ArticlePDF Available

Abstract and Figures

Despite contrasting population trends ranging from –3 to +11% per annum, the annual survival rates of Atlantic puffins Fratercula arctica in the 5 colonies spanning the species range in the east Atlantic were virtually identical over a 10 to 15 yr period, giving no support to the hypothesis that variation in population growth rates is driven by adult survival. The extent to which survival varied among years differed markedly between colonies. Similarities between colonies in the patterns of inter-annual variation in survival did not reflect similarities in wintering areas, as indicated by recoveries of ringed birds, nor the geographic proximity of the colonies. However, survival in 4 of the 5 colonies correlated with sea surface temperatures around each colony 2 yr earlier. The relationship between survival and sea temperature was positively correlated with the effects of sea temperature on recruitment of the Atlantic puffin’s main prey species, the lesser sandeel Ammodytes marinus, the herring Clupea harengus and the capelin Mallotus villosus.
Content may be subject to copyright.
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 297: 283296, 2005
Published August 1
INTRODUCTION
In long-lived birds that have delayed sexual maturity
and a small clutch size, the annual survival of breeding
adults is often seen as the main factor influencing the
rate of population change (Croxall & Rothery 1991, but
see Nur & Sydeman 2000). The Atlantic puffin Frater-
cula arctica is such a species, having an annual sur-
vival rate of adults of 90 to 95%, an age at first breed-
ing of 5 yr, and an invariant single-egg clutch (Harris
1984). It is a highly abundant piscivorous species,
endemic to the colder parts of the North Atlantic
Ocean, breeding from the Bay of Fundy in the west,
Brittany in the east, and north to the high arctic. Popu-
lations in various parts of the range have shown very
different population trajectories over the past few
decades. Thus numbers in colonies in the North Sea
and the far NE of Norway have increased at rates of up
to 11% yr
–1
at the same time that populations in central
Norway have declined at 3% yr
–1
(Barrett 2001, Anker-
Nilssen & Aarvak 2004, Harris & Wanless 2004). The
survival of adult Atlantic puffins has been monitored
for several colonies spanning almost the entire latitudi-
nal range in the E Atlantic, and the results allowed us
to investigate the hypothesis that changes in numbers
are governed by differences in adult survival.
© Inter-Research 2005 · www.int-res.com*Email: mph@ceh.ac.uk
Effect of wintering area and climate on the survival
of adult Atlantic puffins Fratercula arctica in the
eastern Atlantic
Michael P. Harris
1,
*
, Tycho Anker-Nilssen
2
, Robin H. McCleery
3
,
Kjell Einar Erikstad
4
, Deryk N. Shaw
5
, Vladimir Grosbois
6, 7, 8
1
NERC Centre for Ecology and Hydrology, Hill of Brathens, Banchory, Aberdeenshire AB31 4BW, UK
2
Norwegian Institute for Nature Research (NINA), 7485 Trondheim, Norway
3
Edward Grey Institute, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK
4
Norwegian Institute for Nature Research (NINA), The Polar Environmental Centre, 9296 Tromsø, Norway
5
Fair Isle Bird Observatory Trust, Fair Isle, Shetland ZE2 9JU, UK
6
University of Antwerp, Campus Drie Eiken, Department of Biology, 2610 Antwerp, Belgium
7
Lighthouse Field Station, University of Aberdeen, Cromarty, Ross-shire IV11 8YJ, UK
8
Present address: Centre d’Ecologie Fonctionnelle et Evolutive, Unité Mixte de Recherche 5175, 1919 route de Mende,
34293 Montpellier cedex 5, France
ABSTRACT: Despite contrasting population trends ranging from –3 to +11% per annum, the annual
survival rates of Atlantic puffins Fratercula arctica in the 5 colonies spanning the species range in the
east Atlantic were virtually identical over a 10 to 15 yr period, giving no support to the hypothesis that
variation in population growth rates is driven by adult survival. The extent to which survival varied
among years differed markedly between colonies. Similarities between colonies in the patterns of
inter-annual variation in survival did not reflect similarities in wintering areas, as indicated by recov-
eries of ringed birds, nor the geographic proximity of the colonies. However, survival in 4 of the 5
colonies correlated with sea surface temperatures around each colony 2 yr earlier. The relationship
between survival and sea temperature was positively correlated with the effects of sea temperature
on recruitment of the Atlantic puffin’s main prey species, the lesser sandeel Ammodytes marinus, the
herring Clupea harengus and the capelin Mallotus villosus.
KEY WORDS: Adult survival · North Atlantic Oscillation · Sea surface temperature · Population
regulation
Resale or republication not permitted without written consent of the publisher
Mar Ecol Prog Ser 297: 283296, 2005
Little detailed information is available for the
Atlantic puffin when it is away from the breeding
colonies in the E Atlantic, but the species occurs
mainly well offshore, with birds dispersed at very low
density over vast areas of ocean from the Barents Sea
south to the Azores and the Canary Islands, west to
Greenland and Newfoundland and east to Italy in the
Mediterranean (Harris 1984, 2002, Stone et al. 1995,
Bakken et al. 2003). Ringing of Atlantic puffins in
colonies throughout the range suggested that adults
from different populations have differing wintering
areas (Myrberget 1973, Anker-Nilssen & Tatarinkova
2000, Harris 2002) and so might be influenced in differ-
ent ways by climate change. We investigated factors
influencing adult survival in these colonies using 2
approaches. First, we determined the pattern of corre-
lation between time series of adult survival rates in 5
colonies and examined whether this pattern was com-
patible with the degree of overlapping of the wintering
areas of the birds as assessed from ring recoveries
and/or with the pattern of correlation between time
series of local environmental variables in the vicinity of
the 5 colonies. Second, we addressed directly the cor-
relation between adult survival rates and large- and
local-scale environmental variables.
MATERIALS AND METHODS
Study colonies. The survival of Atlantic puffins was
studied for 5 colonies: Skomer (Wales), Isle of May and
Fair Isle (Scotland), and Røst and Hornøya (Norway),
covering almost 20° of latitude (see Fig. 1). Details of
the colonies are given in Table 1. At the beginning of
the study the breeding population size of these
colonies varied from 7000 pairs (at Skomer) to 660 000
pairs (at Røst) and the current population status also
showed contrasting trends from rapid increase (Isle
of May, Fair Isle and Hornøya), through stability
(Skomer) to consistent decline (Røst) (Barrett 1983,
Anker-Nilssen & Røstad 1993, Anker-Nilssen & Øyan
1995, Harris et al. 2003, Harris & Wanless 2004). In all
colonies the diet of chicks (and apparently adults) com-
prises small fishes, notably the lesser sandeel Ammo-
dytes marinus on Skomer, Isle of May and Fair Isle,
young herring Clupea harengus on Røst, and sandeels
and capelin Mallotus villosus on Hornøya (Harris 1984,
Anker-Nilssen 1992, Barrett 2002, Durant et al. 2003,
Anker-Nilssen & Aarvak 2004). Breeding success also
varied greatly between colonies, being high at most
colonies, but lower and more variable on Røst (Harris
1984, Anker-Nilssen & Øyan 1995, Barrett 2002, Durant
et al. 2003, Anker-Nilssen & Aarvak 2004)
Bird capture histories. In each colony, breeding
adults were caught and marked with a numbered
metal ring, and either an individually coded colour-
ring (Røst from 1998 onwards) or a unique combination
of colour rings (other colonies and Røst 1990 to 1997).
In each subsequent year, visual searches were made
for these birds predominantly in the areas where they
had been ringed, but also in other parts of the colony.
Throughout the period, additional birds were marked
to replace those disappearing. The years for which sur-
vival was assessed and the numbers of adults involved
are shown in Table 1. The actual data in the form of
m-arrays are presented in Appendix 1. On the Isle of
May and Røst, most birds were sexed by bill or head +
bill measurements (Harris 1984, Anker-Nilssen &
Brøseth 1998). The sex of ringed individuals from the
other colonies was unknown.
Winter dispersal data. Outside the breeding season,
Atlantic puffins are pelagic and are rarely seen close to
284
Table 1. Fratercula arctica. Characteristics of the studied colonies. References are given in the text. SST: sea surface temperature;
Period covered: period covered by capture-resighting data set; Sample size: size of same data set; Breeding success: this was
lagged 5 yr (growth from fledgling to breeding adult); Breeding pairs: number at start of monitoring period
Colony Coordinates Area for Main prey Period Sample Breeding Breeding Population
local SST in breeding covered size success means pairs trend
data season (range) (% yr
–1
)
Skomer 51° 45’ N 51°–53° N Sandeel 19842002 647 0.76 6700
d
0
18’ W 4°– W (0.670.87)
a
Isle of May 56° 11’ N 55°–57° N Sandeel 19842002 408 0.82 12 000 +11.8
34’ W 0°– W (0.690.93)
a
Fair Isle 59° 32’ N 59°–61° N Sandeel 19862002 277 0.72 20 000 +5.1
38’ W 0°– W (0.570.87)
a
Røst 67° 26’ N 66°–68° N Herring 19902002 373 0.42 660 0000 3.2
11° 52’ E 10°–14° E (0.000.96)
b
Hornøya 70° 27’ N 70°–72° N Sandeel and 19902002 689 0.85 8400 +2.0
31° 9’ E 29°–34° E capelin (0.780.92)
c
a
Chicks reared per egg laid;
b
chicks reared per egg hatching;
c
large chicks alive at the end of the field season per egg laid;
d
1988
Harris et al.: Survival of Atlantic puffins
land (Stone et al. 1995). Recoveries of ringed Atlantic
puffins reported between 1 August and 31 March were
obtained from the British and Norwegian Ringing
Schemes. We only used recoveries of birds ringed as
full-grown birds or as chicks that had survived at least
3 yr after ringing, by which time they are regularly
attending colonies. The few recoveries of long-dead
birds in or near the colonies in August were omitted.
Recoveries of local birds close to the Isle of May in
March were excluded since adults at this colony return
from late February onwards. To increase sample sizes
available for analysis, we added recoveries of Atlantic
puffins ringed at other colonies near to Skomer (within
100 km), Isle of May (100 km), Fair Isle (Shetland
colonies within 150 km), Røst (120 km) and Hornøya
(colonies in Norway along 650 km of coast north of
69°N) to those from the colonies themselves.
Environmental factors underlying temporal varia-
tion in survival probability. The limited data available
suggest that most adult Atlantic puffins die during the
nonbreeding season. For instance, between 1990 and
1994, only 1 of 30 birds (3%) that disappeared from
intensively studied burrows on the Isle of May be-
tween one spring and the next did so during the breed-
ing season; the remainder were present at the end of
July but did not return the next April (C. V. Wernham
pers. comm.). Less intensive observations over the rest
of the period provided additional support for this. We
therefore looked for environmental factors that might
have influenced survival outside the breeding season.
The North Atlantic Oscillation (NAO) is an atmospheric
phenomenon that influences winter climatic conditions
over the whole North Atlantic region (Hurrell et al.
2003). We considered NAO in winter from December
to March (hereafter referred to as wNAO) as a poten-
tial proxy for the climatic conditions during the winter.
Furthermore, since wNAO reflects variations in the
dynamics and composition of several North Atlantic
phytoplankton, zooplankton and fish communities
(Planque & Taylor 1998, Ottersen et al. 2001, Arnott &
Ruxton 2002, Attril & Power 2002, Edwards et al. 2002,
Hjermann et al. 2004), we used it as a proxy for food
abundance for the Atlantic puffin during the winter
and/or the breeding season. Since effects of the wNAO
on lower trophic levels might take time to reach higher
levels, wNAOs lagged by 1 and 2 yr were also consid-
ered. Values for the wNAO were downloaded from
www.cgd.ucar.edu/cas/jhurrell/indices.html.
The influence of sea surface temperatures (SST) on
the abundance of forage fishes that are important for
Atlantic puffins is well documented. In particular,
Arnott & Ruxton (2002) found a negative correlation
between summer recruitment of 0-group (first-year)
sandeels in the North Sea and SST during the sandeel
larval period (January to May). Conversely, a positive
relationship between sea temperature during winter
and recruitment of 0-group (first-year) herring has
been documented for the Norwegian and Barents Seas
(Toresen & Østvedt 2000, Sætre et al. 2002). SST in the
vicinity of the Atlantic puffin colonies during January
to May of Year Y and of Year Y–1 are thus likely to be
associated with 0-group and 1-group (second-year)
abundance of key prey species for the Atlantic puffin
during and after the breeding season of Year Y.
We also considered that local environmental condi-
tions (reflected by SST) prevailing in the vicinity of the
colony during the breeding season (May to July) could
affect subsequent survival of adult Atlantic puffins.
Monthly SST blended from ship, buoy and bias-cor-
rected satellite data at a resolution of 1° longitude by 1°
latitude (Reynolds et al. 2002) were obtained from
http://iridl.ldeo.columbia.edu/SOURCES/.IGOSS/.nm
c/.Reyn_SmithOIv2/.monthly/.sst/. For each colony, 6
to 10 cells of the grid were selected such as to repre-
sent an area of sea of about 40 000 km
2
around the
colony under study, and averages were computed for
these cells over 2 periods: the spawning season and
larval period of the main prey species (January to May)
and the breeding season of the Atlantic puffin (May to
July). The pattern of spatial correlation between local
SST time series was investigated using ‘proc corr’ of
SAS Version 8.02 statistical package (SAS 1999).
Capture history analysis. Capture histories of adult
Atlantic puffins were analysed using specific proce-
dures developed to provide robust estimates of survival
probabilities (hereafter referred to as Φ), while account-
ing for potential biases due to variation in resighting
probabilities (hereafter referred to as p). A logit-link
function was used to constrain the estimates of Φ and p
between 0 and 1. The statistical package MARK 5.0
(White & Burnham 1999) was used to obtain maximum
likelihood estimates of survival and resighting proba-
bilities, and fit statistics, under various models.
Goodness-of-fit: Before using capture-resighting
models for hypothesis-testing, a departure model has
to be defined that adequately fits the data (Lebreton et
al. 1992). To check for the goodness-of-fit (GOF) of
departure models and identify the causes of any lack of
fit (i.e. heterogeneities in survival and/or recapture
probabilities), tests were conducted with software U-
CARE 2.0 (Choquet et al. 2001) that assesses the GOF
of the Cormack-Jolly-Seber model, in which survival
and resighting probabilities are fully time dependent
(i.e. model Φ(year), p(year)). The GOF of model Φ(year)
p(year) was assessed independently for the data sets of
each colony and each sex for the 2 colonies for which
this information was available. For all 7 data sets con-
sidered, model Φ(year), p(year) fitted the data very
poorly (all p < 0.01 for the sum of the components of the
GOF test). Examination of the different components of
285
Mar Ecol Prog Ser 297: 283296, 2005
the test revealed that the lack of fit was due to a trap-
happiness effect (all p < 0.0001). Individuals were more
likely to be resighted if they had been resighted on
the previous occasion. This effect probably reflected
heterogeneity in recapture probability within and/or
between capture-resighting histories rather than a
genuine influence of the resighting process on the
probability of resighting in the following year. Hetero-
geneity between capture histories would not be sur-
prising since the probability of resighting seabirds on
breeding grounds is likely to be related to components
of the breeding performance such as intermittent
breeding, probability of failing at an early stage of
reproduction or nest attendance and since numerous
seabird studies have demonstrated inter-individual
heterogeneity in these components of breeding per-
formance (e.g. Cam et al. 1998). Heterogeneity within
individual capture histories could reflect a positive
temporal autocorrelation within individuals in the
components of breeding performance listed above and
could be explained, for example, by age-related varia-
tion in these components (e.g. Wooller et al. 1990). The
investigation of the mechanisms generating hetero-
geneity in resighting probability is beyond the scope
of the present paper, nonetheless, heterogeneity in
recapture probability was accounted for by introduc-
ing a factor ‘years since last resighting’ in the model,
describing the variation in resighting probability
(Pradel 1993). We considered 4 such ‘years since last
resighting‘ classes (1, 2, 3, and more than 3 yr after last
resighting, referred to as ‘a
4
’) in the departure model
for each colony. The other components of the GOF test
(3SR and 3SM) address heterogeneities in survival
probabilities. They did not show any significant depar-
ture from the null hypothesis (all p > 0.15) except for
the Skomer data set (p = 0.04). This slight lack of fit
was allowed for by scaling down the deviance of the
models built for this data set by an over-dispersion
parameter equal to the ratio of the χ
2
statistic associ-
ated with test 3SR+3SM to the number of degrees of
freedom associated with this statistic: 52.3/36 = 1.45.
Reference models: Initially the effects of year on
survival and resighting probabilities, and of years
since last resighting on resighting probabilities were
assessed independently for each colony. This defined a
reference model for each colony that captured the most
important general sources of variation in survival and
resighting probability, without relying on specific
assumptions concerning the covariates underlying
their temporal variation. The most complex model, i.e.
the one with most parameters, considered was Φ(year)
p(a
4
+ year) except for the Isle of May and Røst, where
it was Φ(year + sex + year × sex) p(a
4
+ year + sex + a
4
× sex + year × sex). Starting from this model, a step-
down selection procedure was performed in order to
define a reduced model that adequately fitted the data.
The set of reduced models comprised those in which
only 3 and 2 ‘yr since last resighting’ classes were con-
sidered for resighting probability. The model selection
criterion used combined the parsimony principle and a
form of Akaike’s information criterion modified for
data sparseness (AICc) and, for the Skomer data, for
over-dispersion (QAICc) (Lebreton et al. 1992, Burn-
ham & Anderson 1998). Models with AICc differing by
less than 2 points were considered to describe the data
equally well and the model with the fewest parameters
was preferred.
Descriptive statistics for adult survival rates: Mean
survival rates over the study period were derived from
a model in which survival was constrained to be con-
stant over years. To avoid an unrealistic estimate (= 1)
caused by convergence problems in the likelihood
optimisation procedure, the estimate for Fair Isle was
obtained from a model where survival probability was
constrained to be constant in 1990 to 1996 and 1997 to
2001 but allowed to be different in 1996 to 1997. Esti-
mates for different colonies were compared within all
possible pairs of colonies using Z-tests for differences
between logit-transformed mean survival estimates
(Lebreton et al. 1992). In order to obtain an estimate
of the temporal process variance in survival probabil-
ity for each colony (referred to as
ˆ
σ
2
) that was cor-
rected for the sampling variance (Gould & Nichols
1988), a variance component procedure was per-
formed based on a model in which the pattern of vari-
ation in survival was estimated without any constraint
(i.e. model Φ[year]). We used the variance decomposi-
tion method implemented in MARK (White & Burn-
ham 1999).
Similarities in pattern of temporal variation in adult
survival rates: The similarity within pairs of colonies in
the pattern of inter-annual variation in adult survival
rate was expressed using the fraction of the total tem-
poral deviance accounted for by a common pattern of
variation. We included the capture histories of the indi-
viduals from 2 colonies and computed models in which
(1) the patterns of variation in survival was allowed to
differ in the 2 colonies (Φ[year + colony + colony × year]);
(2) survival was forced to exhibit parallel variations in
the two colonies (Φ[year + colony]), and (3) survival
was forced to be constant over time in the 2 colonies.
(Φ[colony]). The fraction of deviance accounted for by
the common pattern of variation was computed as the
ratio of (deviance [model Φ(colony)] – deviance [model
Φ(year + colony)]) to (deviance [model Φ(colony)] –
deviance [model Φ(year + colony + colony × year)]).
This statistic is an equivalent for maximum likelihood
models of the coefficient of determination (R
2
) in ana-
lysis of variance models, and is hereafter referred to
as ‘R
2
’.
286
Harris et al.: Survival of Atlantic puffins
Detecting environmental factors underlying inter-
annual variation in adult survival rates: We tested the
effect of the covariates potentially underlying inter-
annual variation in adult survival rates independently
for each colony, after centring covariates reflecting
temporal variation in environmental conditions around
the average and standardising them by the standard
deviation in the time series covering the study period
(White & Burnham 1999). Starting where survival was
constrained to be time-invariant, we performed a
step-up selection of the covariates that showed the
strongest correlation with adult survival rates. R
2
was
used as a selection criterion. At each step of the proce-
dure and for each candidate covariate, we computed
models in which (1) the pattern of variation in survival
was estimated without any constraint (Φ[year]); (2) sur-
vival variation was forced by logit linear relationship(s)
with the focal climatic and/or oceanic factor(s) (Φ[year
+ covariate(s)]), and (3) survival was forced to be con-
stant over time (Φ[cst]). The R
2
of the covariate model
was then computed as the ratio of {deviance Φ[cst] –
deviance [model Φ (covariate[s])]) to (deviance [model
Φ(cst)] – deviance [model Φ(year)]}. A covariate was
selected if its incorporation increased the R
2
associated
with the covariate model by at least 20%. We based
covariate selection on biological significance (reflected
by R
2
) rather than on statistical significance (reflected
by the p-value) because statistical significance de-
pends on the statistical power, which obviously varies
between data sets of different sizes and thus between
colonies. However, we derived ANODEV p-values
(Skalski et al. 1993) as measurements of the degree of
confidence in the existence of the relationships
between adult survival rate and the covariates.
RESULTS
Winter dispersal
There were marked differences in the patterns of
recoveries from birds ringed in different areas (Fig. 1).
Birds from the Isle of May area (340 recoveries) were
mainly found in the North Sea. The majority came from
the east coast of Britain, with relatively few in the SE
North Sea, where Atlantic puffins are generally rare
(Stone et al. 1995), but with some (<10%) in the Faeroe
Islands, the western seaboard of Britain and Ireland
and the Bay of Biscay. In contrast, the main wintering
areas of Atlantic puffins from the Fair Isle (16 recover-
ies) and Skomer (59) areas were to the south of Britain,
presumably mainly in the Atlantic Ocean, with some
entering the Mediterranean and the Norwegian Sea.
However, the few recoveries of birds ringed in the Fair
Isle area (see next paragraph) mean that it is by no
means certain that the bulk of the birds from these 2
areas wintered in the same area. The recoveries from
the Røst area (18) and colonies further north (8) sug-
gest that birds from these areas winter in the southern
Norwegian Sea and the northern North Sea. Too few
287
Fig. 1. Fratercula arctica. Geographical location of 5 colonies studied and locations of recoveries of Atlantic puffins ringed at or
near Skomer (n = 59), the Isle of May (n = 340), Fair Isle (n = 16), Røst (n = 18) and Hornøya (n = 8) reported dead outside breed-
ing season when at least 3 yr old. (
m
) Individual recoveries of birds from Hornøya; (
d
) recoveries of birds from the other
4 locations
Mar Ecol Prog Ser 297: 283296, 2005
recoveries were available to deter-
mine differences in the wintering
areas of Atlantic puffins from Røst
and Hornøya. These general pat-
terns are supported by recent inde-
pendent analyses of the recoveries
of all Atlantic puffins ringed by
the British and Norwegian ringing
schemes (Harris 2002, Bakken et
al. 2003).
There were striking differences
in the recovery rates of birds ringed
at the study colonies. For instance,
totals of 17 462, 12 759 and 2333
Atlantic puffins ringed on Røst, Fair
Isle and Hornøya, respectively, had
resulted in just 9 (0.05%), 8 (0.06%) and 2 (0.09%)
winter recoveries, respectively. In contrast 32 888 and
4859 Atlantic puffins ringed on the Isle of May and
Skokholm (10 km from Skomer) resulted in 170 (0.5%)
and 29 (0.6%) recoveries, respectively, recovery rates
5 to 10 times higher than for the 3 other colonies. Con-
sidering coastline topography and sea currents, this
marked difference in recovery rates is probably due to
the relatively higher chance of a bird dying in the
North Sea and the Bay of Biscay being found and
reported, compared to the presumably far lower
chances of finding birds wintering along the Norwe-
gian coast or far offshore. The chances of recovery are
further reduced to the north, where shores are rocky
and areas more sparsely populated than in mainland
Europe. The paucity of recoveries in the North Sea
from the colonies other than the Isle of May does, how-
ever, suggest that few birds from these 4 colonies win-
ter in this area. The concentration of recoveries around
the Faeroe Islands of birds from all colonies was
undoubtedly due to the high numbers of auks shot
there during the winter, but does indicate an overlap
in wintering areas in those waters.
Pattern of spatial correlation between
local SST time series
The correlations between local SST at different loca-
tions are shown in Table 2, with time series plotted in
Fig. 2. The pattern of correlation did not differ whether
we considered the period covered by the longest sur-
vival time series (1984 to 2000) or only the period cov-
ered by all the survival time series (1990 to 2001). Fur-
thermore, January to May SSTs and May to July SSTs
showed similar spatial correlation patterns, although
correlations for the former were generally higher.
288
Area Isle of May Fair Isle Røst Hornøya
January to May
Skomer 0.73** (0.70*) 0.68** (0.67*) 0.44* (0.08) 0.29 (–0.43)
Isle of May 0.86** (0.91**) 0.58* (0.64*) 0.35 (–0.30)
Fair Isle 0.57* (0.67*) 0.39 (–0.36)
Røst 0.15 (0.09)
May to July
Skomer 0.65** (0.66*) 0.26** (0.45) 0.42* (0.46) 0.24 (–0.37)
Isle of May 0.75** (0.53) 0.53* (0.69**) 0.29 (–0.29)
Fair Isle 0.48* (0.64**) 0.04 (0.09)
Røst –0.05 (–0.02)
Year
North Atlantic Oscillation
December Y – 1 to March Y
Sea surface temperature (°C)
January to May Y
Sea surface temperature (°C)
May to July Y
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
3
0
2
4
6
8
10
12
Skomer Isle of May Fair Isle Røst Hornøya
0
2
4
6
8
10
12
14
16
1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002
Skomer Isle of May Fair Isle Røst Hornøya
Fig. 2. Annual measures of winter North Atlantic Oscillation
(wNAO) and Sea Surface Temperature (SST) in the seas
around the 5 colonies in Year Y (with December of the
previous year included in the wNAO) (see text)
Table 2. Patterns of spatial correlations among local SST time series. Pearson correla-
tion coefficients are given for period covered by longest survival-rate time series
(1984 to 2000) and (in parentheses) for period 1990 to 2001 covered by all time series
of survival. *Significant at 5%; **significant at 1%
Harris et al.: Survival of Atlantic puffins
Strong correlations were found among the 3 southern-
most colonies (Skomer, Isle of May and Fair Isle) and
among the 3 colonies at intermediate latitudes (Isle of
May, Fair Isle and Røst) (Table 2, Fig. 2). SST around
Hornøya, in the Barents Sea, showed no significant
correlation with SSTs around the other colonies.
Adult survival reference-models
The model selected for Skomer, Fair Isle and Isle of
May was Φ(y) p(a
3
+ y). For Hornøya, the model Φ(cst)
p(a
3
+ y) was retained. Finally, for Røst, model Φ(y)
p(a
3
+ y + sex) was retained. For Røst, resighting proba-
bility of females was slightly lower than that of males.
This sex effect was not detected in the Isle of May
colony, where information on the sex of individuals was
also available. Since sex did not appear to influence
survival significantly, it is not discussed further. For all
colonies 3 ‘age since last resighting’ classes and a year
effect were thus retained in the modelling of resighting
probability. Probability of being seen 2 yr after last re-
sighting was lower than after 1 yr, and the probability
after more than 2 yr after last resighting was lower
still (Fig. 3). Except for Hornøya, there were significant
inter-annual variations in survival (Fig. 4, see also
Tables 3 & 5).
289
Resighting probability
Year
One year after last resighting
Two years after last resighting
Three years after last resighting
Skomer
Isle of May
Fair Isle
Røst
Hornøya
Thick: males
Thin: females
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
1991 1993 1995 1997 1999 2001
0
0.2
0.4
0.6
0.8
1
1985
1987
1989 1991 1993 1995 1997 1999 2001
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
0
0.2
0.4
0.6
0.8
1
1993 1995 1997 1999 2001
0
0.2
0.4
0.6
0.8
1
1985 1987 1989 1991 1993 1995 1997 1999
2001
0
0.2
0.4
0.6
0.8
1
Fig. 3. Fratercula arctica. Annual variation in resighting prob-
abilities in the 5 colonies 1, 2 and 3 yr after last re-sighting;
±95% CI shown for 1 yr data
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
Adult survival rate
Hornøya
Røst
Fair Isle
1984
-
1985
1986
-
1987
1988
-
1989
-
1991
1992
-
1993
1994
-
1995
1996
-
1997
1998
-
1999
2000
-
2001
Period
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
0.4
0.5
0.6
0.7
0.8
0.9
1
Isle of May
Skomer
1990
Fig. 4. Fratercula arctica. Annual estimates (±95% CI) of
survival rates in the 5 colonies
Mar Ecol Prog Ser 297: 283296, 2005
Adult survival averages and variability
Mean survival rate estimates over the maximum time
periods showed very little variation among colonies
(0.930 ± SE 0.005 for Skomer, 0.935 ± 0.006 for the Isle
of May, 0.935 ± 0.022 for Fair Isle, 0.935 ±0.013 for
Røst, and 0.935 ± 0.016 for Hornøya) and none of the
inter-colony differences were significant (Z-values 0 to
1.12 for comparison of the logit of mean survival; corre-
sponding p-values 1 to 0.26). The survival rate at
Skomer over the common time period 1990 to 2001 was
slightly lower at 0.925 ± 0.007 (Table 3). The likelihood
optimisation procedure converged towards an unreal-
istic estimate of mean survival rate (= 1) for Fair Isle for
the period 1990 to 2001. By allowing the estimate for
the period 1996 to 1997 to differ from the mean sur-
vival over the periods 1990 to 2001, a more realistic
estimate (0.915 ± 0.015) was obtained. We retained this
value as an estimate of the mean survival rate at this
colony for the period 1990 to 2001. As for Skomer, it is
slightly lower than the estimate of mean survival rate
over the maximum time period (see above). Again
none of the inter-colony differences were significant
over the period 1990 to 2001 (Z-values 0 to 0.88;
p-values 1 to 0.38). Temporal variability in survival rate
(expressed as the estimate of survival standard devia-
tion
ˆ
σ, the square root of the estimate of the survival
variance
ˆ
σ
2
) showed more variation among colonies,
being lowest (0.025) for Hornøya and highest (0.09) for
Fair Isle (Table 3, Fig. 4).
Temporal variation in adult survival rates:
patterns of similarity between colonies
The pattern of similarity between time series of adult
survival rates in the colonies (as reflected by the R
2
of
a common pattern of variation model) did not differ
substantially according to whether one considered the
whole period of overlap between survival time series in
each pair of colonies or only the period 1990 to 2001
covered by all the survival time series (Table 3). The
highest R
2
were found for the pairs Røst/Skomer,
Hornøya/Skomer and Hornøya/Fair Isle, colonies some
2000 km apart that did not show any overlap in the
locations of winter recoveries (see earlier subsection
‘Winter dispersal’). Comparison of Hornøya with other
colonies yielded consistently high fractions of temporal
variation accounted for by a common pattern (R
2
rang-
ing from 65 to 71%), while comparison of Fair Isle with
other colonies yielded consistently lower R
2
(38 to
47%, excluding the comparison with Hornøya); this
was at least partly due to the sharp decline in survival
in the later years.
Temporal variation in survival probabilities:
correlations with environmental factors
For Skomer, Fair Isle, Røst and Hornøya, local SST in
January to May of Year Y–1 explained at least 20% of
the variation in adult survival between the breeding
season of Year Y and that of Year Y+1 (Tables 4 & 5,
Fig. 5). In contrast, local SST in May to July of Year Y
was more important for Atlantic puffins on the Isle of
May, explaining 26% of the variation in adult survival
between the breeding season of Year Y and that of
Year Y+1. The influence of increased SST was positive
for Atlantic puffins on Røst but negative for the birds
at the other colonies. In addition to the effect of SST,
wNAO explained more than 20% of the variation in
adult survival at Skomer and Hornøya, having a posi-
tive effect on Skomer birds in the same winter and a
negative effect on Hornøya puffins 2 yr later (Tables 4
& 5, Fig. 5).
All the relationships detected between environmen-
tal factors and survival were significant at the 5%
level except for those relating to Hornøya. This was
290
Table 3. Fratercula arctica. Adult survival rate means (to the nearest 0.005 ± SE) and process standard deviation (95% CI)
for each colony and covariation among colony time series (period 1990 to 2001 given in second line)
Area Adult Survival rate Fraction (%) of variation explained by common pattern (R
2
)
survival process May Fair Isle Røst Hornøya
rate SD (95% CI) 19842001 19862001 19902001 19902001
Skomer 19842001 0.930 ± 0.005 0.050 (0.0400.075) 53 (63) 39 (47) 77 71
Skomer 19902001 0.925 ± 0.007 0.050 (0.0350.095)
May 19842001 0.935 ± 0.006 0.050 (0.0400.075) 40 (43) 56 66
May 19902001 0.935 ± 0.007 0.035 (0.0150.075)
Fair Isle 19862001 0.935 ± 0.022 0.090 (0.0500.160) 38 71
Fair Isle 19902001 0.915 ± 0.015 0.110 (0.0700.200)
Røst 19902001 0.935 ± 0.013 0.040 (0.0200.085) 65
Hornøya 19902001 0.935 ± 0.016 0.025 (0.0100.060)
Harris et al.: Survival of Atlantic puffins
probably due to the low variability of adult survival at
this colony. Furthermore, a graphical examination of
the relationships between climatic factors and sur-
vival of Hornøya birds (Fig. 5) revealed that the 4 last
estimates of the survival time series were outliers in
the relationships with SST and wNAO. Hence, the
relationships detected for Hornøya must be consid-
ered with caution.
DISCUSSION
The estimated mean annual survival rates of adult
Atlantic puffins of 0.930 to 0.935 were high, although
slightly lower than estimates of 0.95 and 0.975 for
Skomer and the Isle of May, respectively, during the
1970s (Ashcroft 1979, Harris et al. 1997) and of 0.95 for
an introduced population in the western Atlantic
291
Table 4. Fratercula arctica. Selection of climatic/oceanographic covariates influencing survival between Breeding Season Y and
Y+1. Step 1: among those covariates whose inclusion in model resulted in a R
2
20%, that leading to largest R
2
was selected; Step
2: that covariate whose inclusion in model increased R
2
by largest extent, and by at least 20% relative to model arising from
Step 1, was selected. wNAO: winter North Atlantic Oscillation
Breeding season Skomer Fair Isle Isle of May Røst Hornøya
R
2
(%) p R
2
(%) p R
2
(%) p R
2
(%) p R
2
(%) p
Step 1
SST January Y–1 to May Y–1 13 0.16 49 0.004 09 0.24 40 0.04 19 0.17
SST January Y to May Y 04 0.41 20 0.09 20 0.07 28 0.09 04 0.54
SST May Y to July Y 00 0.85 16 0.14 26 0.04 18 0.19 01 0.34
wNAO December Y to 2 to March Y–1 00 0.78 26 0.05 01 0.70 31 0.07 11 0.30
wNAO December Y to 1 March Y 07 0.31 25 0.06 12 0.17 04 0.57 07 0.42
wNAO December Y to March Y+1 20 0.07 08 0.3 02 0.55 35 0.05 10 0.34
Step 2
SST January Y–1 to May Y–1 45 0.02 28 0.51
SST January Y to May Y 26 0.29 54 0.27 27 0.64 46 0.36 22 0.57
SST May Y to July Y 21 0.76 53 0.32 41 0.66 22 0.56
wNAO December Y–2 to March Y–1 23 0.48 49 0.84 28 0.51 50 0.24 43 0.10
wNAO December Y–1 to March Y 22 0.56 61 0.08 28 0.59 40 0.75 19 0.81
wNAO December Y to March Y+1 49 0.9 28 0.56 42 0.59 23
Table 5. Fratercula arctica. Final model for survival between Breeding Season Y and Y+1. Relationship between environmental
covariates and survival described by estimate of the slope, 95% confidence interval for the slope (in parentheses) and ANODEV
p-value for the relationship. AICc (Akaike’s information criterion modified for each data sparseness) of constant and fully time-
dependent models are given for comparison with final covariate model. wNAO: winter North Atlantic Oscillation
Skomer Isle of May Fair Isle Røst Hornøya
19842000 19842000 19862000 19902000 19902000
AICc constant survival model 4867.1 2875.6 2258.1 2494.3 3978.6
AICc fully time dependent survival model 4854.4 2856.6 2230.1 2481.8 3988.8
AICc of the final covariate survival model 4850.8 2864.2 2232.1 2483.3 3978.3
R
2
associated to the covariate model 45 26 49 40 43
0.37 0.68 0.46 1.16
SST January Y–1 to May Y–1 (–0.54; –0.20) (–0.96; –0.40) (0.21; 0.72) (–3.00; 0.68)
p = 0.025 p = 0.004 p = 0.04 p = 0.07
SST January Y to May Y
0.36
SST May Y to July Y (–0.55; –0.16)
p = 0.04
1.1
wNAO December Y–2 to March Y–1 (–3.20; 1.00)
p = 0.1
wNAO December Y–1 to March Y
0.32
wNAO December Y to March Y+1 (0.19; 0.46)
p = 0.01
Mar Ecol Prog Ser 297: 283296, 2005
(Breton et al. in press). Despite highly contrasting pat-
terns of breeding population change among the
colonies, adult survival was remarkably consistent,
giving no support to the hypothesis that survival was
an important demographical driver. There were, how-
ever, marked differences between colonies in the
inter-annual variability of survival rates for which we
have no satisfactory explanation. Breeding success
(lagged 5 yr to cover time between fledging and
becoming adult) was generally high in the 4 stable or
increasing colonies, averaging 0.72 to 0.85 young
fledging per pair laying (Table 1). In contrast, Atlantic
puffins in the decreasing colony at Røst had a much
lower, and more variable, average success of 0.42
young fledging per pair that hatched a young. Since
many breeding attempts fail during incubation (Harris
1984), overall breeding success at Røst will have been
substantially lower than this. Thus it seems likely that
recruitment (the integration of breeding success, sur-
vival to breeding age and natal dispersal) rather than
adult survival is underlying variation in growth
rate among colonies of Atlantic puffins.
Although care is needed in the interpretation of
recoveries of ringed seabirds that spend most of
their lives in the open ocean, where birds that die
have only a very small chance of being found, the
distributions of recoveries during the winter of
birds from the study areas showed very marked
patterns. Adults from the Isle of May were mainly
within the North Sea, whereas those from
Skomer and Fair Isle dispersed widely in the
Atlantic west of Britain, France and Iberia, with
some entering the Mediterranean, and those from
the 2 Norwegian colonies had a much more
northerly distribution. Hence, if environmental
conditions over the wintering areas were the
main driver of variation in Atlantic puffin sur-
vival, the pattern of covariation in adult survival
would be expected to be structured in 3 clusters
(Fair Isle and Skomer/Isle of May/Røst and
Hornøya). This was not the case. Under the
hypothesis that oceanographic or weather condi-
tions in winter strongly influenced Atlantic puffin
adult survival, we would also have expected to
detect unlagged effects of the wNAO, because
this large-scale climate index reflects variation in
weather and oceanographic conditions in winter
over most NE Atlantic regions (Becker & Pauly
1996, Hurrell et al. 2003). Such effect was
detected for only 1 colony (Skomer), suggesting
that oceanographic and weather conditions of
the wintering areas have little influence on
Atlantic puffin survival. The diet of the Atlantic
puffin outside the breeding season is largely
unknown, so the possibility that food abundance
in the wintering areas influences the survival of
adult Atlantic puffins cannot be ruled out.
However, the mismatch between the pattern of
spatial overlap between winter recoveries of
adults from the different study colonies on the
one hand and the pattern of covariation in adult
survival among colonies on the other hand sug-
gests that food abundance in the wintering areas
is unlikely to account for much of the variation
in survival.
292
Residuals
logit survival rate (Y-Y+1)
-2
-1
0
1
2
3
7891011
-5
-4
-3
-2
-1
0
-2 0 2
-5
-3
-1
1
3
024
1
2
3
4
5.4 5.6 5.8 6 6.2
1
2
3
4
5
9 10111213
0
2
4
6
78910
Local -scale factors Large -scale factors
Skomer
Isle of May
Fair Isle
Røst
Residuals
logit survival rate (Y-Y+1)
logit survival rate
(Y
-
Y+1)
SST January Y-1 to May Y-1
SST May Y to July Y
SST January Y-1 to May Y-11
SST January Y-1 to May Y-11
NAO December Y to March Y+1
logit survival rate
(Y
-
Y+1)
logit survival rate
(Y
-
Y+1)
Hornøya
SST January Y-1 to May Y-11 NAO December Y-2 to March Y-1
-2
-1
0
1
2
3
-2 0 2
----
Fig. 5. Fratercula arctica. Relationship between logit survival rate
and environmental variables (SST: sea surface temperature; NAO:
North Atlantic Oscillation) for the 5 colonies. (
s) Survival estimates
at upper boundary (= 1) that must be considered with caution; (
n)
estimates for last 4 time intervals of time series at Hornøya (see
‘Discussion’ for Hornøya results)
Harris et al.: Survival of Atlantic puffins
Whereas we detected an influence of the proxy pre-
sumably reflecting oceanographic and weather condi-
tions in the wintering areas (i.e. unlagged wNAO) in
only 1 out of 5 colonies, oceanographic factors in the
vicinity of the breeding colonies correlated with adult
survival in all colonies. Local SST in January to May
had a significant effect on survival 2 yr later in 4 out
5 colonies. These relationships probably reflect the
importance of the abundance of the main prey during
the breeding season (herring for Røst, sandeel and
capelin for Hornøya and sandeel for all other colonies).
The slopes of the relationships with SST in January to
May are compatible with this interpretation, being
negative where sandeels and capelin are the main
food, and positive where herring are taken. These con-
trasting relationships might be explained by recruit-
ment in sandeels and capelin decreasing with increas-
ing sea temperature and the reverse being true for
herring, where recruitment decreases with increasing
sea temperatures (Arnott & Ruxton 2002, Sætre et al.
2002, Hjermann et al. 2004). In addition to these SST
effects, the equally long lagged (2 yr) effect of wNAO
on survival at Hornøya was also in agreement with the
hypothesis that the main prey species at the colony are
either the key species affecting adult survival of the
Atlantic puffin, or at least that they respond to SST
changes in a similar way. Given a sudden food short-
age in summer, as not infrequently occurs on Røst,
there is no reason to suppose that adults would remain
in the area and die of starvation. In all probability a
combination of winter and summer food availability
influences the survival of adult seabird, since individu-
als entering the winter in poor condition are more
likely to die than those in good condition.
The relationship between SST and adult survival on
the Isle of May differed from that detected in the other
colonies. Here, SST during the breeding season had
the strongest effect on survival during the subsequent
year, suggesting a more direct influence of oceano-
graphic conditions during the breeding season on food
and hence survival the following winter. It is, however,
difficult to separate this effect from that of SST in Jan-
uary to May of the same year. The 2 measures were
highly autocorrelated (Pearson correlation coefficient
r = 0.84) and while SST in May to July showed the
strongest correlation with adult survival (R
2
= 26%;
Table 4), SST in January to May of the same year also
explained a substantial fraction of the variation in adult
survival (R
2
= 20%; Table 4). Thus again it may well
have been a trophic response to food abundance, but
seemingly the birds from Isle of May were most sensi-
tive to the availability of 0- and 1-group fishes (mainly
sandeels in the North Sea) while those from the other
colonies probably depended more heavily on 1- and
2-group fishes.
Although the results of the pattern-oriented analysis
presented here have to be interpreted with caution
(especially for the colony of Hornøya, where little vari-
ation in adult survival was detected), the local SST
effects detected suggest that food abundance during,
or possibly soon after, reproduction influences the
survival of adult Atlantic puffins. A similar conclusion
has been reached for 2 other sandeel consumers, the
black-legged kittiwake Rissa tridactyla and the great
skua Catharacta skua, in Shetland (Oro & Furness
2002, Ratcliffe et al. 2002). The Atlantic puffin, black-
legged kittiwake and great skua are all pelagic outside
the breeding season. The coefficient of determination
of our model that includes all selected covariates
ranged from 26 to 49%. The Atlantic puffin is the the
most numerous seabird in the western Palearctic
(Mitchell et al. 2004), but until we obtain more in-
formation from process-oriented studies outside the
breeding season, we are unlikely to greatly improve
our understanding of the factors influencing the
survival of this pelagic species.
Acknowledgements. We thank the many people who helped
collect the data used in these analyses; the British Trust for
Ornithology and the Norwegian Ringing Centre for supplying
ringing totals and recoveries; R. Barrett for allowing us to use
unpublished data from Hornøya; L. Crespin for considerable
help with the analyses; and S. Wanless, M. Frederiksen, C. M.
Perrins and 4 anonymous reviewers for improving the manu-
script with criticisms. Studies on Skomer, Fair Isle and the Isle
of May were supported by the Joint Nature Conservation
Committee, those on Røst and Hornøya by the Norwegian
Directorate for Nature Management (DN) and the Environ-
mental Protection Departments of the County Governors
in Nordland, Troms and Finnmark. N. C. Stenseth and his
EcoClim project helped initiate this analysis.
LITERATURE CITED
Anker-Nilssen T (1992) Food supply as a determinant of
reproduction and population development in Norwegian
puffins Fratercula arctica. Dr Sci thesis, University of
Trondheim
Anker-Nilssen T, Aarvak T (2004) The population ecology of
the Atlantic puffin at Røst. Status after the breeding
season 2003. NINA Oppdragsmeld 809:144
Anker-Nilssen T, Brøseth H (1998) Long-term studies of the
breeding biology of puffins at Røst. An update with results
from 199597. NINA Fagrapp 32:146
Anker-Nilssen T, Øyan HS (1995) Long-term studies of the
breeding biology of puffins at Røst. NINA Fagrapp 15:
1–48
Anker-Nilssen T, Røstad OW (1993) Census and monitoring of
puffins Fratercula arctica on Røst, N Norway, 19791988.
Ornis Scand 24:19
Anker-Nilssen T, Tatarinkova IP (2000) Atlantic puffin Frater-
cula arctica. In: Anker-Nilssen T, Bakken V, Strøm H,
Golovkin AN, Bianki VV, Tatarinkova IP (eds) The status
of marine birds breeding in the Barents Sea region. Nor
Polarinst Rapp Ser 113:137143
293
Mar Ecol Prog Ser 297: 283296, 2005
Arnott SA, Ruxton GD (2002) Sandeel recruitment in the
North Sea: demographic, climatic and trophic effects. Mar
Ecol Prog Ser 238:199210
Ashcroft RE (1979) Survival rates and breeding biology of
puffins on Skomer Island, Wales. Ornis scand 10:100110
Attril M, Power M (2002) Climatic influence on a marine fish
assemblage. Nature 417:275278
Bakken V, Runde O, Tjørve E (2003) Norsk ringmerkingsatlas
(Norwegian ringing atlas), Vol 1. Stavanger Museum,
Stavanger
Barrett RT (1983) Seabird research on Hornøy, East Finnmark
with notes from Nordland, Troms and West Finnmark
19801983. A preliminary report. Tromsø University
Museum, Tromsø
Barrett RT (2001) Monitoring the Atlantic puffin Fratercula
arctica, common guillemot Uria aalge and black-legged
kittiwake Rissa tridactyla breeding populations on
Hornøya, northeast Norway, 19802000. Fauna Norv Ser
21:110
Barrett RT (2002) Atlantic puffin Fratercula arctica and com-
mon guillemot Uria aalge chick diet and growth as indica-
tors of fish stocks in the Barents Sea. Mar Ecol Prog Ser
230:275287
Becker GA, Pauly M (1996) Sea surface temperature changes
in the North Sea and their causes. ICES J Mar Sci 53:
887898
Breton AR, Diamond AW, Kress SW (in press) Adult survival
from two Atlantic puffin (Fratercula arctica) colonies in the
Gulf of Maine. Auk
Burnham KP, Anderson DR (1998) Model selection and
inference. A practical information-theoretic approach.
Springer-Verlag, New York
Cam E, Hines JE, Monnat JY, Nichols JD, Danchin E (1998)
Are adult nonbreeders prudent parents? The kittiwake
model. Ecology 79:29172930
Choquet R, Reboulet AM, Pradel R, Lebreton JD (2001) U-care
(utilitiescapture-recapture) user’s guide. CNRS (Centre
National de la Recherche Scientifique), Montpellier
Croxall JP, Rothery P (1991) Population regulation of seabirds:
implications of their demography for conservation. In: Per-
rins CM, Lebreton JD, Hirons GJM (eds) Bird population
studies: relevance to conservation and management.
Oxford University Press, Oxford, p 272295
Durant JM, Anker-Nilssen T, Stenseth NC (2003) Trophic
interactions and climate change: the Atlantic puffin as an
example. Proc R Soc Lond Ser B 270:14611466
Edwards M, Beaugrand G, Reid PC, Rowden AA, Jones MB
(2002) Ocean climate anomalies and the ecology of the
North Sea. Mar Ecol Prog Ser 239:110
Gould WR, Nichols JD (1988) Estimation of temporal vari-
ability of survival in animal populations. Ecology 79:
25312538
Harris MP (1984) The puffin. T & AD Poyser, Calton
Harris MP (2002) Atlantic puffin (puffin) Fratercula arctica.
In: Wernham CV, Toms M, Marchant JH, Clark JA, Siri-
wardena GM, Baillie SR (eds) The migration atlas; move-
ments of the birds of Britain and Ireland. T & AD Poyser,
London, p 407409
Harris MP, Wanless S (2004) The Atlantic puffin Fratercula
arctica. In: Mitchell IP, Newton SF, Ratcliffe N, Dunn TE
(eds) Seabird populations in Britain and Ireland. T & AD
Poyser, London, p 392406
Harris MP, Freeman SN, Wanless S, Morgan BJT, Wernham
CV (1997) Factors influencing the survival of puffins
Fratercula arctica at a North Sea colony over a 20-year
period. J Avian Biol 28:287–295
Harris MP, Wanless S, Murray S, Leitch A, Wilson LJ (2003)
Counts of Atlantic puffins Fratercula arctica in the Firth
of Forth, south-east Scotland in 2003. Atl Seabirds 5:
101110
Hjermann DØ, Stenseth NC, Ottersen G (2004) Indirect cli-
matic forcing of the Barents Sea capelin: a cohort effect.
Mar Ecol Prog Ser 273:229238
Hurrell JW, Kushnir Y, Ottersen G, Visbeck M (2003) The
North Atlantic Oscillation: climate significance and envi-
ronmental impact. Geophys Monogr 134:1279
Lebreton JD, Burnham KP, Clobert J, Anderson DR (1992)
Modelling survival and testing biological hypotheses
using marked animals: a unified approach with case
studies. Ecol Monogr 62:37118
Mitchell PI, Newton SF, Ratcliffe N, Dunn TE (2004) Seabird
populations of Britain and Ireland. Results of the seabird
2000 census (19982002). T & AD Poyser, London
Myrberget S (1973) Ringing of shags and puffins at Røst,
north Norway. Sterna (Stavanger) 12:307315
Nur N, Sydeman WJ (2000) Survival, breeding probablity and
reproductive success in relation to population dynamics
of Brandt’s cormorants Phalacrocorax penicillatus. Bird
Study 46:S92103
Oro D, Furness RW (2002) Influences of food availability and
predation on survival of kittiwakes. Ecology 83:25162528
Ottersen G, Planque B, Belgrano A, Post E, Reid PC, Stenseth
NC (2001) Ecological effects of the North Atlantic Oscilla-
tion. Oecologia 128:114
Planque B, Taylor AH (1998) Long-term changes in zooplank-
ton and the climate of the North Atlantic. ICES J Mar Sci
55:644654
Pradel R (1993) Flexibility in survival analysis from recapture
data: handling trap-dependence. In: Lebreton JD, North
PM (eds) Marked individuals in the study of bird popula-
tions. Birkhauser Verlag, Basel, p 2937
Ratcliffe N, Catry P, Hamer KC, Klomp NI, Furness RW (2002)
The effect of age and year on the survival of breeding
adult great skuas Catharacta skua in Shetland. Ibis 144:
384392
Reynolds RW, Rayner NA, Smith TM, Stokes DC, Wang W
(2002) An improved in situ and satellite SST analysis for
climate. J Climate 15:16091625
Sætre R, Toresen R, Anker-Nilssen T (2002) Factors affecting
the recruitment variability of the Norwegian spring-
spawning herring (Clupea harengus L.). ICES J Mar Sci
59:725736
SAS (1999) SAS property software release 8.02. SAS Institute,
Cary, NJ
Skalski JR, Hoffmann A, Smith SG (1993) Testing the signifi-
cance of individual- and cohort-level covariates in animal
survival studies. In: Lebreton JD, North PM (eds) Marked
individuals in the study of bird populations. Birkhauser
Verlag, Basel, p 928
Stone CJ, Webb A, Barton C, Ratcliffe N, Reed TC, Tasker
ML, Camphuysen CJ, Pienkowski MW (1995) An atlas of
seabird distribution in north-west European waters. Joint
Nature Conservation Committee, Peterborough
Toresen R, Østvedt OJ (2000) Variation in abundance of
Norwegian spring-spawning herring (Clupea harengus,
Clupeidae) throughout the 20th century and the influence
of climatic fluctuations. Fish Fish Ser 2000:231256
White GC, Burnham KP (1999) Program MARK: survival esti-
mation from populations of marked animals. Bird Study 46
(Suppl):120139
Wooller RD, Bradley JS, Skira IJ, Serventy DL (1990) Repro-
ductive success of short-tailed shearwaters Puffinus
tenuirostris in relation to their age and breeding experi-
ence. J Anim Ecol 59:161170
294
Harris et al.: Survival of Atlantic puffins
295
Year N First subsequent resightings
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Skomer
1984 198 141 92033410100000000
1985 191 142 8064000011010010
1986 176 133 12 27102011100000
1987 169 125 14 7220010110001
1988 157 104 31 411130001000
1989 158 115 740574001011
1990 239 170 10 3391302010
1991 241 174 13 5 15 5110020
1992 219 135 15 37 13 201101
1993 172 96 48 11 210100
1994 141 100 11 530000
1995 248 194853421
1996 262 172313721
1997 211 1588844
1998 221 145 30 5 3
1999 168 132 7 3
2000 203 158 8
2001 190 160
Isle of May
1984 71 56 62110101000000000
1985 75 63 3010101000001000
1986 73 56 341020000000000
1987 64 50 60012000000000
1988 61 51 3100000000000
1989 63 37 12 00000000000
1990 113 90 10000000000
1991 197 183 2021000000
1992 215 198 600000000
1993 225 210 60000000
1994 221 195 11 010000
1995 209 189511011
1996 220 20071000
1997 218 180 13 0 0 1
1998 197 174 5 1 1
1999 209 176 9 1
2000 190 172 3
2001 195 163
Fair Isle
1986 33 29 100100000000000
1987 140 108 83200000030000
1988 123 103 8000001000000
1989 133 108 700000000000
1990 137 104 21000011100
1991 127 100 2000013000
1992 106 92 001100100
1993 98 74 53000010
1994 78 69 4000000
1995 85 79101000
1996 89 7940210
1997 85 58 10 2 0 2
1998 91
67703
1999 82 53 1 1
2000 81 42 4
2001 45 22
Røst, males
1990 28 25 10000000000
1991 75 67 2001000000
1992 74 66 500000000
1993 68 60 02100100
1994 66 51 5100100
1995 52 44300000
1996 61 5322000
1997 61 426000
1998 57 52100
1999 74 65 2 1
2000 74 52 5
2001 65 58
Appendix 1. Fratercula arctica. M-arrays for the 5 study colonies. N: no. of individuals encountered (initial capture or resighting).
Data on males and females on Røst shown separately because the CMR model selected for describing Røst data included sex
effect on recapture probability
(Appendix continued on next page)
Mar Ecol Prog Ser 297: 283296, 2005296
Year N First subsequent resightings
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
Røst, females
1990 44 34 71100000000
1991 94 86 0110100000
1992 107 92 711001000
1993 93 77 83100000
1994 95 72 5200000
1995 82 58620201
1996 83 6063100
1997 73 497110
1998 84 60912
1999 97 70 8 4
2000 96 67 10
2001 91 73
Hornøya
1990 37 3 20 1012100000
1991 242 179 7266011100
1992 212 132 25 83441000
1993 145 101 16 2011100
1994 431 274 23 19 94030
1995 494 252 28 56 19 4 1 4
1996 288 208134000
1997 164 1287100
1998 220 145 11 7 2
1999 145 110 11 3
2000 130 87 12
2001 100 67
Appendix 1 (continued)
Editorial responsibility: Otto Kinne (Editor-in-Chief),
Oldendorf/Luhe, Germany
Submitted: December 27, 2004; Accepted: March 22, 2005
Proofs received from author(s): July 22, 2005
... Unfortunately, drivers behind spatial variation in population trends are generally poorly understood, although they have important implications for understanding changes in abundance across the range of species and for their population management. In the case of colonially breeding seabirds, there is evidence that they form spatially distinct populations and their demographic traits vary across different spatial scales, with inter-population differences found in parameters such as adult survival, productivity, and population growth rate (Frederiksen et al. 2005a, Harris et al. 2005, Cordes et al. 2015. As central place foragers during the breeding season, seabirds travel back and forth to the sea (or terrestrial habitats) constrained by the need to regularly provision and care for their offspring (Bolton et al. 2019), which radically affects their spatial ecology. ...
... Populations of nearby colonies might have been affected similarly, either because they foraged in the same places or because food availability in the foraging habitats surrounding the colonies changed in the same direction. In any case, this suggests that the composition of potential resources was similar among neighboring colonies but different between colonies further away, which could lead to interpopulation differences in parameters such as survival, reproduction, emigration, immigration, and consequently population growth rates (Frederiksen et al. 2005a, Harris et al. 2005, Cordes et al. 2015. ...
... The largest changes in seabird numbers in the eastern Norwegian Sea are linked to ocean climate variability 13,14 and most likely mediated through substantial changes in prey abundance and availability with dire consequences for reproductive success and recruitment [15][16][17][18][19][20] . To some degree, this has also affected survival rates [21][22][23] , which in addition can occasionally be severely hit by extreme weather events [24][25][26] . Still, an increasing number of studies document effects of other natural and man-induced changes that may also contribute to the variation in seabird breeding performance. ...
Full-text available
Technical Report
The task of the Working Group on Integrated Ecosystem Assessments for the Norwegian Sea (WGINOR) is to advance the understanding of the Norwegian Sea ecosystem and to develop an operational approach for integrated ecosystem assessment (IEA) that is applicable to management. This includes research on functional connections within the ecosystem, compiling relevant time series and develop models suitable for IEA. The current report contains the results from the second meeting of a 3-year term which has six Terms of Reference (ToRs). An interim assessment of the ecosystem (ToR A) shows that since 2016 the water has become fresher and cooled slightly. Primary production has been higher the last 7 years than previously, while zooplankton biomass remains low as it has since 2003. Biomass of the major pelagic stock shows a declining trend and the decline in seabird populations have continued. Further development of the IEA approach include work with classification of trends in time series of physical and biological ecosystem components (to be used for communication) and warning signal analyses based on trend estimation and an outlier detection analysis of the same time series (ToR A). Work has also been done on a framework for forecasting ocean climate based on statistical models (ToR C) and a foodweb based model assessment with hindcast and forecast properties using a chance and necessity modelling approach (ToR D), and a plan has been made for a framework for exploring multispecies harvest control rules for pelagic fish (ToR B). Further work on these issues will be done in the years 2021-2023 through the research project “Sustainable multispecies harvest from the Norwegian Sea and adjacent ecosystems” at the Institute for Marine Research (Norway). Work on revising the Norwegian Sea ecosystem overview (EO) was continued (ToR F). A work�shop involving experts external to WGINOR will follow up this. Science highlights from the meeting includes two model-based studies, one on harvest patterns in zooplankton fisheries and another exploring the population and ecosystem effects of changes in harvest control rules of two target species; mackerel and hake; in the Norwegian Sea and the California Current system.
... The largest changes in seabird numbers in the eastern Norwegian Sea are linked to ocean climate variability 15,16 and most likely mediated through substantial changes in prey abundance and availability with dire consequences for reproductive success and recruitment [17][18][19][20][21][22] . To some degree, this has also affected survival rates [23][24][25] , which in addition can occasionally be severely hit by extreme weather events [26][27][28] . Still, an increasing number of studies document effects of other natural and man-induced changes that may also contribute to the variation in seabird breeding performance. ...
Full-text available
Technical Report
The Working Group on Integrated Ecosystem Assessments for the Norwegian Sea (WGINOR) executes and develops an integrated ecosystem assessment (IEA) for the Norwegian Sea ecoregion. This report summarizes the working group’s progress on annual updates to the IEA and time series, development of an ocean climate forecast, a food-web based assessment of the pelagic ecosystem, evaluation of single and multispecies harvest control rules for pelagic fish, revisions of the Ecosystem Overview completed in 2021, and establishment of dialogue with pelagic fisheries stakeholders and managers in Norway, Faroe Islands, and Iceland. The Norwegian Sea Ecosystem Overview revision included updates to the pressures and their importance, and a re-evaluation of the sector-pressure-component pathways. Other outputs included: (1) a 10-page ecosystem state summary for a non-scientific audience was produced (Annex 3), (2) a framework for identifying extreme values in the time series for in-depth analysis was developed, and (3) a study on multispecies harvest control rules scenarios and another on impact of value- and ecosystem-based management scenarios on Norwegian Spring-spawning herring showed impact of management decisions on the ecosystem. Dialogue was also successfully initiated with pelagic fisheries stakeholders and managers in Iceland. Research on ocean climate impacts on ecosystem productivity indicated Arctic water masses facilitate greater abundance of nutrients and zooplankton compared to Atlantic waters. Preliminary results suggest oceanographic conditions are influenced by many processes operating at different time-scales complicating the development of ocean climate forecast products. Model reconstruction of tropic interactions in the pelagic ecosystem suggests limited competitions between dominant pelagic fish stocks. Comparison of diet estimation methods reveals occurrence-based methods give similar results to weight-based methods but are more robust and cost efficient. Furthermore, sampling fewer specimens at more stations reduces diet variance. WGINOR priorities for the next term is to continue development of robust IEAs to support development of ecosystem-based management and ecosystem-based fisheries management in the Norwegian Sea.
... The proportion of total annual variance in egg volume explained by covariates was calculated as [deviance (model constant)-deviance (model with covariate)] / [deviance (model constant)-deviance (model timedependent)]. The resulting statistics can be used as an equivalent of the coefficient of determination, R 2 (hereafter R 2 ; see [53]). ...
Full-text available
Article
Large-scale climatic indices are extensively used as predictors of ecological processes, but the mechanisms and the spatio-temporal scales at which climatic indices influence these processes are often speculative. Here, we use long-term data to evaluate how a measure of individual breeding investment (the egg volume) of three long-lived and long-distance-migrating seabirds is influenced by i) a large-scale climatic index (the North Atlantic Oscillation) and ii) local-scale variables (food abundance, foraging conditions, and competition). Winter values of the North Atlantic Oscillation did not correlate with local-scale variables measured in spring, but surprisingly, both had a high predictive power of the temporal variability of the egg volume in the three study species, even though they have different life-history strategies. The importance of the winter North Atlantic Oscillation suggests carry-over effects of winter conditions on subsequent breeding investment. Interestingly, the most important local-scale variables measured in spring were associated with food detectability (foraging conditions) and the factors influencing its accessibility (foraging conditions and competition by density-dependence). Large-scale climatic indices may work better as pre-dictors of foraging conditions when organisms perform long distance migrations, while local-scale variables are more appropriate when foraging areas are more restricted (e.g. during the breeding season). Contrary to what is commonly assumed, food abundance does not directly translate into food intake and its detectability and accessibility should be considered in the study of food-related ecological processes.
... Sea surface temperatures (SST) at various time lags have been associated with food supply, adult survival and breeding success in multiple seabird populations [23][24][25][26][27][28]. We included SST in the breeding and non-breeding areas for the current year to reflect proximate effects on plankton productivity. ...
Full-text available
Article
Understanding the points in a species breeding cycle when they are most vulnerable to environmental fluctuations is key to understanding interannual demography and guiding effective conservation and management. Seabirds represent one of the most threatened groups of birds in the world, and climate change and severe weather is a prominent and increasing threat to this group. We used a multi-state capture-recapture model to examine how the demographic rates of a long-lived trans-oceanic migrant seabird, the Manx shearwater Puffinus puffinus , are influenced by environmental conditions experienced at different stages of the annual breeding cycle and whether these relationships vary with an individual’s breeding state in the previous year (i.e., successful breeder, failed breeder and non-breeder). Our results imply that populations of Manx shearwaters are comprised of individuals with different demographic profiles, whereby more successful reproduction is associated with higher rates of survival and breeding propensity. However, we found that all birds experienced the same negative relationship between rates of survival and wind force during the breeding season, indicating a cost of reproduction (or central place constraint for non-breeders) during years with severe weather conditions. We also found that environmental effects differentially influence the breeding propensity of individuals in different breeding states. This suggests individual spatio-temporal variation in habitat use during the annual cycle, such that climate change could alter the frequency that individuals with different demographic profiles breed thereby driving a complex and less predictable population response. More broadly, our study highlights the importance of considering individual-level factors when examining population demography and predicting how species may respond to climate change.
... A paradigmatic example is cycles in small mammal densities and their influence on survival of predators in high-latitude ecosystems (Brommer et al. 2002;Karell et al. 2009;Millon et al. 2014). Spatio-temporal variability in survival may be particularly acute also for longlived migrant birds moving in large geographical areas, and selecting a suitable wintering area may optimise survival probabilities and fitness prospects (Harris et al. 2005;Genovart et al. 2013;Klaassen et al. 2014;Sergio et al. 2014b). Yet, some species may have distinct populations with strategies being either resident or migrant and with consequences for population heterogeneity in individual survival and population dynamics (Sanz-Aguilar et al. 2012. ...
Full-text available
Article
Stochasticity in food availability influences vital rates such as survival and fertility. Life-history theory predicts that in long-lived organisms, survival should be buffered against environmental stochasticity showing little temporal variability. Furthermore , to optimize survival prospects, many animal species perform migrations to wintering areas where food availability is larger. Species with large latitudinal distribution ranges may show populations that migrate and others that are resident, and they may co-occur in winter. One example of these species is the predatory raptor buzzard Buteo buteo. Here, we test whether temporal variability in the density of five small mammal species of prey inhabiting different habitats (shrubland and forests) influences local annual survival of buzzards in a wintering area depending on their age and residency status (resi-dents versus wintering individuals). We found that prey density explained a considerable amount of annual changes in local survival, which was higher for older and resident birds. This difference in local survival likely corresponded to philopatry to the wintering area, which was larger for residents and increased when prey density was larger. The total density of prey inhabiting open shrublands was the variable explaining more variance in temporal variability of local survival, even though the study area is mostly occupied by woodlands. Temporal population dynamics of the different small mammals inhabiting shrublands were not synchronous, which suggests that buzzards preyed opportunistically on the most abundant prey each winter. Generalist predation may buffer the impact of resource unpredictability for pulsed and asynchronous prey dynamics, typical of small mammals in winter.
Technical Report
https://www.miljodirektoratet.no/globalassets/publikasjoner/m396/m396.pdf
Full-text available
Article
Seabirds are undergoing drastic declines globally and spend the non-breeding season at sea, making it challenging to study the drivers of their survival. Harsh weather and changes in climate conditions can have large impacts on seabird population dynamics through increased mortality. The intensity and persistence of extreme events are forecasted to increase with global warming. As shared conditions can induce population synchrony, multi-population studies of key demographic parameters are imperative to explore the influence of climate change. We used long-term mark-recapture data and position data to determine non-breeding stop-over areas of five Atlantic puffin (Fratercula arctica) populations over a latitudinal gradient in the north-eastern Atlantic (56°11’–70°23’N). We investigated synchrony in adult survival in relation to shared stop-over areas. We quantified effects of extreme extra-tropical cyclones (ETCs) specific to populations’ stop-over areas and the North Atlantic Oscillation on adult survival. Populations with overlapping stop-over areas exhibited temporal synchrony in survival rates. Winter ETCs negatively influenced survival in one population, which was the one most exposed to extreme weather, but did not directly influence adult survival in the other four populations. Synchrony among populations with shared stop-over areas highlights the importance of these areas for adult survival, a key life-history rate. However, extreme weather was not identified as a driving factor for four of the populations. This suggests other factors in these areas, likely related to bottom-up trophic interactions, as environmental drivers of synchrony in the survival of Atlantic puffins.
Thesis
Seabirds are particularly vulnerable to the direct and indirect effects of climate change, however little is known about those impacts outside of the breeding season. This lack of knowledge is problematic because the conditions encountered during migration and wintering strongly shape seabird population dynamics. It is therefore essential to understand the effects of climate on their winter distribution and migration routes. Linking the distribution of organisms to environmental factors is therefore a primary task benefiting from the concept of energyscapes (defined as the variation of an organism's energy requirements across space according to environmental conditions) which has recently provided a mechanistic explanation for the distribution of many animals. In this context, we have predicted the current and future winter habitats of five species representing 75% of the seabird community in the North Atlantic (Alle alle, Fratercula arctica, Uria aalge, Uria lomvia and Rissa tridactyla). To this aim, we monitored the movements of more than 1500 individuals to identify the birds' preferred habitats through resource selection functions based on the modeling of their energy expenditure and prey availability. Electronic tracking data were also overlaid with cyclone locations to map areas of high exposure for the seabird community across the North Atlantic. In addition, we explored the energetic consequences of seabird exposure to storms using a mechanistic bioenergetic model (Niche MapperTM). Finally, we examined the impact of total summer sea ice melt from 2050 on Arctic bird migration. Our analyses predict a northward shift in the preferred wintering areas of the North Atlantic seabird community, especially if global warming exceeds 2°C. Our results suggest that cyclonic conditions do not increase the energy requirements of seabirds, implying that they die from the unavailability of prey and/or inability to feed during cyclones. Finally, the melting sea ice at the North Pole may soon allow 29 species of Arctic birds to make new trans-Arctic migrations between the Atlantic and the Pacific. We also estimate that an additional 26 currently migratory species could remain in the Arctic year-round. This work illustrates how climate change could radically alter the biogeography of migratory species and we provide a methodological toolbox to assess and predict these changes by combining movement ecology and energetic physiology.
Full-text available
Article
We report survival probabilities for 148 breeding adult Atlantic Puffins (Fratercula arctica) monitored through capture-mark-resight at two colonies for 11 years (1992–2003). The colonies, Eastern Egg Rock and Seal Island, are ∼42 km apart in the Gulf of Maine. Support for competing models in the program MARK suggests constant survival of 0.95 ± 0.01 (SE) that is independent of colony. Our high survival probability is consistent with published estimates for Atlantic Puffins and other long-lived seabirds. No time-variance contrasts with many long-term seabird studies, which often report high survival in most years, broken occasionally by low-survival events. However, a post-hoc observation of survival estimates from the time-dependent model suggests that there may have been at least two low-survival events in our time-series; sparse data may have precluded detection by our models. In this study, each bird received an individually engraved, plastic, field-readable leg band, as well as the standard metal band. Using an index of band readability, we show that plastic bands wore rapidly, resulting in accumulating losses of engraved characters through time. Degradation and loss of marks is a common source of overdispersion in capture-mark-re-encounter data and results in underestimated sampling variances. In the presence of a 70% reduction in band readability over eight years, an estimate of the adjusted overdispersion factor (ĉ = 1.14) identified very little overdispersion in our data. Overdispersion was avoided by double banding and intensively resighting metal bands. Estimation de la Survie Adulte de Deux Colonies de Fratercula arctica dans le Golfe du Maine
Full-text available
Article
A new census and monitoring method for populations of breeding Puffins was developed and applied to the population on the island of Hernyken. The method is based on a star-shaped model of evenly spaced sampling plots throughout the topographic surface area of the colony. The main advantages of this approach are that all areas are equally represented, the variance of the estimates can be calculated, and it reveals changes in breeding density in any part of the colony as well as changes of the total colony area. The method can easily be used to survey other resources, even when operating on highly different scales. The number of apparently occupied Puffin burrows on Hernyken decreased from 119 700 in 1979 to 43 160 in 1988. This was presumably the result of very few chicks being reared by the R0st Puffins during 1969-1982. In 1983-1987, when no recruitment was expected, the annual population decline of 13.7% was independent of burrow density and probably equalled the adult mortality rate. In 1988, there was a significant recruitment of first breeders, and these birds settled in the most densely populated areas.
Article
A long-term colour-ringing and resighting programme on the Isle of May, Scotland which involved 734 adult Puffins Fratercula arctica was used to estimate annual survival rates over the period 1973-94 and to investigate some of the factors which influenced survival. Except in 1990 when survival was extremely low, estimated annual survival always exceeded 0.89. There was, however, evidence of a temporal change with survival rates being consistently lower during the second half of the study. This pattern was best described by a stepped model with survival rates of 0.975 (SE 0.004) between 1973 and 1980, 0.924 (0.006) between 1981 and 1994 and a separate value of 0.806 (0.026) for 1990. The decline in survival after 1980 was apparent in all age classes and could not be attributed to a slight change in the study area covered shortly afterwards. There was no evidence that survival rates differed between the sexes but we did find a significant age effect with birds estimated to be 20-25 or more years old either surviving or reproducing much less well than younger individuals.
Article
The proportion of eggs that produced free-flying young increased with increasing breeding experience from an initial 0.4-0.45 to a maximum of 0.75-0.8, before falling to 0.55-0.65 in the most experienced birds. Overall, 10-11% of breeding shearwaters were absent, and 15-18% were present but not recorded with an egg, in any year. Both absentee and non-laying rates declined with increasing age. Older 1st-breeding birds had a higher initial reproductive success than those starting younger. Thereafter, there were no consistent differences in reproductive rates related to age of 1st breeding. Individuals that lived for longer had a higher reproductive success, on average, than shorter-lived birds, especially early in their reproductive careers. -from Authors
Article
Between 1980 and 2000, Atlantic puffins Fratercula arctica and common guillemots Uria aalge breeding in the southern Barents Sea fed their chicks on varying proportions of 4 main categories of prey: capelin Mallotus villosus, sandeels Ammodytes sp., I-group herring Clupea harengus and 0-group cod Gadus morhua. The varying proportions in both numbers and masses of the capelin, herring and cod in the seabird diet showed clear responses to the independently measured prey stocks. Amounts of capelin, herring and 0-group cod fed to Atlantic puffin chicks were good indicators of fish availability, whereas only amounts of herring fed to common guillemot chicks were correlated with the biomass of I-group herring in the region. The more general response by the Atlantic puffins probably results from their ability to catch only small fishes. Despite large interannual differences in prey consumption plus gradual changes in meal sizes, the growth rates of the chicks of both species remained near their maximum, suggesting physiological restraints in growth during plentiful years and/or compensatory foraging behaviour by the adults.
Article
Puffins Fratercula arctica were studied on Skomer Island, Wales (51°44′N, 5°19′W) during 1972-77. Annual survival of breeding adults was 95%. Each year, 20-27% adults were without nesting burrows, and 2% were absent from the colony. 64% of pairs with burrows fledged a chick, with 5-16% not laying, 22-25% eggs not hatching, and 5% chick mortality. Much of the egg loss was caused by disturbance from prospecting Manx Shearwaters Puffinus puffinus which competed with Puffins for burrows. Data on the feeding and growth of chicks is given. Young Puffina first returned to the colony at two, or more usually three, years old. Four years was the earliest age for first breeding. At least 10-16% fledglings survived to four years old; it was not clear whether enough survived to replace adult mortality.