Effects of undergrowth clearing on the bird communities of the

Northwestern Mediterranean Coppice Holm oak forests

Jordi Camprodon a,b,*, Lluıs Brotons a,b

a Centre Tecnologic Forestal de Catalunya, Pujada del Seminari s/n 25280 Solsona, Catalonia, Spainb Centre d’Ecologie Fonctionelle et Evolutive-CNRS, 1919 Route de Mende, 34293 Montpellier Cedex, France

Received 23 November 2004; received in revised form 6 September 2005; accepted 3 October 2005

Abstract

Undergrowth clearing is a widespread forest management technique used in manyMediterranean regions to reduce dense vegetation in order to

prevent fire or to facilitate other forest exploitation activities. Here, we analyze the effects of undergrowth clearing on biodiversity by focusing on

the variations in bird diversity in Holm oak forests in Catalonia (north-east Iberian Peninsula) under different forest management regimes: (1)

coppice Holm oak forests where the undergrowth layer has been completely cleared, (2) partially cleared forests and (3) cleared and tree thinned

forests and finally, (4) undisturbed forests. The synchronic comparison approach was used in conjunction with a before–after control impact

(BACI) experiment in which the effects of undergrowth clearing were explicitly measured. Complete undergrowth clearing resulted in the almost

complete disappearance of threeWarbler species (SubalpineWarbler, SardinianWarbler and GardenWarbler). Partial clearing also led to a marked

reduction in the numbers of these three species, but the presence of Sardinian and GardenWarblers was maintained in the treated forests. Complete

undergrowth clearing accompanied by tree thinning also led to a decrease in amount of undergrowth species and involved additional negative

effects for species such as Wren, Robin, Blackbird and Blackcap. Only one species, the Nightjar, appeared to benefit from undergrowth clearing,

while many others increased their numbers only when clearing was applied together with tree thinning: Turtle dove, Mistle thrush and Cirl bunting.

Undergrowth clearing brought about a significant simplification in the vertical structure of the forest, which probably reduced foraging

opportunities and breeding resources for most undergrowth species. These effects became more pronounced when tree thinning was applied

together with undergrowth clearing. To reconcile forest management and bird diversity, undergrowth clearing should be applied selectively to

ensure that a certain number of undergrowth patches, preferably corresponding to different shrub species, are not affected by management

treatments.

# 2005 Elsevier B.V. All rights reserved.

Keywords: Clearing; Undergrowth vegetation; Quercus ilex; Birds; Thinning; Forestry

www.elsevier.com/locate/foreco

Forest Ecology and Management 221 (2006) 72–82

1. Introduction

Mediterranean forests have traditionally been subject to

strong human pressure through activities such as forestry, coal

production or vegetation clearing (Blondel and Aronson, 1999).

However, recent socioeconomic changes in some Mediterra-

nean regions have induced a reduction of most of these

activities typically leading to an accumulation of vegetation,

especially in the shrub and lianas layers of the undergrowth. In

this context, undergrowth clearings are currently the main

management activity effecting forests in many regions

* Corresponding author. Tel.: +34 973 481752; fax: +34 973 481392.

E-mail addresses: jordi.camprodon@ctfc.es (J. Camprodon),

lluis.brotons@ctfc.es (L. Brotons).

0378-1127/$ – see front matter # 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.foreco.2005.10.044

particularly in the north-westernMediterranean basin. Mechan-

ical undergrowth clearing aims to reduce the densely developed

shrub and liana vegetation layers and are mainly used to

eliminate the combustible load but also to control tree

competition and favour sapling regeneration or access to the

forest (Garolera, 1991; CPF, 1992; Meson and Montoya, 1993).

The increase in the number and impact of forest fires during

recent decades (Terradas, 1996) has boosted the use of

undergrowth clearing, although, the effectiveness of this

forestry technique to fight the spread of fire has been

questioned because of its low cost effectiveness (Camprodon,

2001).

Although the influence of forestry, fire or grass management

on biological diversity in the Mediterranean scrubs, garrigues

and maquis has been studied (Blondel, 1981; Prodon, 1988;

Pons, 1998), the effects of undergrowth clearing have been

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–82 73

largely neglected and are not currently used in the criterion

which guides forest management in the region. Clearing is

likely to have a strong impact on biodiversity, since it entails a

significant elimination of forest biomass. In Mediterranean

forests, more than 40% of the total floristic richness, as

quantified by forest plants, is located in the shrub layer.

Furthermore, most of the arthropod biomass inhabiting Holm

oak forest, essential for the undergrowth feeding of diverse

faunal groups, also occurs in the shrub undergrowth (pers. obs.).

Vegetable biomass is often a good indicator of the density and

specific bird composition of forest habitats (MacArthur, 1964;

Wilson, 1974; Levin, 1976; Prodon and Lebreton, 1981; James

and Wamer, 1982). In particular, a recurrent positive relation

between bird richness, density and extension of the under-

growth layer has been found (for example, Prodon and

Lebreton, 1981; Canterbury and Blockstein, 1997; Donald

et al., 1997; Dıaz et al., 1998). Crawford et al. (1981) concluded

that the crown cover and the height of the undergrowth layer

were the most important factors in explaining the distribution of

forest passerines. These studies linking undergrowth vegetation

and bird communities suggest that an intense modification of

the shrub layer either by mechanical means, grass or prescribed

burning, may have a significant impact on bird fauna. In fact,

the situation after a total clearing of the undergrowth can be

compared to one after a forest fire in which trophic availability

and nest site availability are limited, forcing territorial birds to

breed in unusual locations, consequently increasing their

exposure to predation or reducing their food supply (Renwald,

1977; Wright, 1981; Winter and Best, 1985; Petersen and Best,

1987). In spite of these predicted effects, coppice regeneration

may favour rapid nest establishment even before the develop-

ment of a more dense vegetation layer by growth of new seeds

(Brooker and Rowley, 1995; Pons, 1996). However, unlike

burning, clearing of the undergrowth does not affect tree

canopies and therefore, undergrowth recovery tends to be

slower unless clearing is accompanied by tree thinning.

The objective of this study is to determine the association

between forest birds and the structure of the shrub layer in

Mediterranean forests and test the hypothesis that undergrowth

clearing has a strong impact on the occurrence of forest birds,

especially those known to rely on low vegetation layers.

Furthermore, we explored the combined impact of undergrowth

clearing with other forestry practices (i.e. tree thinning) on the

bird community. We conclude by suggesting some recom-

mended guidelines aimed at making forest management

compatible with the conservation of species associated with

Mediterranean shrubby vegetation.

2. Methods

2.1. Study area

The study area, of about 2000 km2, is located in the Catalan

Pre-littoral System (north-east Iberian Peninsula). It has a

relatively rainy coastal climate with a short cool winter (de

Bolos, 2001). Coppice Holm oak forests (Viburno-Quercetum

ilicis) are dominant in most mountain areas up to 1400–1500 m.

The arboreal layer of coppice Holm oak forests is rather short

and the trees never reach a great size: 8.5 m average maximum

height, 12.4 cm average diameter at breast height (dbh), with

few trees reaching 35 cm dbh. Due to their limited size, trees

encounter strong competition for light and soil resources from

the well developed ligneous undergrowth which is dominated

by shrubs such as the Strawberry tree (Arbutus unedo),

Laurustinus (Viburnum tinus), Heather (Erica multiflora),

Lentiscus (Pistacea lentiscus) and Rockrose (Cistus salviifo-

lius) and lianas such as Bindweed (Smilax aspera), Ivy (Hedera

helix) and Bramble (Rubus ulmifolius). Sprouts of Holm oak

may also appear in the coppice forest undergrowth. The

arboreal layer, with close crown contact, is dominated by the

Holm Oak (Quercus ilex), accompanied in some areas by

Evergreen oak (Quercus humilis), Cork oak (Quercus suber) or

more occasionally Aleppo pine (Pinus halepensis).

The main forest management technique applied to Holm oak

forests in the area is sprout selection. This technique reduces

the number of sprouts per tree leaving from between two to four

stumps. These are eventually harvested for firewood once they

reach between 18 and 25 cm dbh (Garolera, 1991; CPF, 1992).

Cut rotation time ranges from 10 to 15 years. Usually, before a

sprout selection, the undergrowth layer is cleared with a chain

saw or a bush breaker. This undergrowth clearing is the main

disturbance affecting the shrub layers in these Holm oak

forests.

A particular case of forest management in areas used for cow

pasture and ewe grass consists of undergrowth clearing applied

in conjunction with the thinning of the Holm oak forest by

means of strong selective cutting. These forests have a low

density, open forest structure, since thinning has eliminated

most stems and only the best stumps have been preserved.

Overall, when taking into account the intensity and time

elapsed since undergrowth management, Holm oak forests vary

in the degree of cover and height of the shrub and liana layers.

When considering undergrowth structure, coppice Holm oak

forests where sprout selection is the main management

technique can be broadly classified a priori into four different

groups (Fig. 1): (1) Coppice Holm oak forests where the

undergrowth layer has been completely cleared. The entire

shrub and lianas layers having been eliminated by mechanical

practices. (2) Partially cleared coppice Holm oak forests which

maintain approximately one-third of the shrubs and lianas. This

clearing is not uniform and often clumps of shrubs and lianas

may have been left behind after undergrowth clearing. (3)

Cleared and thinned Coppice Holm oak forests in which all of

the undergrowth is eliminated and most of the arboreal layer is

felled by selective thinning. (4) Dense coppice Holm oak forest

where neither clearing nor thinning have taken place. These

have a high shrub and lianas cover (70–90%) and dense young

growth stand.

2.2. Location of census plots

Using data from the national forest inventory, forest maps

and administrative data on forest management histories in the

study area (Camprodon, 2001), we selected a number of forest

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–8274

Fig. 1. Example of study habitats in the Finestres Mountain within or adjacent to the natural park of Zona Volcanica de la Garrotxa (Catalan Pre-littoral System, NE

Iberian Peninsula). Upper left, a dense Holm oak forest (DH); upper right, a partially cleared Holm oak forest (PCH); bottom left, a completely cleared Holm oak

forest (CCH); and bottom right, a cleared and thinned Holm oak forest strongly felled by selective thinning (CTH).

tracts with a history of forest management and homogeneous

vegetation structure. We classified such forest tracts into one of

the four forest types described above and made sure they were

uniformly distributed within the study area in order to avoid

spatial biases. Within the central section of each selected forest

tract, we located census plots in which the vegetation structure

and the bird community was sampled. Selected forest tracts

were located in forests larger than 200 ha in size, contained at

least 10 ha of homogenous vegetation of the corresponding

forest type and were located at least 1 km from the nearest

forest tract of any other type containing a census plot. Due to

the difficulty in locating adequate, large cleared, thinned

forests, we located, for this forest type, up to three census plots

per forest tract. In this case, census plots were located a

minimum of 400 m from the nearest plot to avoid pseudo-

replication (Hurlbert, 1984). All census plots were located at

least 500 m from the nearest forest edge to avoid interference

from surrounding non-forest habitats.

2.3. Bird census

We used point counts to estimate bird abundance at each

census plot (Tellerıa, 1986; Bibby et al., 2000). Point counts

lasted for 20 min during which all audible and visual contact

was registered in three radiuses of 100 m around the central

point of the census plot. All point counts were conducted within

3 h from sunrise (between 6:30 and 9:30 a.m.). Censuses were

conducted between the third week of April and the last week of

June during the years 1999–2002. We conducted one point

count per season at each census plot. We estimated that given

the long duration of the sampling (20 min), one point count per

season was enough to detect most species present and therefore

accurately sample different forest types. Point counts were

conducted in a random order across different forest types and

years of study reducing any possible temporal bias in data

collection. Larger species of raptors which were occasionally

observed in or around the census plots (Circaetus gallicus,

Buteo buteo, Accipiter nisus, Strix aluco and Asio otus) were

not included in the analyses.

We used species richness and abundance (number of

individual contacts) per point count as variables in further

analyses. We also classified bird species into different

categories (Table 3) according to the patterns of micro-habitat

choice as described in the literature (Muntaner et al., 1983;

Snow and Perrins, 1998; Shirihai et al., 2001): trunk-foraging

species (Woodpeckers, Nuthatch and Short-Toed Treecreeper),

other cavity nesters (tits, genus Parus), birds using tree

canopies (Wood Pigeon, Golden Oriole, Jay, Phylloscopus

Warblers, Firecrest, Chaffinch and some thrushes), species of

the shrub layer of the undergrowth (Wren, Robin, Blackbird

and most Sylvia Warblers) and birds associated with the agro-

forest open habitats (Nightjar, Turtle Dove, Serin, Greenfinch,

Goldfinch, Cirl Bunting). Due to their strong dependence on the

undergrowth layer, we also used another sub-guild within the

shrub layer category composed only of the species of the genus

Sylvia (Table 3).

2.4. Habitat inventory

We characterized vegetation structure for each forest type by

a set of measurements conducted at each census plot. We

described structural and floristic variables within a radius of

50 m around the central point of the corresponding census plot.

The cover of different vegetation layers was visually measured

at different interval heights (CV25: 0–0.25 m, CV50: 0.25–

0.5 m, CV1: 0.5–1 m, CV2: 1–2 m, CV4: 2–4 m, CV8: 4–8 m,

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–82 75

CV16: 8–16 m), a model proposed by Prodon and Lebreton

(1981). The precision of this visual method is acceptable, with

the coefficient of variation estimated to be around 7% (Prodon,

1988). The cover of cleared, cut or pruned dry branches on the

ground (Bran) and rock cover (Rock) were also recorded using

the same method. The diameter of the trunk at breast height

(dbh) was also measured for trees using diametrical classes (D)

at 5 cm intervals. We grouped dbh classes into four categories:

lesser trees (<D5, dbh < 2.5 cm), small trees (D5–15, dbh

between 2.5 and 17.5 cm), medium sized trees (D20–30, dbh

between 17.5 and 32.5 cm) and big trees (D > 30, dbh from

32.5 cm). Final variables used an expressed number of trees/ha

for each of the corresponding tree size categories.

2.5. Experimental removal of the understory

To further explore the effect of undergrowth clearing on the

bird community, we employed a before–after control impact

approach on a section of Holm oak forest before and after

undergrowth clearing management (BACI, Stewart-Oaten

et al., 1986). In December–February of 1999 a complete

mechanical clearing of all shrub and lianas layers in 12 ha of

Holm oak forest with dense and homogenous undergrowth was

carried out. This forest was located north of the study area, in

the Finestres Mountain at a height of 600 m with an east-

southeast aspect.

We established two controlled forest sections, one adjacent

to the experimental one, and the second, of similar size,

altitude, orientation and vegetable structure, located 2 km from

the cleared section. Bird occurrence during the breeding season

within the sections was monitored for a period of 5 years: 2

years prior to the undergrowth clearing (1997 and 1998) and 3

years following the intervention (1999–2001). The census was

conducted by walking inside the forest section and locating all

bird contacts on a map. We conducted 10 visits to each forest

section and territories were identified on the basis of the spatial

clustering of simultaneous song and call contacts (i.e. mapping

method, IBCC, 1969; Tellerıa, 1986; Bibby et al., 2000). When

territories were not entirely within the studied forest section, we

only considered the area which lay within the boundaries of the

studied section.

In each forest section, we also monitored changes in

vegetation during the 5 years by means of 25 permanent plots

located 50 m from each other. Within each of these plots

vegetation variables were measured and described using the

same methodology employed at bird point counts. Climatic

conditions within forest sections were similar for the 5 years of

the project, with warm temperatures (20 � 1.2 8C as the

average for the month of July) and moderate rainfall

(790 � 207.5 mm annual).

2.6. Data analysis

In order to assess overall differences in species richness and

abundance between the four forest types with regard to the

vegetation structure and management regimes, we used

ANOVA and tested differences between the two groups by

means of the Tukey honest significant difference (HSD) (Sokal

and Rohlf, 1995). In cases in which assumptions of normality

were not met, we used generalized linear modelling (McCul-

lagh and Nelder, 1989) with a Poisson or binomial (for scarcer

species) error distribution.

Principal components analysis (PCA) with a varimax

normalized rotation, which maximizes the correspondence

between the factors and the original variables, was used to

simplify vegetation structure variables. The variables of tree

density were excluded from the factorial analysis because they

were not correlated with each other or with the cover variables.

Presence and abundance of bird species in each plot were

correlated to the habitat structural variables by means of the

canonical correspondence analysis (CCA). This is a statistical

technique derived from the correspondence analysis (CA),

designed to analyze tables of double entrance that contain some

measurement of correspondence between rows (sampling

points) and columns (species) (Greenacre, 1984). Like the

CA, CCA generates gradients in the original data by means of a

maximization of the correlation between the values corre-

sponding to the species and the values corresponding to the

used sampling points. But in the case of the CCA, the values

corresponding to the sampling points are constrained to be

linear combinations of a series of environmental variables of

interest (Prodon, 1992). CCA is especially sensitive to the

possible effects of the colinearity between variables. With the

object of reducing the number of variables in the final analysis,

it proceeds with a forward selection process in which only

significant variables in the identification of environmental

gradients are evaluated. All the census plots (n = 145), 24 bird

species (excluding those with less than a total of three contacts),

and 13 environmental variables were analyzed in the CCA

(Table 1).

To further investigate habitat selection patterns of different

guilds, we used the generalized linear model. This technique

considers the dependency of discrete variables (census of birds)

on a group of independent variables or indicators (i.e. habitat

structure: in this case, the factors obtained by PCA). We used

presence/absence of individual bird species in each census plot

as a dependent variable and used logistic regressions (binomial

error distribution) to evaluate species habitat relationships. In

addition, we used the number of species per plot for each

different guild and used Poisson error distribution to establish

guild–habitat relationships. A forward step-wise procedure was

applied to select significant variables only ( p-level threshold

p = 0.05).

Finally, changes in relative abundance of individual species

after clearing treatments were evaluated using the BACI

approach by applying the method of Stewart-Oaten et al.

(1986). Abundances were log-transformed (ln(x + 1)), to render

the residuals homoscedastic and normally distributed. For each

species, the difference in (log) abundance between each

treatment and the mean value of the two controls for each

sampling period was calculated, and these differences were

then subjected to analysis of variance of repeated measures

(ANOVA). Effects were fitted for treatment (cleared-control),

before/after, species and the interaction between these. Single

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–8276

Table 1

Average values and standard deviations of the more representative environmental variables and results of ANOVA test between different structures from Holm oak

forest

Variable DH PCH CCH CTH F p

CV0.25 (%) 50.6 � 19.62 a 42.7 � 14.91 b 27.2 � 14.16 c 41.9 � 9.22 b 15.23 <0.001

CV0.5 (%) 40.5 � 16.00 a 30.9 � 9.17 b 19.7 � 8.45 c 33.9 � 5.61 b 23.83 <0.001

CV1 (%) 56.5 � 15.94 a 29.5 � 8.09 b 10.1 � 7.02 c 27.1 � 11.07 b 112.09 <0.001

CV2 (%) 61.7 � 14.83 a 26.8 � 9.42 b 11.0 � 5.32 c 13.1 � 7.46 c 101.32 <0.001

CV4 (%) 53.2 � 13.85 a 33.1 � 14.58 b 28.9 � 12.60 b 12.3 � 3.84 c 65.04 <0.001

CV8 (%) 44.6 � 14.34 a 48.8 � 12.85 a 47.4 � 12.68 a 27.9 � 4.58 b 20.46 <0.001

CV16 (%) 4.3 � 9.35 9.7 � 11.12 7.4 � 9.60 4.2 � 5.49 2.97 n.s.

Bran (%) 4.2 � 5.25 b 16.3 � 16.46 a 18.5 � 15.20 a 0.0 � 0.18 b 20.36 <0.001

Rock (%) 5.2 � 5.73 b 3.2 � 6.78 b 5.4 � 7.13 b 15.3 � 7.94 a 20.17 <0.001

D < 5 trees/ha 511.7 � 932.14 a 214.7 � 420.69 ab 191.9 � 438.53 ab 0.0 � 0.00 b 4.75 0.004

D5–15 trees/ha 2312.5 � 1103.68 a 1518.1 � 805.05 b 1706.1 � 8880.97 b 139.0 � 103.22 c 39.53 <0.001

D20–30 trees/ha 67.6 � 85.04 b 151’8 � 106.44 a 131.4 � 104.43 a 162.3 � 69.61 a 7.72 <0.001

D > 30 trees/ha 3.2 � 12.06 7.3 � 13.59 7.1 � 22.97 10.6 � 29.30 1.09 n.s.

TD (trees/ha) 2383.3 � 1064.05 a 1749.9 � 760.60 b 1878.9 � 802.04 ab 311.9 � 112.28 c 40.91 <0.001

n 40 39 36 30

Different letters identify forest types groups according to differences in mean values of a given variable as tested by Tukey HSD. CV0.25–CV16: vegetation cover

classes, Bran: dead branch cover, Rock: rock cover, D < 5: density of smaller trees, D5–15: density of small trees, D20–30: density of medium trees, D > 30: density

of large trees. TD: total density. DH: dense Holm oak forests, PCH partially cleared Holm oak forests, CCH: completely cleared Holm oak forests, CTH: cleared and

thinned Holm oak forests.

degree-of-freedom contrasts were then calculated for each

species, to assess whether there was a significant change in the

difference between the band and control forest sections from

before-to-after the application of the clearing treatment.

3. Results

3.1. Structural differences between forest types

The cover of the understory was significantly related to

forest type. In dense, reference forests, shrub cover reached,

below 2 m in height, values above 50%. This cover decreased to

26–42% in partially cleared forests and to 10–27% in

completely cleared forests (Table 1). Total tree density showed

similar figures between different cleared forests, with

abundance of sprouts, ranging between 1.750 and

2.400 trees/ha, largely within the 5–15 cm class and rarely

Table 2

Principal components analysis performed on variables describing vertical cover of th

(Rock)

F1 (high shrub) F2 (low tree) F3 (high t

CV25 0.156 0.133 0.118

CV50 0.469 �0.289 0.003

CV1 0.795* �0.175 0.000

CV2 0.908* �0.002 0.010

CV4 0.863* 0.371 �0.077

CV8 0.098 0.931* 0.149

CV16 �0.042 0.133 0.982*

Bran �0.085 0.017 �0.046

Rock �0.090 �0.161 �0.058

Expl. var. 2.472 1.180 1.013

% from total 27.5 13.1 11.3

Factor loadings for each individual variable are shown for the six factors obtained u

93.1%.* p < 0.05.

surpassing 30 cm. On the other hand, cleared and thinned Holm

oak forests had, on average, as little as 300 trees/ha, mainly

within the classes 15–30 cm (Table 1). They presented an

overall aspect of grazed forest, with trunks not in direct contact

with the shrub and liana layers.

PCA allowed us to distinguish six independent structural

factors associated with: (1) a decrease in shrub cover (up to

0.5 m high), (2) high shrub cover (between 0.5 and 4 m), (3)

low arboreal layer (between 4 and 8 m) and (4) high arboreal

cover (more than 8 m). Factor (5) was associated with a

gradient in rock cover while factor (6) identified a gradient in

dry branch abundance (Table 2).

3.2. Bird–habitat relationships and forest management

We recorded a total of 30 bird species in the Holm oak

forests we analyzed. In dense forests, we detected 20 species,

e vegetation (CV25–CV16), the cover of dry branches (Bran) and the rock cover

ree) F4 (rock) F5 (dead branches) F6 (low shrub)

�0.002 �0.100 0.918*

�0.038 �0.216 0.683*

0.005 �0.167 0.495

�0.101 �0.077 0.252

�0.048 0.049 �0.009

�0.169 0.014 0.001

�0.057 �0.043 0.094

�0.122 0.969* �0.176

0.972* �0.119 �0.016

1.006 1.048 1.659

11.2 11.6 18.4

sing a varimax normalized rotation. The percentage of accumulated variation is

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–82 77

whereas in totally and partially cleared Holm oak forests 18–19

species were sighted and 26 in cleared and thinned Holm oak

forests. Species associated with the shrub layer were abundant

in dense Holm oak forests and tended to become scarcer in

forests in which the shrub layer had been reduced as a result of

the undergrowth clearing management (Table 3). However,

these trends varied according to the species in question.

Whereas species such as the Wren and the turdids (Robin and

Blackbird) appeared in cleared zones in which the arboreal

layer had been unaffected by the management, Sylvia Warblers

were very scarce in any type of cleared forests. For instance, the

Subalpine Warbler was completely absent from totally and

Table 3

Species abundance, total richness and abundance and richness of each guild by po

Scientific name Common name Guild

Species richness

Abundance

Undergrowth species Und

Sylvia Warblers Syl

Trunk-foraging species Tru

Tits Par

Tree-canopy species Cro

Open habitats species Ope

Columba palumbus (Copa) Wood Pigeon Cro

Streptopelia turtur (Sttu) Turtle Dove Ope

Caprimulgus europaeus (Caeu) Nightjar Ope

Upupa epops (Upep) Hoopoe Ope

Picus viridis (Pivi) Green Woodpecker Tru

Dendrocopos major (Dema) Great Spotted Woodpecker Tru

Troglodytes troglodytes (Trtr) Wren Und

Erithacus rubecula (Erru) Robin Und

Turdus merula (Tume) Blackbird Und

Turdus philomelos (Tuph) Song Thrush Cro

Turdus viscivorus (Tuvi) Mistle Thrush Cro

Sylvia cantillans (Syca) Subalpine Warbler Und Sy

Sylvia melanocephala (Syme) Sardinian Warbler Und Sy

Sylvia borin (Sybo) Garden Warbler Und Sy

Sylvia atricapilla (Syat) Blackcap Und Sy

Phylloscopus bonelli (Phbo) Bonelli’s Warbler Cro

Phylloscopus collybita (Phco) Chiffchaff Cro

Regulus ignicapillus (Reig) Firecrest Cro

Muscicapa striata (Must) Spotted Flycatcher Cro

Aegithalos caudatus (Aeca) Long-tailed Tit Cro

Parus cristatus (Pacr) Crested Tit Par

Parus ater (Paat) Coal Tit Par

Parus caeruleus (Paca) Blue Tit Par

Parus major (Pama) Great Tit Par

Certhia brachydactyla (Cebr) Short-toed Treecreeper Tru

Oriolus oriolus (Oror) Golden Oriole Cro

Garrulus glandarius (Gagl) Jay Cro

Fringilla coelebs (Frco) Chaffinch Cro

Serinus serinus (Sese) Serin Ope

Carduelis chloris (Cach) Greenfinch Ope

Carduelis carduelis (Caca) Goldfinch Ope

Emberiza cirlus (Emcr) Cirl Bunting Ope

n

Different letters identify forest types groups according to differences in mean values

dense Holm oak forests, PCH: partially cleared Holm oak forests, CCH: completely

significant.* p < 0.05.** p < 0.01.*** p < 0.001.

partially cleared forests (Table 3). The Blackcap occupied

totally cleared forests, but it was significantly less abundant in

dense forests (Table 3). When the clearing was accompanied by

selective thinning, the Wren, the Robin and the Blackcap were

less frequent and the rest of the Sylvia Warblers disappeared.

Only Blackbird abundance seemed to remain relatively

unchanged despite the undergrowth vegetation structure

(Table 3).

Common ground foragers in the dense forests like TheWood

Pigeon, the Jay and the Chaffinch, did not increase their use of

forest habitats with reduced shrub cover (Table 3). Only the

Chaffinch increased in abundance when the arboreal canopy

int count, according to the degree of undergrowth clearing or arboreal cutting

DH PCH CCH CTH p

8.30 a 6.36 b 5.44 b 8.60 a ***

10.30 a 7.95 ab 6.86 b 9.97 a ***

5.05 a 2.56 b 1.78 c 1.17 c ***

2.80 a 0.82 b 0.22 c 0.03 c ***

0.13 b 0.35 b 0.26 b 1.50 a ***

0.68 b 1.03 b 0.94 b 1.60 a ***

2.28 2.15 2.14 2.07 n.s.

0.17 a 0.21 a 0.28 a 2.07 b ***

0.03 0.21 0.11 0.13 n.s.

0.20 *

0.14 0.10 *

0.10 ***

0.05 b 0.05 b 0.06 b 0.40 a ***

0.03 b 0.37 a ***

0.63 a 0.41 a 0.39 a 0.03 b ***

1.58 a 1.51 a 1.39 a 0.37 b ***

0.80 0.79 0.69 0.93 n.s.

0.05 0.07 n.s.

0.05 b 0.30 a ***

l 1.13 ***

l 0.98 a 0.28 b ***

l 0.65 a 0.13 b ***

l 0.78 a 0.56 ab 0.31 b 0.03 c ***

0.05 0.13 0.17 0.07 n.s.

0.50 a 0.31 a 0.36 a ***

1.05 a 0.79 a 0.81 a 0.17 b ***

0.20 ***

0.18 0.21 0.19 0.17 n.s.

0.08 b 0.23 b 0.19 b 0.43 a **

0.05 0.06 n.s.

0.28 b 0.36 b 0.28 b 0.43 a ***

0.38 0.41 0.44 0.43 n.s.

0.13 b 0.33 a 0.22 b 0.57 a ***

0.05 b 0.06 b 0.20 a **

0.48 0.44 0.42 0.33 n.s.

0.45 b 0.38 b 0.44 b 1.33 a ***

0.10 b 0.10 b 0.06 b 1.07 a ***

0.05 0.05 0.06 0.20 n.s.

0.05 0.08 0.06 n.s.

0.50 ***

40 39 36 30

of a given variable as tested by Tukey HSD or generalized linear modelling. DH:

cleared Holm oak forests, CTH: cleared and thinned Holm oak forests. n.s.: not

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–8278

Fig. 2. The arrangement of samples in the two first axes of the canonical

correspondence analysis. Dense Holm oak forests are represented by filled

squares; partially cleared Holm oak forests by open circles, completely cleared

Holm oak forests by triangles and cleared and thinned Holm oak forests by

crosses.

was also thinned, together with open habitat ground foragers

which are absent or very rare in dense forests, such as the Turtle

Dove, the Hoopoe, the GreenWoodpecker, theMistle Thrush or

the Cirl Bunting. Only the Nightjar appeared in completely

cleared forests, being entirely absent from dense forests.

Finally, trunk-foraging species and birds from agro-forest open

habitats increased in abundance in cleared and thinned Holm

oak forests remaining less frequent in denser forests (Table 3).

Bird community structure in different types of forests was

significantly captured by the environmental variables used. The

first CCA axis, accounting for 11.2% of total variation in bird

community, represented a gradient based on structural variables

related to cover of shrubs and small trees (i.e. from dense to

cleared forests), whereas the second gradient was positively

related to tree density and negatively to rock cover (Figs. 2 and

3). The two first axes accounted for 14.9% of total variation.

The presence of three shrub layer bird species, the Subalpine

Warbler, the Sardinian Warbler and the Garden Warbler, which

were then more abundant in dense Holm oak forests, was

Fig. 3. The relationship between the bird species and the environmental

variables selected by the canonical correspondence analysis. The variables

have been selected from a progressive evaluation of the meaning of the different

variables, by means of the Monte Carlo test. Only the variables with a p-level of

at least 0.05 are included. In order to interpret the environmental variable and

the species acronym’s, see the text and Table 3, respectively.

strongly associated to the first gradient. The rest of the shrub

layer species, such as the Wren, the Robin or the Blackcap, and

tree-canopy birds were associated with totally or partially

cleared forests. On the other hand, birds from agro-forest open

habitats, such as the Turtle Dove, the Hoopoe or the Cirl

Bunting (Fig. 3), were mainly associated with cleared and

thinned Holm oak forests.

Habitat selection models for undergrowth species showed a

strong association of this group with PCA factors describing the

low and high shrub layer and the low arboreal layer (Tables 2

and 4). Within this group of birds, the Wren and the Blackbird

also selected the accumulation of dead branches on the ground

(Table 4). Furthermore, we detected a negative association of

the occurrence of species such as the Wren, the Blackbird and

the Sardinian Warbler with the highest tree cover and also for

the Wren, the Blackbird, the Blackcap and the Garden Warbler

with rock cover. Trunk foraging species and tits were negatively

associated with the density of small trees. Tits were the only

group which had a positive association with rock cover. We did

not find any significant model for tree-canopy species. Finally,

birds of agro-forest open habitats tended to negatively select

plots with high tree densities and dense sub-arboreal layers

(Table 4).

3.3. Effects of experimental clearing on the bird

community

The average cover of shrubs and lianas before the

undergrowth clearing in the experimental forest section

(12 ha in size) was 57% for the vegetation layers between

0.25 and 4 m. These layers were dominated by Heather,

Strawberry tree, Binweed, Lentiscus, Bramble, Liguster and

Old Man’s Beard. After the clearing, the average cover of these

layers decreased to about 6%, before increasing progressively

again to about 16% in 2002 four years after the treatment.

Before the clearing the experimental forest section hosted

between 2 and 7 territories of different undergrowth bird

species (Fig. 4). We found a highly significant interaction

between the control-impact and the before–after factors

(F1,21 = 224.25, p < 0.001) indicating that while within the

control forest sections the number of occupied territories

remained constant, the number of bird territories within the

treatment section decreased sharply overall after the clearing of

the understory (Fig. 4). We also found that the effect of the

clearing was indeed species specific (F6,21 = 34.18, p < 0.001)

and therefore not all species were eliminated from the plots

after the undergrowth clearing (Fig. 4). During the breeding

season following winter clearing, the Subalpine Warbler, the

Sardinian Warbler and the Garden Warbler apparently

abandoned their territories and remained absent during the

years following the disturbance (1999–2001, Fig. 4). From

circumstantial detection of singing males, territories of Sylvia

Warblers seemed to be maintained in areas surrounding the

experimental forest section and the control sections throughout

these years. In contrast to other Sylvia Warblers, the Blackcap

continued to use to some degree the cleared zone, but this

species modified microhabitat use within the territories,

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–82 79

Fig. 4. Representation of temporal changes in mean number of territories within the experimental (black circles) and control forest sections (white squares) for each

species commonly using the forest undergrowth: species specific contrast tests showed that the interaction between control-impact and before–after factors are

significant for all species except Sylvia atricapilla, Troglodytes troglodytes and Erithacus rubecula. The arrowmarks the point in timewhen the undergrowth clearing

treatment was applied to the experimental forest section during winter 1998–1999.

Table 4

Selection of environmental variables on the undergrowth species and the different guilds, from generalized linear models, by means of the procedure of forward

stepwise (whole p to enter: 0.05; whole p to remove: 0.05)

Species Model x2 d.f. % p

Undergrowth species 1.44 � 0.002D20–30 + 0.33 � high shrub + 0.17

� low tree + 0.20 � high tree �0.15 � rock

92.51 5 47.72 ***

Sylvia Warblers �0.16 + 0.79 � high shrub + 0.28 � low tree

� 0.30 � rock + 0.18 � low shrub

126.76 4 50.57 ***

Trunk-foraging species 0.38 � 0.001D5–15 54.04 1 36.06 ***

Tits 0.49 � 0.001D5–15 + 0.21 � rock 35.19 2 19.04 ***

Open habitats species 0.51 � 0.01D5–15 � 0.58 � low tree

� 0.33 � low shrub

156.86 1 63.08 ***

Wren �0.87 + 0.72 � high shrub + 0.49 � low tree

� 0.67 � high tree � 0.59 � rock + 0.63 � branc

42.05 5 22.67 ***

Robin 0.72 + CD10–15 + 1.78 � high shrub + 0.91 �high tree + 0.72 � low shrub

62.28 4 39.43 ***

Blackbird �1.60 + 0.001D < 5 + 0.01D20–30 � 0.47 � rock

+ 0.39 � branc � 0.48 � high tree

26.54 5 13.66 **

Sardinian Warbler 2.53 + 3.06 � high shrub � 1.20 � high tree 97.54 2 62.71 ***

Subalpine Warbler �1.28 + 2.45 � high shrub + 0.87 � low shrub 86.61 2 48.66 ***

Garden Warbler �1.98 + 0.98 � high shrub � 0.74 � rock 23.67 2 17.75 ***

Blackcap �0.92 + 1.04 � high shrub + 0.70 � low tree + 0.65

� high tree � 0.83 � rock + 0.60 � low shrub

55.54 5 29.53 ***

All the species follow a binominal distribution, while the guilds follow a Poisson distribution.** p < 0.01.*** p < 0.001.

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–8280

showing a tendency to concentrate its activity along the areas

neighbouring non-cleared zones. The Blackcap used the

cleared zones basically for feeding and, secondarily, as singing

sites, but located the nest in the non-cleared zones.

However, the distribution of Wren contacts remained

basically identical before and after the clearing and occupied

vegetation along small dry streams. This species took

advantage of the accumulation of dead branches resulting

from forest clearing and used this substrate for feeding and

singing whereas it bred on the ground or rock margins. The

Robin also maintained a constant use of the forest after

clearing. However, more than half of the audible contacts for

this species along the border of the experimental forest section

corresponded to the non-cleared vegetation. Finally, the

Blackbird used the cleared zones as feeding substrates by

foraging on the ground. Most of the species activity also

concentrated on the non-cleared parts. Nearly half of the

Blackbird contacts in the cleared area corresponded to young

birds. The tree-dwelling species showed no change in numbers

after disturbance compared to the control.

4. Discussion

Change in perceived habitat quality is the key factor

determining birds’ response to disturbances (Sousa, 1984;

Wiens, 1989; Bazzaz, 1991). Prodon and Lebreton (1981)

showed that as small changes in vegetal cover occur, significant

impacts on the bird community may be detected when overall

vegetal cover in the undergrowth is low. However, if ligneous

vegetation is dense enough, clearing of the vegetation may have

negligible effects on birds. In line with these predictions, the

abundance of Sylvia Warblers, birds morphologically adapted

to life within dense shrubs, diminished in cleared forests, but

especially so when clearing of the undergrowth vegetation was

almost complete. The results obtained after the experimental

undergrowth clearing of a forest reinforced the trends observed

in the synchronous study based on point counts regarding

changes in the abundance of Sylvia Warblers following the

clearing disturbance. Among this group of birds, the Subalpine

Warbler appeared to be the most demanding in terms of

undergrowth vegetation structure, since it prefers higher and

denser shrubs and is the less arboreal of the Sylvia Warblers in

the Holm oak forests, whereas the Blackcap was the most

flexible species, probably due to its preferential use of tree

canopies compared to other Sylvia Warblers (Fuller, 1995;

Snow and Perrins, 1998; Shirihai et al., 2001). If the

undergrowth clearing is accompanied by selective thinning,

the combined effects of these two factors on the bird

community appear more intense. This management practice

diminishes the abundance of other undergrowth bird species in

addition to the Sylvia Warblers, but also the thinning induces

changes in the composition of the tree-canopy birds and

benefits other species associated with agro-forest open habitats

(Fig. 3).

The disappearance of the undergrowth vegetation induced

by the clearing results in important limitations for birds. First,

species will encounter a decrease in the availability of trophic

resources within the undergrowth vegetation (i.e. arthropods

and fruits), a basic substrate used for feeding (Snow and

Perrins, 1998). According to our own unpublished data,

arthropod biomass contained in undergrowth ligneous vegeta-

tion can be very high, even higher than that recorded in the

canopy layer of a mono-specific arboreal layer such as that

dominating in Holm oak forests. Even the piles of accumulated

fine dead branches on the ground after the clearing are a

substrate apt for feeding. Second, a reduction in vegetation load

may induce the loss of singing and breeding sites or other

locations that may provide refuge from predators. Sites within

the arboreal layers can act as an alternative for flexible species

but not for selective species which focus on the use of low

shrubs. Finally, changes in vegetation structure lead to changes

in the microclimate of otherwise humid, dense Holm oak

forests. These effects may be important for hygrophylic species

especially when clearing is accompanied by tree thinning

(Wren, Robin and Blackcap). In fact, microclimate has often

been considered as an influential abiotic factor on bird

distribution (Janzen, 1986; Lovejoy et al., 1986; Wiens,

1989), either through changes in the interactions with other

species or by placing restrictions on the species physiological

needs. When thinning is combined with undergrowth clearing,

the loss of trophic resources and refuge substrates for more

flexible undergrowth birds is enhanced and accompanied by a

complete loss of forest interior microclimate. On the other

hand, open conditions lead to the appearance of agro-forest

habitat species, which is favoured by the fact that open

meadows are often found relatively near (500–1000 m) thinned

and cleared forests. This is to facilitate the use by cows of

alternative pasture habitats (pers. obs.).

The structure of the remaining vegetation after clearing is

another basic factor determining post-disturbance habitat use

by birds. The height of the shrub layer between 1 and 2 m

appears to have a critical role for undergrowth species such as

the Sylvia Warblers (see also Blondel, 1981 for ‘‘garrigue’’ bird

assemblages). The low height of the arboreal layer prevailing in

Holm oak forests may help to mitigate the negative effects of

sparse undergrowth for flexible species (i.e. species capable of

using tree canopies to some extent) such as the Robin, the

Blackbird and the Blackcap.

In thinned Holm oak forests, the reduction in tree density

and associated lianas, together with the preservation of the

more developed tree trunks seemed to promote the presence of

trunk-foraging species. The rock cover was negatively

associated with different undergrowth passerines and positively

with the tits. The effect of rock cover on undergrowth birds can

be interpreted as an indirect indicator of shrub cover deficiency

in rocky areas (i.e. undergrowth clearing and intense thinning

leave the open and the rocky outcrops visible) or as indication

of dry conditions in the area. A more direct explanation may be

found for tits since this group makes direct use of rocky

outcrops as substrate for nest building which may be otherwise

scarce in managed Holm oak forests (pers. obs.).

A site tenacity effect has been described several times in

undergrowth birds (first described in Emlen (1970), and ratified

by Prodon and Lebreton (1981), Prodon et al. (1987), Pons

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–82 81

(1996, 1998)). According to this effect, birds may persist in the

original territories after disturbances such as fires. The

existence of tenacious birds may cushion the effects of the

disturbance, since only one part of the population disappears

either by death or dispersion. However, in the experimental

forest clearing, the three Sylvia Warblers species most affected

by the disturbance did not show tenacious behaviour after

clearing of the undergrowth, probably because they found

suitable habitats nearby.

To summarize the main effects of clearing and thinning of

Holm oak forests on birds, it is possible to establish a gradient

of response strength according to the use by species of

undergrowth vegetation: (a) species that require high shrub

cover and which are already scarce after partial clearing: the

Subalpine Warbler, the Sardinian Warbler and the Garden

Warbler; (b) species that tolerate clearing to some degree, but

use, preferably, nearby non-cleared areas as breeding sites: the

Blackbird and the Blackcap; (c) a group of species that remain

indifferent to changes in the undergrowth, but are sensitive to

decreases in undergrowth cover when applied together with tree

thinning: the Wren, the Robin and the Firecrest; (d) species

which tended to favour the undergrowth clearing: Nightjar; and

finally, (e) tree-canopy dwelling species that enter the cleared

and thinned Holm oak forests: the Turtle Dove, the Mistle

Thrush, the Spotted Flycatcher, the Golden Oriole, the Serin,

the Greenfinch and the Cirl Bunting.

5. Applied conclusions

The reduction of the vegetal biomass of the undergrowth is a

convenient forestry practice in particular forest management

circumstances. However, maintaining large extensions of

forests with a low combustible load is a titanic and

economically unrealizable task with obvious ecological

impacts (Meson and Montoya, 1993). Consequently, to

improve biodiversity conservation it is important to ensure a

certain number of patches of the different shrubs and lianas

species and to apply clearing to patchy surfaces (10 ha

maximum). The selection of shrubs and lianas for clearing

should be applied plant by plant, so that the ligneous

undergrowth is uniformly distributed, or distributed into

clusters. However undisturbed small sectors no longer than

1 ha, should be reserved.

To maintain undergrowth Warbler species after clearing

would require a minimum of 30–40% of shrub cover above one

meter in height (preferably 1.5–2 m). Sylvia Warblers are one of

the characteristic birds of Mediterranean woodlands, and

therefore, they have a substantial value to conservation, in spite

of not being a threatened species (Tucker and Heath, 1994). It is

worth noticing, that in addition to benefiting Warbler species,

preservation of the undergrowth in Holm oak forests is related

to other ecological and forestry advantages such as the

protection of young saplings from herbivores and the

prevention of soil erosion (Meson and Montoya, 1993).

The most intense clearings should be concentrated in the

sectors with greater risk of fire, for example along forest edges

near farmland areas. A complementary measurement may be to

pile up part of the dry branches in small dispersed piles, of 1–

2 m in diameter and at least 1 m high. These piles simulate live

shrubs and when distributed sparsely throughout the clearing

zone do not increase the risk of fire. In cases where there is a

lack of living ligneous undergrowth, these piles may act as a

refuge for arthropods, small mammals and some undergrowth

birds, especially the Wren, the Robin and the Blackbird.

Finally, the management of small plots of grazed open Holm

oak forest with an arboreal layer below the 400 trees/ha, might

be beneficial for the protection of tree-dwelling birds, that avoid

both dense Holm oak forests as well as opened spaces without

trees, as long as these grazed Holm oak forests are located in

convenient places (preferably neighbouring meadows) and are

at least a few hectares in size. In this case, trees must grow at

low density without technical rotation, until they attain large

trunks and big crowns. These forests play an interesting role in

the local conservation of some scarce bird species that only

occur in open forests, such as Mistle Thrush, Spotted

Flycatcher, Turtle Dove or Nightjar. However, these open

and grazed Holm oak forests are very different ecologically to

semi-natural or dense Holm oak forests, so they should not be

thought of as a general management model, but as comple-

menting the traditional forestry for firewood and wood

production.

Acknowledgements

This study has been carried out thanks to the financing of the

Environmental Department of the Generalitat de Catalunya

(Catalan Government) and the Forest Technology Centre of

Catalonia. At the same time, to carry out the project we have

been able to count on the collaboration and the support of the

technical services of the Environmental Department, particu-

larly Joan Montserrat and Emili Bassols technicians of the

Natural Park of the Zona Volcanica de la Garrotxa (Catalonia).

Also, Enric Batllori, Joan Carles Abella and Marc Albert

Fernandez have collaborated in the field work. We are grateful

also to Xavier Ferrer, Santi Manosa, Raimon Marine, Pere Pons

and Roger Prodon for their participation in the sampling design

and the text revision. This work has been conducted under the

GDRE project ‘‘Mediterranean and Mountain systems in a

changing world’’. LB benefited from a Ramon y Cajal contract

from the Spanish government and JC from a travel grant from

the CTP and the Universities and Research Department of the

Catalan Government.

References

Bazzaz, F.A., 1991. Characteristics of populations in relation to disturbance in

natural and man modified ecosystems. In: Mooney, H.A., Godron, M.

(Eds.), Disturbance and ecosystems. Components of Response. Ecological

Studies 44. Springer-Verlag, Berlin.

Bibby, C.J., Burgess, N.D., Hill, D.A., 2000. Bird Census Techniques, 2nd ed.

Academic Press, London.

Blondel, J., 1981. Structure and dynamics of bird communities in Mediterra-

nean habitats. In: di Castri, F., Goodall, D.W., Specht, R.L. (Eds.), Me-

diterranean-Type Shrublands. Elsevier, Amsterdam, pp. 361–385.

J. Camprodon, L. Brotons / Forest Ecology and Management 221 (2006) 72–8282

Blondel, J., Aronson, J., 1999. Biology and Wildlife of the Mediterrenean

Region. Oxford University Press, Oxford.

Brooker, M., Rowley, I., 1995. The significance of territory size and quality in

the mating strategy of the splendid fairy-wren. J. Anim. Ecol. 64, 614–617.

Camprodon, J., 2001. Tratamientos forestales y gestion de la fauna vertebrada.

In: Camprodon, J., Plana, E. (Eds.), Conservacion de la biodiversidad y

gestion forestal. Edicions Universitat de Barcelona and Centre Tecnologic

Forestal de Catalunya, Barcelona.

Canterbury, G.E., Blockstein, D.E., 1997. Local changes in a breeding bird

community following forest disturbance. J. Field Ornithol. 68 (4), 537–546.

CPF, 1992. Documentacio Tecnica 1. El bosc mediterrani humit. Centre de la

Propietat Forestal. Barcelona.

Crawford, H.S., Hooper, R.G., Titterington, R.H., 1981. Songbird population

response to silvicultural practices in central Appalachian hardwoods. J.

Wildl. Manage. 45 (3), 680–692.

de Bolos, O., 2001. Vegetacio dels Paısos Catalans. Col�leccio Gaia, vol. 8.

Aster, Terrassa.

Dıaz, M., Carbonell, R., Santos, T., Tellerıa, J.L., 1998. Breeding bird com-

munities in pine plantations of the Spanish plateau: biogeography, land-

scape and vegetation effects. J. Appl. Ecol. 35 (4), 562–574.

Donald, P.F., Haycock, D., Fuller, R.J., 1997. Winter bird communities in forest

plantation in western England and their response to vegetation, growth stage

and grazing. Bird Study 44 (2), 206–219.

Emlen, J.T., 1970. Habitat selection by birds following a forest fire. Ecology 51,

343–345.

Fuller, R.J., 1995. Bird Life of Woodland and Forest. Cambridge University

Press, Cambridge.

Garolera, E., 1991. El bosc mediterrani humit. Explotacio forestal de Moll-

fulleda. Apunts de Silvicultura 4, 15–22.

Greenacre, M.J., 1984. Theory and Applications of Correspondence Analysis.

Academic Press, New York.

Hurlbert, S.H., 1984. Pseudoreplication and the design of ecological field

experiments. Ecol. Monogr. 54 (2), 187–211.

International Bird Census Committee (IBCC), 1969. Recommendations for an

international standard for a mapping method in bird census work. Bird

Study 16, 249–255.

James, F.C., Wamer, N.O., 1982. Relationships between temperate forest bird

communities and vegetation structure. Ecology 63, 159–171.

Janzen, D.H., 1986. The eternal external threat. In: Soule, M.E. (Ed.), Con-

servation Biology. The Science of Scarcity and Diversity. Sinauer,

Sunderland, pp. 286–303.

Levin, S.A., 1976. Population dynamics models in heterogeneous environ-

ments. Ann. Rev. Ecol. Syst. 7, 287–310.

Lovejoy, D.H., et al., 1986. Edge and other effects of isolation on Amazon

fragments. In: Soule, M.E. (Ed.), Conservation Biology. The Science of

Scarcity and Diversity. Sinauer, Sunderland, pp. 257–285.

MacArthur, R.H., 1964. Environmental factors affecting bird species diversity.

Ann. Nat. 98, 387–398.

McCullagh, P., Nelder, J.A., 1989. Generalized Lineal Models, 2nd ed. Capman

& Hall, London.

Meson, M., Montoya, M., 1993. Selvicultura Mediterranea. Mundi-prensa,

Madrid.

Muntaner, J., Ferrer, X., Martınez-Vilalta, A., 1983. Atles dels ocells nidificants

de Catalunya i Andorra. Ketres Editora, Barcelona.

Petersen, K.L., Best, L.B., 1987. Effects of prescribed burning on no game birds

in a sagebrush community. Wildl. Soc. Bull. 15, 317–329.

Pons, P., 1996. Dynamique de l’avifaune apres incendie et brulage dirige en

zone mediterraneenne. Importance de la territorialite. These de Doctorat.

Universite Paris 6.

Pons, P., 1998. Bird site tenacity after prescribed burning in a Mediterranean

scrubland. In: Trabaud, L. (Ed.), Fire Management and Landscape Ecology.

International Association of Wildland Fire, Washington, pp. 261–270.

Prodon, R., 1988. Dynamique des systemes avifauna-vegetation apres deprise

rurales et incendies dans les Pyrenees mediterraneennes siliceuses. These de

Doctorat de Etat. Universite Paris 6.

Prodon, R., 1992. Animal communities and vegetation dynamics: measuring and

modelling animal community dynamics along forest successions. In: Teller,

A., Mathy, P., Jeffers, J.R.F. (Eds.), Responses to Forest Ecosystems to

Environmental Changes. Elsevier Applied Science, London, pp. 126–141.

Prodon, R., Lebreton, J.D., 1981. Breeding avifauna of a Mediterranean

succession: the Holm oak and Cork oak series in the eastern Pyrenees.

1. Analysis and modelling of the structure gradient. Oikos 37, 21–38.

Prodon, R., Fons, R., Athias-Binche, F., 1987. The impact of fire on animal

communities in Mediterranean area. In: Trabaud, L. (Ed.), The Role of Fire

in Ecological Systems. SPBAcademic Publishing, Den Haag, pp. 121–157.

Renwald, J.D., 1977. Effect of fire on Lark Sparrow nesting densities. J. Range

Manage. 30, 283–285.

Shirihai, H., Gargallo, G., Helbig, A., 2001. Sylvia Warblers. In: Identification,

Taxonomy and Phylogeny of the Genus Sylvia, Christopher Helm. A & C.

Black, London.

Snow, D.W., Perrins, C.M., 1998. The Birds of the Western Palearctic, vol. 2:

Passerines, Concise Edition. Oxford University Press, Oxford.

Sokal, R.R., Rohlf, F.J., 1995. Biometry. Freeman, New York.

Sousa, W.P., 1984. The role of disturbance in natural communities. Ann. Rev.

Ecol. Syst. 15, 353–391.

Stewart-Oaten, A., Murdoch, W.W., Parker, K.R., 1986. Environmental-impact

assessment–pseudreplication in time. Ecology 67, 929–940.

Terradas, J. (Ed.), 1996. Ecologia del foc. Edicions Proa, Barcelona.

Tellerıa, J.L., 1986. Manual para el censo de los vertebrados terrestres. Editorial

Raıces, Madrid.

Tucker, G.M., Heath, M.F., 1994. Birds in Europe. Their Conservation Status.

BirdLife Conservation Series No. 3. BirdLife International, Cambridge.

Wiens, J.A., 1989. The Ecology of Bird Communities. Cambridge University

Press, Cambridge.

Wilson, M.F., 1974. Avian community organization and habitat structure.

Ecology 55, 1017–1029.

Winter, B.M., Best, L.B., 1985. Effect of prescribed burning on placement of

Sage Sparrow nests. Condor 87, 294–295.

Wright, J.M., 1981. Response of nesting Lapland longspurs (Calcarius lappo-

nicus) to burned tundra on the Seward Peninsula. Arctic 34, 366–369.