Content uploaded by Olivier Gimenez
Author content
All content in this area was uploaded by Olivier Gimenez on Jul 11, 2016
Content may be subject to copyright.
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 552: 255–269, 2016
doi: 10.3354/meps11730 Published June 23
INTRODUCTION
Seabirds and environmental variability
Understanding species’ sensitivity to environmen-
tal change is essential for assessing their present sta-
tus and predicting future trends —2 key elements of
conservation. Birds are a well-studied group, and
many studies have reported the impacts of climate
and environmental conditions on their life histories
(for reviews see Walther et al. 2002, Parmesan 2006,
Grémillet & Boulinier 2009, Jenouvrier 2013, Cham-
bers et al. 2015). Climate change may affect birds at
different scales, for example by causing changes in
their distribution, phenology, population dynamics
and demographic traits. Together with habitat loss,
pollution, and introduced predators and/or competi-
tors, climate change is considered the major threat to
the persistence of many avian populations (Møller et
al. 2004, Halpern et al. 2007, Hoegh-Guldberg &
© Inter-Research 2016 · www.int-res.com*Corresponding author: yalbores@cicese.mx
Forecasting ocean warming impacts on
seabird demography: a case study on the European
storm petrel
Cecilia Soldatini1, Yuri Vladimir Albores-Barajas1,*, Bruno Massa2,
Olivier Gimenez3
1Unidad La Paz, Centro de Investigación Científica y de Educación Superior de Ensenada, La Paz, Baja California Sur, 23050,
Mexico
2Department of Agriculture and Forest Sciences, University of Palermo, Viale Scienze 13, 90128 Palermo, Italy
3CEFE UMR 5175, CNRS, Université de Montpellier, Université Paul-Valéry Montpellier, EPHE, 1919 Route de Mende,
34293 Montpellier Cedex 5, France
ABSTRACT: Bottom-up climatic forcing has been shown to be influential for a variety of marine
taxa, but evidence on seabird populations is scarce. Seasonal variation in environmental condi-
tions can have an indirect effect on subsequent reproduction, which, given the longevity and
single-brooding of seabirds, may affect population dynamics. Our study focuses on linking the
effect of oceanographic conditions (from 1991 to 2013) to the fecundity and consequently pop -
ulation growth rate of the Mediterranean subspecies of the European storm petrel Hydrobates
pelagicus melitensis. In this study, we examined 23 yr of >5400 capture–mark−recaptures (CMR)
and modelled the probability of skipping reproduction as a function of oceanographic variables
using CMR models. We demonstrate that a decrease in sea surface temperature in the pre-breed-
ing period negatively influences skipping propensity, and therefore hypothesize that this behav-
iour would have significant influence on population abundance over time. For this reason, we ana-
lysed population growth as a function of skipping probability as affected by oceanographic
conditions. We used stochastic demographic models to forecast the fate of the population, and
evaluated contrasted environmental condition scenarios. As a result, we found that a decrease in
frequency of cold winter events would probably reduce skipping propensity, with a positive effect
on the population as a whole.
KEY WORDS: Capture−mark−recapture · Environmental stochasticity · Hydrobates pelagicus ·
Population growth rate · Senescence
Resale or republication not permitted without written consent of the publisher
This authors' personal copy may not be publicly or systematically copied or distributed, or posted on the Open Web,
except with written permission of the copyright holder(s). It may be distributed to interested individuals on request.
Mar Ecol Prog Ser 552: 255–269, 2016
Bruno 2010, Maxwell et al. 2013). In this context,
adaptation, phenotypic plasticity, and homeostasis
are essential defences against extinction (Møller et
al. 2008) and may contribute to population robust-
ness in response to climate change (Jenouvrier 2013),
i.e. the species’ ability to cope with climate change.
However, assessing robustness to climate change is
challenging due to the many factors involved.
Seabird species have evolved in oceanic systems
characterized by cyclic climatic events, but are nega-
tively affected by the rapid increase of environmen-
tal stochasticity resulting from global warming,
which can induce a temporal mismatch between life
history strategies and increasingly erratic environ-
mental oscillations (Jenouvrier et al. 2008, Ainley et
al. 2010, Barbraud et al. 2011, Sydeman et al. 2012).
However, the effects of extreme climatic events are
far from being understood; some cases remain un -
explained and results of studies are contradictory
(Croxall et al. 2002, Forcada & Trathan 2009, Jenou-
vrier et al. 2015, Bailey & van de Pol 2016). Breeding
performance in particular may be affected directly by
sea surface temperature (SST) by changing the distri-
bution and/or abundance of important prey species
(Frederiksen et al. 2007), or indirectly via effects on
prey recruitment (Hedd et al. 2006). Furthermore,
unfavourable climate conditions may induce seabirds
to skip breeding (Jenouvrier et al. 2005a, Olivier et
al. 2005, Cubaynes et al. 2011) or affect nest-site con-
ditions (Chambers et al. 2011, Moreno & Møller 2011,
Soldatini et al. 2014).
Global climatic oscillations and environmental
variability in the Mediterranean
The North Atlantic Oscillation index (NAO) and
the At lantic Multidecadal Oscillation are decadal
climate indices with strong influences on ocean and
terrestrial ecosystems in North America and Europe.
These indices modulate a variety of phenomena such
as precipitation, surface winds, and upwelling, thus
influencing ecosystem states at a macro-scale (from
hundreds to thousands of kilometres) (Stommel 1963,
Nye et al. 2014). On a meso-scale (from tens to hun-
dreds of kilometres), upwelling in the Mediterranean
is driven by current systems and prevailing winds
(Agostini & Bakun 2002). Up welling conditions en -
hance nutrient circulation and positively affect pri-
mary producers, which in turn has a direct effect
on subsequent trophic levels, thereby enhancing
the trophic web. These conditions support predator
populations (Agostini & Bakun 2002, Reul et al.
2005, Coll et al. 2006, Baum & Worm 2009). Overall,
the presence of an upwelling area in the proximity
of seabird colonies has positive effects on the
population.
In the Mediterranean, future temperatures are pre-
dicted to increase and rainfall to decrease (Bladé et
al. 2012). These changes could positively affect nest-
ing seabirds by allowing higher fledging survival
(Soldatini et al. 2014). On the other hand, a general-
ized rise in temperature could affect the entire mar-
ine food web, with unpredictable consequences for
seabird population trends (Paleczny et al. 2015).
The intermittent breeding strategy
Due to the high costs of reproduction, seabirds
often skip reproduction as an adaptive strategy in
response to environmental constraints that favours
their own survival and future reproduction (Jenou-
vrier et al. 2005b, Giudici et al. 2010, Cubaynes et al.
2011, Reed et al. 2015). Intermittent breeding is actu-
ally a widely observed phenomenon, which can be
modelled based on the cost of reproduction (Shaw &
Levin 2013). It is found not only in seabirds but also,
notably, in reptiles and fishes (Tinkle 1962, Thorpe
1994, Solow et al. 2002) which incur lower energetic
costs related to reproduction than birds. Breeding
success tends to follow ecosystem productivity. At
the individual level, the cost of intermittent breeding
corresponds more to a breeding season without re -
productive output than to a loss, while at a popula-
tion level it may result in a temporary loss from a
demographic point of view (Jenouvrier et al. 2005b).
On the other hand it may result in enhancement of
future breeding probabilities, as postulated by the
prudent-parent hypothesis (Le Bohec et al. 2008).
Pre-breeding climatic conditions strongly affect
recruitment age (Soldatini et al. 2014). Here we
extend this research by investigating oceanographic
effects on skipping propensity, with the expectation
that this phenemenon is more frequent in un -
favourable environmental conditions (Cubaynes et
al. 2011). Intermittent breeding was found to be asso-
ciated with a range of factors, including the availa -
bility of food resources, in mammals (Pilastro et al.
2003) and birds (Newton 1995, 1998). In some sea-
bird species, in cluding gulls (Calladine & Harris
1997, Mills et al. 2008), kittiwakes (Cam et al. 1998),
penguins (Jiguet & Jouventin 1999) and petrels (Bar-
braud & Weimerskirch 2005), intermittent breeding
is described in association with environmental con-
straints or, in the case of shearwaters (Sanz-Aguilar
256
Author copy
Soldatini et al.: Forecasting petrel demography
et al. 2011), as a characteristic of individuals with
a lower intrinsic quality. Furthermore, reproduction
skipping was described as a consequence of incom-
plete primary moult in albatross (Langston & Rohwer
1996).
Research approach and hypotheses
Our study focuses on tracing the effect of oceano-
graphic conditions on the demography of the Medi-
terranean subspecies of the European storm petrel
Hydrobates pelagicus melitensis (‘storm petrel’ here-
after). Most of the known colonies of this subspecies
(Mante & Debize 2012) are located close to continen-
tal slopes, with larger colonies closer (5–6 km for
Marettimo, the colony investigated in this study, and
Filfla, the other major colony known for the species)
than smaller colonies (found at a distance of between
20 and 40 km), and all are in areas where up-welling
occurs (Agostini & Bakun 2002, Massetti 2004). Win-
tering storm petrels are thought to move to the Albo-
ran Sea (Soldatini et al. 2014), a particularly nutrient-
rich area that benefits from currents inflowing from
the Atlantic Ocean (Agostini & Bakun 2002, Reul et
al. 2005, Renault et al. 2012).
This is a first attempt at a deeper analysis of the
demography of the Mediterranean storm petrel, con-
sidering population growth as a function of oceano-
graphic conditions (Van Houtan & Halley 2011).
Some demographic traits of the storm petrel were de -
scribed in previous studies (Sanz-Aguilar et al. 2008,
2009, Soldatini et al. 2014), but information on senes-
cence and age effects on the species’ demography is
still lacking. In particular, the differential re sponses
of age classes to environmental constraints shape
survival and reproductive traits at a population level
and probably in fluence the response of populations
to global climate change (Pardo et al. 2013). As a con-
sequence, in order to understand the influence of
future climate on this and other seabird species, an
assessment on how climate is currently related to
population metrics is needed (McClure et al. 2013).
These results can then be projected onto a possible
future scenario in order to model demographic
responses to anticipated climate-related changes, the
magnitude of which is not yet recorded in human his-
tory (McClure et al. 2013). We are thus modelling
demographic traits and their variability in response
to environmental constraints.
Our hypothesis was that changes in population
growth rates are due to variation in skipping fre-
quencies as a consequence of periodically oscillating
climatic conditions. We expected to find a decrease
in population growth rate in cases where there is an
increase in the frequency or magnitude of cold ano -
malies in SST. A similar approach was applied to log-
gerhead sea turtles (Van Houtan & Halley 2011), a
similarly long-lived species whose recent decline
appears to be the result of climatic changes, and for
which further significant declines are predicted in
coming decades, according to available climatic
data. Using more than 20 years of capture− mark−
recapture (CMR) data and > 5400 marked individu-
als, our aim was to investigate the effects of climate
change on this seabird species.
MATERIALS AND METHODS
Study area and species
CMR data were obtained from the capture and
banding of storm petrel chicks and adults on Maret-
timo Island, located in the Sicilian Channel, Italy
(37° 58’ N, 12° 03’ E), between 1991 and 2013. Field-
work was carried out by certified ringers authorized
by the Institute for Research and Environmental Pro-
tection (ISPRA) to handle adults and chicks. The
main colony on the island consists of approximately
2500 pairs (Albores-Barajas et al. 2012). Each year,
the colony was visited at least once between June
and August, and adults and chicks were banded with
stainless steel rings, resulting in a total of 5168
marked individuals (ringing and recapture data avail-
able upon request from the ringing unit of the Univer-
sity of Palermo, www.ornitologiasiciliana.it/ contatti.
htm). Once fledged in September− October, chicks are
not observed at the colony until their first breeding
attempt that usually occurs 1 to 6 yr after fledging.
Recruitment age is negatively affected by cold condi-
tions in the pre-breeding season (Soldatini et al.
2014). Following breeding, adults depart from the
colony. Yearling survival is estimated to be ~22%
and is strongly affected by first year climatic condi-
tions. The survival of juveniles (pre-breeders, 2 to
3 yr old) is also affected by climatic conditions and
estimated to be ~50%. The estimated survival of
breeders is ~92% (Soldatini et al. 2014). Fieldwork
was carried out under permit nos. 1625/2013, 1721/
2012, 3/2011 and unnumbered permits before 2011
from the Marine Protected Area ‘Isole Egadi’. No
ethics committee approval requirements were neces-
sary due to the minimum handling of the individuals
under Italian and Sicilian regional legislation (LN
157/92, LR 105/99 and LR 74/2012).
257
Author copy
Mar Ecol Prog Ser 552: 255–269, 2016
Demographic traits
We applied a multistate CMR analysis in 2 steps.
The first step was to estimate age-specific survival
(age-dependent model) and the second step was to
assess the probability of skipping a breeding event,
and to evaluate the potential effects of time and
environmental covariates on these variables (Lebre-
ton et al. 2009, Frederiksen et al. 2014). For a given
sampling occasion, a storm petrel could be in one of
4 states: a pre-breeder (PB), breeder (B), skipper (S)
or dead (D). The following observations could be
made on a bird in the field: the individual was
detected as chick and coded ‘1’, the individual was
detected as breeding adult and coded ‘2’ or it went
undetected and was coded ‘0’. For the first step we
used data of birds banded as chicks (3342 individu-
als) in the 23 yr of study. We considered the 4 states,
and 2 age classes: young (first year of age) and
adult (older than 1 yr) (for more details see Soldatini
et al. 2014). We tested for heterogeneity in survival
(Pledger et al. 2003) by running the model with and
without heterogeneity. We first ran the model with
heterogeneity (Pradel et al. 2010), using mixture
models with 2 classes of heterogeneity (Péron et al.
2010). Secondly, we ran the age-dependent model
without heterogeneity (only 1 survival class). After-
wards, in both cases, we modelled age as a factor to
ensure maximum flexibility in the modelling of the
survival−age relationship (hence a nonparametric
relationship). We used the Akaike information crite-
rion (AIC) to select among models (Akaike 1973,
Burnham & Anderson 2002). The results of the
goodness of fit (GOF) tests for the multistate model
show the presence of transience and no trap-depen-
dence (Table 1). The transience effect was probably
due to an age effect on survival, which was
accounted for by incorporating age as a qualitative
covariate (Pradel et al. 1997). The CMR models
were fitted using the program E-SURGE (Choquet
et al. 2009b), while GOF tests (Pradel et al. 2005)
were performed using the program U-CARE (Cho-
quet et al. 2009a).
Skipping probability and climatic conditions
We then ran the CMR model again, as a second
step, and considered time-series data of SST in the
Sicilian Channel (Fig. 1) — an area with upwelling
conditions (Massetti 2004) — from 1991 to 2013 as a
proxy for environmental conditions and food avail-
ability (Rayner et al. 2006, Gremillet et al. 2008). We
decided to use March−April SST in the Sicilian
Channel as it is the period when storm petrels are
258
Test component χ2df p
Global test 116.576 55 <0.0001
Test 3G 104.242 37 <0.001
Test M 12.334 18 0.830
Table 1. Results of goodness-of-fit tests for the multistate
capture−mark–recapture model. The model was tested for
transience (Test 3G) and for trap dependence (Test M)
Fig. 1. (a) At-sea sampling points (open and filled circles in
boxes) in the Mediterranean Sea from which time series of
sea surface temperature (SST) were obtained to study the
impact of SST on the demographics of Mediterranean storm
petrels; (b) the 5 sampling points in the Alboran Sea (tri -
angles); (c) the 13 random at-sea sampling points in the
Sicilian Channel subdivided into 3 sub-sets: 5 points from the
southern part of the channel (PB5SST; black dots), 7 points
from the northern part (PB7SST; medium grey circles) and 10
points taken from the total area considered (PB10SST; white
circles); note that sites can belong to multiple groups
Author copy
Soldatini et al.: Forecasting petrel demography
prospecting the colony, and therefore living in the
area and using local resources. Storm petrels feed
on Mysidacea and larval stages of fishes (Albores-
Barajas et al. 2011), which are directly affected by
SST. Hence, sea conditions of March−April will indi-
rectly affect the body condition of storm petrels in
late May, when they usually start breeding. We did
not consider SST in winter in the Sicilian Channel as
in previous analyses it was not correlated to breeding
and survival (Soldatini et al. 2014).
In order to test for carry-over effects from wintering
to breeding periods we considered the winter North
Atlantic Oscillation (Climate Prediction Center 2008b),
due to its broad geographic coverage, and winter
range SST from the Alboran Sea (Soldatini et al.
2014). In this second step analysis, we used the com-
plete CMR dataset of birds ringed as chicks and
adults (5168 individuals).
We defined 3 environmental covariates to examine
the direct effects of sea conditions during pre- breeding
and winter periods on the propensity of skipping a
breeding event: pre-breeding period SST (PBSST),
the winter North Atlantic Oscillation (WNAO), and
SST at wintering locations in the Alboran Sea (wAl).
We selected PBSST data from 3 sets of samples taken
at random at-sea sampling points during the pre-
breeding period (from March to April) from 1991 to
2013. As a proxy of winter conditions, we analysed
North Atlantic Oscillation index data from December
to February and used the average value of the
selected period to represent WNAO (Climate Predic-
tion Center 2008b). The covariate wAl is the principal
component 1 (PC1) factor (82% of ex plained vari-
ance) of a principal component analysis (PCA) of SST
based on 5 sampling points in the Albo ran Sea (Sol-
datini et al. 2014). We also used the El Niño Southern
Oscillation index (ENSO) for January to March (Cli-
mate Prediction Center 2008a).
Data were from NASA Earth Observations prod-
ucts: SST data from the NOAA/NASA Advanced Very
High Resolution Radiometer (AVHRR) Path finder Pro-
ject and imagery processed by NASA Ocean Color
Group from Moderate Resolution Imaging Spectrora-
diometer (MODIS) on NASA’s Aqua satellite; all freely
available at http://neo.sci.gsfc. nasa. gov/view.php?
data setId=MYD28M.
In order to identify the area where oceanic condi-
tions during the pre-breeding period may directly
affect storm petrel reproductive behaviour, we selec -
ted 13 random at-sea sampling points in the Sicilian
Channel and then subdivided the area into 3 subsets:
5 sampling points from the southern half of the Sicil-
ian Channel (PB5SST), 7 sampling points from the
northern half (PB7SST) and 10 sampling points taken
from the total area considered in the study (PB10SST;
Fig. 1). We expressed oceanographic covariates (SST
of the 3 sets of at-sea sampling points) as anomalies
relative to the mean [anomaly = (value − mean)/
mean] (Jenouvrier 2013). We performed 3 PCAs, one
for each set of sampling points, in order to define
variables to be used in the model (Grosbois et al.
2008) (see Supplement 1 at www.int-res. com/ articles/
suppl/m552p255_supp.pdf). The pre- bree d ing period
SST was summarized through PCAs as 3 external
covariates: PC1 of the PB10SST (94.7% of explained
variance), PC1 of PB7SST: (95.4% of ex plained vari-
ance) and PC1 of PB5SST (93.8% of explained vari-
ance). We then used the PC1 factors, representatives
of SST variation in the areas considered, as explana-
tory environmental covariates in the CMR models.
PC factors embody cumulative direction of variation
of the variables considered; in this study, negative
PC1 values correspond to higher SST and positive
values to lower SST (see Supplement 1).
In order to choose the best result from multiple sig-
nificance tests of the 3 different covariates represent-
ing SST recorded in the 3 at-sea sampling-point sets,
we applied the false discovery rate (FDR) for multiple
testing following the method of Benjamini & Hoch -
berg (1995). We then applied the analysis of deviance
(ANODEV) to assess significance of each covariate
and calculated the proportion of deviance explained
by these covariates using the R2statistic (Grosbois et
al. 2008).
We estimated the probability of breeders entering
the skipping state (transition from state B to state S).
We tested the effects of the environmental covariates
both on survival and on the probability of skipping;
together with linear effects, we tested quadratic ef-
fects of covariates. We also tested for correlation be-
tween pre-breeding external covariates and number
of researchers’ visits to the colony in order to ex clude
the possibility that the number of visits to the colony
may have independently influenced the results.
PCAs, correlation, FDR, ANODEV and R2statistics
were computed using R v. 2.15.1 (R Development
Core Team 2014) with significance defined as p <
0.05 for all analyses.
Population growth rates and extinction
probabilities
Using survival and reproduction probabilities esti-
mated in this and a previous study (Soldatini et al.
2014) we built Leslie matrices (Tuljapurkar 1993,
259
Author copy
Mar Ecol Prog Ser 552: 255–269, 2016
Caswell 2001, Tuljapurkar et al. 2003) for 11 scenar-
ios and calculated population growth rate (λ) to
investigate the effect of age and climate variability.
We constructed and analysed age-structured matrix
population models (see Supplement 2 at www. int-
res. com/ articles/ suppl/ m552p255_ supp. pdf). Survival
rates in the first year and subsequent years were set
according to CMR model results and age of first
reproduction was set at Year 2 (Soldatini et al.
2014). We did not set an age of last reproduction as
we found 23 year old breeders; thus the possible
breeding period goes beyond the limits of our
dataset.
We modelled different scenarios embodying our
predictions under future climate conditions. In Sce-
nario A, we accounted for higher mortality at Years 6
to 8 and possible senescence after Year 16 (using
demographic parameters from this study, see Fig. 2),
while in Scenario B we ignored the age effect and
only considered average survival of juveniles and
adults (demographic parameters from Soldatini et al.
2014). The subsequent 9 scenarios accounted for
senescence (Scenario A) and considered environ-
mental variability. In other words, we modified Sce-
nario A by manipulating the proportion of breeding
individuals according to the skipping model results
(the 9 scenarios and their characteristics are summa-
rized in Table 2).
As a result of the first part of our analysis, we ob -
tained, in some years (1997, 1998, 2003, 2011 and
2012), a strong effect between cold SST anomaly
events (average anomaly −0.37°C) and increased
probability of skipping. We recorded 3 such events
during the study period, approximately once every
5 yr. We extrapolated the SST anomaly threshold of
−0.37°C as representative of extreme conditions cal-
culated as the mean of SST anomalies in the 3 years
with higher skipping probability. We then used a
Bernoulli distribution in order to implement the fre-
quency of SST anomaly events and skipping prob -
ability variation. For each scenario we built a sto-
chastic matrix model and calculated the stochastic
population growth rate and extinction probability
(through simulations) using a pseudo extinction
threshold of 100 individuals and time spans of 100
and 200 yr. Scenarios included real and simulated
skipping probabilities as an effect of the strength and
frequency of SST anomalies. We simulated a situa-
tion with no SST anomaly events and a 15% skipping
probability and then with the actual frequency and
strength of SST anomaly events with 45% skipping
probability, and finally with colder SST anomaly
events and 65% skipping probability. We modified
the percentage of breeding females in the Leslie
matrix using previous values with different frequen-
cies simulating actual conditions (every 5 yr) and
more and less frequent SST anomaly events (i.e.
every 3 and 10 yr; see Table 2).
Data were analysed using the ‘popbio’ (Stubben et
al. 2012) and ‘popdemo’ (Stott et al. 2011, 2012) pack-
ages available for R v. 2.15.0 (R Development Core
Team 2014). We also wrote our own R (partly
reported in Supplement 2) code to run stochastic
demographic models.
RESULTS
Age-dependent survival
We found no heterogeneity in our study popula-
tion (ΔAIC = 6.000, Models 5 and 6 in Table 3). We
ob served a time effect on survival. Senescence
started after Year 16 (Fig. 2) in an age-dependent
model run without heterogeneity on 23 yr of CMR
data. Before Year 16, average survival in adults was
0.95, while from Years 17 to 21 average survival was
0.36, suggesting senescence in the population. We
also observed lower survival probability in Years 6
to 8 (average 0.78) compared to subsequent ages
(Fig. 2).
260
Sce- Skipping Fre- R/M λs after Extinction
nario per- quency 100/200 yr probability
centage of events after
(yr) 100/200 yr
1 65 5 R-R 0.984/0.985 0/0.304
2 65 3 R-M 0.974/0.975 0.126/1
3 65 10 R-M 0.991/0.993 0/0
4 15 5 M-R 0.995/0.997 0/0
5 15 3 M-M 0.993/0.995 0/0
6 15 10 M-M 0.997/1 0/0
7 45 5 M-R 0.989/0.990 0/0
8 45 3 M-M 0.982/0.983 0/0
9 45 10 M-M 0.993/0.995 0/0
Table 2. Description of 9 scenarios where the percentage of
Mediterranean storm petrels skipping reproduction and the
frequency of SST anomaly events in the Sicilian Channel
were manipulated. The frequency of events is shown as the
average return period, e.g. ‘3’ indicates that an event occurs
once every 3 yr. The column headed ‘R/M’ indicates
whether conditions are real or manipulated, the first letter
referring to percentage of skippers and the second to fre-
quency of events. Stochastic population growth rates (λs)
and extinction probability were calculated for each scenario
after 100 and 200 yr
Author copy
Soldatini et al.: Forecasting petrel demography
Skipping probability and oceanographic conditions
We used PB10SST, PB7SST and PB5SST as covari-
ates in the model testing the influence of oceano-
graphic conditions on skipping probability. The FDR
test showed that all 3 covariates had a significant ef-
fect (Table 4). We thus selected the PB7SST data
based on results of the ANODEV (F1, 28 = 4.364, p =
0.049, see Table 4 for more details) as the pre-breed-
ing period external covariate for subsequent analysis.
Winter conditions were represented by wAl and
WNAO. External covariates were not correlated
(PB7SST vs. wAl: r = 0.419, p = 0.176; PB7SSt vs.
WNAO: r = −0.133, p = 0.683; wAl vs. WNAO: r = 0,
p = 1).
The highest-ranked model is the one that included
oceanographic forcing. In fact, the best model indi-
cated a PBSST condition-dependent probability of en-
tering the skipping state for breeders. Age of individ-
uals and winter condition in the Alboran Sea
influenced survival while local SST in March−April
influenced the probability of breeding both in the
case of recruiting individuals and adult breeders.
Both covariates were demonstrated to have a quad-
ratic effect respectively on survival
and skipping propensity. Local SST
anomaly measured at 7 sampling
points (PB7SST) had an inversely pro-
portional effect on recruitment age
and on the probability of entering the
skipping state for breeders. SST varia-
tion in the northern part of the Sicilian
Channel explained 18% of the tempo-
ral variation in skipping probabilities
(Table 4); this suggests that SST influ-
ences breeding decisions in Marettimo
colonies. High skipping probabilities
were synchronized with years corre-
sponding to peaks in the SST PC1 val-
ues (Fig. 3). We observed lower skip-
ping probabilities (on average 0.11%)
in years with warmer SST in the northern Sicilian
Channel (corresponding to an average ano maly of
+0.37°C). On the other hand, higher skipping proba-
bilities (on average 0.65%) were related to colder
March− April SST (corresponding to an average anom-
aly of −0.37°C) (Fig. 3). However, in years 1997 and
1998 there were high percentages of skipping indi-
viduals not properly predicted by SST (Fig. 3).
261
No. Model Deviance NP AIC
1Φt Pt (no het) 3174.3 45 3264.3
2Φa Pt (no het) 3203.7 45 3293.7
3Φt P (no het) 3278.8 24 3326.8
4Φa P (no het) 3349.2 24 3397.2
5ΦP (no het) 3414.6 4 3422.6
6ΦP (het) 3414.6 7 3428.6
Table 3. Age-dependent model selection, focusing on sur-
vival and encounter probabilities for Mediterranean storm
petrels. Only the best models are presented. The effects of
survival (Φ), time (t), and age (a) were considered on en-
counter probability (P). het = heterogeneity. Deviance, the
number of parameters (NP) and Aikake information criteria
(AIC) values (Burnham & Anderson 2002) are reported. The
preferred model is in bold
Model AIC De- NP ANODEV R2p FDR
viance test (%) p
ΦtΨt10988.4 10932.4 28
ΦtΨPB10SST 11241.1 11225.1 8 0.006 0 0.937 0.010
ΦtΨPB5SST 11196.6 11180.6 8 3.590 15 0.072 0.006
ΦtΨPB7SST 11188.7 11172.7 8 4.364 18 0.049 0.003
ΦiΨi11239.2 11225.2 7
Table 4. Model selection, focusing on transition probabilities (from ‘breeder’ to
‘skipper’) for Mediterranean storm petrels. Sea surface temperature recorded
at 5 (PB5SST), 7 (PB7SST) and 10 (PB10SST) points in the Sicilian Channel.
Time (t), and constant effects (i) were considered on the probability of transi-
tion from ‘breeder’ to ‘skipper’ (Ψ). AIC values, deviance, the number of para -
meters (NP), results of analysis of deviance (ANODEV), R2tests, and p-values
based on the F-distribution and the false discovery rate (FDR) are reported.
The preferred model is in bold
Fig. 2. Estimated sur-
vival probability for
Mediterranean storm
petrels as a function of
age (yr). Grey dotted
lines show 95% confi-
dence intervals. The es-
timates are from the
best mo del, Φt Pt (no
het) (Model 1 in Table 3)
Author copy
Mar Ecol Prog Ser 552: 255–269, 2016
When considering wintering conditions effects at
the probable wintering range or on a wider basis, we
detec ted an additive effect of age and wintering
range conditions (i.e. wAl) on survival, while pre-
breeding pe riod SST (i.e. PB7SST) influenced skip-
ping probability (Table 5, Fig. 4). WNAO and ENSO
had a negligible influence on skipping probability.
Our results were robust in capturing effort variabil-
ity in the study pe riod, as we found no correlation be -
tween PB7SST and catch effort (r = 0.121).
Population growth rate
We found a difference in long term
population growth rate between Sce-
narios A and B (λA= 1.005 and λB=
1.032) indicating that the population
is viable in a deterministic environ-
ment according to the minimum
viable population definition, i.e. when
the smallest population has at least a
95% chance of persistence in 200 yr
(Newton 1998).
In the 9 scenarios analyzed, as a
modification of Scenario A obtained
by accounting for real and po tential
environmental variability constraints,
we estimated stochastic population
growth rates from 0.974 to 0.997
(Table 2) in the next 100 yr with gen-
erally no pseudo-extinction probabil-
ity, with the exceptions of Scenario 1
considering the next 100 yr and Sce-
narios 1 and 2 considering the next
200 yr.
In fact, Scenarios 1 and 2 were the
exception, where the continuation of
the current situation (Scenario 1) or
an increase in frequency of SST
anomaly events would lead the
colony to extinction. The analysis of
these scenarios suggested that this
population may decline in the future.
262
MDL NP Deviance AIC ΔAIC
Φa+wAl2ΨPB7SST2Pt 53 9726.134 9832.134 75.507
Φa+wAl ΨPB7SST Pt 53 9801.641 9907.641 489.578
Φa+wAl ΨPB7SST Pa 53 10291.219 10397.219 41.791
Φa+wAl ΨPB7SST 46 10347.010 10439.010 2.602
Φa+PB7sst ΨwAl 46 10349.613 10441.613 6.324
Φa+PB7sst2ΨwAl2 46 10355.938 10447.938 170.252
Φa+wAl ΨPB7SST+ENSO Pt 56 10510.11 10618.19.13 3.94
Φa+wAl2ΨPB7SST2+ENSO2Pt 56 10510.11 10622.13 13.436
Φt Ψt 85 10465.566 10635.566 111.989
Φa Ψa 64 10619.555 10747.555 1.536
ΦwAl ΨPB7SST Pt 31 10687.091 10749.091 234.602
ΦPB7SST Ψa 44 10895.693 10983.693 4.709
ΦΨt 28 10932.402 10988.402 41.344
Φa Ψ1aΨ2t 64 10901.746 11029.746 158.986
ΦΨPB7SST 8 11172.732 11188.732 34.575
ΦΨWNAO 8 11207.307 11223.307 15.866
Φi Ψi 7 11225.173 11239.173 2
ΦΨwAl 8 11225.173 11241.173 28.318
Φa ΨWNAO 44 11181.491 11269.491 1.682
ΦΨ1a Ψ2wAl 23 11225.173 11271.173 0
Φa ΨPB7SST 23 11225.173 11271.173 1.879
ΦwAl ΨwAl 24 11225.052 11273.052 0
ΦwAl Ψ1a Ψ2PB7SST 24 11225.052 11273.052 8.933
Φa ΨPB7SST 45 11191.985 11281.985 0.382
Φa Ψ1PB7SSTΨ2wAl 44 11194.367 11282.367 0
Φa ΨwAl 44 11194.367 11282.367
Table 5. Influence of environmental variables on Mediterranean storm petrel
demographics: model selection results. The effects of sea surface temperature
recorded at 7 sampling points in the Sicilian channel (PB7SST), winter NAO
(WNAO) recorded in December through February, SST in the Alboran Sea,
the potential wintering range recorded in December through February (wAl),
El Niño Southern Oscillation index (ENSO), age (a), time (t), and constant ef-
fects (i) on survival (Φ), the transition probability from state ‘breeder’ to ‘skip-
per’ (Ψ) and encounter probability (P). Number of parameters (NP), deviance,
AIC and ΔAIC values are reported. The preferred model is in bold
Fig. 3. Linear regression of the skipping probability
for Mediterranean storm petrels estimated by the time-
dependent (dep.) model (Φt Ψt, see Table 5 for details) as a
function of values of the first principal component factor
(PC1) of a principal component analysis (PCA), based on
SST of 7 at-sea sampling points during the pre-breeding
period (PB7SST). The regression line is shown in gray,
points represent single year values
Author copy
Soldatini et al.: Forecasting petrel demography
Present conditions, with SST anomalies recorded
once every 5 yr period, will lead to a population
reduction during the following de cade (Scenario 1)
and to probable extinction in the next 200 yr. More
frequent (every 3 yr) SST ano malies would reduce
population growth rate and probably lead to a high
risk of extinction (probability of extinction 12.6% in
100 yr and 100% in 200 yr; see Scenario 2 in Table 2).
Scenario 3, with SST anomalies simulated to occur
every 10 yr, resulted in the highest population
growth rate out of the first 3 scenarios, although the
value is lower than that estimated for constant envi-
ronmental conditions (Scenario A). Simulating an
improvement in sea condition (i.e. reduction in fre-
quency of SST anomalies) with reference to this
warm-condition selected species, we obtained a
smaller proportion of skippers (15 or 45 % of breed-
ers) and population growth rates lower than those in
a constant environment (i.e. no in crease or decrease
in SST anomalies), but in no case leading to pseudo-
extinction in the next 100 or 200 yr. An SST anomaly
event every 10 yr would probably reduce the popula-
tion growth rate without leading it to extinction (Sce-
narios 3, 6 and 9). Scenarios 4, 5 and 6 simulated
lower skipping probabilities, which could occur as a
result of the adaptation of the species to SST anom-
alies or a reduction of SST cold anomalies. This
would result in an improvement of population
growth rates and a zero probability of extinction. On
the other hand, a reduction in the proportion of skip-
pers, in the case of favorable change of oceano-
graphic conditions or as a consequence of a new
adaptive strategy of the population to un favourable
conditions (e.g. shifting wintering range or diet),
would result in an increase in population growth
rates (Scenarios 5, 6 and 9).
DISCUSSION
Storm petrel mortality and senescence
Demographic and ecological traits of Medi ter -
ranean storm petrels have recently been studied
263
Probability of skipping
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Winter NAO
–2.0
–1.5
–1.0
–0.5
0.0
0.5
1.0
1.5
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Local pre-breeding SST (°C)
14.8
15.0
15.2
15.4
15.6
15.8
16.0
16.2
16.4
1998–99
1999–00
2000–01
2001–02
2002–03
2003–04
2004–05
2005–06
2006–07
2007–08
2008–09
2009–10
Pot. winter range SST (°C)
14.2
14.4
14.6
14.8
15.0
15.2
15.4
15.6
ab
d
c
Fig. 4. Temporal trends in skipping probability for Mediterranean storm petrels and environmental conditions: (a) skipping
probability estimated from the model Φa+wAl ΨPB7SST Pt (see Table 5 for details) that accounts for the effect of age on sur-
vival and wintering site plus the effect of pre-breeding local site SST on skipping probability; (b) winter NAO; (c) local pre-
breeding period SST and (d) SST in the potential winter range, in the Alboran Sea. Panels (b) and (d) represent winter period
datasets (December to February), so each label along the x-axis indicates 2 yr
Author copy
Mar Ecol Prog Ser 552: 255–269, 2016
thanks to the availability of long time series of
CMR data (Sanz-Aguilar et al. 2008, 2009, 2010,
Soldatini et al. 2014). This has provided information
on differential age class survival and recruitment
of the species (Sanz-Aguilar et al. 2009, Soldatini
et al. 2014), as the basis for further demographic
analysis. In our study, we observed a reduction in
survival probability after Year 16. This suggests
that senescence starts at this age and thus that the
population can be divided into 3 age classes: juve-
niles, adults and senescing individuals, as in other
long-lived seabird species (Pardo et al. 2013).
Senescing individuals have higher mortality com-
pared to younger adults and could also be influ-
enced by reproductive senescence as a conse-
quence of altered foraging capacities of older
individuals, in accordance with similar findings for
black-browed albatross (Pardo et al. 2013). We
found similar λin Scenarios A and B (respectively
accounting or not for senescence); this may be
explained by the fact that there were very few
females alive after Year 16. Moreover, these
females had a low reproductive value, therefore
contributed little to the population growth rate and
in turn had a small sensitivity. The estimates of the
ef fect of senescence on population growth rate
should be considered with caution and a more
detailed description of the effect of senescence on
population dynamics including reproductive senes-
cence and its covariation with actuarial senescence
would help in completing the picture (Robert et al.
2015).
Pre-breeding period environmental conditions
influence skipping propensity
In addition to the mortality factor, population
growth rate is obviously constrained by factors af -
fecting reproductive behaviour, which thus helps
shape storm petrel population dynamics. Of parti -
cular relevance to reproductive output is the pre-
breeding period, when individuals accumulate the
resources required to afford breeding. Our results
suggest that temperature is a good indicator of eco-
system state for these birds as they are sensitive to
average anomalies of less than 1°C. In particular, SST
in the northern Sicilian Channel in March−April is
important for the species, producing conditions that
are suitable or not for breeding: high SST (average
anomaly of +0.37°C) coincides with high breeding
probability, while low SST (average anomaly of
−0.37°C) will induce breeders to skip reproduction.
This latter situation probably coincides with un -
favourable conditions in terms of prey availability in
the Sicilian Channel. Although primary productivity
at lower trophic levels may be increased by deep/
cool water upwelling enhancing the circulation of
nutrients, negative SST anomalies may have a nega-
tive physiological influence, delaying fish develop-
ment (Lafuente et al. 2002). Studies on fish popula-
tion in this area have shown that sardine and
anchovy biomass are negatively correlated with the
mean SST in periods corresponding to larval and
juvenile growth, resulting in a reduction in recruit-
ment success and thereby also population size (e.g.
Lafuente et al. 2002). In particular, Marettimo storm
petrel colonies are located at the western inlet of the
Sicilian Channel, where the 2 large Mediterranean
sub-basins meet, resulting in the bifurcations of the
surface and intermediate-depth water transports
(Pierini & Rubino 2001). The area is characterized by
seasonal wind-induced water movements, particu-
larly in the spring period when winds from the north
over the western basin and winds from the south-east
over the the eastern basin meet, generating Ekman
transport (Bakun & Agostini 2001). Geostrophic fronts
in the area facilitate the pumping of nutrients from
deep layers, and warm SSTs speed up the develop-
ment of eggs and larvae (Lafuente et al. 2002). When
waters are colder, this process may be altered, affect-
ing storm petrels whose diet depends on larval and
juvenile stages of fishes (Albores-Barajas et al. 2011).
Further investigation on the sensitivity of the marine
food web and possible mismatches caused by vari-
able oceanographic conditions (Ramírez et al. 2016)
would clarify this point.
Only strong climatic oscillations influence skipping
propensity
An interesting point is that in years 1997 and
1998 we observed a high percentage of skipping
not correlated to SST. We did not observe a signif-
icant effect of ENSO on our study population but
we hypothesize that petrels do detect an ENSO
signal but only during exceptionally strong events,
given that strong ENSO years correspond to
higher skipping probability. This observation may
support the results of climatological studies show-
ing that exceptionally strong El Niño events have
an influence on North Atlantic and Mediterranean
rainfall regimes and SST (Pozo-Vázquez et al.
2001, Mariotti et al. 2002, Shaman & Tziperman
2011).
264
Author copy
Soldatini et al.: Forecasting petrel demography
Carryover effect of winter conditions
We found that winter environmental conditions af -
fected survival, in addition to constraints posed by
pre-breeding SST on skipping. A possible explana-
tion is that survivors are likely recovering from a bad
winter, of which low SST in the Sicilian Channel is
probably a consequence. Thus, birds are more likely
to skip breeding after cold winters and in cold pre-
breeding periods. Our findings that the western
Mediterranean basin conditions during winter and
spring affect the probability of breeding in the Medi-
terranean storm petrel is in agreement with studies
on other species (Reed et al. 2015).
Forecasting the impacts of ocean warming
In the case of SST rise as a consequence of climate
change (Kovats et al. 2014), changes in fluid circula-
tion may have a strong effect on storm petrel demo -
graphy. The fact that >60 % of the potential breeders
skip reproduction in response to a negative SST
anomaly of −0.37°C is alarming. The high degree of
philopatry suggests that changes negatively affect-
ing reproduction in the Marettimo population, the
largest known colony of these storm petrels (Mante &
Debize 2012), may have catastrophic consequences
for the subspecies. Seabirds that have adapted to
periodic oscillations of the marine environment may
not move from their breeding areas even if these
have become sub-optimal. Instead, they may opt to
skip reproduction (Cubaynes et al. 2011). Our find-
ings are in accordance with the prudent parent
hypothesis (Drent & Daan 1980, Le Bohec et al. 2007)
as birds are synchronous in skipping reproduction in
years with unfavourable conditions. This strategy
may enhance adaptation to a changing environment,
as a form of bet-hedging (Crean & Marshall 2009,
Ripa et al. 2010). Alternatively it may have a link to
breeder intrinsic quality as in other Mediterranean
seabird species (Sanz-Aguilar et al. 2011).
By considering SST anomalies as the cause of
reduction in fecundity, we found that the increase in
skipping frequency directly and negatively affected
the population growth rate. Given that Scenario 1
closely reflects current conditions, we may expect the
Marettimo population to decrease and face risk of
extinction, i.e. falling below the minimum viable
population threshold (Newton 1998), in the next
200 yr. The decreasing trend projected for this popu-
lation (around 22% in 100 yr) is in accordance with
general seabird population trends, which have
recorded a decrease of 69.7 % in the last 60 yr
(Paleczny et al. 2015). In addition to environmental
stochasticity, human activities such as commercial
fisheries may have a direct or indirect effect on
petrels by depleting the standing and spawning bio-
mass of prey species. Taking into account this human
impact on the trophic web, and its likely indirect
effect on the skipping propensity of the storm petrel,
would complete the picture, probably revealing that
the situation is worse than previously thought.
Changes in the eastern Mediterranean SST show a
~5% annual increase over the period 1990 to 2014
(Van Houtan et al. 2015). In the case of an increase in
frequency of cold winter conditions, the Marettimo
population would manifest an increase in skipping
years with an overall reduction in reproductive out-
put, while the population may not be negatively
affected by SST changes if the present trend is con-
firmed. In fact, a concrete improvement in the demo-
graphic trend would result from a decrease in the fre-
quency of cold SST anomalies and a general increase
in temperatures, as forecasted (Parry et al. 2007,
Giorgi & Lionello 2008, Kovats et al. 2014). Further-
more, in years with warmer conditions, lower chick
mortality is also expected due to drier conditions in
nesting caves (Soldatini et al. 2014). On the other
hand, future climate projections are complex, and
comprise changes other than warming, especially for
upwelling systems. In fact, a cautious approach is
needed, as high breeding success during present
warm SST conditions in the Mediterranean does not
mean that the trophic dynamics in the future will
support a similar food web. Storm petrels, and sea-
birds in general, would be negatively affected by
an upwelling displacement or shutdown due to a
warmer climate (Karnauskas et al. 2015).
Concluding remarks
This study probably underestimates λvalues, as we
only considered data from accessible nests (~10 %) of
a colony of about 2500 individuals (Albores-Barajas
et al. 2012), and for this reason, our results may be
biased toward a worst-case scenario. The high philo -
patry and resultant sensitivity to climatic niche varia-
tion of the species suggest that the population has a
generally low level of robustness to climate change
(Jenouvrier 2013). With particular reference to the
forecasted conditions, the climatic issue may not
negatively affect the species, although the effects of
anthropogenic overexploitation of sea resources re -
main uncertain.
265
Author copy
Mar Ecol Prog Ser 552: 255–269, 2016
Our particular goal was to unravel demographic
traits (senescence and skipping behaviour) of a poorly
known seabird species and link them to environmen-
tal variability using a forecasting approach. Our find-
ings confirm that population demography in marine
systems is sensitive to environmental variation
through interactions between mortality and repro-
duction-dependent effects, in agreement with studies
from other geographic areas (Barbraud & Weimer-
skirch 2003). Our results provide strong evidence for
the influence of oceanographic variables on seabird
skipping propensity, while climatic indi ces are less
significantly related. Other Mediterranean seabird
species such as Scopoli’s shearwater Calonectris dio -
medea and Yelkouan shearwater Puffinus yelkouan,
are most probably facing similar reproductive con-
straints due to environmental variation (Hamer 2010,
Sanz-Aguilar et al. 2011). Confinement to the Medi-
terranean basin may present a problem for these sea-
bird species, as the Mediterranean allows only limited
north−south shifts of habitat and in phenology, in con-
trast with other seas (Weimerskirch et al. 2003, Bar-
braud & Weimerskirch 2006), thus requiring a high
level of species resilience to environmental variation.
Acknowledgements. We are particularly grateful to David
Koons for making available the codes for the analysis of
environmental stochasticity, and for his comments that
greatly improved the manuscript. We thank Kyle Van
Houtan and 3 anonymous referees for their useful and
encouraging revision of the manuscript. We are also grateful
to Emanuela Canale, Fabio Lo Valvo, Paolo Lucido and Mar-
cello Tagliavia for helping in the field. Our thanks go to Ale-
jandro Ramos, Dimitri Giunchi and Carlo Catoni for their
valuable help in spatial data processing and to Andrew
O’Reilly-Nugent for language editing. This study was con-
ducted using NASA Earth Observations products. Cecilia
Soldatini was involved in the project CICESE 691111.
LITERATURE CITED
Agostini VN, Bakun A (2002) ‘Ocean triads’ in the Medi -
terra nean Sea: physical mechanisms potentially structur-
ing reproductive habitat suitability (with example appli-
cation to European anchovy, Engraulis encrasicolus).
Fish Oceanogr 11: 129−142
Ainley D, Russell J, Jenouvrier S, Woehler E, Lyver POB,
Fraser WR, Kooyman GL (2010) Antarctic penguin re -
sponse to habitat change as Earth’s troposphere reaches
2°C above preindustrial levels. Ecol Monogr 80: 49−66
Akaike H (1973) Information theory and an extension of the
maximum likelihood principle. In: Petrov BN, Csaki BF
(eds) Proc 2nd Int Symp on Information Theory. Aka -
demiai Kiado, Budapest, p 267−281
Albores-Barajas YV, Riccato F, Fiorin R, Massa B, Torricelli
P, Soldatini C (2011) Diet and diving behaviour of the
Mediterranean storm petrel (Hydrobates pelagicus meli -
tensis). Bird Study 58: 208−212
Albores-Barajas YV, Massa B, LoCascio P, Soldatini C (2012)
Night surveys and smell, a mixed method to detect un -
known colonies of storm petrel (Hydrobates pelagicus).
Avocetta 36: 95−96
Bailey LD, van de Pol M (2016) Tackling extremes: chal-
lenges for ecological and evolutionary research on ex -
treme climatic events. J Anim Ecol 85: 85−96
Bakun A, Agostini VN (2001) Seasonal patterns of wind-
induced upwelling/downwelling in the Mediterranean
Sea. Sci Mar 65: 243−257
Barbraud C, Weimerskirch H (2003) Climate and density
shape population dynamics of a marine top predator.
Proc R Soc B 270: 2111−2116
Barbraud C, Weimerskirch H (2005) Environmental condi-
tions and breeding experience affect costs of reproduc-
tion in blue petrels. Ecology 86: 682−692
Barbraud C, Weimerskirch H (2006) Antarctic birds breed
later in response to climate change. Proc Natl Acad Sci
USA 103: 6248−6251
Barbraud C, Rivalan P, Inchausti P, Nevoux M, Rolland V,
Weimerskirch H (2011) Contrasted demographic respon -
ses facing future climate change in Southern Ocean sea-
birds. J Anim Ecol 80: 89−100
Baron JP, Le Galliard JF, Ferrière R, Tully T (2013) Inter -
mittent breeding and the dynamics of resource allocation
to reproduction, growth and survival. Funct Ecol 27:
173−183
Baum JK, Worm B (2009) Cascading top down effects of
changing oceanic predator abundances. J Anim Ecol 78:
699−714
Benjamini Y, Hochberg Y (1995) Controlling the false dis-
covery rate: a practical and powerful approach to multi-
ple testing. J R Stat Soc B 57: 289−300
Bladé I, Liebmann B, Fortuny D, van Oldenborgh GJ (2012)
Observed and simulated impacts of the summer NAO in
Europe: implications for projected drying in the Medi -
terranean region. Clim Dyn 39: 709−727
Burnham KP, Anderson DR (2002) Model selection and mul-
timodel inference: a practical information-theoretic ap -
proach. Springer, New York, NY
Calladine J, Harris M (1997) Intermittent breeding in the
herring gull Larus argentatus and the lesser black
backed gull Larus fuscus. Ibis 139: 259−263
Cam E, Hines JE, Monnat JY, Nichols JD, Danchin E (1998)
Are adult nonbreeders prudent parents? The kittiwake
model. Ecology 79: 2917−2930
Caswell H (2001) Matrix population models. Sinauer, Sun-
derland, MA
Chambers LE, Devney CA, Congdon BC, Dunlop N, Woeh -
ler EJ, Dann P (2011) Observed and predicted effects of
climate on Australian seabirds. Emu 111: 235−251
Chambers LE, Patterson T, Hobday AJ, Arnould JP, Tuck
GN, Wilcox C, Dann P (2015) Determining trends and
environmental drivers from long-term marine mammal
and seabird data: examples from Southern Australia. Reg
Environ Change 15: 197−209
Choquet R, Lebreton JD, Gimenez O, Reboulet AM, Pradel
R (2009a) U-CARE: utilities for performing goodness of
fit tests and manipulating CApture-REcapture data.
Ecography 32: 1071−1074
Choquet R, Rouan L, Pradel R (2009b) Program E-SURGE: a
software application for fitting multievent models. In:
Thompson DL, Cooch EG, Conroy MJ (eds) Modelling
demographic processes in marked populations. Springer,
New York, NY, p 845−865
266
Author copy
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
Soldatini et al.: Forecasting petrel demography
Climate Prediction Center (2008a) El Niño Southern Oscilla-
tion (ENSO). Accessed 20 November 2015. www. cpc.
ncep. noaa.gov/products/precip/CWlink/MJO/enso.shtml
Climate Prediction Center (2008b) North Atlantic Oscillation
(NAO). Accessed 30 July 2015. www. cpc. ncep. noaa. gov/
products/precip/CWlink/pna/nao.shtml
Coll M, Shannon LJ, Moloney CL, Palomera I, Tudela S
(2006) Comparing trophic flows and fishing impacts of a
NW Mediterranean ecosystem with coastal upwelling
systems by means of standardized models and indicators.
Ecol Modell 198: 53−70
Crean AJ, Marshall DJ (2009) Coping with environmental
uncertainty: dynamic bet hedging as a maternal effect.
Philos Trans R Soc Lond B Biol Sci 364: 1087−1096
Croxall JP, Trathan PN, Murphy EJ (2002) Environmental
change and Antarctic seabird populations. Science 297:
1510−1514
Cubaynes S, Doherty PF Jr, Schreiber EA, Gimenez O (2011)
To breed or not to breed: a seabird’s response to extreme
climatic events. Biol Lett 7: 303−306
Drent RH, Daan S (1980) The prudent parent: energetic
adjustment in avian breeding. Ardea 68: 225−252
Forcada J, Trathan PN (2009) Penguin responses to climate
change in the Southern Ocean. Glob Change Biol 15:
1618−1630
Frederiksen M, Furness RW, Wanless S (2007) Regional vari-
ation in the role of bottom-up and top-down processes in
controlling sandeel abundance in the North Sea. Mar
Ecol Prog Ser 337: 279−286
Frederiksen M, Lebreton JD, Pradel R, Choquet R, Gimenez
O (2014) Identifying links between vital rates and envi-
ronment: a toolbox for the applied ecologist. J Appl Ecol
51: 71−81
Giorgi F, Lionello P (2008) Climate change projections for
the Mediterranean region. Global Planet Change 63:
90−104
Giudici A, Navarro J, Juste C, Gonzalez-Solis J (2010)
Physio logical ecology of breeders and sabbaticals in a
pelagic seabird. J Exp Mar Biol Ecol 389: 13−17
Grémillet D, Boulinier T (2009) Spatial ecology and conser-
vation of seabirds facing global climate change: a re -
view. Mar Ecol Prog Ser 391: 121−137
Grémillet D, Lewis S, Drapeau L, van Der Lingen CD and
others (2008) Spatial match-mismatch in the Benguela
upwelling zone: should we expect chlorophyll and sea-
surface temperature to predict marine predator distribu-
tions? J Appl Ecol 45: 610−621
Grosbois V, Gimenez O, Gaillard JM, Pradel R and others
(2008) Assessing the impact of climate variation on sur-
vival in vertebrate populations. Biol Rev Camb Philos Soc
83: 357−399
Halpern BS, Selkoe KA, Micheli F, Kappel CV (2007) Evalu-
ating and ranking the vulnerability of global marine eco-
systems to anthropogenic threats. Conserv Biol 21:
1301−1315
Hamer KC (2010) The search for winners and losers in a sea
of climate change. Ibis 152: 3−5
Hedd A, Bertram DF, Ryder JL, Jones IL (2006) Effects of
interdecadal climate variability on marine trophic inter-
actions: rhinoceros auklets and their fish prey. Mar Ecol
Prog Ser 309: 263−278
Hoegh-Guldberg O, Bruno JF (2010) The impact of climate
change on the world’s marine ecosystems. Science 328:
1523−1528
Jenouvrier S (2013) Impacts of climate change on avian pop-
ulations. Glob Change Biol 19: 2036−2057
Jenouvrier S, Barbraud C, Cazelles B, Weimerskirch H
(2005a) Modelling population dynamics of seabirds:
importance of the effects of climate fluctuations on
breeding proportions. Oikos 108: 511−522
Jenouvrier S, Barbraud C, Weimerskirch H (2005b) Long-
term contrasted responses to climate of two Antarctic
seabird species. Ecology 86: 2889−2903
Jenouvrier S, Tavecchia G, Thibault JC, Choquet R, Bretag-
nolle V (2008) Recruitment processes in long-lived spe-
cies with delayed maturity: estimating key demographic
parameters. Oikos 117: 620−628
Jenouvrier S, Péron C, Weimerskirch H (2015) Extreme cli-
mate events and individual heterogeneity shape life-
history traits and population dynamics. Ecol Monogr 85:
605−624
Jiguet F, Jouventin P (1999) Individual breeding decisions
and long term reproductive strategy in the king penguin
Aptenodytes patagonicus. Ibis 141: 428−433
Karnauskas KB, Jenouvrier S, Brown CW, Murtugudde R
(2015) Strong sea surface cooling in the eastern equato-
rial Pacific and implications for Galápagos penguin con-
servation. Geophys Res Lett 42: 6432−6437
Kovats RS, Valentini R, Bouwer LM, Georgopoulou E and
others (2014) Europe. In: Barros VR, Field CB, Dokken
DJ, Mastrandrea MD and others (eds) Climate change
2014: impacts, adaptation, and vulnerability. Part B:
Regional aspects. Contribution of Working Group II to
the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change. Cambridge University Press,
Cambridge, p 1267–1326
Lafuente JG, Garcia A, Mazzola S, Quintanilla L, Delgado J,
Cuttita A, Patti B (2002) Hydrographic phenomena influ-
encing early life stages of the Sicilian Channel anchovy.
Fish Oceanogr 11: 31−44
Langston NE, Rohwer S (1996) Molt-breeding tradeoffs in
albatrosses: life history implications for big birds. Oikos
76: 498−510
Le Bohec C, Gauthier-Lerc M, Grémillet D, Pradel R, Béchet
A, Gendner JP, Le Maho Y (2007) Population dynamics in
a long lived seabird: I. Impact of breeding activity on sur-
vival and breeding probability in unbanded king pen-
guins. J Anim Ecol 76: 1149−1160
Le Bohec C, Durant JM, Gauthier-Clerc M, Stenseth NC and
others (2008) King penguin population threatened by
Southern Ocean warming. Proc Natl Acad Sci USA 105:
2493−2497
Lebreton JD, Nichols JD, Barker RJ, Pradel R, Spendelow JA
(2009) Modeling individual animal histories with multi-
state capture-recapture models. Advances Animal Res
41: 87−173
Mante A, Debize E (2012) Mediterranean storm petrel,
Hydrobates pelagicus melitensis: updated state of know -
ledge and conservation of the nesting populations of the
Mediterranean small islands. Initiative PIM
Mariotti A, Zeng N, Lau KM (2002) Euro-Mediterranean
rainfall and ENSO — a seasonally varying relationship.
Geophys Res Lett 29: 1621, doi: 10.1029/2001GL014248
Massetti L (2004) Identification and analysis of Mediterra -
nean upwelling areas. MSc thesis, University of Florence
Maxwell SM, Hazen EL, Bograd SJ, Halpern BS and others
(2013) Cumulative human impacts on marine predators.
Nat Commun 4: 2688, doi: 10.1038/ncomms3688
McClure MM, Alexander M, Borggaard D, Boughton D and
others (2013) Incorporating climate science in applica-
267
Author copy
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
Mar Ecol Prog Ser 552: 255–269, 2016
tions of the US Endangered Species Act for aquatic spe-
cies. Conserv Biol 27: 1222−1233
Mills JA, Yarrall JW, Bradford Grieve JM, Uddstrom MJ,
Renwick JA, Merilä J (2008) The impact of climate fluc-
tuation on food availability and reproductive perform-
ance of the planktivorous red billed gull Larus novae -
hollandiae scopulinus. J Anim Ecol 77: 1129−1142
Møller AP, Fiedler W, Berthold P (2004) Birds and climate
change. Advances Ecol Res 35: 235−243
Møller AP, Rubolini D, Lehikoinen E (2008) Populations of
migratory bird species that did not show a phenological
response to climate change are declining. Proc Natl Acad
Sci USA 105: 16195−16200
Moreno J, Møller AP (2011) Extreme climatic events in rela-
tion to global change and their impact on life histories.
Curr Zool 57: 375−389
Newton I (1995) The contribution of some recent research on
birds to ecological understanding. J Anim Ecol 64:
675−696
Newton I (1998) Population limitations in birds. Academic
Press, London
Nye JA, Baker MR, Bell R, Kenny A and others (2014) Eco-
system effects of the Atlantic Multidecadal Oscillation.
J Mar Syst 133: 103−116
Olivier F, van Franeker JA, Creuwels JCS, Woehler EJ
(2005) Variations of snow petrel breeding success in rela-
tion to sea-ice extent: detecting local response to large-
scale processes? Polar Biol 28: 687−699
Paleczny M, Hammill E, Karpouzi V, Pauly D (2015) Popula-
tion trend of the world’s monitored seabirds, 1950−2010.
PLoS One 10: e0129342
Pardo D, Barbraud C, Authier M, Weimerskirch H (2013)
Evidence for an age-dependent influence of environ-
mental variations on a long-lived seabird’s life-history
traits. Ecology 94: 208−220
Parmesan C (2006) Ecological and evolutionary responses to
recent climate change. Annu Rev Ecol Evol Syst 37:
637−669
Parry ML, Canziani OF, Palutikof JP, van der Linden PJ,
Hanson CE (2007) Climate change 2007: impacts, adap-
tation and vulnerability. Contribution of Working Group
II to the fourth assessment report of the Intergovernmen-
tal Panel on Climate Change. Cambridge University
Press, Cambridge
Péron G, Crochet PA, Choquet R, Pradel R, Lebreton JD,
Gimenez O (2010) Capture−recapture models with
hetero geneity to study survival senescence in the wild.
Oikos 119: 524−532
Pierini S, Rubino A (2001) Modeling the oceanic circulation
in the area of the Strait of Sicily: the remotely forced
dynamics. J Phys Oceanogr 31: 1397−1412
Pilastro A, Tavecchia G, Marin G (2003) Long living and
reproduction skipping in the fat doormouse. Ecology 84:
1784−1792
Pledger S, Pollock KH, Norris JL (2003) Open capture−
recapture models with heterogeneity: I. Cormack−Jolly−
Seber model. Biometrics 59: 786−794
Pozo-Vázquez D, Esteban-Parra M, Rodrigo F, Castro-Diez
Y (2001) The association between ENSO and winter
atmo spheric circulation and temperature in the North
Atlantic region. J Clim 14: 3408−3420
Pradel R, Hines JE, Lebreton JD, Nichols JD (1997) Cap-
ture−recapture survival models taking account of tran-
sients. Biometrics 53: 60−72
Pradel R, Gimenez O, Lebreton JD (2005) Principles and
interest of GOF tests for multistate capture−recapture
models. Anim Biodivers Conserv 28: 189−204
Pradel R, Choquet R, Lima M, Merritt J, Crespin L (2010)
Estimating population growth rate from capture−recap-
ture data in presence of capture heterogeneity. J Agric
Biol Environ Stat 15: 248−258
R Development Core Team (2014) R: a language and envi-
ronment for statistical computing. R Foundation for Sta-
tistical Computing, Vienna
Ramírez F, Afán I, Tavecchia G, Catalán IA, Oro D, Sanz-
Aguilar A (2016) Oceanographic drivers and mistiming
processes shape breeding success in a seabird. Proc R
Soc B 283, doi: 10.1098/rspb.2015.2287
Rayner NA, Brohan P, Parker DE, Folland CK and others
(2006) Improved analyses of changes and uncertainties
in sea surface temperature measured in situ since the
mid-nineteenth century: the HadSST2 Dataset. J Clim
19: 446−469
Reed T, Harris M, Wanless S (2015) Skipped breeding in
common guillemots in a changing climate: restraint or
constraint? Front Ecol Evol 3, doi: 10.3389/fevo.2015.
00001
Renault L, Oguz T, Pascual A, Vizoso G, Tintore J (2012) Sur-
face circulation in the Alboran Sea (western Mediterra -
nean) inferred from remotely sensed data. J Geophys Res
Oceans 117: C08009, doi:
10.1029/2011JC007659
Reul A, Rodríguez V, Jiménez-Gómez F, Blanco JM and oth-
ers (2005) Variability in the spatio-temporal distribution
and size-structure of phytoplankton across an upwelling
area in the NW-Alboran Sea, (W-Mediterranean). Cont
Shelf Res 25: 589−608
Ripa J, Olofsson H, Jonzén N (2010) What is bet-hedging,
really? Proc R Soc B 277: 1153−1154
Robert A, Chantepie S, Pavard S, Sarrazin F, Teplitsky C
(2015) Actuarial senescence can increase the risk of
extinction of mammal populations. Ecol Appl 25: 116−124
Sanz-Aguilar A, Tavecchia G, Pradel R, Minguez E, Oro D
(2008) The cost of reproduction and experience-dependent
vital rates in a small petrel. Ecology 89: 3195−3203
Sanz-Aguilar A, Massa B, Lo Valvo F, Oro D, Minguez E,
Tavecchia G (2009) Contrasting age specific recruitment
and survival at different spatial scales: a case study with
the European storm petrel. Ecography 32: 637−641
Sanz-Aguilar A, Tavecchia G, Minguez E, Massa B and oth-
ers (2010) Recapture processes and biological inference
in monitoring burrow-nesting seabirds. J Ornithol 151:
133−146
Sanz-Aguilar A, Tavecchia G, Genovart M, Igual JM, Oro D,
Rouan L, Pradel R (2011) Studying the reproductive skip-
ping behavior in long-lived birds by adding nest inspec-
tion to individual-based data. Ecol Appl 21: 555−564
Shaman J, Tziperman E (2011) An atmospheric teleconnec-
tion linking ENSO and southwestern European precipi-
tation. J Clim 24: 124−139
Shaw AK, Levin S (2013) The evolution of intermittent
breeding. J Math Biol 66: 685−703
Soldatini C, Albores-Barajas YV, Massa B, Gimenez O
(2014) Climate driven life histories: the case of the Medi-
terranean storm petrel. PLoS One 9: e94526
Solow AR, Bjorndal KA, Bolten AB (2002) Annual variation
in nesting numbers of marine turtles: the effect of sea
surface temperature on re-migration intervals. Ecol Lett
5: 742−746
Stommel H (1963) Varieties of oceanographic experience.
Science 139: 572−576
268
Author copy
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
Soldatini et al.: Forecasting petrel demography
Stott I, Hodgson DJ, Townley S (2011) popdemo: provides
tools for demographic modelling using projection matri-
ces. https: //cran.r-project.org/web/packages/ popdemo/
index.html
Stott I, Hodgson DJ, Townley S (2012) popdemo: an R pack-
age for population demography using projection matrix
analysis. Methods Ecol Evol 3: 797−802
Stubben C, Milligan B, Nantel P (2012) popbio: construction
and analysis of matrix population models. https: //cran.r-
project.org/web/packages/popbio/index.html
Sydeman WJ, Thompson SA, Kitaysky A (2012) Seabirds
and climate change: roadmap for the future. Mar Ecol
Prog Ser 454: 107−117
Thorpe JE (1994) Reproductive strategies in Atlantic salmon,
Salmo salar L. Aquacult Res 25: 77−87
Tinkle DW (1962) Reproductive potential and cycles in
female Crotalis atrox from northwestern Texas. Copeia
1962: 306−313
Tuljapurkar S (1993) Entropy and convergence in dynamics
and demography. J Math Biol 31: 253−271
Tuljapurkar S, Horvitz CC, Pascarella JB (2003) The many
growth rates and elasticities of populations in random
environments. Am Nat 162: 489−502
Van Houtan KS, Halley JM (2011) Long-term climate forc-
ing in loggerhead sea turtle nesting. PLoS One 6:
e19043
Van Houtan KS, Halley JM, Marks W (2015) Terrestrial
basking sea turtles are responding to spatio-temporal
sea surface temperature patterns. Biol Lett 11: 20140744,
doi:10.1098/rsbl.2014.0744
Walther GR, Post E, Convey P, Menzel A and others (2002)
Ecological responses to recent climate change. Nature
416: 389−395
Weimerskirch H, Inchausti P, Guinet C, Barbraud C (2003)
Trends in bird and seal populations as indicators of a sys-
tem shift in the Southern Ocean. Antarct Sci 15: 249−256
269
Editorial responsibility: Scott Shaffer,
San Jose, California, USA
Submitted: August 28, 2015; Accepted: April 11, 2016
Proofs received from author(s): May 30, 2016
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
Author copy
➤
➤
➤
➤
➤
➤
➤
➤
➤
➤
