Content uploaded by Morten Frederiksen
Author content
All content in this area was uploaded by Morten Frederiksen on Feb 05, 2015
Content may be subject to copyright.
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
Vol. 454: 119–133, 2012
doi: 10.3354/meps09520 Published May 21
INTRODUCTION
Phenology (the timing of seasonally recurring bio-
logical events) plays a crucial role in linking organisms
to their biotic and abiotic environments (Forrest &
Miller-Rushing 2010). Recent climate warming has
significantly altered the phenology of a wide range of
taxa across ecosystems (Thackeray et al. 2010), but re-
sponses frequently vary among species, potentially
disrupting the synchronisation of key ecological inter-
actions (Visser & Both 2005). In particular, failure of a
predator to overlap the period of peak resource de-
© Inter-Research 2012 · www.int-res.com*Email: sburthe@ceh.ac.uk
Phenological trends and trophic mismatch across
multiple levels of a North Sea pelagic food web
Sarah Burthe1,*, Francis Daunt1, Adam Butler2, David A. Elston3,
Morten Frederiksen1, 4, David Johns5, Mark Newell1, Stephen J. Thackeray6,
Sarah Wanless1
1Centre for Ecology & Hydrology, Bush Estate, Penicuik, Midlothian EH26 0QB, UK
2Biomathematics and Statistics Scotland, The Kings Buildings, Edinburgh EH9 3JZ, UK
3Biomathematics and Statistics Scotland, Craigiebuckler, Aberdeen AB15 8QH, UK
4Dept. of Bioscience, Aarhus University, Frederiksborgvej 399, 4000 Roskilde, Denmark
5Sir Alister Hardy Foundation for Ocean Science, The Laboratory, Citadel Hill, Plymouth PL1 2PB, UK
6Centre for Ecology & Hydrology, Lancaster Environment Centre, Bailrigg, Lancaster LA1 4AP, UK
ABSTRACT: Differential phenological responses to climate among species are predicted to dis-
rupt trophic interactions, but datasets to evaluate this are scarce. We compared phenological
trends for species from 4 levels of a North Sea food web over 24 yr when sea surface temperature
(SST) increased significantly. We found little consistency in phenological trends between adjacent
trophic levels, no significant relationships with SST, and no significant pairwise correlations
between predator and prey phenologies, suggesting that trophic mismatching is occurring. Finer
resolution data on timing of peak energy demand (mid-chick-rearing) for 5 seabird species at a
major North Sea colony were compared to modelled daily changes in length of 0-group (young of
the year) lesser sandeels Ammodytes marinus. The date at which sandeels reached a given
threshold length became significantly later during the study. Although the phenology of all the
species except shags also became later, these changes were insufficient to keep pace with sandeel
length, and thus mean length (and energy value) of 0-group sandeels at mid-chick-rearing
showed net declines. The magnitude of declines in energy value varied among the seabirds, being
more marked in species showing no phenological response (shag, 4.80 kJ) and in later breeding
species feeding on larger sandeels (kittiwake, 2.46 kJ) where, due to the relationship between
sandeel length and energy value being non-linear, small reductions in length result in relatively
large reductions in energy. However, despite the decline in energy value of 0-group sandeels dur-
ing chick-rearing, there was no evidence of any adverse effect on breeding success for any of the
seabird species. Trophic mismatch appears to be prevalent within the North Sea pelagic food web,
suggesting that ecosystem functioning may be disrupted.
KEY WORDS: Timing of breeding · Climate change · Prey size · Ammodytes marinus · Winter
NAO · Long-term studies · Zooplankton · Phytoplankton
Resale or republication not permitted without written consent of the publisher
Contribution to the Theme Section ‘Seabirds and climate change’
O
PEN
PEN
A
CCESS
CCESS
Mar Ecol Prog Ser 454: 119–133, 2012
mand (typically breeding) with peak prey availability
may lead to ‘trophic mismatch,’ and such decoupling
may alter food web structure and eco systems (Cushing
1990, Edwards & Richardson 2004). A recent review
found that trophic mismatch was widespread, with
predator phenology shifting too little or too much in
response to that of prey (Visser & Both 2005).
Marine systems are vulnerable to trophic mismatch
because they exhibit highly seasonal pulses of pri-
mary productivity upon which the fitness of higher
trophic levels depends (Cushing 1990). Studies
across multiple trophic levels are rare, mainly be -
cause phenological data at the appropriate temporal
and spatial resolution are lacking. Most
studies have investigated single species
responses and are unable to explicitly test for
mismatch (Leggett & DeBlois 1994). Studies
have also compared phenology of a focal
consumer with climate data such as sea-
surface temperature (SST; e.g. Durant et al.
2003, Frederiksen et al. 2004a, Shultz et al.
2009) that may indicate variation in prey
availability, including phenology, or be used
as a cue by predators to predict key pheno-
logical events in their prey (Frederiksen et
al. 2004a, Moe et al. 2009). Several studies
have suggested that mismatch may be an im -
portant determinant of fitness in seabirds
(e.g. Durant et al. 2006, Hipfner 2008, Wata -
nuki et al. 2009).
Here we examine phenological changes
across 4 trophic levels of a pelagic food web
in the north-western North Sea from 1983
through 2006. This system has a ‘wasp-waist’
structure (Cury et al. 2000), with high species
richness at upper and lower trophic levels
but markedly lower richness at the mid-
trophic position linking secondary producers
(zooplankton) and top predators (mammals,
fish and birds). Lesser sandeel Ammodytes
marinus occupies this key mid-trophic posi-
tion (Daan et al. 1990). Over the study period
there have been significant hydro-biological
changes and increased sea temperatures in
this area (Edwards et al. 2002). A major
ecosystem regime shift occurred in the late
1980s (Beaugrand 2004), and there have
been profound changes in plankton commu-
nities (Edwards et al. 2002) and fish distribu-
tions (Perry et al. 2005). Previous studies in
this area have investigated phenological
changes in species from primary producers
to top predators and found contrasting pat-
terns suggestive of trophic mismatch (Edwards &
Richardson 2004, Wanless et al. 2009, Frederiksen et
al. 2011). However, none have ad opted an integrated
ap proach and compared multiple trophic levels
within the same area over the same time period. A
major aim of our study was to use a standardised
approach to quantify changes in the timing of key
events for species or taxonomic groups across all 4
trophic levels (Fig. 1a). Disparity in phenological
trends would be indicative of trophic mismatch in the
system. We also assessed whether phenology was
related to climate variables (SST and winter North
Atlantic Oscillation, wNAO) and if climate res pon ses
120
Fig. 1. Schematic illustration of the different stages of analysis under-
taken: (a) analyses of the 4 trophic levels to examine trends in phe-
nology; trends between phenology and climate, and relationships be-
tween phenologies of adjacent trophic levels; (b) detailed analyses of
the seabird and sandeel data to examine phenological trends and the
impact of mismatch on breeding success. SST: sea surface tempera-
ture; wNAO: winter North Atlantic Oscillation; T: month (day of year)
of central tendency
Burthe et al.: North Sea trophic mismatch
were similar in terms of their magnitude and direc-
tion across the different trophic levels.
To investigate in more detail how the relative tim-
ing of trophically linked events have changed over
time, we focussed on 5 seabird species and their
major sandeel prey (Furness & Tasker 2000, Daunt et
al. 2008). Although trophic mismatch theory is most
often applied to the timing of peak prey abundance,
prey size is a key component and is known to be
important for seabirds in our study area (Wanless et
al. 2005). We therefore compared modelled annual
length-at-date of 0-group sandeels (Frederiksen et
al. 2011; sandeel hatched in the current year) with
the timing of peak energy demand in each seabird
species, which we assumed corresponded to the mid-
chick-rearing period (Drent & Daan 1980) (Fig. 1b).
Under this modified version of the mismatch hypo -
thesis our prediction was that mid-chick-rearing
should coincide with sandeels having attained a
threshold size, since individuals are expected to bal-
ance increasing prey quality through the season with
the fitness advantages of breeding as early as possi-
ble (Daunt et al. 2007, Harris et al. 2007). We
assessed whether the timing of chick-rearing had
become decoupled from seasonal changes in sandeel
length, and estimated mean size of fish at mid-chick-
rearing to quantify the consequences of mismatch on
prey energy value. Finally, we used the mismatch
index to explore the fitness consequences of mis-
matching on seabird breeding success.
MATERIALS AND METHODS
Climate
Monthly average SST data were obtained from
NOAA Pathfinder Version 5.0 (Kilpatrick et al. 2001)
for an area of the North Sea (55 to 58°N, 3°W to 0°E)
between 1983 and 2006. Since spring events were our
main interest for comparison across the trophic levels
we focussed SST analysis on February and March
values (hereafter winter/spring SST). June and July
(hereafter summer SST) values were also modelled
with seabird breeding phenology as this period over-
lapped with mid-chick-rearing. As large-scale sea-
sonal measures of climate have been found to be use-
ful predictors of ecological processes (Hallett et al.
2004) and as 4 of the 5 seabird species being consid-
ered may be distributed outside the western North
Sea during winter, we also considered wNAO indices
(www.cgd.ucar.edu/cas/jhurrell/ in dices.html) for the
winter prior to spring phenology events.
Phenology data
Phenology data were available across all trophic
levels between 1983 and 2006, and analyses were
thus restricted to this period.
Phytoplankton (primary producers) and
zooplankton (primary consumers)
The continuous plankton recorder (CPR) survey
is an upper layer plankton monitoring programme
(Richardson et al. 2006). We analysed a subset of
plankton data that are important in the diet of
sand eels Ammodytes marinus. Phytoplankton and
copepod nauplii are the main prey of larval sand -
eels (Monteleone & Peterson 1986), while older
stages of calanoid copepods (particularly Temora
spp. and Calanus spp.) are important for postlarval
stages, <10 cm in length (Macer 1966). According -
ly, we focussed on spring-peaking copepod species
(sandeel hatch date occurred mainly in February to
March) and ana lysed data for C. helgolandicus, C.
finmarchicus, T. longicornis, Calanus spp. Stages
I to IV, and copepod nauplii. Calanus spp. nauplii
feed preferentially on diatoms (Soreide et al.
2008). As there was no evidence that particular
diatom species were important for copepods, we
analysed the total summed monthly abundances
of the 10 most abundant diatoms in spring (indi-
vidual species data are presented in Table S1 of
the supplement at www. int-res.com/articles/suppl/
m454 p119_ supp. pdf) and a colour in dex of phyto-
plankton. Data were obtained from an area of the
North Sea (55 to 58° N, 3° W to 0° E; Johns 2009)
that provided a balance between sampling reso lu -
tion and proximity to the Isle of May, Scotland
(56° 11’ N, 2° 33’ W), the focal point of sea bird
observations.
For plankton, the phenology measure was the
month of central tendency, T, (see Edwards & Rich -
ardson 2004) converted to day of year for compari-
son with other taxa. The average monthly abun-
dances over the 24 yr period for each species were
used to determine whether the seasonal pattern was
unimodal or bimodal (spring and autumn). For uni-
modal taxa, Twas calculated using data from the
entire year, whereas, for bimodal taxa, it was calcu-
lated using data from the first 6 mo of the year (see
Ed wards & Richardson 2004). Due to missing
monthly values for all species of plankton in 1995,
this year was omitted from plankton phenology
analyses.
121
Mar Ecol Prog Ser 454: 119–133, 2012
Sandeels (secondary consumers)
Estimates of sandeel phenology (hatch dates) were
obtained from a statistical model implemented using
Markov Chain Monte Carlo procedures that com-
bined 2 time series of sandeel size at date data, i.e.
from larval fish captured during CPR surveys and 0-
group fish obtained from foraging puffins (see Fred-
eriksen et al. 2011 for full details).
Seabird predators
Analysis focussed on 5 species of seabirds for
which 0-group sandeels are an important dietary
component for adults and/or chicks on the Isle of
May (Daunt et al. 2008): common guillemot Uria
aalge (hereafter guillemot), razorbill Alca torda,
European shag Phalacrocorax aristotelis (hereafter
shag), black-legged kittiwake Rissa tridactyla (here-
after kittiwake) and Atlantic puffin Fratercula arctica
(hereafter puffin). Median egg dates were recorded
in guillemots and razorbills from daily checks of
monitoring plots (mean of ca. 800 and ca. 100 breed-
ing sites, respectively). For shags, annual median
ringing dates of chicks were analysed (mean ca. 800
chicks ringed mid-way through the chick-rearing
period at a mean age of 20 to 25 d) since median lay-
ing dates (estimated from weekly checks of ca. 100
pairs) were only available for a subset of years and
were strongly correlated with median egg date (r =
0.94, df = 20, p < 0.001). Kittiwake first egg dates
were analysed (from daily checks of the whole colony
of >3000 pairs) as median egg dates were only
recorded in a subset of years (from 5 d checks of ca.
200 pairs; correlation between first and median:
r = 0.90, df = 8, p < 0.001). First egg dates were also
analysed for puffins, back-calculated from daily
checks of the entire colony for adults bringing fish
back to chicks (>10 000 pairs; see Wanless et al. 2009
for details) since median egg dates (based on back-
calculation from wing and bill measurements of
chicks from a mean of ca. 30 individuals; see Harris &
Wanless 2011 for details) were only available for a
subset of years (correlation between first and
median: r = 0.61, df = 13, p = 0.015). Even though first
egg dates are likely to be subject to a higher level of
stochastic variation than median egg dates (Wanless
et al. 2009), they were assumed to be reliable indica-
tors of the timing of breeding for kittiwakes and
puffins because of the correlation with median egg
dates in the subset of years where both were
recorded.
Phenological regressions
Standard linear regressions with year were used to
investigate temporal trends in SST, wNAO and phe-
nology of trophic levels. In order to avoid false detec-
tion of significant correlations due to multiple testing,
we applied the Benjamini & Hochberg (1995) correc-
tion factor to this set of models. In all cases we report
the uncorrected p-values. Phenology data for sandeel
hatch date showed evidence of a break-point (Fred-
eriksen et al. 2011), and were therefore analysed us-
ing a piecewise regression model employing the seg-
mented package in R (Muggeo 2008). In order to
assess whether the phenologies of consumers and
their prey covaried over time, we examined whether
there were pairwise correlations between phenolo-
gies of taxa across successive trophic levels. We in-
vestigated whether phenology was correlated with
climate by regressing the phenology of each species
against SST or wNAO. We analysed trends using sim-
ple linear regression, without taking temporal auto-
correlation into account, and thus assume that con-
secutive years are independent. Autocorrelation plots
of model residuals were examined and, in general,
showed no apparent evidence of autocorrelation,
suggesting that the assumption of independence was
reasonable. There was only 1 case (linear regression
of kittiwake first egg phenology and wNAO) in which
the regression coefficient was significant and the
model residuals showed evidence of autocorrelation.
Trophic mismatch in seabirds and sandeels
To evaluate phenological mismatch in greater
detail we focussed on interactions between seabirds
and 0-group sandeels, since, not only was the tempo-
ral resolution of these data markedly better than for
lower trophic levels, but information on other aspects
of performance such as sandeel growth rates and
seabird breeding success was also available. We
focussed on the peak period of energy demand for
the seabirds, i.e. the mid-point of the chick-rearing
period (Drent & Daan 1980). For shags this was esti-
mated directly from median chick ringing date as this
occurs midway through the chick-rearing period. For
guillemots and razorbills we used the species- and
year-specific median laying date plus the average
incubation period plus the average chick-rearing
period/2, while for kittiwakes and puffins we used
first egg date plus average difference between first
and median egg dates plus average incubation
period plus the average chick-rearing period/2. Val-
122
Burthe et al.: North Sea trophic mismatch
ues for incubation and fledging periods were ob -
tained from Cramp & Simmons (1978, 1983). The
average difference between first and median egg
dates was 11 d for kittiwake (range: 6 to 18 d, n = 10)
and 12 d for puffin (range: 6 to 17 d, n = 15).
The sandeel model (Frederiksen et al. 2011) esti-
mated mean hatch dates and growth rates, from
which daily size of juvenile sandeels was then esti-
mated. We compared relationships between the tim-
ing of mid-chick-rearing in seabirds and 2 phenolog-
ical metrics for sandeels: hatch dates and the date
each year that 0-group sandeels reached a predicted
mean threshold length of 55 mm. This was under-
taken because many phenological studies investi-
gate the timing of appearance of prey (hatch dates),
but we also wanted to test whether phenology of
sandeel size and hence prey quality was more rele-
vant to seabird predators. Other threshold sizes were
also analysed (Fig. S1 in the supplement at www.
int-res. com/articles/suppl/m454p119_supp. pdf), and
model fit was found to increase with sandeel size,
with 55 mm being the largest threshold size that
sandeels attained in all years of the study. Thus, this
was a useful measure to compare with timing of mid-
chick-rearing for each seabird species. Linear
regression with year was used to assess whether the
date sandeels reached 55 mm showed evidence of a
temporal trend.
In order to evaluate whether the timing of seabird
mid-chick-rearing had become decoupled from the
timing of availability of quality sandeel prey over the
study period, a ‘mismatch index’ was calculated as
the difference (in days) between the date of the mid-
point of chick-rearing for each seabird species and
the date that mean sandeel size was predicted to
reach the threshold of 55 mm. We emphasise that as
the mismatch index is based on a threshold sandeel
size, a mismatch index of 0 does not indicate perfect
matching of predator and prey timing. Instead posi-
tive values of the index indicate that the seabirds’
peak energy demand occurred after sandeels
reached 55 mm, while negative values indicated that
peak demand preceded this threshold being at tained.
Thus, more positive values potentially indicated bet-
ter matching with higher quality prey (larger fish)
and more negative values indicated peak demand co-
inciding with poorer quality prey. We assessed tem-
poral changes in this mismatch index for each seabird
species using linear regressions with year. Finally, we
estimated mean sandeel size at mid-chick-rearing to
quantify the effects of mismatch on prey energy value
using the equation relating sandeel length to energy
value from Hislop et al. (1991). Due to the non-linear
nature of this relationship, declines in size of large
fish are more energetically costly than similar
declines in smaller fish. Linear regression with year
assessed whether there had been any systematic
change in sandeel size at this time.
Generalised linear modelling was used to evaluate
whether changes in seabird breeding success were as-
sociated with sandeel phenology and mismatch para-
meters, selecting models by Akaike’s information cri-
terion corrected for small sample sizes (AICc; Hurvich
& Tsai 1989, Burnham & Anderson 2002). Breeding
success was defined to be the ratio of the total number
of chicks fledged to the total number of chicks that
could potentially have fledged (a proportion). The
total number of chicks that could potentially have
fledged is equal to the maximum brood size multiplied
by the total number of nests at which eggs were laid.
Maximum brood size is invariably 1 for guillemots, ra-
zorbills and puffins, and typically 3 for kittiwake and 4
for shag (Cramp & Simmons 1978, 1983). Details of
sample sizes and field methodology for monitoring
breeding success are given in Harris et al. (2005). In
addition to the sandeel phenology and mismatch para -
meters (sandeel hatch date, sandeel growth rate,
length of sandeels at the mid-point of chick-rearing
and mismatch index) we also included in model selec-
tion the following extrinsic factors that have previously
been shown to correlate with breeding success for
these seabird species on the Isle of May (Frederiksen
et al. 2004b, 2006): lagged sandeel biomass index (an
annual index modelled from the probability of sand eel
larvae occurring in CPR samples and summed mass of
larvae in a sample; see Frederiksen et al. 2006 for de-
tails); lagged SST (previous year) and sandeel fishery
presence (kittiwake only; Frederiksen et al. 2008).
As the sandeel variables (apart from sandeel bio-
mass index) arise from the same statistical model
(Frederiksen et al. 2011; see Table S3 in the supple-
ment at www.int-res.com/articles/suppl/m454 p119_
supp.pdf) we inclu ded at most one of these variables
in each model. We used summed Akaike weights to
calculate the relative strength of support for each of
the potential predictors. Note that sandeel phenology
and mismatch parameters each appeared in 8 of the
40 models within the candidate set (a prior weight
of 0.2), whereas the variables relating to extrinsic fac-
tors each appeared in 20 of the 40 models (a prior
weight of 0.5) —this difference must be taken into ac-
count when drawing comparisons between the 2
groups of variables.
Regression models were applied to logit-trans-
formed data on breeding success for guillemot, razor-
bill and puffin, and to log-transformed data for
123
Mar Ecol Prog Ser 454: 119–133, 2012
kittiwake and shag, as well as being applied to
untransformed data for all species. The same models
(i.e. the same sets of explanatory variables) were
selected by AICc for both transformed and untrans-
formed data; we present the results for the untrans-
formed data solely in order to allow direct compari-
son with the results of Frederiksen et al. (2006). We
considered the inclusion of quadratic, as well as lin-
ear, relationships and sandeel parameters lagged by
1 yr, but found no support for inclusion of these
terms. Spurious relationships between an explana-
tory variable and the response variable can arise if
both are correlated with a third variable, particularly
time (Grosbois et al. 2008). We therefore included
year as an explanatory variable in order to assess
whether the same best model was selected once year
was included, and whether the addition of year
improved model fit.
RESULTS
Climate
Winter/spring SST increased by an average (±SE)
of 0.056 ± 0.014°C yr−1; p < 0.001, an in crease of
1.34°C over the study period. Summer SST also in -
creased significantly (overall increase: 1.42°C; 0.059
± 0.016°C yr−1; p = 0.002). In contrast, there was no
significant trend in the wNAO (estimate: −0.097 ±
0.060; p = 0.118; Fig. 2).
Phenological trends
For primary producers, neither the timing of the
seasonal peak of overall summed phytoplankton
abundance, nor that of the colour index showed a sta-
tistically significant trend (Fig. 3). Similarly, phenol-
ogy of primary consumers appeared to be largely
unchanged and only 1 species, Temora longicornis,
showed a significant advancement in timing. Sand -
eel Ammodytes marinus hatch date showed, within
the piecewise regression model, a highly significant
break-point in 1995 (95% CI from 1991 to 1998,
p = 0.001; model R2= 0.452), with hatching initially
becoming later and then becoming earlier (Fig. 3).
With the exception of shags, whose timing varied
greatly from year to year, seabird breeding tended to
become later, with significant trends for guillemot
and kittiwake (Fig. 3). All significant trends re -
mained significant after the Benjamini and Hochberg
correction factor was applied to this set of models.
124
Winter/spring SST
4
6
8
Summer SST
10
12
14
wNAO
1990 2000
–4
–2
0
2
4
Year
wNAO index Mean SST (ºC)
Trend (days per decade)
–30
–20
–10
0
10
20
30
✶
✶
✶
✶
✶
Phytoplankton colour index
Summed phytoplankton
Calanus stages I-IV
Temora longicornis
Calanus finmarchicus
Calanus helgolandicus
Copepod nauplii
Sandeel hatch date (1983 to 1995)
Sandeel hatch date (1995 to 2006)
Guillemot median egg
Razorbill median egg
Shag median ringing
Kittiwake first egg
Puffin first egg
Fig. 3. Phenology trends (negative values below the line in-
dicate timing becoming earlier, and positive values above
the line indicate timing becoming later; n = 23 for plankton
species and n = 24 for sandeel and seabirds) for species/
groups from the 4 trophic levels with standard errors
(trophic levels are shaded differently: palest grey: primary
producers; darkest grey: top predators) in a North Sea
pelagic food web between 1983 and 2006. ✶Significant
trends after correction (p < 0.05)
Fig. 2. Mean winter/spring and summer sea-surface temper-
ature (SST) values, and the winter North Atlantic Oscillation
(wNAO) index score over the study period. Fitted lines show
significant regressions
Burthe et al.: North Sea trophic mismatch
Phenological regressions with climate
Overall there was little evidence that trends in phe-
nology were associated with either of the climate
variables. Significant relationships between phenol-
ogy and winter/spring SST (see Table 1) for Temora
longicornis and Calanus spp. Stages I to IV were
apparent, but did not remain significant after apply-
ing the Benjamini and Hochberg correction to this set
of models. None of the regressions were significant
between seabird egg-laying phenology and winter/
spring SST (Table 1) or summer SST (guillemot:
t= 0.50, p = 0.62; razorbill: t= −1.22, p = 0.24; shag:
t= −0.69, p = 0.50; kittiwake: t= 0.911, p = 0.37;
puffin: t= 0.40, p = 0.70). Phytoplankton colour index,
but not summed abundance, was positively related to
wNAO, while timing of guillemots, razorbills and
kittiwakes showed a negative relationship. Only the
regression with kittiwake pheno logy remained
significant once the correction factor had been
applied to this set of models. How-
ever, this model also showed some
evidence of autocorrelation when
model resi duals were examined and
hence should be interpreted with
caution.
Comparisons of phenological
change among trophic levels
There was no evidence that pre -
dator and prey phenologies were
related, with no significant pairwise
regressions between any of the tro -
phic comparisons (the significant re -
lationship between sandeel hatch
and timing of peak abundance
of Cala nus helgo landicus was no
lon ger significant after correction
(Table 2 and Table S2 in the supple-
ment at www.int-res.com/ articles/
suppl/ m454 p119_supp. pdf).
125
Phenology measure n Day of year Spring SST wNAO
Mean SD Slope SE p R2Slope SE p R2
estimate (%) estimate (%)
Phytoplankton colour T23 103.00 11.09 −1.89 3.82 0.63 1.16 2.54 1.04 0.02 22.20
Summed phytoplankton T23 98.35 11.24 −1.32 3.88 0.74 0.55 −0.86 1.18 0.47 2.50
Calanus spp. Stages I to IV T23 115.65 15.73 −10.40 4.95 0.05 17.35 1.18 1.65 0.48 2.39
Temora longicornis T 23 120.35 15.72 −12.64 4.70 0.01 25.66 0.29 1.67 0.86 0.14
C. finmarchicus T 23 96.09 14.65 −3.66 5.01 0.47 2.48 −0.37 1.55 0.82 0.26
C. helgolandicus T 23 66.70 26.82 12.64 8.88 0.17 8.81 3.97 2.71 0.16 9.26
Copepod nauplii T23 123.87 13.00 0.80 4.50 0.86 0.15 −0.28 1.38 0.84 0.20
Sandeel hatch date 24 71.44 9.03 0.52 3.12 0.87 0.13 0.93 0.90 0.31 4.65
Date sandeels reach 55 mm 24 162.17 16.45 7.62 5.45 0.18 8.18 −1.88 1.62 0.26 5.79
Guillemot median egg 24 128.13 3.79 0.71 1.30 0.59 1.33 −0.93 0.33 0.01 26.28
Razorbill median egg 24 130.17 3.69 −0.76 1.26 0.55 1.63 −0.76 0.34 0.03 18.93
Shag median ringing 22 132.55 19.82 −7.14 5.96 0.24 6.13 0.71 1.80 0.70 0.70
Kittiwake first egg 24 131.50 8.46 1.37 2.91 0.64 0.99 −2.35 0.70 0.00 34.14
Puffin first egg 24 99.88 5.10 1.59 1.73 0.37 3.68 −0.70 0.50 0.17 8.41
Table 1. Linear regressions of phenology against mean winter/spring sea-surface temperature (SST) and winter North Atlantic
Oscillation (wNAO), together with uncorrected p-values. Significant regressions (at the 5% level, after applying the Benjamini
and Hochberg correction factor) are highlighted in bold. T: central tendency for monthly plankton data
Response Explanatory Slope SE t p
estimate
Calanus finmarchicus T Phytoplankton colour T −0.476 0.269 −1.768 0.092
C. helgolandicus T Phytoplankton colour T −0.119 0.527 −0.227 0.823
Calanus stages T Phytoplankton colour T 0.277 0.304 0.914 0.371
Temora longicornis T Phytoplankton colour T 0.282 0.303 0.929 0.363
Copepod nauplii T Phytoplankton colour T 0.258 0.250 1.032 0.314
Sandeel hatch C. finmarchicus T 0.038 0.111 0.344 0.734
Sandeel hatch C. helgolandicus T 0.128 0.054 2.377 0.027
Sandeel hatch Calanus stages T 0.096 0.101 0.949 0.354
Sandeel hatch T. longicornis T −0.011 0.103 −0.110 0.913
Sandeel hatch Copepod nauplii T −0.175 0.119 −1.469 0.157
Sandeel hatch Phytoplankton colour T −0.056 0.146 −0.380 0.708
Guillemot median egg Sandeel hatch 0.034 0.089 0.376 0.711
Razorbill median egg Sandeel hatch 0.075 0.086 0.874 0.392
Shag median ring Sandeel hatch −0.332 0.414 −0.802 0.431
Kittiwake first egg Sandeel hatch 0.088 0.199 0.440 0.664
Puffin first egg Sandeel hatch 0.106 0.118 0.893 0.382
Table 2. Linear regressions of phenology of upper trophic level species/group against
that of the relevant lower trophic level, together with uncorrected p-values. Results for
summed total phytoplankton abundance were similar to the colour index (see Table S2
in the supplement at www.int-res.com/articles/suppl/m454p119_ supp. pdf). Tis the
central tendency for monthly plankton data
Mar Ecol Prog Ser 454: 119–133, 2012
Trophic mismatch in seabirds and sandeels
The date at which 0-group sandeels reached a pre-
dicted mean threshold length of 55 mm became sig-
nificantly later, at an average rate of 13.1 d decade−1
over the study period (Fig. 4). The date of mid-chick-
rearing was significantly related to the estimated date
this threshold length was reached for 4 of the seabird
species (guillemot, razorbill, puffin and kittiwake;
Fig. 5), and relationships for guillemot, razorbill and
kittiwake remained significant even when year was
also included in the model. However, the regression
coefficient (slope) of mid-chick-rearing against the
date sandeel threshold length was reached was sub-
stantially less than unity for these species (Fig. 5). This
indicated that although mid-chick-rearing had be-
come later in these 4 species, the shift in timing of
sandeel size had been even faster. Razorbills showed
the slowest rate of change in timing of breeding, and
kittiwakes, the fastest. In contrast, shags showed no
temporal trend in breeding phenology or relationship
with sandeel size phenology (Fig. 5).
In addition to differing rates of phenological
change, there was also interspecific variation in the
126
1985 1990 1995 2000 2005
1 Jun 1 Jul 1 Aug
Slope = 13.1 days per decade
R2 = 31.6%
p = 0.004
Year
Date sandeels reached 55 mm
■
■
■
■
■■
■
■
■
■
■■
■
■
■
■
■
■
■■■
■■
■
1 Jun 1 Jul 1 Aug
1 Jun 1 Jul 1 Aug
Slope = 1.5 days per decade
R2 = 42.5%
p < 0.001
Guillemot
Slope = 1.2 days per decade
R2 = 30.5%
p = 0.005
Razorbill
p = 0.73
Shag
1 Jun 1 Jul 1 Aug 1 Jun 1 Jul 1 Aug
1 Jun 1 Jul 1 Aug 1 Jun 1 Jul 1 Aug 1 Jun 1 Jul 1 Aug
Slope = 3.3 days per decade
R2 = 42.3%
p < 0.001
Kittiwake
Slope = 1.4 days per decade
R2 = 21.7%
p = 0.022
Puffin
Date predicted mean sandeel sizes reached 55 mm threshold
Date of mid chick-rearing
Fig. 4. Regression between the date that mean sizes of
sandeel Ammodytes marinus are predicted to attain a
threshold of 55 mm and year. The line shows the significant
fitted regression. Relationship remained significant even
when the latest date sandeels reached 55 mm (in 2004,
dotted circle) was removed from the analysis
Fig. 5. Regressions between the
date of peak chick demand and the
date that predicted mean sizes of
sandeel Ammodytes marinus reach
a threshold of 55 mm. Lines show
significant fitted regressions, and
relationships remained significant
even when the latest date sandeels
reached 55 mm (in 2004, dotted cir-
cle) was removed from the analysis
Burthe et al.: North Sea trophic mismatch
absolute timing of breeding. Ranking of the timing of
breeding for 4 of the 5 seabird species was generally
constant across the study period, with puffins and
guillemots breeding earliest, followed by razorbills,
and with kittiwakes breeding last. In contrast, shag
mid-chick-rearing was highly variable, being the
earliest studied bird in 3 of the years and the latest in
9 of the years.
These disparities in absolute timing and rates of
change in seabird breeding schedules and sandeel
sizes were integrated in the mismatch index (Fig. 6).
Thus, in the 1980s, mid-chick-rearing for all the
species considered occurred well after 0-group
sandeels reached 55 mm (positive values of the mis-
match index), whereas, in recent years, mid-chick-
rearing coincided with (mismatch index around 0)
or oc curred before (negative values) the date at
which sand eels had attained this size in 4 of the 5
seabird species. Kittiwakes were the exception; they
bred latest and hence had generally higher mis-
match index values than the other species (Fig. 6).
As a result, the mean length of 0-group sandeels at
mid-chick-rearing has significantly decreased over
the study period in all seabird species (Fig. 7). The
total estimated decrease over 24 yr and rates of de -
cline (±SE) were as follows: for guillemots: 10.2 mm,
−0.44 (±0.16) mm yr−1; for razorbills: 12.4 mm, −0.54
(±0.16) mm yr−1; for shags: 21.7 mm, −0.94 (±0.28)
mm yr−1; for kittiwakes: 10.2 mm, −0.44 (±0.19) mm
yr−1; and for puffins: 10.7 mm, −0.47 (±0.15) mm
127
Chick rearing
before sandeels after sandeels
= 55 mm = 55 mm
Guillemot
Sandeel size
Razorbill
1985 1990 1995 2000 2005
Shag Kittiwake
1985 1990 1995 2000 2005
Puffin
1985 1990 1995 2000 2005
1985 1990 1995 2000 2005 1985 1990 1995 2000 2005
Year
Mismatch index (days)
80
40
0
–40
80
40
0
–40
Fig. 6. Temporal changes in mismatch index (difference in days between timing of seabird mid-chick-rearing and the mean
date that sandeels Ammodytes marinus were predicted to reach a threshold of 55 mm). As shown in upper left panel, a nega-
tive index means that mid-chick-rearing occurred before sandeels attained a threshold length of 55 mm; hence, sandeels
would have been 55 mm and smaller during chick-rearing. Positive index values indicate that mid-chick-rearing occurred af-
ter this threshold; hence, sandeels would have been 55 mm or larger during chick-rearing. Solid lines: significant fitted regres-
sions; dashed line (kittiwakes): non-significant. We emphasise that because the mismatch index is based on a threshold size of
sandeel, a value of 0 for the mismatch index does not indicate perfect matching between seabirds and sandeels; rather the
higher the mismatch index the larger the sandeels available
Mar Ecol Prog Ser 454: 119–133, 2012
yr−1. Shags, with the steepest rate of predicted
decline, showed the highest net reduction in en -
ergetic value of fish over the 24 yr study period
(4.80 kJ, a 70.4% decline from 1983). Despite kitti-
wakes having the lowest rate of decline in sand eel
size, this species showed the next highest overall
reduction in net energetic value of 2.46 kJ (42.2%
decline from 1983). As the latest breeding seabird,
mid-chick-rearing of kittiwakes occurred when
sandeels were predicted to be larger and, due to the
non-linear nature of the relationship between sand -
eel length and energetic content, declines in the
length of large fish are more ener getically costly
than equivalent declines in smaller fish. Net de -
clines in the energetic value of fish were 1.70 kJ
(46.7%) for guillemot, 2.21 kJ (52.4%) for razorbill
and 1.79 kJ (48.2%) for puffin.
Consequences of mismatch for seabird
breeding success
Despite the energetic implications associated with
the pronounced decline in 0-group sandeel length
at mid-chick-rearing, there was no evidence that
this or sandeel hatch date had a significant effect on
the breeding success of any of the 5 species of
seabirds considered once other significant variables
were in cluded in the models. The breeding success
of guillemot, razorbill and shag was, however, posi-
tively related to sandeel growth rates (slope esti-
mates ± SE: 0.849 ± 0.204, 0.341 ± 0.193, and 0.729 ±
0.333, respectively) such that in years with poorer
sandeel growth these species had poorer breeding
success.
Breeding success for shags, kittiwakes and puffins
was poorer in years following a year with warmer
winter/spring SST (slope estimates ± SE: −0.101 ±
0.042, −0.126 ± 0.030, −0.108 ± 0.028, respectively)
and higher following years with a high sandeel bio-
mass index (slope estimates ± SE: 0.047 ± 0.020, 0.035
± 0.018, and 0.032 ± 0.013, respectively). All species
ex cept kittiwake showed evidence of a linear trend
(negative for all species except shag) in success over
time that was not accounted for by the other model
variables (Table 3). The inclusion of quadratic or
lagged sandeel terms did not lead to models with
lower AICc values.
128
Guillemot
R2 = 27.3%
p = 0.009
Razorbill
R2 = 35.2%
p = 0.002
Shag
1985 1990 1995 2000 2005
R2 = 35.7%
p = 0.003
Kittiwake
1985 1990 1995 2000 2005
R2 = 20.6%
p = 0.026
Puffin
1985 1990 1995 2000 2005
1985 1990 1995 2000 2005 1985 1990 1995 2000 2005
R2 = 30.5%
p = 0.005
Year
Year
Size of sandeels at mid-point of chick rearing (mm)
90
80
70
60
50
90
80
70
60
50
Fig. 7. Predicted size (mm) of
Ammo dytes marinus sandeel at date
of mid-chick-rearing for each sea -
bird species over the study period
Burthe et al.: North Sea trophic mismatch
DISCUSSION
Phenological changes across trophic levels
To our knowledge, this is the first direct comparison
of phenologies across multiple trophic levels of the
North Sea pelagic food web. We found contrasting
trends across the 4 trophic levels that were suggestive
of trophic mismatch, supporting the assertion from
other studies that this phenomenon is widespread
(Visser & Both 2005). However, in contrast to mis-
match theory for marine systems (Cushing 1990) and
empirical results from terrestrial studies (Visser &
Both 2005, Both et al. 2009), we found no evidence of
strong pairwise relationships between predator and
prey phenologies, except for timing of mid-chick-
rearing in some seabird species and threshold size of
0-group sandeels Ammodytes marinus, and hence no
evidence of matching in this system. This suggests
that phenologies of North Sea species may have been
mismatched since at least 1983, which is possible as
most of our study followed the abrupt increase in SST
and regime shift in the late 1980s (Beaugrand 2004).
Contrasting phenological responses can arise for
several reasons. For example, if organisms at differ-
ent trophic levels vary in their ability to respond to
climate warming or in the extent to which their phe-
nology is influenced by alternative drivers. The ab -
sence of significant relationships be tween any of the
phenology metrics considered and SST, apart from
suggestive relationships with Cala nus spp. stages
and Temora longicornis, suggests that differential re -
s ponses to climate warming may be unimportant, at
least in this part of the western North Sea. Alterna-
tively winter/ spring or summer SST may not be at the
appropriate temporal or spatial scale for elucidating
such responses. Indeed, there was evidence that
timing of breeding for 3 of the 4 sea birds that dis-
perse outside the North Sea in winter (kittiwakes,
guillemot and razorbills, but not puf fins) was related
to broader scale climate cues (wNAO), particularly in
the case of the kittiwake which ex hibits the widest
dispersal (Wernham et al. 2002, Bogdanova et al.
2011). High abundance and/or broad peaks of sea-
sonal prey can potentially mask phenological rela-
tionships (Durant et al. 2005), with phenological
matching likely to be particularly apparent when
prey abundance is reduced or only available for a
short duration of time. In the North Sea, dramatic
changes in abundance and spatial distributions of
phytoplankton and copepods (Beaugrand et al. 2009)
have been observed. However, patterns of change
are not consistent within trophic levels, with in -
creases apparent for some species (for example
C. helgolandicus), while others are decreasing (e.g.
C. finmarchicus) (Planque & Fromentin 1996). Thus it
is unclear to what extent changing abundance may
be masking phenological matching in our system.
Finally, there may potentially be lagged responses of
predators to prey phenology such that comparison of
phenologies in the same year does not elucidate rela-
tionships. For example, phenology of juvenile zoo-
plankton abundance is related to the reproductive
timing of parent generations (Ellertsen et al. 1987).
129
Predictor No. of models Guillemot Razorbill Shag Kittiwake Puffin
Sandeel size 8 / 40 0.001 0.125 0.083 0.135 0.125
Sandeel hatch date 8 / 40 0.025 0.090 0.099 0.071 0.171
Sandeel growth rate 8 / 40 0.966 0.364 0.319 0.176 0.097
Mismatch index 8 / 40 0.004 0.135 0.110 0.019 0.121
Sandeel biomass index lagged 20 / 40 0.169 0.237 0.560 0.872 0.679
SST lagged 20 / 40 0.177 0.116 0.477 0.985 0.979
Year (linear time trend) 20 / 40 0.956 0.924 0.507 0.585
Sandeel fishery presence 20 / 40 0.847
R2for model with lowest AICc (%) 71.80 50.85 49.9 65.38 71.30
n (yr) 2424212224
Table 3. Importance of variables associated with seabird breeding success, based upon summed Akaike weights (range from 0
to 1; high values indicate strong support), were calculated using the full candidate set of n = 40 models (see Table S4 in the
supplement at www.int-res.com/articles/suppl/m454p119_supp.pdf. Note that parameters derived from the sandeel model
(hatch date, growth rate, size and mismatch index) were present in 8 of the 40 models within the candidate set, whereas the
remaining predictors were present in 20 models —the prior weights for these variables are therefore 0.2 and 0.5, respectively,
and this difference needs to be taken into account when drawing comparisons between the 2 groups of variables. Predictors
included in the model with the lowest value of the corrected Akaike’s information criterion (AICc) for each species are shown
in bold, and we report the overall R2values for these models. SST: sea-surface temperature
Mar Ecol Prog Ser 454: 119–133, 2012
It is also possible that sampling differences be -
tween the trophic levels could potentially result in
phenology measures that are too crude to identify
correlations. The plankton data were at a lower tem-
poral resolution and broader spatial scale than the
sandeel and seabird data, and central tendency esti-
mates of plankton phenology are known to have low
sensitivity if timing shifts are small (Ji et al. 2010).
Moreover, copepods may respond to timing of critical
abundance thresholds of diatoms, rather than to sea-
sonal peaks (Runge et al. 2005). In addition, we con-
sidered mean phenological values at a broad spatial
scale, whereas changes in prey distributions can lead
to localised spatial mismatch (Schweiger et al. 2008).
Ideally future analyses of phenological trends and
mismatch in this system should account for annual
variation in prey abundance by investigating overlap
between distributions of prey availability and preda-
tor peak energetic demands rather than treating
them as point estimates.
Trophic mismatch between seabirds and sandeels
A key finding of the present study was that, for most
of the Isle of May seabird populations considered, the
timing of seabird mid-chick-rearing tracked the timing
of 0-group sandeels attaining a threshold size, rather
than sandeel hatch dates per se. The rate at which
mid-chick-rearing was delayed varied among the spe-
cies, but in all cases was insufficient to keep pace with
the delayed date at which sandeels attained a thresh-
old size. The net result was that for all 5 seabird spe-
cies the size of 0-group sand eels around the mid-point
of chick-rearing, when energetic requirements are
likely to be greatest, has significantly declined.
The seabird species we considered differed to
some extent as to the importance of 0-group sand eels
in adult and chick diets, with the likely ranking in
decreasing order of reliance being kittiwake, puffin,
razorbill, guillemot and shag (Daunt et al. 2008). This
ranking reflects the species- specific differences
apparent in mismatching, with kittiwakes, which
show the greatest reliance on the 0-group, tracking
changes in size most closely, while shags, for which
the 0-group is a minor part of the diet, showed no
trend in their breeding phenology. However, the pre-
dicted net energetic re duction in 0-group sandeel
prey was higher for kittiwakes than for the other
seabird species that were also tracking changes in
sandeel size, despite kittiwakes showing the greatest
phenological shifts in re sponse to changing sandeel
timing. This is be cause variation in net energetic
costs of reductions in sandeel size also depends on
the absolute tim ing of breeding. As the consistently
latest bree d ing seabird on the Isle of May, kittiwake
mid-chick-rearing occurred when sandeels were
predic ted to be larger and, due to the non-linear
nature of the relationship between sandeel length
and energetic content, declines in the length of large
fish are more energetically costly than equivalent
declines in smaller fish.
Although the results presented here pertain to
seabirds from the Isle of May, the order and annual
timing of breeding for the species considered were
consistent and varied in parallel with another North
Sea colony 90 km distant (Wanless et al. 2009). This
suggests consistency, at least at the local scale, with
seabirds from these 2 colonies potentially foraging on
the same sandeel stock. However, qualitative data
from other UK seabird colonies suggests that breed-
ing phenology may vary considerably between loca-
tions (Wanless et al. 2009). Moreover, sandeel popu-
lations also exhibit significant regional variation in
growth rates, length-at-age and abundance (Boulcott
et al. 2006), and further work is needed to identify
whether interactions between seabirds and sandeel
phenology vary regionally.
Very few other studies have considered the role of
prey quality in trophic mismatch. However, Beau-
grand et al. (2003) found a mismatch between the
size of larval cod Gadus morhua relative to the size of
their calanoid copepod prey, resulting in poorer cod
survival. Similarly, in a terrestrial system, caribou
Rangifer tarandus births had become mismatched
from the onset of newly emergent nutrient-rich plant
growth, resulting in reduced offspring survival (Post
& Forchhammer 2008). There are several potential
reasons why Isle of May seabirds are failing to track
changes in the timing of 0-group sandeel size. Firstly,
the cues used by the birds to time their breeding may
not accurately predict 0-group sandeel phenology.
Secondly, there may be trade-offs between the bene-
fits of delaying breeding to maintain 0-group sandeel
size and the fitness benefits of early breeding (Daunt
et al. 2007, Harris et al. 2007). Thirdly, the birds may
be constrained in their ability to alter their phenology
due to having a photoperiodically controlled physio-
logical window of breeding timing (Dawson 2008).
Finally, adult birds may compensate for the lower
energy value of 0-group sandeels by increasing for-
aging effort and/or switching to alternative prey. It is
probable that some, possibly all, of these factors are
operating in our study system.
Despite the apparently serious implications in
terms of the reduced energy value of prey associated
130
Burthe et al.: North Sea trophic mismatch
with changes in 0-group sandeel length at the time of
chick-rearing, particularly for kittiwakes, there was
no evidence that this was related to poorer breeding
success. This contrasts with studies from a Norwe-
gian seabird colony where breeding success of
puffins was positively related to the size of herring
prey (Durant et al. 2003) and from a Japanese colony
where rhinoceros auklet Cerorhinca monocerata
breeding success was poorer in years when birds
were mismatched from the availability of their prey,
Japa nese anchovy Engraulis japonicus (Watanuki et
al. 2009). The reason for this disparity is currently
unclear, but may well be linked to differences in
the life-history traits of the species involved, hydro-
biological characteristics of the study systems — with
both the mismatch examples given above being from
a conveyor-type system where prey availability is
affected by the timing of currents — and whether
other factors such as predation or severe weather
exert a major effect on breeding success.
It is also plausible that breeding success of some
Isle of May seabirds is more closely linked to the
scheduling and/or abundance of one or more alterna-
tive prey species. The main alternative prey species
for Isle of May seabirds are 1+-group sandeels, clu-
peids (predominantly sprats Sprattus sprattus) and
butterfish Pholis gunellus (Daunt et al. 2008). Time
series data on abundance or phenology for these spe-
cies are lacking; thus, it is impossible to repeat the
approach used to look for matching between seabird
breeding and 0-group sandeels. Previous findings
suggest that the timing of 1+-group sandeel abun-
dance is potentially important for kittiwakes, guille-
mots and shags (Frederiksen et al. 2004b). Our
analyses of breeding success included a range of
potentially important variables, and results indicated
that these differed among the species. Sandeel abun-
dance as indicated by a sandeel biomass index and
the sandeel growth rate, which may be linked with
behavioural changes and hence availability of sand -
eel to seabirds, emerged as being consistently im -
portant. However, we emphasise that investigating
breed ing phenology in relation to prey phenology
assumes that this is the single critical activity under
selection (Visser & Both 2005). In reality, the entire
life-cycle is under selection and responding to multi-
ple environmental drivers, such that breeding phe-
nology may be the outcome of trade-offs between
several selection pressures. Ideally, in future work,
multiple life-history traits should be investigated at
the individual level to understand whether predators
are adequately responding to changes in prey phe-
nology and the fitness costs of mismatch.
CONCLUSIONS
Quantifying the shift in the phenology of a prey
species has been suggested as an appropriate yard-
stick for interpreting whether a predator is shifting its
phenology adequately to match the change in its
environment often as a result of climate change
(Visser & Both 2005). Our study followed this ap -
proach to assess phenological changes across multi-
ple trophic levels of the pelagic food web in the west-
ern North Sea. At a broad spatial scale we found
contrasting phenological trends between trophic
levels that may indicate that the system is currently
experiencing trophic mismatching. By developing a
novel approach we also explored finer scale data on
the timing of peak energy for avian predators in rela-
tion to temporal changes in the energy value of an
important piscivorous prey. This analysis highlighted
significant changes over 24 yr in temporal matching
of the chick-rearing periods of 5 species of seabirds
and in the size of 0-group sandeels such that prey
length, and hence energy value of individual items,
have declined significantly. To date, there is no evi-
dence that these changes are impacting on the
breeding success of any of the seabird species con-
sidered, but further changes, particularly if alterna-
tive prey are also affected adversely, could well have
population level consequences for the North Sea
seabird community.
Acknowledgements. The present study was funded by the
Centre for Ecology & Hydrology Environmental Change
Integrating Fund project SPACE (Shifting Phenology:
Attributing Change across Ecosystems). SST data were pro-
vided courtesy of the NOAA National Oceanographic Data
Center and the University of Miami. We thank Mike Harris
and many others who collected seabird data, and all
past/present members and supporters of the Sir Alister
Hardy Foundation for Ocean Science for establishment and
maintenance of the CPR dataset and the owners and crews
of ships towing the CPRs. We are grateful to the Natural
Environmental Research Council and the Joint Nature Con-
servation Committee for supporting the long-term seabird
studies and to the Scottish Natural Heritage for allowing
work on the Isle of May. Five anonymous referees provided
constructive criticism which significantly improved the orig-
inal manuscript.
LITERATURE CITED
Beaugrand G (2004) The North Sea regime shift: evidence,
causes, mechanisms and consequences. Prog Oceanogr
60: 245−262
Beaugrand G, Brander KM, Souissie S, Reid PC (2003)
Plankton effect on cod recruitment in the North Sea.
Nature 426: 661−664
131
➤
➤➤
Mar Ecol Prog Ser 454: 119–133, 2012
Beaugrand G, Luczak C, Edwards M (2009) Rapid biogeo-
graphical plankton shifts in the North Atlantic Ocean.
Glob Change Biol 15: 1790−1803
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
Bogdanova MI, Daunt F, Newell M, Phillips RA, Harris MP,
Wanless S (2011) Seasonal interactions in the black-
legged kittiwake, Rissa tridactyla: links between breed-
ing performance and winter distribution. Proc Biol Sci
278: 2412−2418
Both C, van Asch M, Bijlsma RG, van den Burg AB, Visser
ME (2009) Climate change and unequal phenological
changes across four trophic levels: constraints or adapta-
tions? J Anim Ecol 78: 73−83
Boulcott P, Wright PJ, Gibb FM, Jensen H, Gibb IM (2006)
Regional variation in maturation of sandeels in the North
Sea. ICES J Mar Sci 64: 369−376
Burnham KP, Anderson DR (2002) Model selection and mul-
timodel inference: a practical information-theoretical
approach. Springer-Verlag, New York, NY
Cramp S, Simmons K (1978) Handbook of the birds of
Europe, the Middle East and North Africa: the birds of
the western Palearctic, Vol 1. Oxford University Press,
Oxford
Cramp S, Simmons K (1983) Handbook of the birds of
Europe, the Middle East and North Africa: the birds of
the western Palearctic, Vol 3. Oxford University Press,
Oxford
Cury P, Bakun A, Crawford RJM, Jarre A, Quinones RA,
Shannon LJ, Verheve HM (2000) Small pelagics in
upwelling systems: patterns of interaction and structural
changes in ‘‘wasp-waist’’ecosystems. ICES J Mar Sci 57:
603−618
Cushing DH (1990) Plankton production and year-class
strength in fish populations—an update of the match
mismatch hypothesis. Adv Mar Biol 26: 249−293
Daan N, Bromley PJ, Hislop JRG, Nielsen NA (1990) Ecol-
ogy of North Sea fish. Neth J Sea Res 26: 343−386
Daunt F, Wanless S, Harris M, Money L, Monaghan P (2007)
Older and wiser: improvements in breeding success are
linked to better foraging performance in European
shags. Funct Ecol 21: 561−567
Daunt F, Wanless S, Greenstreet SPR, Jensen H, Hamer KC,
Harris MPH (2008) The impact of the sandeel fishery
closure on seabird food consumption, distribution, and
productivity in the northwestern North Sea. Can J Fish
Aquat Sci 65: 362−381
Dawson A (2008) Control of the annual cycle in birds: endo -
crine constraints and plasticity in response to ecological
variability. Philos Trans R Soc Lond B Biol Sci 363:
1621−1633
Drent RH, Daan S (1980) The prudent parent: energetic
adjustments in avian breeding. Ardea 68: 225−252
Durant JM, Anker-Nilssen T, Stenseth NC (2003) Trophic
interactions under climate fluctuations: the Atlantic puf-
fin as an example. Proc Biol Sci 270: 1461−1466
Durant JM, Hjermann DO, Anker-Nilssen T, Beaugrand G,
Mysterud A, Pettorelli N, Stenseth NC (2005) Timing
and abundance as key mechanisms affecting trophic
interactions in variable environments. Ecol Lett 8:
952−958
Durant JM, Anker-Nilssen T, Stenseth NC (2006) Ocean cli-
mate prior to breeding affects the duration of the nestling
period in the Atlantic puffin. Biol Lett 2: 628−631
Edwards M, Richardson AJ (2004) Impact of climate change
on marine pelagic phenology and trophic mismatch.
Nature 430: 881−884
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: 1−10
Ellertsen B, Fossum P, Solemdal P, Sundby S, Tilseth S
(1987) The effect of biological and physical factors on the
survival of the Arctic-Norwegian cod and the influence
on recruitment variability. In: Loenge H (ed) The effect of
oceanographic conditions on distribution and population
dynamics of commercial fish stocks in the Barents Sea.
Institute of Marine Research, Bergen, p 101−126
Forrest J, Miller-Rushing AJ (2010) Toward a synthetic
understanding of the role of phenology in ecology and
evolution. Philos Trans R Soc Lond B Biol Sci 365:
3101−3112
Frederiksen M, Harris MP, Daunt F, Rothery P, Wanless S
(2004a) Scale-dependent climate signals drive breeding
phenology of three seabird species. Glob Change Biol
10: 1214−1221
Frederiksen M, Wanless S, Harris MP, Rothery P, Wilson LJ
(2004b) The role of industrial fisheries and oceano-
graphic change in the decline of North Sea black-legged
kittiwakes. J Appl Ecol 41: 1129−1139
Frederiksen M, Edwards M, Richardson AJ, Halliday NC,
Wanless S (2006) From plankton to top predators: bot-
tom-up control of a marine food web across four trophic
levels. J Anim Ecol 75: 1259−1268
Frederiksen M, Jensen H, Daunt F, Mavor RA, Wanless S
(2008) Differential effects of a local industrial sand lance
fishery on seabird breeding performance. Ecol Appl 18:
701−710
Frederiksen M, Elston DA, Edwards M, Mann AD, Wanless
S (2011) Mechanisms of long-term decline in size of
lesser sandeels in the North Sea explored using a growth
and phenology model. Mar Ecol Prog Ser 432: 137−147
Furness RW, Tasker ML (2000) Seabird−fishery interactions:
quantifying the sensitivity of seabirds to reductions in
sandeel abundance, and identification of key areas for
sensitive seabirds in the North Sea. Mar Ecol Prog Ser
202: 253−264
Grosbois V, Giminez 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
Hallett TB, Coulson T, Pilkington JG, Clutton-Brock TH,
Pemberton JM, Grenfell BT (2004) Why large-scale cli-
mate indices seem to predict ecological processes better
than local weather variables. Nature 430: 71−75
Harris MP, Wanless S (2011) The puffin. T & AD Poyser,
London
Harris MP, Wanless S, Murray S, Mackley E (2005) Isle of
May seabird studies in 2004. JNCC Report 375, Joint
Nature Conservancy Committee, Peterborough
Harris MP, Frederiksen M, Wanless S (2007) Within- and
between-year variation in the juvenile survival of com-
mon guillemots Uria aalge. Ibis 149: 472−481
Hipfner JM (2008) Matches and mismatches: ocean climate,
prey phenology and breeding success in a zooplanktivo-
rous seabird. Mar Ecol Prog Ser 368: 295−304
Hislop JRG, Harris MP, Smith JGM (1991) Variation in the
calorific value and total energy content of the lesser
sandeel (Ammodytes marinus) and other fish preyed on
by seabirds. J Zool 224: 501−517
132
➤
➤➤
➤
➤
➤
➤
➤
➤
➤
➤➤
➤➤
➤
➤➤
➤➤
➤
➤➤
➤
➤
➤➤
➤➤
➤
➤
➤➤
➤➤
➤
➤
➤
Burthe et al.: North Sea trophic mismatch
Hurvich CM, Tsai CL (1989) Regression and time series
model selection in small samples. Biometrika 76: 297−307
Ji R, Edwards M, Mackas DL, Runge JA, Thomas AC (2010)
Marine plankton phenology and life history in a chang-
ing climate: current research and future directions.
J Plankton Res 32: 1355−1368
Johns D (2009) Monthly averaged data for phyto- and zoo-
plankton (55−58°N, 3−0°W) 1983−2007 recorded by the
Continuous Plankton Recorder [to 13/11/2009]. Sir Alis-
ter Hardy Foundation for Ocean Science, Plymouth
Kilpatrick KA, Podesta GP, Evans R (2001) Overview of the
NOAA/NASA Advanced Very High Resolution Radio -
meter Pathfinder algorithm for sea surface temperature
and associated matchup database. J Geophys Res C 106:
9179−9197
Leggett WC, DeBlois E (1994) Recruitment in marine
fishes —Is it regulated by starvation and predation in the
egg and larval stages? Neth J Sea Res 32:119–134
Macer CT (1966) Sandeels (Ammodytidae) in the south-
western North Sea; their biology and fishery. Fish Inves-
tig Ser II Mar Fish GB Minist Agric Fish Food 24(6):1–55
Moe B, Stempniewicz L, Jakubas D, Angelier F and others
(2009) Climate change and phenological responses of
two seabird species breeding in the high-Arctic. Mar
Ecol Prog Ser 393: 235−246
Monteleone DM, Peterson WT (1986) Feeding ecology of
American sand lance Ammodytes americanus larvae
from Long Island Sound. Mar Ecol Prog Ser 30: 133−143
Muggeo VMR (2008) Segmented: an R Package to fit re -
gression models with broken-line relationships. R News,
8/1, 20−25. Available at: http: //cran.r-project.org/doc/
Rnews/ (accessed Feb 2011)
Perry AL, Low PJ, Ellis JR, Reynolds JD (2005) Climate
change and distribution shifts in marine fishes. Science
308: 1912−1915
Planque B, Fromentin JM (1996) Calanus and environment
in the eastern North Atlantic. I. Spatial and temporal pat-
terns of C. finmarchicus and C. helgolandicus. Mar Ecol
Prog Ser 134: 101−109
Post E, Forchhammer MC (2008) Climate change reduces
reproductive success of an Arctic herbivore through
trophic mismatch. Philos Trans R Soc Lond B Biol Sci 363:
2369−2375
Richardson AJ, Walne AW, John AWG, Jonas TD and others
(2006) Using continuous plankton recorder data. Prog
Oceanogr 68: 27−64
Runge JA, Franks PJS, Gentleman WC, Megrey BA and oth-
ers (2005) Diagnosis and prediction of variability in sec-
ondary production and fish recruitment processes: devel-
opments in physical-biological modeling. In: Robinson
AR, Brink K (eds) The global coastal ocean; multi-scale
interdisciplinary processes, Vol 13. Harvard University
Press, Cambridge, MA, p 413−473
Schweiger O, Settele J, Kudrna O, Klotz S, Kuhn I (2008) Cli-
mate change can cause spatial mismatch of trophically
interacting species. Ecology 89: 3472−3479
Shultz MT, Piatt JF, Harding AMA, Kettle AB, Van Pelt TI
(2009) Timing of breeding and reproductive perfor-
mance in murres and kittiwakes reflect mismatched sea-
sonal prey dynamics. Mar Ecol Prog Ser 393: 247−258
Soreide JE, Falk-Petersen S, Hegseth EN, Hop H, Carroll
ML, Hobson KA, Blachowiak-Samolyk K (2008) Seasonal
feeding strategies of Calanus in the high-Arctic Svalbard
region. Deep-Sea Res II 55: 2225−2244
Thackeray SJ, Sparks TH, Frederiksen M, Burthe S and
others (2010) Trophic level asynchrony in rates of pheno-
logical change for marine, freshwater and terrestrial
environments. Glob Change Biol 16: 3304−3313
Visser ME, Both C (2005) Shifts in phenology due to global
climate change: the need for a yardstick. Proc Biol Sci
272: 2561−2569
Wanless S, Harris MP, Redman P, Speakman JR (2005) Low
energy values of fish as a probable cause of a major
seabird breeding failure in the North Sea. Mar Ecol Prog
Ser 294: 1−8
Wanless S, Frederiksen M, Walton J, Harris MP (2009) Long-
term changes in breeding phenology at two seabird
colonies in the western North Sea. Ibis 151: 274−285
Watanuki Y, Ito M, Deguchi T, Minobe S (2009) Climate-
forced seasonal mismatch between the hatching of
rhino ceros auklets and the availability of anchovy. Mar
Ecol Prog Ser 393: 259−271
Wernham CV, Toms MP, Marchant JH, Clark J, Siriwardena
G, Baillie S (eds) (2002) The migration atlas: movements
of the birds of Britain and Ireland.T & AD Poyser,
London
133
Submitted: March 30, 2011; Accepted: November 21, 2011 Proofs received from author(s): February 29, 2012
➤
➤➤
➤
➤
➤
➤➤
➤
➤
➤
➤➤
➤
➤
➤
➤➤
➤
➤
