Hole Selection by Nesting Swifts in Medieval City-Walls of Central Spain

Abstract

Disturbance and predation risks from terrestrial animals decline the higher up the nest sites selected by birds that nest in wall cavities are located. Terrestrial predators can also negate the protective quality of higher nesting sites by approaching from above in walls. It is unknown how terrestrial predation risks from below and above walls determine nest site selection in cavity-nesting species. In relation to this situation, we describe nest-site selection in common swifts Apus apus in the medieval city walls of Ávila, Spain. We recorded the entry size, hole depth and the horizontal and vertical positions of cavities. Most cavities were empty despite their size being suitable for nesting. Swifts nested in cavities at least 12 cm deep and with an entry between 3.5 cm and 13 cm wide. Nests were 3.5 m above the ground and 1.7 m below the top of the wall, although there were suitable cavities at the lower and higher extremes, respectively. Higher predation risks and disturbances could explain why suitable cavities were empty at lower and higher heights. The distances to the ground and to the top of the wall, as well as the distance to the nearest corner, accounted for about one-tenth of the probability that a cavity was used for nesting. Our data do not indicate a possible reason for nesting near corners, but weather is an obvious candidate.

Full text

HOLE SELECTION BY NESTING SWIFTS
IN MEDIEVAL CITY-WALLS OF CENTRAL SPAIN
SELECCIÓN DE LAS CAVIDADES DE NIDIFICACIÓN
POR EL VENCEJO COMÚN EN MURALLAS MEDIEVALES
DEL CENTRO DE ESPAÑA
Luis CORRALES1*, Luis Miguel BAUTISTA2, Tomás SANTAMARÍA1
and Pedro MAS1
SUMMARY.—Disturbance and predation risks from terrestrial animals decline the higher up the nest
sites selected by birds that nest in wall cavities are located. Terrestrial predators can also negate the
protective quality of higher nesting sites by approaching from above in walls. It is unknown how
terrestrial predation risks from below and above walls determine nest site selection in cavity-nesting
species. In relation to this situation, we describe nest-site selection in common swifts Apus apus in the
medieval city walls of Ávila, Spain. We recorded the entry size, hole depth and the horizontal and
vertical positions of cavities. Most cavities were empty despite their size being suitable for nesting.
Swifts nested in cavities at least 12 cm deep and with an entry between 3.5 cm and 13 cm wide. Nests
were 3.5 m above the ground and 1.7 m below the top of the wall, although there were suitable cavities
at the lower and higher extremes, respectively. Higher predation risks and disturbances could explain
why suitable cavities were empty at lower and higher heights. The distances to the ground and to the top
of the wall, as well as the distance to the nearest corner, accounted for about one-tenth of the probability
that a cavity was used for nesting. Our data do not indicate a possible reason for nesting near corners,
but weather is an obvious candidate.
Key words:Apus apus, castle, cavity-nesting, common swift, nest selection.
RESUMEN.—Las molestias y el riesgo de depredación por animales terrestres determinan que las aves
que emplean paredes para anidar seleccionen cavidades a mayor altura. También podrían seleccionar
el lugar del nido a menor altura si los depredadores terrestres se encuentran presentes encima de la pa-
red, pero se desconoce cual es la selección de cavidades para anidar cuando hay riesgo de depreda-
ción terrestre en ambos lados de la pared. Para dilucidarlo, estudiamos la selección de cavidades en el
vencejo común Apus apus en las murallas medievales de Ávila, España. Registramos el tamaño de la
entrada, la profundidad de orificio y las posiciones horizontal y vertical de las cavidades. La mayor
parte de las cavidades se encontraron vacías, incluso cuando su tamaño era apropiado para anidar.
Los vencejos anidaron en cavidades con una profundidad de al menos 12 cm y orificios de entrada no
más pequeños de 3,5 cm ni mayores de 13 cm. Las oquedades con nidos no se encontraron a menos de
Ardeola 60(2), 2013, 291-304 DOI: 10.13157/arla.60.2.2013.291
1Universidad Católica de Ávila, C/ Canteros s/n, 05005 Ávila, Spain.
2National Museum of Natural History, CSIC. José Gutiérrez Abascal 2, 28006 Madrid, Spain.
*Corresponding author: recogemensajes@yahoo.es

INTRODUCTION
Cavity-nesting birds that cannot excavate
their own holes depend on the availability of
cavities. Consequently, the location and den-
sity of breeders should conform to the dis-
tribution of suitable holes (Newton, 1998).
However, this distribution pattern is not
realised in all species: some cavity-nesting
bird species prefer buildings with many
cavities in the roof and walls (Franco et al.,
2005), but a shortage of nest-sites may not
have a detrimental effect on the population
dynamics in other species (Forero et al.,
1996). Where cavities abound, some birds
prefer higher nests to avoid terrestrial preda-
tors (Fisher and Wiebe, 2006) but other
species avoid the upper cavities when nesting
on cliffs, for the same reason (Penloup et al.,
1997). Therefore, it may be that some cavity-
nesting bird species select holes in the mid-
dle height range to avoid terrestrial predators
and disturbances from both above and below.
For instance, small tree-nesting bird species
often avoid extreme positions at the bottom
and top of the tree, and do not select the less
dense, external positions within the tree, in
order to reduce predation risk (e.g., Alonso
et al., 1991). However, as far as we know,
previous studies have not reported the same
effects in bird species that nest in buildings.
Not surprisingly, most terrestrial predators
cannot reach nests from above because they do
not have access to building roofs. However,
old medieval city walls are an exception be-
cause disturbance caused by people walking
near and above the walls, and nocturnal pre-
dation by terrestrial vertebrates such as ro-
dents, can occur from both above and below
the nests and are thus plausible determinants
of nest-site selection.
Birds that nest in wall cavities, such as
some sparrows, swifts and kestrels, may
tolerate conspecifics nesting nearby where
there are several cavities in the same wall.
Common swifts Apus apus nesting in me-
dieval city walls are good candidates with
which to study the effectsofdisturbance
and predation risk on the nest-site selection
of a cavity-nesting species. Common swifts
nest inawider variety of sites than any
other species of Apus (Lack, 1956a). Swifts
usually nest in buildings but they can also
be found nesting in holes in trees, cliffs and
crevices, and even in nestboxes (Beaud, 2010;
Froehlich, 2006; Kharkova and Boehme,
2005; Cortes, 2005; Guenther et al., 2009;
Beschow, 2003; Cameliti and Boano, 2002;
Roger and Fossé, 2001; García and Álvarez,
1996; Holmgren, 1993; Huber, 1990; Gámez,
1989; Ferrero et al., 1983; Vielliard, 1978;
Herrera and Ramírez, 1975; Lack, 1956a). It
would be expected that there exist plausible
reasons for choosing buildings, and at least
three such reasons have been identified
(Lack, 1956b; Bernis, 1988a): (a) they offer
an abundance of available cavities of suitable
size; (b) most have a minimum height facili-
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
292
3,5 m del suelo ni a menos de 1,7 m a la cima de la pared, aunque hubiera cavidades convenientes tan-
to a menor como a mayor altura. El mayor riesgo de depredación y las molestias tanto cerca del suelo
como de las almenas podrían explicar que hubiese cavidades adecuadas para nidificar que sin embar-
go estaban vacías en ambos extremos de la muralla. Las distancias verticales al suelo y al borde supe-
rior de la muralla, junto con la distancia horizontal al rincón más cercano determinaron alrededor de
una décima parte de la probabilidad de anidar en una cavidad de dimensiones adecuadas. Los datos
registrados no explican las causas de la preferencia para anidar cerca de los rincones, aunque la meteo-
rología sea un candidato obvio.
Palabras clave:Apus apus, cavidad para anidar, muralla, vencejo común, nido.

tating take-off; and (c) the minimum open
space in front of the nesting hole is sufficient
for flight. These three requirements may vary
between breeding colonies and therefore their
quantification in more than one colony may
be informative. This study aims to shed light
on the first, taking into account other factors
that can determine the preferred nesting
height in walls, such as human disturbance
and the risk of predation by terrestrial preda-
tors from both above and below.
Swifts usually enter their nesting holes
with direct flight, and take-off is charac-
terised by an initial free-fall. Therefore,
nesting holes should be easily accessible
from open space and located at a minimum
height. Previous studies of swifts breeding
in old buildings have described the height
constraint (Pellantová, 1975; Bernis, 1988a):
in Spain the nests are 3-5 m up in small towns
(Bernis, 1988b), at 12-16 m in houses in the
Czech Republic (Pellantová, 1975) and above
6 m in Oxford (some near Oxford were only
at 3 m; Lack, 1956b). Given the variability
in the minimum heights, it may be that take-
off does not require a height greater than
3 m, although higher nests may serve to
conserve energy at the beginning of flight.
The availability of cavity heights is limited
by the maximum height of the building and
the vertical distribution of the cavities.
Therefore, it is unclear whether the swifts
select holes at the maximum height, or sim-
ply avoid holes below 3 m. The effect of
the distance to the nearest obstacle, and its
interaction with the minimum height of
the nests, was not analysed in this study
because the sectors of the medieval city
walls included were free of nearby obstacles,
and therefore flight was not restricted by
the distance to the nearest obstacle in front
of the nests.
Higher holes may be unsuitable for nesting
because of a potentially higher risk of preda-
tion from terrestrial predators approaching
from above the wall. It is unsurprising that
the effect of predation risk from above has
not been explored in swifts, because most
buildings do not allow roof access to terres-
trial predators. The medieval walls, with easy
access to the tops of the walls from the in-
side face, may allow cavity access to rodents
from above. Recording rodent presence
would indicate whether they can access the
wall from both extremes. The presence of
mice, as a surrogate measure of the risk
of predation, would explain the avoidance
by swifts of nest-site locations near the tops
and bottoms of walls.
People on the walkway located behind
the crenelations of the medieval wall may
also disturb the swifts. The walkways are a
popular tourist attraction in the city of Ávila:
c. 19,000 people walk along them every
month (Troitiño et al., 2012) during the
swift breeding season. Human disturbance
and predation risk are not mutually exclu-
sive and both were considered possible
explanations for avoidance by swifts of the
cavities located on the upper end of the wall.
Our aim was to compare the size and po-
sition of the cavities with and without swift
nests, describing the distribution of the cavi-
ties in the towers and walls of a fortified me-
dieval city. We also recorded the presence/ab-
sence of mouse droppings in cavities suitable
for nesting, testing the potential effect of mice
on the use of cavities by swifts. Although
the risk of predation and the nest height are
probably associated, the former is also re-
lated to the maximum size of the hole, as
described in other species of passerines
(e.g., Cordero, 1986; Van Balen et al., 1982).
Therefore, we also measured the size of the
cavities.
STUDY AREA AND METHODS
The study was conducted in the fortified
city of Ávila, central Spain. The city is
surrounded by 2,516m ofwalls and 87
Ardeola 60(2), 2013, 291-304
NEST-SITE SELECTION BY SWIFTS IN WALL CAVITIES 293

towers. The walls are about 3 m thick, 11 m
high and 20 m wide; towers are about 6-7 m
thick, 15 m high and 27 m wide. The exter-
nal side of the walls and towers offer small
cavities, some used by cavity nesting bird
species such as common swifts. In Ávila the
population size of swifts is approximately
12,400 birds (L. Corrales, unpublished).
Swift density in the town is about 1,700
birds/km
–
2(L. Corrales, unpublished), larger
than in other towns of similar size (Bernis,
1988a). In this study we selected two sectors
of the medieval walls: between and including
towers 2 and 3 (East) and towers 41 to 45
(West, fig. 1). The total length of these
towers was 119 m and the total length of
the intermediate walls was 81 m. Overall
we looked for cavities in 923 m2of walls and
2,401 m2of towers. Cavities were sampled
between May and June 2011, when the walls
and towers were being restored. An observer
used scaffolding erected from the bottom
to the top of the walls and towers to inspect
the cavities. Each hole was visited only
once, because the scaffolding was moved
between sites and never returned to a pre-
vious location.
We recorded the vertical distance to the
ground of each cavity, the horizontal dis-
tance to the nearest corner, and the height
and length of walls and towers (fig. 2). We
measured the maximum and minimum length
of the entry to the hole (max and min, cm),
and calculated the product of max ×min
(cm2) as an index for the size of the entry.
Hole depth (D, cm) was also measured.
Cavities were classified as occupied and
empty. Empty cavities of similar dimen-
sions to occupied ones were labelled as
‘suitable’, and empty cavities of different
dimensions to occupied ones were labelled
as ‘unsuitable’.
We recorded the presence/absence of
excrement in cavities suitable for nesting.
Mouse abundance was not sampled because
the erection of scaffoldings and maintenance
work would have affected mouse presence.
We did not record the frequency of people
walking above or below the medieval walls,
because this was recorded by the city tourism
department. The medieval city walls were
open to public access every day during the
study period. People may disturb the birds
through the battlements, which are short and
small in size.
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
294
FIG. 1.—Map of the city walls of Ávila (Central
Spain) with the two sampling sites highlighted on
the east (towers 2-3) and the west (towers 41-45).
[Plano de las murallas de Ávila (centro de Espa-
ña), con dos lugares de muestreo destacados en
el este (torres 2-3) y el oeste (torres 41-45).]

Statistical analyses
The randomness of the frequency distribu-
tions of nests (presence, absence) according
to the factors ‘building type’ (tower, wall)
and ‘height of cavity’ (centred, extreme) was
tested with the chi-square test. The statistical
effects of the two previous factors on the dis-
tances and sizes of the holes were analysed
with Generalised Linear Models (GLM)
in JMP (SAS Institute, 2009). Continuous
data were modelled in GLM with a normal
distribution and an identity link function.
The descriptive statistics are shown in the
text as the mean ± standard deviation and
in the figures as the mean ± standard error.
The means calculated with small sample
sizes (N < 10) were compared with the
non-parametric Wilcoxon-Mann Whitney W
statistic test.
We calculated two types of regression
models (stepwise and logistic ordinal) to
compute the probability of nesting as a func-
tion of entry size, cavity depth, distance
from a cavity to the nearest corner, and
cavity height above ground. Nesting was
coded as an ordinal variable (0, 1). Predictors
were selected following a mixed forward-
backward stepwise process, with P-values to
enter and leave of 0.25. Selected variables
were subjected to logistic analysis to calcu-
late the variable regression coefficients, the
Ward statistic and the R-square of the model.
Stepwise and logistic regression models
were calculated twice: first with all holes,
and second with unsuitable holes excluded.
The first analysis aimed to identify the
physical variables that best predicted the
use of any cavity for nesting. The second
sample was selected to identify the physi-
Ardeola 60(2), 2013, 291-304
NEST-SITE SELECTION BY SWIFTS IN WALL CAVITIES 295
FIG. 2.—Diagram of the measurements taken for cavities in walls and towers: height of hole
(
Hh
)
, dis-
tance to the nearest corner
(
C
)
, wall height
(
Hw
)
and wall length
(
W
)
.
[Diagrama de las distancias medidas en cavidades de paredes y torres. Las distancias fueron la altura
del agujero
(
Hh
)
, la distancia a la esquina más cercana
(
C
)
, la altura de la pared
(
Hw
)
y la longitud de
la pared
(
W
)
.]

cal variables that best predicted the proba-
bility of nesting in cavities with suitable
dimensions.
RESULTS
We found 300 empty cavities and 63 with
nests. The empty cavities were shallower
than those with nests, and the entry size
was smaller in the former than in the latter
(table 1). The depth in 64 empty cavities was
no greater than 9 cm. The minimum depth
in cavities with a nest had been recorded in a
previous study (Corrales et al., 2013). In 135
empty cavities the entry was narrower than
3 cm and in eight empty cavities it was greater
than 13 cm. Both distances of 3 cm and 13 cm
were, respectively, the minimum and maxi-
mum entry widths of cavities with nests
(Corrales et al., 2013). Overall there were
171 empty cavities with one or more dimen-
sions outside the range measured in cavities
with nests. We defined these 171 cavities as
unsuitable for nesting. We looked for addi-
tional differences between the remaining
129 cavities that were empty but apparent-
ly suitable for nesting and the 63 cavities
with nests. As expected, there were no sig-
nificant differences in orifice measurements
between empty but apparently suitable cavi-
ties and cavities with nests (means are shown
in table 1).
In walls we found 22 cavities with a nest
and 35 suitable but empty cavities. In towers
we found 41 cavities with a nest and 94
suitable but empty cavities. Suitable cavities,
whether or not they had a nest, were propor-
tionally more frequent in towers
(
c2
1= 9.7,
P = 0.002
)
taking into account the length of
walls and towers. The cavities were no more
frequent in towers than in walls when the
surface area explored was taken into account
(
c2
1= 0.3, P = 0.555
)
. Density of cavities was
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
296
TABLE 1
Mean dimensions (± SD) for cavities with (Used, N = 63) and without (Empty, N = 300) nests. The
ranges (minimum-maximum) are included. Statistical results for comparison of means are the Student
t-test, degrees of freedom (d.f.), and P-value (P). Measurements of empty cavities are shown in two
subgroups: empty cavities with dimensions ranges similar to those with a nest (Suitable, N = 129) and
empty cavities with at least one variable outside the observed range of cavities with nest (Unsuitable,
N = 171). Suitable cavities are split in two subgroups according to their height: empty cavities at
heights within the range of nests (Central, N = 72) and empty cavities beyond the height limits of
nests (Extreme, N = 57).
Variable Used Empty t
Maximum entry distance (cm) 7.8 ± 2.1 (4.0-12.0) 6.5 ± 3.4 (1.0-27.0) 3.0
Minimum entry distance (cm) 4.8 ± 1.2 (3.0-7.9) 3.3 ± 1.7 (0.5-8.0) 6.5
Entry size (cm) 39.1 ± 17.0 (12.0-77.0) 24.7 ± 20.4 (0.7-132.0) 5.2
Hole depth (cm) 24.3 ± 6.7 (14.0-50.0) 19.0 ± 10.8 (1.0-56.0) 3.8

0.06 holes/m2. The proportion of holes with
a nest did not differ between walls and
towers
(
c2
1= 0.3, P = 0.271
)
.
Although suitable cavities were present
between 1 and 16 m above the ground, nests
were located only in cavities above 3.5m
and below 12.2 m (fig. 3). The minimum
distance of a nest to the top of the wall was
1.7 m, four times greater than the equivalent
distance of a suitable but empty hole (0.4 m).
This was not an exception, but rather a statis-
tical pattern: the distance of the 5% of nests
nearest to the top was 1.8 ± 0.1m (N = 3),
a greater distance than the 1.1 ± 0.4 m as
calculated with the 5% nearest suitable
but empty cavities (N = 6, Wilcoxon Mann
Whitney rank test, W = 6, P < 0.012). The
same result was observed in cavities at the
bottom: the height of the lowest nest was
3.5 m, four times greater than the height of
the lowest suitable but empty hole (0.4 m).
Again, this was not an exception but a statis-
tical pattern: the mean height of the 5% of the
lowest nests was 3.8 ± 0.2 m (N = 3), greater
than the 5% lowest suitable but empty cavi-
ties (1.0 ± 0.0 m, N = 6, Wilcoxon Mann
Whitney rank test, W = 6, P < 0.012).
Within the height range of nests (3.5-
12.2 m), there were 72 suitable but empty
holes (‘centred’ cavities hereafter, fig. 3).
More nests were found in towers than in
walls, at heights between 3.5 and 12.2 m
(41 and 22, respectively), but swifts did not
show a preference for towers after the num-
ber of centred cavities available in each site
was taken into account
(
c2
1= 0.7, P = 0.411
)
.
Swifts nested in cavities located at 7.5 ±
2.3 m, a mean height close to half the height
of the towers and walls (7.3 m, Student t-test,
t= 0.8, df = 191, P = 0.399). In towers the
mean height of nests (8.0 ± 2.4 m, N = 41)
and the mean height of suitable but empty
cavities (7.3 ± 3.9 m, N = 94 cavities, cen-
tred and extreme combined) did not differ
Ardeola 60(2), 2013, 291-304
NEST-SITE SELECTION BY SWIFTS IN WALL CAVITIES 297
TABLE 1 (cont.)
[Valores medios ± SD de las cavidades con nidos (N = 63) y sin nidos (N = 300). Son incluidos los ran-
gos (mínimo-máximo). Los resultados estadísticos para la comparación de las medias son la t de Student,
los grados de libertad (d.f.) y el valor de significación (P). Las medidas de las cavidades vacías se mues-
tran en dos subgrupos: cavidades vacías con dimensiones dentro del rango de las cavidades con nido
(Adecuadas, N = 129) y cavidades vacías con al menos una variable fuera de los rangos observados
en cavidades con nido (Inadecuadas, N = 171). Las cavidades adecuadas son divididas en dos subgru-
pos según su altura: cavidades vacías con alturas dentro del rango de las que tienen nidos (Central,
N = 72) y cavidades vacías más allá de los límites de altura de las que tienen nidos (Extremo, N = 57).]
Empty
Suitable
d.f. P Central Extreme Unsuitable Variable
361 0.003 8.0 ± 1.8 8.0 ± 1.9 5.3 ± 3.9 Maximum entry distance (cm)
361 <0.001 4.8 ± 1.1 4.9 ± 1.1 2.2 ± 1.2 Minimum entry distance (cm)
361 <0.001 39.0 ± 14.7 39.1 ± 14.4 13.8 ± 17.2 Entry size (cm)
361 <0.001 25.7 ± 7.7 23.9 ± 7.5 14.4 ± 10.6 Hole depth (cm)

(
GLM, c2
1= 1.0, P = 0.309
)
. The same re-
sult was seen in walls: the mean height of
nests (6.6 ± 1.8 m, N = 22) and the mean
height of empty holes (5.5 ± 2.6 m, N = 35,
centred and extreme combined) did not differ
(
GLM, c2
1= 3.6, P = 0.06
)
. The frequency
distribution of the heights of holes suggests
that swifts avoided the highest and lowest
holes in towers and as well as the lowest holes
in walls (fig. 3). The statistical analysis
of variance between heights of nests and
heights of empty but suitable holes (centred
and extreme combined) indeed showed that
variances were significantly different in the
towers and as well as the walls (Levene test
of variances, F1, 133 = 12.5, P < 0.001 and
F1, 55 = 10.4, P = 0.002, towers and walls
respectively). After excluding suitable but
empty holes at the extreme heights, we
found that the variance of the cavity height
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
298
FIG. 3.—Frequency distribution and boxplots of occupied cavities heights and empty but apparently
suitable cavity heights in towers and walls. Empty holes at extreme heights are shown with squares.
The range of heights (minimum-maximum) of towers and walls are displayed with a grey band. Notice
that some extreme cavities (squares) had the same height as centred cavities (circles) because they were
located in towers or walls of different heights.
[Distribución de frecuencias y diagramas de cajas de alturas de cavidades ocupadas y cavidades apa-
rentemente adecuadas pero vacías en torres y paredes. Los cuadrados muestran los agujeros vacíos
en alturas extremas. Se muestra con una banda gris el rango de alturas (mínimo-máximo) de torres y
paredes. Algunas cavidades extremas (cuadrados) tenían la misma altura que las cavidades centradas
(círculos) porque fueron localizadas en torres o paredes de altura diferente.]

in towers did not differ between the holes
with nests and the empty but suitable holes
(Levene test of variances, F1, 90 = 2.8, P =
0.100). The same result was obtained in the
walls (Levene test of variances, F1, 41 = 1.5,
P = 0.227). We concluded that swifts did
not show active selection of cavity height
within the central range of heights, because
the variances of the used cavities and the
suitable cavities were not different. This
finding is further supported by comparing
the mean heights of the used and the empty
suitable cavities. The mean height of centred
suitable but empty holes in towers (8.0 ± 2.1
m, N = 51) did not significantly differ from
the mean height of holes with nests
(
GLM,
c2
1= 0.0, P = 0.969). The same result was
seen in walls: the mean height of suitable but
empty holes in the central subsample (6.9 ±
1.4 m, N = 21) did not significantly differ
from the mean height of holes with nests
(
GLM, c2
1= 0.2, P = 0.622). Note that these
comparisons are not as trivial as they seem
because the limits of the height in the cen-
tral suitable but empty holes were set by the
height range of nests (minimum and maxi-
mum heights), and not by their variance.
Nests were located at 3.5 ± 2.6 m from the
nearest corner, a shorter distance than that
observed in the suitable but empty holes
(4.3 ± 2.8 m, Student t-test: t = 2.1, df = 190,
P = 0.04). The statistical difference increased
when compared to the subsample of the
extreme suitable but empty holes (fig. 4).
The nests in towers were not closer to the
corner than the nests in walls (Student t-test:
Ardeola 60(2), 2013, 291-304
NEST-SITE SELECTION BY SWIFTS IN WALL CAVITIES 299
FIG. 4.—Distance to the nearest corner (mean ± SE) in towers (empty circles) and walls (filled circles)
for each cavity type.
[Distancia a la esquina más cercana (media ± SE) en torres (círculos vacíos) y paredes (círculos llenos)
de cada tipo de cavidad.]

t= 0.5, df = 61, P = 0.30). The distance to
the nearest corner did not significantly differ
between towers and walls in any of the three
categories of empty holes (suitable and cen-
tred, suitable at extreme heights, and unsuit-
able). A statistical analysis of both factors
combined, the type of hole (used, centred
and extreme) and the type of building (tower
vs wall), on the distance to the corner did
not modify the previous result: although the
full model was significant
(
N = 192 holes,
GLM, c2
3= 8.3, P = 0.041
)
, the mean cor-
ner distances remained not significantly
different between towers and walls
(
GLM,
c2
1= 2.1, P = 0.151
)
, as did the interaction
be tween both factors
(
GL M, c2
2= 0. 3,
P = 0.876
)
. Removing the interaction did
not change the result: the building type had
no significant effect on the distance to the
nearest corner
(
GLM, c2
1= 2.1, P = 0.145
)
.
Mean distances to the corner were signifi-
cantly different according only to the type of
hole in the combined model
(
GLM, c2
2= 6.8,
P = 0.033, fig. 4
)
. We concluded that nests
were located closer to the corners than central
and extreme holes, with no effect according
to the type of building.
The vertical and horizontal positions of
cavities significantly predicted the likelihood
of nesting in the medieval city of Ávila
when unsuitable holes were either included
or excluded from the analyses (table 2). Bear
in mind that we included the distance to the
ground twice
(
height and height2
)
to account
for the non-linear relationship with the
probability of nesting (fig. 3). When all holes
were included in the analysis, the stepwise
regression also selected the minimum length
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
300
TABLE 2
Coefficients (B) and statistics of the logistic function to assess the likelihood of swifts nesting in a cavi-
ty in the medieval walls of the city of Ávila. Analyses were calculated (a) with all cavities and (b) with
unsuitable cavities excluded.
[Coeficientes (B) y estadísticos de la función logística para evaluar la probabilidad de anidar de los
vencejos en una cavidad en las paredes medievales de la ciudad de Ávila. Los análisis fueron calcula-
dos (a) con todas las cavidades y (b) con cavidades inadecuadas excluidas.]
Variable B SE (B) Wald P
(a) All cavities (N = 363)
Minimum entry distance 0.543 0.105 26.7 <0.001
Height 1.406 0.326 18.6 0.022
Height squared
–
0.092 0.022 17.6 < 0.001
Distance to the corner
–
0.142 0.062 5.3 0.022
Depth 0.029 0.017 2.8 0.093
(b) Suitable and used cavities (N = 192)
Height 1.314 0.337 15.2 < 0.001
Height squared
–
0.083 0.022 13.8 < 0.001
Distance to the corner
–
0.129 0.064 4.1 0.043

of the orifice entry and the cavity depth as
predictors, in addition to the cavity distances
to the corner and to the ground. The logistic
function explained 24.3% of the probability
of nesting
(
N = 363, c2
5= 81.2, P < 0.001
)
when all holes were included, and 12.1%
when unsuitable holes were excluded
(
N =
192, c2
3= 29.5, P < 0.001
)
.
Scats from mice were found in ten out of
129 suitable but empty cavities, and in one
out of 63 cavities with nests. Thus, mouse
scats were marginally more frequent in
empty than in used cavities
(
N = 192, c2
1=
3.7, P = 0.056
)
. The percentage of suitable
but empty cavities with mouse scats was
6.9% in centred cavities and 8.8% in extreme
cavities. Thus, mouse scats were not more
frequent in cavities at extreme heights
(
N =
129, c2
1= 0.1, P = 0.701
)
. The mean height
of holes with mouse scats was 8.4 ± 3.7 m
(N = 11), no higher than the holes without
mouse scats (7.0 ± 3.3 m, ANOVA F1, 190 =
2.0, P = 0.160). Cavities with mouse scats
did not differ in size from cavities with swift
nests (ttests of maximum and minimum
entry distances, entry size and hole depth,
df = 53, allP>0.128).
DISCUSSION
Nesting swifts in the medieval city walls
of Ávila avoided cavities situated at extreme
heights, those too shallow to support a nest
and those with entry sizes that were too
narrow or too wide. These are common ob-
servations in cavity-nesting species, which
usually select an entry width that can limit
the size of potential predators entering the
cavity but with a diameter large enough for
the resident bird to enter (Wesolowski, 2002).
Both limits were observed in unsuitable cavi-
ties: entry size was either smaller or larger
than that of cavities with a nest.
Other bird species compete to some extent
with swifts, breeding in the same holes. Com-
mon starlings Sturnus vulgaris may displace
swifts in territorial disputes for nestboxes
(Lack, 1956b). In the study area, there is a
large population of spotless starlings S. uni-
color (Santamaría et al., 2009) but they were
not observed breeding in the medieval walls.
We did not observe any aggressive interaction
between swifts and starlings. House sparrows
Passer domesticus were not found, perhaps
because swifts easily displace them (Lack,
1956b), especially when the colony size is
large (Genton, 2009). A few nesting sparrows
were observed elsewhere in the city walls
(< 5% of holes, Corrales unpublished), in
the medieval towers where the curve of the
wall was greatest, and also in smaller holes
than those occupied by swifts. The small
percentage of holes with sparrows can be
explained by displacement by swifts, and
also because the typical cavity size is smaller
in sparrows: 3 ×9 cm and 2 ×9 cm inner
surfaces, for the house sparrow Passer do-
mesticus and tree sparrow P. montanus,
respectively (Cordero, 1986). No bats were
found in holes selected by swifts, nor in
any unsuitable holes. Although unsuitable
holes were too small for swifts, they were
too large overall for bats (for example: min.
1.8 cm in Pipistrellus and 2.5 cm in Seroti-
nus (Huitema, 2010).
Swift nest size (12 ×9 cm; Corrales et al.,
2013) requires cavities at least 9 cm deep to
hold the nest and some additional distance
to allow arrivals and departures of adults. The
minimum depth found in this study was 14
cm. This would indicate that adults avoided
holes with sufficient depth to hold the nest
but insufficient distance to land and take-off
between the nest and the cavity entry. Nest
sizes in Ávila did not differ when compared
to nests in other parts of Europe (Corrales et
al., 2013), perhaps because the body sizes of
swifts across Europe were not meaningfully
different either (Lack and Lack, 1951).
However, nests were located deeper in the
nesting cavity at other European sites. For
Ardeola 60(2), 2013, 291-304
NEST-SITE SELECTION BY SWIFTS IN WALL CAVITIES 301

instance, in the Czech Republic they were at
20-50 cm (Tr
ˇebon
ˇ) and 20-70 cm (Hrotovice)
from the cavity entry (Pellantová, 1975). In
Germany the depths ranged between 24 and
29 cm (Guenther and Hellmann, 2002).
Temperatures at those sites are lower than in
Ávila, and rain is also heavier during the
breeding season, hence it seems likely that
swifts positioned the nests deeper in the
holes to avoid harsher weather conditions
than those in Ávila. Cavity orientation should
determine to some extent cavity selection
because weather is an obvious consideration
to explain why swifts nested near corners.
Wall orientation effects on cavity selection
in Ávila should be explored further.
Swifts nested in about half (48%) of the
suitable holes located within the central strip
of the medieval walls. That percentage of
use was greater than in other swift species
(A. pallidus and A. melba; Brichetti et al.,
1988). The high percentage of use in suitable
centred holes and the large population of
swifts in Ávila (between 6,222 and 8,091
birds•km
–
2in the historic district between the
years 2006-2009; L. Corrales, unpublished),
suggests that nest-site competition in the
medieval city walls of Ávila is strong.
We did not measure breeding success
and therefore we cannot relate nest location
to breeding success. Several studies of other
(Gil-Delgado and Barba, 1987; Nilsson, 1984;
Osborne and Osborne, 1980; Tenaza, 1971;
Best, 1978) show that higher nests suffer
lower predation risks. Birds breeding early in
the season often achieve higher breeding suc-
cess. Hence, one approach in future studies is
to explore the link between fitness and nest se-
lection by sampling throughout the breeding
season to determine whether early and late
breeders are similarly distributed in the walls.
On the other hand, studies of cliff-nesting
birds report a significant effect of the dis-
tance from the top of the cliff on hatching
success (Harris et al., 1997). Perhaps pre-
dation risk at the nest is not a main concern
in swift species, as has been shown in the
american black swift Cypseloides níger
(Hirshman et al., 2007), but we cannot dis-
regard predation risk as the main explana-
tion for the height of cavity selection in the
Ávila walls as swift nests were not found in
high and low holes suitable for breeding. A
plausible explanation is that swifts avoid the
highest and lowest holes due to the presence
of mice, a potential predator of eggs in other
species of Apus (Penloup and Martin, 1995;
Penloup et al., 1997). However, we did not
find more mice in empty cavities at extreme
heights, and therefore we cannot conclude
that swifts avoided extreme heights to reduce
predation risk from mice. Simple experi-
ments can be done to test predation pressure
in relation to distance from the wall bottom
or top by measuring the survival of artificial
eggs placed at different heights. In the same
vein,itwould be interesting to examine
the distribution of nests in towers or walls
with restricted human access. The upperside
of the walls was open to pedestrians and
tourists: 22,762 tourists accessed these areas
in April and 15,431 in May, 2011 and an
average of 21,526 for April and 16,725
for May in the last decade (Troitiño et al.,
2012). This activity could cause distur-
bance to breeding swifts. Disturbance from
pedestrians may be detrimental to breeding
sparrow s (Sac a r rao a n d S o a res, 1975;
Nilsson, 1984), and is also likely to be detri-
mental to breeding swifts. On the other hand,
we did not find any nest in holes below 3.5 m,
the same height as in other studies (Bernis,
1988b; Lack, 1956b).
ACKNOW LE DGEMENTS.—We thank the City
Council of Ávila for allowing us to access the
nests in the medieval walls surrounding the city.
The company Construcción y Desarrollo de Ser-
vicios S.A. (VOLCONSA) restored the towers
and walls included in this study, and helped us to
locate the holes before they performed the main-
tenance work on the medieval city walls. We also
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
302

thank Pablo Corrales for assisting in fieldwork.
Ernest Garcia and Sarah Young kindly reviewed
the English translation.
BIBLIOGRAPHY
ALONSO, J. A., MUÑOZ-PULIDO, R., BAUTISTA, L.
M. and ALONSO, J. C. 1991. Nest-site selection
and nesting success in the azure-winged magpie
in central Spain. Bird Study, 38: 45-51.
BEAUD, M. 2010. Common swift Apus apus,
nesting on the molasse cliffs of the river Sarine
(Fribourg, Switzerland). A recapitulation of
breeding at natural sites throughout Switzer-
land. Nos Oiseaux, 57: 265-276.
BERNIS, F. 1988a. Los Vencejos. Su biología, su
presencia en las mesetas españolas como aves
urbanas. Universidad Complutense de Madrid.
Madrid.
BERNIS, F. 1988b. Aves de los medios urbano y
agrícola en las mesetas españolas. Monogra-
fías nº 2. SEO/BirdLife. Madrid.
BESCHOW, R. 2003. Tree-breeding swifts Apus
apus in the city of Spremberg. Result of a sur-
vey of swift population 2003. Ornithologische
Jahresberichte des Museum Heineanum, 21:
47-55.
BEST, L. B. 1978. Field sparrow reproductive
success and nesting ecology. Auk, 95: 9-22.
BRICHETTI, P., FOSCHI, U. F. and GELLINI, S. 1988.
Distribuzione e consistenza delle colonie di
Apodidae del Promontorio del Gargano (Pu-
glia). Rivista Italiana di Ornitologia, 58: 53-58.
CAMELITI, G. and BOANO, G. 2002. Nesting of the
swift Apus apus in tree cavities in Turin. Picus,
28: 105-107.
CORRALES, L., SANTAMARÍA, T., MAS, P. and CAM-
POS, F. 2013. Morfología y composición de los
nidos de vencejo común (Apus apus) en Ávila
(España). Cuadernos Abulenses, 41: 47-69.
CORDERO, P. J. 1986. Aspectos de la ecoetología
de la nidificación en el gor rión moli nero
(Passer m. montanus, (L.)) y el gorrión común
(Passer d. domesticus, (L.)) en Cataluña. Tesis
Doctoral, Universidad de Barcelona.
CORTES, J. 2005. Nesting of common swifts Apus
apus in palm trees. Gibraltar Bird Report, 4:
54-56.
FERRERO, J. J., NEGRO J. J., ROMÁN, J. A., CA-
BRERO, M., GUTIÉRR EZ , A. and PI ZA RRO, V.
M. 1983. Cría de vencejo común en árboles.
Ardeola, 30: 121.
FISHER, R. J. and WIEBE, K. L. 2006. Nest site
attributes and temporal patterns of northern
flicker nest loss: effects of predation and com-
petition. Oecologia, 147: 744-753.
FORERO, M. G., TELLA, J. L., DONÁZAR, J. A. and
HIRALDO, F. 1996. Can interspecific compe-
tition and nest site availability explain the de-
crease of lesser kestrel Falco naumanni popu-
lations? Biological Conservation, 78: 289-293.
FRANCO, A. M. A., MARQUES, J. T. and SUTHER-
LAND, W. J. 2005. Is nest-site availability limit-
ing lesser kestrel populations? A multiple scale
approach. Ibis, 147: 657-666.
FROEHLICH, C. 2006. Swifts (Apus apus) visiting
tree holes in a natural forest reserve in the
south western Palatinate (Rhineland-Palati-
nate). Fauna und Flora in Rheinland-Pfalz, 10:
1401-1402.
GÁMEZ, I. 1989. Nidificación en árboles. Ardeola,
36: 256.
GARCÍA, J. A. and ÁLVAREZ, C. 1996. Cría proba-
ble en robles. Ardeola, 43: 253.
GENTO N, B. 2009. Inter-specific relationships
between the common swift Apus apus and the
house sparrow Passer domesticus, some origi-
nal ideas in favour of the common swift. Nos
Oiseaux, 56: 67-86.
GIL-DELGADO, J. A. and BARBA, E. 1987. Aves
nidificantes en huecos de los naranjos. Medi-
terránea Serie Biología, 9: 29-40.
GUENTHER, E. and HELLMANN, M. 2002. Strong
decrease of population of tree-breeding swift
Apus apus in the northeastern Harz Mountains
(Sachsen-Anhalt)
-
was it the racoon Procyon
lotor?. Ornithologische Jahresberichte des
Museum Heineanum, 20: 81-98.
GUENTHER, E. and HELLMANN, M. 2009. The su-
per hole
-
common swifts Apus apus breeding in
a treehole in the course of 25 years. Ornitholo-
gische Jahresberichte des Museum Heineanum,
27: 79-83.
HARRIS, M. P., WANLESS, S., BARTON, T. R. and
ELSTON, D. A. 1997. Nest site characteristics,
duration of use and breeding success in the
guillemot Uria aalge. Ibis, 139: 468-476.
Ardeola 60(2), 2013, 291-304
NEST-SITE SELECTION BY SWIFTS IN WALL CAVITIES 303

HERRERA, C. M. and RAMÍREZ, A. 1975. El ven-
cejo común nidificando en árboles. Ardeola,
22: 146.
HIRSHMAN, S. E., GUNN, C. and LEVAD, R. G.
2007. Breeding phenology and success of black
swifts in Box Canyon, Ouray, Colorado. Wilson
Journal of Ornithology, 119: 678-685.
HUBER, H. 1990. Breeding site of the swift (Apus
apus) in rocks of shell limestone. Ornitholo-
gische Jahrbuch Baden-Württemberg, 6: 91-94.
HUITEMA, H. 2010. Verborgen dierenleven onder
dak. Dakenraad, 95: 58-61.
KHARKOVA, O. Y. and BOEHME, I. R. 2005. Pat-
terns of location of bird nests in an oak forest in
the Nature Reserve “Les na Vorskle” (Russia).
Berkut, 14: 201-213.
HOLMGREN, J. 1993. Young common swifts roost-
ing in foliage of trees. British Birds,86: 358-369.
LACK, D. 1956a. A review of the genera and
nesting habits of swifts. Auk, 73: 1-32.
LACK, D. 1956b. Swifts in a Tower. Methuen.
London.
LACK, D. and LACK, E. 1951. The breeding biolo-
gy of the swift Apus apus. Ibis, 93: 501-546.
NEWTON, I. 1998. Population Limitation in Birds.
San Diego Academic Press.
NILSSON, S. G. 1984. The evolution of nest-site
selection among hole-nesting birds: The im-
portance of nest predation and competition.
Ornis Scandinavica, 15: 167-175.
OSBORNE, P. and OSBORNE, L. 1980. The contri-
bution on nest site characteristics to breeding
success among Black birds (Turdus merula).
Ibis, 122: 512-517.
PELLANTOVÁ, J. 1975. The course of breeding of
the swift (Apus apus Linn.) Zoologické Listy,
24: 249-262.
PENLOUP, A. and MARTI N, J. L. 1995. Consé-
quences de la prédation des nids par le rat noir
sur la distribution du martinet pâle (Apus palli-
dus) dans la Bouches de Bonifacio. Parc natu-
rel régional de Corse, 56: 49-69.
PENLOUP, A., MARTIN, J. L., GORY, G., BRUNSTEIN,
D. and BR ETAGNOLLE, V. 1997. Distribution
and breeding success of pallid swifts, Apus
pallidus, on Mediterranean islands: nest pre-
dation by the roof rat, Rattus rattus, and nest
site quality. Oikos, 80: 78-88.
ROGER , T. and FOS SÉ, A. 2001. Nidifications
arboricole et rupertre du martinet noir Apus
apus en Maine-et-Loire. Crex, 6: 21-29.
SANTAMARÍA, T., CORRALES, L., CAMPOS, F. and
ALONSO, C. 2009. Los Estorninos en la Ciudad
de Ávila. Situación Actual y Problemática. Uni-
versidad Católica de Ávila. Ávila.
SACARRAO, G. F. and SOARES, A. A. 1975. Algu-
mas observações sobre a biología de Passer
hispaniolensis (Temm.) em Portugal. Estudos
sobre a Fauna Portuguesa, 8: 1-20.
TENAZA, R. 1971. Behaviour and nesting success
relative to nest location in Adelie Penguins
(Pygoscelis adeliae). Condor, 73: 81-92.
TRO I T I ÑO, M. Á.,GAR C Í A , M.,CALLE, M. ,
MÍNGUEZ, M. C., TROITIÑO, L. and VELASCO,
I. 2012. Boletín Informativo del Observatorio
Turístico de la Ciudad de Ávila, nº 28. Ayunta-
miento de Ávila. Ávila.
VAN BALEN, J. H., BOOY, C. J. H., VAN FRANEKER,
J. A. and OSIECK, E. R. 1982. Studies on hole-
nesting birds in natural nest sites. Ardea, 70:
1-24.
VIELLIARD, J. 1978. Le Djebel Babor et sa sittelle
Sitta ledanti Vielliard. Alauda, 46: 1-42.
WESOLOWSKI, T. 2002. Anti-predator adaptations
in nesting marsh tits Parus palustris: the role
of nest-site security. Ibis, 144: 593-601.
Received: 22 April 2013
Accepted: 4 June 2013
Editor: Jesús M. Avilés
Ardeola 60(2), 2013, 291-304
CORRALES, L., BAUTISTA, L. M., SANTAMARÍA, T.and MAS, P.
304