ArticlePDF Available

Abstract and Figures

The first Catalan Breeding Bird Atlas (CBBA) published in the early eighties, covered, with a 10x10 km UTM grid, an area of 32,000 km 2 in north-east Spain and stands as a pioneer landmark in bird mapping in the Mediterranean region. Here, we present the context, novelties and main results of the new CBBA conducted 20 years later, between 1999 and 2002. This new atlas has included a series of methodological innovations amongst which we should emphasize the establishment of small scale timed censuses on a sample of 1x1 km UTM squares within each of the original squares of the 10x10 km UTM grid. These small-scale censuses (about 3,200) allowed a consistent sampling of the territory by covering uniformly about 10% of the total study area, and make possible the production of species distribution maps at local scales (e.g. 500 m resolution). These maps were generated using spatial logistic regressions in the framework of niche based modelling approaches. Such models included information from 45 environmental variables ranging from land use and relief to direct human influence and climate, as well as information on the spatial structure of the data collected for each sampled species data to account for spatial autocorrelation. Currently, the CBBA stands amongst the best current quantitative mapping of bird distribution in Europe.We also used data from the 1x1 km UTM squares to develop species-time curves and model the number of species found in each 10x10 km squares per unit of time. The application of these models to the first and the new CBBAs allowed an accurate estimation of changes in bird distribution between the two atlases by taking into account local differences in sampling effort (e.g. time spent on a given square) between the two periods. Overall, the new CBBA provides one of the best large-scale pictures of the changes that occurred during the last 20 years in southern Europe. The results obtained strongly support the hypothesis that strong changes in bird community patterns have occurred in large areas of the country, often associated with changes in land use patterns.
Content may be subject to copyright.
Revista Catalana d’Ornitologia 24:118-137, 2008
The Catalan Breeding Bird Atlas (CBBA):
methodological aspects and ecological
implications
Lluís Brotons, Sergi Herrando, Joan Estrada, Vittorio Pedrocchi &
Jean Louis Martin
The first Catalan Breeding Bird Atlas (CBBA) published in the early eighties, covered, with
a 10x10 km UTM grid, an area of 32,000 km2 in north-east Spain and stands as a pioneer
landmark in bird mapping in the Mediterranean region. Here, we present the context,
novelties and main results of the new CBBA conducted 20 years later, between 1999 and
2002. This new atlas has included a series of methodological innovations amongst which we
should emphasize the establishment of small scale timed censuses on a sample of 1x1 km
UTM squares within each of the original squares of the 10x10 km UTM grid. These small-
scale censuses (about 3,200) allowed a consistent sampling of the territory by covering
uniformly about 10% of the total study area, and make possible the production of species
distribution maps at local scales (e.g. 500 m resolution). These maps were generated using
spatial logistic regressions in the framework of niche based modelling approaches. Such models
included information from 45 environmental variables ranging from land use and relief to
direct human influence and climate, as well as information on the spatial structure of the
data collected for each sampled species data to account for spatial autocorrelation. Currently,
the CBBA stands amongst the best current quantitative mapping of bird distribution in
Europe.We also used data from the 1x1 km UTM squares to develop species-time curves and
model the number of species found in each 10x10 km squares per unit of time. The application
of these models to the first and the new CBBAs allowed an accurate estimation of changes in
bird distribution between the two atlases by taking into account local differences in sampling
effort (e.g. time spent on a given square) between the two periods. Overall, the new CBBA
provides one of the best large-scale pictures of the changes that occurred during the last 20
years in southern Europe. The results obtained strongly support the hypothesis that strong
changes in bird community patterns have occurred in large areas of the country, often
associated with changes in land use patterns.
Key words: bird distribution mapping, habitat suitability modelling, distribution changes,
monitoring programme.
Lluís Brotons*, Sergi Herrando, Joan Estrada & Vittorio Pedrocchi, Catalan Ornithological
Institute, Museu de Ciències Naturals de la Ciutadella, Passeig Picasso s/n, 08003 Barcelona,
Catalonia, Spain.
Lluís Brotons* & Jean Louis Martin, Centre d’Ecologie Fonctionelle et Evolutive-CNRS, 1919
Route de Mende, 34293 Montpellier Cedex, France.
Lluís Brotons* (present address), Centre Tecnològic Forestal de Catalunya, Àrea de
Biodiversitat, Ctra. Sant Llorenç, km. 2, 25280 Solsona, Catalonia, Spain.
* Corresponding author: lluis.brotons@ctfc.cat
In 1984 the Atlas of Breeding Birds of Catalo-
nia and Andorra was published (Muntaner et
al. 1984). This publication was one of the first
extensive atlases to be completed in Spain and
Southern Europe. This atlas allowed, for the first
time, the accurate mapping of the of the 214
breeding bird species observed in Catalonia on
a 10x10 square grid in the Universal Transversal
Mercator (UTM) geographic projection. The
data used to map all the species were collected
The Catalan Breeding Bird Atlas
119
in the period 1975-1983, although most of the
sampling effort was concentrated during the
years 1981-1982. The development of the atlas
project involved 129 people.
As species distribution is unlikely to remain
constant for all species through time, the
Catalan Ornithological Institute put forward a
proposal to conduct a revision of the first breed-
ing bird atlas in order to analyse temporal trends
in bird distribution during the last 20 years and
therefore update the knowledge on the coun-
try’s bird fauna. The present number of Catalan
ornithologists and their high identification skills
and commitment, suggested that additional
mapping of spatial variation in species abun-
dance could be a possibility as it had been in-
cluded in other recent atlases (Gibbons et al.
1993, Schmid et al. 1998). Indeed, the collec-
tion of objective data on the spatial patterns of
bird species would advance knowledge of their
ecology and therefore help in defining priori-
ties for biodiversity conservation strategies.
The present Catalan Breeding Bird Atlas
includes information on the distribution of spe-
cies for which breeding evidence for the 1999-
2002 period has been recorded in a spatial grid
of 10x10 km UTM squares. The use of the same
grid allows a direct spatial comparison with the
previous breeding bird atlas of Catalonia and
therefore allows the identification of new and
extinct species as well as temporal trends in the
spatial distribution of the rest. Maps of relative
abundance for most species have been included
in this atlas. These maps have been developed
through information gathered in timed censuses
conducted in a sampling of 1x1 km UTM
squares throughout the study area. This infor-
mation has been used together with geographic
information on main environmental variables
in different areas to allow the estimation of prob-
ability of occurrence maps for the species for
the whole of Catalonia. This analysis has also
allowed the gathering of preliminary but quan-
titative and comparable information about the
main environmental variables that may be im-
portant in explaining the distribution of Catalan
breeding species. This information may also be-
come a useful tool in the design of conserva-
tion policies. In the forthcoming sections, we
will briefly describe the information included in
the new Catalan Breeding Bird Atlas and pro-
vide information on the methodology employed.
Bird atlas data provide the researcher with
a powerful tool to broadly identify changes and
state hypothesis more likely to be behind the
detected changes in bird distribution (Donald
& Fuller 1998). In this paper, we aim to sum-
marise the main patterns of distributional
change in bird species between the two Catalan
breeding bird atlases, which are 20 years apart.
We go on to discuss these changes in light of
different hypothesis acting as drivers of change
in bird populations. Although many factors act
in conjunction to affect bird populations, land
use changes are expected to be especially im-
portant at the temporal and spatial scales used
by this atlas. Therefore, we concentrate our dis-
cussion on the main land use patterns and their
changes as major drivers of species change
(Tucker & Evans 1997).
Methodology
General organisation of the Atlas
In 1997, to coincide with the 20th anniversary
of the Atlas of the Breeding Birds of Catalonia
and Andorra 1975-1983, the ICO contacted a
number of ornithologists and organisations to
Figure 1. Study area and UTM grid used for plotting
data gathered for this atlas. Square size is 10x10 km.
Àrea d’estudi i reticle UTM emprat per recopilar les
dades d’aquest atles. Les dimensions del quadrat són
10x10 km.
L. Brotons et al.
120
Revista Catalana d’Ornitologia 24 (2008)
produce a new, up-to-date atlas of the breeding
birds of Catalonia. It was decided to use the
same 10x10 UTM Universal Transverse Mer-
cator grid as was employed in the previous atlas
to allow for direct spatial comparisons between
species (Figure 1). Catalonia corresponds to
zone 31T. A total of 386 10x10 UTM squares
were surveyed for this atlas. Unlike in the first
atlas, which included Andorra and areas with-
in Aragon and France in cross-border squares
(Muntaner et al. 1984), in the present atlas the
study area was strictly limited to the adminis-
trative territory of Catalonia. Hence, in those
border squares the surveyed area was restricted
to the Catalan portion of the square.
Criteria for species inclusion
and taxonomy
Regardless of their origins (established or other-
wise), species for which some breeding evi-
dence was gathered during the sampling peri-
od (1999-2002) have been included in this
atlas. Non-established (mostly exotic) species
have also been included as their populations
may stabilise in Catalonia and data regarding
these colonisation processes may prove useful
in the future.
In the main part of the atlas all breeding
species and the exotic species for which there is
evidence of breeding and continuous presence
during the period 1999-2002 are considered.
Exotic species showing occasional evidence of
breeding, not regularly observed within the sam-
pling period have been included in a Comple-
mentary Species section that also covers those
species that only bred during the period between
the two atlases (1984-1998) and the summer
species that did not show any evidence of breed-
ing within the period 1984-2002.
Sampling methods
The sampling methodology aimed to detect ei-
ther by sight or by sound the maximum evidence
of breeding for all the species found in each
10x10 UTM square. The assignment of the
10x10 UTM squares was aimed at guaranteeing
that the territory was covered as thoroughly as
possible. Volunteers were asked which areas they
preferred to survey and were then given a series
of 10x10 UTM squares still to be assigned to
choose from. Those not showing a geographical
preference were asked to survey one of the
squares with a low density of ornithologists. If a
volunteer could only survey a square that had
already been assigned, then he/she was still ac-
cepted as a participant in the project. For this
reason, some 10x10 UTM squares were surveyed
by more than one person.
Evidence of breeding
The assignation of the different categories of
evidence of breeding follows the recommenda-
tions of the European Ornithological Atlas
Committee (EOAC) that has been working
from 1992 as part of the European Bird Census
Council (EBCC) (Hagemeijer & Blair 1997).
Given that some observers tended to place in
the ‘possible breeding species’ category some
individuals surely not breeding in the area, it
was decided to add two new categories in order
to correctly classify these observations: ‘non-
breeding summer visitor’ and ‘migrant’. These
additions allowed new data to be collected
whilst avoiding confusing breeding species with
those birds that would not breed in that partic-
ular square. Thus, the ‘non-breeding summer
visitor’ category allows the real summer distri-
bution of many species to be defined.
Timed censuses in UTM 1x1 squares
The use of timed censuses as a mean of produc-
ing abundance maps is one of the most signifi-
cant features of the present atlas. Besides the
generic sampling conducted for each UTM
10x10 square, a set of sample 1x1 km squares
was established inside the former units: this
approach is similar to that used, for example, in
the British and Swiss ornithological atlases (Gib-
bons et al. 1993, Schmid et al. 1998). These 1x1
km units were established by dividing each of
the UTM 10x10 squares into 100 UTM 1x1
squares. In order to reproduce as accurately as
possible the environmental heterogeneity of
each UTM 10x10 square, a number of 10 non-
adjacent 1x1 squares had to be selected by the
responsible observer in such a way that togeth-
er they were proportionally representative of the
habitats present inside the UTM 10x10 square.
Those squares that were insufficiently covered
and therefore had to be surveyed semi-profes-
The Catalan Breeding Bird Atlas
121
sionally, the minimum coverage was lowered to
five 1x1 squares.
For nocturnal species, a sub-sample of five
out of the 10 previously selected UTM 1x1
squares were selected to be surveyed at night.
This figure was lowered to two UTM 1x1 squares
when surveyed semi-professionally. For both di-
urnal and nocturnal censuses, two 1-hour sur-
veys were conducted for each one of the selected
UTM 1x1 squares during which every square was
entirely surveyed and every species recorded. The
number of individuals was not noted.
Sampling period
As a general rule, the gathering of breeding ev-
idence was restricted to the period March-July
inclusive; for ‘confirmed evidence of breeding’,
data obtained in dates previous or posterior to
the prescribed period were accepted, especially
for those species with very long reproductive
seasons. The surveying period for nocturnal rap-
tors was brought backward to February so that
the territorial songs of species that principally
call in winter could be detected (for example,
Eagle Owl, Long-eared Owl and Tawny Owl).
If possible, surveys in the UTM 1x1 sample
squares had to be conducted between sunrise and
11 am, and between 6 pm and sunset. Timeta-
bles and dates were approximate and could be
adapted to local conditions provided they coin-
cided with peak bird activity. The first survey was
conducted in March-April to detect resident spe-
cies and the second in May-June to detect sum-
mer visitors. For nocturnal species, the first sur-
vey was conducted in February for the reasons
described above, and the second survey was con-
ducted in May in order to detect the remaining
nocturnal raptors and other nocturnal or crep-
uscular species such as the Common and Red-
necked Nightjars and Stone Curlew. Nocturnal
surveys were conducted at any time between sun-
set and sunrise, although the majority were car-
ried out during the first hours of darkness.
Data Analysis
Distribution maps
Data collected during the field observations and
surveys were marked on maps based on the
UTM 10x10 km grid, showing the ‘breeding
evidence’ distribution in the period 1999-2002.
Breeding evidence on the maps is classed as ‘pos-
sible’, ‘probable’, ‘confirmed’ and ‘summer non-
breeding visitor’; migrants are not represented
on the maps, although all relevant information
has been stored in the atlas database.
In the background the previously known
distribution of the species (Muntaner et al. 1984)
is given allowing changes to be seen at a glance.
Data from the previous atlas (including infor-
mation on rare and/or sensitive species whose
real distributions had been hidden in the origi-
nal maps) were directly provided by its coordi-
nator and then added to the database of the
current atlas. Those data were also adapted into
a single three categories breeding code (‘possi-
ble’, ‘probable’ and ‘confirmed’, so that only the
best possible evidence of breeding per species
and per sheet was considered.
Changes in species’ distribution:
variations in the occupation of UTM
10x10 squares between atlases
Animal populations are not always stable and
changes may affect their spatial distribution.
Since one of the main goals of an atlas is to ac-
curately identify these hypothetical changes, it
was essential to differentiate between real
changes in distribution and simple variations in
the sampling effort between atlases. This is par-
ticularly relevant for bird atlases because these
large-scale projects usually incorporate data
from different sources, with subsequent tempo-
ral and spatial variations in the data-collection
effort. For this reason, several bird atlases have
tried to increase the reliability of their estimates
of spatial (and temporal) variations in distribu-
tion by placing a great deal of emphasis on the
standardisation of applied field methods (i.e.
Schmid et al. 1998). Despite possible biases, such
as an heterogeneous spatial effort within atlas
data affecting the reliability of the comparison
with future atlases, the usefulness of these anal-
yses (Donald & Fuller 1998) makes it necessary
to use methodologies that overcome limitations
in data collection.
For a particular species, the quantification
of changes in its distribution can be estimated
by way of an analysis of the changes in occupied
UTM 10x10 squares. However, a temporal vari-
L. Brotons et al.
122
Revista Catalana d’Ornitologia 24 (2008)
ation in sampling effort may create a significant
bias in the estimates of distribution changes.
While only 129 observers took part in data col-
lection during the first atlas (Muntaner et al.
1984), the number climbed up to 497 during
the second period. Unfortunately, both the data
collected for the first atlas and, to some degree,
the data from the present atlas were not (com-
pletely) standardised in relation to the sampling
effort applied to each UTM 10x10 square. De-
spite the timed censuses conducted in the UTM
10x10 squares, a large portion of the data ob-
tained for the new atlas comes from observa-
tions gathered in a non-standardised way from
a time effort point of view. Thus, changes in a
single species distribution may be a consequence
of the differences in sampling effort between the
two data sets or more realistically due to the
general higher sampling effort in the second at-
las. In these circumstances, the perceived ex-
pansion of a particular species may be related
both to biological causes and/or to more thor-
ough fieldwork during the second atlas.
Changes in a species’ distribution from one
atlas to the other have been analysed using es-
timates of the change in the total number of
UTM 10x10 squares occupied by the species,
corrected for using information gathered from
species accumulation curves. This approach
used timed censuses conducted on the sample
of UTM 1x1 squares located within each of the
UTM 10x10 squares. Species’ accumulation
Figure 2. Diagram of the process for obtaining the effective sampling time for each UTM 10x10 square from
curves of richness-effort controlling the effect of environmental heterogeneity. 1) Production of the basal
curve richness-sampling effort (hours) from the information gathered in the timed censuses conducted in
UTM 1x1 squares. 2) The first step allows us to obtain a preliminary relationship between effective sampling
time and the number of detected species. 3) Generation of a second regression equation that allows us to
control the effect of environmental heterogeneity in the UTM 10x10 squares by means of the number of
species detected in five UTM 1x1 squares. 4) Application of the generated equation of richness values per
UTM 10x10 square in order to obtain the effective sampling time (in hours)
Diagrama del procés utilitzat per a l’obtenció del temps de mostratge efectiu realitzat en cada quadrat UTM
10x10 a partir de corbes riquesa-esforç, controlant l’efecte de l’heterogeneïtat ambiental. 1) Producció de la
corba base entre riquesa i esforç de mostratge (en hores) a partir de la informació recollida en els censos de
temps controlat portats a terme en quadrats UTM 1x1. 2) Aquest primer pas ens permet obtenir una associació
preliminar entre el temps efectiu de mostratge i el nombre d’espècies detectades. 3) Generació d’un segon
model de regressió que permeti controlar l’efecte de l’heterogeneïtat ambiental a cada UTM 10x10 a través
del nombre d’espècies detectades en 5 quadrats UTM 1x1. 4) Aplicació d’aquesta darrera equació als valors
de riquesa obtinguts en els quadrats UTM 10x10 per tal d’obtenir el temps de mostratge efectiu (en hores).
The Catalan Breeding Bird Atlas
123
curves were created from these data and by re-
versing the process the effective surveying time
for particular species richness was estimated
(Figure 2). The independent evaluation of the
effects derived from the sampling effort (effec-
tive census time) and from the period (first vs.
second atlas) on the presence of each species
was carried out by means of a logistic regression
for repeated measures (Genmod SAS software
module). The number of UTM 10x10 squares
in which every species was detected, either in
the first atlas (1 vs. 0), in the present one (0 vs.
1) or in both (1 vs. 1), was used.
Once survey effort was estimated, these sta-
tistics allowed us to determine the significance
of the change in the number of squares occu-
pied by every species in the period between at-
lases. It also enabled us to obtain corrected es-
timates of change, expressed in the results as
the percentage increase or decrease in the
number of UTM 10x10 squares occupied dur-
ing the period between atlases. Importantly, this
analysis only indicates changes in the number
of squares, which means a quantitative varia-
tion in the size of the distribution area of the
species. If this variation is not significant it does
not necessarily imply that the species’ distribu-
tion has remained the same, since the number
of occupied squares may be unchanged even if
their location may have varied. Thus, not in-
frequently the disappearance from one area and
appearance at a similar scale in another area
implies an overall stability in the number of
squares but also important qualitative changes
in distribution.
On average, the effective sampling time (es-
timated form species accumulation curves) per
UTM 10x10 square was 66 hours (standard de-
viation = 44 hours) for the present atlas and
43 hours (standard deviation = 43 hours) for
the previous atlas. Thus, from a general point
of view, any UTM 10x10 square was surveyed
for 23 more hours during work on this atlas than
work on the previous one. The spatial pattern
of change in the sampling effort from one atlas
to another is far from homogeneous (Figure 3).
Patterns of temporal change
across species groups
The main objective of an atlas is to provide in-
formation on the distribution of the bird spe-
cies present in a given region as well as the eco-
Figure 3. Spatial representation of the sampling effort invested in each UTM 10x10 square in Catalonia for
both breeding bird atlases. The unit employed is the estimated effective number of hours invested in bird
surveying, not necessarily the total time spent by observers in every square.
Representació espacial de l’esforç de mostratge invertit en cada quadrat UTM 10x10 de Catalunya en els dos
atles dels ocells nidificants. La unitat emprada és el nombre estimat d’hores invertides cercant ocells, les
quals no corresponen necessàriament al temps total que els observadors han passat en el quadrat.
L. Brotons et al.
124
Revista Catalana d’Ornitologia 24 (2008)
logical factors determining such distributions
(Gibbons et al. 1993, Schmid et al. 1998). How-
ever, given that a historic atlas exists in Catalo-
nia conducted 20 years ago, an equally impor-
tant objective has been to quantify the temporal
changes in bird distribution occurred between
the two atlases (e.g. changes in the number of
10x10 UTM squares occupied by a particular
species) and try to provide information on the
likely causes behind such changes.
In order to identify hypotheses behind dis-
tribution changes in a Mediterranean context,
we conducted analyses on the spatial patterns
of distribution changes in Catalonia in differ-
ent bird groups according to their broad habi-
tat selection patterns and discuss their associa-
tions with land use variables generated from
satellite imagery. We then (1), analyse the over-
all trends in species changes according to their
habitat preferences, and (2) map the overall
patterns in spatial trends of such changes among
some of the species groups showing significant
variations in distribution. Due to the pioneer-
ing initiative of the first Catalan atlas, the re-
sults of this new Atlas stand amongst the first
large-scale temporal comparison of main ten-
dencies in bird distribution within the Mediter-
ranean region.
To analyse the patterns of distribution
changes in the Catalan breeding bird fauna, we
first classified the species considered in the
present atlas according to their main habitat used
during the breeding season and their migratory
status. According to their main habitat selec-
tion as derived from the literature (Muntaner et
al. 1983, Cramp & Perrins 1994), species were
divided in seven categories: forests, shrublands,
wetlands and rivers (including coastal and in-
land wetlands), farmland, steppes (as particular
extensive dry farmland habitats), alpine habitats
located in the Pyrenees (including both
subalpine forests and alpine non-forested areas
above the tree line), and a final group of other
species, which selects habitats at larger spatial
scales, including most raptors and aerial species
weakly associated to terrestrial habitats.
Migratory status was assigned according to
whether the species winters south of the Sahara
(long distance migrant), makes short to medium
distance movements during winter (short dis-
tance migrant) or always remains in or around
the area where it breeds (resident).
Changes in species distributions between the
two atlases were calculated after taking into
account the sampling effort per 10 km square
and comparing the number of squares occupied
in the two different periods (see section Changes
in species distribution: variations in occupation of
UTM 10x10 squares between atlases). For analy-
ses involving several species, we used a distri-
bution change index = (range2nd atlas –
range1statlas)/((range2nd atlas+ range1st atlas)/
2). This index allowed us to measure distribu-
tional increases and declines symmetrically
around 0, with maximum decline in the case of
extinction as -2, and maximum increase in case
of colonisation as +2 (BohningGaese & Bauer
1996). Changes in distribution were also
weighted because extinctions and colonisations
are not equally reliable in rare and abundant
species. Weights were calculated as the loga-
rithm of the mean number of squares occupied
in the two atlas (BohningGaese & Bauer 1996).
Species consistently using two different habi-
tats were included in the two habitat categories
and we assigned to each category half of the
corresponding weight in statistical analyses.
It is important to underline the fact that
given a low sample size, it was not possible to
calculate the distribution change index for some
species. Therefore, species lists from the present
chapter may appear shorter than in other sec-
tions of this Atlas.
Abundance index maps: modelling
habitat quality
Mapping spatial bird distribution accurately and
consistently is not a straightforward task. First-
ly, detecting a species requires considerable ef-
fort, which raises logistical limitations when a
large area (as usually happens in bird atlases) is
under study. Therefore, a trade-off is likely to
appear between the spatial coverage of the at-
las and the effort put in to cover it. In general,
this compromise has led to bird atlases giving
complete coverage of the study region but de-
scribing occurrence (verified presence vs. in-
ferred absence) at a relatively coarse resolution
of 10x10 km (e.g. Martí & del Moral 2003) or
50x50 km (e.g. Hagemeijer & Blair 1997).
Niche-based models are a simple way of ob-
taining directly a species-presence probability
value for each point in space from available
The Catalan Breeding Bird Atlas
125
environmental information. Niche-based mod-
els make no use of additional species data other
than those that collected during the species sam-
pling; the process is based on the fact that spe-
cies use a restricted combination of environmen-
tal parameters that allow us to model their
ecological niches (Guisan & Zimmermann
2000). These niche-based models are based on
modelling species’ response to a set of environ-
mental variables and on the subsequent predic-
tion of their presence in unsampled areas using
values for environmental variables (Figure 4).
This approach has been widely applied to small
spatial scales, however until now, niche-based
models have only had limited applications in
the mapping of species’ distribution at large
scales (Osborne & Tigar 1992, Parker 1999,
Suarez-Seoane et al. 2002).
The modelling approach used in
the Catalan Breeding Birds Atlas
For this atlas, we estimated the probability of
occurrence of a species by applying niche-based
models to the data collected during 1x1 UTM
square censuses (3,077 squares in the case of
diurnal birds and 1,204 for nocturnal birds). The
models developed allowed us to estimate each
species’ response to the selected environmen-
tal variables and thereby obtain the predicted
probability of occurrence for each species for a
particular combination of those environmental
variables.
In addition to species-presence data, the
availability of good quality environmental data
is essential for implementing niche-based mod-
els. Fortunately, over recent years a great deal
of work has been done by different institutions
such as the Catalan Government, Centre for
Ecological Research and Forestry Applications
(CREAF: Centre de Recerca Ecològica i
Aplicacions Forestals) and the Catalan Carto-
graphic Institute (ICC: Institut Cartogràfic de
Catalunya) on building up high-quality digital
databases of environmental information. We
used these databases in order to generate perti-
nent environmental variables that were used as
predictors of species’ occurrence. These vari-
ables were selected to incorporate those factors
known a priori to determine bird distribution at
different spatial scales (Wiens 1989). The dif-
ferent groups of variables (a total number of 45)
included in the analysis included habitat/land
use variables, climate and relief variables and
human influence (see Brotons et al. 2004a for
more information on the variables included).
A way to account for the unknown variables
that may be relevant to species distribution is
to use measures of spatial autocorrelation. Spa-
tial autocorrelation depends on the fact that the
presence of a species in a given area is not usu-
ally independent of whether the species occurs
or not in surrounding areas (Vaughan &
Ormerod 2003). In these cases, the information
on the degree of spatial contagion in species
occurrence may be used as a surrogate for envi-
ronmental information missing from the model.
Examples of such missing environmental vari-
ables are historical factors. For instance, recently
lost areas of a species’ distribution that are still
suitable for the species may be difficult to iden-
tify if they are environmentally similar to oth-
ers where the species occurs (Brotons et al.
2004a). In this case, given that the species would
be completely absent from such areas, the use
of contagion information may help us include
information about the species’ current absence
in the model. In the Catalan atlas, we used three
different contagion variables that are auto-
covariables as defined in Augustin et al. (1996).
These contagion variables were calculated for
all species and represented an estimate of the
mean occurrence of the species around each 1x1
UTM square (Figure 4) including a different
number of neighbours in the calculation.
Statistical modelling approach
Different methods based on the estimation of
species’ ecological niches exist for generating
species habitat-suitability maps. A major differ-
ence between all these methods is the quality
of data needed. Preliminary studies with differ-
ent methodologies suggested that in our data
set the additional information provided by ab-
sence data tended to increase the predictive
accuracy of niche-based models (Brotons et al.
2004a). Therefore, we selected a method that
uses both data indicating the presence of spe-
cies in 1x1 UTM squares and data indicating
their absence. We used generalised linear mod-
els (GLM) to estimate the probability of a spe-
cies occurring at a given locality and thus ob-
tain estimates of its relative abundance. GLMs
have proven to be robust in a multitude of pre-
vious modelling exercises and their predictive
L. Brotons et al.
126
Revista Catalana d’Ornitologia 24 (2008)
accuracy is generally comparable to that of the
much more complex methods mentioned above
(Osborne & Tigar 1992, Fielding & Haworth
1995).
In this atlas, GLM were used as predictive
rather than explanatory tools; hence the ac-
curacy of model projections is more important
than the significance of a particular ecological
term (Legendre & Legendre 1998). The main
consequence of this approach is that the best
environmental models did not always lead to
easily interpretable ecological hypotheses.
Niche-based models need a minimum set of
data to correctly estimate response curves. The
selection of minimum sample sizes for devel-
oping these models is not easy since the
number of observations may interact with sam-
ple size to determine model accuracy. In this
analysis, we have included niche-based mod-
els for species detected at least in 10 1x1 UTM
atlas squares.
Evaluation of niche-based models
We used a cross-validation procedure to evalu-
ate the accuracy of model predictions (Guisan
& Zimmermann 2000). This procedure consist-
ed of dividing the data (1x1 UTM square sur-
veys) into two different sets by randomly assign-
ing 70% of occurrence values for each species
to a calibration data-set and the remaining 30%
of occurrences to an evaluation data-set. The
calibration dataset was used to develop the
niche-based model. The evaluation process con-
sisted of measuring quantitatively to what de-
gree predictions from the models fitted the in-
dependent observations that were not used for
the development of the model.
A powerful approach is to assess model suc-
cess across a range of dichotomies from differ-
ent cut-off points using the receiver operating
characteristic (ROC) plots. The ROC plot is
based on a series of misclassification matrices
computed for a range of cut-offs from 0 to 1. It
Figure 4. Schematic representation of the methodology used in this atlas to generate abundance index maps
(see text for further explanation).
Representació esquemàtica de la metodologia utilitzada en aquest atles per generar els mapes d’índex
d’abundància (vegeu text per a més explicacions).
Environmental
data
Samples
Species and
Environmental
data
1
×
1 UTM sampling
Statistical modeling
Projection to the
whole Catalonia
Final generation
of maps (filtering)
Evaluation
Contagion
123
4
5
The Catalan Breeding Bird Atlas
127
then plots on the y-axis the true positive frac-
tion against the false positive fraction from the
same misclassification matrix (Fielding & Bell
1997). The area under the ROC curve (AUC)
is a convenient measure of overall accuracy, and
commonly varies between 0.5 (for chance per-
formance) and 1 (perfect discrimination). We
obtained the AUC and its standard error with a
customised function in the S-Plus software.
Good models are those that reliably predict
species presence in unknown sites. The AUC
can be interpreted here as a measure that indi-
cates the percentage of classification errors that
the model is likely to make once the predictions
are compared to the observed occurrences of
the evaluation data set. According to previously
defined ratings we classified model performance
as poor (0.5-0.7), fair (0.7-0-8), good (0.8-0.9)
and excellent (0.9-1). The AUC was used as
the main indicator of model performance and
only models with an AUC higher than 0.7 were
included in this atlas.
Final generation of abundance index maps
Once they had been calibrated and evaluated,
we proceeded to extrapolate our niche-based
models to the unsampled 1x1 UTM squares for
which environmental information was availa-
ble. We projected species-specific niche-based
models (logistic formulas) on the whole of Cata-
lonia by means of a Geographical Information
System (G.I.S., Arc-View 3.2). The grid maps
generated initially had a cell size of 1x1 km (the
same resolution as the sampling grid). A new
version of the map was then created by reclas-
sifying each pixel grid map to a resolution of
500 m and by applying smoothing that reclassi-
fied each value by assigning the mean value of
their adjacent neighbours. This smoothing step
assumes that the probability of occurrence of a
species varies linearly between two points, a
seemingly valid assumption given the spatial
scale involved.
Many authors consider maps generated by
niche-based models to be equivalent to poten-
tial distribution maps and not to real distribu-
tion maps (Guisan & Zimmermann 2000). As
previously commented, due to unrecorded en-
vironmental variables, historical factors or habi-
tat fragmentation (among other factors), niche-
based models may predict species presence
where a given species is certainly known to be
absent. Although our models predicted very
accurately the occurrence of most species, we
decided to include an additional step in the fi-
nal map production to ensure that the final rela-
tive abundance maps corresponded as accu-
rately as possible to real rather than potential
distribution maps. This additional step consisted
of the application of a filter that would delete
zones where the model predicted a species as
present outside its known 10x10 UTM distri-
bution range (i.e. taking into consideration the
squares were the species was known to be
present and all neighbouring squares in direct
contact). This step was taken in order to allow
for the inclusion in the abundance maps of ar-
eas of low relative abundance at the edges of
species’ distributions where the non-recording
of the species in the 10x10 UTM sampling may
have occurred for stochastic reasons. All rela-
tive abundance maps were filtered and we made
sure that the hypothetical occurrence of a spe-
cies did not appear outside its known distribu-
tion range.
The final maps at a resolution of 500 m cor-
respond to the final abundance index maps and
measure the estimated probability of occurrence
(from 0 to 1) in each 500x500 m unit of the
territory within the known distribution of the
species in question. Such a probability of oc-
currence is assumed a surrogate for species abun-
dance and as such is interpreted in the present
atlas as a reliable estimator of a species’ relative
abundance in each square. This assumption
seems justified since abundance patterns at
small spatial scales are often strongly related to
the probability of the occurrence of the species
during sampling.
Ecological requirements: graphs and
tables
The main objective of any bird atlas is to un-
ravel the distribution of birds in the area under
study. Species’ ecological requirements, which
vary in strictness from one species to another,
are described in terms of ranges in temperature,
altitude, slope gradients, forest cover, salinity
and so forth. In the texts for each species vari-
ous experts discuss some of the species’ main
ecological requirements; also included in this
section (when available) is information regard-
ing the selection of altitudinal ranges (graph of
L. Brotons et al.
128
Revista Catalana d’Ornitologia 24 (2008)
altitude) and the use and selection of different
landscape types measured at 1x1 km scale
(graph of landscape types).
In a mountainous country such as Catalo-
nia, height above sea level is an essential factor
in a species’ distribution, since it strongly influ-
ences climate, vegetation and human activity.
The other significant group of environmental
characteristics is related to habitats and how
they combine to form landscape types. There-
fore, both the species’ main habitats and the
selections they make, as well as abundances in
the principal landscape types, are shown when-
ever possible.
One of the most frequently used parameters
in the study of a species’ ecological requisites is
its abundance or density in different environ-
mental conditions. The launching of the Com-
mon Bird Survey in Catalonia (SOCC) in 2002
provided useful data for estimating bird densi-
ties in the main landscape types of the country.
Data from transects were converted into den-
sity estimates and are shown in a table as the
mean, maximum and minimum densities for 16
pre-defined landscape types: alpine and
subalpine pastures and rocky outcrops,
subalpine forests, montane thickets and mead-
ows, montane pine forests, montane deciduous
forests, montane mosaics, sub-Mediterranean
and continental Mediterranean landscape mo-
saics, humid Mediterranean landscape mosaics,
dry Mediterranean landscape mosaics, Mediter-
ranean farmlands, Holm and Cork Oak forests,
Mediterranean pine forests, Mediterranean
shrublands, irrigated farmlands in the Ebro de-
pression, steppes in the Ebro depression, and
built-up areas.
Population estimates
One of the main objectives of the present atlas
was to generate reliable population estimates for
the different bird species breeding in Catalonia
and thus take an important step towards know-
ing the conservation status of these species. Gen-
erally, due to their scarcity some species have
been the target of greater conservation efforts
and research institutes, governments and indi-
Figure 5. Examples of species accounts, the Red-legged Partridge Alectoris rufa. Main text in Catalan with
English summary.
Exemple de fitxa d’espècie, la perdiu roja Alectoris rufa. El text principal està en català i el resum en anglès.
The Catalan Breeding Bird Atlas
129
viduals have devoted an important amount of
time to try to estimate their populations. This is
the case of a number of raptors, wetland and
steppe birds, all of which are birds of special con-
servation concern in our country. The total pop-
ulations of these species are often known very
accurately and are re-evaluated regularly, often
on a yearly basis. This atlas is an attempt to uni-
fy all the work carried out in these specific mon-
itoring schemes by different research groups.
In spite of this, specific procedures to evalu-
ate population size only exist for a few Catalan
breeding species. Thus, new methodologies have
had to be designed to address this issue and ei-
ther consists of data collected by atlas contribu-
tors or the combining of data from other moni-
toring projects such as the Common Bird Survey
in Catalonia (SOCC). One of this new meth-
odologies used was the the Atlas-SOCC model.
The initial hypothesis under this methodologi-
cal procedure is that the probability or frequency
of occurrence of a species in an area is related
to its absolute abundance (see Herrando et al.
2008, this proceedings). In cases in which the
Atlas-SOCC could not be estimated and accu-
rate population estimates based on direct spe-
cies censuses were not available, population es-
timates came from atlas contributors’ field data.
Atlas contributors were also asked to esti-
mate the population of each bird species in each
10x10 UTM surveyed. This methodology tries
to capture quantitatively contributors’ impres-
sions of bird numbers, since, despite not having
to count bird numbers, all spent many hours
covering the 10x10 square in search of birds. It
should be remembered that this method is very
subjective and different observers may give dis-
tinct values for the same species and square, and
it is not necessarily true that overestimates and
underestimates compensate one another.
On average, Catalan 10x10 km squares have
estimates for 81% of bird species, and 313
squares have more than 75% of their species
with at least one population estimate. In squares
without estimates for a particular species, we
assessed values using the mean of the number
of pairs of that species in all Catalan squares.
Then, the Catalan population was assessed fol-
lowing the methodology used in the European
atlas (Hagemeijer & Blair 1997) based on geo-
metric means as the most reliable estimator of
each square’s population.
General considerations on population
estimates
Population estimates shown in the atlas come
from three different sources depending on the
species: from specific estimates from particular
monitoring projects, from the Atlas-SOCC
model or from estimates made on a basis of at-
las contributors’ field data. We believe that a
hierarchy in the accuracy of the estimates ex-
ists, with estimates being generally better from
the Atlas-SOCC model and poorer from the
atlas field estimates. In the species texts the ac-
curacy of each estimate is only given for the
Atlas-SOCC model estimates (acceptable, good,
very good), whereas every atlas field estimate is
taken to be acceptable. In the group of specific
estimates by experts, accuracy varies according
to the level of current knowledge of the spe-
cies; this will be much greater in the case of in-
tensively monitored species such as the Bonel-
li’s Eagle than in many poorly monitored
passerine species. Nevertheless, in many cases
expert specific estimates have the substantial
advantage of being focussed on scarce and lo-
calised species, whose total populations can of-
ten be estimated very accurately.
Another aspect to bear in mind is the differ-
ent interpretation of the accuracy of estimates
(closeness to the real value) and their precision
(estimate range); this largely depends on
methodology. The atlas field estimates are usu-
ally much more precise than the other two esti-
mates, although they are much less accurate.
Species conservation status
Until recently the conservation status of a giv-
en species was often decided by using partial cri-
teria based on particular aspects of a species’ bi-
ology or its population status, rather than a
combination of all this information. On some
occasions, the conservation status of a species
may have been determined by purely non-sci-
entific or social criteria that have more to do
with the attractive distinctiveness of the species
than its scarcity. Thus, there has been a marked
tendency to protect species belonging to bird
groups such as the larger raptors or wetland birds,
leaving other groups such as small passerines un-
protected even in cases where clear and urgent
conservation action is required to ensure short-
term survival. This bias in the assignment of con-
L. Brotons et al.
130
Revista Catalana d’Ornitologia 24 (2008)
servation status is widespread and is not unique
to Catalonia, where the Catalan government’s
Law 3/1988 on wildlife protection and its revi-
sion Law 22/2003 tend to give greater conserva-
tion priority to spectacular species such as Grey
Heron, White Stork, Lammergeier, Griffon Vul-
ture, Golden Eagle, Peregrine and Eagle Owl.
However, all of these species except for the Lam-
mergeier have at present relatively secure popu-
lations in Catalonia and thus should be of low
conservation concern. On the other hand, de-
clining species of critical conservation concern
such as Dupont’s Lark, Bearded Tit, Lesser Grey
Shrike and Reed Bunting have very small popu-
lations and hence should have a higher conser-
vation status.
The data in the atlas provide a standardised
and objective assessment of the conservation
status of the bird species that breed in Catalo-
nia. We decided to follow the IUCN (Interna-
tional Union for Conservation of Nature) cri-
teria applied worldwide (UICN 2001), as well
as the regional correctors proposed by
Gardenfors et al. (2001) and other correctors
for species such as raptors whose Catalan den-
sities and population are very small (for the
whole procedure, see Estrada et al. 2004). The
status of each bird species were further discussed
with 43 Catalan ornithologists, all experts in
different groups of birds and with historical vi-
sions of the changes in bird populations that
are occurring in Catalonia, and their conclu-
sions were also compared with data provided
by the authors of each text.
The categories and main criteria considered
for Catalonia are:
Extinct (EX): A species is considered to be
Extinct in Catalonia when there is no doubt
that it once bred regularly in Catalonia and
that it has now disappeared. Likewise, exhaus-
tive surveys in known or prospective habitats
at appropriate times (diurnal, seasonal, an-
nual) throughout its historic range have failed
to record any individual.
Critically Endangered (CR): A species is con-
sidered to be Critically Endangered in Cata-
lonia when it is thought to be facing an ex-
tremely high risk of extinction, either because
of an extremely low population (less than 25
breeding pairs), a population of 25-125 breed-
ing pairs in serious decline, or a larger popu-
lation in drastic decline with no possibility of
spontaneous immigration from nearby popu-
lations reversing the trend.
Endangered (EN): A species is considered to
be Endangered in Catalonia when it is thought
to be facing a very high risk of extinction, ei-
ther because of very low population levels (less
than 125 breeding pairs), a population of 125-
500 breeding pairs in decline, or a larger pop-
ulation in drastic decline with no possibility of
spontaneous immigration from nearby popu-
lations reversing the trend. Those species that
could be considered as Critically Endangered
but whose populations could be reinforced by
the spontaneous immigration from nearby pop-
ulations are also classified as Endangered.
Vulnerable (VU): A species is considered to
be Endangered in Catalonia when it is thought
to be facing a high risk of extinction, either
as a result of low population levels (less than
500 breeding pairs), or by having a higher
population but a large decline, or by having a
very small distribution, with no possibility of
spontaneous immigration from nearby popu-
lations reversing the trend. Those species that
could be considered as Endangered but whose
populations could definitely be reinforced by
the spontaneous immigration from nearby
populations are also classified as Vulnerable.
Near Threatened (NT): A species is consid-
ered to be Near Threatened in Catalonia
when it does not qualify at present for the
Critically Endangered, Endangered or Vulner-
able categories, but is close to doing so at
present or will be in the near future, either
because of its small population, a marked de-
cline or a distribution over-concentrated in
one area. Those species that would be con-
sidered as Vulnerable but whose populations
could certainly be reinforced by the sponta-
neous immigration from nearby populations
are classified as Near Threatened. Finally, rap-
tors whose Catalan populations presently
have low numbers as a result of their position
in trophic chains and their great spatial re-
quirements have been considered only as Near
Threatened if no other threatening factors
also occur.
Least Concern (LC): A species is considered
to be of Least Concern in Catalonia when it
does not qualify for the Critically Endangered,
Endangered or Vulnerable categories or, in
The Catalan Breeding Bird Atlas
131
case of being initially classified as Near
Threatened, if it has numerically important
populations in Catalonia and it is thought
certain to be reinforced by spontaneous im-
migration from nearby populations.
Data Deficient (DD): A species is considered
to be Data Deficient when available informa-
tion does not allow for a direct or indirect as-
sessment of its conservation status and/or risk
of extinction.
Not Evaluated (NE): None of the species that
have been artificially introduced into Cata-
lonia since 1900, nor those species that have
never bred regularly in Catalonia have been
evaluated.
Results and discussion
Patterns of bird distribution and land
use changes
In order to generate consistent hypothesis about
the main processes affecting changes in bird dis-
tribution, consistent patterns across species
should be used (BohningGaese & Bauer 1996).
Determining the causes of changes in distribu-
tion of bird populations is not a trivial task due
to the number of not mutually exclusive mech-
anisms likely to be involved. Different factors
may affect bird populations and eventually lead
to changes in their spatial distribution. These
factors can be of natural origin such as genetic
or ecological factors, but also strong climatic
events such as cold winters or mild summers
may have a strong impact on species demogra-
phy. Major causes behind changes in bird spe-
cies distributions have often a direct or indirect
human related origin. Human activities may a)
directly affect individuals through hunting or
contamination, b) transform habitats via exploi-
tation or species translocations or c) indirectly
contribute to large scale climatic patterns, all
of which may have different impact on bird pop-
ulations (Thomas & Lennon 1999, Gaston et
al. 2003). Recent reviews have identified glo-
bal warming, biological invasions and changes
in land use patterns as major threats to biodi-
Figure 6. Frequency distributions of changes in the number of 10x10 UTM squares occupied by each species
between the two atlases.
Distribucions de freqüència dels canvis en el nombre de quadrats UTM 10x10 ocupats per cada espècie entre
els dos atles.
L. Brotons et al.
132
Revista Catalana d’Ornitologia 24 (2008)
versity at a global scale during the coming years.
These studies point out also that Mediterrane-
an systems may be especially sensitive to these
changes because future projections point to-
wards an especially strong and simultaneous
impact of many of these factors (Tucker & Evans
1997, Sala et al. 2000).
Patterns of change in breeding birds
A remarkable result of the present atlas is the
number of introduced species detected in Cata-
lonia. Up to 10 species were detected, although
only three of them were present in more than
10 UTM squares.
Amongst indigenous species from which
changes in distribution could be estimated us-
ing changes in the number of occupied 10x10
UTM, we suggest that Catalan bird fauna has
suffered remarkable changes during the last 20
years (Figure 6). Changes in distribution did not
show a homogeneous pattern of change, and
rather were dependent on the different habitat
selection of the considered groups (F5, 176=4.04,
p<0.01). While species of shrublands, steppe-
like extensive agricultural areas, forests and spe-
cially wetland habitats showed a clear tendency
to expand their distributions, farmland and al-
pine species showed a marginal tendency to
shrink their distributional range (Figure 7).
Other species (mainly raptors) showed a con-
sistent tendency to increase in distribution area
(Figure 7). We did not detect a significant effect
of the migratory behaviour in the overall spe-
cies distribution change (F2,176=1.08, n.s.), nor
within each habitat category (F10, 188=0.88, n.s.).
Spatial patterns of bird distribution
changes according to habitat use
Analysis of the distributional changes of differ-
ent bird groups in Catalonia between the two
atlases supports the hypothesis that changes
have not been homogeneous throughout the
territory and some areas have suffered more
marked variations than others.
We describe the spatial pattern of distribu-
tion change for some example groups including
species that either have increased their overall
distribution (expanding species group, having
significantly p<0.05 increased the number of
occupied 10x10 UTM squares, or by at least
having a 20% increase) or that have decreased
their overall distribution (contracting species
group, having significantly p<0.05 decreased
the number of occupied 10x10 UTM squares,
or by at least having a 20% decrease). Alterna-
tive hypotheses behind the observed patterns
of change are discussed. We have restricted the
analyses described above to forest and farmland
species groups.
Forest species
Increasing forests species
Forest species that experienced significant in-
creases in range in Catalonia according to the
comparison between atlases (Table 2) have ex-
panded most in central and eastern parts of the
territory. These species have only shown minor
changes in the western half of Catalonia (Fig-
ure 8). Abandonment of traditional farming
activities leads to forest colonisation, but at the
same time, creates a lower rate of exploitation
of forest resources due to low economic bene-
fits. Therefore regions have more forest, but
more important is the increase in maturity and
Figure 7. Mean index of distribution change according
to the habitat categories of the species. Change index
= (range2nd atlas – range1statlas)/((range2nd atlas+
range1st atlas)/2). This index measures distributional
increase or decline symmetrically around 0, with
maximum decline in the case of extinction as -2, and
maximum increase in case of colonisation as +2.
Índex de la mitjana de canvi en la distribució d’acord
amb la categoria d’hàbitat de les espècies. Índex de
canvi: (distribució2n atles – distribució1er atles)/
((distribució2n atles+ distribució1er atles)/2). Aquest
índex permet mesurar els increments i decrements
en la distribució de forma simètrica al voltant de 0:
-2 significaria un valor de regressió màxim (extinció)
i +2 un valor d’increment màxim (colonització).
The Catalan Breeding Bird Atlas
133
complexity of vegetation structure. This matu-
ration may have allowed strict forest species
such as the nuthatch or some woodpecker spe-
cies to colonise new forest areas that might have
been not used due to lack of appropriate vege-
tation structure (Camprodon 2003). Increases
in bird populations inhabiting core forest habi-
tats may in turn result in a surplus of individu-
als that disperse to other areas not necessarily
of recent origin but just happen to be near pop-
ulation source areas (Donald & Fuller 1998).
Overall the results suggest further investigation
and careful examination of forest species expan-
sion patterns in Catalonia in the light of cur-
rent and predicted expansion and maturation
of forested habitats in the area (Preiss et al.
1997).
Decreasing forests species
Forest species contracting in range showed a
rather homogeneous spatial pattern with the
decreases being in large areas of the coast, and
western pre-Pyrenees. Only some areas located
in central Catalonia, north-east coast and in-
land south, escape the general decreasing ten-
dency (Figure 8).
Decreasing forest species (e.g. Stock Dove,
Turtle Dove, Green Woodpecker) seem to be
species that to some extent depend on the avail-
ability of nearby open habitats to gather some
essential resources, especially food (Brotons et
al. 2004). Given the importance of open habi-
tats for forest species, distributional changes may
be related to changes occurring in those open
habitats (either agricultural, shrubs or both)
surrounding the forests where the species breed,
or to variation in landscape heterogeneity and
the relative composition of different habitat type
in the landscape (Dunning et al. 1992).
Farmland species
Increasing farmland species
Farmland species with expanding distributions
concentrated increases in agriculture areas in
western and inland Catalonia and in two coast-
al areas nearer the central and northern coast-
line (Figure 9). Minor decreases of otherwise in-
creasing farmland species seemed apparent in
southern areas and in more local areas in the
north. Positive changes for this group seemed to
have concentrated in areas with lower propor-
tions of wooded farmland in the 80’s indicating
that expansions have mostly occurred in mosaic
farmland areas due to changes occurring within
these habitats (Estrada et al. 2004). To some
degree, these changes may be associated with
changes in agricultural activities leading to in-
tensification in the area as suggested by the pos-
itive relationship between farmland bird expan-
Figure 8. Spatial representation of changes in species distribution of (A) increasing and (B) decreasing
forest species between the two breeding bird atlases (periods 1975-1983 and 1999-2002).
Representació espacial dels canvis en la distribució entre els dos atles d’ocells nidificants (períodes 1975-
1983 i 1999-2002): (A) espècies forestals en augment i (B) espècies forestals en disminució.
L. Brotons et al.
134
Revista Catalana d’Ornitologia 24 (2008)
Figure 9. Spatial representation of changes in species distribution of (A) increasing and (B) decreasing
farmland species between the two breeding bird atlases (periods 1975-1983 and 1999-2002).
Representació espacial dels canvis en la distribució entre els dos atles d’ocells nidificants (períodes 1975-
1983 i 1999-2002): (A) espècies de medis agrícoles en augment i (B) espècies de medis agrícoles en disminució.
sion and the amount of urban and suburban
habitats in 1987. Expanding farmland species
tended to increase more in areas with relatively
high proportion of forests in the 80’s indicating
that generalist species already using diverse land-
scapes with a high proportion of forests seemed
to suffer less from the changes occurred and rath-
er may have partially benefited from them.
Decreasing farmland species
Farmland species with significant range contrac-
tions showed a very consistent pattern in distri-
butional change, with apparent higher loss of
species in the Mediterranean coastal and pre-
coastal area. Inland, losses appeared to be high
in western regions (Figure 9). Decreases in farm-
land species were larger in open herbaceous
farmland with relatively higher amounts of for-
est habitat in the landscape (Estrada et al. 2004).
This suggests that decreases in habitat quality
of the already marginal farmland habitats may
have been important here or alternatively, that
decreases in the permeability of the landscape
to farmland birds due to forestation may play
an important role. An additional predictor of
farmland species loss in Catalonia appears to
be the amount of tree crop cover in 1987. These
results suggest that agricultural areas contain-
ing tree crops may prove to be important habi-
tats for decreasing farmland species. In fact, re-
cent changes in agriculture have meant large
changes in the tree species cropped in Mediter-
ranean areas (loss of almond trees, increase in
olive trees). The results suggest that socioeco-
nomic processes occurring within these farm-
land habitats may be critical to a number of
farmland species which certainly merits further
investigation (Calvet et al. 2004).
The results of the present atlas on distribu-
tion changes of farmland birds in Catalonia are
in line with the results found elsewhere in cen-
tral Europe on the negative effects of agricul-
tural intensification on farmland bird commu-
nity (Fuller et al. 1995, Gates & Donald 2000,
Donald et al. 2001). However, particularities of
Mediterranean agricultural mosaics should be
considered here. The pattern of change seems
to be less intense than the one found in the pre-
vious studies, stressing the fact that agricultural
practices in the Mediterranean may be less
harmful for birds due to its less intensive char-
acter. Future research should identify the proxi-
mate reasons affecting populations at different
stages of life cycle for different focal species in
these habitats, with a special focus on the role
of habitat heterogeneity on species occurrence,
which has been recently identified as a key fac-
tor behind biodiversity loss in farmland systems
(Benton et al. 2003).
The Catalan Breeding Bird Atlas
135
Conclusions
The distribution of Catalan avifauna has been
far from stable during the last 20 years and sig-
nificant changes have occurred in a large
number of species. Changes were often concen-
trated in particular areas, which greatly differed
according to the habitat selection of the differ-
ent groups and their general distribution trends.
This result stresses the importance of taking into
account variation in sampling effort to analyse
temporal variations in distributions.
Differences in sampling effort were by far the
most significant single predictor of the detected
differences between the two atlases (Estrada et
al. 2004). However, and in spite of the broad
character of the categorisation employed, the
results also provide strong evidence that the
observed changes in distribution patterns be-
tween the two atlases are to some degree linked
to parallel changes in landscape composition
during this period or to changes in the quality
of specific habitats. In particular, wetland man-
agement, forest maturation, fire regime and
changes in agricultural practices appear amongst
the most powerful driving factors of bird distri-
bution changes in Catalonia during the 20-year
interval.
The methodology used in this atlas is a pow-
erful tool for mapping complete species distri-
bution from a detailed standardised grid. Fur-
thermore, it will allow for a simple visual
interpretation of changes in species’ distribution
at a local scale in future atlases. Since changes
in land-use and climatic patterns will be cap-
tured by future GIS layers, new niche-based
models generated from new data and future dis-
tribution maps depicting new conditions will be
able to be produced.
Although not entirely new, the modelling
approach used by this atlas represents one of
the first applications of niche-based models to
the generation of high-resolution species distri-
bution maps. The accuracy of the model’s pre-
dictions and the possibility of objective model
evaluation make this methodology a robust tool
that will undoubtedly be incorporated into
faunal cartography.
In spite of the promising results obtained,
our analysis is clearly a tentative first step to-
wards the identification of causal hypotheses
behind distribution changes in Catalan bird
populations. In addition to land use patterns and
their temporal changes, other factors are surely
critical to explain the changes in the distribu-
tion of a particular species group. Amongst these
factors, climate change and the impact of inva-
sive species should be considered further. Other
unforeseen factors are likely to affect particular
species and therefore should be treated on a
species per species basis. The identification and
quantification of causes behind population and
distribution changes in birds is essential to help
in their understanding and conservation. We
hope that the data included in the atlas and this
preliminary analysis help to guide future ap-
proaches.
These future studies should be conducted
on focal species or groups to further explore
main patterns and the mechanisms causing
them. Combination of different data from dif-
ferent bird monitoring programs may help fill
gaps in our knowledge. The availability of other
monitoring programmes lead by ICO such as
SOCC (Common Bird Survey), SYLVIA (Con-
stant Effort Ringing Sites), PERNIS (Raptor
Migration) and available environmental infor-
mation including satellite data will crucially
contribute to this objective.
Acknowledgements
The results described in this manuscript depend on
the work of more than five hundred volunteers. We
would like to thank their invaluable effort. We would
also thank for the support given by the Government
of Catalonia by means of Departament de Medi Am-
bient i Habitatge and by Obra Social Caixa Catalunya.
This research is a contribution to the Montpellier-
Barcelona LEA “Mediterranean Ecosystems in a
Changing World” and has been supported by the
Marie Curie programme within the 6th European
Community Framework.
Resum
Atles dels ocells nidificants de
Catalunya: aspectes metodològics i
implicacions ecològiques
El primer atles dels ocells nidificants de Catalunya,
publicat a principis dels vuitanta del segle passat, va
cobrir en una quadrícula UTM de 10x10 km una
àrea de 32.000 km2 situada al nordest d’Espanya,
constituint una experiència pionera en relació a la
L. Brotons et al.
136
Revista Catalana d’Ornitologia 24 (2008)
cartografia d’ocells en la conca Mediterrània. Ara es
presenten les novetats i els principals resultats del
nou atles dels ocells nidificants de Catalunya, portat
a terme 20 anys després, en el període 1999-2002.
Aquest nou atles va incorporar una sèrie
d’innovacions metodològiques entre les quals cal
ressaltar els censos de temps controlat en una mostra
de quadrats UTM 1x1 km dintre de cadascun dels
quadrats UTM 10x10 km. Aquests censos a petita
escala (al voltant de 3.200) van permetre un mostreig
uniforme del voltant del 10% de l’àrea d’estudi, i van
possibilitar així la producció de mapes de distribució
d’espècies a escala local (a 500x500 m de resolució).
Aquests mapes van ser generats a través de
regressions logístiques en el marc dels models de
nínxol ecològic. En el nostre cas aquests models van
incloure informació de 45 variables ambientals que
van des dels usos del sòl i el relleu a la influència
humana i el clima; d’altra banda es va incorporar
informació sobre l’estructura espacial de les dades
per tal d’incorporar l’efecte de l’autocorrelació es-
pacial. Actualment, aquest atles està entre els millors
treballs de mapatge quantitatiu de les distribucions
d’ocells a Europa. D’altra banda, es van utilitzar les
dades dels censos en quadrats UTM 1x1 km per a
produir corbes d’acumulació d’espècies en el temps i
estimar així el nombre d’espècies que es poden trobar
en cada quadrat UTM 10x10 km per unitat de temps.
L’aplicació d’aquests models per al primer i el nou
atles va permetre una estimació dels canvis en la
distribució dels ocells tenint en compte les diferències
en l’esforç de mostreig (temps emprat en el cens d’un
determinat quadrat) entre els dos períodes. En gene-
ral, el nou atles proporciona una de les millors imat-
ges dels canvis a gran escala ocorreguts en l’avifauna
del sud d’Europa durant els últims 20 anys. Els resul-
tats obtinguts donen suport fermament a la hipòtesi
que en aquest període hi ha hagut canvis de distri-
bució molt destacats, sovint associats a modificacions
en els usos del sòl.
Resumen
Atlas de las aves reproductoras de
Catalunya: aspectos metodológicos e
implicaciones ecológicas
El primer atlas de las aves reproductoras de Catalu-
ña, publicado a principios de los años ochenta del
siglo pasado, cubrió en una cuadrícula UTM de
10x10 km un área de 32.000 km2 situada en el nor-
deste de España, constituyendo una experiencia pio-
nera en relación a la cartografía de aves en la cuen-
ca Mediterránea. Ahora presentamos las novedades
y los principales resultados del nuevo atlas de aves
reproductoras de Cataluña, llevado a cabo 20 años
después, en el período 1999-2002. Este nuevo atlas
incorporó una serie de innovaciones metodológicas
entre las que hay que resaltar los censos de tiempo
controlado en una muestra de cuadrículas UTM 1x1
km dentro de cada una de las cuadrículas UTM
10x10 km. Estos censos a pequeña escala (alrededor
de 3.200) permitieron un muestreo uniforme de al-
rededor del 10% del área de estudio, posibilitando
así la producción de mapas de distribución de espe-
cies a escala local (a 500x500 m de resolución). Es-
tos mapas fueron generados a través de regresiones
logísticas en el marco de los modelos de nicho
ecológico. En nuestro caso estos modelos incluye-
ron información de 45 variables ambientales que van
desde los usos del suelo y el relieve a la influencia
humana y el clima; por otro lado se incorporó infor-
mación sobre la estructura espacial de los datos para
incorporar el efecto de la autocorrelación espacial.
Actualmente, este atlas está entre los mejores tra-
bajos de mapeo cuantitativo de las distribuciones de
aves en Europa.
Asimismo, se utilizaron los datos de los censos
en cuadrículas UTM 1x1 km para producir curvas
de acumulación de especies en el tiempo y estimar
así el número de especies que se pueden encontrar
en cada cuadrícula UTM 10x10 km por unidad de
tiempo. La aplicación de estos modelos para el primer
y el nuevo atlas permitió una estimación de los
cambios en la distribución de aves teniendo en
cuenta las diferencias en esfuerzo de muestreo
(tiempo empleado en el censo de una determinada
cuadrícula) entre los dos períodos. En general, el
nuevo atlas proporciona una de las mejores imágenes
de los cambios a gran escala ocurridos en la avifauna
del sur de Europa durante los últimos 20 años. Los
resultados obtenidos apoyan firmemente la hipótesis
de que en este período ha habido cambios de
distribución muy destacados, a menudo asociados a
modificaciones en los usos del suelo.
References
Augustin, N.H. Mugglestone, M.A., & Buckland,
S.T. 1996. An autologistic model for the spatial
distribution of wildlife. J. Appl. Ecol. 33: 339
347.
Benton, T.G., Vickery, J.A. & Wilson, J.D. 2003.
Farmland biodiversity: is habitat heterogeneity
the key? Trends Ecol. Evol. 18: 182188.
BohningGaese, K. & Bauer, H.G. 1996. Changes
in species abundance, distribution, and diversity
in a central European bird community. Conserv.
Biol. 10: 175187.
Brotons, L., Herrando S. & Martin, J.L. 2004. Bird
assemblages in forest fragments within Mediter-
ranean mosaics created by wild fires. Landscape
Ecol. 19 (6): 663675.
Calvet, J., Estrada, J., Mañosa, S., Moncasí, F.,
Solans, J. & West, S. 2004. Els ocells de la plana
de Lleida. Lleida: Pagès editors.
The Catalan Breeding Bird Atlas
137
Camprodon, J. 2003. Estructura dels boscos i gestió
forestal al nord-est ibèric: efecte sobre la
composició, abundància i conservació dels ocells.
PhD thesis. Barcelona: Universitat de Barcelona.
Cramp, S. & Perrins, C.M. 1994. The Birds of the
Western Palearctic. Vol. VIII. Oxford: Oxford
University Press.
Donald, P.F. & Fuller, R.J. 1998. Ornithological atlas
data: a review of uses and limitations. Bird Study
45: 129145.
Donald, P.F., Green, R.E. & Heath, M.F. 2001.
Agricultural intensification and the collapse of
Europe’s farmland bird populations. Proc. Royal
Soc. London (Ser. B Biol. Sci.) 268: 2529.
Dunning, J.B., Danielson, B.J. & Pulliam, H.R.
1992. Ecological Processes That Affect Populations
in Complex Landscapes. Oikos 65: 169175.
Estrada, J., Pedrocchi, V., Brotons, L. and
Herrando, S. (eds.) 2004. Atles dels ocells
nidificants de Catalunya 1999-2002. Barcelona:
ICO-Lynx Editions.
Fielding, A.H., Bell, J.F. 1997. A review of meth-
ods for the assessment of prediction errors in
conservation presence/absence models. Environ.
Conserv. 24: 3849.
Fuller, R.J., Gregory, R.D., Gibbons, D.W.,
Marchant, J.H., Wilson, J.D., Baillie, S.R. &
Carter, N. 1995. Population declines and range
contractions among lowland farmland birds in
Britain. Conserv. Biol. 9: 14251441.
Gardemfors, U., Hilton-Taylor, C., Mace, G.M.
and Rodríguez, J.P. 2001. The application of
IUCN Red List Criteria at regional levels. Conserv.
Biol. 15 (5): 12061212.
Gaston, K.J., Blackburn, T.M. & Goldewijk, K.K.
2003. Habitat conversion and global avian
biodiversity loss. Proc. Royal Soc. London (Ser.
B Biol. Sci.) 270: 12931300.
Gates, S. & Donald, P.F. 2000. Local extinction of
British farmland birds and the prediction of fur-
ther loss. J. Appl. Ecol. 37: 806820.
Gibbons, D.W., Reid, J.B. & Chapman, R.A. 1993.
The New Atlas of Breeding Birds in Britain and
Ireland: 1988-1991. London: Poyser.
Guisan, A. Zimmermann, N.E. 2000. Predictive
habitat distribution models in ecology. Ecological
Modelling 135: 147186.
Hagemeijer, W.J.W. Blair, M.J. 1997. The EBCC
atlas of European breeding birds: their distribu-
tion and status. London: Poyser.
Herrando, S., Brotons, L., Estrada, J. and
Pedrocchi, V. The Catalan common bird survey
(SOCC): A tool to estimate species population
numbers. In press. (this proceedings).
Legendre, P. Legendre, L. 1998. Numerical ecol-
ogy. Amsterdam: Elsevier.
Martí, R. del Moral, J.C. 2003. Atlas de las Aves
Reproductoras de España. Madrid: Dirección Gen-
eral de Conservación de la Naturaleza - Sociedad
Española de Ornitología.
Muntaner, J., Ferrer, X. & Martínez-Vilalta, A.
1983. Atlas dels ocells nidificants de Catalunya i
Andorra. Barcelona: Ketres.
Osborne, P.E. Tigar, B.J. 1992. Interpreting Bird
Atlas Data Using Logistic-Models - an Example
from Lesotho, Southern Africa. J. Appl. Ecol. 29:
5562.
Parker, V. 1999. The use of logistic regression in
modelling the distributions of bird species in
Swaziland. S. Afr. J. Zool. 34: 3947.
Preiss, E., Martin, J.L. & Debussche, M. 1997.
Rural depopulation and recent landscape changes
in a Mediterranean region: Consequences to the
breeding avifauna. Landscape Ecol. 12: 5161.
Sala, O.E., Chapin, F.S., Armesto, J.J., Berlow,
E., Bloomfield, J., Dirzo, R., Huber-Sanwald,
E., Huenneke, L.F., Jackson, R.B., Kinzig, A.,
Leemans, R., Lodge, D. M., Mooney, H.A.,
Oesterheld, M., Poff, N.L., Sykes, M.T.,
Walker, B.H., Walker, M. & Wall, D.H. 2000.
Biodiversity - Global biodiversity scenarios for the
year 2100. Science 287: 17701774.
Schmid, H., Luder, R., Naef-Daenzer, R., Graf,
R. & Zbinden, N. 1998. Atlas des oiseaux
nicheurs de Suisse. Distribution des oiseaux
nicheurs en Suisse et du Liechtenstein en 1993-
1996. Sempach: Station Ornithologique Suisse.
Suarez-Seoane, S. Osborne, P.E. Alonso, J.C.
2002. Large-scale habitat selection by agricul-
tural steppe birds in Spain: identifying species-
habitat responses using generalized additive
models. J. Appl. Ecol. 39: 755771.
Thomas, C.D. & Lennon, J.J. 1999. Birds extend
their ranges northwards. Nature 399: 213213.
Tucker, G.M. & Evans, M.I. 1997. Habitats for birds
in Europe: a conservation strategy for a wider
environment. Cambridge: BirdLife International
(BirdLife Conservation Series nº 3).
Vaughan, I.P. Ormerod, S.J. 2003. Improving the
quality of distribution models for conservation by
addressing shortcomings in the field collection of
training data. Conserv. Biol. 17: 16011611.
Wiens, J. A. 1989. The Ecology of Bird Communi-
ties. Cambridge: Cambridge University Press.
... Data was requested from the Catalan Ornithological Institute (ICO), which provided presence and absence records of M. calandra from the Catalan Breeding Bird Atlas (CBBA) [51]. CBBA surveys were conducted during the breeding season (1 March to 30 July) in the years 1999–2002. ...
... Surveys were conducted by experienced professionals between sunrise and 11 am and between 6 pm and sunset. The survey plots were 1 km × 1 km UTM squares in which two 1-hour surveys were conducted and the presence or absence of the species was recorded [51]. The total number of M. calandra locations in the study area was 339, out of which, 251 (74%) were presence points and 88 (26%) were absence points. ...
... The total number of M. calandra locations in the study area was 339, out of which, 251 (74%) were presence points and 88 (26%) were absence points. Further details on the field sampling methodology can be found in Brotons et al. [51]. ...
Article
Full-text available
The efficacy of integrating open access geospatial data to produce habitat suitability maps for the corn bunting (Miliaria calandra) was investigated. Landsat Enhanced Thematic Mapper Plus (ETM+), Shuttle Radar Topography Mission (SRTM) and Corine (Coordination of Information on the Environment) land cover data for the year 2000 (CLC2000) were processed to extract explanatory variables and divided into three sets; Satellite (ETM+, SRTM), CLC2000 and Combined (CLC2000 + Satellite). Presence-absence data for M. calandra, collected during structured surveys for the Catalan Breeding Bird Atlas, were provided by the Catalan Ornithological Institute. The dataset was partitioned into an equal number of presence and absence points by dividing it into five groups, each composed of 88 randomly selected presence points to match the number of absences. A logistic regression model was then built for each group. Models were evaluated using area under the curve (AUC) of the receiver operating characteristic (ROC). Results of the five groups were averaged to produce mean Satellite, CLC2000 and Combined models. The mean AUC values were 0.69, 0.81 and 0.90 for the CLC2000, Satellite and the Combined model, respectively. The probability of M. calandra presence had the strongest positive correlation with land surface temperature, modified soil adjusted vegetation index, coefficient of variation for ETM+ band 5 and the fraction of non-irrigated arable land.
... Catalan Breeding Birds Atlas (CBBA data), which reports information on breeding bird distribution in Catalonia between 1999 and 2002 based on intensive surveys of 3077 grid cells at 1-km resolution (Brotons et al 2008). Among the 214 bird species that breed in Catalonia, we focused on those that are expected to be affected by fire-vegetation dynamics and climate change to illustrate how the combined effect of these drivers can affect the performance of the PAs Regos et al. 2018). ...
Article
Full-text available
1.The combined effects of climate change and other factors, such as land use change or fire disturbance, pose daunting challenges for biodiversity conservation worldwide. 2.We predicted the future effectiveness of the Natura 2000 (N2000), the current network of protected areas (PA) in Europe, at maintaining and representing suitable environmental conditions for a set of 79 bird species between 2000 and 2050 in a fire‐prone area, strongly affected by land abandonment processes in North East Spain. We then compared PA performance with a set of alternative priority areas for conservation, which consider fire‐vegetation dynamics, selected by using a conservation planning tool (Marxan). Fire‐vegetation dynamics were modelled using a process‐based model (Medfire Model) under alternative fire management and climate change scenarios. Bird assemblage distributions were predicted using the spatially‐explicit species assemblage modelling framework (SESAM) using distribution models from individual species that hierarchically integrate climate change and wildfire‐vegetation dynamics. 3.The amount of suitable environmental conditions within the N2000 network was predicted to fall by around 15%, on average, over the next decades in relation to the initial conditions but could be partially modulated by fire management policies in future. The efficiency of the current PA system was predicted to decrease from 17.4 to 15% over the next decades. However, a more efficient PA system could be achieved with a conservation planning approach that explicitly considers fire‐vegetation dynamics and their management. 4.Synthesis and applications. Our findings show: (1) how the current Natura 2000 could still hold an important bird conservation value by 2050; (2) how the relocation of some protected areas should be considered in order to substantially increase bird conservation effectiveness; and (3) how the integration of fire‐vegetation dynamics, fire management policies and their objectives within conservation planning provide ‘win‐win’ solutions for bird conservation and fire prevention in fire‐prone abandoned landscapes.
... Thus, we adopted a more conservative outlook of habitat change for both species, which is appropriate given the inherent uncertainty of modelling approaches. This method has been successfully applied in identifying priority areas for bird conservation, and selecting marine Important Bird and Biodiversity Areas (IBAs) in Spain [56][57][58]. HSH can be interpreted in the context of the European Birds directive as the priority areas for conservation of these species. ...
Article
Full-text available
Monitoring protected areas (PAs) is essential for systematic evaluation of their effectiveness in terms of habitat protection, preservation and representativeness. This study illustrates how the use of species distribution models that combine remote sensing data and information about biodiversity surrogates can contribute to develop a systematic protocol for monitoring PAs. In particular, we assessed the effectiveness of the Natura 2000 (N2000) network, for conserving and preserving the representativeness of seven raptor species in a highly-dynamic landscape in northwest Spain between 2001 and 2014. We also evaluated the cost-effectiveness of the N2000 network by using the total area under protection as a proxy for conservation costs. Overall, the N2000 network was found to poorly represent the habitats of the raptor species. Despite the low representativeness, this network showed a high degree of effectiveness due to increased overall habitat availability for generalist and forest specialist species between 2001 and 2014. Nevertheless, additional protected areas should be established in the near future to increase their representativeness, and thus ensure the protection of open-habitat specialist species and their priority habitats. In addition, proactive conservation measures in natural and semi-natural ecosystems (in particular, montane heathlands) will be essential for long-term protection of Montagu’s harrier (species listed in the Annex I of the Bird Directive), and thus complying with the current European Environmental Legislation. This study sheds light on how the development and application of new protected area indices based on the combined use of freely-available satellite data and species distribution models may contribute substantially to the cost-efficiency of the PA monitoring systems, and to the ‘Fitness Check’ process of EU Nature Directives.
... Model accuracy was evaluated for test data using the area under the receiver operating characteristic (ROC) curve, commonly denoted as area under the curve (AUC). Area under the curve values of 0.5 imply that model accuracy is no better than random, while AUCs of 0.8 or more are considered good, and values of 0.9 or more are considered excellent (Brotons, Herrando, Estrada, Pedrocchi, & Martin, 2008). The AUC for test data was calculated in package pROC of program R (Robin et al., 2011). ...
Article
Full-text available
To assess the importance of variation in observer effort between and within bird atlas projects and demonstrate the use of relatively simple conditional autoregressive (CAR) models for analyzing grid-based atlas data with varying effort. Pennsylvania and West Virginia, United States of America. We used varying proportions of randomly selected training data to assess whether variations in observer effort can be accounted for using CAR models and whether such models would still be useful for atlases with incomplete data. We then evaluated whether the application of these models influenced our assessment of distribution change between two atlas projects separated by twenty years (Pennsylvania), and tested our modeling methodology on a state bird atlas with incomplete coverage (West Virginia). Conditional Autoregressive models which included observer effort and landscape covariates were able to make robust predictions of species distributions in cases of sparse data coverage. Further, we found that CAR models without landscape covariates performed favorably. These models also account for variation in observer effort between atlas projects and can have a profound effect on the overall assessment of distribution change. Accounting for variation in observer effort in atlas projects is critically important. CAR models provide a useful modeling framework for accounting for variation in observer effort in bird atlas data because they are relatively simple to apply, and quick to run.
... This study was conducted in Catalonia, a Mediterranean climate area in north-eastern Spain. Data was obtained on the occurrence (presence/absence) of bird species at two different spatial extents and resolutions from 1) the atlas of European breeding birds (Hagemeijer and Blair 1997), which documents the occurrence of breeding bird species in Europe between the late 1980s and early 1990s in 3165 grid cells at 50 km resolution, to encompass the widest possible climatic range of the target species (Del Hoyo et al. 2006, Huntley et al. 2007, and 2) the Catalan breeding bird atlas (Brotons et al 2008), which reports information on breeding bird distribution in Catalonia between 1999 and 2002 based on intensive surveys of 3077 grid cells at 1 km resolution. ...
Article
Full-text available
Community-level climate change indicators have been proposed to appraise the impact of global warming on community composition. However, non-climate factors may also critically influence species distribution and biological community assembly. The aim of this paper was to study how fire–vegetation dynamics can modify our ability to predict the impact of climate change on bird communities, as described through a widely-used climate change indicator: the community thermal index (CTI). Potential changes in bird species assemblage were predicted using the spatially-explicit species assemblage modelling framework – SESAM – that applies successive filters to constrained predictions of richness and composition obtained by stacking species distribution models that hierarchically integrate climate change and wildfire–vegetation dynamics. We forecasted future values of CTI between current conditions and 2050, across a wide range of fire–vegetation and climate change scenarios. Fire–vegetation dynamics were simulated for Catalonia (Mediterranean basin) using a process-based model that reproduces the spatial interaction between wildfire, vegetation dynamics and wildfire management under two IPCC climate scenarios. Net increases in CTI caused by the concomitant impact of climate warming and an increasingly severe wildfire regime were predicted. However, the overall increase in the CTI could be partially counterbalanced by forest expansion via land abandonment and efficient wildfire suppression policies. CTI is thus strongly dependent on complex interactions between climate change and fire–vegetation dynamics. The potential impacts on bird communities may be underestimated if an overestimation of richness is predicted but not constrained. Our findings highlight the need to explicitly incorporate these interactions when using indicators to interpret and forecast climate change impact in dynamic ecosystems. In fire-prone systems, wildfire management and land-use policies can potentially offset or heighten the effects of climate change on biological communities, offering an opportunity to address the impact of global climate change proactively. This article is protected by copyright. All rights reserved.
Article
Full-text available
Landscape dynamics pose substantial challenges to biodiversity conservation and could compromise the future effectiveness of protected areas (PAs). Multiple interacting drivers of landscape dynamics, including land-cover change, wildfire, and climate change, are usually studied separately. This limits our ability to respond effectively to conservation needs through appropriate measures. Using Catalonia (North-east Spain) as a case study, we evaluated the future representativeness of PAs in covering the distributions of bird species of conservation interest under the effects of different global change drivers and identified priority areas for their conservation in the future. We also evaluated the role of different global change drivers in defining future conservation needs. Under current management practices, land-cover change appears to be a more relevant driver than climate change. Forested habitats are predicted to increase at the expense of early successional habitats, and it appears that fires will not compensate for this loss by creating new early successional habitats. Consequently, early successional species are likely to suffer the greatest decline in distribution. Consequently, additional priority for future management should be given to the last available early successional habitats, mainly outside PAs, as their selection will be marginally driven by climate change or wildfires. The conservation of early successional habitats in the European Union can benefit from existing policies, particularly the 2030 Biodiversity Strategy or the Common Agriculture Policy. However, further evaluation is needed to explore whether these policies are suitable under dynamic landscape conditions or if new policy is needed to ensure effective conservation efforts.
Article
Full-text available
Habitat loss and degradation induced by human development are among the major threats to biodiversity worldwide. In this study, we tested our ability to predict the response of bird communities (128 species) to land‐use changes in southern Québec (c. 483,100 km2) over the last 30 years (between ‘1984‐89’ and ‘2010‐2014’) by using species distribution models (299 302 occurrences in 30 408 locations) from a hindcasting perspective. Results were grouped by functional guilds to infer potential impacts on ecosystem services, and to relate model transferability (i.e., ability of our models to be generalized to other times and scales) to specific functional and life‐history traits. Overall, our models were able to accurately predict, both in space and time, habitat suitability for 69% of species, especially for granivorous, non‐migrant, tree‐nesting species, and species that are tied to agricultural areas under intensive use. These findings indicate that model transferability depends upon specific functional and life‐history traits, providing further evidence that species’ ecologies affect the ability of models to accurately predict bird distributions. Declining bird species were mostly short‐distance migrants that were associated with open habitats (agricultural and non‐productive forest) with aerial insectivorous or granivorous diets, which may be related to agricultural intensification and land abandonment. Land‐use changes were positive for some forest bird species, which were mainly associated with mixed and deciduous forests, generalist diets and tree‐nesting strategies. Yet cavity‐nesting birds have suffered substantial reductions in their distributions, suggested that cumulative effects of intensive logging and wildfires on mature forests pose a threat for forest‐specialist species. Habitat suitability changes predicted by our course‐scale species distribution models partially agreed with the long‐term trends reported by the North American Breeding Bird Survey. Our findings confirm land‐use change as a key driving force for shaping bird communities in southern Québec, together with the need to explicitly incorporate it into global change scenarios that better inform decision‐makers on conservation and management.
Article
Full-text available
Estimates of total population sizes of individual bird species are widely used in macroecology and conservation biology. Current estimates of population sizes of Czech birds are based on atlas mapping data. However, this mapping has primarily been focused on breeding bird distribution and not on species abundance. The aim of our study is to compare these atlasbased estimates with extrapolations from point count data. For this purpose, we collected data on breeding bird abundances not adjusted for species detectability, counted along a 400-km point count transect. The atlas estimates were strongly correlated with transect estimates suggesting that the assessment of species relative commonness and rarity is quite independent of the method of the estimate. Although the total estimate of the number of all individuals for all species together was similar using both types of data, the atlas estimates of species population sizes were lower than the transect-based ones in most but the commonest species. Unfortunately, we cannot judge which kind of data is closer to reality. Since the estimates often differed by an order of magnitude for a given species, it is possible that both of them are substantially biased. We urge to focus on achieving more reliable estimates using modern bird census techniques employing representative sampling and quantification of species detectability.
Article
Aim The paradigm that species’ patterns of distribution, abundance and coexistence are the result of adaptations of the species to their niches has recently been challenged by evidence that similar patterns may be generated by simple random processes. We argue here that a better understanding of macroecological patterns requires an integration of both ecological and neutral stochastic approaches. We demonstrate the utility of such an integrative approach by testing the sampling hypothesis in a species–energy relationship of forest bird species. Location A Mediterranean biome in Catalonia, Spain. Methods To test the sampling hypothesis we designed a metacommunity model that reproduces the stochastic sampling from a regional pool to predict local species richness variation. Four conceptually different sampling procedures were evaluated. Results We showed that stochastic sampling processes predicted a substantial part (over 40%) of the observed variation in species richness, but left considerable variation unexplained. This remaining variation in species richness may be better understood as the result of alternative ecological processes. First, the sampling model explained more variation in species richness when the probability that a species colonises a new locality was assumed to increase with its niche width, suggesting that ecological differences between species matter when it comes to explaining macroecological patterns. Second, extinction risk was significantly lower for species inhabiting high-energy regions, suggesting that abundance–extinction processes play a significant role in shaping species richness patterns. Main conclusions We conclude that species–energy relationships may not simply be understood as a result of either ecological or random sampling processes, but more likely as a combination of both.
Thesis
Full-text available
Twenty-five years after the implementation of the Birds Directive in 1979, Europe‟s farmland bird species and long-distance migrants continue to decrease at an alarming rate. Farmland supports more bird species of conservation concern than any other habitat in Europe. Therefore, it is imperative to understand farmland species‟ relationship with their habitats. Bird conservation requires spatial information; this understanding not only serves as a check on the individual species‟ populations, but also as a measure of the overall health of the ecosystem as birds are good indicators of the state of the environment. The target species in this study is the corn bunting, Miliaria calandra, a bird whose numbers in northern and central Europe have declined sharply since the mid-1970s. This study utilizes public domain data, namely Landsat imagery and CORINE land cover, along with the corn bunting‟s presence-absence data, to create a predictive distribution map of the species based on habitat preference. Each public domain dataset was preprocessed to extract predictor variables. Predictive models were built in R using logistic regression. Three models resulted from the regression analysis; one containing the satellite-only variables, one containing the land cover variables and a combined model containing both satellite and land cover variables. The final model was the combined model because it exhibited the highest predictive accuracy (AUC=0.846) and the least unexplained variation (RD=276.11). The results have shown that the corn bunting is strongly influenced by land surface temperature and the modified soil adjusted vegetation index. Results have also shown that the species strongly prefers non-irrigated arable land and areas containing vegetation that has high moisture content while avoiding areas with steep slopes and areas near human activity. This study has shown that the combination of public data from different sources is a viable method in producing models that reflect species‟ habitat preference. The development of maps that are comprised of information from both satellites and land cover datasets are of importance for species whose habitat requirements are poorly known.
Article
Full-text available
2 ) to the probability of occurrence (ranging from 0 to 1) for the sample of 1x1 km UTM squares in which transects were located. Model fit was highly significant in many cases, thus allowing the projection of species densities to non surveyed areas from which the probability of occurrence had been assessed. The Atlas-SOCC models were used to predict densities for each of the 1x1 km UTM squares of Catalonia, and then we added all these values in order to obtain population estimates for the whole country. Our approach stresses the interest of combining results from different monitoring programmes to obtain data useful to establish bird conservation strategies.
Article
Full-text available
We analyzed possible causes of changes in species abundance, range size, and diversity as well as extinctions and colonizations in a central European bird community. Using data from the semiquantitative "Lake Constance" breeding bird atlas, we demonstrated that changes in regional abundances from 1980-1981 to 1990-1992 of 151 coexisting bird species were influenced by breeding habitat and migratory status. Significant declines were found in populations of farmland species and long-distance migrants. Farmland species lost parts of their ranges but hardly changed in local abundance in sites where they still occurred. In contrast, declines in long-distance migrants were caused by significant declines in local abundance with only slight loss of occupied range. Regional extinctions and colonizations were predictable from overall population trends. For example, all species that went extinct were either farmland species or long-distance migrants. Avian community composition was influenced by disproportionate declines of abundant species. This led to declines in the total number of breeding pairs and in community biomass and to increases in community evenness, but to only slight declines in species richness. Future conservation efforts in Europe need to focus more on farmland species and on understanding causes for the declines of long-distance migrants.
Article
The method of logistic regression was used to model the observed geographical distribution patterns of bird species in Swaziland in relation to a set of environmental variables. Reporting rates derived from bird atlas data are used as an index of population densities. This is justified in part by the success of the modelling process. For each species the variables which were significantly related to its distribution were identified. Various methods for coding environmental variables from maps were investigated. A jack-knifing technique was used to demonstrate the predictive power of the logistic models. A criterion for assessing the goodness of fit of the logistic models was derived.
Article
The populations of farmland birds in Europe declined markedly during the last quarter of the 20th century, representing a severe threat to biodiversity. Here, we assess whether declines in the populations and ranges of farmland birds across Europe reflect differences in agricultural intensity, which arise largely through differences in political history. Population and range changes were modelled in terms of a number of indices of agricultural intensity. Population declines and range contractions were significantly greater in countries with more intensive agriculture, and significantly higher in the European Union (EU) than in former communist countries. Cereal yield alone explained over 30% of the variation in population trends. The results suggest that recent trends in agriculture have had deleterious and measurable effects on bird populations on a continental scale. We predict that the introduction of EU agricultural policies into former communist countries hoping to accede to the EU in the near future will result in significant declines in the important bird populations there.
Article
With the rise of new powerful statistical techniques and GIS tools, the development of predictive habitat distribution models has rapidly increased in ecology. Such models are static and probabilistic in nature, since they statistically relate the geographical distribution of species or communities to their present environment. A wide array of models has been developed to cover aspects as diverse as biogeography, conservation biology, climate change research, and habitat or species management. In this paper, we present a review of predictive habitat distribution modeling. The variety of statistical techniques used is growing. Ordinary multiple regression and its generalized form (GLM) are very popular and are often used for modeling species distributions. Other methods include neural networks, ordination and classification methods, Bayesian models, locally weighted approaches (e.g. GAM), environmental envelopes or even combinations of these models. The selection of an appropriate method should not depend solely on statistical considerations. Some models are better suited to reflect theoretical findings on the shape and nature of the species’ response (or realized niche). Conceptual considerations include e.g. the trade-off between optimizing accuracy versus optimizing generality. In the field of static distribution modeling, the latter is mostly related to selecting appropriate predictor variables and to designing an appropriate procedure for model selection. New methods, including threshold-independent measures (e.g. receiver operating characteristic (ROC)-plots) and resampling techniques (e.g. bootstrap, cross-validation) have been introduced in ecology for testing the accuracy of predictive models. The choice of an evaluation measure should be driven primarily by the goals of the study. This may possibly lead to the attribution of different weights to the various types of prediction errors (e.g. omission, commission or confusion). Testing the model in a wider range of situations (in space and time) will permit one to define the range of applications for which the model predictions are suitable. In turn, the qualification of the model depends primarily on the goals of the study that define the qualification criteria and on the usability of the model, rather than on statistics alone.
Article
We used extensive atlas and census data to assess trends in the distribution and population levels of birds on lowland farmland in Britain between the late 1960s and early 1990s. Many species of farmland birds have become less widespread or have declined in numbers, or both, but few have become more wide-spread or have increased. Of the 28 species classified as farmland birds the distributions of 24 contracted between 1970 and 1990. Of the 18 farmland species for which it was possible to assess population change, 15 were less abundant in 1990 than in 1970. Seven of the species were estimated to have undergone population decreases of at least 50%. Farmland species showing the largest population declines tended also to show substantial range contractions. Farmland species underwent an appreciably larger contraction of distribution than species associated with any other habitat. Furthermore, farmland species tended to decrease in abundance, whereas woodland species tended to increase. Population declines among farmland birds became evident in the mid- to late 1970s, a period when several fundamental changes were taking place in British agricultural practices. These included a great reduction in the spring sowing of cereals, a simplification of crop rotations, increased use of chemical pesticides and inorganic fertilizers, and more-intensive grassland management. We suggest that the declines of farmland bird species have been caused or aggravated by this pervasive intensification of agriculture. Existing research on declining farmland birds, however, indicates that there is no single mechanism underlying the population changes. We identify priorities for research, focusing mainly on relationships between bird populations and agricultural practices, but we also recognize a need for a better understanding of the role of predation.