Habitat preferences and causes of population decline for Barn Owls Tyto alba: A multi-scale approach

Abstract

Aims: Habitat preferences of Barn Owls was studied in two areas of Spain undergoing large-scale habitat alteration. Location: Alicante (dry cultivations) and Valencia (irrigated cultivations) in eastern Spain. Methods: Habitat composition around occupied and unoccupied territories in dry cultures and irrigated cultures was compared (n = 71, 1989-2000). This study also described differences in habitat composition between occupied and deserted territories after major habitat alterations started in 1996. Generalized Linear Models were used to examine patterns of habitat preference at three different spatial scales: nest site, home range and landscape. Results: The study population declined by 69% in both study areas. At the nest site scale, Barn Owls preferred undisturbed areas with high availability of cavities, mainly in man-made structures. At the home range and landscape scales, Barn Owls occupied undisturbed areas with a high availability of cavities and high percentages of edges and ditches. Territory desertion was prompted by the modernisation or disappearance of man-made structures, depletion of edges and ditches, expansion of the road network and persecution. Accordingly, the spatial distribution of territories in irrigated cultures changed from uniform to random after habitat alteration. The availability of cavities alone does not account for all of the explained deviance, i.e., Barn Owls occupy structurally complex landscapes. Conclusions: Compensation measures for habitat loss such as nest-box programs, usually proposed within the framework of environmental impact assessment, are discouraged unless habitat restoration and effective control of persecution are promoted first.

Full text

Ardeola 51(2), 2004, 303-317
HABITAT PREFERENCES AND CAUSES OF POPULATION
DECLINE FOR BARN OWLS TYTO ALBA:
A MULTI-SCALE APPROACH
Jose Antonio MARTÍNEZ*1& Íñigo ZUBEROGOITIA**
SUMMARY.—Habitat preferences and causes of population decline for Barn owls Tyto alba: a multi-sca-
le approach.
Aims: Habitat preferences of Barn Owls was studied in two areas of Spain undergoing large-scale habitat al-
teration.
Location: Alicante (dry cultivations) and Valencia (irrigated cultivations) in eastern Spain.
Methods: Habitat composition around occupied and unoccupied territories in dry cultures and irrigated cul-
tures was compared (n = 71, 1989-2000). This study also described differences in habitat composition between
occupied and deserted territories after major habitat alterations started in 1996. Generalized Linear Models
were used to examine patterns of habitat preference at three different spatial scales: nest site, home range and
landscape.
Results: The study population declined by 69% in both study areas. At the nest site scale, Barn Owls prefe-
rred undisturbed areas with high availability of cavities, mainly in man-made structures. At the home range
and landscape scales, Barn Owls occupied undisturbed areas with a high availability of cavities and high per-
centages of edges and ditches. Territory desertion was prompted by the modernisation or disappearance of
man-made structures, depletion of edges and ditches, expansion of the road network and persecution. Ac-
cordingly, the spatial distribution of territories in irrigated cultures changed from uniform to random after ha-
bitat alteration. The availability of cavities alone does not account for all of the explained deviance, i.e., Barn
Owls occupy structurally complex landscapes.
Conclusions: Compensation measures for habitat loss such as nest-box programs, usually proposed within the
framework of environmental impact assessment, are discouraged unless habitat restoration and effective
control of persecution are promoted first.
Key words: Agriculture abandonment, environmental impact assessment, strategic environmental as-
sessment, persecution, nest-boxes, Barn Owl, Tyto alba.
RESUMEN.—Preferencias de hábitat y causas de declive poblacional para la Lechuza Común Tyto alba:
una aproximación a varias escalas.
Objetivos: Estudiar las preferencias de hábitat de la Lechuza Común en dos zonas de sometidas a grandes
transformaciones del suelo.
Localidad: Alicante (cultivos de secano) y Valencia (cultivos de regadío) en el este de España.
Métodos: Se comparó la composición del hábitat alrededor de territorios ocupados y no ocupados en cultivos
de secano y de regadío (n = 71, 1989-2000). También se describió las diferencias en la composición del há-
bitat entre los territorios ocupados y los abandonados tras el pico en la tasa de alteración del paisaje de 1996.
Hemos utilizado Modelos Generales Linearizados para examinar los patrones de preferencias a tres escalas es-
paciales: lugar de nidificación, área de campeo y paisaje.
Resultados: La población total de estudio decreció en un 69%. A la primera escala, las Lechuzas prefirieron
áreas poco alteradas con gran disponibilidad de cavidades, principalmente en edificios. A la segunda y tercera
escalas, prefirieron áreas poco alteradas con gran disponibilidad de cavidades, altos porcentajes de ecotonos
y acequias con cañaverales desarrollados. El abandono de territorios está favorecido por la modernización o
la desaparición de edificios con cavidades, acequias y ecotonos así como por la expansión de la red de ca-
rreteras y la caza ilegal. En virtud de estos factores, la distribución espacial de los territorios en el regadío ha
cambiado de uniforme a aleatoria.
Conclusiones: La restauración del hábitat y el control efectivo de la caza ilegal son las medidas prioritarias a
tomar para la conservación de esta especie en el área de estudio, desaconsejándose los programas de coloca-
ción de cajas anidaderas que no incluyan mejora o preservación de hábitat. La colocación de dichas cajas, su-
*C/ Juan de la Cierva 43, El Campello, E-03560, Alicante, Spain. E-mail: qvcocotiers@hotmail.com
** Lab. Zoología, Dpto. Zoología. Facultad de Ciencias. Universidad del País Vasco. Aptdo 644.
E-48080, Bilbao, Spain. E-mail: inigo.zuberogoitia@wanadoo.es
1 Corresponding author: E-mail address: qvcocotiers@hotmail.com

INTRODUCTION
The landscape in the Mediterranean basin
has been subjected to a relentless process of
alteration for centuries (Pain & Pienkowski,
1997). The rate of habitat alteration has steeply
increased over the last fifty years so that little
remains of the original or agricultural landsca-
pe, especially in coastal areas (Agencia del Me-
dio Ambiente, 1997). In particular, the rate of
habitat loss to large housing developments and
extended road networks in the east coast of
Spain peaked in 1996 (Simarro, 2002). Habitat
loss occurred mostly at the expense of dry and
irrigated cultures, which are otherwise being
replaced by large fruit/vegetable greenhouses
favoured by the reform of the Common Agri-
cultural Policy (CAP) (Pain & Pienkowski,
1997). Contrary to other areas across Europe
(Penteriani et al., 2002), abandonment of agri-
cultural lands has not lead to a substantial ex-
pansion of the surface of forest mainly because
of the rapid changes on the suitability for buil-
ding of the abandoned lands and subsequent
urbanisation (Simarro, 2002).
Interestingly, human-maintained semi-natu-
ral systems have the potential to host a higher
richness and diversity of species than extensive
cultures or large forests as long as traditional
agricultural practices are put into practice (Rico
1997; Sánchez-Zapata & Calvo, 1999; Zubero-
goitia, 2000, Zuberogoitia, 2002). Indeed, the
bulk of the raptor and owl population in Ali-
cante (east of Spain) inhabits the agro-pastoral
complex (Rico et al., 2001; Martínez et al.,
2003). Paradoxically, birds of prey are protec-
ted but the conservation status of their preferred
nesting and foraging habitats is low because
they are located out of the network of Natural
Parks. Therefore, the question is raised as to
how the shrinking availability of the agro-pas-
toral complex would affect the probability of
having occupied territories of listed species.
An approach combining long temporal data
series with multi-scale descriptions of the habi-
tat preferences of the target species can produ-
ce meaningful results with regard to the res-
ponse of animals to habitat loss (Penteriani et
al., 2001; Marchesi et al., 2002; Martínez et
al., 2003). The multi-scale approach to the
study of habitat preferences is mostly based on
the conceptual framework proposed by Johnson
(1980), whose basic assumption is that animals
are capable of making decisions regarding re-
sources at consecutively smaller scales. Accor-
dingly, general habitat selection can be consi-
dered as a hierarchical process regarding, for
example, a suitable patch for breeding at a
small scale and apt areas for foraging at a bro-
ader scale (Martínez et al., 2003). The multi-
scale approach may be especially appropriate to
identify key factors involved in habitat prefe-
rence of owls because they have large home
ranges consisting of different patches for bree-
ding and foraging (Mikkola, 1983).
The Barn Owl Tyto alba is one of the ar-
chetypal inhabitants of the agri-pastoral mo-
saics of Spain (Zuberogoitia, 2002). In com-
mon with other countries across Europe, a
negative trend in population size has been re-
ported in Spain, the ultimate causes of which
are still poorly documented (Tucker & Heath,
1994; Martínez & Zuberogoitia, 2003a). Ho-
wever, several studies have pointed out the
owl’s sensitivity to small-scale habitat changes
(Taylor, 1994; Ramsden, 1998; Zuberogoitia,
2002), which suggests the importance of moni-
toring how changes in the traditional land uses
affect the abundance of Barn Owls. Further-
more, recently released fauna catalogues such
as the Decreto 32/2004 DOGV, 27 February
(Comunidad Valenciana, East of Spain) do not
list the Barn Owl as protected, an assessment
based on the subjective criteria of the board
because no large-scale survey has ever been
performed in the area. On these grounds, we
aim to: (1) catalogue relevant environmental
characteristics affecting habitat preferences of
Barn Owls at three different spatial scales (nest
areas, home ranges and landscape) and (2) as-
ses how habitat loss may be influencing the
sustainability of a Barn Owl population.
304 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317
gerida tanto en los estudios de impacto ambiental como dentro de evaluaciones estratégicas ambientales de
planes de desarrollo en la zona de estudio, se considera una medida compensatoria errónea, pues los modelos
muestran que las Lechuzas Comunes ocupan paisajes estructuralmente más complejos que los determinados
por la simple disponibilidad de cavidades para anidar.
Palabras clave: Abandono de la agricultura, estudios de impacto ambiental, evaluaciones estratégicas am-
bientales, caza ilegal, cajas anidaderas, Lechuza Común, Tyto alba.

MATERIAL AND METHODS
Study area
The study was carried out between 1989 and
2000 in two study areas: (1) dry cultures, in the
province of Alicante, and (2) irrigated cultures, in
the province of Valencia (Fig. 1). In dry cultures
the climate varies from semi-arid meso-medite-
rranean to sub-humid Mediterranean. Average
annual rainfall is about 400 mm, and annual ave-
rage temperature is about 19°C. The landscapes
is dominated by dry cultivated fields, mainly Ca-
rob Ceratonia siliqua, Almond Prunus amygda-
lus, Olive trees Olea europaea, vineyards, bar-
ley, sunflowers, wheat, scrubland and pine
forests (Pinus halepensis and Pinus pinea) and
medium-sized cities. Altitude varies between 0
and 1500 m a.s.l. In the irrigated cultures area,
the main land uses are citrus groves interspersed
with vegetable smallholdings. Average annual
rainfall is about 500 mm, and annual average
temperature is about 18°C. Altitude varies bet-
ween 0 and 300 m a.s.l. See Martínez (1998)
and Martínez & López (1999) for further details.
Survey methods
In order to locate Barn Owls, nests, moulted
feathers, droppings and food remains were loo-
ked for in man-made structures, ephemeral ri-
vers and trees (Shawyer, 1987; Taylor, 1994;
Ramsden, 1998; Martínez & López, 1999; Zu-
berogoitia & Campos, 1998; Toms et al.,
2001). As a complementary method, recordings
of the male’s main territorial call were broad-
cast (Zuberogoitia & Campos, 1998). All te-
rritories (and random sites, see below) were vi-
sited four times each year to ascertain their
occupancy according to the presence or absen-
ce of nests, pellets, droppings, moulted feat-
hers, sightings or responses to playback (Tay-
lor, 1994).
Selection of scales
Following the rationale of Johnson (1980),
Martínez et al. (2003) and Martínez & Zubero-
goitia (2004), three different spatial scales were
used to study habitat preferences of Barn Owls:
a) Nest site scale. Core areas during the bre-
eding season can be encircled by a circle 1 km
radius from the nests (Taylor, 1994) including
nests, diurnal or nocturnal roosts and hunting
grounds (Taylor, 1994). Hence, an average core
area was assumed to be a circular 317 ha plot
around the nests (1 km radius) (Taylor, 1994).
b) Home range scale. Barn Owl home ranges
vary in length from 2 to 5 km (Taylor, 1994).
From these studies, an average territory was
HABITAT LOSS AND BARN OWL POPULATION DECLINE 305
Ardeola 51(2), 2004, 303-317
FIG.1.—Percentage of deviance explained at three spatial scales by the Generalized Linear Models for the
probability of presence of Barn Owls comparing occupied vs. non occupied territories in dry areas.
[Porcentajes de devianza explicados a tres escalas espaciales por los Modelos Generales Linearizados
para la probabilidad de presencia de Lechuzas Comunes comparando territorios ocupados y no ocupados en
áreas de cultivos de secano.]
Unexplained
bosque]

assumed to be a circular area of 2826 ha around
centres of activity (3 km radius).
c) Landscape scale. Since landscape ecology
addresses the relationships between animal dis-
tribution and mosaics of ecosystems (Forman
& Gordon, 1986) tests were made for a possi-
ble response of Barn Owls to habitat composi-
tion at a broad landscape level. Thus, an area of
100 km2around nests (5.6 km radius) was cho-
sen because in Alicante and Valencia it is likely
to find substantial changes in landscape com-
position within this radius (Agencia del Medi
Ambient, 1997). Furthermore, other studies
have reported responses of birds of prey and
owls to this landscape scale in the study areas
(Rico et al., 2001; Martínez et al., 2003; Martí-
nez & Zuberogoitia, 2004).
Selection of variables
From 1:2000 aerial photographs and 1:2000
maps, an a priori set of environmental varia-
bles related to topography were selected (Burn-
ham & Anderson, 2002) (1), human disturban-
ce (3), land use (6), edges (3), linear structures
(1) and nesting or roosting requirements (1)
(see Appendix 1). Two studies have singled
out illegal hunting as one of the most important
proximate causes of mortality of raptors and
owls in the study areas (Martínez et al., 1996;
Martínez et al., 2001). From these studies and
from unpublished data maps were developed
which depicted the location of every raptor or
owl casualty attributed to persecution. Then,
the variable «persecution» was constructed as
the presence (1) or absence (0) of records of
persecution in the three circular sampling areas
around centres of activity. The aim of the study
was to test if territory abandonment is more li-
kely to occur in areas where persecution takes
place. During the course of routine visits the
number and location of potential roosting or
breeding sites was plotted on aerial photo-
graphs or detailed maps.
Centres of activity are defined as nests or as
the most frequently used roost judging from
the amount of signs of activity (Ramsden,
306 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317
APPENDIX 1
Variables used to characterize centres of activity at the nest, home range and landscape scales.
[Variables utilizadas para caracterizar el centro de actividad a escala del nido, territorio y paisaje.]
Physiography
RELIEF, number of 100 m contours cut by four lines starting from the centre of the area in directions N, S, E
and W.
Human disturbance
PAVED ROADS, metres around the centre of activity.
HOUSING DEVELOPMENTS, %
PERSECUTION, presence (1) or absence (0) of records of persecution in the circular sampling area.
Land use (%)
CAROB, OLIVE, ALMOND CULTURES (TREE CULTURES)
CEREAL CULTIVATIONS
CITRUS GROVES
VEGETABLE SMALLHOLDINGS
EPHEMERAL RIVERS
FOREST
Edges (m)
CERAL-FOREST
CITRUS GROVES-SET ASIDE CULTIVATIONS
CITRUS GROVES-VEGETABLE SMALLHOLDINGS
Linear structures (m)
DITCHES
Nesting/roosting requirements
NUMBER OF BUILDINGS WITH CAVITIES

1998). Habitat composition around 71 active
Barn Owl territories (31 in dry cultures; 40 in
irrigated cultures) was compared with 82 sites
chosen randomly (41 in dry cultures, 41 in irri-
gated cultures). Random sites were located at a
minimum distance of 5 km from each other or
from occupied sites. All non-occupied random
sites remained empty throughout the study pe-
riod. These analyses were referred to the year
2000, using cartography for that year.
Major habitat alteration started in 1996 in
both study areas involved: (1) renovation and
corresponding inaccessibility of man-made
structures (e.g., screening of belfries, obstruc-
tion of niches of cemeteries), (2) abandonment
of traditional agricultural practices and corres-
ponding decay of man-made structures used
for breeding (e.g., old warehouses, dovecots),
misuse or decay of irrigation ditches and di-
sappearance of edges and (3) construction of
large housing developments and expansion of
the road network. Forty-nine territories became
vacant between 1996 and 2000 (23 in dry cul-
tures; 26 in irrigated cultures). Thus, habitat
composition around abandoned centres of acti-
vity was compared with habitat composition
around these occupied before (1995) and after
(2000) large-scale habitat loss started. Fittingly,
the cartography used corresponds to the same
years.
Analytical procedures
Generalised Linear Models (GLMs, McCu-
llagh & Nelder, 1989) were used to obtain the
mathematical descriptions of habitat preferen-
ces. GLMs allow for the use of appropriate
error formulations from the exponential family
distributions, hence avoiding some of the limi-
tations of the conventional regression models.
Generalised Linear Models consists of a linear
predictor, an error function and a link function.
The linear predictor (LP) is defined as: LP =
a+ bx1 + cx2 + … where ais the intercept, b,
c… are the parameter estimates to be obtained
from the observed data, and x1, x2,… are the
explanatory variables. The error and link func-
tions depend on the nature of the data. The pre-
sence of owls follows a binomial distribution
(binary response variable: presence = 1, absen-
ce = 0). Therefore, a logit link was used (Bus-
tamante, 1997; Martínez et al, 2003a). Six se-
parate GLMs were conducted for the environ-
mental description of data. Each variable was
tested for significance in turn, and only those
variables that contributed to the largest signifi-
cant change in deviance were retained. Only
variables significant at the 1% level were in-
cluded in the models (Nicholls, 1989). The fi-
nal models were selected by likelihood ratio
tests for type I analysis (SAS Institute, 1996).
Recommendations have been recently made
that ecologists reduce their reliance on predic-
tion success as a performance measure in pre-
sence-absence modelling when independent
sets of data are not available to validate habitat
models (Manel et al., 2001). In these circums-
tances, Kappa statistics (Titus et al., 1984) are
recommended to test whether model discrimi-
nation significantly improves chance classifi-
cations. Kappa is a robust statistic that is spe-
cially adequate as a measure of the proportion
of all possible cases of presence or absence that
are predicted by a model after accounting for
chance effects, offering a meaningful numerical
variable for intercomparison between models
or between different statistical algorithms (Ma-
nel et al., 2001). The output variables (i.e., the
predicted values) in each case have a value bet-
ween 0 and 1, and presence for all samples was
accepted at a threshold probability of 0.5. For
Kappa, values of prediction success of 0-40%
are considered to indicate slight to fair model
performance, values of 40-60% moderate, 60-
80% substantial and 80-100% almost perfect
(Landis & Koch, 1977; Manel et al., 2001).
The breeding ecology of the Barn Owl in
irrigated cultures has been described in full el-
sewhere (Martínez, 1998; Martínez & López,
1999). Here, tests were made for a relationship
between the total length edges and ditches and
productivity (number of young per successful
pair), because Taylor (1994) has pointed out
the importance of edges in explaining the bre-
eding outcome for Barn Owls. Regularity of
dispersion of centres of activity in irrigated
cultures was assessed by means of the G-sta-
tistic (Brown, 1975), calculated as the ratio of
the geometric mean to arithmetic mean of the
squared nearest neighbour distances. The G-
statistic ranges from 0 to 1, with values over
0.65 indicating regularity of spacing. It is pos-
sible that 1-2 territories went unnoticed in dry
cultures, thus precluding calculation of the G-
statistic.
HABITAT LOSS AND BARN OWL POPULATION DECLINE 307
Ardeola 51(2), 2004, 303-317

RESULTS
Overall, a 69% population decline was found
in both study areas. Two sets of models of ha-
bitat preferences were produced. Firstly, occu-
pied versus non occupied territories in dry and
irrigated cultures were compared. The first set
of models showed that occupied territories were
characterized by a greater availability of cavi-
ties, linear structures and edges and a less de-
veloped road network than in non occupied te-
rritories (Tables 1 and 2). Secondly, occupied
versus deserted territories in dry and irrigated
cultures were compared. The second set of mo-
dels showed that deserted territories were more
likely to be found in areas were the road net-
work has been extended, housing developments
have been constructed, roosting or nesting pla-
ces become scarce, edges between traditional
land uses have been depleted and persecution
of birds of prey and owls takes place (Tables 3
and 4). The values of prediction success are
acceptable, ranging between 72-97% (Tables
1 to 4). The percentage of deviance explained
at each spatial scale by the GLMs is shown in
the figures 2 to 5. It was a combination of va-
riables that accounted for the explained de-
viance of the models, rather than a predomi-
nant variable (Fig. 2 to 5).
The spacing of centres of activity in irrigated
cultures was uniform before 1996 (G = 0.87).
However, the value of the G-statistic (0.25)
shows random dispersion of centres of activity
after 1996. A significant relationship was found
between the total length of edges and ditches
and the productivity before 1996 at the home
range scale (r2= 0.88, df = 34 , P = 0.001).
DISCUSSION
Barn Owls in dry cultures breed and roost
mainly in rural houses, dovecots and abando-
ned rabbit (Oryctolagus cuniculus) furrows in
the banks of ephemeral rivers, whereas in irri-
gated cultures they use niches in cemeteries,
belfries, rural houses and old warehouses (Mar-
tínez, 1998; Martínez & López, 1999). Unlike
other European populations (Shawyer, 1987;
Taylor, 1994), trees are very seldom used for
breeding, although mature carobs, olives, dense
pines and cypresses are frequently used as diur-
nal or nocturnal roosts. This probably reflects
the increasing scarcity of old trees resulting
from the conversion of cultures into urbanised
areas. Moreover, the abandonment of traditio-
nal agricultural practices has prompted the di-
sappearance of old trees through over-matu-
rity, disease or decay (Simarro, 2002). In
irrigated cultures, orange and lemon trees are
trimmed so that no cavities are available. Po-
llard trees, which occurred in the margins of
citric cultivations at the beginning of the study
period, have been completely depleted. Occu-
pied territories occurred in less disturbed areas
(as measured by the length of roads) than ran-
dom sites in both study areas.
Variables reflecting the feeding habits of the
owls enter the models at the home range scale
(Tables 1 and 2). This probably reflects the in-
creased density of prey in edges between habi-
tats (McCollin, 1998). Indeed, Rico (1997)
recommends a network of cereal patches in-
terspersed in Mediterranean forest and scru-
bland in order to increase the length of frontiers
between habitats and therefore small mammal
density in Alicante. Herbaceous edges in both
study areas are mowed or burnt periodically to
prevent overgrowth that would cause labour
difficulties, obstruct ditches or else «attract
rats» (Simarro, 2002), which probably enhan-
ces detection of prey by owls (Ille & Grinschgl,
2000). Barn Owls also prefer areas with high
availability of ditches. This traditional irrigation
system is frequently used by rats (Rattus spp),
which are the staple prey of Barn Owls in the
study areas (Martínez & López, 1999). Howe-
ver abundant ditches are in dry cultures, they
are mostly in a state of decay because, firstly,
of the abandonment of agricultural practices
and, secondly, because drip irrigation slowly
takes over as a more efficient irrigation sys-
tem, for instance in the fruit/vegetable green-
houses (Simarro, 2002). Furthermore, old acti-
ve ditches usually become overgrown with
small reed beds where small mammals are
abundant. All the same, reeds have been al-
most completely depleted (Simarro, 2002).
Again, occupied territories occurred in undis-
turbed areas and with a high availability of ca-
vities.
At the landscape level, the amount of man-
made structures with cavities entered the model
with positive value, and the increasing length of
paved roads emerges as a negative factor af-
fecting the probability of settlement of Barn
308 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317

HABITAT LOSS AND BARN OWL POPULATION DECLINE 309
Ardeola 51(2), 2004, 303-317
TABLE 1
(A) Generalized Linear Models for the probability of presence of Barn Owls (comparing occupied vs. non oc-
cupied territories) in dry areas. (B) Explanatory power of the models (percentage of territories correctly clas-
sified, percentage of improvement over a classification by chance and test of significance). Level of signifi-
cance: ***<0.001.
[(A) Modelos Generales Linearizados para la probabilidad de presencie de Lechuzas Comunes (comparan-
do territorios ocupados con no ocupados) en cultivos de secano (B) Poder explicativo de los modelos (por-
centaje de territorios clasificados correctamente, porcentaje de mejora respecto a una clasificación al azar
y prueba de significación. Nivel de significación: ***<0,001.]
(A) Nest scale Home range-scale Landscape scale
[A escala del nido] [A escala del territorio] [A escala del paisaje]
Factor bSE (b)
χ
2bSE (b)
χ
2bSE (b)
χ
2
Intercept 20.331 0.672 –14.7 1.004 –1.22 0.123
[Punto de corte]
Buildings with cavities 0.06 0.01 48.52*** 0.106 0.011 57.999*** 0.048 0.016 25.008***
[Edificios con cavidades]
Roads –0.043 0.001 27.532*** –0.03 0.014 38.891*** –0.06 0.002 21.562***
[Carreteras]
Cereal-forest edges 0.266 0.007 58.011*** 0.043 0.011 9.905***
[Ecotono cereal-bosque]
Ditches 0.15 0.005 50.987***
[Acequias]
Residual deviance 39.08 10.05 39.87
[Devianza residual]
Variance explained by 60.1 92.3 58.3
model [Varianza explicada
por el modelo]
(B) % of correct Classification better Kappa Test (Z)
classification than chance (%)
Nest scale
[A escala del nido]
Occupied territories 87 77 8.1***
[Territorios ocupados]
non occupied territories 88
[Territorios no ocupados]
Home range scale
[A escala del territorio]
Occupied territories 97 97 9.1***
non occupied territories 98
Landscape scale
[A escala del paisaje]
occupied territories 89 78 7.7***
non occupied territories 85

310 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317
TABLE 2
(A) Generalized Linear Models for the probability of presence of Barn Owls (comparing occupied vs. non oc-
cupied territories) in irrigated areas. (B) Explanatory power of the models (percentage of territories co-
rrectly classified, percentage of improvement over a classification by chance and test of significance). (Level
of significance ***< 0.001).
[(A) Modelos Generales Linearizados para la probabilidad de presencie de Lechuzas Comunes (comparan-
do territorios ocupados con no ocupados) en cultivos de regadío (B) Poder explicativo de los modelos
(porcentaje de territorios clasificados correctamente, porcentaje de mejora respecto a una clasificación al
azar y prueba de significación. Nivel de significación: ***<0,001.]
(A) Nest scale Home range-scale Landscape scale
[A escala del nido] [A escala del territorio] [A escala del paisaje]
Factor bSE (b)
χ
2bSE (b)
χ
2bSE (b)
χ
2
Intercept 16.109 0.073 18.47 0.802 –1.52 0.09
[Punto de corte]
Buildings with cavities 0.019 0.001 59.622*** 0.077 0.04 52.06*** 0.066 0.04 40.121***
[Edificios con cavidades]
Roads –0.044 0.002 42.321*** -0.043 0.002 66.176*** –0.09 0.001 38.099***
[Carreteras]
Smallholdings-citric edges 0.031 0.001 66.93*** 0.051 0.001 69.072*** 0.099 0.001 40.541***
[Ecotono minifundios-
cítricos]
Ditches 0.033 0.02 39.799***
[Acequias]
Residual deviance 52.1 24.1 35.3
[Devianza residual]
Variance explained by 49.5 79.1 65.2
model [Varianza explicada
por el modelo]
(B) % of correct Classification better Kappa Test (Z)
classification than chance (%)
Nest scale
[A escala del nido]
Occupied territories 77 80 8.01***
[Territorios ocupados]
non occupied territories 73
[Territorios no ocupados]
Home range scale
[A escala del territorio]
Occupied territories 93 92 10.22***
non occupied territories 89
Landscape scale
[A escala del paisaje]
occupied territories 83 72 6.36***
non occupied territories 71

HABITAT LOSS AND BARN OWL POPULATION DECLINE 311
Ardeola 51(2), 2004, 303-317
TABLE 3
(A) Generalized Linear Models for the probability of presence of Barn Owls (comparing occupied vs. deser-
ted territories) in dry areas. (B) Explanatory power of the models (percentage of territories correctly classified,
percentage of improvement over a classification by chance and test of significance). (Level of significance:
***< 0.001).
[(A) Modelos Generales Linearizados para la probabilidad de presencie de Lechuzas Comunes (comparan-
do territorios ocupados con abandonados) en cultivos de secano (B) Poder explicativo de los modelos (por-
centaje de territorios clasificados correctamente, porcentaje de mejora respecto a una clasificación al azar
y prueba de significación. Nivel de significación: ***<0,001]
(A) Nest scale Home range-scale Landscape scale
[A escala del nido] [A escala del territorio] [A escala del paisaje]
Factor bSE (b)
χ
2bSE (b)
χ
2bSE (b)
χ
2
Intercept 25.14 0.661 –9.58 0.157 2.42 1.15
[Punto de corte]
Buildings with cavities 0.092 0.033 64.63*** 0.072 0.003 41.013*** 0.088 0.004 31.32***
[Edificios con cavidades]
Roads –0.041 0.000 28.144*** –0.081 0.05 23.33*** –0.063 0.002 50.006***
[Carreteras]
Housing developments –0.063 0.000 49.555*** –0.079 0.004 49.02*** –0.052 0.001 66.446***
[Urbanizaciones]
Cereal-forest edges 0.098 0.002 62.871***
[Ecotono cereal-bosque]
Persecutio –0.069 0.023 22.505***
[Caza ilegal]
Residual deviance 11.4 10.7 26.7
[Devianza residual]
Variance explained by 85.3 89.9 73.4
model [Varianza explicada
por el modelo]
(B) % of correct Classification better Kappa Test (Z)
classification than chance (%)
Nest scale
[A escala del nido]
Occupied territories 86 89 6.78***
[Territorios ocupados]
non occupied territories 93
[Territorios no ocupados]
Home range scale
[A escala del territorio]
Occupied territories 90 94 9.04***
non occupied territories 96
Landscape scale
[A escala del paisaje]
occupied territories 86 73 8.44***
non occupied territories 83

312 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317
TABLE 4
(A) Generalized Linear Models for the probability of presence of Barn Owls (comparing occupied vs. deser-
ted territories) in irrigated areas. (B) Explanatory power of the models (percentage of territories correctly clas-
sified, percentage of improvement over a classification by chance and test of significance).(Level of signi-
ficance: ***< 0.001).
[(A) Modelos Generales Linearizados para la probabilidad de presencie de Lechuzas Comunes (comparan-
do territorios ocupados con abandonados) en cultivos de regadío (B) Poder explicativo de los modelos (por-
centaje de territorios clasificados correctamente, porcentaje de mejora respecto a una clasificación al azar
y prueba de significación. Nivel de significación: ***<0,001.]
(A) Nest scale Home range-scale Landscape scale
[A escala del nido] [A escala del territorio] [A escala del paisaje]
Factor bSE (b)
χ
2bSE (b)
χ
2bSE (b)
χ
2
Intercept –5.688 0.722 29.71 0.367 2.05 0.209
[Punto de corte]
Buildings with cavities –0.044 0.001 43.644*** –0.068 0.006 45.715*** –0.072 0.06 44.135***
[Edificios con cavidades]
Roads –0.087 0.001 30.691*** –0.047 0.003 42.719*** –0.099 0.009 46.336***
[Carreteras]
Housing developments –0.093 0.002 39.164*** –0.098 0.004 39.064*** –0.064 0.02 63.254***
[Urbanizaciones]
Smallholdings-citric edges 0.009 0.000 56.892*** 0.053 0.004 79.038***
[Ecotono minifundios-
cítricos]
Persecutio –0.093 0.027 29.111***
[Caza ilegal]
Residual deviance 19.9 6.2 21.9
[Devianza residual]
Variance explained by 80.2 94 78.2
model [Varianza explicada
por el modelo]
(B) % of correct Classification better Kappa Test (Z)
classification than chance (%)
Nest scale
[A escala del nido]
Occupied territories 86 89 6.51***
[Territorios ocupados]
non occupied territories 80
[Territorios no ocupados]
Home range scale
[A escala del territorio]
Occupied territories 98 98 8.02***
non occupied territories 97
Landscape scale
[A escala del paisaje]
occupied territories 77 77 6.25***
non occupied territories 79

HABITAT LOSS AND BARN OWL POPULATION DECLINE 313
Ardeola 51(2), 2004, 303-317
FIG.2.—Percentage of deviance explained at three spatial scales by the Generalized Linear Models for the
probability of presence of Barn Owls comparing occupied vs. non occupied territories in irrigated areas.
[Porcentajes de devianza explicados a tres escalas espaciales por los Modelos Generales Linearizados
para la probabilidad de presencia de Lechuzas Comunes comparando territorios ocupados y no ocupados en
áreas de cultivos de regadío.]
FIG. 3.—Percentage of deviance explained at three spatial scales by the Generalized Linear Models for the
probability of presence of Barn Owls comparing occupied vs. deserted territories in dry areas.
[Porcentajes de devianza explicados a tres escalas espaciales por los Modelos Generales Linearizados
para la probabilidad de presencia de Lechuzas Comunes comparando territorios ocupados y abandonados en
áreas de cultivos de secano]

Owls. Yet again, a set of variables describing
the length of edges between traditional agri-
cultural land uses enter the models with positi-
ve value.
The spatial dispersion of centres of activity
in irrigated cultures was uniform before major
habitat changes started, but the pattern was at
random after 1996. The models suggest that
this change is due mainly to the desertion of
territories through habitat change or to the de-
ath of individuals in the expanded road net-
work. Indeed, the decline of the Barn Owl po-
pulation reported in this study is the result of a
complex combination of factors that emerge
selectively at different spatial scales (Fig. 2
and 5). Nonetheless, the variable «number of
man-made structures with cavities» entered all
the models. The Barn Owl is sensitive to
small-scale habitat change, as shown by the
finding that the loss of a single occupied site
can lead to the desertion of other nearby sites
(the «knock-on effect»; Ramsden, 1998). Furt-
hermore, the disappearance of the preferred
roost can lead to territory desertion even if the
nest is left untouched (Martínez, 1998). Ac-
cordingly, the increasing loss of traditional
nesting or roosting sites may have prompted
territory desertion in our study areas. In an
non-exclusive way, it could be suggested that
owls died in the extended road network, as
suggested by the fact that the variable «roads»
enters every single model with negative value
(Tables 1 to 4) (see Van der Hut et al., 1992;
Taylor, 1994).
The loss of foraging areas (edges, ditches)
could have a major bearing in explaining the
absence of multiple broods and the reduction in
productivity recorded after 1996 in irrigated
cultures (Martínez, 1998; Martínez & López,
1999; see also Van der Hut et al., 1992; Taylor,
1994; De Jong, 1995; Alasdair et al., 2000). In
the absence of habitat change, the Barn Owl
population in this area has been found to be
self-maintaining, as shown by the finding that
the observed productivity before 1996 is no-
tably higher than the critical productivity (Mar-
tínez, 1998). However, the critical productivity
is not reached after habitat loss started and no
imports take over vacancies (Martínez, 1998;
Martínez & López, 1999). Bruijn (1994) has
also documented variations in the ability to
compensate mortality through reproduction bet-
ween two Barn Owl populations.
Interestingly enough, the likelihood of ha-
ving a deserted territory increases when records
of persecution are available within a 5.6 km
radius around the centres of activity in both
areas. This is amenable to the finding that wild-
314 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317
FIG. 4.—Percentage of deviance explained at three spatial scales by the Generalized Linear Models for the
probability of presence of Barn Owls comparing occupied vs. deserted territories in irrigated areas.
[Porcentajes de devianza explicados a tres escalas espaciales por los Modelos Generales Linearizados
para la probabilidad de presencia de Lechuzas Comunes comparando territorios ocupados y abandonados en
áreas de cultivos de regadío]

life protection laws have not been effectively
put into practice over the last decade in the
East of Spain (Martínez & López, 1995; Martí-
nez et al., 1996; Martínez et al., 2001). A prio-
ri, the variable «persecution» could have been
regarded as a bad predictor, especially at the
high-resolution spatial scales (i.e., nest-site sca-
le) because there are few data on raptor and
owl casualties available in relation to the area
surveyed. The fact that it enters the occupation
versus desertion models at the low-resolution
scale stresses the importance of illegal hunting
in Spain.
Subsequent to the success of various Euro-
pean population reinforcement programs (Van
der Hut et al., 1992; Bruijn, 1994; Taylor,
1994), the erection of nest-boxes in alternative
sites is a commonly proposed measure within
the framework of environmental impact stu-
dies in order to compensate for habitat loss for
Barn Owls in Spain. The models show that the
loss of cavities accounts for between 10 to 39%
of the explained deviance (Fig. 4 and 5), the
rest being accounted for by habitat loss and
persecution. Territory occupancy by Barn Owls
is the result of a complex combination of fea-
tures, which suggests that erecting nests-boxes
would achieve limited success in our study are-
as unless managers promote policies encoura-
ging the preservation or restoration of the agri-
pastoral network in designed core areas as well
as the effective control of persecution. The im-
plementation of such policies would be also
beneficial for the guild of Strigiformes and Fal-
coniformes in the East of Spain (Rico et al.,
2001; Sánchez-Zapata & Calvo, 1999; Martí-
nez et al., 2003; Martínez & Zuberogoitia,
2004), where negative population trends have
been already reported for owls (Martínez &
Zuberogoitia, 2003b, 2004; Alonso et al.,
2003).
In conclusion, it is suggested that: (1) the
current distribution and abundance of Barn
Owls in the study areas are artefacts of anthro-
pogenic land transformation, (2) population
reinforcement schemes would benefit from mo-
delling pre and post-transformation habitat af-
finities of owls, (3) environmental studies pro-
tecting only small areas around occupied Barn
Owl nests may not be sufficiently meeting the
habitat requirements of this species, as shown
by the relatively high percentages of deviance
explained by the GLMs at broad spatial scales
(Tables 1 and 2), (4) cataloguing of owls would
achieve optimal efficiency if based on scientific
evidence and (5) should the population decline
caused by habitat loss persist in the study areas,
a review for possible re-cataloguing of the Barn
Owl will be needed in brief.
ACKNOWLEDGEMENTS.—We are most indebted to
Fernando Falcó, Luis Rico, HELIACA, Roque Be-
lenguer, Alejandro Izquierdo, Juanjo Izquierdo, Fran,
G.E.R.-Valencia, RONCADELL, Javier Simarro,
Mavi Corell, Javier Martínez-Valle, Domingo, Ma-
nuel Carrascosa, Angel, Paco Segarra, Luis Fidel,
Alfonso Lario, the C.E.P.M.N., and the S.E.R.
Thanks are due to AMBARTEC S.L. for their logis-
tic support. A previous study with David Serrano
provided us with useful ideas for this study. Vincen-
zo Penteriani and the Editorial Board of Ardeola
made valuable comments on the manuscript. Javier
Seoane greatly enhanced our understanding of the
statistics in this paper.
BIBLIOGRAPHY
AGENCIA DEL MEDIO AMBIENTE. 1997. Mapa geo-
científico de la Provincia de Alicante. Generalitat
Valenciana. Valencia.
ALASDAIR, R., WEBBON, CH., GLUES, D. & HARRIS,
S. 2000. Changes in the food of Barn Owls (Tyto
alba) between 1974 and 1997. Mammal Review,
30: 107-129.
ALONSO, R., OREJAS, P. ZUBEROGOITIA. I. & MARTÍ-
NEZ, J. A. 2003. El Autillo Europeo (Otus scops).
En, R. Martí & J. C. del Moral (Eds). Atlas Na-
cional de las Aves Reproductoras de España. pp.
314-315. Dirección General de Conservación de la
Naturaleza- SEO. Madrid.
BURNHAM, K. P., & ANDERSON, R. 2002. Model se-
lection and multimodel inference. A practical in-
formation-theoric approach. Springer-Verlag,
New York.
BROWN, D. 1975. A test of randomness of nesting
spacing. Wildfowl 26, 102-103.
BRUIJN, O. 1994. Population Ecology and Conser-
vation of the Barn Owl Tyto alba in farmlands
habitats in Liemers and Achterhoek (The Nether-
lands). Ardea, 82: 1-109.
BUSTAMANTE, J. 1997. Predictive models for Lesser
Kestrel Falco naumanni distribution, abundance
and extinction in southern Spain. Biological Con-
servation,80: 153-160.
DEJONG, J. 1995. De Kerkui en Adere in Nederland
Voorkomende Ulien. Teknigen van Ad Cameron.
Friese Pers Boekerij. Leeuwarden.
FORMAN, R. T. & GORDON, M. 1986. Landscape
Ecology. John Wiley and Sons.
HABITAT LOSS AND BARN OWL POPULATION DECLINE 315
Ardeola 51(2), 2004, 303-317

ILLE, R. & GRINSCHGL, F. 2000. Little Owl in Aus-
tria. Habitat characteristics and population den-
sity. Athenews, 4: 2.
JOHNSON, D. H. 1980. The comparison of usage and
availability measurements for evaluating resource
preference. Ecology, 61: 65-71.
LANDIS, J. R. & KOCH, G. G. 1977. The measure-
ments of observer agreement for categorical data.
Biometrics, 33: 159-174.
MANEL, S., WILLIAMS, H. & ORMEROD, S. 2001. Eva-
luating presence–absence models in ecology: the
need to account for prevalence. Journal of Ap-
plied Ecology,38: 921-931.
MARCHESI, L., SERGIO, F. & PEDRINI, P. 2002. Costs
and benefits of living in human-altered landsca-
pes for the Eagle Owl Bubo bubo. Ibis, 144: 164-
177.
MARTÍNEZ, J. A. 1998. Biología reproductora de la
Lechuza común Tyto alba en los cultivos de rega-
dío de Valencia. Consellería de Medi Ambient.
Valencia.
MARTÍNEZ, J. A. & LÓPEZ, G. 1995. Dispersal and
causes of mortality for the Barn Owl Tyto alba in
Spain. Ardeola, 42: 29-38.
MARTÍNEZ, J. A. & LÓPEZ, G. 1999. Breeding eco-
logy of the Barn Owl (Tyto alba) in Valencia
(East of Spain). Journal für Ornithologie, 140:
93-99.
MARTÍNEZ, J. A. & ZUBEROGOITIA, I. 2003a. La Le-
chuza Común (Tyto alba). En, R. Martí y J. C. del
Moral (Eds.). Atlas Nacional de las Aves Repro-
ductoras de España, pp. 312-313. Dirección Ge-
neral de Conservación de la Naturaleza- SEO. Ma-
drid.
MARTÍNEZ, J. A. & ZUBEROGOITIA, I. 2003b. La ur-
banización litoral en Alicante reduce la población
del Mochuelo Europeo (Athene noctua). Quercus,
203: 52.
MARTÍNEZ, J. A. & ZUBEROGOITIA, I. 2004. Habitat
preferences for Long-eared owls (Asio otus) and
Little owls (Athene noctua) in semi-arid environ-
ments at three spatial scales. Bird Study, 51: 163-
169.
MARTÍNEZ, J. A., IZQUIERDO, A. & ZUBEROGOITIA, I.
2001. Causes of admission of raptors in rescue
centres of the East of Spain as proximate causes of
mortality. Biota, 2: 163-169.
MARTÍNEZ, J. A., SERRANO, D. & ZUBEROGOITIA, I.
2003. Predictive models of habitat preferences for
the Eurasian Eagle Owl Bubo bubo: a multi-scale
approach. Ecography, 26: 21-28.
MARTÍNEZ, J. A., FALCÓ, F., RICO, L., IZQUIERDO, A.
L., IZQUIERDO, J., MARTÍNEZ-VALLE, R. & CALA-
TAYUD, P. 1996. Causas de mortalidad de las ra-
paces nocturnas en la Comunidad Valenciana.
Quercus,126: 18-19.
MCCOLLIN, D. 1998. Forest edges and habitat selec-
tion in birds: a functional approach. Ecography,
21: 247-260.
MCCULLAGH, P. & NELDER, J. A. (1989). Generali-
sed Linear Modelling. Chapman and Hall.
MIKKOLA, H. 1983. Owls of Europe. Poyser, Carton.
London.
NICHOLLS, A. O. 1989. How to make biological sur-
veys go further with generalised linear models.
Biological Conservation, 50: 51-75.
PAIN, D. J. & PIENKOWSKI, M. W. 1997. Farming
and Birds in Europe: the Common Agricultural
Policy and its Implications for Bird Conservation.
Academic Press, Cambridge.
PENTERIANI, V. GALLARDO, V. ROCHE, M. & CAZAS-
SUS, H. 2001. Effects of landscape spatial structu-
re and composition on the settlement of the Eura-
sian Eagle Owl Bubo bubo in a Mediterranean
habitat. Ardea, 89: 331-340.
PENTERIANI, V. GALLARDO, & M. ROCHE, P. 2002.
Habitat structure and prey density affect Eagle
Owl Bubo bubo density and breeding performan-
ce: a case of intrapopulation heterogeneity. Jour-
nal of Zoology London, 257: 365-372.
RAMSDEN, D. 1998. Effect of Barn Owl conversion
on local populations of Barn Owl Tyto alba. Bird
Study, 45: 68-76.
RICO, L. 1997. Estudio de base sobre la situación
de las comunidades faunísticas en el Parque Na-
tural del Carrascal de la Font Rotja mediante
la selección de bioindicadores. 1997. Consellería
de Medi Ambient. Generalitat Valenciana. Ali-
cante.
RICO, L., MARTÍNEZ, J. A., MORÁN, S. NAVARRO, J.
R. & RICO, D. 2001. Preferencias de hábitat del
Águila-azor Perdicera Hieraaetus fasciatus en Ali-
cante a dos escalas espaciales. Ardeola, 48: 55-
62.
SÁNCHEZ-ZAPATA, J. A. & CALVO, J. F. 1999. Raptor
distribution in relation to landscape composition in
semi-arid Mediterranean habitats. Journal of Ap-
plied Ecology, 36: 245-262.
SHAWYER, C. 1987. The Barn Owl in the British Is-
les. Its past, present and future. The Hawk Trust.
London.
SIMARRO, P. 2002. Los cultivos de secano de la
provincia de Alicante. Patrimonio Cultural. San
Juan.
TAYLOR, I. 1994. Barn Owls. Predation-prey rela-
tionships and conservation. Cambridge Univer-
sity Press.
TITUS, K., MOSHER, J. A. & WILLIAMS, B. K. 1984.
Chance corrected classification for use in discri-
minant analysis and ecological applications. Ame-
rican Midland Naturalist, 111: 1-7.
TOMS, M. P., CRICK, H. & SHAWYER, C. R. 2001.
The status of breeding Barn Owls Tyto alba in the
United Kingdom 1995-1997. Bird Study, 48: 32-
37.
TUCKER, G. M. & HEATH, M. F. 1994. Birds in Eu-
rope: their conservation status. BirdLife Conser-
vation Series, n.o3. BirdLife International.
316 MARTÍNEZ, J. A. & ZUBEROGOITIA, Í.
Ardeola 51(2), 2004, 303-317

VAN DER HUT, DE JONG, J. & OSIECK, E. 1992. Bio-
logie en bescherming van der Kerkuil Tyto alba:
aanzet tot het beschermingsplan. Technical Rap-
port. Vogelbescherming 7.
ZUBEROGOITIA, I. 2000. La influencia de los factores
meteorológicos sobre el éxito reproductor de la
Lechuza Común. Ardeola, 47: 49-56.
ZUBEROGOITIA, I. 2002. Eco-etología de la comuni-
dad de rapaces nocturnas de Bizkaia. PhD Dis-
sertation. Bask Country University.
ZUBEROGOITIA, I & CAMPOS, L. F. 1998. Censusing
owls in large areas: a comparison between met-
hods. Ardeola, 45: 47-53.
Jose Antonio Martínez is a freelance ecologists that
started his own long-term project on habitat selection
for the guild of raptors and owls in the East of Spain
in 1989. Iñigo Zuberogoitia carries out research on
the community of raptors, owls and carnivores in
the North of Spain (since 1989) as a member of the
E. M. Icarus s.l. environmental consulting. They joi-
ned efforts in 1996 to produce a join scheme aimed
at evaluating the effects of habitat loss on animal
communities, among others.
[Recibido: 11-12-03]
[Aceptado: 02-06-04]
HABITAT LOSS AND BARN OWL POPULATION DECLINE 317
Ardeola 51(2), 2004, 303-317