Page 1
Global impacts of the 1980s regime shift
PHILIP C. REID1 , 2 , 3, RENATA E. HARI4, GR?EGORY BEAUGRAND1 , 5, DAVID M.
LIVINGSTONE4, CHRISTOPH MARTY6, DIETMAR STRAILE7, JONATHAN
BARICHIVICH8 , 9, ERIC GOBERVILLE1 , 5, RITA ADRIAN1 0, YASUYUKI AONO1 1, ROSS
BROWN1 2, JAMES FOSTER1 3, PAVEL GROISMAN1 4 , 1 5, PIERRE H?ELAOU€ET1, HUANG-
HSIUNG HSU1 6, RICHARD KIRBY2, JEFF KNIGHT1 7, ALEXANDRA KRABERG1 8, JIANPING
LI1 9 , 2 0, TZU-TING LO2 1, RANGA B. MYNENI2 2, RYAN P. NORTH4 , 2 3, J. ALAN POUNDS2 4,
TIM SPARKS2 5 , 2 6 , 2 7 , 2 8, REN?E ST€UBI2 9, YONGJUN TIAN3 0 , 3 1, KAREN H. WILTSHIRE1 8,
DONG XIAO3 2and ZAICHUN ZHU33,34
1Sir Alister Hardy Foundation for Ocean Science, The Laboratory, Citadel Hill, Plymouth PL1 2PB, UK,2Marine Institute,
Plymouth University, Drake Circus, Plymouth PL4 8AA, UK,3Marine Biological Association of the UK, The Laboratory, Citadel
Hill, Plymouth PL1 2PB, UK,4Eawag, Swiss Federal Institute of Aquatic Science and Technology,€Uberlandstrasse 133, CH-8600
D€ ubendorf, Switzerland,5Centre National de la Recherche Scientifique, Laboratoire d’Oc? eanologie et de G? eosciences (LOG), UMR
8187 LOG, Universit? e des Sciences et Technologies de Lille, BP 80, 62930 Wimereux, France,6WSL Institute for Snow and
Avalanche Research SLF, Fl€ uelastrasse 11, CH-7260 Davos, Switzerland,7Department of Biology, Limnological Institute,
University of Konstanz, 78464 Konstanz, Germany,8Climatic Research Unit, School of Environmental Sciences, University of
East Anglia, Norwich NR4 7TJ, UK,9Laboratoire des Sciences du Climat et de l’Environnement, CEA-CNRS-UVSQ, L’Orme des
Merisiers, 91191 Gif-sur-Yvette, France,10Department of Ecosystem Research, Leibniz- Institute of Freshwater Ecology and Inland
Fisheries, M€ uggelseedamm 301, D-12587 Berlin, Germany,11Graduate School of Life and Environmental Sciences, Osaka
Prefecture University, Sakai 599-8531, Japan,12Climate Research Division, Science and Technology Branch, Environment Canada
Ouranos, 550 Sherbrooke St. West, 19th Floor, Montr? eal, QC H3A 1B9, Canada,13Code 917, NASA/Goddard Space Flight Center,
Greenbelt, MD 20771, USA,14National Centers for Environment Information - Center for Weather and Climate, Federal Building,
151 Patton Avenue, Asheville, NC 28801, USA,15P.P. Shirshov Institute for Oceanology, RAS, 36 Nakhimovsky Avenue, 117997
Moscow, Russia,16Research Center for Environmental Changes, Academia Sinica, 128 Academia Road, Section 2, Nankang, Taipei
115,Taiwan,17Met Office, Hadley Centre, FitzRoy Road, Exeter, Devon EX1 3PB, UK,18Alfred-Wegener Institute for Polar and
Marine Research, Biologische Anstalt Helgoland, Kurpromenade 201, 27498 Helgoland, Germany,19College of Global Change and
Earth System Science (GCESS), Beijing Normal University, Beijing 100875, China,20Joint Center for Global Change Studies,
Beijing 100875, China,21Weather Forecast Center, Central Weather Bureau, 64 Gongyuan Road, Taipei 10048, Taiwan,
22Department of Earth and Environment, Boston University, 685 Commonwealth Avenue, Boston, MA 02215, USA,23Helmholtz-
Zentrum Geesthacht, Institute of Coastal Research, Max-Planck-Str. 1, D-21502 Geesthacht, Germany,24Monteverde Cloud Forest
Preserve, Tropical Science Center, Santa Elena, Puntarenas 5655-73, Costa Rica,25Institute of Zoology, Poznan ´ University of Life
Sciences, Wojska Polskiego 71 C, 60-625 Poznan ´, Poland,26Faculty of Engineering, Environment and Computing, Coventry
University, Coventry CV1 5FB, UK,27Fachgebiet fu ¨r O¨koklimatologie, Technische Universita ¨t Mu ¨nchen, Hans-Carl-von-
Carlowitz-Platz 2, 85354 Freising, Germany,28Institute for Advanced Study, Technische Universita ¨t Mu ¨nchen, Lichtenbergstrasse
2a, 85748 Garching, Germany,29Federal Office of Meteorology and Climatology, MeteoSwiss, Ch. de l’Ae ´rologie 1, CH-1530
Payerne, Switzerland,30Fisheries College, Ocean University of China, Yushan Road 5, Qingdao 266003, China,31Japan Sea
National Fisheries Research Institute, Fisheries Research Agency, Chuo-ku, Niigata 951-8121, Japan,32Chinese Academy of
Meteorological Sciences, Beijing 100081, China,33State Key Laboratory of Remote Sensing Science, Institute of Remote Sensing
and Digital Earth, Chinese Academy of Sciences, Beijing 100101, China,34Center for Applications of Spatial Information
Technologies in Public Health, Beijing 100101, China
Abstract
Despite evidence from a number of Earth systems that abrupt temporal changes known as regime shifts are impor-
tant, their nature, scale and mechanisms remain poorly documented and understood. Applying principal component
analysis, change-point analysis and a sequential t-test analysis of regime shifts to 72 time series, we confirm that the
1980s regime shift represented a major change in the Earth’s biophysical systems from the upper atmosphere to the
depths of the ocean and from the Arctic to the Antarctic, and occurred at slightly different times around the world.
Using historical climate model simulations from the Coupled Model Intercomparison Project Phase 5 (CMIP5) and
statistical modelling of historical temperatures, we then demonstrate that this event was triggered by rapid global
warming from anthropogenic plus natural forcing, the latter associated with the recovery from the El Chich? on
Correspondence: Philip C. Reid, tel. +441752 633269, fax +441752 600015, e-mail: pcre@sahfos.ac.uk
1
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use,
distribution and reproduction in any medium, provided the original work is properly cited.
Global Change Biology (2015), doi: 10.1111/gcb.13106
Page 2
volcanic eruption. The shift in temperature that occurred at this time is hypothesized as the main forcing for a cas-
cade of abrupt environmental changes. Within the context of the last century or more, the 1980s event was unique in
terms of its global scope and scale; our observed consequences imply that if unavoidable natural events such as major
volcanic eruptions interact with anthropogenic warming unforeseen multiplier effects may occur.
Keywords: climate, Earth systems, global change, regime shift, statistical analysis, time series, volcanic forcing
Received 23 July 2015 and accepted 3 September 2015
Introduction
Regime shifts are abrupt, substantial and persistent
changes in the state of natural systems. Such shifts have
been observed in the atmosphere (Lo & Hsu, 2010; Xiao
et al., 2012), ecosystems (Hastings & Wysham, 2010)
and human social systems (Campbell & Allen, 2001).
Three regime shifts (1970s, 1980s and 1990s), distin-
guished by marked increases in temperatures or by
abrupt temporal changes across different biophysical
systems, have been identified in the last few decades
(Hare & Mantua, 2000; Reid et al., 2001; Gong & Ho,
2002; Yasunaka & Hanawa, 2002; Peterson & Schwing,
2003; Beaugrand et al., 2013). Documented until now at
ocean basin or regional scales, the mechanisms behind
these events, their environmental interactions, and the
synchrony and scale of their effects around the globe
are poorly understood. There is thus a considerable
research gap with many disparate observations by dif-
ferent scientific disciplines, but no comprehensive over-
all assessment. Here, we address this gap by focusing
on the 1980s regime shift and show, using three inde-
pendent statistical methods that this shift took place on
a planetary scale and involved the carbon cycle (Beau-
lieu et al., 2012b); disease (Vezzulli et al., 2012); and bio-
tic, physical and chemical components of land (Myneni
et al., 1997; Brandt et al., 2013), freshwater (Hari et al.,
2006), precipitation (Tao et al., 2015), marine (M€ ollmann
& Diekmann, 2012; Beaugrand et al., 2015) as well as
cryospheric (Brown & Robinson, 2011) and atmospheric
(Lo & Hsu, 2010; Xiao et al., 2012) Earth systems. A total
of 72 time series was processed and analysed statisti-
cally to represent as many natural systems as possible,
and to illustrate shiftlike abrupt changes in a ‘1980s per-
iod of interest’ (1983 to 1990) between the volcanic
eruptions of El Chich? on and Pinatubo (see Materials
and methods).
To explore possible mechanisms behind the 1980s
regime shift, we used historical climate model simula-
tions from the Coupled Model Intercomparison Project
Phase 5 (CMIP5) (Jones et al., 2013) together with statis-
tical modelling (Folland et al., 2013) of the main anthro-
pogenic radiative forcing and natural (volcanic and
solar) factors influencing global mean surface tempera-
ture. Using these approaches, we show that the rapid
cooling of the Earth’s surface (Robock, 2000), and
especially of the oceans (Church et al., 2005), initiated
by the El Chich? on volcanic eruption of 1982 was fol-
lowed by a recovery reinforced by anthropogenic
warming. It is the scale and speed of these combined
heating effects that we propose contributes to the syn-
chronization of the regime shift between different sys-
tems. Although temperature appears to be the main
forcing factor, volcanic and anthropogenic aerosols and
their interactions with clouds (IPCC Chap. 7, 2013) and
the brightening effect described by Wild (2009) may
also have contributed. The 1980s regime shift is an
example of unforeseen compounding effects that may
occur if unavoidable natural events such as major vol-
canic eruptions interact with anthropogenic warming.
Materials and methods
Data
Time series selection. Long time series of variables represent-
ing the various key components of the climate system
(drivers) and a wide range of environmental and ecological
indicators (responders) were used for our study. Half the 72
time series represent global and hemispheric (26) to local (10)
land, sea and freshwater temperatures, with the other half
covering, at global to local scales, the carbon cycle (3) and the
following natural systems: atmosphere (9), cryosphere (6),
marine hydrosphere (2), marine biosphere (3), terrestrial
hydrosphere (3) and terrestrial biosphere (10). We have used a
variety of different forms of measurements with one value per
year: for example, averaged over single months, seasons or
yearly, or in the case of phenology the timing date. Some of
the time series presented are averages of data sets where our
analyses showed the shift in most of the members, for exam-
ple 18 river water temperatures averaged for Switzerland or
many gridded temperature data sets; others are representa-
tives of vertical profiles in the atmosphere or ocean that show
similar shifts at other different heights or depths. The majority
of the time series are from the Northern Hemisphere. Few
unbroken long-term time series from the tropics exist that
have been produced using the same standard protocols over
time. Furthermore, time series data, other than for land sur-
face temperature (LST) and sea surface temperature (SST)
from ocean areas outside the North Atlantic and North Pacific,
are very sparse.
Considerable effort was put into the search for long-term
time series and included the following: submissions from rec-
ommendations by the authors, downloading from open-access
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
2 P. C. REID et al.
Page 3
online databases, requesting data from original literature
sources after an extensive search for long-term time series and
a number of exploratory searches for time series to cover
perceived gaps by theme, for example natural system and geo-
graphical region. In addition, data sets were processed to
cover large geographical regions, for example continents and
ocean basins, independently of global and hemispheric data
that are often readily available. All acquired time series were
processed in the same way. The resulting compilation of data
was dependent on sampling, monitoring location and avail-
ability, and does not claim to be comprehensive or geographi-
cally representative of the whole world.
For inclusion in the analysed database, continuous data sets
needed to be within the period 1946 to 2012 and start by 1980
at the latest. The year 1946 was chosen, when possible, as the
start date to exclude the poor sampling during World War II.
There were two exceptions, one to allow satellite information
to be included using the Normalized Difference Vegetation
Index (NDVI), start 1982 (Myneni et al., 1997) and the other
mesopelagic fish eggs, start 1981 (Fujino et al., 2013).
Data sources. The 72 time series selected for more detailed
study were chosen to yield as wide a global geographical cov-
erage as possible; they include data from single sites, sets of
data, and data averaged at global, hemispheric, continental
and ocean scales. The data used in the study can be obtained
from Table S1 or from the corresponding author. Citations for
all the time series analysed are given in Table S2. The years
from 1983 to 1990, between the major volcanic eruptions of El
Chich? on (1982) and Pinatubo (1991), were focused on as a ‘pe-
riod of interest’ to study the 1980s regime shift.
Two data sources for temperature were used to derive
continental, oceanic, global and hemispheric means. First, pre-
calculated 5° grid annual mean anomalies with respect to
1961–1990 produced by the Hadley Centre of the UK Met
Office and the Climatic Research Unit, University of East
Anglia (http://www.metoffice.gov.uk/hadobs/index.html),
were downloaded: HadCRUT4.3.0.0 (combined LST and SST);
HadSST3.1.1.0 (SST); and CRUTEM4.3.0.0 (LST), the latter cal-
culated from ~5500 monthly meteorological station tempera-
tures. Second, monthly mean data were downloaded from the
NASA Global Historical Climatology Network (GHCN) data
set (ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v3). While the
first set was prepared by the provider, with this data set we
were able to preprocess the data ourselves to take into account
gaps in sampling in time and space before calculating global,
hemispheric and continental means.
Processing the GHCN global temperature data set. We used
data from the GHCN data set to provide an independent
temperature time series that is based on real and not gridded
data. After preliminary investigations, the data set used for
further study was limited to land meteorological stations from
continental regions (6449 stations). The distribution of these
stations was not random and did not systematically cover all
regions of the world with a similar density, and many had
missing data. To address these issues, the sequence of
subsequent processing followed the order:
1 To reduce the number of stations with missing data, new
regional mean time series were calculated for areas where
adjacent weather stations were likely to show similar
temperatures. A sensitivity analysis was carried out to
check the similarity in data from stations enclosed within
circles of diameter 50, 100, 150 and 200 km. A circle with a
diameter of 100 km gave the best result, so ‘circle time ser-
ies’ were calculated by averaging the data from stations
within circles of this diameter.
2 To decide which stations to include within each circle time
series, a threshold test was applied to filter the station time
series for missing years. Seven options were tested: with 5,
10, 15, 20, 25 or 30 missing years, or with no filtering. A
threshold with a maximum of 10 years’ missing data was
selected, as using this threshold the signal was not altered
by missing values.
3 For the calculation of annual means, at least nine sampled
months were required, otherwise a missing value was
attributed.
4 A final set of time series was then generated for each
selected 100-km-diameter circle to give 302 circles in Eur-
ope, 624 in Asia, 228 in Australia, 97 in Africa, 1932 in North
America and 108 in South America, with 2977 for the North-
ern Hemisphere, 314 for the Southern Hemisphere and 3291
for the world. Of these 3291 time series, the maximum num-
ber of circles that overlap is 37.
5 All the GHCN time series were anomalized with respect to
their own mean over the period 1961–1990 for comparison
with the Hadley and CRU time series.
6 A control sensitivity analysis was applied to determine
whether any change in the number of circle time series per
year coincided with observed regime shifts in the data.
There were no major changes in the number of missing sta-
tions until 1989/1990 in some regions of the world and
more globally in 2000; the former reduction in stations coin-
cided with the break-up of the USSR. This result gives
increased confidence to the regime shifts identified in the
GHCN data prior to 1989.
Statistical analysis
Identifying regime shifts. Three different statistical methods
were used to identify regime shifts.
First, a standardized principal component analysis (PCA)
(Beaugrand et al., 2002) was applied to the whole data set. The
period 1968 to 2010 was selected for the analysis as at least 58
of the 72 time series (80%) had data throughout this interval.
PCA is not sensitive to temporal autocorrelation. Results of
the analysis are shown in Fig. 1 and Table 1.
Second, we used a change-point analysis (Taylor, 2000) to
identify the timing of the shift in the first principal component;
a technique that is simple, is not sensitive to outliers and takes
into account the effects of temporal autocorrelation when cal-
culating the probability of the shift. The cumulative sums of
the time series (i.e. the first principal component) were calcu-
lated (Iba~ nez et al., 1993; Kirby et al., 2009); then, the first dif-
ference in the cumulative sums was estimated, followed by a
Monte Carlo test to determine the probability that a shift had
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
3
Page 4
occurred based on the number of times the simulated first dif-
ference exceeded the observed amplitude. About 100,000 runs
were performed, and the simulated time series were retained
if their order-1 autocorrelation was higher or equal to the one
observed in the original time series.
Third, for standardization and comparison of individual
time series, a ‘multiple’ sequential t-test analysis of regime
shifts (STARS) based on the Rodionov method (Rodionov,
2004; Rodionov & Overland, 2005) was applied. This method
was used individually on each time series; the results in the
period of interest are given in Figs 2–6 and Table 2.
Variant of the Rodionov method. Generally, a t-test examines
whether the means of two sample populations differ signifi-
cantly, based on the standard deviations and on the number
of values in each sample. The existence of a trend in one or
both samples is not excluded.
The STARS method (Rodionov, 2004; Rodionov & Over-
land, 2005) tests whether the end of one period (regime) of a
certain length is different from a subsequent period (new
regime). The cumulative sum of normalized deviations from
the hypothetical mean level of the new regime is calculated,
and then compared with the mean level of the preceding
regime. A shift year is detected if the difference in the mean
levels is statistically significant according to a Student’s t-test.
In his third paper, Rodionov (2006) shows how autocorrela-
tion can be accounted for. From each year of the time series
(except edge years), the rules are applied backwards and for-
wards to test that year as a potential shift year. The method is,
therefore, a running procedure applied on sequences of years
within the time series.
The multiple STARS method used here repeats the proce-
dure for 20 test-period lengths ranging from 6 to 25 years that
are, for simplicity (after testing many variations), of the same
length on either side of the regime shift. The last year of the
−5
0
5
(a)
First principal
component
1970 1980199020002010
−5
0
5(b)
Second principal
component
Fig. 1 A standardized principal component analysis (PCA) of
long-term changes in 72 time series. (a) First principal compo-
nent (49% total variance) and (b) second principal component
(12% total variance). The red dashed line marks a significant
regime shift year in 1987 (P ≤ 0.05), identified by both principal
component and change-point analyses.
Table 1
of an analysis performed on the whole 72 time series as an
entity for the period 1968–2010 where 80% of the time series
had no gaps. The first two normalized eigenvectors show the
correlation between each variable and the first two principal
components. Colour code for eigenvector contribution: posi-
tive red 0.50–0.75, red bold 0.76–1; negative the same in green
(See also Fig. 1). See Table 2 legend for ’system’ acronyms
Standardized principal component analysis. Results
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
4 P. C. REID et al.
Page 5
old regime is defined as the shift year, and all results shown
are for P ≤ 0.05. Our approach allows us to assess the strength
of a regime shift by the number of test-period lengths with sig-
nificant results, while Rodionov uses a regime shift index to
weight his results.
Not all 20 test-period lengths can be applied to all the
time series as the length of the time series, and the location
of the shift year in the time series can sometimes make this
impossible. For example, it would not be possible to apply
the longest test-period length of 25 years to a short time ser-
ies. For comparative purposes, we therefore determine the
number of significant results as a percentage of the maxi-
mum possible, and set a minimum threshold of 20% to
accept a significant shift year. This approach favours the
detection of changes between long regimes and discards
changes between short regimes, such as, for example, that in
the NAO from 1989 to 1995, which lasted only 7 years
(Fig. 2i).
Detection of a shift year depends very much on the test-pe-
riod used. Extreme values in a time series can create a barrier
for short test-periods and limit a shift year result to a small
part of the whole time series. On the other hand, long test-pe-
riods may integrate these values to give significant shift years
for longer periods. These observations emphasize the impor-
tance of varying the test-period length.
All acquired time series were analysed using the multiple
STARS method. In some cases, because we used various test-
period lengths, two adjacent regime shift years were identified.
Possible reasons for this duplication, other than the distribu-
tion of the values in the time series, include the following: that
the shift event takes time to evolve over more than 1 year; that
some examples of shifts occur in the last months of 1 year and
in the first months of the next; and that several different shift
years occur in a large area that has been averaged over.
A comparison of real data with artificial time series. To
demonstrate that the identification of regime shifts in our
Table 1
(Continued)
Table 1
(Continued)
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
5
Page 6
−55
−52Swiss stratospheric T
(a)
−21
−19
Swiss tropospheric T
(b)
0.03
0.08Meridional wind 60−75°N
(c)
4
6
Zonal wind 60−75°N
(d)
2
6
China spring dust storms
(e)
600
900Global tropical hurricanes
(f)
−1
1NH SV−NAM Index
(g)
−1
1Arctic Oscillation AO
(h)
−5
2Arctic land & sea T
(j)
5N−Atlantic Osc. NAO winter
(i)
0
1960 19802000
1012
1016Arctic sea level pressure
(k)
−2
−1Alaska atm. CO2, Apr−Sep
(l)
1
2
Alaska atm. CO2, Oct−Mar
(m)
−1
1
Global CO2 net land uptake
(n)
0
200 NH T growing season
(o)
−30
−26
W−antarctic land T
(p)
1.2
1.6
W−antarctic sea−ice extent
(q)
32
36
NH spring snow extent
(r)
0
50
Swiss snow days, Dec−Mar
(s)
100
300
Baltic Sea sea−ice extent
(t)
8
16Arctic sea−ice volume Sep
(u)
1960 19802000
4
6
NH sea−ice extent Sep
(v)
0
40
W−USA Wildfire duration
(w)
18
19
NH satellite vegetation
(x)
−5
5 NH start growing season
(y)
0
10NH length growing season
(z)
0
5
NH end growing season
(aa)
90
105Japan cherry blossom
(ab)
75
90
UK sand martin arrival
(ac)
210
240Germany grape ripening
(ad)
300
600Baltic river winter flow
(ae)
10
11
Swiss river T
(af)
19601980 2000
7.9
8.1
Swiss river pH
(ag)
1
2
North Sea phytoplankton
(ah)
−0.5
0.5North Sea T
(ai)
34.5
35N−Sea 50 m depth salinity
(aj)
12
14Japan Sea T 50 m depth
(ak)
20
40
N−Pac. Kuroshio current
(al)
0
15Japan Sea tuna catch
(an)
50
Japan Sea deep fish eggs
(am)
10
15
20Germany lake algal bloom
(ao)
11
12Swiss groundwater T
(ap)
6
7
Swiss Lake Zürich T
(aq)
1960 19802000
−1
1
SH Annular Mode SAM
(ar)
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
6 P. C. REID et al.
Page 7
analyses, based on the multiple STARS method, is not spuri-
ous, we compared the results obtained from the real time ser-
ies with the results derived from two sets of artificial time
series. The procedure used to create these autocorrelated time
series is fully described in Beaugrand et al. (2014). First, we
examined the 72 real time series with an autocorrelation func-
tion. The resulting autocorrelograms include information on
the order if the correlation and the value for each lag, and
showed a pronounced autocorrelation for some of our real
time series (Fig. S1a). Second, the artificial time series were
produced with two types of temporal autocorrelation, ‘high’
and ‘medium’, to approximate reality. For each type, 70 time
series with 67 hypothetical years were simulated to corre-
spond to the maximum period of the real time series (1946–
2012, a total of 4690 years each). In the real data, some of the
72 time series extended over a shorter period (a total of
4104 years or 88% of the maximum possible).
The two types of autocorrelation were as follows: (1) High
temporal autocorrelation, where 70 time series were con-
structed with a linear trend of magnitude 100 (arbitrary units)
and random temporal fluctuations of magnitude 120 (green in
Fig. S1b). These time series had an autocorrelation that corre-
sponded to the maximum autocorrelation observed in the real
time series (upper fine lines in Fig. S1a). (2) Medium temporal
autocorrelation, where 70 time series were constructed with a
linear trend of magnitude 100 (arbitrary units) and random
temporal fluctuations of magnitude 45 (blue in Fig. S1b). These
time series had an autocorrelation that corresponded to the
medium autocorrelation in the observed time series (central
fine lines in Fig. S1a). The thresholds of 45 and 120 were cho-
sen to correspond to the two main situations (i.e. high and
medium levels of autocorrelation) encountered in the time ser-
ies. The total number of significant test-periods and shift years
for the real and the two artificial time series sets are shown in
Fig. S2.
A further comparison between the results for the real
and artificial data sets shows the longest significant test-pe-
riods observed (Fig. S3a–c), where the triangular shape of
the plots reflects the longest possible test-period length at
the edges of the time series, from 6 years at the bottom to
25 years at the top. The strength of shift years in a specific
calendar year is shown as a percentage of the number of
significant test-periods against the total number of possible
test-periods (Fig. S3d–f). In the middle of long time series
the latter is 20 test-periods, but at the edges it is less. A
cut-off of 20% was applied to eliminate the majority of only
one or two significant test-periods, which are more likely to
be artefacts.
Modelling
CMIP5 historical climate modelling. Global mean historical
CMIP5 temperature data (Jones et al., 2013) were used to attri-
bute the changes in observed global temperature from the
HadCRUT4.3.0.0 data set (Morice et al., 2012). Two multi-
model ensembles were selected, firstly with natural (solar and
volcanic) forcings (with a total of 46 members), and secondly
with both natural and anthropogenic forcings (99 members).
The multimodel ensemble mean is taken to represent the best
estimate of the response of global mean temperature to the
forcings in each ensemble. The difference between the two
ensemble means is used as an estimate of the effect of anthro-
pogenic forcings on global mean temperature. Linear least-
squares fitting was used to obtain running 7-year trends in the
various global mean time series.
Statistical modelling. Estimates of the response of global
mean temperature to solar and volcanic forcings individually
were produced based on statistical reconstructions (Folland
et al., 2013). This latter study used a cross-validated multiple
regression approach on data from 1891 to 2011 to estimate the
effect of a range of known influences on global mean tempera-
ture. The sum of the solar and volcanic reconstructions was
calculated to provide an equivalent to the ensemble mean
temperature in the CMIP5 natural forcing ensemble. Calcula-
tion of anomalies and trends was performed using similar
techniques to those used for the CMIP5 data.
Results
Identification of the 1980s regime shift in a wide range of
Earth systems
Many statistical techniques exist to identify regime
shifts (e.g. Beaugrand, 2004; Mantua, 2004; Rodionov
& Overland, 2005; Rodionov, 2006; Beaulieu et al.,
2012a; Varotsos et al., 2013). All have strengths and
drawbacks, but the three we use here are complemen-
tary. To determine the main long-term patterns of
variability, we applied a standardized principal com-
ponent analysis (PCA) (Beaugrand et al., 2002) to the
data set as an entity. The results (Fig. 1, Table 1)
showed a clear shift in 1987 for the first component,
which accounts for almost half (49%) of the total vari-
ance. A large number of time series, especially of vari-
Fig. 2 Significant regime shifts in time series representing a range of different Earth systems. Vertical lines denote regime shift years
(P ≤ 0.05), coloured throughout the paper to reflect the ‘1980s period of interest’ (1983–1990): 1984 (blue), 1985 (green), 1986 (orange),
1987 (red), 1988 (brown), 1989 (purple), 1991 (pink as a lagged effect); grey solid lines mark the earlier and later regime shifts (1976 and
1996), and grey dashed lines other significant shift years outside the 1980s. Horizontal lines mark the longest test-period with a signifi-
cant result. Triangles (coloured as per the shift year) point up or down to indicate the direction of a significant trend before or after the
shift year: here 3 before, 12 after the regime shift and 26 time series with a shift in the ‘period of interest’ but no trends. The plots are
presented in the following sequence: atmosphere (a–p) and (ar), cryosphere (q–v), terrestrial biosphere (w–ad), hydrosphere (ae–aq).
Further details for each of the time series are given in Tables 3 and S3 (origin, units and shift years).
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
7
Page 8
ables related to temperature and vegetation, are
strongly correlated with the first component. The sec-
ond component, which accounts for 12% of the total
variance, is correlated, for example, with the Arctic
Oscillation, North Atlantic Oscillation, Arctic sea-level
pressureandzonal wind.
change-point analysis (Taylor, 2000) to identify step-
wise shiftsalongthe first
(Fig. 1a); the analysis detected a significant regime
shift in 1987 (P ≤ 0.05).
For standardization and a more detailed analysis of
regime shifts in individual time series, the multiple
version of the STARS method (Rodionov, 2004, 2006;
Rodionov & Overland, 2005) described above (see Ma-
terials and methods) was employed. The well recog-
nized and documented Rodionov method has now
been used in several dozen papers (e.g. Luczak et al.,
2011; Jaffr? e et al., 2013; Litzow & Mueter, 2014). Multi-
ple STARS results for 44 time series representative of
sixnatural systems (terrestrial
spheres/hydrospheres,the
sphere) are shown in Fig. 2, with more information
on the time series given in Table 3 and Table S3. The
multiple STARS method identifies a shift from a
mostly stationary state to a new state, which in a
third of cases contains a trend after the regime shift
(Table 2). Note that our definition allows a significant
linear trend within a regime.
The results of the multiple STARS method stress
again the importance of the 1980s event compared to
the smaller regime shifts in the 1970s and 1990s. The
dominant 1980s regime shift also shows synchronous
timing for systems and geographical regions. A total of
165 step changes were identified in the 72 time series
over the analysis period from 1946 to 2012. Of these,
11% occurred in the late 1970s (1973–1980), 40% in the
late 1980s (1983–1990) and 25% in the late 1990s
(1993–2000). A comparison between the scale of the
1970s, 1980s and 1990s regime shifts can also be made
by noting the maximum hemispheric step change in
temperature shown by the GHCN LST time series with
differences between the old and new regimes of 0.38 °C
in 1976 (Southern Hemisphere), 0.80 °C in
(Northern Hemisphere, Table 2) and 0.66 °C in 1997
(Northern Hemisphere).
Within the ‘1980s period of interest’ (1983–1990), 66
step changes were found in 59 of the 72 time series
(82%). In some of these, none were found, and in others
more than one. Of the 66 detected shift years, very
strong shifts (100% of the possible test-periods signifi-
cant) were found in 27% and strong shifts (≥50% of the
possible options) in 74%, the latter divided into 35%
and 39% between temperature (for all air, sea and
freshwater data sets) and a grouping of all other time
series (36 each). Of the 44 time series presented in
Fig. 2, 41 show a step-change in the mid-1980s, only
two (Fig. 2v, ac) in the mid-1970s, and only 11 (Fig. 2d,
We applied Taylor’s
principal component
and marine
and
bio-
cryo-atmosphere
1985
90
110
(a)
Japan Kyoto cherry blossom
80
100
120
(b)
Switzerland Liestal cherry blossom
1880 1900 19201940 19601980 2000
80
100
(c)
USA Washington D.C. cherry blossom
Fig. 3 Long-term time series of cherry blossom blooming dates
from three different continents. Units: day of the year. Vertical
lines denote significant regime shift years (P ≤ 0.05), coloured
for the ‘1980s period of interest’ (1983–1990) throughout the
paper, in this case: 1988 (brown) and 1985 (green); grey dashed
lines mark other significant shift years outside the 1980s. Hori-
zontal lines mark the longest test-period possible with a signifi-
cant result. (a) The date on which the blossom of the Japanese
cherry (Prunus jamasakura) comes into full bloom in Kyoto,
Japan, 35°N, 136.67°E (1873–2012). Data: updated and revised
from Yasuyuki Aono. Shift years: 1898, 1940, 1988. (b) The start
date of the flowering of blossoms on a cherry tree (Prunus
avium) at Liestal, Switzerland, 47.48°N, 7.44°E (1894–2012). The
tree is checked with a telescope every day at the beginning of
the flowering season; the date when 25% of the blossoms are
open marks the start. Data: Andreas Buser, Landwirtschaftliches
Zentrum Ebenrain, Sissach and MeteoSwiss, Switzerland. Shift
year: 1988. (c) The peak bloom date of the blossoms on the
Yoshino cherry trees (Prunus x yedonensis) in the tidal basin,
Washington D.C., USA, 38.88°N, 77.04°W (1921–2011). Peak
bloom date is defined as the day when 70% of the blossoms on
the trees in the basin are open. Data: from http://www.nps.-
gov/cherry/upload/Cherry-Festival-dates.pdf. Shift year: 1985.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
8 P. C. REID et al.
Page 9
j, k, u, v, x, aa and all the indices Fig. 2g, h, i, ar) in the
mid-1990s. The timing of the 1980s event in all natural
systems ranges from 1984 to 1989, with most changes
occurring in 1987 or 1988. Two examples are shown
where the timing of the shift may reflect a lagged
response (Swiss river pH and tuna catch in the Japan
Sea, Fig. 2ag, an).
The change can be expressed as a percentage for
24ofthe 72timeseries
anomalies, indices, day of year or pH). In 19 of
these 24 time series, the change between the two
levels of the longest test-period with a result was
≥10%, in 7 ≥ 50% and in 3 ≥ 100%. In one – wildfire
duration inthe western
increase of >400%. This means, from our calcula-
tions, that Western US wildfires lasted on average
(not fortemperature,
USA
–
therewas an
29 days
5.5 days between 1970 and 1985 (Table 2).
If the regimes (of n years length and for the longest
test-period with a result) to either side of the shifts
are considered, with the term ‘old’ applied to the per-
iod before and ‘new’ to that after the regime shift,
there are few significant trends in the old regimes,
but many more in the new, emphasizing the dynamic
nature of the change after the 1980s. Significant
trends in the old regimes before the shifts were only
found in 11% of the 66 shifts in the ‘period of inter-
est’ and in the new regimes after the shifts 41%,
while 55% had no significant trends either before or
after the shifts. Only three time series (Australia,
Fig. 4l; GHCN land SH, HadSST3 sea SH, Fig. 5c,i)
show trends in both old and new regimes, which
between1986and 2003compared to
0
1 (a)
Arctic Ocean
0
0.5
(b)
North Pacific
0
1 (i)
North America
0
0.5 (c)
North Atlantic
−1
1 (j)
Europe
0
1 (k)
Asia
0
0.5
(e)
South Pacific
1960 19802000
0
1
(l)
South America
−0.5
0.5
(f)
South Atlantic
1960 19802000
−1
1 (m)
Africa
−0.5
0.5 (d)
North Indian Ocean
19601980 2000
−1
1 (n)
Australia
196019802000
−0.5
0
(h)
Southern Ocean
196019802000
0
0.5 (g)
South Indian Ocean
Fig. 4 Time series of annual mean land/sea surface temperature for six continents and eight ocean basins. Anomalies calculated with
respect to their own mean over the period 1961–1990 (continents 1946–2011, oceans 1946–2012). Vertical and horizontal lines, symbols
and colours as per the legend for Fig. 2. Data for land from ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/v3/, processed by Eric Gober-
ville. Units: °C. (c) North America, shift years: 1985, 1997; (e) Europe: 1987; (f) Asia: 1988; (h) South America: 1976, 1984, 1998; (j) Africa:
1996; (l) Australia: 1956, 1971, 1978, 1987, 1996. Data for the sea processed by Jonathan Barichivich from http://www.metoffice.gov.uk/
hadobs/hadsst3/data/download.html. All grid boxes with data for a given region were selected, averaged, and weighted by the cosine
of the latitude to account for the changing size of grid cells towards the poles. Units: °C. (a) Arctic Ocean, shift years: 1961, 1962, 1999;
(b) North Pacific Ocean: 1969, 1970, 1985, 1989; (d) North Atlantic Ocean: 1969, 1970, 1986, 1994, 1996, 1997, 2002; (g) South Pacific
Ocean: 1978, 1994; (i) South Atlantic Ocean: 1958, 1971, 1997; (k) North Indian Ocean: 1986, 2000; (m) Southern Ocean: 1974, 1998, 2006;
(n) South Indian Ocean: 1976, 1986, 1996, 1997.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
9
Page 10
means for these three out of 72 another method
would be more appropriate.
To demonstrate that the identification of regime
shifts determined by the multiple STARS method is not
spurious and to address autocorrelation, we compared
the results from the real data with those from two sets
of artificial time series (Figs S1–S3). The sum of signifi-
cant test-periods or shift years in the ‘1980s period of
interest’ is much greater in the real time series than
achieved by the two sets of artificial time series
throughout the analysed time period and conclusively
confirms the reality of the 1980s regime shift and the
validity of the multiple STARS method. In the real data,
a total of 165 shift years were detected with 40% in the
‘period of interest’; in the data sets with high and
medium autocorrelation, the number of shift years
were 348 and 172, of which only 16% and 12%, respec-
tively, occurred in the ‘period of interest’. We expected
that the shift years would be evenly distributed over
the analysed time period in the artificial time series and
more concentrated in the real data, and this is clearly
evident. The probability of detecting a spurious shift is
thus likely to be the same in all of the time series. Both
sets of the artificial time series have a symmetrical pat-
tern to either side of the middle of the analysed time
period (Fig. S3b, c, e, f), whereas the real data are
strongly biased towards the second half of the time
series and especially within the ‘period of interest’
(Fig. S3a, d). An edge effect is evident in all the plots,
originating in the start and end phase of the time series
where shift years cannot be detected.
A long-term context for the 1980s shift is provided by
three centennial-scale time series of the flowering date
of cherry trees in Japan, Switzerland and the USA
(Fig. 3). The 1980s step changes at Liestal in 1988 and
Washington in 1985 are the only significant shifts in
these time series in at least 80 years. Flowering date cal-
ibrated against springtime air temperature showed that
the earliest timing of the bloom and warmest period in
over 1000 years in Kyoto (Aono & Kazui, 2008)
followed the shift in 1988.
In continental averages of LST (Fig. 4c, e, f, h, j, l), the
1980s regime shift is evident for all continents except
Africa, ranging from 1984 in South America to 1988 in
Asia. However, for mean oceanic SST (Fig. 4a, b, d, g, i,
k, m, n), the shift is only shown for the North Pacific in
1985, the North Atlantic in 1986 and the south and
north basins of the Indian Ocean in 1986. For both LST
and SST, the evidence for shifts in other decades is
limited. Our analyses confirm earlier findings that SST
has increased in steps over the last century (Reid &
Beaugrand, 2012; Varotsos et al., 2013) and that the pro-
nounced upward trend in global combined land and
sea temperatures seen in the decadal means of
Figure SPM.1 in the IPCC Summary for Policy Makers
(IPCC SPM, 2013; see also Fig. S4) started in the 1980s.
Land and sea surface temperatures averaged over the
entire globe and over the Northern and Southern Hemi-
spheres separately (Fig. 5) record rapid increases as sig-
nificant regime shifts centred on approximately 1976,
1986 and 1996. The relative increase in temperature is
much greater over land than over the sea, and also
larger over the Northern Hemisphere than over the
Southern Hemisphere. The 1970s regime shift was first
described for the North Pacific (Hare & Mantua, 2000),
0
1 (a)
Global
Land 1 [°C]
(b)
Northern H.
(c)
Southern H.
0
1 (d)
Land 2 [°C]
(e) (f)
0
1 (g)
Sea [°C]
(h) (i)
1960 2000
0
1 (j)
Land 2 & sea
19602000
(k)
19602000
(l)
Fig. 5 Time series of annual mean global and hemispheric land
and sea surface temperatures. Periods: 1946–2011/2012. (a–c)
GHCN land surface temperature. (d–f) CRUTEM4.3.0.0 land
surface temperature. (g–i) HadSST3.1.1.0 sea surface tempera-
ture. (j–l) HadCRUT4.3.0.0 combined land and sea surface tem-
perature. Vertical and horizontal lines, symbols and colours as
per the legend for Fig. 2. All data series are anomalized with
respect to their own mean over the period 1961–1990. Data: (a–
c) Station-based fromftp://ftp.ncdc.noaa.gov/pub/data/
ghcn/v3. (d–l) 5° grid based from Hadley Centre, Met Office,
UK http://www.metoffice.gov.uk/hadobs/ or from Climatic
Research Unit, Universityof
www.cru.uea.ac.uk/cru/data/. Units: °C. Shift years: (a) 1985,
1997; (b) 1985, 1997; (c) 1956, 1971, 1976, 1986, 1987; (d) 1976,
1978, 1979, 1993, 1994, 1996; (e) 1979, 1986, 1993, 1994, 1996,
2004; (f) 1976, 1996; (g) 1986, 1996; (h) 1970, 1986, 1989, 1996,
2002; (i) 1968, 1976, 1986, 1996; (j) 1978, 1989, 1994, 1996; (k)
1986, 1996; (l) 1976, 1996.
EastAnglia, UK http://
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
10 P. C. REID et al.
Page 11
but in our results is evident only in the Southern Hemi-
sphere. By contrast, the 1980s and 1990s regime shifts
are seen in both hemispheres, although the 1980s event
is stronger in the Northern Hemisphere.
The atmosphere underwent a major transformation
around 1988 (Lo & Hsu, 2010; Xiao et al., 2012), with
changes in temperature, meridional and zonal wind
patterns, and pressure (Fig. 2a–f, j, k, o, p). Changes in
zonal winds, tropical storms and arctic sea level pres-
sure during the 1980s regime shift (Fig. 2d, f, k) and a
subsequent reversal, in part associated with the 1996
regime shift (Xiao et al., 2012), reflect a pattern that can
also be seen in the dominant general circulation modes
in the Northern Hemisphere (Fig. 2g–i). However, the
1980s regime shift appears unique in that it does not
show a relationship with El Ni~ no (Yasunaka &
Hanawa, 2003).
Many step changes in the 1980s are apparent in
cryospheric records (Fig. 2q–v). In West Antarctica and
the Arctic, an approximately synchronous regime shift
in sea-ice extent occurred in the 1980s, with modelled
sea-ice volume in the Arctic declining linearly by 21%
from 1989 to 2012 (Fig. 2u). In marine systems the
1980s regime shift is well documented (M€ ollmann &
Diekmann, 2012) (Fig. 2ah-an) and coincided with the
Eastern Mediterranean Transient in 1987, which initi-
ated profound hydrographic changes that appear to be
unique in the last hundred years (Roether et al., 2014).
Measured since 1982, satellite observations of vegeta-
tion greenness/plant biomass (Normalized Difference
Vegetation Index, NDVI) reveal a step increase in plant
growth in the Northern Hemisphere in 1987/88
(Fig. 2x). This index, derived from Advanced Very
High Resolution Radiometer (AVHRR) satellite ima-
gery, is a measure of the photosynthetically active radi-
ation absorbed by chlorophyll in the leaves of plants.
Other satellite- and ground-based observations of the
start and length (Fig. 2y, z), but not the end (Fig. 2aa),
Fig. 6 Global map with magnified insets: locations of time series plotted in Figs 2 and 4. The latitudinal extent of data averaged on a
global or zonal basis is shown as arrows on the right (the letters above the arrows refer to the time series in Fig. 2). The time series are
grouped into six system categories (symbols, top right); the terrestrial biosphere group also includes one freshwater biological category
(Fig. 2ao). Individual sites are denoted by solid coloured symbols and regions by hollowed coloured symbols. The colours represent
the regime shift year in each time series (see key, bottom right). White lettering with a black border (e.g. Africa) indicates time series
with no significant regime shifts in the 1980s. The areas covered by the six regions averaged to produce ‘Global’ tropical storm days are
labelled A to F (coordinates in Table S4). A list of the system allocation and shift years of each of the time series from Fig. 2 is given in
Table S5; for oceans and continents see Fig. 4.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
11
Page 12
Table 2
Results of the multiple sequential t-test analyses of regime shifts (STARS) in Figs 2–5 for the ‘period of interest’ (1983–1990). Introductory information is listed in col-
umns 1–5. This is followed by two sets of results in sequence (columns 6a–13a and 6b–13b) as some of the 72 time series have two shift years in the ‘period of interest’. Introduc-
tory text: (1) brief title; (2) temperature/system category (T temperature, At atmosphere, TB terrestrial biosphere, MB marine biosphere, CO2carbon cycle, CR cryosphere, TH
terrestrial hydrosphere, MH marine hydrosphere); (3) whether the data are anomalies A or absolute values D; (4) the start year; and (5) the end year of the time series; (6a,b) the
‘shift year’ in bold to highlight the colours that are the same as in the legend for Fig. 2; (7a,b) the ‘longest test-period’ [years] which gives a significant result; (8a,b) the strength
of the change as ‘% significant results’, bold because it is a measure of the strength of the shift (A percentage is used here as it is not possible to include all the 20 default test-per-
iods in the calculation for short time series.); (9a,b) the ‘difference’ (in the units of each time series) as the size of the change between the means of the ‘longest test-period’ to
either side of the shift; (10a,b) the difference between the two means as a ‘% change’ where possible (not for temperature, anomalies, indices, day of year, pH); (11a,b) the ‘P-
value’ of the shift; (12a,b) the ‘trend old regime’ over the ‘longest test-period’ prior to the shift; and (13a,b) the ‘trend new regime’ over the ‘longest test-period’ after the shift
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
12 P. C. REID et al.
Page 13
Table 2
(continued)
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
13
Page 14
Table 2
(continued)
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
14 P. C. REID et al.
Page 15
Table 2
(continued)
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
15
Page 16
of the vegetation growing season and the cumulative
temperatures over this period (Fig. 2o) show that the
increase in productivity has closely tracked the rate of
warming in the northern extratropics (Barichivich et al.,
2013) and associated reductions in snow cover (Fig. 2r,
s). This evidence is reinforced by additional extensive
in situ measurements of changes in vegetation composi-
tion and biomass (Sturm et al., 2001; Mann et al., 2012;
Brandt
et al.,2013) and
changes in terrestrial and freshwater systems (Fig. 2ab-
ag, ao-aq).
Annual emissions of CO2from fossil fuel and land use
sources either accumulate in the atmosphere (atmo-
spheric growth rate) or are removed from the atmo-
sphere and absorbed by the land and oceans (sinks). The
net land uptake is calculated from emission data, minus
the atmospheric CO2growth rate and minus modelled
ocean uptake. A rapid increase in photosynthetic activity
in the late 1980s is consistent with the step decrease in
summer atmospheric CO2concentration in northern lati-
tudes and the simultaneous increase in winter atmo-
spheric CO2 (Fig. 2l, m) (Barichivich et al., 2013), the
latter a likely consequence of increased ecosystem respi-
ration (Barichivich et al., 2012). This linkage is confirmed
by carbon cycle observations and models (Sarmiento
et al., 2010) that show a sudden increase of ~1 Pg C yr?1
in net uptake by the land around 1988 (Fig. 2n). The
regime shift coincided with a sudden decline in the
annual growth rate of atmospheric CO2(Beaulieu et al.,
2012b). This large increase in the net land carbon sink is
evident in the global carbon budget despite increased
carbon emissions from anthropogenic sources and from
greater fire activity (Fig. 2w).
The PCA and the change-point analysis results sug-
gest that the main shift took place in 1987. The 72 indi-
vidual time series show shifts within the ‘period of
interest’ in the proportions 1983 (0), 1984 (2), 1985 (8),
1986 (12), 1987 (23), 1988 (17), 1989 (4), 1990 (0) = 66,
with three shifts in 1991 that are likely to be lagged
effects. Regionally, the results are generally closely
grouped and show a degree of consistency with regard
to the year of their occurrence. For example, in North
America: wildfires, atmospheric summer CO2concen-
tration, the Washington Cherry and the mean LST all
showed a shift in 1985; and in Europe eight time series
from the cryosphere, fresh water and vegetation sys-
tems all showed a shift in 1987, with a shift in a ninth
series, the UK sand martin arrival in 1988. These regio-
nal differences are mapped in Fig. 6 by ocean and con-
tinent for SST and LST (Fig. 4) with superimposed shift
years for some of the time series plotted in Fig. 2. With
the exception of dust storms in Asia (Fig. 2e), the signal
seems to have started in the early 1980s in South
America (1984), spreading to the North Pacific and
byothersynchronous
North America (1985), to the North Atlantic Ocean
(1986) and Europe (1987), and then on to Asia (1988),
with a possible, weaker, second signal in the North
Pacific in 1989 (see Fig. 4b and Table 2). In the Southern
Hemisphere, it seems to have extended eastwards to
the Indian Ocean (1986) and to Australia (1987).
Using CMIP5 and statistical modelling to examine
mechanisms
The global scale of the 1980s shift documented here
suggests that a fundamental shift in the climate system
took place at this time. It is clear that rapidly increasing
temperature is central to the shifts, with rising concen-
trations of greenhouse gases contributing to a net
warming of global climate since the late 1970s (Fig-
ure SPM 1, in IPCC SPM, 2013). Temperature is also
important for the carbon cycle as well as vice versa, as
reflected in the correlated changes in global combined
land and sea surface temperature and the atmospheric
growth rate of CO2over the last ~50 years (Fig. 7).
Over the period 1975 to 1995, the anthropogenic
warming from CMIP5 simulations of 0.19 °C per dec-
ade is close to the observed warming rate of 0.16 °C per
decade (Fig. 8a).Witha
(Fig. 8b), however, it does not explain the abrupt and
substantial temperature shift in the 1980s. Natural forc-
ing, in contrast, induced a marked peak in short-term
warming of 0.34 °C per decade in 1986 (in the 7-year
relativelysteady trend
1960 1970198019902000 2010
−1
0
1
2
0
0.5
CO2 growth rate [ppm yr−1]
Temperature [°C]
Fig. 7 Annual temporal development of CO2growth rate and
global temperature. Correlation: (r2= 57%, P < 10?10). Period:
1959–2014. Mauna Loa CO2annual growth rate (black) plotted
against HadCRUT4.3.0.0 annual global combined land and sea
surface temperature (red), both as anomalies to 1961–1990.
Mauna Loa data from: ftp://aftp.cmdl.noaa.gov/products/
trends/co2/co2_gr_mlo.txt and Hadley data from: http://
www.metoffice.gov.uk/hadobs/hadcrut4/data/current/down-
load.html.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
16 P. C. REID et al.
Page 17
trend belonging to the period 1983 to 1989, Fig. 8b),
temporarily exceeding the anthropogenic warming rate.
Therefore, anthropogenic and natural forcing factors
combined in the mid-1980s to produce a sudden accel-
eration in global warming.
Statistical modelling of historical temperatures (Fol-
land et al., 2013) indicates that this large natural forcing
has a volcanic origin rather than being linked to solar
radiation, the other possible natural forcing factor
(Fig. 8c,d). In the ‘period of interest’, the statistically
derived natural forcing has a similar pattern to that
from the CMIP5 model simulations but is somewhat
smaller in amplitude (Fig. 8b,d).
The major volcanic eruption of El Chich? on in 1982
was responsible for an estimated cooling of 0.2–0.3 °C
per decade, offsetting anthropogenic warming and
resulting in relatively small global mean temperature
trends in the early 1980s. By the mid-1980s to late
1980s, however, recovery from the climatic impacts of
the eruption, including a reduction in stratospheric
aerosol concentrations, led to a natural warming, rein-
forcing anthropogenic warming and producing a rapid
increase in global mean temperature on a higher level
than before the eruption (Fig. 7). As a result, the global
climate shifted to a warmer state in just a few years, set-
ting in motion a cascade of responses in natural sys-
tems.
Discussion
Based on evidence from a wide range of Earth system
components, we show that a global and approximately
synchronous shift occurred in the 1980s that is strongly
evident in global, hemispheric and some local tempera-
tures. Since the regime shift, decadal temperatures have
shown a steep increase compared to the previous per-
iod of little change and an earlier smaller rise from 1920
to 1950 (Fig. S4).
Temperature appears to be the main forcing factor
behind the shift: it is fundamental to most chemical,
physical and biological processes. Independent of our
study, there is substantial evidence that changes in the
heat structure of the world are profoundly affecting
regional climate variability, the cryosphere, terrestrial
systems, sea level, ocean hydrodynamics and the bio-
geochemistry, ecosystems and living resources of the
world. In a biological context, temperature modulates
all processes, including the physiology, reproduction,
development, occurrence, behaviour, disease and phe-
nology of organisms, at cellular to ecosystem scales.
The changes in temperature show great similarity
across the world; this commonality provides a plausible
explanation for the synchrony of the changes we
observed. The abrupt increase in temperature during
the regime shift may have initiated the intensification
of environmental impacts, for example storms, floods,
forest fires and the spread of pests seen over the last
few decades.
The increase in temperature of the 1980s has also
been linked to diverse biological changes on tropical
mountains of the New World. In a Costa Rican cloud
forest, declines of amphibian and reptile populations
and shifts in the altitudinal distribution of birds are
associated with a decrease in mist frequency and other
local climatic changes that appear to have crossed an
important biological threshold in 1987 (Pounds et al.,
1999). Extinction of harlequin frog species across Cen-
tral and South America, which, along with the disap-
pearance of the golden toad from Costa Rica, have been
associated with disease outbreaks and were the first
species-level extinctions in which global warming was
implicated, accelerated in the mid-1980s (Pounds et al.,
2006).
−0.2
0
0.2
(a)
−0.4
0
0.4
(b)
1980 1990
−0.1
0
(c)
19801990
−0.3
0
0.3 (d)
Fig. 8 Attribution of the global mean temperature shift in the
1980s. The ‘period of interest’ (1983–1990) is highlighted in
white. (a) Observed annual global mean temperature [°C] (blue
fine) with its centred running 3-year means (blue bold), and the
response [°C] to anthropogenic (red) and natural (green) forc-
ings from CMIP5 climate model simulations. (b) Centred run-
ning 7-year trends [°C per decade] corresponding to the
variables in (a). (c) Statistical reconstructions of the impact of
solar (orange) and volcanic (brown) forcings on global mean
temperature, plus their total (green). (d) Centred running 7-year
trends [°C per decade] corresponding to the variables in (c). All
data series are anomalized with respect to their own mean over
the period 1985–1988. In reading the figure, note that the RUN-
NING means as well as the RUNNING trends begin per defini-
tion to rise/decline before the El Chich? on eruption starts.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
17
Page 18
Table 3
Summary details of the time series presented in Fig. 2. Additional information in Table S3
a-b
°C Stratospheric air temperature measured by radiosondes launched at Payerne, Switzerland.
(a) at 20 hPa (~26 km above sea level), (b) at 500 hPa (~5 km above sea level)
Meridional wind speed at 500 hPa
Zonal wind speed at 500 hPa
Number of spring (March - May) dust storms per year, averaged for 48 observing stations in
northwest China
Number of tropical cyclone storm days per year (hurricanes plus tropical storms)
Seasonally varying Northern Hemisphere annular mode (SV-NAM) index
Arctic Oscillation (AO) index: the dominant pattern of winter (Nov-Apr) sea level pressure
variation north of 20°N
North Atlantic Oscillation (NAO, December - March) index: a basin-scale alternation of
atmospheric mass over the North Atlantic between high pressure in the subtropical Atlantic
and low pressure around Iceland
Combined land-surface air temperature and sea-surface water temperature for the zonal
band >64°N. Anomalies relative to 1951–1980
Sea-level pressure from the 20th Century Reanalysis version 2 data averaged over the Arctic
Interannually detrended atmospheric CO2concentration at Point Barrow, Alaska: (l) during the
warm (April - September), and (m) during the cold (October - March) seasons
Global net CO2land uptake
Time-integrated air temperature over the thermal growing season for the extratropical (>35°N)
Northern Hemisphere. Thermal growing season if temperature >5 °C
A reconstructed time series of air surface temperature for the Byrd meteorological station in
Western Antarctica
Sea-ice extent in the Bellingshausen/Amundsen Seas, Western Antarctica
Northern Hemisphere spring (March–April) snow extent
Number of snow days on the north side of the Swiss Alps between December and March
Maximum sea-ice extent in the Baltic Sea and Kattegat
Modelled total sea-ice volume in September for the Arctic Ocean
Northern Hemisphere (September) sea-ice extent
Mean number of burning days per wildfire event in forests of the western USA
Normalized Difference Vegetation Index (NDVI) averaged for the growing season in northern
latitudes (>35°N)
Thermal growing season (temperature >5 °C) for the extratropical (>35°N) Northern Hemisphere,
(y) start (day of the year), (z) length (days), (aa) end (day of the year)
The flowering date on which the Japanese cherry, Prunus jamasakura, comes into full bloom in
Kyoto, Japan. A plot of the time series starting in 1873 is shown in Fig. 3
Mean first arrival dates of the sand martin, Riparia riparia, averaged for eight locations in the UK
V? eraison (colour change and initial maturation) of M€ uller-Thurgau grapes harvested from
vineyards in Franconia, Germany
Winter flow (December–February) in the >1000 km long River Daugava measured at the
Daugavpils hydrological station in Latvia
River water temperature averaged for 18 hydrological stations that are representative of >80% of
the river outflow from Switzerland
River water pH averaged for 6 hydrological stations that are representative of >80% of the river
outflow from Switzerland
Phytoplankton biomass: a visual estimate of chlorophyll sampled at ~10 m depth by the
Continuous Plankton Recorder and averaged for the North Sea
Temperature of the North Sea averaged for the full depth within grid cells with centres enclosed
by 50°N-61°N and 3°W-9°E, anomalies relative to 1971–2000
Mean salinity at 50 m depth at the deepest station on the Torungen–Hirtshals hydrographic
section between Norway and Denmark in the North Sea, Skagerrak
Sea temperature at 50 m depth based on monthly measurements taken in the Japan Sea
Mean summer (July–September) volume transport of the Kuroshio Current in the Western
North Pacific
c
d
e
m s?1
m s?1
No. of days yr?1
f
g
h
No. of days yr?1
-
-
i-
j
°C
k
l-m
hPa
ppm
n
o
Pg C yr?1
°C
p
°C
q
r
s
t
u
v
w
x
106km2
106km2
Snow days
106km2
103km3
103km2
No. of days
–
y-aaDay of the year/days
ab Day of the year
ac
ad
Day of the year
Day of the year
aem3s?1
af
°C
ag
–
ahColour categories
ai
°C
aj
–
ak
al
°C
106m3s?1
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
18 P. C. REID et al.
Page 19
The numerous individual processes behind sudden
changes (regime shifts) in a temporal context as defined
in this study are still poorly understood and to our
knowledge have not been replicated in climate models
(Lo & Hsu, 2010). However, analyses using CMIP5
model scenarios and patterns of global temperature
and precipitation have been carried out and estimate
that pronounced geographical changes will occur in cli-
mate regimes and vegetation types by 2100 (Feng et al.,
2014). This study of vegetation types confirms the key
role of temperature, highlights the regional importance
of precipitation and emphasizes the large size and
potential impact of future change (a ~31 to 46% increase
in warmer and drier climate types by 2070–2100 based
on RCP4.5 and RCP8.5 scenarios).
The long and precipitous decline in the CO2growth
rate in the period after 1988 (Keeling et al., 1995; Fig. 7),
despite higher temperatures in 1989/90, coincides with
the step increase in the terrestrial carbon sink (Fig. 2n).
This decline in the CO2growth rate and the net increase
in the terrestrial sink from 1988 started well before the
Pinatubo eruption in 1991. The sink was mostly in
northern temperate boreal regions, as increases in the
tropics were counterbalanced by deforestation and
changes in land use (Sarmiento et al., 2010). Likely rea-
sons for the enhanced biospheric carbon sequestration
after the regime shift are increased photosynthesis from
CO2fertilization, an earlier and longer thermal growing
season in the Northern Hemisphere (Fig. 2o, y, z),
expansion of forests (IPCC Chap. 6, 2013) and the
increase in surface solar radiation known as global
brightening (Wild, 2009). To quantify approximately
the change in the carbon cycle, decadal means from
table 4 of Le Qu? er? e et al. (2013) were used to calculate
averages for the periods 1960 to 1989 and 1990 to 2009
of three variables: total anthropogenic emissions, atmo-
spheric growth rate and carbon uptake by the com-
bined land and ocean sinks (Table 4). The rate of
carbon uptake by the combined land and ocean sinks
(carbon storage) increased between these periods by
~1.65 Pg C yr?1, a 52% increase over the earlier period.
To place this regime shift in carbon storage in context,
it is close to double the rise in atmospheric growth rate
of CO2(0.92 Pg C yr?1) and equals 64% of the increase
in total anthropogenic emissions between the same
periods (2.57 Pg C yr?1) (Le Qu? er? e et al., 2013). If a sink
of this magnitude reversed to a CO2source, it would
markedly accelerate the rise in atmospheric CO2
growth rate and temperature.
The mechanisms behind the apparent easterly move-
ment seen in the timing of the 1980s regime and possi-
ble interhemispheric transfer of the year of the shift
(e.g. South to North America) around the world are not
understood. The pattern is best seen in the Northern
Hemisphere where it in part alternates between land
and ocean, starting in the Pacific in 1985 to Asia in 1988
(Fig. 4), reflecting the dominant flow of mid-latitude
winds and possibly the progressive movement of
waves along the eastward flowing polar and subtropi-
cal jet streams.
What is the trigger that initiated the sudden increase
in temperature from the 1980s until 2010 (Fig. S4)? Did
the Earth’s climate jump to a new ‘equilibrium’ state
due to its sensitivity to past radiation forcing (Hansen
amNo. of eggs m?2
Egg abundance per m2of the deep water (mesopelagic) fish Maurolicus japonicus sampled by
regular ichthyoplankton net surveys in May in the eastern Japan Sea. The eggs are found below
100 m and are used as an index of the biomass of the adult population
Tuna catch in Japanese waters of the Japan Sea
Timing of the spring algal bloom in Lake M€ uggelsee, Berlin, Germany
Mean groundwater temperature of two Swiss aquifers fed by riverbank filtration
Volume-weighted mean temperature of Lower Lake Zurich, Switzerland
The Southern Annular Mode (SAM) index is a zonal feature that reflects the main variability of
atmospheric circulation in the Southern Hemisphere extratropics and high latitudes
an
ao
ap
aq
ar
103tons
Calendar week
°C
°C
–
Table 4
Change in CO2-sink after the 1980s regime shift. Based on data from table 4 in (Le Qu? er? e et al. (2013)
*Fossil fuel emissions + cement production + land-use change. Text in red to emphasize the scale of the change.
Table 3 (continued)
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
19
Page 20
et al., 1984), or did its systems cross a threshold
(Barnosky et al., 2012) possibly linked to the rise in the
concentration of atmospheric CO2? Our results based
on CMIP5 and statistical modelling indicate that a com-
bination of warming from natural and anthropogenic
forcing was responsible as a rebound from the cooling
that followed the El Chich? on eruption together with a
reduction in stratospheric aerosols, although the rela-
tive contribution of each forcing factor is still unclear.
While the present generation of climate models can
reproduce the general features of global temperature
change and the cooling that follows major volcanic
eruptions (IPCC Chap. 9, 2013), up to now they have
not been able to replicate observed regime shifts (Lo &
Hsu, 2010) or simulate the dynamic response to exter-
nal forcings that includes the effects of volcanism (Dris-
coll et al., 2012).
If the above-combined anthropogenic and natural
forcing thesis is correct, why was there no equivalent
regime change after the larger Pinatubo eruption?
Changes in direct and diffuse radiation as measured at
Mauna Loa (Robock, 2000) were much greater for El
Chich? on in 1982 than for the eruption of Mount Pina-
tubo almost a decade later (Fig. S5). The recovery from
El Chich? on was also more rapid, and the aerosol plume
initially covered a greater area than for the following
Pinatubo eruption (to 30°N compared to 15°N) (Robock,
2000), which favours a stronger shift impulse from the
former. This fits with measurements that show that
much of the aerosol plume from Pinatubo moved
rapidly to the south of the equator after the eruption
whereas most of the El Chich? on plume remained in the
Northern Hemisphere (McCormick et al., 1995). The
seasonal timing of the eruptions (figure 7 in Post et al.,
1996) and differences in stratospheric wind direction
and strength, linked to the quasi-biennial oscillation
(McCormick et al., 1995), have also contributed.
A reversal in trend from global dimming to global
brightening, with stronger surface solar radiation from
the late 1980s, preceded and possibly masked some of
the effects of the Pinatubo eruption. The enhanced
radiation has been attributed to a substantial reduc-
tion in anthropogenic sulphur aerosols (Wild, 2009)
that occurred at this time (Stern, 2006). Parallel and
likely associated changes occurred in precipitation,
reflecting a more active global hydrological cycle
(Wild, 2009).
In the last decade, there has been an increasing
debate on the possible need to deliberately geoengi-
neer the climate to compensate for greenhouse gas
induced global warming by either removing CO2from
the atmosphere or by reducing solar irradiance (Shep-
herd et al., 2009). Our results have considerable rele-
vance to this debate. The speed, scale and global
extent of the changes that took place in the 1980s have
not been recognized until now, and therefore could
not have been taken into account by geoengineering
proposals. Current understanding that major volcanic
eruptions only cause short-term cooling of the Earth
(Cole-Dai, 2010) is contradicted by our demonstration
of a longer-term warming effect that involves the
interaction of major volcanism with global warming.
The cascading effects of the 1980s regime shift empha-
size the vulnerability of the Earth to large scale human
climate intervention.
The importance of the 1980s regime shift is shown
here to have been unparalleled within at least the last
century. It has been little recognized in the past due to
a paucity of long-term time series that are maintained
and sampled in the same consistent way over decades,
and a compartmentalization of science with insuffi-
cient communication between different disciplines.
Many factors related to atmospheric and oceanic circu-
lation, volcanism, latent and sensible heat transport,
cloudiness, aerosol effects, and shortwave and long-
wave radiation are likely to have been involved. Given
the scale and global extent of the shift, the public, pol-
icy makers and the scientific community need to be
made more aware of the importance of such events.
We need to improve our ability to forecast and model
the occurrence,magnitude
regime shifts and include their effects in risk assess-
ments for proposed geoengineering approaches to
modify the climate. The enormous impact of the
regime shift is seen especially in the land and ocean
carbon sinks; a key issue for humanity is how these
ecosystem services will behave in the future. The wide
range of changes associated with the 1980s regime shift
supports a threshold thesis that moved the whole glo-
bal system into a new, rapidly warming state, with
compounding consequences.
and consequencesof
Acknowledgements
We wish to thank all who contributed the time series data used
in the figures (See Table S3) and others for assistance or for data
not presented here, including Stephan Bader and Regula Gehrig
Bichsel, MeteoSwiss, Zurich, Switzerland; Adrian Jakob, Edith
Oosenbrug and Mich? ele Oberh€ ansli, BAFU Hydrology, Switzer-
land; and Reto Ruedy, NASA Goddard Institute for Space Stud-
ies, New York, USA. PG acknowledges support from Grant
14.B25.31.0026 of the Russian Ministry of Education and
Science. We are grateful to Eawag and its Department of Sys-
tems Analysis, Integrated Assessment and Modelling for host-
ing a workshop in Zurich, Switzerland.
References
Aono Y, Kazui K (2008) Phenological data series of cherry tree flowering in Kyoto,
Japan, and its application to reconstruction of springtime temperatures since the
9th century. International Journal of Climatology, 28, 905–914.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
20 P. C. REID et al.
Page 21
Barichivich J, Briffa KR, Osborn TJ, Melvin TM, Caesar J (2012) Thermal growing sea-
son and timing of biospheric carbon uptake across the Northern Hemisphere. Glo-
bal Biogeochemical Cycles, 26, GB4015.
Barichivich J, Briffa KR, Myneni R et al. (2013) Large-scale variations in the vegetation
growing season and annual cycle of atmospheric CO2at high northern latitudes
from 1950 to 2011. Global Change Biology, 19, 3167–3183.
Barnosky AD, Hadly EA, Bascompte J et al. (2012) Approaching a state shift in Earth’s
biosphere. Nature, 486, 52–58.
Beaugrand G (2004) The North Sea regime shift: evidence, causes, mechanisms and
consequences. Progress in Oceanography, 60, 245–262.
Beaugrand G, Reid PC, Iba~ nez F, Lindley JA, Edwards M (2002) Reorganisation of
North Atlantic marine copepod biodiversity and climate. Science, 296, 1692–
1694.
Beaugrand G, McQuatters-Gollop A, Edwards M, Goberville E (2013) Long-term
responses of North Atlantic calcifying plankton to climate change. Nature Climate
Change, 3, 263–267.
Beaugrand G, Harlay X, Edward M (2014) Detecting plankton shifts in the North Sea:
a new abrupt ecosystem shift between 1996 and 2003. Marine Ecology Progress Ser-
ies, 502, 85–104.
Beaugrand G, Conversi A, Chiba S et al. (2015) Synchronous marine pelagic regime
shifts in the Northern Hemisphere. Philosophical Transactions of the Royal Society of
London, Series B: Biological Sciences, 370, 20130272.
Beaulieu C, Chen J, Sarmiento JL (2012a) Change-point analysis as a tool to detect
abrupt climate variations. Philosophical Transactions of the Royal Society A: Mathemat-
ical, Physical and Engineering Sciences, 370, 1228–1249.
Beaulieu C, Sarmiento JL, Mikaloff Fletcher SE, Chen J, Medvigy D (2012b) Identifica-
tion and characterization of abrupt changes in the land uptake of carbon. Global
Biogeochemical Cycles, 26, GB1007.
Brandt JS, Haynes MA, Kuemmerle T, Waller DM, Radeloff VC (2013) Regime shift
on the roof of the world: alpine meadows converting to shrublands in the southern
Himalayas. Biological Conservation, 158, 116–127.
Brown RD, Robinson DA (2011) Northern Hemisphere spring snow cover variability
and change over 1922-2010 including an assessment of uncertainty. The Cryosphere,
5, 219–229.
Campbell JL, Allen MP (2001) Identifying shifts in policy regimes: cluster and inter-
rupted time-series analyses of U.S. income taxes. Social Science History, 25, 187–216.
Church JA, White NJ, Arblaster JM (2005) Significant decadal-scale impact of volcanic
eruptions on sea level and ocean heat content. Nature, 438, 74–77.
Cole-Dai J (2010) Volcanoes and climate. Wiley Interdisciplinary Reviews: Climate
Change, 1, 824–839.
Driscoll S, Bozzo A, Gray LJ, Robock A, Stenchikov G (2012) Coupled Model Inter-
comparison Project 5 (CMIP5) simulations of climate following volcanic eruptions.
Journal of Geophysical Research: Atmospheres, 117, D17105.
Feng S, Hu Q, Huang W, Ho C-H, Li R, Tang Z (2014) Projected climate regime shift
under future global warming from multi-model, multi-scenario CMIP5 simula-
tions. Global and Planetary Change, 112, 41–52.
Folland CK, Colman AW, Smith DM, Boucher O, Parker DE, Vernier J-P (2013) High
predictive skill of global surface temperature a year ahead. Geophysical Research
Letters, 40, 761–767.
Fujino T, Goto T, Shimura T et al. (2013) Decadal variation in egg abundance of a
mesopelagic fish, Maurolicus japonicus, in the Japan Sea during 1981-2005. Journal of
Marine Science and Technology, 21, 58–62.
Gong D-Y, Ho C-H (2002) Shift in the summer rainfall over the Yangtze River valley
in the late 1970s. Geophysical Research Letters, 29, 78-1–78-4.
Hansen J, Lacis A, Rind D et al. (1984) Climate sensitivity: analysis of feedback mech-
anisms. Geophysical Monograph, 29, 130–163.
Hare SR, Mantua NJ (2000) Empirical evidence for North Pacific regime shifts in 1977
and 1989. Progress in Oceanography, 47, 103–145.
Hari RE, Livingstone DM, Siber R, Burkhardt-Holm P, G€ uttinger H (2006) Conse-
quences of climatic change for water temperature and brown trout populations in
Alpine rivers and streams. Global Change Biology, 12, 10–26.
Hastings A, Wysham DB (2010) Regime shifts in ecological systems can occur with no
warning. Ecology Letters, 13, 464–472.
Iba~ nez F, Fromentin JM, Castel J (1993) Application de la m? ethode des sommes
cumul? ees ? a l’analyse des s? eries chronologiques en oc? eanographie. Comptes Rendus
de l’Acad? emie des Sciences de Paris, Sciences de la Vie, 316, 745–748.
IPCC Chap. 6, Ciais P, Sabine C et al. (2013) Carbon and other biogeochemical cycles.
In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to
the Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds
Stocker TF, Qin D, Plattner G-K et al.), pp. 465–570. Cambridge University Press,
Cambridge, UK and New York, NY, USA.
IPCC Chap. 7, Boucher O, Randall D et al. (2013) Clouds and aerosols. In: Climate
Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth
Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker TF,
Qin D, Plattner G-K et al.), pp. 571–658. Cambridge University Press, Cambridge,
UK and New York, NY, USA.
IPCC Chap. 9, Flato G, Marotzke J et al. (2013) Evaluation of climate models. In: Cli-
mate Change 2013: The Physical Science Basis. Contribution of Working Group I to the
Fifth Assessment Report of the Intergovernmental Panel on Climate Change (eds Stocker
TF, Qin D, Plattner G-K et al.), pp. 741–866. Cambridge University Press, Cam-
bridge, UK and New York, NY, USA.
IPCC SPM (2013) Summary for policymakers. In: Climate Change 2013: The Physical
Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Inter-
governmental Panel on Climate Change (eds Stocker TF, Qin D, Plattner G-K et al.), pp.
4. Cambridge University Press, Cambridge, UK and New York, NY, USA.
Jaffr? e M, Beaugrand G, Goberville E et al. (2013) Long-term phenological shifts in rap-
tor migration and climate. PLoS ONE, 8, e79112.
Jones GS, Stott PA, Christidis N (2013) Attribution of observed historical near-surface
temperature variations to anthropogenic and natural causes using CMIP5 simula-
tions. Journal of Geophysical Research: Atmospheres, 118, 4001–4024.
Keeling CD, Whorf TP, Wahlen M, van der Plicht J (1995) Interannual extremes in the
rate of rise of atmospheric carbon dioxide since 1980. Nature, 375, 666–670.
Kirby RR, Beaugrand G, Lindley JA (2009) Synergistic effects of climate and fishing in
a marine ecosystem. Ecosystems, 12, 548–561.
Le Qu? er? e C, Andres RJ, Boden T et al. (2013) The global carbon budget 1959–2011.
Earth System Science Data, 5, 165–185.
Litzow MA, Mueter FJ (2014) Assessing the ecological importance of climate regime
shifts: an approach from the North Pacific Ocean. Progress in Oceanography, 120,
110–119.
Lo T-T, Hsu H-H (2010) Change in the dominant decadal patterns and the late 1980s
abrupt warming in the extratropical Northern Hemisphere. Atmospheric Science
Letters, 11, 210–215.
Luczak C, Beaugrand G, Jaffr? e M, Lenoir S (2011) Climate change impact on Balearic
Shearwater through a trophic cascade. Biology Letters, 7, 702–705.
Mann DH, Scott Rupp T, Olson MA, Duffy PA (2012) Is Alaska’s boreal forest now
crossing a major ecological threshold? Arctic, Antarctic, and Alpine Research, 44,
319–331.
Mantua NJ (2004) Methods for detecting regime shifts in large marine ecosystems: a
review with approaches applied to North Pacific data. Progress in Oceanography,
60, 165–182.
McCormick MP, Thomason LW, Trepte CR (1995) Atmospheric effects of the Mt Pina-
tubo eruption. Nature, 373, 399–404.
M€ ollmann C, Diekmann R (2012) Chapter 4 - Marine ecosystem regime shifts induced
by climate and overfishing: a review for the Northern Hemisphere. Advances in
Ecological Research, 47, 303–347.
Morice CP, Kennedy JJ, Rayner NA, Jones PD (2012) Quantifying uncertainties in glo-
bal and regional temperature change using an ensemble of observational esti-
mates: the HadCRUT4 data set. Journal of Geophysical Research: Atmospheres, 117,
D08101.
Myneni RB, Keeling CD, Tucker CJ, Asrar G, Nemani RR (1997) Increased plant
growth in the northern high latitudes from 1981 to 1991. Nature, 386, 698–702.
Peterson WT, Schwing FB (2003) A new climate regime in northeast pacific ecosys-
tems. Geophysical Research Letters, 30, 1896.
Post MJ, Grund CJ, Weickmann AM, Healy KR, Willis RJ (1996) Comparison of
Mount Pinatubo and El Chich? on volcanic events: Lidar observations at 10.6 and
0.69 lm. Journal of Geophysical Research: Atmospheres, 101, 3929–3940.
Pounds JA, Fogden MPL, Campbell JH (1999) Biological response to climate change
on a tropical mountain. Nature, 398, 611–615.
Pounds JA, Bustamante MR, Coloma LA et al. (2006) Widespread amphibian
extinctions from epidemic disease driven by global warming. Nature, 439, 161–
167.
Reid PC, Beaugrand G (2012) Global synchrony of an accelerating rise in sea surface
temperature. Journal of the Marine Biological Association of the United Kingdom, 92,
1435–1450.
Reid PC, Borges M, Svendsen E (2001) A regime shift in the North Sea circa 1988
linked to changes in the North Sea horse mackerel fishery. Fisheries Research, 50,
163–171.
Robock A (2000) Volcanic eruptions and climate. Reviews of Geophysics, 38, 191–219.
Rodionov SN (2004) A sequential algorithm for testing climate regime shifts. Geophys-
ical Research Letters, 31, L09204.
Rodionov SN (2006) Use of prewhitening in climate regime shift detection. Geophysi-
cal Research Letters, 33, L12707.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
GLOBAL IMPACTS OF THE 1980S REGIME SHIFT
21
Page 22
Rodionov SN, Overland JE (2005) Application of a sequential regime shift detection
method to the Bering Sea ecosystem. ICES Journal of Marine Science, 62, 328–332.
Roether W, Klein B, Hainbucher D (2014) The Eastern Mediterranean Transient: evi-
dence for similar events previously? In: The Mediterranean Sea: Temporal Variability
and Spatial Patterns (eds Borzelli GLE, Gacic M, Lionello P, Malanotte-Rizzoli P),
chap. 6, pp. 75–83, John Wiley & Sons, Inc., Oxford.
Sarmiento JL, Gloor M, Gruber N et al. (2010) Trends and regional distributions of
land and ocean carbon sinks. Biogeosciences, 7, 2351–2367.
Shepherd J, Caldeira K, Cox P et al. (2009) Geoengineering the Climate. The Royal Soci-
ety, London. pp. i–ix, 1–82.
Stern DI (2006) Reversal of the trend in global anthropogenic sulfur emissions. Global
Environmental Change, 16, 207–220.
Sturm M, Racine C, Tape K (2001) Increasing shrub abundance in the Arctic. Nature,
411, 546–547.
Tao H, Borth H, Fraedrich K, Schneidereit A, Zhu X (2015) Hydrological extremes in
the Aksu-Tarim River Basin. Part I: climatology and regime shift. Climate Dynam-
ics. doi: 10.1007/s00382-015-2649-3.
Taylor WA (2000) Change-point analysis: a powerful new tool for detecting changes.
Available at: http://www.variation.com/cpa/tech/changepoint.html (accessed 4
February 2014).
Varotsos C, Franzke CE, Efstathiou M, Degermendzhi A (2013) Evidence for two
abrupt warming events of SST in the last century. Theoretical and Applied Climatol-
ogy, 114, 365–373.
Vezzulli L, Brettar I, Pezzati E, Reid PC, Colwell RR, Hofle MG, Pruzzo C (2012)
Long-term effects of ocean warming on the prokaryotic community: evidence from
the vibrios. The ISME Journal, 6, 21–30.
Wild M (2009) Global dimming and brightening: a review. Journal of Geophysical
Research: Atmospheres, 114, D00D16.
Xiao D, Li J, Zhao P (2012) Four-dimensional structures and physical process of the
decadal abrupt changes of the northern extratropical ocean–atmosphere system in
the 1980s. International Journal of Climatology, 32, 983–994.
Yasunaka S, Hanawa K (2002) Regime shifts found in the Northern Hemisphere SST
field. Journal of the Meteorological Society of Japan, 80, 119–135.
Yasunaka S, Hanawa K (2003) Regime shifts in the northern hemisphere SST field:
revisited in relation to tropical variations. Journal of the Meteorological Society of
Japan, 81, 415–424.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Figure S1. Autocorrelograms (a) for observed and (b) for
simulated time series.
Figure S2. Comparison of multiple STARS on real and artifi-
cial time series: shift years.
Figure S3. Comparison of multiple STARS on real and artifi-
cial time series: shift strength.
Figure S4. Replotted figure SPM 1a from IPCC Summary for
Policy Makers (2013).
Figure S5. Effects of the El Chich? on and Pinatubo eruptions
on radiation, redrawn from Robock (2000).
Table S1. Excel Database that includes all the annual values
of the 72 analysed time series.
Table S2. Supporting source, background and methodology
citations.
Table S3. Additional information and notes on the time ser-
ies presented in Fig. 2.
Table S4. Coordinates for six regions of tropical hurricanes/
storms in Fig. 6.
Table S5. Shift year of the time series from Fig. 2 included
in the regions of Fig. 6.
A compendium of all the supplementary figures and Tables
2–5 with their legends plus additional references for Table
S2.
© 2015 The Authors. Global Change Biology Published by John Wiley & Sons Ltd., doi: 10.1111/gcb.13106
22 P. C. REID et al.
Page 23
1
Supporting Information
Fig. S1. Autocorrelograms (a) for observed and (b) for simulated time series: 70 highly
autocorrelated time series (green) and 70 medium autocorrelated time series (blue) that
correspond respectively to the maximum and medium autocorrelation observed in the real
time series.
Page 24
2
Fig. S2. Comparison of multiple STARS on real and artificial time series: shift years. Data
plotted for 72 real time series (red), and 70 artificial time series (with medium – blue, and
high autocorrelation - green) for the period from 1946-2012. The ‘period of interest’ (1983-
1990) is marked by vertical lines. (a) Number of significant t-tests per year (p ≤ 0.05). (b)
Number of significant shift years per year (p ≤ 0.05).
Page 25
3
Fig. S3. Comparison of multiple STARS on real and artificial time series: shift strength. Real
(a,d) and artificial (b,c,e,f) time series. Data and period analysed as in Supporting
Information Fig. S2. (a-c) The length of the longest significant test-period for each shift year
out of the 20 test-periods applied is between 6 to 25 years (y-axis). (d-f) The percentage of
the number of significant t-tests (p ≤ 0.05) against the total number of possible t-tests for each
shift year (y-axis). Results below 20% discarded.
Page 26
4
Fig. S4. Replotted figure SPM 1a from the IPCC Summary for Policy Makers (2013). Global
temperature changes per decade: (a) Decadal global means of combined land and sea surface
temperatures (HadCrut4.3.0.0) as anomalies relative to 1961-1990 for the period 1850 to
2009. Recalculated and re-plotted after figure SPM 1a in the IPCC Summary for Policy
Makers (IPCC SPM, 2013); data courtesy Colin P. Mollis, Hadley Centre, Exeter, UK. (b)
The change per decade.
Page 27
5
Fig. S5. Effects of the El Chichón and Pinatubo eruptions on radiation [W m-2].
Direct and diffuse broadband radiation measurements from the Mauna Loa Observatory
against years. Redrawn from (Robock, 2000) with superimposed dashed lines for both
volcanic eruptions showing the extent and its value as a number of (a) the reduction of direct
(red) and (b) the enhancement of diffuse (blue) radiation immediately after the eruptions of
El Chichón at the end of March 1982 and Pinatubo in June 1991 (dashed lines). The lower
numbers for Pinatubo are 78% (direct) and 72% (diffuse) of the change seen after the earlier
eruption. The data were obtained with a tracking pyrheliometer and shade disk pyranometer
on mornings with clear skies at a solar zenith angle of 60°, equivalent to two relative air
masses. Data from NOAA/ESRL/GMD/GRAD Radiation Archive:
ftp://aftp.cmdl.noaa.gov/data/radiation/baseline/ .
Page 28
6
Table S1. Supporting Information Database
Excel file containing the data for all 72 time series analysed .
24. November 2015:
http://onlinelibrary.wiley.com/doi/10.1111/gcb.13106/epdf > open below menue
"Supplements" > right mouse click on Table S1: Save target as …….
Page 29
7
Table S2. Supporting source, background and methodology citations. The table addresses the
time series presented in Figs. 2-5.
Graph title Supporting citations Graph title Supporting citations
Fig. 2
a
Swiss ~26 km stratospheric
temperature
Swiss ~5 km tropospheric
temperature
Meridional wind speed 60-
75°N ~5 km a.s.l.
Zonal wind speed 60-75°N
~5 km a.s.l.
China spring dust storm
frequency
Global tropical
hurricane/storm days
NH seasonally varying
Index (SV-NAM)
Arctic Oscillation Index
(AO)
North Atlantic Oscillation
(NAO DJFM)
Arctic temperature (SST
and LST)
Arctic sea level pressure
Brocard et al., 2013
w
Western USA Wildfire
duration (days)
NH satellite vegetation
Westerling et al., 2006
b
Brocard et al., 2013
x
Myneni et al., 1997
c
Xiao et al., 2012
y
NH start of thermal
growing season
NH length of thermal
growing season
NH end of thermal growing
season
Japan Kyoto cherry
blossom
UK sand martin arrival
Barichivich et al., 2013;
Barichivich et al., 2012
Barichivich et al., 2013
d
Xiao et al., 2012
z
e
Ding et al., 2005
aa
Barichivich et al., 2013;
Barichivich et al., 2012
Aono and Kazui, 2008
f
Webster et al., 2005
ab
g
Ogi et al., 2004
ac
Sparks and Tryjanowski,
2007
Bock et al., 2011
h
Thompson and Wallace,
1998
Hurrell, 1995
ad
Germany grape vine
ripening date
Baltic river Daugava winter
flow
Switzerland river
temperature
Switzerland river pH
i
ae
Klavins et al., 2009
j
Hansen et al., 2006
af
Hari et al., 2006;
Jakob et al., 2002
Jakob et al., 2002;
Sigg and Stumm, 2011
Reid et al., 1998;
Raitsos et al., 2014
Ingleby and Huddleston,
2007
Danielssen et al., 1996
k
Compo et al., 2011
ag
l
Alaska atmospheric CO2,
Apr-Sep
Alaska atmospheric CO2,
Oct-Mar
Global CO2 net land uptake Beaulieu et al., 2012;
Barichivich et al., 2012;
Thoning et al., 1989
Barichivich et al., 2012;
Thoning et al., 1989
ah
North Sea phytoplankton
biomass
North Sea temperature
m
ai
n
Sarmiento et al., 2010
Barichivich et al., 2013
aj
North Sea Skagerrak 50 m
depth salinity
Japan Sea temperature at 50
m depth
North Pacific Kuroshio
current flow
Japan Sea deep living fish
(egg numbers)
Japan Sea tuna catch
Germany lake algal spring
bloom
Switzerland groundwater
temperature
Switzerland Lake Zürich
temperature
SH annular Mode Index
(SAM)
o
NH time-integrated temp.
of thermal growing season
Western Antarctica
temperature (LST)
Western Antarctica sea-ice
extent
NH spring snow extent
Switzerland snow days,
Dec-Mar
Baltic Sea sea-ice extent
ak
Tian et al., 2008
p
Bromwich et al., 2013
al
Japan Meteorological
Agency 2006
Fujino et al., 2013
q
Parkinson and Cavalieri,
2008
Brown and Robinson, 2011
Marty 2008
am
r
s
an
ao
Tian et al., 2008
Gerten and Adrian, 2000
t
Axell and Lindquist, 2005
ap
Figura et al., 2011
u
Arctic sea-ice volume (Sep) Lindsay et al., 2009
aq
North et al., 2013
v
NH sea-ice extent (Sep) Rayner et al., 2003
ar
Marshall, 2003
Fig. 3
a
Japan Kyoto cherry
blossom blooming
Switzerland Liestal cherry
blossom
USA Washington D.C.
cherry blossom
Fig. 4
Continents GHCN v3
Oceans HadSST3
Fig. 5
a-c
GHCN v3
d-f
CRUTEM4
Aono and Kazui, 2008
b
Defila and Clot, 2001
c
Chung et al., 2011
Lawrimore et al., 2011
Kennedy et al., 2011
Lawrimore et al., 2011
Jones et al., 2012
g-i
j-l
HadSST3
HadCRUT4
Kennedy et al., 2011
Morice et al., 2012
Page 30
8
Table S3. Additional information and notes on the time series presented in Fig. 2.
Dimensionless time series (i. e. NDVI, salinity, pH and indices) have no units; this is
indicated with a dash ‘-’. Order as per Fig. 2: atmosphere, cryosphere, terrestrial, hydrophere
(ocean and freshwater).
ATMOSPHERE
a Swiss ~26 km stratospheric temperature
b Swiss ~5 km tropospheric temperature
1,2 René Stübi, from Swiss Federal Office of Meteorology and Climatology (MeteoSwiss), Payerne, Switzerland
c Meridional wind speed 60-75°N ~5 km a.s.l. 360° circumglobal band 60°N-
75°N
d Zonal wind speed 60-75°N ~5 km a.s.l.
360° circumglobal band 60°N-
75°N
1,2 Dong Xiao, from Chinese Academy of Meteorological Sciences, Beijing, China
e China spring dust storm frequency
35°N-48°N, 75°E-102°E
1 Ruiqiang Ding
Time series title
1 Processed by / Data Originator / Source
2 Organisation
3 Notes
Latitude / Longitude Time period Units: Shift years
46.82°N, 6.95°E
46.82°N, 6.95°E
1976-2011
1976-2011
°C
°C
1986
1988
1948-2010
m s-1 1956, 1988
1948-2010
m s-1 1988, 1997
1960-2011
No. of days
of dust
storms
1984
2 State Key Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics (LASG), Beijing,
China
Global tropical hurricane/storm days
See Supp. Info. table S4
1 Peter J. Webster, Violeta E. Toma & Hai-Ru Chang
2 Georgia Institute of Technology, Atlanta, USA
3 Global total derived from six-hourly reports summed from six ocean basins.
NH SV−NAM Index
> 40°N
1 http://wwwoa.ees.hokudai.ac.jp/people/yamazaki/SV-NAM/index.html
2 Hokkaido University, Japan
3 The index is the standardised score of the leading empirical orthogonal function (EOF) of the monthly and zonally averaged
geopotential height fields poleward of 40°N from 1000 hPa to 200 hPa.
Arctic Oscillation Index (AO)
> 20°N
1 http://www.esrl.noaa.gov/psd/data/climateindices/list/#AO
2 NOAA Earth System Research Laboratory, Boulder, USA
3 Determined as the leading orthogonal empirical function at the height of the 1000 hPa surface (approximately sea level
pressure) and plotted as anomalies relative to 1979-2000.
North Atlantic Oscillation (NAO, DJFM)
38.71-65.07°N, 9.14-22.72°W
1 https://climatedataguide.ucar.edu/sites/default/files/climate_index_files/nao_station_djfm.ascii
2 Climate Analysis Section, NCAR, Boulder, USA
3 The winter (December-March) NAO index used here is the normalised sea level pressure (SLP) between Lisbon, Portugal and
Stykkisholmur/Reykjavik as anomalies relative to 1864-1983.
Arctic temperature (SST and LST)
> ~64°N
1 GISS Land-Ocean Temperature Index (LOTI), http://data.giss.nasa.gov/gistemp/tabledata_v3/ZonAnn.Ts+dSST.txt
2 NASA Goddard Institute for Space Studies (GISS), New York, USA
Arctic sea level pressure
> 70°N
1 http://www.esrl.noaa.gov/psd/data/gridded/data.20thC_ReanV2.monolevel.mm.html
2 NOAA Earth System Research Laboratory, Boulder, USA
Alaska atmospheric CO2, Apr−Sep
71.32°N, 156.61°W
Alaska atmospheric CO2, Oct−Mar
71.32°N, 156.61°W
1 Jonathan Barichivich and Renata E. Hari; ftp://ftp.cmdl.noaa.gov/ccg/co2/GLOBALVIEW
2 Cooperative Atmospheric Data Integration Project – CarbonDioxide, Boulder, USA
Global CO2 net land uptake
90°S-90°N
1,2 Claudie Beaulieu, from Ocean and Earth Science, University of Southampton, UK
3 The net land uptake (NLU) is calculated as a difference between annually reported global fossil fuel emissions, the growth rate
of atmospheric CO2 as a mean of measurements at Mauna Loa and the South Pole and the ocean uptake as a mean calculated
from four ocean biogeochemical models. The NLU, as opposed to the land uptake, does not require specification of land use
sources, which have remained approximately constant from 1959 to 2006.
NH integrated temperature growing season > 35°N
1,2 Jonathan Barichivich, from University of East Anglia, UK
3 The thermal growing season is defined as the period of the year with daily mean air temperatures > 5°C.
Plotted as anomalies relative to 1961-1990.
Western Antarctica temperature (LST)
80°S, 120°W
1 David H. Bromwich, http://polarmet.osu.edu/Byrd_recon/
2 The Ohio State University, Columbus, USA
f
g
1970-2004
No. of days
per year
1988
1948-2009
- 1987, 1994
h
1950-2012
- 1987, 1988, 1994
i
1946-2011
hPa 1961, 1971, 1972,
1995
j
k
l
m
n
1946-2012
°C 1987, 1999, 2004
1946-2010
hPa 1987, 1995
1972-2010
1972-2010
ppm
ppm
1985, 1988
1986, 1988
1966-2003
Pg C year-1 1988
o
1950-2011
°C 1987, 2004
p
1957-2012
°C 1986
Page 31
9
CRYOSPHERE
q Western Antarctica sea−ice extent
1 Claire L. Parkinson
2 NASA Goddard Space Flight Center, Greenbelt, USA
r NH spring snow extent
1 Ross Brown, see also: http://www.the-cryosphere.net/5/219/2011/tc-5-219-2011-supplement.zip
2 Climate Research Division, Environment Canada, Montreal, Canada
s Swiss snow days, Dec−Mar
1,2 Christoph Marty, from WSL Institute for Snow and Avalanche Research, Davos, Switzerland and the Swiss Federal Office of
Meteorology and Climatology (MeteoSwiss)
3 Mean of seven low-altitude stations (201–800 m). For this altitudinal band a snow day is when the snow depth exceeds a
threshold of 5 cm.
t Baltic Sea sea−ice extent
53.91°N-65.91°N, 9.43°E-30.3°E,
and Kattegat up to the tip of
Skagen at 57.75°N
1 Lars B. Axell and Karin Borenäs
2 Sveriges Meteorologiska och Hydrologiska Institut (SMHI), Norrköping and Gothenburg, Sweden
3 From 1957 to 2012 the time series is based on digitized hand-drawn ice charts and prior to 1957 on a least-squares method
applied to observations along the Swedish coast. The total area evaluated covers 420·103 km2 with the maximum ice extent due
to the methodology prior to 1957 equal to ~351·103 km2.
u Arctic sea−ice volume (Sep)
> 65°N
1 Ron Lindsay, from the PIOMAS ice-ocean coupled model
2 Polar Science Center, University of Washington
v NH sea−ice extent (Sep)
0-90°N
1 Ron Lindsay, from the Hadley Centre Global Sea Ice and Sea Surface Temperature (HadISST1)
gridded dataset http://www.metoffice.gov.uk/hadobs/hadisst/
2 Polar Science Center, University of Washington, Seattle, USA
TERRESTRIAL
w Western USA Wildfire duration (d)
31°N-49°N, 102°W-125°W
1 Anthony L. Westerling
2 University of California, Merced, USA
3 The time series is based on 1166 large (> 400 ha) forest wildfires.
x NH satellite vegetation
> 45°N
1 NDVI3g (third generation Global Inventory Modeling and Mapping System (GIMMS) NDVI
y NH start thermal growing season
> 35°N
z NH length thermal growing season
> 35°N
aa NH end thermal growing season
> 35°N
1 Jonathan Barichivich
2 University of East Anglia, UK
3 The thermal growing season is defined as the period of the year with daily mean air temperatures > 5°C. Plotted as anomalies
relative to 1961-1990.
ab Japan Kyoto cherry blossom
35°N, 136.67°E
1 Yasuyuki Aono, data updated and revised
2 Osaka Prefecture University, Japan
ac UK sand martin arrival
50.72°N-54.19°N, 2.58°W-1.75°E
1 Tim Sparks
2 Coventry University, UK
ad Germany grape vine ripening date
49.83° N, 9.87° E
1 Anna Bock
2 Technische Universität München, Freising, Germany
3 Harvested from the vineyards of the Landesanstalt für Weinbau und Gartenbau (the regional office for viticulture and
horticulture) Veitshöchheim, Franconia, Germany.
Time series title
1 Processed by / Data Originator / Source
2 Organisation
3 Notes
Latitude / Longitude Time periodUnits: Shift years
50°S-75°S, 60°W-130°W
or coast of Antarctica
1979-2010
106 km2 1987
0-90°N
1946-2010
106 km2 1953, 1987
45.82°N-47.81°N, 5.96°E-10.49°E 1946 to 2012
snow days 1987
1946-2012
106 km2 1987
1950-2012
103 km3 1960, 1980, 1988,
1992, 1994, 1997,
2004
1960-2011
103 km2 1978, 1989, 1998,
2001, 2004
1970-2003
No. of days
per fire
1985
1982-2010
- 1987, 1988, 1996
1950-2011 day of the yr
1950-2011
1950-2011 day of the yr
1972, 1988
1987, 2004
1993, 2002
days
1946-2012 day of the yr 1988
1950-2005 day of the yr 1976, 1980, 1988
1968-2010 day of the yr 1987, 1991
Page 32
10
HYDROSPHERE (OCEAN AND FRESHWATER)
ae Baltic river Daugava winter flow
1 Maris Klavins
2 University of Latvia, Riga, Latvia
3 Basin area: 64'500 km2.
af Swiss river temperature
1 Renata E. Hari
2 Swiss Federal Office for the Environment (BAFU), Hydrology Division
ag Swiss river pH
1 Renata E. Hari
2 Swiss Federal Office for the Environment (BAFU), Hydrology Division
3 Legislation to reduce phosphate inputs to lakes and rivers was introduced in Switzerland in 1986. This would have had the
opposite effect to the observed increase in pH: less phosphate → less algal growth → more CO2 →lower pH. It is more likely
that the higher pH reflects increased algal growth due to higher temperatures, more sunshine and a higher CO2 concentration or
increased weathering.
ah North Sea phytoplankton biomass
51°N-61°N, 3°W-10°E
1 Sir Alister Hardy Foundation for Ocean Science (SAHFOS).
2 SAHFOS, Plymouth, UK
3 Unit details: Four colour categories calibrated by acetone extracts and fluorescence
ai North Sea temperature
50°N-61°N and 3°W-9°E
1 Simon A. Good
2 UK Met Office Hadley Centre, Exeter; EN3: quality controlled subsurface ocean temperature and salinity dataset. See:
http://www.metoffice.gov.uk/hadobs/en3/
aj North Sea 50 m depth salinity
58.13°N, 9.18°E
1 Else Juul Green, http://ocean.ices.dk/HydChem/HydChem.aspx?plot=yes
2 International Council for the Exploration of the Seas, Copenhagen, Denmark
3 The sampling location is equivalent to the Norwegian station Z220, which is 20 miles from the Norwegian coast.
Measurements have been taken approximately once a month.
ak Japan Sea temperature at 50 m depth
33-38°N, 130-136°E
1 Yongjun Tian from the Japan Sea National Fisheries Research Institute.
2 Japan Sea National Fisheries Research Institute, Niigata, Japan.
3 Monthly measurements taken in the Japan Sea. to cover the path of the Tsushima Current between Wakasa Bay in Kyoto
Prefecture and Yamaguchi Prefecture, Japan and averaged for the area within 33-38°N, 130-136°E.
al North Pacific Kuroshio current flow
across 137°E between 3-34°N
1 Yongjun Tian from theJapan Meteorological Agency (JMA), Tokyo, Japan.
2 Japan Sea National Fisheries Research Institute, Niigata, Japan.
3 Current flow estimated from geostrophic calculations based on temperature and salinity profiles taken twice a year (summer =
Jul-Sep and winter = Jan-Mar) on a standard north to south section during research cruises of the Japan Meteorological Agency
(JMA).
am Japan Sea deep living fish (eggs)
between 34.45°N-41.17°N and
1 Yongjun Tian from theJapan Meteorological Agency (JMA), Tokyo, Japan.
2 Japan Sea National Fisheries Research Institute, Niigata, Japan.
3 Sampling stations were located within approximately 185 km of the coast of Japan. This abundant mesopelagic (1000-100 m
deep) species normally swims between 150 to 250 m during the daytime and migrates to shallower depths during the night.
an Japan Sea tuna catch
34°N-41.5°N, 131°E-141°E
1 Yongjun Tian from theJapan Meteorological Agency (JMA), Tokyo, Japan.
2 Japan Sea National Fisheries Research Institute, Niigata, Japan.
3 Mostly comprising warm-water bluefin, albacore, and yellowfin tuna.
ao Germany lake algal spring bloom
13.65°E, 52.43°N
1 Rita Adrian
2 Leibniz- Institute of Freshwater Ecology and Inland Fisheries, Berlin, Germany.
3 The timing refers to the calendar week of the year when maximum total phytoplankton biomass developed after ice-off.
ap Swiss groundwater temperature
Pump-stations: Kiesen 46.80°N,
7.57°E;Neuhausen 47.68°N,
8.61°E
2 Energie Wasser Bern and Städtische Werke Schaffhausen und Neuhausen am Rheinfall
aq Swiss Lake Zürich temperature
47.37°N-47.20°N, 8.53°E-8.82°E
1 Ryan P. North.
2 Oliver Köster from Wasserversorgung der Stadt Zürich
ar SH Annular Mode Index (SAM)
40°S and 65°S
1 http://www.nerc-bas.ac.uk/icd/gjma/sam.html
2 British Antarctic Survey, Cambridge, UK
3 The index used here is derived from a proxy zonal mean sea level pressure for 40°S and 65°S calculated from twelve
meteorological stations that approximate to each of these latitudes, anomalies relative to 1971-2000.
Time series title Latitude / Longitude Time period Units: Shift years
55.2°N-57.4°N, 24°E-28.2°E
45.82°N-47.81°N, 5.9°E-10.49°E
1946-2010
m3 s-1 1987
1978-2011
°C 1987
45.82°N-47.81°N, 5.9°E-10.49°E
1977-2010
- 1991
1946-2011
Colour
categories
1951, 1985
1950-2011
°C 1987, 2001
1965-2007
- 1988
1964-2008
°C 1987
1972-2011
1 Sv =
106 m3 s-1
1987
1981-2005
No. of eggs
per m2
1988
1964-2004
103 tons 1991
1980-2010
Calendar
week
1987
1970-2005
°C 1987
1 Simon Figura
1946-2005
°C 1987
1957-2012
- 1992, 1996
Page 33
11
Table S4. Coordinates for six regions of tropical hurricane/storms in Fig. 6. The six regions
(A-F, outlined in Fig. 6) are averaged to give a ‘global’ total for tropical hurricane/storm
days.
Ocean region
Minimum
latitude
5°N
5°N
5°N
5°S
5°N
5°S
Maximum
latitude
20°N
25°N
20°N
20°S
20°N
20°S
Minimum
longitude
90°W
20°W
55°E
50°E
120°E
155°E
Maximum
longitude
120°W
90°W
90°E
115°E
180°E
180°E
A Eastern North Pacific
B North Atlantic
C North Indian
D South Indian
E Western North Pacific
F Southwest Pacific
Page 34
12
Table S5. Shift year of the time series from Fig. 2 included in the regions of Fig. 6.
EUROPE INSET
2b Switzerland Payerne ~5 km tropospheric air temperature, atmosphere
2t Baltic Sea sea-ice extent, cryosphere
2ac UK sand martin arrival date, terrestrial biosphere
2ad Germany grape vine ripening date, terrestrial biosphere
2ae Baltic river Daugava winter flow, terrestrial hydrosphere
2af Switzerland river temperature, terrestrial hydrosphere
2ah North Sea phytoplankton biomass, marine biosphere
2ai North Sea temperature, marine hydrosphere
2ao Germany Lake Müggelsee algal bloom spring timing, terrestrial biosphere
JAPAN INSET
2ak Japan Sea temperature at 50m depth, marine hydrosphere
2al Western North Pacific Kuroshio current flow, marine hydrosphere
2ab Japan Kyoto cherry blossom blooming, terrestrial biosphere
GLOBAL MAP
North America
2m Alaska Point Barrow atmospheric CO2 concentration (Apr-Sep), atmosphere
2n Alaska Point Barrow atmospheric CO2 concentration (Oct-Mar), atmosphere
2w Western USA Wildfire duration (days), terrestrial biosphere
Asia
2e China dust storm frequency (March-May), atmosphere
Tropics
2f Global tropical hurricane/storm days (mean areas A-F), atmosphere
Antarctica
2p Western Antarctica air surface temperature Byrd station, atmosphere
2q Western Antarctica sea-ice extent, cryosphere
GLOBAL AND HEMISPHERIC TIME SERIES
Arrows left to right:
2n Global CO2 net land uptake, atmosphere
2r Northern Hemisphere spring snow extent, cryosphere
2x Northern Hemisphere vegetation from satellites, terrestrial biosphere
2j Arctic combined sea and air surface temperature, atmosphere
2k Arctic sea level pressure, atmosphere
2c Meridional wind speed 60-75°N (360°) ~5 km above sea level, atmosphere
2d Zonal wind speed 60-75°N (360°) ~5 km above sea level, atmosphere
1988
1987
1988
1987
1987
1987
1985
1987
1987
1987
1987
1988
1985
1986
1985
1984
1988
1986
1987
1988
1987
1987
1987
1987
1988
1988
Page 35
13
Additional References for Supporting Information table S2
Aono Y, Kazui K (2008) Phenological data series of cherry tree flowering in Kyoto, Japan,
and its application to reconstruction of springtime temperatures since the 9th century.
International Journal of Climatology, 28, 905-914.
Axell L, Lindquist K (2005) Reconstruction of Annual Maximum Ice Extent in the Baltic Sea
1660-2005. In: Fifth Workshop on BalticSea Ice Climate. (eds Schrum C, Schmelzer
N), pp. 79-80, Bundesamt für Seeschifffahrt und Hydrographie, Hamburg, Germany.
Barichivich J, Briffa KR, Myneni R et al. (2013) Large-scale variations in the vegetation
growing season and annual cycle of atmospheric CO2 at high northern latitudes from
1950 to 2011. Global Change Biology, 19, 3167-3183.
Barichivich J, Briffa KR, Osborn TJ, Melvin TM, Caesar J (2012) Thermal growing season
and timing of biospheric carbon uptake across the Northern Hemisphere. Global
Biogeochemical Cycles, 26, GB4015.
Beaulieu C, Sarmiento JL, Mikaloff Fletcher SE, Chen J, Medvigy D (2012) Identification
and characterization of abrupt changes in the land uptake of carbon. Global
Biogeochemical Cycles, 26, GB1007.
Bock A, Sparks T, Estrella N, Menzel A (2011) Changes in the phenology and composition
of wine from Franconia, Germany. Climate Research, 50, 69-81.
Brocard E, Jeannet P, Begert M, Levrat G, Philipona R, Romanens G, Scherrer SC (2013)
Upper air temperature trends above Switzerland 1959–2011. Journal of Geophysical
Research: Atmospheres, 118, 4303-4317.
Bromwich DH, Nicolas JP, Monaghan AJ, Lazzara MA, Keller LM, Weidner GA, Wilson
AB (2013) Central West Antarctica among the most rapidly warming regions on
Earth. Nature Geoscience, 6, 139-145.
Brown RD, Robinson DA (2011) Northern Hemisphere spring snow cover variability and
change over 1922-2010 including an assessment of uncertainty. The Cryosphere, 5,
219–229.
Chung U, Mack L, Yun JI, Kim S-H (2011) Predicting the timing of cherry blossoms in
Washington, DC and Mid-Atlantic states in response to climate change. PLoS One, 6,
e27439.
Compo GP, Whitaker JS, Sardeshmukh PD et al. (2011) The Twentieth Century Reanalysis
Project. Quarterly Journal of the Royal Meteorological Society, 137, 1-28.
Danielssen DS, Svendsen E, Ostrowski M (1996) Long-term hydrographic variation in the
Skagerrak based on the section Torungen–Hirtshals. ICES Journal of Marine Science,
53, 917–925.
Defila C, Clot B (2001) Phytophenological trends in Switzerland. International Journal of
Biometeorology, 45, 203-207.
Ding R, Li J, Wang S, Ren F (2005) Decadal change of the spring dust storm in northwest
China and the associated atmospheric circulation. Geophysical Research Letters, 32,
L02808.
Figura S, Livingstone DM, Hoehn E, Kipfer R (2011) Regime shift in groundwater
temperature triggered by the Arctic Oscillation. Geophysical Research Letters, 38,
L23401.
Fujino T, Goto T, Shimura T et al. (2013) Decadal variation in egg abundance of a
mesopelagic fish, Maurolicus japonicus, in the Japan Sea during 1981-2005. Journal
of Marine Science and Technology, 21, 58-62.
Page 36
14
Gerten D, Adrian R (2000) Climate-driven changes in spring plankton dynamics and the
sensitivity of shallow polymictic lakes to the North Atlantic Oscillation. Limnology
and Oceanography, 45, 1058-1066.
Hansen J, Sato M, Ruedy R, Lo K, Lea DW, Medina-Elizade M (2006) Global temperature
change. Proceedings of the National Academy of Sciences of the United States of
America, 103, 14288-14293.
Hari RE, Livingstone DM, Siber R, Burkhardt-Holm P, Güttinger H (2006) Consequences of
climatic change for water temperature and brown trout populations in Alpine rivers
and streams. Global Change Biology, 12, 10-26.
Hurrell JW (1995) Decadal trends in the North Atlantic Oscillation: regional temperatures
and precipitation. Science, 269, 676-679.
Ingleby B, Huddleston M (2007) Quality control of ocean temperature and salinity profiles —
Historical and real-time data. Journal of Marine Systems, 65, 158-175.
Jakob A, Binderheim-Bankay E, Davis JS (2002) National long-term surveillance of Swiss
rivers. Verhandlungen Internationale Vereinigung für theoretische und angewandte
Limnologie, 28, 1101-1106.
Japan Meteorological Agency (2006) The Medical Examination of the Sea: a Synthetic
Examination (First Edition, in Japanese), pp. 196.
Jones PD, Lister DH, Osborn TJ, Harpham C, Salmon M, Morice CP (2012) Hemispheric
and large-scale land-surface air temperature variations: An extensive revision and an
update to 2010. Journal of Geophysical Research: Atmospheres, 117, D05127.
Kennedy JJ, Rayner NA, Smith RO, Parker DE, Saunby M (2011) Reassessing biases and
other uncertainties in sea surface temperature observations measured in situ since
1850: 1. Measurement and sampling uncertainties. Journal of Geophysical Research:
Atmospheres, 116, D14103.
Klavins M, Briede A, Rodinov V (2009) Long term changes in ice and discharge regime of
rivers in the Baltic region in relation to climatic variability. Climatic Change, 95, 485-
498.
Lawrimore JH, Menne MJ, Gleason BE, Williams CN, Wuertz DB, Vose RS, Rennie J
(2011) An overview of the Global Historical Climatology Network monthly mean
temperature data set, version 3. Journal of Geophysical Research: Atmospheres, 116,
D19121.
Lindsay RW, Zhang J, Schweiger A, Steele M, Stern H (2009) Arctic sea ice retreat in 2007
follows thinning trend. Journal of Climate, 22, 165-176.
Marshall GJ (2003) Trends in the Southern Annular Mode from observations and reanalyses.
Journal of Climate, 16, 4134-4143.
Marty C (2008) Regime shift of snow days in Switzerland. Geophysical Research Letters, 35,
L12501.
Morice CP, Kennedy J.J, Rayner NA, Jones PD (2012) Quantifying uncertainties in global
and regional temperature change using an ensemble of observational estimates: The
HadCRUT4 dataset. Journal of Geophysical Research: Atmospheres, 117, D08101.
Myneni RB, Keeling CD, Tucker CJ, Asrar G, Nemani RR (1997) Increased plant growth in
the northern high latitudes from 1981 to 1991. Nature, 386, 698-702.
North RP, Livingstone DM, Hari RE, Köster O, Niederhauser P, Kipfer R (2013) The
physical impact of the late 1980s climate regime shift on Swiss lakes and rivers.
Inland Waters, 3, 341-350.
Ogi M, Yamazaki K, Tachibana Y (2004) The summertime annular mode in the Northern
Hemisphere and its linkage to the winter mode. Journal of Geophysical Research:
Atmospheres, 109, D20114.
Page 37
15
Parkinson CL, Cavalieri DJ (2008) Arctic sea ice variability and trends, 1979–2006. Journal
of Geophysical Research: Oceans, 113, C07003.
Raitsos DE, Pradhan Y, Lavender SJ, Hoteit I, Mcquatters-Gollop A, Reid PC, Richardson
AJ (2013) From silk to satellite: half a century of ocean colour anomalies in the
Northeast Atlantic. Global Change Biology, 20 2117-23.
Rayner NA, Parker DE, Horton EB et al. (2003) Global analyses of sea surface temperature,
sea ice, and night marine air temperature since the late nineteenth century. Journal of
Geophysical Research: Atmospheres, 108, 4407.
Reid PC, Edwards M, Hunt HG, Warner AJ (1998) Phytoplankton change in the North
Atlantic. Nature, 391, 546.
Sarmiento JL, Gloor M, Gruber N et al. (2010) Trends and regional distributions of land and
ocean carbon sinks. Biogeosciences, 7, 2351- 2367.
Sigg L, Stumm W (2011) Aquatische Chemie. Einführung in die Chemie natürlicher
Gewässer.Zürich, vdf Hochschulverlag AG an der ETH Zürich.
Sparks T, Tryjanowski P (2007) Patterns of spring arrival dates differ in two hirundines.
Climate Research, 35, 159-164.
Thompson DWJ, Wallace JM (1998) The Arctic Oscillation signature in the wintertime
geopotential height and temperature fields. Geophysical Research Letters, 25, 1297-
1300.
Thoning KW, Tans PP, Komhyr WD (1989) Atmospheric carbon dioxide at Mauna Loa
Observatory: 2. Analysis of the NOAA GMCC data, 1974–1985. Journal of
Geophysical Research: Atmospheres, 94, 8549-8565.
Tian Y, Kidokoro H, Watanabe T, Iguchi N (2008) The late 1980s regime shift in the
ecosystem of Tsushima warm current in the Japan/East Sea: evidence from historical
data and possible mechanisms. Progress in Oceanography, 77, 127-145.
Webster PJ, Holland GJ, Curry JA, Chang H-R (2005) Changes in tropical cyclone number,
duration, and intensity in a warming environment. Science, 309, 1844-1846.
Westerling AL, Hidalgo HG, Cayan DR, Chang H-R (2006) Warming and earlier spring
increase western U.S. forest wildfire activity. Science. 313, 940-943.
Xiao D, Li J, Zhao P (2012) Four-dimensional structures and physical process of the decadal
abrupt changes of the northern extratropical ocean–atmosphere system in the 1980s.
International Journal of Climatology, 32, 983-994.
Download full-text 
-for-observed-and-(b)-for-simulated-time-series.-Figure_small.png "Figure S1 . Autocorrelograms (a) for observed and (b) for simulated...")
.-Global_small.png "Fig. S4 . Replotted figure SPM 1a from the IPCC Summary for Policy...")
