Trophic level asynchrony in rates of phenological changefor marine, freshwater and terrestrial environmentsS T E P H E N J . T H A C K E R AY *, T I M O T H Y H . S PA R K S w 1 , M O R T E N F R E D E R I K S E N z, § , S A R A H

B U R T H E § , P H I L I P J . B A C O N } , J A M E S R . B E L L k, M A R C S . B O T H A M **, T O M M .

B R E R E T O N w w , PA U L W. B R I G H T zz, L A U R E N C E C A R VA L H O § , T I M C L U T T O N - B R O C K § § ,

A L I S T A I R D A W S O N § , M A R T I N E D WA R D S } } , J . M A L C O L M E L L I O T T kk, R I C H A R D

H A R R I N G T O N k, D AV I D J O H N S } } , I A N D . J O N E S *, J A M E S T . J O N E S ***, D AV I D I .

L E E C H w w w , D AV I D B . R O Y **, W. A N D Y S C O T T *, M AT T S M I T H zzz, R I C H A R D J .

S M I T H E R S § § § , I A N J . W I N F I E L D * and S A R A H WA N L E S S §

*Centre for Ecology and Hydrology Lancaster, Lancaster Environment Centre, Library Avenue, Bailrigg, Lancaster, LA1 4AP, UK,

wInstitute of Zoology, Poznan University of Life Sciences, Wojska Polskiego 71 C, 60-625 Poznan, Poland, zNERI, Aarhus

University, Frederiksborgvej 399, 4000 Roskilde, Denmark, §Centre for Ecology and Hydrology Edinburgh, Bush Estate, Penicuk,

Midlothian EH26 0QB, UK, }Marine Scotland, Freshwater Laboratory, Faskally, Pitlochry, Perthshire, PH16 5LB, UK,

kRothamsted Insect Survey, Department of Plant and Invertebrate Ecology, Rothamsted Research, Harpenden, Hertfordshire AL5

2JQ, UK, **Centre for Ecology and Hydrology Wallingford, Maclean Building, Benson Lane, Crowmarsh Gifford, Wallingford,

Oxfordshire OX10 8BB, UK, wwButterfly Conservation, Manor Yard, East Lulworth, Wareham, Dorset BH20 5QP, UK, zzSchool of

Biological Sciences, Royal Holloway University of London, Egham, Surrey TW20 0EX, UK, §§Department of Zoology, University of

Cambridge, Downing Street, Cambridge CB2 3EJ, UK, }}The Sir Alister Hardy Foundation for Ocean Science, The Laboratory,

Citadel Hill, Plymouth PL1 2PB, UK, kkThe Freshwater Biological Association, The Ferry Landing, Far Sawrey, Ambleside,

Cumbria LA22 0LP, UK, ***People’s Trust for Endangered Species, 15 Cloisters House, 8 Battersea Park Road, London SW8 4BG,

UK, wwwBritish Trust for Ornithology, The Nunnery, Thetford IP24 2PU, UK, zzzNational Pollen and Aerobiology Research Unit,

University of Worcester, Henwick Road, Worcester WR2 6AJ, UK, §§§The Woodland Trust, Autumn Park, Dysart Road, Grantham,

Lincolnshire NG31 6LL, UK

Abstract

Recent changes in the seasonal timing (phenology) of familiar biological events have been one of the most

conspicuous signs of climate change. However, the lack of a standardized approach to analysing change has

hampered assessment of consistency in such changes among different taxa and trophic levels and across freshwater,

terrestrial and marine environments. We present a standardized assessment of 25 532 rates of phenological change for

726 UK terrestrial, freshwater and marine taxa. The majority of spring and summer events have advanced, and more

rapidly than previously documented. Such consistency is indicative of shared large scale drivers. Furthermore,

average rates of change have accelerated in a way that is consistent with observed warming trends. Less coherent

patterns in some groups of organisms point to the agency of more local scale processes and multiple drivers. For the

first time we show a broad scale signal of differential phenological change among trophic levels; across environments

advances in timing were slowest for secondary consumers, thus heightening the potential risk of temporal mismatch

in key trophic interactions. If current patterns and rates of phenological change are indicative of future trends, future

climate warming may exacerbate trophic mismatching, further disrupting the functioning, persistence and resilience

of many ecosystems and having a major impact on ecosystem services.

Keywords: climate, linear mixed effects models, meta-analysis, phenology, traits, trophic mismatch

Received 9 October 2009 and accepted 9 December 2009

Introduction

Recent climate warming has dramatically altered the

distribution, abundance and population dynamics of

many aquatic and terrestrial organisms (IPCC, 2007).

Acting in concert with other ecological stressors, cli-

matic change is projected to weaken the resilience of

many ecosystems, leading to sudden re-organizations

of communities and drastic alterations to ecosystem

structure and function (Scheffer & Carpenter, 2003).

Recorded changes in the seasonal timing, or phenology,

of life history events have formed a key part of this

assessment, with spring and summer events generally

Correspondence: Stephen J. Thackeray, tel. 1 44 01524 595852, fax

1 44 01524 61536, e-mail: sjtr@ceh.ac.uk

1Formerly: Centre for Ecology and Hydrology Monks Wood,

Abbots Ripton, Huntingdon, Cambridgeshire PE28 2LS, UK.

Global Change Biology (2010) 16, 3304–3313, doi: 10.1111/j.1365-2486.2010.02165.x

3304 r 2010 Blackwell Publishing Ltd

becoming earlier as temperatures have risen. Species-

specific variation in phenological responses to climate

can disrupt the synchrony of ecological interactions

(Harrington et al., 1999; Visser & Both, 2005) and poten-

tially affect community persistence. To assess the poten-

tial risk of phenological asynchrony affecting community

structure, it is crucial to determine how rates of pheno-

logical change vary among species, functional groups

and trophic levels.

Assessments of broad-scale taxonomic variation in

phenological change have generally involved meta-

analyses in which previously published results on

changes in the timing of biological events are extracted

from the literature and combined to provide overall

rates of change. While there has been broad unanimity

that spring events in the northern hemisphere have

become earlier, with estimated mean advances ranging

from 2.3 days per decade (Parmesan & Yohe, 2003) to

5.5 days per decade (Root et al., 2003), there has been a

lack of consensus on which taxonomic groups show the

most, and least, change. Differences in the data selection

criteria adopted by different meta-analyses have been

cited as a contributory factor to this lack of consistency

(Englund et al., 1999; Parmesan, 2007). Reliance on

published studies creates two further potential pro-

blems. Firstly, positive publication bias (Gurevitch

et al., 2001) leads to the under-representation of taxa

showing little or no phenological change. Researchers

have sought to overcome this by restricting analyses to

multispecies studies that include nonresponding spe-

cies (Parmesan & Yohe, 2003), though there is no

assurance that such assemblages are representative of

the wider community. Secondly, despite the fact that

rates (and even directions) of phenological change are

known to vary over time (Crick & Sparks, 1999; Thack-

eray et al., 2008), the lengths and periods of phenologi-

cal time series used in meta-analyses are rarely

standardized (Badeck et al., 2004; Parmesan, 2007).

Thus, apparent taxonomic differences in rates of change

can be generated by differences in time series start year,

end year and length, rather than by biologically relevant

characteristics.

To date analyses overcoming such problems by using

raw time series data and consistent time scales rather

than published values have been mainly restricted to

species comparisons within major taxonomic or func-

tional groups, such as plankton (Edwards & Richard-

son, 2004), terrestrial plants (Fitter & Fitter, 2002;

Menzel et al., 2006), insects (Roy & Sparks, 2000; Stefa-

nescu et al., 2003) and birds (Cotton, 2003; Both &

Artemyev, 2004). Here, we apply this approach across

environments to obtain the first directly comparable

estimates of phenological change for UK terrestrial,

freshwater and marine taxa between 1976 and 2005.

This has been a period of major environmental change

with factors such as eutrophication (including atmo-

spheric deposition of nutrients), acid rain, land-use

change, overfishing and, of course, climate change

impacting on natural systems (Smart et al., 2003; Mil-

lenium Ecosystem Assessment, 2005).

A number of functional traits might be expected to

influence the rate of phenological change demonstrated

by organisms. It might be expected that the mode of

thermal physiology would be particularly influential.

Developmental rates of ectotherms are, by definition,

directly affected by habitat warming (Precht et al., 1973).

One might hypothesize therefore that they have shown

the most rapid mean rates of phenological advance. It

has been argued that generation time should affect

evolutionary rates in response to selective pressures

imposed by environmental change (Rosenheim &

Tabashnik, 1991; Berteaux et al., 2004) and so we might

expect this trait to affect rates of phenological change,

assuming a role for microevolution. At the scale of

single ecosystems, differential phenological change

among trophic levels has frequently been observed

(Harrington et al., 1999; Visser & Both, 2005) though it

is not clear whether mean rates of change differ sig-

nificantly among trophic levels at larger scales of eco-

logical organization (Blenckner & Hillebrand, 2002), or

whether these patterns are idiosyncratic. Also, choice of

phenological metric (Miller-Rushing et al., 2008; Van

Buskirk et al., 2009) may affect observed rates of change.

The aims of this study were to (i) provide a fully

standardized and unbiased assessment of rates of phe-

nological change in the United Kingdom and (ii) to

resolve differences in rates of change among environ-

ments and major taxonomic and functional groups. We

explicitly considered the following functional attri-

butes; thermal physiology, generation time, environ-

ment (where each species feeds/gathers resources

during the focal phenological event) and trophic level.

The latter attribute allowed us to compute average rates

for each trophic level and thus evaluate, across the three

environments, likely risks of trophic mismatching

i.e. reduced synchrony between the seasonal occurrence

of food resources and the peak demand for those

resources.

Materials and methods

Quantifying rates of change

All rates of change were calculated from the raw observational

data for each nonoverlapping 10-year period between 1976

and 2005, and for the whole 30-year period.

We selected only records indicating the beginning or middle

of phenological events, occurring on average between January

P H E N O L O G I C A L C H A N G E A C R O S S M A J O R E N V I R O N M E N T S 3305

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

and August. Thus, our emphasis was on spring and summer

events rather than those occurring in autumn or winter.

All phenological indicators were the standard metrics

used by each of the monitoring programmes represented in

the analysis (supporting information, Table S1). Linear regres-

sion was used to quantify phenological trends over each

specified time period. Annual seasonal timing (day of year)

of phenological events was analysed as a function of year

by linear regression. Slope coefficients quantified rates of

change in days per year. Regressions were performed for

each nonoverlapping 10-year period between 1976 and 2005

and for the whole 30-year period. Trends were only retained

for meta-analysis if phenological data were available for 80%

or more of the years within the respective time period. Con-

sidering all taxa, sites and time periods this yielded 25 532

rates of phenological change for 726 distinct taxa (supporting

information, Table S2) including terrestrial plants, freshwater

and marine phytoplankton and zooplankton, insects, amphi-

bians, birds, fish and mammals (supporting information, Table

S2). Phenological records included dates of flowering and

leafing, plankton population growth, insect flight periods,

births and migration. The analysis included data from across

the whole United Kingdom, spanning the latitudinal range

49.9–60.91N.

Categorization of phenological trends

For the meta-analysis, phenological trends were categorized

by major taxonomic group (plants/phytoplankton, inverte-

brates, vertebrates) and by functional attributes: environment

(marine, freshwater, terrestrial), trophic level (primary produ-

cer, primary consumer, secondary consumer), thermal physiol-

ogy (ectothermic, endothermic) and generation time (o1

week, 1 week–1 month, 1 month–1 year, 41 year). Previous

studies have shown a tendency for first events to shift more

rapidly than middle or peak events (Miller-Rushing et al., 2008;

Van Buskirk et al., 2009), and so we also included this dichot-

omy as a factor in our analyses. Species that used more than

one major environment (e.g. amphibians, dragonflies, sea-

birds) were assigned to the environment with which they

were most closely associated for feeding, during the focal

phenological event.

Differences in time series lengths among monitoringprogrammes

Some of the monitoring programmes represented could only

provide phenological data for a single decade from the 30-year

period (supporting information, Table S1). It is possible that

among-decade differences in phenological change were a

statistical artefact of having a different assemblage of monitor-

ing programmes represented in each decade. We therefore

calculated mean rates of change separately for more finely

resolved taxonomic groups that were each represented by a

single monitoring programme in all three decades. Patterns of

phenological change present in these more consistently re-

corded taxa were used to confirm patterns of change observed

in the whole dataset.

Summarizing rates of change

For each major taxonomic group and habitat type, both the

mean rate of change and variability in rates of change were

calculated. To test whether trends have on balance shifted

either towards earliness or lateness, two-tailed binomial tests

were used to test the null hypothesis that trends were symme-

trically distributed about zero. Differences in trend variability

among groups were assessed using Levene’s tests, after An-

derson–Darling tests showed that data were not normally

distributed.

Linear mixed effects modelling

Linear mixed effects models were used to test for significant

differences in mean rates of change among taxonomic groups,

and with organism functional traits. This was carried out in

SAS 9.1 using PROC MIXED. The following random effects were

specified: data source, species nested in data source, site

nested in data source, and brood/generation nested in data

source. The collated datasets, each of which focussed on a

particular taxonomic or functional group, varied in size. By

including data source as a random effect, each dataset was

given equal weight in the analysis so that results were not

unduly influenced by any one, well represented, group. In

addition, this explicitly allowed nonindependence among

trends from closely related organisms, recorded using the

same sampling protocols. Data points (i.e. regression slopes)

were weighted by inverse variance. Tested fixed effects were

major taxonomic group, metric type (beginning or middle

event) and organism functional attributes. Denominator

degrees of freedom for fixed effects were estimated using

Satterthwaite’s approximation. All significance levels given

are from type III tests. Figures show marginal least-squares

means � 1 SE.

Results

Widespread phenological change across environments

Rapid climate change occurred over the study period:

all three environments experienced very similar warm-

ing trends (0.04–0.05 1C yr�1), with temperatures fre-

quently above the 1976–2005 mean since the late 1980s

(Fig. 1). During this period, the seasonal timing of

biological events in all major taxonomic groups in UK

terrestrial, freshwater and marine environments ad-

vanced, on average, by 0.39 d yr�1 (equivalent to 11.7

days over the whole period). Overall, 83.8% of pheno-

logical trends were towards earlier seasonal timing. In

all groups the proportion of negative trends (i.e. to-

wards earliness) was significantly greater than expected

if trends were symmetrical about zero (Fig. 2a) and the

percentage of statistically significant phenological

trends exceeded that expected by chance alone (5%) at

the P 5 0.05 significance level.

3306 S . J . T H A C K E R AY et al.

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

A taxonomically resolved analysis showed that mean

rates of change were significantly different among

taxonomic groups, though not among environments

(linear mixed effects model, LME, n 5 3419, interaction:

F4,71.8 5 1.47, P 5 0.22, environment: F2,152 5 0.34, P 5 0.71,

taxonomic group: F2,73.4 5 3.58, P 5 0.033). Leafing,

flowering and fruiting dates of terrestrial plants showed

the most rapid mean rate of change (0.58 d yr�1, Fig. 2b)

and the highest percentage of advancing trends (92.5%,

Fig. 2a). Freshwater plants (phytoplankton bloom tim-

ings) had the slowest rate of response (0.23 d yr�1 ear-

lier, Fig. 2b) and showed the lowest percentage of

advancing trends (62.2%, Fig. 2a). There was also

significant heterogeneity of variance in phenological

trends among taxa (Levene’s test, W8,3411 5 39.73,

Po0.001) with marine and freshwater plants and in-

vertebrates exhibiting much higher variability in rates

of change (SD range 0.74–0.98) than other groups (SD

range 0.30–0.53).

Decadal variation in rates of change

Examination of phenological changes at a finer tempor-

al resolution indicated that at the decadal scale, trends

varied markedly among taxonomic groups and envir-

onments (Fig. 3). For many taxa, rates of phenological

change have accelerated during recent decades. In gen-

eral, timing advanced least rapidly before 1986 except

for marine plants (phytoplankton bloom timings) and

marine vertebrates (migration and spawning/egg lay-

ing in marine fish and marine birds). In some cases

there was a reversal of phenological trends between the

1976–1985 and 1986–1995 periods i.e. a period of phe-

nological delay was followed by a period of phenolo-

gical advance (marine and freshwater invertebrates,

freshwater plants). For some groups mean rates of

advance were consistently higher during 1986–2005

than during 1976–1985 (e.g. freshwater and terrestrial

plants, marine and freshwater invertebrates), for others

rates of advance declined considerably post 1996 (e.g.

terrestrial vertebrates). These results were robust when

only analysing taxa that were consistently recorded

throughout the whole 30-year period (Table 1). Results

Fig. 1 January–August mean temperatures from 1976 to 2005

for the United Kingdom. Presented are UK mean air tempera-

tures (blue line), mean sea surface temperature 101W–51E, 501N–

601N (red line) and mean lake surface water temperatures

(brown line). All series presented as anomalies from the 1976

to 2005 mean. Air temperatures provided by the UK Met Office

and sea surface anomalies by the UK Met Office Hadley Centre.

Lake temperatures are means of those recorded at the sites

included in this study: the northern and southern basins of

Windermere, Esthwaite Water and Loch Leven.

Fig. 2 Phenological change for the UK flora and fauna from

1976 to 2005. (a) percentages of advancing (below horizontal)

and delaying (above horizontal) trends for each taxon–environ-

ment combination. Statistically significant advancing and delay-

ing trends are indicated by black shading. Nonsignificant trends

are indicated by white shading. The number of trends analysed

for each taxon-environment combination (n) is given above each

bar. Also shown is the significance level (P) of a two-tailed

binomial test of the null hypothesis that negative and positive

trends are equally likely. (b) Mean � SEM rates of change for

plants/phytoplankton (plant; green bars), invertebrates (invert;

yellow bars) and vertebrates (vert; blue bars) in marine, fresh-

water and terrestrial environments. All mean trends are nega-

tive, indicating an advance of phenological events. The taxa

included in each taxonomic group-environment combination,

and the number of trends per taxon, are given in supporting

information, Table S3.

P H E N O L O G I C A L C H A N G E A C R O S S M A J O R E N V I R O N M E N T S 3307

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

based on these taxa confirmed the acceleration of phe-

nological change observed in the full dataset: rates of

advancing phenology were again lowest during 1976–

1985, with higher rates of advance occurring during

1986–1995 (e.g. butterfly first flight and bird arrival

dates) or 1986–2005 (e.g. freshwater plankton and ter-

restrial plants). Flowering dates and aphid flight dates

advanced particularly rapidly, by more than 1.0 d yr�1

(10.0 days per decade) after 1986, as did moth flight

dates during 1986–1995. The major exception to this

overall trend was marine vertebrates, which showed

little phenological change after 1986.

Trophic and functional variation in rates of change

Phenological trends varied significantly among trophic

levels over the whole 30-year period (LME, n 5 3419,

F2,63.7 5 4.16, P 5 0.020). In particular, events associated

with secondary consumers advanced less rapidly than

those for both primary producers and primary consu-

mers (Fig. 4a). The nonsignificant interaction between

trophic level and environment (LME, n 5 3419,

F4,139 5 0.99, P 5 0.42) would suggest that this has been

a general phenomenon in marine, freshwater and ter-

restrial ecosystems. Analysis of decadal phenological

trends confirmed the difference in rates of change

among trophic levels (LME, n 5 22113, F2,116 5 7.46,

P 5 0.0009). There was a significant acceleration in over-

all rates of advance post 1986 (LME, n 5 22113,

F2,16*103 5 37.7, Po0.0001) but this was less pronounced

Fig. 3 Decadal variations in rates of phenological change. Dec-

adal mean � SEM rates of change for plants/phytoplankton

(plant; green bars), invertebrates (invert; yellow bars) and verte-

brates (vert; blue bars) in marine, freshwater and terrestrial

environments. For each taxon–environment combination, dec-

ades appear chronologically: 1976–1985 (1), 1986–1995 (2) and

1996–2005 (3). Negative trends indicate advance, and positive

trends delaying, of phenological events. The number of trends

analysed for each taxon–environment combination, in each

decade, is given above each bar.T

ab

le1

Mea

n�

SE

Mtr

end

s(s

amp

lesi

ze)

for

var

iou

sm

on

ito

rin

gsc

hem

esan

dp

hen

olo

gic

alm

etri

cs.

Tre

nd

sar

esu

mm

ariz

edfo

rea

cho

fth

e10

-yea

rp

erio

ds

1976

–198

5,19

86–

1995

and

1996

–200

5,an

do

ver

all

for

the

30-y

ear

per

iod

1976

–200

5

Rec

ord

ing

sch

eme

Ph

eno

log

ical

met

ric

1976

–198

5(d

yr�

1)

1986

–199

5(d

yr�

1)

1996

–200

5(d

yr�

1)

1976

–200

5(d

yr�

1)

RIS

:A

ph

ids

Fir

stfl

igh

t0.

10�

0.18

(114

)�

2.35�

0.18

(114

)�

1.64�

0.19

(114

)�

0.87�

0.04

(114

)

5th

per

cen

tile

cum

ula

tiv

ean

nu

alca

tch

0.53�

0.14

(114

)�

1.51�

0.14

(114

)�

1.28�

0.12

(114

)�

0.59�

0.05

(114

)

UK

BM

S:

Bu

tter

flie

sF

irst

flig

ht

�0.

58�

0.08

(104

6)�

0.98�

0.05

(227

9)�

0.58�

0.03

(541

0)�

0.42�

0.02

(115

6)

Mea

nfl

igh

td

ate

�0.

11�

0.05

(104

4)�

0.85�

0.03

(227

2)�

0.51�

0.02

(539

1)�

0.35�

0.01

(115

4)

BT

O:

Bir

ds

Mea

nfi

rst

egg

dat

e0.

05�

0.10

(51)

�0.

47�

0.09

(63)

�0.

27�

0.11

(55)

�0.

19�

0.03

(56)

Co

un

tyb

ird

rep

ort

sF

irst

arri

val

�0.

26�

0.07

(238

)�

0.64�

0.07

(254

)�

0.10�

0.06

(250

)�

0.31�

0.02

(239

)

CE

H:

Fre

shw

ater

pla

nk

ton

50%

max

.m

on

thly

mea

nab

un

dan

ce0.

42�

0.40

(114

)�

0.13�

0.30

(150

)�

0.60�

0.38

(130

)�

0.23�

0.09

(126

)

Blo

om

cen

tre

of

gra

vit

y0.

75�

0.34

(94)

�0.

38�

0.27

(129

)�

0.48�

0.32

(120

)�

0.22�

0.09

(108

)

RIS

:M

oth

sD

ayo

fm

edia

nca

tch

0.14�

0.09

(39)

�1.

01�

0.13

(39)

�0.

53�

0.09

(38)

�0.

33�

0.04

(39)

SA

HF

OS

:M

arin

ep

lan

kto

nB

loo

mce

ntr

eo

fg

rav

ity

�0.

20�

0.31

(101

)�

0.54�

0.26

(152

)�

0.51�

0.26

(164

)�

0.38�

0.07

(136

)

UK

PN

:P

lan

tsF

irst

flo

wer

ing

�0.

02�

0.07

(270

)�

1.31�

0.09

(300

)�

1.75�

0.22

(105

)�

0.57�

0.07

(50)

UK

BM

S:

UK

Bu

tter

fly

Mo

nit

ori

ng

Sch

eme,

BT

O:

Bri

tish

Tru

stfo

rO

rnit

ho

log

y,C

EH

:C

entr

efo

rE

colo

gy

and

Hy

dro

log

y,R

IS:

Ro

tham

sted

Inse

ctS

urv

ey,

SA

HF

OS

:S

irA

list

er

Har

dy

Fo

un

dat

ion

for

Oce

anS

cien

ce,

UK

PN

:U

KP

hen

olo

gy

Net

wo

rk

3308 S . J . T H A C K E R AY et al.

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

for secondary consumers than for the remaining trophic

levels (LME, trophic level–decade interaction, n 5 22113,

F4,4392 5 12.9, Po0.0001, Fig. 4b). Both consumer cate-

gories showed similar rates of advance post 1986.

Therefore, the relatively weak acceleration of phenolo-

gical change for secondary consumers would seem

to account for their lower mean rates of change over

the whole 30-year period.

Over the 30-year period we found no significant effect

of environment (LME, n 5 3419, F2,152 5 0.48, P 5 0.62)

and the decadal acceleration of phenological advance

was similar across environments (Fig. 4c). There was no

significant effect of generation time on rates of changing

phenology over the 30-year period (LME, n 5 3419,

F3,33.2 5 1.92, P 5 0.15) but variability in rates of change

was significantly higher for organisms with a genera-

tion time of o1 week (Levene’s test, W3,3416 5 84.09,

Po0.001). The effect of generation time interacted with

decade (LME, n 5 22113, F6,7323 5 16.4, Po0.0001), with

the increase in rates of change post 1986 not apparent in

organisms with a generation time shorter than 1 week

(Fig. 4d).

While mean rates of change indicated a more pro-

nounced advance in ectotherms than endotherms over

the 30-year period (0.40 and 0.29 d yr�1 respectively),

variation in rates of change was considerable (SD 0.59

for ectotherms, 0.31 for endotherms). There was no

significant effect of thermal physiology on rates of

change over the whole 30-year period (LME, n 5 3419,

F1,56.4 5 0.75, P 5 0.39) or in the decadal analysis (LME,

n 5 22113, F1,121 5 0.19, P 5 0.67). Rates of change were

more rapid for phenological first events than for middle

and peak events in both the 30-year analysis (LME,

n 5 3419, F1,2912 5 28.0, Po0.0001) and the decadal ana-

lysis (LME, n 5 22113, F1,19*103 5 41.6, Po0.0001).

Discussion

Phenological changes and developing asynchronies be-

tween trophic pairings (e.g. predators and their prey)

Fig. 4 Mean trends in phenology over the 30-year period 1976–2005, based on linear mixed effects models. (a) Secondary consumers

have advanced less than primary producers and primary consumers over the whole period. Patterns of change have varied among

decades for major functional groups. (b) Trophic levels. (c) Major ecosystems. (d) Generation time. Values shown are marginal least-

squares means � 1 SE.

P H E N O L O G I C A L C H A N G E A C R O S S M A J O R E N V I R O N M E N T S 3309

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

have been linked to reductions in individual fitness and

declines in the population size of focal species (Platt

et al., 2003; Winder & Schindler, 2004; Both et al., 2006;

M�ller et al., 2008), increasing the risk of population

extinctions and biodiversity loss. Such changes are also

of great economic and societal importance due to the

influence of phenological synchrony on processes such

as pollination (Elzinga et al., 2007), fisheries production

(Cushing, 1990) and herbivory by agricultural pests

(Harrington et al., 2007). Our analysis of the most

comprehensive UK phenological data collection to date

has shown that the timing of spring and summer events

has become earlier for the majority of taxa, and at a

more rapid rate than previously reported. The phenom-

enon is widespread across the terrestrial, freshwater

and marine environments and rates of change differ

significantly among trophic levels. Our results therefore

suggest a heightened risk that phenological asynchrony

may disrupt the stability and functioning of aquatic and

terrestrial ecosystems, and the delivery of key ecosys-

tem services.

Previous analyses, restricted to taxa showing signifi-

cant trends in the northern temperate zone, suggest

mean rates of advance of 4.4–5.5 days per decade (Root

et al., 2003; Root et al., 2005). A recent synthesis (Parme-

san, 2007) calculated an ‘unbiased’ estimated mean

phenological advance of 2.8 days per decade by analys-

ing published rates of change from taxa showing both

significant and nonsignificant phenological trends. Our

unbiased, standardized analysis reports a substantially

higher rate of advance. This comes contrary to expecta-

tion: our avoidance of potentially biased published

results, and inclusion of trends from ‘nonresponding’

taxa, should result in a lower mean rate of change. The

discrepancy probably arises due to the standardization

of time series lengths, and the focus on a period with a

consistent warming trend. In previous meta-analyses

some taxa were represented over longer time scales,

including periods of stable and decreasing mean tem-

peratures. Since warming trends are projected to inten-

sify (IPCC, 2007), we believe our results to be indicative

of a period of continuing, possibly increasing, phenolo-

gical change.

For the first time we have explicitly demonstrated

that there has been systematic variation in rates of

phenological change among trophic levels, and that this

has been apparent across marine, freshwater and ter-

restrial environments in the United Kingdom. Specifi-

cally, phenological advances have been slowest for

secondary consumers over the 30-year period. This

appears to result from a slower acceleration of pheno-

logical advance for this group, compared with the lower

trophic levels. Numerous potential trophic interactions

exist between the primary and secondary consumers

represented in our analysis e.g. amphibians and fish

consume both plankton and aquatic invertebrate larvae

such as dragonflies and alderflies, adult dragonflies can

consume members of the Lepidoptera and Hymeno-

ptera, many of the terrestrial bird species provision

their chicks on Lepidoptera, Hemiptera and Coleoptera

while marine fish and birds utilize larval fish such as

sandeels Ammodytes marinus. Our analysis would there-

fore suggest an increasing likelihood of trophic mis-

matching. Although developing trophic asynchronies

have been observed within single communities (Har-

rington et al., 1999; Visser & Both, 2005), generalizations

regarding rates of change among trophic levels have

proven elusive. Patterns of change within food webs

have appeared idiosyncratic, with different trophic

levels showing the most rapid rates of change in differ-

ent communities and ecosystems (Visser & Both, 2005).

This has prohibited the detection of systematic variation

in the effects of climate on phenological change at

different trophic levels (Blenckner & Hillebrand, 2002).

Our standardized, unbiased analysis has made

such detection possible. We also show for the first time

that accelerating rates of change, previously identified

for single species or small groups of species (Crick &

Sparks, 1999; Thackeray et al., 2008), have been cohe-

rent across ecosystems and have potentially widened

existing trophic asynchronies. The timing of accelera-

tion corresponds with a transition between periods

when temperatures were frequently below, and fre-

quently above, the 1976–2005 mean. This suggests

that this increase in rates of phenological advance has

been a response to climate warming over the 30-year

period.

There is much variability in the direction and magni-

tude of phenological trends. This occurs because taxa/

populations (i) differ in the extent to which life history

events are able to accelerate with warming, (ii) experi-

ence different warming trends due to variations in

mean seasonal timing and microhabitat use, (iii)

vary in the extent to which their phenological responses

are driven/constrained by factors other than increases

in temperature and (iv) may respond to changing

climate in other ways, e.g. distributional changes.

Alternative drivers of phenological change will vary

regionally and include atmospheric nutrient deposition

and precipitation (Penuelas et al., 2002; Cleland et al.,

2006), nutrient enrichment of aquatic ecosystems

(Thackeray et al., 2008) and variations in popula-

tion size (Langvatn et al., 2004; Miller-Rushing et al.,

2008) and age structure (Langvatn et al., 2004; Gillet &

Dubois, 2007). Resolution of the mechanisms under-

pinning variations in rates of change is crucial to

projecting impacts of these changes. We adopted a

functional approach to this problem, though the sources

3310 S . J . T H A C K E R AY et al.

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

of variation outlined above and correlations among

functional traits render mechanistic understanding a

major challenge. We examined expectations that rates

of change would be highest for ectotherms and short

generation time organisms; attributes which should

enhance the potential for phenological advance with

warming.

Ectotherms had a higher mean rate of change than

endotherms, as might be expected. The phenology of

ectotherm life history events will be directly influenced

by warming while endotherms will respond more in-

directly, via ecosystem effects. The difference will be

exacerbated because birds, the most well-represented

endotherms in the analysis, demonstrate constrained

phenological plasticity due to photoperiodic induction

of gonad maturation and migration, especially for long-

distance migrants (Dawson, 2008). Despite the apparent

difference in rates of change between ectotherms and

endotherms, there was considerable variability around

the mean rates of change for each thermal physiology

category and the difference was not significant. Varia-

bility was highest for phytoplankton, which also

showed the highest incidences of phenological delays.

Despite having the shortest generation times of the

analysed taxa, the phytoplankton showed compara-

tively low mean rates of change over the 30-year period.

While individual phytoplankton taxa can show very

rapid phenological shifts, great trend variability ap-

pears to have prevented mean rates of change from

being exceptional. Additionally, although phytoplank-

ton growth is directly affected by temperature, bloom

timings reflect population dynamics that are also influ-

enced by light and nutrient availability, grazing and

sedimentation (Thackeray et al., 2008) and interspecific

interactions that have potentially unpredictable out-

comes (Beninca et al., 2008).

In contrast to previous meta-analyses (Parmesan,

2007), we found that terrestrial plants have demon-

strated the fastest mean phenological advance and the

highest proportion of phenological advances. This per-

haps indicates the primacy of temperature as a limiting

factor for plant growth at mid northern latitudes (Ba-

deck et al., 2004) and that temperature change has not

been great enough to induce delaying effects such as

heat stress in summer (Sherry et al., 2007) or insufficient

winter chilling (Thompson & Clark, 2008). However,

warming is also believed to be the predominant driver

of change for other phenological metrics, such as butter-

fly flight periods (Roy & Sparks, 2000). The rapid and

consistent phenological advances for terrestrial plants

cannot therefore be explained by the fact that they

respond the least to alternative drivers. Plant develop-

ment may simply accelerate more rapidly with warm-

ing than developmental rates for other taxa. It is also

possible that, since individual plants are sessile, they

cannot modulate the warming they experience by mov-

ing among microhabitats. This might result in stronger

and more consistent phenological trends for this

group.

It is possible that effects of thermal physiology and

generation time were masked by correlations among

functional traits. Trophic level and thermal physiology

are, to an extent, correlated. Endotherms were repre-

sented only in the consumer categories, such that the

fastest rates of phenological change and acceleration

occurred in the only trophic level occupied solely by

ectotherms. It is plausible that consumers showed lower

rates of advance and acceleration because of the influ-

ence of endotherms, but that the independent effect of

thermal physiology could not be resolved in the analy-

sis. Generation time covaries strongly with trophic level

in aquatic ecosystems, generally increasing higher up

the food web, but it would seem that this trait had a

limited influence on rates of change. The taxonomic

group demonstrating the most rapid change, terrestrial

plants, did not have the shortest generation times.

Furthermore, evidence for rapid microevolution in re-

sponse to warming is equivocal, with phenotypic plas-

ticity playing a dominant role in phenological

adjustment (Berteaux et al., 2004; Charmantier et al.,

2008). Interestingly, the expected decrease in rates of

change with increasing generation time appears to be

maintained post 1986, during the period of most rapid

overall phenological change, and for organisms with

generation times in excess of 1 week. Given the correla-

tions among functional traits, a major challenge for

phenology research is to determine the traits responsi-

ble for rates of change and to establish the ecological

scale at which independent effects of these traits can be

disentangled.

Climate change has been identified as one of the most

critical threats to the maintenance of global biodiversity.

Our findings indicating earlier occurrence of many

spring and summer biological events across a wide

range of terrestrial, freshwater and marine taxa should

aid current and future assessment of the impacts of

climate change. Although the precise consequences of

phenological changes on ecosystem functioning are not

clear, our study highlights the potential risk of desyn-

chronizing trophic linkages between primary and sec-

ondary consumers. Such disruption could have

catastrophic environmental consequences and affect

delivery of important ecosystem services, particularly

food production. Given the broad taxonomic and en-

vironmental scales at which our analyses were con-

ducted, it is reasonable to expect that our findings

may generalize to other mid latitude regions and have

relevance at a global level.

P H E N O L O G I C A L C H A N G E A C R O S S M A J O R E N V I R O N M E N T S 3311

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

Acknowledgements

We thank the following for collecting and allowing access todata: Miss Ursula Allitt, Belfast HSC Trust, Cardiff School ofHealth Sciences UWIC, Prof. Jean Emberlin, Environmental andPublic Protection Offices Islington, the Freshwater BiologicalAssociation, Lancashire Immunology Service (Royal PrestonHospital), the National Dormouse Monitoring Programme, Uni-versity of Plymouth School of Geography. We thank the follow-ing organizations for funding: BBSRC, Countryside Council forWales, Defra, Forestry Commission, Joint Nature ConservationCommittee (JNCC), The Lawes Agricultural Trust, Natural Eng-land, NERC, Northern Ireland Environment Agency, ScottishNatural Heritage. The Nest Record Scheme is supported by theJNCC/BTO partnership. We thank Suzanne Clark and PeterRothery for assistance with coding statistical models, and BjornBeckmann for assistance with data extraction. We are indebted tothe many thousands of volunteer recorders who have contrib-uted records to many of the datasets we have analysed. Theauthors have no conflicts of interest to declare. We thank Prof.Mark Bailey and Dr Alex Elliott for their comments on an earlierversion of the manuscript. This analysis was funded by theCentre for Ecology and Hydrology Environmental Change In-tegrating Fund project SPACE (Shifting Phenology: AttributingChange across Ecosystems).

References

Badeck FW, Bondeau A, Bottcher K, Doktor D, Lucht W, Schaber J, Sitch S (2004)

Responses of spring phenology to climate change. New Phytologist, 162, 295–309.

Beninca E, Huisman J, Heerkloss R et al. (2008) Chaos in a long-term experiment with a

plankton community. Nature, 451, 822–825.

Berteaux D, Reale D, Mcadam AG, Boutin S (2004) Keeping pace with fast climate

change: can arctic life count on evolution? Integrative and Comparitive Biology, 44,

140–151.

Blenckner T, Hillebrand H (2002) North Atlantic Oscillation signatures in aquatic and

terrestrial ecosystems - a meta-analysis. Global Change Biology, 8, 203–212.

Both C, Artemyev AV (2004) Large-scale geographical variation confirms that climate

change causes birds to lay earlier. Proceedings of the Royal Society of London Series B-

Biological Sciences, 271, 1657–1662.

Both C, Bouwhuis S, Lessells CM, Visser ME (2006) Climate change and population

declines in a long-distance migratory bird. Nature, 441, 81–83.

Charmantier A, McCleery RH, Cole LR, Perrins C, Kruuk LEB, Sheldon BC (2008)

Adaptive phenotypic plasticity in response to climate change in a wild bird

population. Science, 320, 800–803.

Cleland EE, Chiariello NR, Loarie SR, Mooney HA, Field CB (2006) Diverse responses

of phenology to global changes in a grassland ecosystem. Proceedings of the National

Academy of Sciences of the United States of America, 103, 13740–13744.

Cotton PA (2003) Avian migration phenology and global climate change. Proceedings of

the National Academy of Sciences of the United States of America, 100, 12219–12222.

Crick HQP, Sparks TH (1999) Climate change related to egg-laying trends. Nature, 399,

423–424.

Cushing DH (1990) Plankton production and year class strength in fish populations –

an update of the match mismatch hypothesis. Advances in Marine Biology, 26,

249–293.

Dawson A (2008) Control of the annual cycle in birds: endocrine constraints and

plasticity in response to ecological variability. Philosophical Transactions of the Royal

Society B-Biological Sciences, 363, 1621–1633.

Edwards M, Richardson AJ (2004) Impact of climate change on marine pelagic

phenology and trophic mismatch. Nature, 430, 881–884.

Elzinga JA, Atlan A, Biere A, Gigord L, Weis AE, Bernasconi G (2007) Time after time:

flowering phenology and biotic interactions. Trends in Ecology and Evolution, 22,

432–439.

Englund G, Sarnelle O, Cooper SD (1999) The importance of data-selection criteria:

meta-analyses of stream predation experiments. Ecology, 80, 1132–1141.

Fitter AH, Fitter RSR (2002) Rapid changes in flowering time in British plants. Science,

296, 1689–1691.

Gillet C, Dubois JP (2007) Effect of water temperature and size of females on the timing

of spawning of perch Perca fluviatilis L. in Lake Geneva from 1984 to 2003. Journal of

Fish Biology, 70, 1001–1014.

Gurevitch J, Curtis PS, Jones MH (2001) Meta-analysis in ecology. Advances in Ecological

Research, 32, 199–247.

Harrington R, Clark SJ, Welham SJ et al. (2007) Environmental change and the

phenology of European aphids. Global Change Biology, 13, 1550–1564.

Harrington R, Woiwod I, Sparks T (1999) Climate change and trophic interactions.

Trends in Ecology and Evolution, 14, 146–150.

IPCC. (2007) Summary for policymakers. In: Climate Change 2007: Impacts, Adaptation

and Vulnerability. Contribution of working group II to the Fourth Assessment Report of the

Intergovernmental Panel on Climate Change (eds Parry ML, Canziani OF, Palutikof JP,

Van Der Linden PJ, Hanson CE), Cambridge University Press, Cambridge.

Langvatn R, Mysterud A, Stenseth NC, Yoccoz NG (2004) Timing and synchrony of

ovulation in red deer constrained by short northern summers. American Naturalist,

163, 763–772.

Menzel A, Sparks TH, Estrella N et al. (2006) European phenological response to

climate change matches the warming pattern. Global Change Biology, 12, 1969–1976.

Millenium Ecosystem Assessment. (2005) Ecosystems and human well-being: synthesis.

Island Press, Washington, DC.

Miller-Rushing AJ, Inouye DW, Primack RB (2008) How well do first flowering dates

measure plant responses to climate change? The effects of population size and

sampling frequency. Journal of Ecology, 96, 1289–1296.

M�ller AP, Rubolini D, Lehikoinen E (2008) Populations of migratory bird species

that did not show a phenological response to climate change are declining.

Proceedings of the National Academy of Sciences of the United States of America, 105,

16195–16200.

Parmesan C (2007) Influences of species, latitudes and methodologies on estimates of

phenological response to global warming. Global Change Biology, 13, 1860–1872.

Parmesan C, Yohe G (2003) A globally coherent fingerprint of climate change impacts

across natural systems. Nature, 421, 37–42.

Penuelas J, Filella I, Comas P (2002) Changed plant and animal life cycles from 1952 to

2000 in the Mediterranean region. Global Change Biology, 8, 531–544.

Platt T, Fuentes-Yaco C, Frank KT (2003) Spring algal bloom and larval fish survival.

Nature, 423, 398–399.

Precht H, Christophersen J, Hensel H, Larcher W (1973) Temperature and life. Springer-

Verlag, Berlin.

Root TL, MacMynowski DP, Mastrandrea MD, Schneider SH (2005) Human-modified

temperatures induce species changes: joint attribution. Proceedings of the National

Academy of Sciences of the United States of America, 102, 7465–7469.

Root TL, Price JT, Hall KR, Schneider SH, Rosenzweig C, Pounds JA (2003) Finger-

prints of global warming on wild animals and plants. Nature, 421, 57–60.

Rosenheim JA, Tabashnik BE (1991) Influence of generation time on the rate of

response to selection. American Naturalist, 137, 527–541.

Roy DB, Sparks TH (2000) Phenology of British butterflies and climate change. Global

Change Biology, 6, 407–416.

Scheffer M, Carpenter SR (2003) Catastrophic regime shifts in ecosystems: linking

theory to observation. Trends in Ecology and Evolution, 18, 648–656.

Sherry RA, Zhou XH, Gu SL et al. (2007) Divergence of reproductive phenology under

climate warming. Proceedings of the National Academy of Sciences of the United States of

America, 104, 198–202.

Smart SM, Clarke RT, Van De Poll HM, Robertson EJ, Shield ER, Bunce RGH, Maskell

LC (2003) National-scale vegetation change across Britain; an analysis of sample-

based surveillance data from the Countryside Surveys of 1990 and 1998. Journal of

Environmental Management, 67, 239–254.

Stefanescu C, Penuelas J, Filella I (2003) Effects of climatic change on the phenology of

butterflies in the northwest Mediterranean Basin. Global Change Biology, 9,

1494–1506.

Thackeray SJ, Jones ID, Maberly SC (2008) Long-term change in the phenology of

spring phytoplankton: species-specific responses to nutrient enrichment and

climatic change. Journal of Ecology, 96, 523–535.

Thompson R, Clark RM (2008) Is spring starting earlier? Holocene, 18, 95–104.

Van Buskirk J, Mulvihill RS, Leberman RC (2009) Variable shifts in spring and autumn

migration phenology in North American songbirds associated with climate change.

Global Change Biology, 15, 760–771.

Visser ME, Both C (2005) Shifts in phenology due to global climate change: the need for

a yardstick. Proceedings of the Royal Society of London Series B-Biological Sciences, 272,

2561–2569.

Winder M, Schindler DE (2004) Climate change uncouples trophic interactions in an

aquatic ecosystem. Ecology, 85, 2100–2106.

3312 S . J . T H A C K E R AY et al.

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313

Supporting Information

Additional Supporting Information may be found in the online version of this article:

Table S1. Data sources for the meta-analysis of changing phenology. The column ‘data period’ indicates the longest analytical time

period to which each scheme contributed data. Where this is a multi-decadal block, the scheme also contributed data to each of the

constituent decades.

Table S2. Taxa and other phenological indicators included in meta-analyses of changing phenology. The functional categorizations

given to each taxon in the meta-analysis are shown as (Environment, Trophic level, Thermal physiology, Generation time). En-

vironment 5 M, marine; F, freshwater; T, terrestrial. Trophic level 5 PP, Primary producer; PC, Primary consumer; SC, Secondary

consumer. Thermal physiology 5 En, Endotherm; Ec, Ectotherm. Generation time 5 ow, o1 week; w-m, 1 week-1 month; m-y, 1

month-1year; 4y, 41 year. Taxa have been assigned to Orders according to the Catalogue of Life, 2009 annual checklist (http://

www.catalogueoflife.org).

Table S3. Taxa and number of trends per taxon for the analysis of phenological change from 1976–2005, categorized by environment

and major taxonomic group. NA is given where a taxon does not have a common name, or a phenological indicator does not have an

appropriate scientific name.

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors.

Any queries (other than missing material) should be directed to the corresponding author for the article.

P H E N O L O G I C A L C H A N G E A C R O S S M A J O R E N V I R O N M E N T S 3313

r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 3304–3313