SM

Ga

b

c

a

ARR1A

KHHLW

I

ad(mashpeTdo

1h

Journal for Nature Conservation 22 (2014) 103–112

Contents lists available at ScienceDirect

Journal for Nature Conservation

j o ur nal homepage: www.elsev ier .de / jnc

patially explicit conservation issues for threatened bird species inediterranean farmland landscapes

ianpasquale Chiatantea,∗, Mattia Brambillab,c, Giuseppe Bogliania

Department of Earth and Environmental Sciences, University of Pavia, via Ferrata 9, 27100 Pavia, ItalyMuseo delle Scienze, Sezione di Zoologia dei Vertebrati, Corso del Lavoro e della Scienza 3, I-38123 Trento, ItalyFondazione Lombardia per l’Ambiente, Settore Biodiversità e aree protette, Largo 10 luglio 1976 1, I-20822 Seveso, MB, Italy

r t i c l e i n f o

rticle history:eceived 7 January 2013eceived in revised form9 September 2013ccepted 26 September 2013

eywords:abitat managementabitat suitability modelesser Grey Shrike

oodchat Shrike

a b s t r a c t

Coupling habitat models based on GIS and on ground variables could help identify suitable areas (bymeans of landscape models obtained by GIS variables) to concentrate management actions for species’conservation. In this study, the habitat requirements of Lesser Greys (LGS) and Woodchat Shrikes (WS),two threatened farmland bird species declining in Europe, were assessed in Apulia (south-eastern Italy)by means of binary logistic regression at two different levels: landscape (using GIS-measured variables);and, territory (using ground-measured variables) scales. The LGS occurrence at landscape scale was cor-related to steppe-like areas and cereal crops. At the territory level, significant effects were detected fordeciduous forests and the presence of isolated trees and shrubs. The WS occurrence at landscape scalewas promoted by steppe-like areas and cereal crops, whereas, at the territory level significant effectswere detected for steppe-like areas positively and suburban areas negatively. The landscape model wasextrapolated to the entire region. Within highly suitable areas (occurrence probability higher than 0.66according to the landscape model), we measured average habitat features and compared them with theoptimal mosaic depicted by the territory level models. This allowed us to give spatially explicit and site-specific management recommendations for these two threatened species. LGS will mostly benefit from

an increase in isolated shrubs and trees; whereas for WS, the most widespread recommendations are toincrease steppe-like habitat and to prevent further urbanisation.

Coupling “coarse” landscape models with the species ecology provided by fine-scaled models can inte-grate relevant information on species potential distribution and territory level requirements, makingplanning fine-tuned habitat management (within potentially suitable landscapes) in a spatially explicit

way possible.

ntroduction

Farmland bird species represent a large proportion of Europeanvifauna, and the populations of several species have suffered aramatic decline in recent decades, especially in Western EuropeDonald et al. 2002). The causes of this decline have been identified

ostly in the changes of agricultural practices, such as heavy mech-nisation, increased fertiliser inputs, and a temporal shift of cerealowing from spring to autumn. In addition, the loss of landscapeeterogeneity, through the destruction of hedgerows, shrubs, treeatches, and other natural areas, following intensification (Bentont al. 2003; Donald et al. 2002; Fuller et al. 1995; Newton 2004).

hese changes have led both to the reduction of refuge and repro-uction areas and to the decrease in invertebrate prey, the latterf which is also prompted by the increase in biocide use (Benton

∗ Corresponding author. Tel.: +39 3331868129.E-mail address: harrier84@libero.it (G. Chiatante).

617-1381/$ – see front matter © 2013 Elsevier GmbH. All rights reserved.ttp://dx.doi.org/10.1016/j.jnc.2013.09.006

© 2013 Elsevier GmbH. All rights reserved.

et al. 2002; Boatman et al. 2004; Genghini et al. 2006; Wilson et al.1999). A further cause of farmland species’ decline is represented byland abandonment (Donald et al. 2002; Rippa et al. 2011; Suarez-Seone et al. 2002), which is threatening important farmland birdpopulations in mountain areas (Brambilla et al. 2010).

A set of agri-environmental policies (AEPs) has been initi-ated to make safeguarding agro-ecosystems and their dependentspecies possible. Targeting is, therefore, required to direct agri-environment funding to those areas and actions which will providethe greatest environmental results (Thompson et al. 1999; Webster& Felton 1993). Consequently, knowledge of the distribution andcharacterisation of preferred species’ habitats is essential to achiev-ing this goal and, hence, species’ conservation and managementplanning.

Resource Selection Functions can be used to synthesise the pro-

cess of habitat selection into a mathematical expression in orderto predict the distribution of a species (Boyce & McDonald 1999;Boyce et al. 2002; Jedrzejewski et al. 2008). Geographical Infor-mation Systems (GIS) are often used for this purpose (Chow et al.

1 ture C

2Gfia2v(ibra(et(e

tpt2

aasoorodhacaafis(BsscipohiaMssgfic

omtwfiwatsah

04 G. Chiatante et al. / Journal for Na

005; Jedrzejewski et al. 2008; Olivier & Wotherspoon 2005), butIS environmental layers, although capable of operating at veryne scales in some specific cases, often only give coarse informationbout a habitat at landscape level (Jaberg & Guisan 2001; Numa et al.009; Westphal et al. 2003), which may not be sufficient to pro-ide detailed knowledge of species’ ecological needs at a finer scaleBrambilla et al. 2009). As a consequence, it is often necessary tomprove GIS-based definitions of suitable areas for a given speciesy integrating them with a fine-scaled study of habitat prefe-ences based on variables measured in the field in order to plann operational habitat management scheme for the study speciesBoyce 2006; Brambilla et al. 2009; Parody & Milne 2004; Suzukit al. 2008). Effective habitat management can be developed only ifhe habitat preferences of a species are known with sufficient detailHinsley & Bellamy 2000; Hobbs 1997; Mabry et al. 2010; Moreirat al. 2005).

A spatially explicit map of species’ distribution in terms of habi-at suitability could assist conservation managers by identifyingotentially suitable areas within which to concentrate conserva-ion or management actions (Smith et al. 2007; Traill & Bigalke006).

Conversely, when the factors affecting species occurrencend/or abundance operate at a fine-scale and the underlying vari-bles are poorly represented as GIS layers (see above), mappingpecies’ potential distribution can be difficult. In this case, carryingut field surveys of the actual distribution of the species in a portionf the area concerned, in parallel with the measurement of the envi-onmental variables at a small scale, may allow for the identificationf potentially suitable areas within which species occurrence isetermined by the presence/absence of fine-scale determinants ofabitat selection. Combining the use of GIS models with fine-scalednalyses of habitat preferences and habitat features of study areasould, therefore, offer opportunities to plan effective managementctions within suitable landscapes. This type of an approach wouldllow for the identification of the measures needed to increasene-scaled habitat suitability within “structurally” suitable land-capes and bridge the gap between coarse landscape structureand relative distribution models) and within-territory habitat (seerambilla et al. 2009). Previous works suggested a set of mea-ures to be applied at the fine scale within landscapes potentiallyuitable at the large scale (Brambilla et al. 2009). This approachan be further developed to promote full integration between thenformation provided by the two different evaluations of habitatreferences by explicitly mapping the area-specific managementptions required at the territory-level on the basis of fine-scaleabitat preferences within areas suitable at the landscape scale and

dentified on the basis of the GIS model within which the variablesffecting fine-scale habitat preferences have also been quantified.easuring fine-scale variables within potentially suitable land-

capes and the relative departure from optimal values for targetpecies would allow for an effective planning of habitat mana-ement with site-specific recommendations aimed at maximisingne-scale habitat suitability within areas potentially suitable at theoarse-scale.

The aim of this research was to carry out a two-level assessmentf habitat preferences and to propose spatially explicit manage-ent recommendations in two species of conservation concern:

he lesser grey shrike (Lanius minor J.F. Gmelin, 1788); and, theoodchat shrike (Lanius senator Linnaeus, 1758). Firstly, we identi-ed potentially suitable landscapes at the regional scale. Secondly,e identified fine-scaled habitat variables affecting occurrence

t the territory level in order to develop a conservation strategy

hrough dedicated habitat management in farmland habitats inouthern Italy for these two threatened species. After the two-levelssessment of habitat preferences on the basis of the fine-scaleabitat traits of the potentially suitable areas, we considered

onservation 22 (2014) 103–112

specific fine-scale features of each potentially suitable area andproposed management recommendations for each individual areawhich aim at reducing the discrepancy between actual and optimalhabitat composition within the areas. This three-step to a spatiallyexplicit definition of management recommendations based on theintegration between landscape models (which identified the areas)and fine-scaled models (which identified habitat features associ-ated with species occurrence) to identify limiting factors withineach area specifically.

Methods

Study model

Shrikes (Laniidae) are highly associated with farmed landscapesand have, therefore, suffered significant declines in their populationdistribution and size (Yosef 1994; Yosef & Lohrer 1995; Yosef &Lohrer 1998; Yosef et al. 2000).

Within this family, considerable attention has been given tothree of the five species breeding in western Europe: Lanius col-lurio (see e.g. Brambilla & Ficetola 2012; Brambilla et al. 2009,2010; Ceresa et al. 2012; Goławski & Meissner 2008; Söderström& Karlsson 2011) and the two species Lanius excubitor and Laniusmeridionalis (e.g. Karlsson 2002; Keynan & Yosef 2010; Kuczynskiet al. 2010; Olsson et al. 2010; Padilla et al. 2009). The other twospecies, L. senator and L. minor, have been poorly investigated (butsee Giralt & Valera 2007; Guerrieri & Castaldi 2000; Guerrieri &Castaldi 2010; Guerrieri et al. 1995; Hernandez 1994; Isenmann &Debout 2000; Moskát & Fuisz 2002).

L. minor inhabits open areas with small woods and scatteredtrees and is strongly associated with traditional (low intensity)agricultural landscapes (Cramp & Perrins 1993; Harris & Franklin2000; Lefranc & Worfolk 1997). Since the early part of the 20thCentury, this species has suffered a steady decline with the extinc-tion and the reduction of several European populations (BirdLifeInternational 2004; Lefranc 1995; Lefranc & Worfolk 1997). Con-sequently, the species is listed on Annex I of the Birds Directive(2009/147/EC). One of the main causes of their decline is the inten-sification of agriculture by landscape alteration, loss of hedgesand rows, and the heavy use of fertilisers and biocides. Climatechanges in reproductive areas and drought in the wintering areasare thought to be other causes of the species’ decline (Kristin &Lefranc 1997; Lefranc 1995).

L. senator is a smaller species mainly associated with Mediter-ranean areas where it breeds in grassland with shrubs and scatteredtrees, arid steppes, and semi-desert. This species also breeds inplantations, particularly olive groves. In the northern part of itsdistribution, it is a typical inhabitant of traditional orchards withscattered trees (Cramp & Perrins 1993; Harris & Franklin 2000;Lefranc & Worfolk 1997). It has undergone a steady decline in pastdecades throughout its entire range due to agriculture intensifi-cation, forestry, and fires (BirdLife International 2004; Hernandez1997; Lefranc & Worfolk 1997). Climate change leading to more fre-quent wet springs is also thought to be a major threat to this shrikespecies. Prolonged droughts and changes in agricultural practicesin wintering areas are thought to be the other causes of populationdecline. In Italy, Spain, and North Africa the species is illegally butregularly poached (Harris & Franklin 2000; Hernandez 1997).

Study area

The study area includes the whole territory of the Apulia regionin south-eastern Italy (19,358 km2) (Fig. 1). The area is dominatedby lowland plains with hills and small mountains in the north-western portion of the region (highest peak 1151 m a.s.l.). The

G. Chiatante et al. / Journal for Nature Conservation 22 (2014) 103–112 105

(2) Su

lvtt

F

s

TTr

Fig. 1. Apulia region. The sample areas are shown in grey: (1) Gargano;

andscape is composed of cereal crops (42.4%), olive groves (24.2%),ineyards (7.5%), grasslands (5.6%), and deciduous forests (5.4%). Inhe present study, five sampling areas were checked (Table 1) for aotal area of about 4500 km2.

ieldwork

Between June 1st and July 31st in 2009 and 2010 the twohrike species were censused through territory mapping (Bibby

able 1opographic and land use characteristics of the five sampling areas studied for thisesearch.

Sampling area

Gargano Mean altitude (m a.s.l.) 528Main orientation (% cover) S-SE (35.2)Steppe-like areas (% cover) 26.5Deciduous woodlands (% cover) 21.7Cereal crops (% cover) 12.0Olive groves (% cover) 10.4

Sub-Appennino Dauno Mean altitude (m a.s.l.) 364Main orientation (% cover) N-SE (54.6)Cereal crops (% cover) 76.5

Tavoliere Mean altitude (m a.s.l.) 135Main orientation (% cover) flat (51.0)Cereal crops (% cover) 65.3Vineyards (% cover) 15.8Olive groves (% cover) 11.7

Murge plateau Mean altitude (m a.s.l.) 377Main orientation (% cover) flat (25.8)Cereal crops (% cover) 45.4Olive groves (% cover) 12.2Deciduous woodlands (% cover) 11.4Steppe-like areas (% cover) 11.1

Salento Mean altitude (m a.s.l.) 43Main orientation (% cover) flat (67.6)Olive groves (% cover) 39.8Cereal crops (% cover) 26.8Vineyards (% cover) 9.6

b Appennino Dauno; (3) Tavoliere; (4) Murge plateau; and, (5) Salento.

et al. 2000). Each study area was surveyed a minimum of twoand a maximum of five times with the number of visits dictatedaccording to the area, terrain, and number of previous contactswith the target species. Using this method, pairs were locatedand the territory boundaries were defined by the observation ofbreeding behaviours, such as courtship, copulation, nest building,and nestling rearing. This technique has been used elsewhere withother Lanius species and is based on the highly territorial and con-spicuous behaviour of these species (Brambilla & Ficetola 2012;Brambilla et al. 2009; Brambilla et al. 2010; Ceresa et al. 2012;Guerrieri & Castaldi 2010; Karlsson 2004).

Model building

All territories mapped during the two breeding seasons wereused to develop two models for each species in order to identifyenvironmental variables likely to be important for their habitatselection at the landscape and fine scale with the latter respec-tively corresponding to the territory level. The models obtained atthe landscape scale were then extrapolated to the whole region toobtain habitat suitability maps displaying the potential distribu-tion of the two species in Apulia. Spatial data was processed witha GIS platform (ArcGIS 9.2, ESRI, Redlands, CA). To develop the mo-dels, nests or the main centre of territories and an equal number ofcontrol points (selected differently for the landscape and fine scalemodels, respectively) were used (see below for details). The centreof the territory was defined as the mid-point of all the territorialcontacts from a given pair.

To identify the most important environmental factors for habitatselection at the landscape level and to formulate a model of habi-tat suitability at a regional scale for the two species, we consideredland-use variables (urban areas, suburban areas, cereal crops, vine-

yards, plantations, woods, steppe-like areas with scattered trees,and shrublands) derived from land use maps at the fourth level ofCORINE 1:5000 (SIT – Regione Puglia) as well as topographic vari-ables including altitude (m a.s.l.), orientation (degrees from North),

1 ture C

aM

frs12e

Poidoohans2tdtWndeTsfai

wgoiiuoioclopoh

S

bttmmABRmn2Mr

06 G. Chiatante et al. / Journal for Na

nd slope (degrees), obtained from 8 m resolution Digital Elevationodel (SIT – Regione Puglia).As orientation is a circular variable expressed in degrees

rom the North, we transformed it into a linear factor usingeclassification scores indicating the different degree of expo-ure to the sun (0–22.5◦ = 0; 22.5–67.5◦ = 1; 67.5–112.5◦ = 2;12.5–157.5◦ = 3; 157.5–202.5◦ = 4; 202.5–247.5◦ = 3;47.5–292.5◦ = 2; 292.5–337.5◦ = 1; 337.5–360◦ = 0) (Brambillat al. 2009).

Field observations and the work of previous authors (Cramp &errins 1993; Lefranc & Worfolk 1997) indicated that the territoriesf the two species have different dimensions. L. minor may breedn aggregations with pairs breeding a few metres from each otherefending only the immediate vicinity of the nest with territoryverlap (Cramp & Perrins 1993). This condition only occurred inne instance during this research. In general, L. minor territoriesave a variable size from 1.9 to 23 ha (Harris & Franklin 2000),lthough, most prey are captured within a radius of 150 m from theest (Giralt et al. 2008; Lefranc & Worfolk 1997). L. senator defendsmaller territories ranging from 0.5 to 12 ha (Harris & Franklin000) and varying in size according to the size of the overall habi-at block (Harris & Franklin 2000) and to the reproductive phase;uring the rearing of nestlings, the territories are larger because ofhe increased food requirements to feed the nestlings (Lefranc &

orfolk 1997). In this research, buffers were constructed aroundests or territory centres and, for an equal number of control pointsistributed randomly throughout the region. The points were gen-rated with the “Generate Random Points” procedure from Hawth’sools extension, within which environmental variables were mea-ured. For L. minor, a 150 m radius buffer was used (7.1 ha), whileor the L. senator the radius of the buffer was set at 75 m (1.7 ha)ccording to the different average size of the respective territoriesndicated in the literature.

To model habitat preferences at the fine scale for the two species,e used data obtained from habitat variables measured on the

round. Within a buffer radius of 75 m centred on the territoriesf both species, an equal number of points were randomly placedn the whole sampling area, and we measured variables describ-ng the vegetation structure in the territory/random plots. We alsosed a radius of 75 m for L. minor, because estimating the coverf different vegetation types in the field was too complicated andmprecise for larger plots. We considered topographic (altitude,rientation, and slope) and land-use cover variables (urban areas,ereal crops, pastures, fallows, steppe-like areas, vineyards, shrub-ands, olive groves, mixed plantations with olive and almond trees,ther plantations, and deciduous forests), along with the length andresence of specific habitat elements (length of paved roads, lengthf unpaved roads, length of electrical cables, and the presence ofedges, rows, isolated trees, and isolated shrubs).

tatistical analyses

Habitat features of territories and control points were comparedy means of a Mann–Whitney U test (Fowler & Cohen 2002) both athe landscape and fine scales. To compare binomial variables, a �2

est was used (Fowler & Cohen 2002). To develop habitat preferenceodels for shrikes at the landscape and fine scales, we carried out aodel selection procedure according to an Information-Theoreticpproach (Anderson & Burnham 2002; Anderson et al. 2000, 2001;urnham & Anderson 2002; Garamszegi 2011; Rayner et al. 2007;ichards et al. 2011; Whittingham et al. 2006) particularly byeans of the second-order AIC (AICc) that should be used when

/k < 40 (in which k is the parameters number) (Anderson et al.000, 2001; Burnham & Anderson 2002; Gray et al. 2010; Merli &eriggi 2006; Tsaparis et al. 2009). This method stems from the

ecognition that data seldom provides absolute support for a single

onservation 22 (2014) 103–112

hypothesis; rather, data can only influence the extent to which wefeel that any given hypothesis is supported (Burnham & Anderson2002). For habitat variables, we checked for a possible quadraticeffect. The ability of the models to distinguish between occupiedand unoccupied sites was tested by the means of the area underthe curve of the Receiver Operating Characteristic (ROC) plot (Boyceet al. 2002; Fawcett 2006; Gray et al. 2010; McAlpine et al. 2008;Pearce & Ferrier 2000). This area provides a measure of discrimina-tion ability varying from 0.5 for a model with discrimination abilityno better than random, to 1.0 for a model with perfect discrimina-tory ability. All the statistical analyses were carried out in R version2.12.1 (www.cran.r-project.org).

Mapping distribution of potentially suitable habitat

The landscape scale models allowed for the production of habi-tat suitability maps that graphically represent the probability ofspecies occurrence in the region. These maps were produced usingthe Kriging interpolation method (see also Brambilla et al. 2009,2010) carried out in ArcGIS 9.2, after calculating the occurrenceprobability of the two species in 20,000 points randomly scat-tered over the whole region. The occurrence probability was thenclassified into three classes: scarce suitability (0–0.33); mediumsuitability (0.33–0.66); and, high suitability (0.66–0.99).

Spatially explicit recommendations for habitat management

The fine scale model allowed for an assessment of the optimalhabitat requirements of both L. minor and L. senator. Then, utilisingonly the highly suitable areas identified by the landscape model,we compared (i) the habitat composition of each cell in a 300 m-spaced grid with the optimal habitat mosaic for a L. minor territory,(ii) the habitat composition of a 150 m spaced grid with the opti-mal habitat mosaic for a L. senator territory. Grid-cell sizes wereset to be comparable with territory size and with the radius consi-dered for model building (see above). Optimal habitat compositionwas defined on the basis of the average characteristics of selectedhabitats in the study area (in terms of habitat occurrence and rel-ative cover of preferred and avoided habitats). In this way, it waspossible to plan and map the most important habitat managementrecommendations for improving the habitat of the target species;within suitable landscapes, fine-scaled habitat suitability can beachieved or improved by (re)creating the optimal habitat mosaic forthe target species. By comparing optimal and actual habitat com-position within grid cells, we suggested what management optionsare needed within each cell to increase fine-scale habitat suitability.

Results

Lanius minor

In 2009 and 2010, we mapped 29 and 31 L. minor territoriesrespectively. Significant differences were found between the sam-ple areas (�2 = 100.167, df = 4, P < 0.001), with 70% of breeding pairson the Murge plateau.

At the landscapes scale, altitude, orientation, and percent coverof steppe-like areas, urban areas, plantations, and cereal cropswere significantly different between territories and control plots(Table 2). At the fine scale, the percent cover of vineyards andthe length of electric cables were significantly different betweenterritories and control plots (Table 2).

The most parsimonious models according to the AICc-basedranking at the two scale levels are shown in Table 3. The area underthe curve of the ROC plot for the landscape and the fine scale modelwere 0.923 (P < 0.001) and 0.802 (P < 0.001) respectively.

G. Chiatante et al. / Journal for Nature Conservation 22 (2014) 103–112 107

Table 2Comparison between habitat features of L. minor territories and random points at landscape and fine scale. The values are presented as mean ± SD.

Variables Territories (N = 60) Random points (N = 60) P

Landscape scaleAltitude (m) 353.55 ± 17.99 220.51 ± 24.03 <0.001Orientation (◦N) 154.18 ± 8.82 92.90 ± 8.73 <0.001Steppe-like areas (% cover) 15.30 ± 3.50 2.77 ± 1.70 0.002Urban areas (% cover) 3.22 ± 0.36 6.85 ± 2.18 0.023Plantation (% cover) 14.58 ± 2.93 29.03 ± 4.46 0.025Cereal crops (% cover) 57.32 ± 3.90 40.50 ± 4.94 0.011

Fine scaleVineyards (% cover) 2.02 ± 1.34Electric cables (m) 118.98 ± 11.79

Table 3The most parsimonious models at landscape and fine scales obtained by theInformation-Theoretic Approach for L. minor.

Variables ± s.e.

Landscape scaleIntercept −6.64 ± 1.46Altitude 0.02 ± 0.01Altitude2 −2.3 × 10−5 ± 1.1 × 10−5

Slope −0.47 ± 0.18Orientation 0.02 ± 0.01Steppe-like areas (% cover) 19.78 ± 7.34Steppe-like areas2 (% cover) −15.13 ± 7.29Cereal crops (% cover) 8.87 ± 3.63Cereal crops2 (% cover) −6.55 ± 3.28

Fine scaleIntercept −1.29 ± 0.51Deciduous forests (% cover) 9.68 ± 5.82Plantations (% cover) −19.57 ± 12.57Length of electric cables 0.01 ± 3.0 × 10−3

Presence of isolated shrubs 2.05 ± 0.73 (for shrubs occurrence)Presence of isolated trees 0.79 ± 0.48 (for tree occurrence)

Slope −0.17 ± 0.11

Fig. 2. Habitat suitability map

5.47 ± 1.94 0.01659.47 ± 10.31 <0.001

The model formulated at the landscape scale allowed us to builda habitat suitability map for L. minor in Apulia (Fig. 2). The meanvalue of habitat suitability for this species in the region is 0.24 withnull or scarce suitability in 71.7% of the territory, medium suitabilityin 21.8% and high suitability in 6.5%.

Considering the output of the fine-scaled model for L. minorterritories, we assumed that the optimal territory for this speciesshould include an average land cover equal to 2% of deciduousforests which is the mean value of the percent cover of this land usein the L. minor territories (min. 0%, max. 30%; higher cover is likelyto have a negative effect given the ecology of the species) withoutplantations and with isolated trees and shrubs. Then, consideringthe highly suitable areas identified on the basis of the landscapescale model and comparing their own features with the ones ofthe optimal habitat mosaic (see above), the following recommen-dations are given: in 9.3% of the highly suitable areas, forests aretoo widespread, in 0.04%, both forests and plantations should bereduced, in 10.6%, isolated shrubs should be increased, and in 4.4%,isolated trees should be increased (see Table 6 and Fig. 4 as an

example of management recommendations).

for L. minor in Apulia.

108 G. Chiatante et al. / Journal for Nature Conservation 22 (2014) 103–112

Table 4Comparison between habitat features of L. senator territories and random points at landscape and fine scale. The values are presented as mean ± SD.

Variables Territories (N = 54) Random points (N = 54) P

Landscape scaleAltitude (m) 327.22 ± 20.75 210.33 ± 27.96 <0.001Orientation (◦N) 117.44 ± 9.16 89.46 ± 9.10 0.014Slope (◦) 4.01 ± 0.45 3.16 ± 0.60 0.012Urban areas (% cover) 4.33 ± 0.54 6.85 ± 2.27 <0.001

Fine scaleIsolated trees (% frequency) 59.26

Length electric cable (m) 115.50 ± 13.64

Table 5The most parsimonious models at landscape and fine scales obtained by theInformation-Theoretic Approach for L. senator.

Variables ± s.e.

Landscape scaleIntercept −4.65 ± 1.02Altitude 0.02 ± 0.01Altitude2 −3.2 × 10−5 ± 9.3 × 10−6

Orientation 0.01 ± 4.2 × 10−3

Steppe-like areas (% cover) 3.32 ± 1.87Cereal crops (% cover) 7.97 ± 2.73Cereal crops2 (% cover) −7.20 ± 2.64

Fine scaleIntercept −0.10 ± 0.71Altitude −3.4 × 10−3 ± 1.8 × 10−3

Slope 0.29 ± 0.17Slope2 −0.02 ± 0.01

L

tdP

identified on the basis of the landscape scale model and compa-

Steppe-like areas (% cover) 2.28 ± 1.06Suburban areas −6.99 ± 5.47Length of electric cables 0.01 ± 2.5 × 10−3

anius senator

In 2009 and 2010, we mapped 37 and 17 territories, respec-ively, and found a 54.1% reduction of breeding pairs. Significantifferences were found between sample areas (�2 = 104.148, df = 4,

< 0.001), with 74.1% of the breeding pairs on the Murge plateau.

Fig. 3. Habitat suitability map

37.04 0.03469.34 ± 11.43 0.008

At landscape scale, the percent cover of urban areas was signi-ficantly different between territories and control plots as altitude,orientation, and slope (Table 4). At the fine level, the presence ofisolated trees and the length of the electric cables were significantlydifferent between territories and control plots (Table 4).

The most parsimonious models according to the AICc-basedranking at the two scale levels are shown in Table 5. The area underthe curve of the ROC plot for the landscape and the fine scale modelwere 0.846 (P < 0.001) and 0.761 (P < 0.001) respectively.

On the base of the landscape scale model, it was possible toreclassify the whole regional territory to obtain a habitat suitabilitymap for L. senator (Fig. 3). The mean suitability value of the regionis 0.37, with a null or scarce suitability in 50.4% of the territory,medium suitability in 28.9% and high suitability in 20.7%.

Considering the output of the fine-scaled model for L. senatorterritories, we assumed that the optimal territory for this speciesshould include at least 78% of steppe-like areas, which is the meanvalue of the percent cover of this land use in the L. senator territories(min. 0%, max. 100%), and should have no or limited suburban areas,because the latter was demonstrated as having a negative effecton species occurrence. Thus, considering the highly suitable areas

ring their own features with the ones of the optimal habitat mosaicaccording to both scale models, the following recommendations aregiven: in 56.4% of the areas, the cover of steppe-like habitat should

for L. senator in Apulia.

G. Chiatante et al. / Journal for Nature Conservation 22 (2014) 103–112 109

Fig. 4. A detail of the recommended actions for habitat managem

Table 6Summary of management recommendations formulated according to the specificdifferences between optimal and actual habitat composition in each of the five studyareas.

Sampling areas Habitat recommendations % cover

Lanius minorGargano Less forests 1.6

Less forests, less plantations 0.4More isolated trees, moreisolated shrubs

90.9

No management actions 7.1Sub Appennino Dauno Less forests, less plantations 1.0

More isolated trees, moreisolated shrubs

29.8

No management actions 69.2Tavoliere More isolated trees, more

isolated shrubs100.0

Murge plateau Less forests 10.4Less forests, less plantations 1.9More isolated trees, moreisolated shrubs

49.5

No management actions 38.2

Lanius senatorGargano More steppe-like areas 39.7

Less suburban areas 8.4More steppe-like areas, lesssuburban areas

9.0

No management actions 42.9Sub Appennino Dauno More steppe-like areas 72.7

Less suburban areas 1.5More steppe-like areas, lesssuburban areas

22.7

No management actions 3.1Tavoliere More steppe-like areas 79.6

Less suburban areas 2.0More steppe-like areas, lesssuburban areas

16.4

No management actions 2.0Murge plateau More steppe-like areas 50.9

Less suburban areas 4.7More steppe-like areas, lesssuburban areas

29.7

No management actions 14.7

ent targeted at increasing habitat suitability for L. minor.

be increased to reach the optimal condition, in 3.8%, suburbs aretoo widespread so it is recommended to avoid further urbanisa-tion, in 27.9%, suburban areas are too widespread whilst steppe-likehabitat should be increased, and in 11.9%, no habitat managementaction should be realised (see Table 6 as an example of managementrecommendations).

Discussion

There are numerous examples of spatial targeting used in natureconservation (Bailey et al. 2006; Lee & Thompson 2005; Lee et al.,2002; Thompson et al. 1999; Webster & Felton 1993). Much of thisresearch has used Geographical Information Systems (GIS), becausethey are appropriate for manipulating large quantities of data andfor combining data collected from various sources and from a va-riety of appropriate spatial scales. Information on significant habi-tat associations could be used to encourage species to settle at siteswhere those species do not yet exist (or have low numbers) and,thus, support management decisions.

In this study, the habitat preferences of two poorly known andthreatened farmland birds were assessed by the combined use ofa landscape model and of a fine-scale model. The first one enabledus to identify the most suitable areas at the regional scale for thetarget species in terms of land use and topography. The fine-scalemodel was deemed important as it yielded complementary andfiner detailed knowledge about the habitat requirements of thetarget species, thus, providing the basic information for habitatimprovement by means of specific management recommendations.As a consequence of the outputs of the landscape model, these re-commendations could be explicitly mapped and related to specificareas or portion of areas. Our approach led to the identificationof what and where habitat management is required to conservethreatened species by improving their habitat on a site by site basis.

Our results show that, in Apulia, both L. minor and L. senatorinhabit areas with similar characteristics. They prefer territoriesbetween 200 and 600 m above sea level and relatively flat terrainon south-facing slopes. On the Tyrrhenian coast of Italy, L. minor

1 ture C

p1Rda(1WsPaaiita1tiar(hfifswtifts

bml1lmc

ciifec

C

nmp

ss(tltrafs

10 G. Chiatante et al. / Journal for Na

refers areas between sea level and 200 m (Castaldi & Guerrieri,995) or between 400 and 600 m a.s.l. (Mastronardi, Di Sarra, &icchi, 1996); the negative influence of slope for L. minor is alsoemonstrated in Latium (Guerrieri & Castaldi 2010). Open areasre generally reported as the appropriate habitat for the speciesCramp & Perrins 1993; Giralt et al. 2008; Guerrieri & Castaldi999; Guerrieri & Castaldi 2010; Harris & Franklin 2000; Lefranc &orfolk 1997; Moga et al. 2010); as a consequence, open habitats

hould be extended to guarantee the necessary trophic resources.erch sites such as electric cables and isolated trees and shrubsre essential to the species which use them to defend territoriesnd to forage in their typical “sit-and-wait” hunting technique. Themportance of the availability of perch sites has also been reportedn other shrike species, because their availability could affect bothhe territory size and the reproductive success (e.g. a greater avail-bility of perches results in a reduction of territory size) (Yosef,993; Yosef & Grubb, 1994). Accordingly, adult energy expenditureo provide food for offspring decreases while reproductive successncreases (Yosef, 1993; Yosef & Grubb, 1994). Further, increasedvailability of suitable areas results in a decrease of shrikes’ territo-ies size, which could then enable other breeding pairs to establishYosef, 1993; Yosef & Grubb, 1994). Pronounced differences inabitat association between the two species clearly emerge at thene scale. L. minor inhabits areas with small patches of deciduous

orests, which represent source areas for the invertebrates that con-titute much of their diet. A similar situation was observed in Spainhere natural vegetation hosts a richer and more abundant inver-

ebrate community than arable lands, because of the use of biocidesn the latter (Giralt et al. 2008). L. minor avoids plantations exceptor extensive olive and almond groves, because the availability ofrophic resources is low on plantations due to frequent pesticidepraying.

L. senator avoids suburban areas, likely because of human distur-ance. The potential distribution of the two species is similar, butore restricted for L. minor than for L. senator, which appears to be

ess selective as observed in other sites in Italy (Guerrieri & Castaldi999). The specific habitat requirements of L. minor make them

ess adaptable to environmental changes; thus, habitat improve-ent for this threatened species is an important component of their

onservation management.The recommendations for habitat management provided here

an inform specific agro-environmental schemes to be includedn the next regional rural development program. In particular,ncreasing steppe-like habitats through the conversion of intensivearming systems and the conservation and re-creation of marginallements such as isolated trees and shrubs are key-points for theonservation of those threatened species.

onclusions

Undoubtedly, to safeguard species and to ensure cost effective-ess of agri-environmental schemes, it is beneficial to concentrateanagement actions around existing highly suitable habitat

atches (Bailey et al. 2006).Our modeling approach showed how coupling “coarse” land-

cape models (obtained by using GIS-measured variables) withpecies specific ecological data provided by fine-scaled modelsobtained via variables measured directly on the ground) can allowhe predicted species distribution to be integrated with territoryevel requirements. The results of this approach not only identifiedhe land parcels of high habitat suitability for the species’ occur-

ence, but, also, identified the habitat structure for the species. Thispproach makes it possible to plan fine-tuned habitat managementor the target species within potentially suitable landscapes in apatially explicit way.

onservation 22 (2014) 103–112

In the case of the two threatened shrike species we studied,management should primarily consider mechanisms to increaseperching and nesting sites (isolated trees and shrubs) in areas withnatural vegetation in open landscapes, low levels of urbanisation,at an elevation of between 200 and 600 m a.s.l., and on flat terrainwith a southwards aspect.

Acknowledgements

We thank the anonymous referees for reviewing the manuscript.We appreciate the improvements in English usage made by KelseyHorvath.

References

Anderson, D. R., & Burnham, K. P. (2002). Avoiding pitfalls when using information-theoretic methods. Journal of Wildlife Management, 66, 912–918.

Anderson, D. R., Burnham, K. P., & Thompson, W. L. (2000). Null hypothesis test-ing: Problems, prevalence and an alternative. Journal of Wildlife Management,64, 912–923.

Anderson, D. R., Link, W. A., Johnson, D. H., & Burnham, K. P. (2001). Suggestionfor presenting the results of data analyses. Journal of Wildlife Management, 65,373–378.

Bailey, N., Lee, J. T., & Thompson, S. (2006). Maximising the natural capital benefits ofhabitat creation: Spatially targeting native woodland using GIS. Landscape andUrban Planning, 75, 227–243.

Benton, T. G., Bryant, D. M., Cole, L., & Crick, H. Q. P. (2002). Linking agriculturalpractice to insect and bird populations: A historical study over three decades.Journal of Applied Ecology, 39, 673–687.

Benton, T. G., Vickery, J. A., & Wilson, J. D. (2003). Farmland biodiversity: Is habitatheterogeneity the key? Trends in Ecology and Evolution, 18, 182–188.

Bibby, C. J., Burgess, N. D., Hill, D. A., & Mustoe, S. (2000). Bird census techniques.London: Academic Press.

BirdLife International. (2004). Birds in Europe: Population estimates, trends and con-servation status. Cambridge: BirdLife International Conservation Series.

Boatman, N. D., Brickle, N. W., Hart, J. D., Milsom, T. P., Morris, A. J., Murray, A. W.A., et al. (2004). Evidence for the indirect effects of pesticides on farmland birds.Ibis, 46(Suppl. 2), 131–143.

Boyce, M. S. (2006). Scale for resource selection function. Diversity and Distributions,12, 269–276.

Boyce, M. S., & McDonald, L. L. (1999). Relating population to habitats using resourceselection functions. Tree, 14, 268–272.

Boyce, M. S., Vernier, P. R., Nielsen, S. E., & Schmiegelow, F. K. A. (2002). Evaluatingresource selection functions. Ecological Modelling, 157, 281–300.

Brambilla, M., & Ficetola, G. F. (2012). Species distribution models as a tool toestimate reproductive parameters: A case study with a passerine bird species.Journal of Animal Ecology, 81, 781–787.

Brambilla, M., Casale, F., Bergero, V., Crovetto, G. M., Falco, R., Negri, I., et al. (2009).GIS-models work well, but are not enough: Habitat preferences of Lanius collurioat multiple levels and conservation implications. Biological Conservation, 142,2033–2042.

Brambilla, M., Casale, F., Bergero, V., Bogliani, G., Crovetto, G. M., Falco, R., et al. (2010).Glorious past, uncertain present, bad future? Assessing effects of land-usechanges on habitat suitability for a threatened farmland bird species. BiologicalConservation, 143, 2770–2778.

Burnham, K. P., & Anderson, D. R. (2002). Model selection and multimodel inference:A practical information-theoretic approach (2nd ed.). NY: Springer-Verlag.

Castaldi, A., & Guerrieri, G. (1995). Distribuzione altitudinale del genere Lanius nelLazio. Avocetta, 19, 136.

Ceresa, F., Bogliani, G., Pedrini, P., & Brambilla, M. (2012). The importance of keymarginal habitat features for birds in farmland: An assessment of habitat pref-erences of Red-backed Shrike Lanius collurio in the Italian Alps. Bird Study, 59,327–334.

Chow, T. E., Gaines, K. F., Hodgson, M. E., & Wilson, M. D. (2005). Habitat and exposuremodelling for ecological risk assessment: A case study for the raccoon on theSavannah River Site. Ecological Modelling, 189, 151–167.

Cramp, S., & Perrins, C. M. (Eds.). (1993). The birds of the western palearctic (Vol. 7).Oxford: Oxford University Press.

Donald, P. F., Pisano, G., Rayment, M. D., & Pain, D. J. (2002). The Common Agricul-tural Policy, EU enlargement and the conservation of Europe’s farmland birds.Agriculture, Ecosystem and Environment, 89, 167–182.

Fawcett, T. (2006). An introduction to ROC analysis. Pattern Recognition Letters, 27,861–874.

Fowler, J., & Cohen, L. (2002). Statistica per ornitologi e naturalisti. Bologna: MuzzioEditore.

Fuller, R. J., Gregory, R. D., Gibbons, D. W., Marchant, J. H., Wilson, J. D., Baillie, S.

R., et al. (1995). Population declines and range contractions among lowlandfarmland birds in Britain. Conservation Biology, 9, 1425–1441.

Garamszegi, L. Z. (2011). Information-theoretic approaches to statistical analysis inbehavioural ecology: An introduction. Behavioral Ecology and Sociobiology, 65,1–11.

ture C

G

G

G

G

G

G

G

G

G

H

H

H

H

H

I

J

J

K

K

K

K

K

L

L

L

L

M

M

M

G. Chiatante et al. / Journal for Na

enghini, M., Gellini, S., & Gustin, M. (2006). Organic and integrated agriculture: Theeffects on bird communities in orchard farms in northern Italy. Biodiversity andConservation, 15, 3077–3094.

iralt, D., & Valera, F. (2007). Population trends and spatial synchrony in peripheralpopulations of the endangered Lesser grey shrike in response to environmentalchange. Biodiversity and Conservation, 16, 841–856.

iralt, D., Brotons, L., Valera, F., & Kristin, A. (2008). The role of natural habitats inagricultural systems for bird conservation: The case of the threatened LesserGrey Shrike. Biodiversity and Conservation, 17, 1997–2012.

oławski, A., & Meissner, W. (2008). The influence of territory characteristics andfood supply on the breeding performance of the Red-Backed Shrike (Lanius col-lurio) in an extensively farmed region of eastern Poland. Ecological Research, 23,347–353.

ray, T. N. E., Phan, C., & Long, B. (2010). Modelling species distribution at multiplespatial scales: Gibbon habitat preferences in a fragmented landscape. AnimalConservation, 13, 324–332.

uerrieri, G., & Castaldi, A. (1999). Status e distribuzione del genere Lanius nel Lazio(Italia centrale). Rivista italiana di Ornitologia, 69, 63–74.

uerrieri, G., & Castaldi, A. (2000). Selezione di habitat e riproduzione dell’averlacapirossa, Lanius senator, nel Lazio – Italia Centrale. Avocetta, 24, 85–93.

uerrieri, G., & Castaldi, A. (2010). Selezione di habitat e riproduzione dell’averlacenerina, Lanius minor, in un’area costiera del Lazio (Monti della Tolfa). Rivistaitaliana di Ornitologia, 80, 35–56.

uerrieri, G., Pietrelli, L., & Biondi, M. (1995). Status and reproductive habitat selec-tion of three species of shrikes, Lanius collurio, L. senator and L. minor, in aMediterranean area. Proceedings of Western Foundation of Vertebrate Zoology, 6,167–171.

arris, T., & Franklin, K. (2000). Shrikes and Bush-shrikes: Including wood-shrikes,helmet shrikes, flycatcher-shrikes, philentomas, batises and wattle-eyes. London:Cristopher Helm, A&C Black.

ernandez, A. (1994). Seleccion de habitat en tres especies simpatricas dealcaudones (real, Lanius excubitor L., dorsirrojo, Lanius collurio L. y comun, Laniussenator L.): Segregacion interespecifica. Ecologia, 8, 395–413.

ernandez, A. (1997). In W. J. M. Hagemeijer, & M. J. Blair (Eds.), The EBCC Atlasof European Breeding Birds: Their distribution and abundance. London: T. & A.D.Poyser.

insley, S. A., & Bellamy, P. E. (2000). The influence of hedge structure, managementand landscape context on the value of hedgerows to birds: A review. Journal ofEnvironmental Management, 60, 33–49.

obbs, R. (1997). Future landscapes and the future landscape ecology. Landscape andUrban Planning, 37, 1–9.

senmann, P., & Debout, G. (2000). Vineyards harbour a relict population of LesserGrey Shrike (Lanius minor) in Mediterranean France. Journal of Ornithology, 141,435–440.

aberg, C., & Guisan, A. (2001). Modelling the distribution of bats in relation to land-scape structure in a temperate mountain environment. Journal of Applied Ecology,38, 1169–1181.

edrzejewski, W., Jedrzejewska, B., Zawadzka, B., Borowik, T., Nowak, S., & Mysla-jek, R. W. (2008). Habitat suitability model for Polish wolves base on long-termnational census. Animal Conservation, 11, 377–390.

arlsson, S. (2002). Analyses on prey composition of overwintering Great GreyShrike Lanius excubitor in southern Finland. Ornis Fennica, 79, 181–189.

arlsson, S. (2004). Season-dependent diet composition and habitat use of red-backed shrikes Lanius collurio in SW Finland. Ornis Fennica, 81, 97–108.

eynan, O., & Yosef, R. (2010). Temporal changes and sexual differences of impalingbehavior in Southern Grey Shrike (Lanius meridionalis). Behavioural Processes, 85,47–51.

ristin, A., & Lefranc, N. (1997). In W. J. M. Hagemeijer, & M. J. Blair (Eds.), The EBCCAtlas of European Breeding Birds: Their distribution and abundance.. London: T. &A.D. Poyser.

uczynski, L., Antczak, M., Czechowski, P., Grzybek, J., Jerzak, L., Zabłocki, P., et al.(2010). A large scale survey of the great grey shrike Lanius excubitor in Poland:Breeding densities, habitat use and population trends. Annales Zoologici Fennici,47, 67–78.

ee, J. T., & Thompson, S. (2005). Targeting sites for habitat creation: An investigationinto alternative scenarios. Landscape and Urban Planning, 71, 17–28.

ee, J. T., Bailey, N., & Thompson, S. (2002). Using Geographical Information Systemsto identify and target sites for creation and restoration of native woodlands: Acase study of the Chiltern Hills, UK. Journal of Environmental Management, 64,25–34.

efranc, N. (1995). Decline and current status of the Lesser Grey Shrike (Lanius minor)in Western Europe. In R. Yosef, & F. E. Lohrer (Eds.), Shrikes (Lanidae) of the world:Biology and conservation. Proc. of the 1st international shrikes symposium (6) (pp.167–171). Western Foundation of Vertebrate Zoology.

efranc, N., & Worfolk, T. (1997). Shrikes. A guide to the shrikes of the world. EastSussex: Pica Press.

abry, C. M., Brudvig, L. A., & Atwell, R. C. (2010). The confluence of landscape contextand site-level management in determining Midwestern savanna and woodlandbreeding bird communities. Forest Ecology and Management, 260, 42–51.

astronardi, D., Di Sarra, G., & Ricchi, I. (1996). La distribuzione dei Lanidi in Cam-pania in relazione all’antropizzazione del territorio. In ASOIM (Ed.), L’avifauna

degli ecosistemi di origine antropica: zone umide artificiali, coltivi, aree urbane (pp.83–85). Napoli: Electa.

cAlpine, C. A., Rhodes, J. R., Bowen, M. E., Lunney, D., Callaghan, J. C., Mitchell, D.L., et al. (2008). Can multiscale models of species’ distribution be generalized

onservation 22 (2014) 103–112 111

from region to region? A case study of the koala. Journal of Applied Ecology, 45,558–567.

Merli, E., & Meriggi, A. (2006). Using harvest data to predict habitat-populationrelationship of the wild boar Sus scrofa in Northern Italy. Acta Theriologica, 51,383–394.

Moga, C. I., Hartel, T., Öllerer, K., & Szapanyos, Á. (2010). Habitat use by the endan-gered Lesser Grey Shrike Lanius minor in Central Romania. Belgian Journal ofZoology, 140, 223–226.

Moreira, F., Beja, P., Morgado, R., Reino, L., Gordinho, L., Delgado, A., et al. (2005).Effects of field management and landscape context on grassland winter-ing birds in Southern Portugal. Agriculture, Ecosystems and Environment, 109,59–74.

Moskát, C., & Fuisz, T. I. (2002). Habitat segregation among the woodchat shrike,Lanius senator, the red-backed shrike, Lanius collurio, and the masked shrike,Lanius nubicus, in NE Greece. Folia Zoologica, 51, 103–111.

Newton, I. (2004). The recent declines of farmland bird populations in Britain: Anappraisal of causal factors and conservation actions. Ibis, 146, 579–600.

Numa, C., Verdú, J. R., Sánchez, A., & Galante, E. (2009). Effect of landscape structureon the spatial distribution of Mediterranean dung beetle diversity. Diversity andDistributions, 15, 489–501.

Olivier, F., & Wotherspoon, S. J. (2005). GIS-based application of resource selec-tion functions to the prediction of snow petrel distribution and abundance inEast Antarctica: Comparing modelsat multiple scales. Ecological Modelling, 189,105–129.

Olsson, U., Alström, P., Svensson, L., Aliabadian, M., & Sundberg, P. (2010). The Laniusexcubitor (Aves, Passeriformes) conundrum – Taxonomic dilemma when molec-ular and non-molecular data tell different stories. Molecular Phylogenetics andEvolution, 55, 347–357.

Padilla, D. P., Gonzáles-Castro, A., Nieves, C., & Nogales, M. (2009). Trophicecology of the Southern Grey Shrike (Lanius meridionalis) in insular environ-ments: The influence of altitude and seasonality. Journal of Ornithology, 150,557–568.

Parody, J. M., & Milne, B. T. (2004). Implications of rescaling rules for multi-scaledhabitat models. Landscape Ecology, 19, 691–701.

Pearce, J., & Ferrier, S. (2000). Evaluating the predictive performance of habitatmodels developed using logistic regression. Ecological Modelling, 133, 225–245.

Rayner, M. J., Clout, M. N., Stamp, R. K., Imber, M. J., Brunton, D. H., & Hauber, M. E.(2007). Predictive habitat modelling for the population census of a burrowingseabird: A study of the endangered Cook’s petrel. Biological Conservation, 138,235–247.

Richards, S. A., Whittingham, M. J., & Stephens, P. A. (2011). Model selection andmodel averaging in behavioural ecology: The utility of the IT-AIC framework.Behavioral Ecology and Sociobiology, 65, 77–89.

Rippa, D., Maselli, V., Soppelsa, O., & Fulgione, D. (2011). The impact of agro-pastoralabandonment on the Rock Partridge Alectoris graeca in the Apennines. Ibis, 153,721–734.

Smith, C. S., Howes, A. L., Price, B., & McAlpine, C. A. (2007). Using a Bayesian beliefnetwork to predict suitable habitat of an endangered mammal – The Julia Creekdunnart (Sminthopsis douglasi). Biological Conservation, 139, 333–347.

Söderström, B., & Karlsson, H. (2011). Increase reproductive performance of Red-Backed Shrikes Lanius collurio in forest clear-cuts. Journal of Ornithology, 152,313–318.

Suarez-Seone, S., Orborne, P. E., & Baudry, P. E. (2002). Responses of birds of differentbiogeographic origins and habitat requirements to agricultural land abandon-ment in northern Spain. Biological Conservation, 105, 333–344.

Suzuki, N., Olson, D. H., & Reilly, E. C. (2008). Developing landscape habitat mod-els for rare amphibians with small geographic ranges: A case study of SiskiyouMountains salamanders in the western USA. Biodiversity and Conservation, 17,2197–2218.

Thompson, S., Larcom, A., & Lee, J. T. (1999). Restoring and enhancing rare and threat-ened habitats under agri-environment agreements: A case study of the ChilternHills Area of Outstanding Natural Beauty, UK. Land Use Policy, 16, 93–105.

Traill, L. W., & Bigalke, R. C. (2006). A presence-only habitat suitability model forlarge grazing African ungulates and its utility for wildlife management. AfricanJournal of Ecology, 45, 347–354.

Tsaparis, D., Katsanevakis, S., Ntolka, E., & Legakis, A. (2009). Estimating dung decayrates of roe deer (Capreolus capreolus) in different habitat types of a Mediter-ranean ecosystem: An information theory approach. European Journal of WildlifeResearch, 55, 167–172.

Webster, S., & Felton, M. (1993). Targeting for nature conservation in agriculturalpolicy. Land Use Policy, 10, 67–82.

Westphal, M. I., Field, S. A., Tyre, A. J., Paton, D., & Possingham, H. P. (2003). Effectsof landscape pattern on bird species distribution in the Mt. Lofty Ranges, SouthAustralia. Landscape Ecology, 18, 413–426.

Whittingham, M., Stephens, P. A., Bradbury, R. B., & Freckleton, R. P. (2006). Whydo we still use stepwise modelling in ecology and behaviour? Journal of AnimalEcology, 75, 1182–1189.

Wilson, J. D., Morris, A. J., Arroyo, B. E., & Clark, S. C. (1999). A review of the abun-dance and diversity of invertebrate and plant foods of granivorous birds innorthern Europe in relation to agricultural change. Agriculture, Ecosystem andEnvironment, 75, 13–30.

Yosef, R. (1993). Influence of observation posts on territory size in Northern Shrikes.The Wilson Bulletin, 105, 180–183.

Yosef, R. (1994). Evaluation of the global decline in the true shrikes (family Laniidae).The Auk, 111, 228–233.

1 ture C

Y

Y

12 G. Chiatante et al. / Journal for Na

osef, R., & Grubb, T. C. (1994). Resource dependence and territory size in LoggerheadShrikes (Lanius ludovicianus). The Auk, 111, 465–469.

osef, R., & Lohrer, F. E. (Eds.). (1995). Shrikes (Laniidae) of the World: Biology andConservation. Proceedings of the Western Foundation of Vertebrate Zoology, 6,1–343.

onservation 22 (2014) 103–112

Yosef, R., & Lohrer, F. E. (Eds.). (1998). Shrikes of the World II: Conservationand implementation. International Birdwatching Center Technical Publications, 7,1–119.

Yosef, R., Lohrer, F. E., Van Nieuwenhuyse, D., & Busse, P. (Eds.). (2000). Proceedingsof the 3rd International Shrike Symposium. Ring, 22, 1–217.