekla lark, black redstart Phoenicurus

ochruros Gmelin and chaffinch). In RDA2, increasing

cover by open pastoral woodland was associated with

increasing abundance of woodlark, serin, chaffinch and

blue tit Parus caeruleus L., and decreasing abundance

of golden plover, lapwing, red-legged partridge,

skylark, calandra lark and goldfinch. No landscape

variable remained significant after controlling for the

effects of field characteristics and spatial trends

The spatial RDA highlighted a significant spatial

trend, with positive scoring sites in the RDA1

distributed in the study area across a band running

north-west to south-east, corresponding to fallows

with higher abundances of golden plover, lapwing,

red-legged partridge, robin Erithacus rubecula L.,

southern grey shrike Lanius meridionalis Temminck,

crested/thekla lark, Dartford warbler Sylvia undata

Boddaert and song thrush Turdus philomelos Brehm.

Sites with high negative scores along RDA1 were

primarily on the north-east corner, indicating higher

relative abundances of hoopoe Upupa epops L.,

woodlark, white wagtail Motacilla alba L. and blue tit.

The RDA2 reflected basically a east-west gradient,

along which there was increasing abundance of

skylarks and calandra larks, and declining abundances

of red-legged partridge, crested/thekla lark, white

wagtail, stonechat, starlings and linnet. After control-

ling for the effect of field characteristics and landscape

context, only the east-west spatial gradient remained

close to significance.

Partial RDA indicated that the three sets of

variables together accounted for 40.6% of variation

in the bird species data matrix. Of this explained

variation, 20.0% was uniquely related to field

characteristics, 6.6% to the landscape context and

12.1% to large-scale spatial trends (Table 3). However,

only the unique contribution of field was significant,

though the spatial component was close to signifi-

cance. The proportion of variation shared among

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 69

variables was always small (<2%). The negative value

of shared variation between field and spatial variables

indicates that these two sets have contradictory

influences on the assemblage.

4. Discussion

The wintering birds using agricultural grasslands in

pseudosteppe landscapes of Southern Portugal were

numerically dominated by a core group of widespread

and abundant species, which are commonly recorded

wintering in open farmland habitats of the Iberian

Peninsula (Tellerıa, 1988). Except for resident corn

buntings, this core group comprised species that are

largely absent from these grasslands during the

breeding season (Delgado and Moreira, 2000),

including both seed-eaters (skylark, linnet and gold-

finch) and invertebrate-eaters (lapwing, golden plover

and meadow pipit). Some of these birds probably

originated from northern latitudes (Wernham et al.,

2002), profiting from the abundance of seeds and

invertebrates that are seasonally available in agricul-

tural grasslands of Southern Iberia (Tellerıa, 1988;

Dıaz and Telleria, 1994; Suarez et al., 1997). Species

such as the linnet and the goldfinch probably included

also birds breeding in nearby habitats, which use

agricultural grasslands only during the winter. Density

estimates were similar to the ones recorded by other

authors in the Iberian Peninsula, supporting the

general trend for higher densities of granivorous birds

in farmland in relation to shrublands and forests,

probably due to the higher food availability (Tellerıa,

1988; Santos and Tellerıa, 1985).

4.1. Landscape and field management effects

Wintering birds showed responses to local field

management, landscape context and large-scale

spatial trends, underlining the value of investigating

multiple spatial scales to understand the effects of

agriculture on bird populations. Although this has

been recognised previously (e.g. Milson et al., 1998;

Soderstrom and Part, 2000; Best et al., 2001), there

have been very few attempts to isolate the unique

contributions of different scales to bird–agriculture

relationships (but see Freemark and Kirk, 2001). The

variation partitioning approach adopted in here

statistically isolated such unique contributions, show-

ing local variation in species richness to be primarily

influenced by landscape variables, whereas abun-

dances were primarily related to local field conditions.

These results probably reflect the hierarchical nature

of assemblage organisation in which the set of species

available to colonise a site results from factors

operating at large scales, whereas the actual species

abundances are largely influenced by local conditions

(Gaston and Blackburn, 2000).

After removing local and landscape effects, there

was still significant spatial structure in assemblage

variation, which may be a consequence of unmea-

sured, spatially structured field or landscape effects, or

alternatively, may relate to unmeasured environmental

variables or population processes acting at scales

above that normally considered in studies assessing

bird–agriculture relationships (e.g. Soderstrom and

Part, 2000; Best et al., 2001).

The shared components of explained variation were

largely caused by correlations among sets of variables

measured at different scales, making it impossible to

disentangle the corresponding scale-dependent

effects. For instance, tree cover influenced bird

assemblages at both the local and landscape levels,

but their effects were difficult to tell apart because the

presence of trees in fields was greatest in the most

wooded landscapes. Likewise, spatially structured

variation in field management and landscape context

caused major overlaps between the spatial component

and both the field and landscape sets of variables.

Species richness was generally highest in diverse

landscapes, with high stream density and great cover

by woods and shrub vegetation, which favoured the

local occurrence of species characteristic of these

habitats (e.g. woodlark, serin, chaffinch and blue tit).

Accordingly, richness was smallest in fields sampled

within the largest arable patches, probably because

they were farther away from forests and shrubland.

These patterns suggest that the observed landscape

effects could at least partly be attributed to the

distribution and abundance of non-agricultural habi-

tats acting as sources of species to grassland fields.

These results agree with studies suggesting that the

richest farmland bird assemblages are associated with

diverse landscapes, including woods, hedgerows and

other non-agricultural habitats (e.g. Freemark and

Kirk, 2001).

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7470

The composition and configuration of landscapes

also influenced bird assemblages through effects on the

core group of species typical of open farmland. These

species were positively associated with continuous

expanses of arable land and less wooded landscapes

(skylark, calandra lark, golden plover, lapwing and corn

bunting). These results agree with observations show-

ing that many grassland species such as plovers,

lapwings and skylarks avoid small fields, enclosed by

hedgerows or tree lines (Tucker, 1992; Milson et al.,

1998; Perkins et al., 2000; Donald et al., 2001). Inverse

relationships between shrub cover and densities of

wintering seed-eating birds were reported by Dıaz and

Telleria (1994). These patterns are frequently inter-

preted as resulting from anti-predator responses of birds

(e.g. Donald et al., 2001). Another potential explana-

tion to the patterns observed, though not necessarily

exclusive, may be related to variation in food supply

among landscape types. It is possible that large,

continuous patches of arable land provided increased

foraging opportunities to farmland birds, thus permit-

ting higher local densities than landscapes with a larger

proportion of non-agricultural habitats (Soderstrom and

Part, 2000). Whatever the ultimate causes, these results

support the view that the encroachment of trees and

shrubs into open habitats is a type of fragmentation that

can be detrimental to grassland specific birds (Cop-

pedge et al., 2001; Fuhlendorf et al., 2002)

There was a tendency for the lowest richness and

abundance of resident invertebrate-eaters to be found

in game estates. This may eventually be related to

predator control, as this group includes corvid species

(carrion crow, raven, magpie, azure-winged magpie)

often killed by gamekeepers. Agro-environmental

management showed consistent positive effects on the

abundance of resident birds, both seed-eaters and

invertebrate-eaters, but it had no effect on winter

visitors. This may be because management prescrip-

tions in this region have been largely targeted at the

breeding season (e.g. harvesting schedule, restrictions

in agro-chemicals and livestock densities, planting of

wildlife crops), though some of the measures also

improve winter habitats (e.g. preservation of stubble

fields). These measures were previously shown to

affect breeding birds positively (Borralho et al., 1999).

At the field scale, old fallows regularly grazed by

livestock were associated with the highest abundances

of both invertebrate-eaters (e.g. lapwing and golden

plover) and seed-eaters (e.g. skylark, calandra lark and

goldfinch), particularly that of winter visitors. The

positive association of invertebrate-feeding birds with

permanent pastures and long-term leys has been

reported in many previous studies, likely reflecting

particularly favourable feeding conditions (e.g. Fuller

and Youngman, 1979; Tucker, 1992; Wilson et al.,

1996). Invertebrate abundance was probably highest in

old fallows because cultivation reduces invertebrate

soil populations, which may then take time to recover

after the fields are left uncultivated (Tucker, 1992).

Grazing causes soil disturbance and creates a patchy

and open sward, which encourages greater inverte-

brate abundance and diversity, and also provides easier

access to the soil surface for feeding birds (Ausden

and Treweek, 1995; Wilson et al., 1996; Vickery et al.,

2001). Furthermore, dung creates localized sources of

invertebrates and promotes sub-surface invertebrate

populations (Tucker, 1992; Vickery et al., 2001).

The positive association of several seed-eating

species with grazed old fallows may be regarded as

unexpected, as these birds are usually perceived to

avoid grasslands and old fields (Wilson et al., 1996;

Dıaz and Telleria, 1994). Furthermore, this study

confirmed only partly the strong preferences reported

elsewhere of seed-eaters for stubble fields (Wilson

et al., 1996; Dıaz and Telleria, 1994; Moorcroft et al.,

2002), as these habitats showed a positive influence on

the abundance of seed-eating winter visitors but not on

the overall assemblage composition. These results

probably reflect patterns in the availability of food

resources. In the Iberian Peninsula, food accessibility

may be more important than the actual seed abundance

in influencing the distribution of wintering seed-eaters

because their food requirements are one order of

magnitude lower than the abundance of seeds in

extensive farmland habitats (Dıaz and Telleria, 1994).

Therefore, although stubbles attract seed-eaters

because they usually provide a rich source of spilt

crop product and the seeds of grasses and arable

weeds, there may be other adequate feeding grounds.

In old fallows grazed by sheep, the short and open

swards with large proportion of bare ground may be

particularly favourable feeding habitats, as they

provide easy access to food resources for species

such as skylarks (Wakeham-Dawson and Aebischer,

1998; Perkins et al., 2000). In summary, the results

suggest that grazed old fallows may be at least as

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 71

important as stubble fields to sustain populations of

wintering seed-eating birds in the Iberian Peninsula.

4.2. Implications for conservation

Results from this study strengthen the view that

Iberian agricultural grasslands are important for a wide

range of wintering birds typical of open farmland,

including migrant species such as skylarks that have

declined markedly in their breeding quarters. Further-

more, these habitats are important for resident species

with unfavourable conservation status, such as

calandra larks and little bustards (Moreira, 1999;

Delgado and Moreira, 2000, this study). Therefore,

conservation management of these grasslands should

have the primary objective of improving habitat

conditions for these open farmland species, both

migratory and resident. One major obstacle to this goal

is the ongoing scrub encroachment associated with

agricultural abandonment and forest plantation in

agricultural land, encouraged through European Union

subsidy schemes (e.g. Suarez et al., 1997). Although

these processes will increase local diversity of

breeding birds, the conservation value actually declines

because farmland species of unfavourable conserva-

tion status are replaced by widespread shrubland and

forest species (Dıaz et al., 1998; Santos, 2000; Suarez-

Seoane et al., 2002). Comparable losses were

documented in this study for wintering birds, with

negative responses of farmland birds to the prevalence

of non-agricultural habitats over a range of scales, from

the local presence of individual trees and shrubs in

grassland fields, to the landscape occupation by forests

and shrubland. Accordingly, the preservation of large

blocks of open farmland habitat should be a priority in

pseudosteppe landscapes, thus avoiding the negative

effects on open grassland birds of habitat loss and

fragmentation resulting from scrub encroachment and

afforestation (Herkert, 1994; Coppedge et al., 2001;

Fuhlendorf et al., 2002). Definition of minimum areas

required for the different bird species should be the

subject of future studies (e.g. Herkert, 1994).

Conservation management should also strive to

maintain a patchwork of both stubble fields and old

grazed fallows. Maintenance of stubble fields can

probably be made with the support of agro-environ-

mental subsidy schemes similar to that available in the

study area, which have helped maintaining a rotation

scheme including stubbles, and preventing stubble

burning after harvest to increase soil fertility. Old,

grazed fallows may be more at risk due to the

abandonment of the less productive land and the

consequent scrub encroachment. The preservation of

these habitats is strongly dependent on the subsidies

for livestock production, and thus can disappear very

quickly if they become unavailable or strongly

reduced. Changes in the livestock subsidy scheme

are already affecting the farming system in some

areas, with a trend for an increasing importance of

cattle at the expense of sheep production. The effects

of such changes should be evaluated, as there are

usually marked differences in sward structure and

floristic composition between pastures grazed by

cattle and sheep (Vickery et al., 2001).

This study adds to growing evidence supporting the

value of explicitly addressing conservation manage-

ment over multiple spatial scales (e.g. Freemark,

1995). In the particular case of Portuguese grasslands,

it was apparent that considering only the field scale

would be ineffective in securing the conservation

values, given the apparent responses of many species

to habitat fragmentation at the landscape scale.

However, considering only the landscape level would

miss the key role of field management in improving

habitat conditions for several species with unfavour-

able conservation status. Future agri-environment

programs should thus be combined with effective

land-planning policies, whereby subsidies to indivi-

dual farmers for maintaining favourable farming

practices should be conditional on the maintenance

of adequate landscape contexts at larger spatial scales.

Acknowledgements

This study was carried out within the scope of

project PRAXIS XXI/C/AGR/11063/98 ‘‘Determi-

nants of biodiversity in fallows of pseudosteppes:

implications for the definition of agri-environmental

management rules.’’ A. Delgado and F. Moreira were

financed by grants BD/4811/2001, BPD/22102/99 and

BPD/12213/03 from Fundacao para a Ciencia e

Tecnologia, and Pedro Beja was financed by a grant

from Agencia de Inovacao. Thanks are due to Chris

Stoate, Paul Donald and two anonymous referees for

commenting the paper.

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7472

Appendix A

Percentage occurrence (%occ) and mean (�S.D.) abundances (birds/10 ha) of wintering birds recorded in 42

fallow fields sampled in Southern Portugal. Species are classified according to migratory status (M; R, resident; W,

winter visitor) and feeding habits (F; I, invertebrates; S, seeds). Codes are given for species included in multivariate

analyses.

Common name (scientific name) Code MF %occ Mean � S.D.

Meadow pipit (Anthus pratensis) APRA WI 100.0 10.4 � 7.8

Corn bunting (Miliaria calandra) MCAL RS 100.0 8.3 � 16.0

Skylark (Alauda arvensis) AARV WS 95.2 29.2 � 27.8

Linnet (Carduelis cannabina) CCAN WS 92.9 6.3 � 8.05

Lapwing (Vanellus vanellus) VVAN WI 88.1 4.8 � 6.2

Goldfinch (Carduelis carduelis) CCAR WS 78.6 6.2 � 14.0

Crested/thekla lark (Galerida cristata/theklae) GALE RS 76.2 1.8 � 2.0

White wagtail (Motacilla alba) MALB WI 73.8 1.9 � 4.3

Golden plover (Pluvialis apricaria) PAPR WI 59.5 4.4 � 10.2

Stonechat (Saxicola torquata) STOR RI 57.1 0.5 � 0.7

Southern grey shrike (Lanius meridionalis) LMER RI 50.0 0.2 � 0.2

Common/spotless starling (Sturnus vulgaris/unicolor) SUNI WI 42.9 2.5 � 7.6

Red-legged partridge (Alectoris rufa) ARUF RS 40.5 1.1 � 2.5

Calandra lark (Melanocorypha calandra) MELA RS 33.3 2.0 � 5.7

Greenfinch (Carduelis chloris) CCHL WS 33.3 0.5 � 1.7

Chiffchaff (Phylloscopus collybita) PCOL WI 26.2 0.1 � 0.3

Hoopoe (Upupa epops) UEPO RI 26.2 0.1 � 0.2

Little bustard (Tetrax tetrax) TTET RS 21.4 2.0 � 8.7

Serin (Serinus serinus) SERI WS 21.4 0.3 � 1.4

Song thrush (Turdus philomelos) TPHI WI 21.4 0.2 � 0.7

Dartford warbler (Sylvia undata) SUND RI 19.0 0.1 � 0.4

Black redstart (Phoenicuros ochruros) POCH RI 19.0 0.1 � 0.2

Sardinian warbler (Sylvia melanocephala) SMEL RI 19.0 0.1 � 0.2

Woodlark (Lululla arborea) LARB RS 16.7 0.1 � 0.3

Chaffinch (Fringilla coelebs) FCOE WS 16.7 0.1 � 0.2

Robin (Erithacus rubecula) ERUB WI 14.3 0.1 � 0.2

Blue tit (Parus caeruleus) PCAE RI 11.9 0.2 � 0.7

Raven (Corvus corax) CCOR RI 11.9 0.1 � 0.1

Great bustard (Otis tarda) RS 9.5 0.4 � 1.9

House/Spanish sparrow (Passer domesticus/hispaniolensis) RS 9.5 0.4 � 0.2

Little owl (Athene noctua) RI 9.5 0.02 � 0.08

Great tit (Parus major) RI 7.1 0.1 � 0.5

Black-bellied sandgrouse (Pterocles orientalis) RS 4.8 0.07 � 0.34

Fan-tailed warbler (Cisticola juncidis) RI 4.8 0.04 � 0.19

Stone curlew (Burhinus oedicnemus) RI 4.8 0.03 � 0.16

Snipe (Gallinago gallinago) WI 4.8 0.01 � 0.05

Carrion crow (Corvus corone) RI 4.8 0.01 � 0.06

Azure-winged magpie (Cyanopica cyanus) RI 2.4 0.59 � 3.85

Blackbird (Turdus merula) RI 2.4 0.01 � 0.04

Magpie (Pica pica) RI 2.4 0.01 � 0.08

Rock sparrow (Petronia petronia) RS 2.4 0.01 � 0.06

Grey wagtail (Motacilla cinerea) WI 2.4 0.01 � 0.05

Rock bunting (Emberiza cia) RS 2.4 0.01 � 0.04

White stork (Ciconia ciconia) RI 2.4 0.01 � 0.07

Short-toed treecreper (Certhia brachydactyla) RI 2.4 0.01 � 0.10

Siskin (Carduelis spinus) WS 2.4 0.01 � 0.01

Mistle thrush (Turdus viscivorus) RI 2.4 0.003 � 0.023

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–74 73

References

Anderson, M.J., Gribble, N.A., 1998. Partitioning the variation

among spatial, temporal and environmental components in a

multivariate data set. Aust. J. Ecol. 23, 158–167.

Ausden, M., Treweek, J., 1995. Grasslands. In: Sutherland, W.J.,

Hill, D.A. (Eds.), Managing Habitats for Conservation. Cam-

bridge University Press, Cambridge, pp. 197–229.

Best, L.B., Bergin, T.M., Freemark, K.E., 2001. Influence of land-

scape composition on bird use of row crop fields. J. Wildl.

Manage. 65, 442–449.

Bignal, E.M., McCracken, D., 1996. Low-intensity farming in

the conservation of the countryside. J. Appl. Ecol. 33, 413–

424.

Borralho, R., Carvalho, C.R., Stoate, C., Araujo, M., Reino, L.M.,

1999. Avaliacao intermedia do impacte do plano zonal de Castro

Verde na Avifauna. In: Beja, P., Catry, F., Moreira, F. (Eds.), II

Congresso de Ornitologia. Sociedade Portuguesa para o Estudo

das Aves, Lisboa, pp. 52–54.

Buhl, M.L., 1996. Type II statistical errors in environmental

science and the precautionary principle. Mar. Pollut. Bull.

32, 528–531.

Coppedge, B.R., Engle, D.M., Masters, R.E., Gregory, M.S., 2001.

Avian responses to landscape change in fragmented southern

Great Plains grasslands. Ecol. Appl. 11, 47–59.

Delgado, A., Moreira, F., 2000. Bird assemblages of an Iberian

cereal steppe. Agric. Ecosyst. Environ. 78, 65–76.

Dıaz, M., Carbonell, R., Santos, T., Telleria, J.L., 1998. Breeding

bird communities in pine plantations of the Spanish plateaux:

biogeography, landscape and vegetation effects. J. Appl. Ecol.

35, 562–574.

Dıaz, M., Telleria, J.L., 1994. Predicting effects of agricultural

changes in central Spanish croplands on seed-eating overwinter-

ing birds. Agric. Ecosyst. Environ. 49, 289–298.

Donald, P.F., Buckingham, D.L., Moorcroft, D., Muirhead, L.B.,

Evans, A.D., Kirby, W.B., 2001. Habitat use and diet of skylarls

Alauda arvensis wintering on lowland farmland in southern

Britain. J. Appl. Ecol. 38, 536–547.

Estades, C.F., Temple, S.A., 1999. Deciduous-forest bird commu-

nities in a fragmented landscape dominated by exotic pine

plantations. Ecol. Appl. 9, 573–585.

Evans, A., 1997. The importance of mixed farming for seed-eating

birds in the UK. Farming and birds in Europe. In: Pain, D.J.,

Pienkowski, M.W. (Eds.), The Common Agricultural Policy and

its Implications for Bird Conservation. Academic Press, London,

pp. 331–357.

Freemark, K., 1995. Assessing effects of agriculture on terrestrial

wildlife: developing a hierarchical approach for the US EPA.

Landscape Urban Plan 31, 99–115.

Freemark, K.E., Kirk, D.A., 2001. Birds on organic and conven-

tional farms in Ontario: partitioning effects of habitat and

practices on species composition and abundance. Biol. Cons.

101, 337–350.

Fuhlendorf, S.D., Woodward, A.J.W., Leslie Jr., D.M., Shackford,

J.S., 2002. Multi-scale effects of habitat loss and fragmentation

on lesser prairie-chicken populations of the US Southern Great

Plains. Landscape Ecol. 17, 617–628.

Fuller, R.J., Youngman, R.E., 1979. The utilization of farmland by

Golden Plovers wintering in southern England. Bird Study 26,

37–46.

Gaston, K.J., Blackburn, T.M., 2000. Pattern and Process in Macro-

ecology. Blackwell Science, Oxford.

Gutzwiller, K.J., Barrow, W.C., 2001. Bird-landscape relations in the

Chihuahuan desert: coping with uncertainties about predictive

models. Ecol. Appl. 11, 1517–1532.

Herkert, J.R., 1994. The effects of habitat fragmentation on Mid-

western grassland bird communities. Ecol. Appl. 4, 1448–1458.

Legendre, P., Legendre, L., 1998. Numerical Ecology, second ed.

Elsevier, Amsterdam.

Milson, T.P., Ennis, D.C., Haskell, D.J., Langton, S.D., McKay,

H.V., 1998. Design of grassland feeding areas for waders during

winter: the relative importance of sward, landscape factors and

human disturbance. Biol. Cons. 84, 119–129.

Moorcroft, D., Whittingham, M.J., Bradbury, R.B., Wilson, J.D.,

2002. The selection of stubble fields by wintering granivorous

birds reflect vegetation cover and food abundance. J. Appl. Ecol.

39, 535–547.

Moreira, F., 1999. Relationships between vegetation structure and

breeding bird densities in fallow cereal steppes of Castro Verde,

Portugal. Bird Study 46, 309–318.

Pain, D.J., Pienkowski, M.W. (Eds.), 1997. Farming and Birds in

Europe. The Common Agricultural Policy and its Implications

for Bird Conservation. Academic Press, London.

Peach, W.J., Siriwardena, G.M., Gregory, R.D., 1999. Long-term

changes in over-winter survival rates explain the decline of reed

buntings Emberiza schoeniclus in Britain. J. Appl. Ecol. 36,

798–811.

Perkins, A.J., Whittingham, M.J., Bradbury, R.B., Wilson, J.D.,

Morris, A.J., Barnett, P.R., 2000. Habitat characteristics affect-

ing use of lowland agricultural grassland by birds in winter. Biol.

Cons. 95, 279–294.

Potts, G.R., 1986. The Partridge: Pesticides, Predation and Con-

servation. Collins, London.

Robinson, R.A., Sutherland, W.J., 2000. The winter distribution of

seed-eating birds: habitat structure, seed density and seasonal

depletion. Ecography 22, 447–454.

Santos, C.P., 2000. Succession of breeding bird communities after

the abandonment of agricultural fields in south-east Portugal.

Ardeola 47, 171–181.

Santos, T., Tellerıa, J.L., 1985. Patrones generales de la distribucion

invernal de passeriformes en la Penınsula Iberica. Ardeola 32,

17–30.

Soderstrom, B., Part, T., 2000. Influence of landscape scale on

farmland birds breeding in semi-natural pastures. Cons. Biol.

14, 522–533.

Stoate, C., Boatman, N.D., Borralho, R.J., Rio Carvalho, C., de

Snoo, G.R., Eden, P., 2001. Ecological impacts of arable inten-

sification in Europe. J. Environ. Manag. 63, 337–365.

Suarez, F., Garza, V., Morales, M.B., 2003. The role of extensive

cereal crops, dry pastures and shrub-steppe in determining

skylark Alauda arvensis densities in the Iberian Peninsula.

Agric. Ecosyst. Environ. 95, 551–557.

Suarez, F., Naveso, M.A., De Juana, E., 1997. Farming in the

drylands of Spain: birds of the pseudosteppes. In: Pain, D.J.,

F. Moreira et al. / Agriculture, Ecosystems and Environment 109 (2005) 59–7474

Pienkowski, M.W. (Eds.), Farming and Birds in Europe. The

Common Agricultural Policy and its Implications for Bird

Conservation. Academic Press, London, pp. 297–330.

Suarez-Seoane, S., Osborne, P., Baudry, J., 2002. Responses of birds

of different biogeographic origins and habitat requirements to

agricultural land abandonment in northern Spain. Biol. Cons.

105, 333–344.

ter Braak, C.J.F., 1995. Ordination. In: Jongman, R.H.J., ter Braak,

C.J.F.,van Tongeren, O.F.R. (Eds.),Data Analysis in Community

and Landscape Ecology. Cambridge University Press, Cam-

bridge, pp. 91–173.

Tellerıa, J.L., 1988. Invernada de Aves en la Peninsula Iberica.

Sociedade Espanola da Ornitologia, Madrid.

Tucker, G.M., 1992. Effects of agricultural practices on field use

by invertebrate-feeding birds in winter. J. Appl. Ecol. 29, 779–

790.

Underwood, A.J., 1997. Environmental decision-making and the

precautionary principle: what does this principle mean in envir-

onmental sampling practice? Landscape Urban Plan 37, 137–

146.

Vickery, J.A., Tallowin, J.R., Feber, R.E., Asteraki, E.J., Atkinson,

P.W., Fuller, R.J., Brown, V.K., 2001. The management of

lowland neutral grasslands in Britain: effects of agricultural

practices on birds and their food resources. J. Appl. Ecol. 38,

647–664.

Wakeham-Dawson, A., Aebischer, N.J., 1998. Factors determining

winter densities of birds on environmentally sensitive area arable

reversion grassland in southern England, with special reference

to skylarks (Alauda arvensis). Agric. Ecosyst. Environ. 70, 189–

201.

Wernham, C., Toms, M., Marchant, J., Clark, J., Siriwardena, G.,

Baillie, S., 2002. The Migration Atlas: Movements of the birds

of Britain and Ireland. Poyser, London.

Wilson, J.D., Taylor, R., Muirhead, L.B., 1996. Field use by farm-

land birds in winter: an analysis of field type preferences using

resampling methods. Bird Study 43, 320–332.