What can Palaeoclimate Modelling do for you?

Abstract

In modern environmental and climate science it is necessary to assimilate observational datasets collected over decades with outputs from numerical models, to enable a full understanding of natural systems and their sensitivities. During the twentieth and twenty-first centuries, numerical modelling became central to many areas of science from the Bohr model of the atom to the Lorenz model of the atmosphere. In modern science, a great deal of time and effort is devoted to developing, evaluating, comparing and modifying numerical models that help us synthesise our understanding of complex natural systems. Here we provide an assessment of the contribution of past (palaeo) climate modelling to multidisciplinary science and to society by answering the following question: What can palaeoclimate modelling do for you? We provide an assessment of how palaeoclimate modelling can develop in the future to further enhance multidisciplinary research that aims to understand Earth’s evolution, and what this may tell us about the resilience of natural and social systems as we enter the Anthropocene.

Full text

Vol.:(0123456789)
1 3
Earth Systems and Environment
https://doi.org/10.1007/s41748-019-00093-1
ORIGINAL ARTICLE
What can Palaeoclimate Modelling doforyou?
A.M.Haywood1 · P.J.Valdes2· T.Aze1· N.Barlow1· A.Burke3· A.M.Dolan1· A.S.vonderHeydt4· D.J.Hill1·
S.S.R.Jamieson5· B.L.Otto‑Bliesner6· U.Salzmann7· E.Saupe8· J.Voss9
Received: 19 March 2019 / Accepted: 13 April 2019
© The Author(s) 2019
Abstract
In modern environmental and climate science it is necessary to assimilate observational datasets collected over decades with
outputs from numerical models, to enable a full understanding of natural systems and their sensitivities. During the twentieth
and twenty-first centuries, numerical modelling became central to many areas of science from the Bohr model of the atom
to the Lorenz model of the atmosphere. In modern science, a great deal of time and effort is devoted to developing, evalu-
ating, comparing and modifying numerical models that help us synthesise our understanding of complex natural systems.
Here we provide an assessment of the contribution of past (palaeo) climate modelling to multidisciplinary science and to
society by answering the following question: What can palaeoclimate modelling do for you? We provide an assessment of
how palaeoclimate modelling can develop in the future to further enhance multidisciplinary research that aims to understand
Earth’s evolution, and what this may tell us about the resilience of natural and social systems as we enter the Anthropocene.
Keywords Climate· Model· Palaeoclimate· Global change· Environmental change· Earth history
1 Introduction
Complex climate models, and latterly Earth System Models
(ESMs), are in the vanguard of attempts to assess the effects,
risks and potential impacts associated with the anthropo-
genic emission of greenhouse gases (GHG: IPCC 2013).
Climate predictions underpin scientific assessments of miti-
gation and societal adaptation pathways (IPCC 2013).
The use of models to understand the evolution of our
planet’s climate, environment and life (Fig.1), collectively
known as past (palaeo) climate modelling, has matured in
its capacity and capability since the first simulations using
a General Circulation Model (GCM) were published in the
1970s for the Last Glacial Maximum (e.g., Gates 1976).
Since then it has become apparent that to fully appreciate the
complex interactions between climate and the environment,
and to use this knowledge to address societal challenges, it
is necessary to adopt multidisciplinary scientific approaches
capable of robustly testing long-standing hypotheses that
describe the sensitivity/resilience of our planet and the life
forms that inhabit it. Multidisciplinary studies have provided
unique ways of evaluating the efficacy of climate and ESM
predictions in reproducing large-scale climate changes that
occurred in the past (Haywood etal. 2013), and this has
* A. M. Haywood
earamh@leeds.ac.uk
1 School ofEarth andEnvironment, University ofLeeds,
Woodhouse Lane, LeedsLS29JT, UK
2 School ofGeographical Sciences, University ofBristol,
University Road, BristolBS81SS, UK
3 Laboratoire d’Ecomorphologie et de Paleoanthropologie,
Universite de Montreal, Departement d’Anthropologie,
C.P. 6128, Centre-Ville, Montreal, QCH3C3J7, Canada
4 Department ofPhysics, Centre forComplex Systems Science,
Utrecht University, Princetonplein 5, 3584CCUtrecht,
TheNetherlands
5 Department ofGeography, Durham University, South Road,
DurhamDH13LE, UK
6 Climate andGlobal Dynamics Laboratory, National Center
forAtmospheric Research, 1850 Table Mesa Drive, Boulder,
CO80305, USA
7 Department ofGeography andEnvironmental Sciences,
Northumbria University, Newcastle City Campus, 2 Ellison
Place, NewcastleuponTyneNE18ST, UK
8 Department ofEarth Sciences, University ofOxford, South
Parks Road, OxfordOX13AN, UK
9 School ofMathematics, University ofLeeds, Woodhouse
Lane, LeedsLS29JT, UK

A.M.Haywood et al.
1 3
provided valuable out-of-sample tests for the tools used to
predict future climate and environmental change.
The march towards multidisciplinary assessment of past
climate and environmental states has accelerated through the
construction of models that have more complete representa-
tions of the Earth system at higher spatial resolution. From
relatively simple three-dimensional representations of the
atmosphere, models have developed to include representa-
tions of the oceans and land cover, and incorporate the inter-
actions between atmosphere, oceans, and the land and ice
sheets. They have developed to enable dynamic simulation
of the distribution of past vegetation cover, ice sheet distri-
bution and variability, and ocean/terrestrial biogeochemical
cycles (Prinn 2013). Each development has brought with
it opportunities to form new research collaborations with
observational-based scientists to test hypotheses for Earth
evolution in novel and exciting ways, and to relate this
knowledge towards addressing societal challenges.
Whilst some of the contributions made by palaeoclimate
modelling to wider research efforts are obvious, the util-
ity of, and access to, model simulations has grown to such
a degree that many of the connections between palaeocli-
mate modelling and other disciplines are not appreciated.
Unsurprisingly, the way in which palaeoclimate modelling
addresses societal needs, as generally expressed through
UN SDGs and scientific grand challenges, is not fully
appreciated either. Here we address this issue through the
exploration of palaeoclimate modelling’ s (using complex
numerical models) contribution to the better understanding
of climate sensitivity, data-model comparison and geological
proxy interpretation, life and its resiliency, glacial and sea-
level history, hydrology, anthropology and natural resource
exploration as well as energy-based research. We also dis-
cuss potential avenues for the future that have the capability
to enhance the contribution of palaeoclimate modelling to
other disciplines and to better address societal needs.
2 The Climate Sensitivity Grand Challenge
Studies of climate sensitivity quantify changes in global
mean temperature in response to variations in atmospheric
CO2 concentration. The concept of equilibrium climate
states has been crucial in this respect. Equilibrium Climate
Sensitivity (ECS) is the temperature difference in response
to a doubling of CO2, where the climate is assumed to be
in equilibrium before and after the CO2 perturbation (e.g.,
Von der Heydt etal. 2016). An important aim of quantifying
Fig. 1 Global annual mean temperature variation of the Earth through
time (last 400 million years) predicted by the Hadley Centre Coupled
Climate Model version 3 (HadCM3), compared with geologically
derived estimates of temperature variability over the same period [the
Royer etal. 2004 temperature record, the Zachos etal. 2008; Lisiecki
and Raymo 2005 benthic oxygen isotope stack, as well as the EPICA
and NGRIP ice core records; Jouzel et al. 2007 and NGRIP Mem-
bers 2004. Geological epochs include the Devonian (D), Carbon-
iferous (C) Permian (P), Triassic (Tr), Jurassic (J) Cretaceous (K),
Eocene (Eoc), Oligocene (Oli.), Miocene (Mio), Pliocene and Pleis-
tocene (Pleist.)] Future predictions of temperature change are based
on HadCM3 simulations using different Representative Concentration
Pathways (RCPs). Horizontal blue lines represent geological evidence
for ice sheets in the northern (NH) and southern (SH) hemispheres.
Major evolutionary characteristics and events over the last 400 mil-
lion years represented by cartoon silhouettes

What can Palaeoclimate Modelling doforyou?
1 3
ECS has always been to predict future climate change, where
ECS plays a role in quantifying the expected warming in
the year 2100. Moreover, in view of recent plans to limit
future global warming to between 1.5 and 2°C (Paris Agree-
ment), establishing ECS is crucial to determining how to
cap greenhouse gas emissions to limit warming to within
this range and contribute to objectives described under the
climate action SDG.
In addition to the direct radiative effect caused by a
change in CO2 concentration, surface temperature responds
to feedbacks operating in the climate system. These feed-
backs can act on different timescales and amplify (or
dampen) the initial temperature change as a result of CO2
forcing. Certain fast(er) feedback processes, such as surface
albedo-temperature feedbacks, tend to lead to an amplified
climate response to CO2-induced radiative forcing. ECS esti-
mates have mostly been derived using climate models that
represent fast(er) feedbacks, where fast means fast enough
to approach an equilibrium climate state within a century.
Together with observations of the instrumental period,
ECS incorporating fast(er) feedbacks is estimated to range
between 1.5 and 4.5°C (Solomon 2007). This range has
changed little since the first estimates of ECS (Charney etal.
1979).
However, since 1979 our scientific understanding of the
stability of ECS, and how slow(er) feedbacks may alter
it, has grown substantially. This is in no small part due to
palaeoclimate modelling. The concept of longer term cli-
mate sensitivity, or Earth system sensitivity, emerged from
studying the way climate varied in response to variations
in atmospheric CO2 concentration (Hansen etal. 2008).
One of the most salient observations made by palaeocli-
matology is that the magnitude of reconstructed climate
change in the past can be hard to reconcile with the abso-
lute CO2 forcing at a given time, and from fast(er) climate
feedbacks alone. This draws attention to an important
limitation of a scientific focus that is restricted to mod-
ern and recent climate states, as it is incapable of provid-
ing the kind of broader perspective needed to determine
how climate responds to CO2 forcing in the longer term
(multi-centennial to millennial timescales). It has been
possible to reconcile the magnitude of past climate change
to direct CO2 forcing, in part by considering the contri-
bution to temperature change that can be derived from
slower responding components of the Earth system, such
as the response of ice sheets and vegetation cover (Hansen
etal. 2008; Lunt etal. 2010a; Rohling etal. 2012; Hay-
wood etal. 2013). In addition, palaeoclimate modelling
has highlighted that ECS itself may not be a constant. The
nature of the climate system, which can affect feedback
processes, may influence how the surface temperature
responds to CO2-based forcing. However, the degree to
which ECS variations according to base state are influ-
enced by the specific model chosen remains unknown. As
such, palaeoclimate modelling has made an important con-
tribution towards understanding the complexity of deriv-
ing ECS. More broadly, it is helping us to understand how
the sensitivity of global temperature to CO2 variation may
have changed in the past in response to the first order con-
trols of palaeogeography (see Fig.2).
Fig. 2 Global mean annual surface air temperature as a function of
atmospheric CO2 simulated by the Community Climate Model Ver-
sion 4 (CCSM4) at the National Center for Atmospheric Research.
Red dots show the simulated global temperature response to rising
CO2 concentration based when using modern geography, ice sheets
and vegetation in the model. Green dots show the simulated global
temperature response to rising CO2 concentration when using modern
geography, Pliocene ice sheets and vegetation in the model. Blue dots
show the simulated global temperature response to rising CO2 con-
centration when using Eocene or Cretaceous geography, no ice sheets
and prescribed palaeo vegetation (Bitz etal. 2012; Brady etal. 2013;
Baatsen etal. 2018; Tabor etal. 2016; Feng etal. 2017)

A.M.Haywood et al.
1 3
3 Model/Data Comparison: Veracities,
Uncertainties andSynergies
Proxy data-based environmental reconstructions play a
central role in evaluating the ability of climate models to
simulate past, present and future climate change. Over the
last few decades, several paleoclimate modelling intercom-
parison projects have provided compilations of terrestrial
and marine biological and geochemical data to facilitate
global data-model comparisons for different time intervals
in Earth history (e.g., Kageyama etal. 2018). For qualita-
tive and quantitative comparison, climate models are either
used in “forward mode” (i.e., models are capable of simulat-
ing proxy systems, such as biomes or isotopes) or “inverse
mode” where proxy data measurements are translated into
the same climatological values produced by climate mod-
els (temperature/precipitation, etc.). One of the greatest
strengths of palaeoclimate simulations is their ability to
provide process-based explanations for past environmental
change. Testing the importance of feedback mechanisms
through palaeoclimate modelling was a major step towards
identifying and understanding non-linear responses of
the environment to climate change. A prominent example
includes the analysis of vegetation, ocean and soil feed-
backs simulated in palaeoclimate models to understand
the strong response of the African monsoon and associated
rapid “greening” of the Sahara during the Holocene Afri-
can Humid Period (AHP). The AHP is recorded in mul-
tiple archaeological and geological records, but cannot be
explained by orbital forcing alone (Claussen etal. 1999;
Tjallingii etal. 2008; Tierney etal. 2017).
The majority of data-model comparison studies have
focused on the most recent geological past (such as the
AHP), and their outcomes and benefits for the understand-
ing of Holocene and Pleistocene environments have been
discussed elsewhere (Braconnot etal. 2012; Harrison etal.
2016). However, palaeoclimate modelling has also improved
our understanding of warm climates in the deeper geological
past, which were primarily controlled by elevated green-
house gas concentrations, providing an additional framework
for understanding future climate change. Whilst pre-Qua-
ternary warm climates hold the key to understanding how
environments respond to CO2-induced warming in the long
term, the uncertainties in defining geological boundary con-
ditions and reconstructing past environments increase with
geological age. Furthermore, disagreements between climate
model simulations and available geological data in the polar
regions remain, with models underestimating the degree of
warming (e.g., Masson-Delmotte etal. 2006; Haywood etal.
2013; Huber and Caballero 2011; Dowsett 2013).
The analysis of congruence between proxy data and
model simulations is of mutual benefit in that it has the
potential to improve the assessment of model performance,
and the robustness of proxy data-based environmental
reconstructions. Data assimilation, which incorporates
observations into numerical modelling, have been shown
to be a promising new technique in pre-Quaternary global
biome mapping projects to regionally improve model sim-
ulations and to increase the spatial and temporal resolution
of data-based vegetation reconstructions (Salzmann etal.
2008; Pound etal. 2012). In addition, the resolution of
the so-called “cool tropics paradox” is a prominent exam-
ple where palaeoclimate model outputs challenged sea
surface temperature (SST) estimates (e.g., D’Hondt and
Arthur 1996). Early estimates of tropical SSTs for the Cre-
taceous were far cooler than climate model simulations.
However, newer exceptionally well-preserved Palaeogene
microfossils (Sexton etal. 2006) led to a revision to higher
estimated SSTs bringing greater agreement between the
data and model estimates of tropical SSTs (Pearson etal.
2001). Furthermore, a long-standing discrepancy between
model simulations of atmospheric CO2 (pCO2), ice sheet
extent and the geological record of ice sheet and sea-level
variability during the icehouse of the Palaeozoic (330Ma)
prompted the generation of new high-resolution proxy
records of pCO2 that reconcile the geological archives and
model outputs (Montanez etal. 2016).
The spatial and temporal resolution, and accuracy of deep
time, pre-Quaternary reconstructions, have significantly
improved as science has progressed. The outputs from vari-
ous new international pre-Quaternary model intercompari-
son initiatives, for example, PlioMIP (Haywood etal. 2016)
and DeepMIP (Lunt etal. 2017) and proxy data syntheses,
for example, PlioVar PAGES (McClymont etal. 2015) and
PRISM3 (Dowsett etal. 2016) now enable reconstructions
of terrestrial and marine environmental change over multiple
time intervals during the last 65 million years at a global
scale (see Fig.3). Coincidentally, an increasing number
of exceptionally dated, high-resolution deep-time geologi-
cal records spanning several millions of years are becom-
ing available (Brigham-Grette etal. 2013; Herbert etal.
2015; Panitz etal. 2018). These allow, for the first time,
combined model-data approaches to analyse the role and
importance of climate extremes, astronomical cycles, non-
linear responses and feedback mechanisms, and non-modern
analogue environments.
4 Palaeoclimate Modelling
andUnderstanding Life onEarth: Past,
Present andFuture
There is increasing concern over how Earth’s biota will
respond to the rapid climatic changes already underway
(Urban 2015; Thomas etal. 2004; Barnosky etal. 2011).

What can Palaeoclimate Modelling doforyou?
1 3
These concerns are highlighted in the UN Sustainability
Goals of preserving and protecting biodiversity for the
maintenance of ecosystem goods and services, both on land
and in the water. However, preservation and maintenance
of biodiversity rely on accurate understanding and predic-
tions of climate-life dynamics on both short and long time
scales (Finnegan etal. 2015; McKinney 1997; Dawson etal.
2011). Species’ interactions with climate on longer time
scales provide necessary insights into biotic responses to
differing rates of environmental change, non-analogue cli-
mate scenarios, and extreme warmth (Barnosky etal. 2011;
Finnegan etal. 2015; Williams and Jackson 2007), all of
which have relevance to changes that are occurring today
(Williams etal. 2007).
Predictive models of biotic responses to climate change
can be sourced from the integration of fossils and palaeocli-
mate data. Palaeoclimate models are essential to disentangle
biotic responses to climate change, because they provide a
spatially explicit framework in which to test hypotheses. In
a perfect world, science would have access to palaeo-proxy
data that provide accurate estimates of past environmental
conditions for every point on Earth throughout Earth history.
In reality, palaeo-proxy data are spatially discontinuous, and
while it can provide robust palaeoenvironmental constraints
on local scales, it is often temporally limited. To generate
longer term environmental records, data are compiled such
that they represent globally averaged signals (e.g., Zachos
etal. 2008), and thus it can be challenging to disentangle the
causal processes responsible for regional biological patterns
from such global compilations.
Palaeoclimate models fill proxy data-deficient gaps, pro-
viding higher resolution spatial and temporal constraints on
the biotic responses to climate. When coupled with eco-
logical niche modelling (ENM: Myers etal. 2015; Peterson
etal. 2011; Svenning etal. 2011), palaeoclimate models also
provide critical insight on both the rate at which species are
Fig. 3 Distribution of biomes
as displayed on a hypothetical
“supercontinent” (after Troll
1948) for present-day (after
Klink 2008) and the geologi-
cal past. The distribution was
created through combination of
vegetation simulation informed
by palaeoclimate simulations
as well as palaeobotanical data
to create a spatial reconstruc-
tion of global vegetation for
the Miocene (Tortonian; Pound
etal. 2011), the Pliocene (Pia-
cenzian; Salzmann etal. 2008)
and the Last Glacial Maximum;
Kageyama etal. 2012; Tarasov
etal. 2000)

A.M.Haywood et al.
1 3
able to respond to changing conditions, and on those species
most vulnerable to them. The record of responses to differ-
ing rates of environmental change in the past is capable of
elucidating whether a given species can survive the rapid
and unprecedented rate of present-day climate change via
either adaptation and/or environmental range shifts (Dawson
etal. 2011; Harrison and Prentice 2003; Davis and Shaw
2001; Parmesan and Yohe 2003; Chen etal. 2011; Saupe
etal. 2014; Lawing and Polly 2011). Palaeoclimate models
can also be used to estimate a species’ traits, such as their
abiotic niches, to examine whether these traits result in dif-
ferential extinction risk (Saupe etal. 2015) and to question
the role of climate in influencing evolutionary, ecological
and biogeochemical processes at varying spatial and tem-
poral scales (e.g., Svenning etal. 2015). Such models have
been used to study the co-evolution of Earth and life, with
focus on how climate regulates the tempo and mode of spe-
ciation, extinction and adaptation. Examples in the latter
category include work that aims to quantify rates of within-
lineage abiotic niche evolution (Saupe etal. 2014; Jackson
and Blois 2015; Stigall 2014; Veloz etal. 2012), also of
relevance to the UN goal of conserving biodiversity.
Palaeoclimate models have the ability to contribute to
debates regarding the role of climate in regulating past biotic
events, particularly major extinctions. For example, the late
Pleistocene and early Holocene witnessed the extinction of
more than 97 genera of megafauna (animals > 44kg; Bar-
nosky etal. 2004), but the kill mechanism(s) for this event
are debated. Over hunting by humans and climate change
have been proposed as the two primary mechanisms (Sven-
ning etal. 2011), and the latter hypothesis has been tested
by climate models, helping to produce estimates of the
degree to which suitable habitats for various taxa changed
as climate warmed. Results are variable, with some studies
finding available habitat increased for taxa (Martinez-Meyer
etal. 2004; Varela etal. 2010) and others finding that it
decreased (Nogués-Bravo etal. 2008), potentially reflect-
ing where each taxa was distributed latitudinally (Svenning
etal. 2011).
Ecological patterns and processes may also be influenced
by climate, and palaeoclimate models can test the extent to
which climate controls patterns of distribution, dispersal,
community composition and assembly (Lawing and Polly
2011; Gavin etal. 2014). Moritz etal. (2009), for example,
used palaeoclimate models to examine the origins of a suture
zone—shared regions of secondary contact between long-
isolated lineages—in the Australian Wet Tropics rainforest.
The authors found that the zones of contact were clustered
in a corridor between two major Quaternary refugia, sug-
gesting it was unsuitable for the species during the mid-
Holocene, and that the current suture zone was established
only within the last few 1000 years.
Understanding how climate regulates the biological con-
trols of major element cycling, in particular carbon, is of
critical importance for accurate estimations of the effects
of elevated atmospheric CO2 on global temperatures, natu-
ral carbon sources and sinks, future ocean chemistry and
ecosystem responses (e.g., Cox etal. 2000; Le Quéré etal.
2009; Sarmiento etal. 2004). The palaeontological perspec-
tive allows us to ground truth our understanding of these
systems. At the broadest scale Earth System Models of Inter-
mediate Complexity (EMICs), or even simpler models, have
been used to investigate the impact major biological innova-
tions such as evolution of photosynthesis ~ 2.5 billion years
ago on ocean chemistry (Lenton and Daines 2017), and the
colonisation of the land by plants and its effect on weather-
ing and atmospheric CO2 (Berner 1998). The integration of
biogeochemical processes into palaeoclimate models also
allows us to reconstruct the influence of changing climate
and biogeochemistry on shorter timescales and gain a greater
understanding of thresholds, sensitivity and tipping points.
More temporally constrained work allows us to investigate
the impact of glacial–interglacial climates on ocean chem-
istry and carbon cycling on timescales relevant to humans
(Buchanan etal. 2016; Adloff etal. 2018).
The geological record provides a direct source of infor-
mation about biological processes against a backdrop of
varying climate, allowing us to investigate system/species
baselines, resiliencies and failure points in different climate
states. To use this rich resource from a modelling perspec-
tive, we require higher spatial resolution transient climate
simulations that will provide greater spatial and temporal
constraints on speciation, extinction dynamics, niche evo-
lution through time, and dispersal corridors and refugia in
the face of rapidly changing environments. This will facili-
tate fundamental knowledge capable of informing strategies
for the management of future biodiversity. Where transient
simulations with high spatial resolution are computationally
prohibitive, running snapshots over geologically and evolu-
tionarily meaningful timescales, particularly around peri-
ods with major climate transitions and aberrations, provide
highly valuable initial benchmarks.
5 Melting Ice Sheets andSea‑Level Change
A major scientific and societal challenge is understanding
the response of ice sheets to warming, and the resulting
rates, magnitudes and impacts of regional sea-level change
in the next 100years and beyond (Church etal. 2013). By
reconstructing past ice sheet variability, changes over cen-
tennial to millennial (and longer timescales) can be under-
stood. This is central to understanding societal impacts and
risks associated with future climate change.

What can Palaeoclimate Modelling doforyou?
1 3
The palaeo record helps constrain the drivers of ice-sheet
change, and therefore, associated sea-level change, during
differing climate states. Over the last 65 million years, major
climate transitions associated with growth and decay of ice
sheets were superimposed upon a gradual trend of atmos-
pheric cooling (Zachos etal. 2008). Due to the sparse nature
of geological evidence for ice sheet extent and sea-level his-
tory, ice sheet and palaeoclimate models have fundamen-
tally changed our understanding of ice-sheet growth in both
hemispheres, disentangling the role of CO2 versus the role of
tectonics in driving ice sheet expansion over major climate
transitions. For example, DeConto and Pollard (2003) found
that Antarctic Ice Sheet growth at the Eocene Oligocene
Transition (~ 34Ma) was driven by decreasing atmospheric
CO2, countering the prevailing view that the opening of the
Drake Passage and subsequent thermal isolation of the con-
tinent was responsible (Kennett 1977). Tectonics and moun-
tain building have also been implicated in the gradual onset
of northern hemisphere glaciation between 3.6 and 2.4Ma
(e.g., Mudelsee and Raymo 2005). Lunt etal. (2008) estab-
lished that decreasing atmospheric CO2, rather than tecton-
ics, was the dominant control on Greenland ice sheet growth.
Models also highlight that the scale of growth is sensitive to
whether the ice sheet was growing from an entirely ice-free
state or not (Contoux etal. 2015).
Whilst palaeoclimate modelling has been informative in
understanding the growth of ice sheets, of more pressing
concern is the scale and rate of future ice sheet mass loss
in a higher CO2 world. Instrumental records (e.g., satellite
data) of glacier extent only systematically capture the last
4 decades of change and, therefore, limit our capability to
understand large-scale, long-term changes in ice volume.
Global mean sea level (GMSL) rise from 1901 to 2010 has
been 1.7mm/year (Church etal. 2013). However, during
the last deglaciation (ca. 21–7ka) magnitudes and rates of
GMSL rise were significantly larger. Combined palaeocli-
mate and ice sheet modelling has identified an acceleration
in ice mass loss at 14.5ka, triggered by abrupt warming,
driving separation of the North American Ice Sheet into
regional ice domes (a process termed saddle collapse) and
contributing 5–6m to GMSL at a rate of ~ 14.7 to 17.6mm/
year (Gregoire etal. 2016; Gregoire etal. 2012). This pro-
vides a mechanism to explain a significant proportion of the
rise in GMSL at 14.5ka. Given that sea level is projected
to rise well beyond 2100 (Clark etal. 2016) assessing the
ability of models to reproduce rates of sea-level change on
centennial to millennial timescales, and potential mecha-
nisms for rapid collapse, is central to have confidence when
applying them to long-term future projections.
Models have helped refine our understanding of potential
rates and scales of sea-level rise, which can be attributed to
specific processes during previous climatically warm periods.
During the last interglacial (LIG) (ca. 129–116ka), GMSL is
thought to have been 6–9m above present (Dutton etal. 2015)
when the climate was 3–5°C warmer at polar latitude (Capron
etal. 2014). Moreover, it is likely that there was a period dur-
ing the LIG in which GMSL rose at a 1000-year average rate
exceeding 3mm/year (Kopp etal. 2010), but it is important
to understand which ice sheet(s) contributed to this rapid rate.
Coupled palaeoclimate-ice sheet simulations, consistent with
geologic data, indicate a retreat of ice in Greenland during the
LIG (Fig.4) leading to a GMSL rise of ca. 1.4m. Models also
suggest that Antarctica could have also contributed 3–4m to
the LIG highstand (Goelzer etal. 2016), with one study sug-
gesting that with > 2–3°C of Southern Ocean warming there
is the potential for complete collapse of the West Antarctic
ice sheet (Sutter etal. 2016), the recurrence of which is a key
concern in the context of future climate change.
The LIG, and the even warmer mid-Pliocene Warm
Period (mPWP, 3–3.3Ma), have been used as analogues to
understand future Earth system responses to warming at the
poles. DeConto and Pollard (2016) calibrated a palaeocli-
mate and ice sheet-modelling framework against the geo-
logical record of sea-level change during these time periods
to predict future ice mass loss from Antarctica. To reconcile
past records of GMSL with modelled ice mass loss, they
invoked a new mechanism that enhances the sensitivity of
the ice sheet where it meets the ocean. If this is applied
under future climate scenarios, Antarctica has the potential
to contribute more than 1m to GMSL by 2100 and more
than 15m by 2500. A challenge to these future predictions
is that we require supporting empirical evidence that these
processes operated in the past (Ritz etal. 2015).
Palaeoclimate modelling has been critical in improv-
ing understanding of how ice sheets, and thus sea level,
respond to increasing greenhouse gases. Progress towards
addressing the UN SDGs and the WCRP grand challenges
will come from fully coupling palaeoclimate and ice sheet
models to perform transient simulations at higher resolutions
than previously possible so that feedbacks between these
components of the Earth system can be better quantified.
Current modelling efforts focus largely on ice-sheet contri-
butions to GMSL, but in the future regional sea level will
significantly deviate from the global mean. Incorporation of
other controls on sea level such as ocean-density changes,
glacio-isostatic adjustment, dynamic topography and ero-
sion and sediment transport (Church etal. 2013) will help
reduce uncertainty in long-term (centennial to millennial)
projections.
6 Palaeoclimate Models andHydrology
Much of the climate change debate, particularly when
discussing the past, is focussed on changes in tempera-
ture. However, the WCRP grand challenges highlight the

A.M.Haywood et al.
1 3
importance of water supply for food production as well as the
role of extreme hydrological events (floods and droughts).
Both of these aspects are also closely linked to the UN Sus-
tainability Goals of ending hunger and improving food secu-
rity as well as delivering clean water and sanitation.
Until recently, most palaeoclimate modelling has
focussed on improving our understanding and ability to
model the mean changes in temperature/precipitation. This
type of modelling includes long timescale changes, such
as the role of Tibetan uplift in enhancing the South Asian
monsoon system (e.g., Manabe and Terpstra 1974; Ramstein
etal. 1997; Lunt etal. 2010b) or evaluation of simulated
monsoon changes resulting from orbital changes in the late
Quaternary and their impact on lake levels (e.g., Kutzbach
and Street-Perrott 1985). In addition to orbital enhance-
ment of summer monsoons, recent advances in computing
power allow palaeoclimate modelling of transient changes,
indicating the importance of changes in CO2 and meltwater
during the Quaternary as having affected the evolution of
rainfall patterns (Otto-Bliesner etal. 2014). Thus, the late
Quaternary provides a challenging test for models for forc-
ings relevant to the present.
Providing a clean water supply can also be facilitated by
learning from the past. For instance, there is considerable
concern about the recent decreases in the area of Lake Chad
(e.g., Lemoalle etal. 2012). However, in the Holocene and
Pliocene Lake Chad was much larger (the so-called Mega-
Chad). Palaeoclimate modelling (e.g., Sepulchre etal. 2008;
Contoux etal. 2013; Haywood etal. 2004) has shown that
this is a result of modest shifts in the position of the ITCZ
and hence implies that communities must expect and adapt
to high variability in Lake Chad on decadal and longer time-
scales. Considerably more work is needed to expand these
studies to other hydrological systems. Similarly, many areas
Fig. 4 Simulated Greenland Ice Sheet minimum extent for a the mid-
Pliocene warm period (mPWP ~ 3 to 3.3 Ma) and b the Last Inter-
glacial (LIG ~ 125 Ka) simulated by ice sheet models with multiple
climate model forcings for each period. The shading indicates the
number of model simulations that predict ice being present at a given
location. Nine models simulations are included for the LIG (Otto-
Bliesner et al. 2006; Solomon 2007; IPCC 2013; Yau et al. 2016)
and eight models are included for the mPWP (Dolan et al. 2015).
The combination of climate and ice sheet modelling can lead to new
insights regarding ice sheet extent and variability for time periods
where direct geological evidence is sparse or entirely missing

What can Palaeoclimate Modelling doforyou?
1 3
of Africa rely on groundwater sources, and some of these
reservoirs still contain water accumulated many thousands
of years ago. We have a poor understanding of many of these
systems and future work must target improvements in this
area.
Palaeoclimate modelling that directly targets the grand
challenge areas of hydrological extremes and water supply
are at an early stage of development, but should become one
of the major priorities for research. Until recently, extreme
events were hard to simulate but improvements in the spatial
resolution of models are allowing palaeoclimate models to
tackle such issues (see outlook section). Initial work (e.g.,
Haywood etal. 2004) used regional models to show that the
hydrological cycle associated with extreme warm periods
operated very differently, and this affected the interpretation
of the sedimentary structures found for such periods. More
recent work is increasingly focussing directly on the science
of palaeo-tempestology and extreme events. For instance,
Peng etal. (2014) modelled severe and persistent droughts
in China during the last millennium and suggested that these
droughts (and the East Asian monsoon system) could have
been modulated by variations in solar output.
7 Palaeoclimate Modelling andHuman
Systems
Palaeoanthropologists and archaeologists have a long his-
tory of collaboration with climate modellers. From a mod-
elling perspective, palaeoclimate proxies (e.g., pollen data,
microfauna, malacofauna) obtained from dated archaeo-
logical deposits allow climatologists to test model perfor-
mance in non-analogue situations (Braconnot etal. 2012).
From an archaeological perspective, palaeoenvironmental
reconstruction and palaeoclimate modelling provide essen-
tial context for understanding past human adaptations. Pal-
aeoanthropology, firmly rooted in evolutionary ecology,
has long recognised that climate change has an impact on
hominin evolution (Vrba 1995) and palaeoclimate mod-
els feature prominently in palaeoanthropological debates.
Palaeoclimate models are also increasingly integrated into
archaeological models that seek to understand the pattern of
hominin dispersals out of Africa, for example, or to explore
how past climate conditions affected the spatial distribution
and structure of human populations, altering the course of
cultural evolution. The pioneering ‘Stage 3 Project’ (Van
Andel and Davies 2003) is an example of the interdiscipli-
nary nature of archaeological research, demonstrating the
integration of palaeoclimate models and archaeological data
to design research that sheds light on the dynamics of human
populations in the past.
Early human evolution is currently framed as a series
of adaptive responses to environmental changes linked to
orbital forcing mechanisms. Within this framework, climate
models are used in conjunction with palaeoenvironmental
data to interpret the paleontological record (Grove 2011).
For example, although the origins of bipedalism (which
defines the hominin lineage) extend further back in time,
the evolution of obligate bipedalism during the Pliocene
is linked to transformations of the African landscape and
the expansion of C4-dominated grasslands. This event and
others like it (e.g., the emergence of the genus Homo) are
thought to have been triggered ultimately by orbital forc-
ing (Maslin and Christensen 2007). Climate models have
also been used to assess the impact of climate variability on
hominin populations. The variability selection hypothesis,
for example, suggests that trends in variability during the
Plio-Pleistocene resulted in a selection for plasticity that
characterises our lineage (Potts and Faith 2015), which could
explain why humans have dispersed more widely than any
other primate species.
Our understanding of the mechanisms shaping the pattern
of hominin dispersals, which are major events in the history
of our species, is framed in terms of environmental change.
The earliest hominin dispersals, for example, likely coin-
cided with climate events that reshaped the biogeographi-
cal map of Africa (Larrasoaña etal. 2013). Later dispersals,
such as the dispersal of modern humans into Eurasia during
the late Pleistocene, are also best understood from a climate
perspective (Hughes etal. 2007; Timmermann and Friedrich
2016; Eriksson etal. 2012). Modern human dispersals to
Australia and the New World coincide with megafaunal
extinctions and climate models provide us with the data we
need to contextualise this information, attributing causal-
ity where it is due (Prescott etal. 2012). If climate change
shaped the pattern of human dispersals in the past, climate
variability has been shown to affect modern societies too,
increasing conflict (O’Loughlin etal. 2014), which is linked
to modern population displacements.
High-resolution palaeoclimate models have been used to
study the response of human systems to climate instability
(Banks etal. 2013), to assess the impact of climate events
such as the Last Glacial Maximum on population structure
and demography (Tallavaara etal. 2015), and test the sensi-
tivity of human systems to climate predictors such as ecolog-
ical risk (Burke etal. 2017). Modelling the complex interac-
tions between human systems and the environment allows us
to appreciate how demographic patterns such as population
size and connectivity, which are affected by climate change,
can drive technological innovation and cultural complex-
ity. By developing spatially explicit models that incorpo-
rate climate models or simply make use of model outputs,
archaeologists gain a richer and more dynamic appreciation
of the environmental response to climate change and the
various mechanisms that allow human systems to adapt.
These archaeological models, in turn, hold lessons for the

A.M.Haywood et al.
1 3
future as we attempt to gauge the resilience of small-scale, or
“traditional” societies and judge what is required to preserve
human cultural diversity.
Collaborations between palaeoclimate modellers, archae-
ologists and palaeoanthropologists have provided rich
opportunities for modelling human/environment interactions
as well as contributing to improving climate model design.
However, difficulties arise because of differences in the reso-
lution of model outputs and signals from the palaeoclimate
record that limits the application of palaeoclimate models
to the study of early human evolution, for example. Further-
more, human populations perceive and respond to environ-
mental change at a wide range of temporal and spatial scales.
Increased capability and capacity in palaeoclimate research
and improvements to the scale of resolution of model out-
puts, as well as greater efforts towards improving the acces-
sibility of climate model outputs for non-specialists, will
improve this situation in the future.
8 Palaeoclimate Modelling, Industry
andInnovation
There is a strong focus on the impact of contemporary cli-
mate change on various aspects of industry. However, the
fact that many important aspects of society’s requirements
need a longer term perspective either for the future or of the
deep past is often overlooked.
One of the most challenging demands of modern society
is the use of the Earth’s geological resources and reserves.
The growth of the world economy is demanding greater sup-
plies of many metals such as aluminium, as well as ever
increasing demands for fossil fuels. The geographical distri-
bution of aluminium’s chief ore (bauxite) and organic-rich
source rocks for hydrocarbons both depend on past climates.
Hence prediction (or retrodiction) of past climates can help
in frontier exploration for these resources. Since the earliest
days of palaeoclimate modelling, efforts have been made
to retrodict source rocks. Parrish and Curtis (1982) and
Scotese and Summerhayes (1986) developed a conceptual
and a computer model of atmospheric circulation patterns
to predict where oceanic upwelling occurs. Such regions
are typically associated with high organic productivity that
subsequently is buried and potentially becomes source rocks.
Further work with palaeoclimate models extended these
predictions by quantifying them (e.g., Barron 1985). More
recent work (Harris etal. 2017) continues this research with
full ESMs, making use of simulated atmospheric and ocean
circulation (including aspects such as storminess and solar
radiation) as well as aspects of the modelled carbon cycle
to make very specific regional predictions of source rocks.
These are used for frontier exploration.
Palaeoclimate modelling also plays a major role in risk
assessment of the long-term storage of nuclear waste. Any
site proposed as a nuclear waste repository requires a risk
assessment measured up to 100,000years into the future. On
such long time scales, future orbitally forced climate change
becomes as important as anthropogenic forcing. Early stud-
ies (e.g., Goodess etal. 1990) simply extrapolated past long-
term changes into the future, but more recent work has made
extensive use of more detailed palaeoclimate models. The
latest approaches (e.g., Lindborg etal. 2005) use a com-
bination of simple and full complexity climate models to
provide detailed predictions of site-specific climate up to
200,000years into the future, using methodologies identical
to many palaeoclimate modelling studies.
The methodologies of palaeoclimate modelling have been
utilised in some aspects of geoengineering research. This is
because palaeoclimate modelling has often pioneered the
use of ESMs, including detailed representations of the car-
bon cycle. Hence, many ideas of carbon sequestration have
used palaeoclimate modelling tools to evaluate their efficacy.
For example, Taylor etal. (2016) discussed the artificial
acceleration of rock weathering as a potential method for
enhanced drawdown of CO2 and reduced ocean acidification.
Many aspects of society are also vulnerable to extreme
events. Almost by definition (e.g., 100year return period),
these extreme events are beyond the observational record
and palaeoclimate studies are required to give context to
any event. Palaeoclimate proxy observations of storm events
have frequently been used in infrastructure planning (such as
flooding events and the location of nuclear power stations),
but more recently palaeoclimate modelling of extreme
events has also helped in planning process. For instance,
model predictions of storm events during the last millennium
(Kozar etal. 2013) were considered as a part of the evidence
based in the New York City Panel on Climate Change 2015
Report (Horton etal. 2015). These include estimating the
frequency of extreme events for flood protection, the risk
assessment for nuclear power stations and long-term storage
of nuclear waste.
9 Outlook
9.1 Overcoming Current Methodological/
Technological Limitations
A thorough understanding of physical processes, the robust
application of mathematics and statistical techniques, the
availability of accurate geological boundary conditions and
forcing estimates, combined with the required research-
intensive computer facilities and appropriate computational
and software engineering support, are central to the overall
capability of palaeoclimate modelling (Fig.5).

What can Palaeoclimate Modelling doforyou?
1 3
One of the most fundamental strengths of palaeoclimate
modelling is that it provides a unique way to examine the
response of the Earth system to forcing mechanisms in an
integrated way. Key uncertainties associated with future cli-
mate change projection stem from the strength of positive
feedbacks associated with components of the climate system
that respond to forcing over medium to long timescales (e.g.,
ocean circulation and ice sheets; IPCC 2013), and the geo-
logical record is uniquely capable of preserving signals of
change associated with slower responding components of the
Earth system (Haywood etal. 2013). ESMs that incorporate
the representation of many earth system processes, and their
associated feedbacks on climate, are now available and can
be run at higher and higher spatial resolutions (Peng etal.
2014). Such models are capable of simulating the response
and longer term climate feedbacks stemming from a wide
array of earth system processes, and from climate-relevant
processes that operate over medium to long timescales.
Fig. 5 Summary illustration showing a key data and technological/
knowledge requirements that underpin palaeoclimate modelling, b
key areas of contribution to understanding different physical systems
and life on Earth, c the human value components of the contributions
shown in b, and d, e the direction of travel required to address out-
standing critical research questions with significant human impor-
tance

A.M.Haywood et al.
1 3
However, with increasing resolution and model complexity
comes increasing computational demand and cost. In addi-
tion, using high-resolution ESMs to simulate the past comes
with its own unique scientific and technological challenges
that can dramatically increase the computational expense
and time associated with producing simulations.
For example, uncertainties in geological boundary con-
ditions often necessitate the production of an ensemble of
climate simulations for a specific interval of time (Hay-
wood etal. 2013). In addition, reconfiguring ESMs so that
they can simulate the deeper past successfully is extremely
challenging. Such models are not developed with the spe-
cific needs of palaeoclimate modellers in mind. As such,
the reconfiguration of the land/sea mask, land elevation,
ocean bathymetry, land cover, etc., creates challenges that
require dedicated software engineering support to overcome,
which is difficult to resource adequately. In addition, with
increasing model capability comes increasing demand for
appropriate boundary conditions and forcing datasets so
that the potential of these new models can be fully realised.
For instance, models that incorporate complex representa-
tions of atmospheric chemistry and/or atmospheric dust/
aerosol-climate interactions may require information on
the initial concentration of CH4 in the atmosphere, or dust
emission sources and emissions of Volatile Organic Com-
pounds (VOC’s). Unless ESMs are developed so that the
model dynamically predicts such parameters, rather than
requiring their initial prescription, it may lead to increased
uncertainty in boundary conditions and forcings within pal-
aeoclimate simulations, as these parameters may be poorly
constrained geologically. Also given the radically different
(from modern) climates such models are applied to, and the
major changes to boundary conditions that are necessary,
palaeoclimate simulations require substantial spin-up time
insofar as they include a dynamic ocean, which can require
several thousand simulated years to fully adjust, though
atmospheric spin-up time is much faster (several decades to
a century of simulation). Here computational efficiency and
scalability of the model code (across computer processors)
become paramount. Any model that cannot reliably achieve
at least 10–30 model years per wall clock day with a reason-
able total CPU demand/cost will be very challenging, if not
practically impossible, to apply effectively to palaeoclimate
applications and the assessment of uncertainty in past cli-
mate simulation.
The majority of latest generation full complexity ESMs
do not meet the requirements for palaeoclimate modelling.
Model development is carried out in isolation from the pal-
aeoclimate community’s specific requirements. There seems
little scope that will change, meaning that the palaeoclimate
modelling community’s future interests could be best served
by adopting a more tailored strategy towards model devel-
opment. Examples of the development of EMICs (Earth
System Models of Intermediate complexity) as well as other
current large-scale research initiatives such as the climate
modelling initiative called PalMod (Paleo Modelling), which
is funded by the German Federal Ministry of Education and
Science to understand climate system dynamics and vari-
ability during the last glacial cycle. Fundamentally, more
considered and flexible strategies will be required to deter-
mine what spatial and vertical resolution and model com-
plexity is actually needed to answer specific challenges in
palaeoclimate science.
9.2 Enhanced Integration ofStatistical
Methodologies toAssess Uncertainty
While current climate models seek to optimally represent
physical processes that determine weather and climate, this
representation is not perfect and models have been tuned to
provide acceptable simulations of modern climate regime.
For palaeoclimate simulations, where boundary conditions
such as atmospheric CO2 concentration differ from mod-
ern climate simulations, the approaches used to ensure that
models deliver the best possible simulation for the mod-
ern (observed) climate may no longer hold. A commonly
employed solution to this problem is to consider ensembles
of models to understand the commonalities and differences
between possible model outputs. Statistical techniques such
as Bayesian Model Averaging (Hoeting etal. 1999) can be
used to compute ensemble averages where greater weight is
given to models that are most compatible with the available
data.
In addition, the impacts associated with global warm-
ing cannot be fully characterised by a change in the spa-
tial and temporal averages of specific climate variables.
To capture such changes it is necessary to model how the
distribution of a climate variable (instead of just the mean)
depends on changing boundary conditions (such as GHG
concentrations). A variety of statistical techniques have been
used for this purpose: quantile regression is a generalisa-
tion of linear regression, which allows for the estimation
of arbitrary quantiles of the distribution of a climate vari-
able instead of just the mean (e.g., Janson and Rajaratnam
2014). For example, extreme value theory describes the
tails of a distribution, with the aim of predicting extremes
beyond what has been observed in the available time series
of data (e.g., Mannshardt etal. 2013). Some authors have
also used specialised techniques to simultaneously capture
the correlations between proxy variables and climate vari-
ables across space and time (e.g., the GraphEM method;
Guillot etal. 2015), although such beneficial approaches are
not commonly used, and this highlights the need for further
integration between palaeoclimate modellers and applied
statisticians.

What can Palaeoclimate Modelling doforyou?
1 3
9.3 Removing Barriers toData Sharing
andMultidisciplinary Collaboration
Access to appropriate palaeoclimate model outputs is a
significant limitation for other research disciplines. The
progress already made towards widening access can be
attributed in part to research council requirements to make
publicly funded science widely available in national data
repositories (e.g., the British Atmospheric Data Centre).
In addition, journal requirements for the uploading of data
sets associated with specific publications have had a positive
impact, as well as internationally promoted output standards
and software libraries such as those adopted by the Coupled
Model Intercomparison Project (i.e., CMOR: the Climate
Model Output Rewriter). CMOR ensures that a standard set
of model variables for different climate model experiments
are available on Earth System Grid Federation repositories
(https ://esgf.llnl.gov).
New community-based efforts are underway to support
data sharing across disciplines. For example, the PaleoClim
database provides pre-processed climate data to support eco-
logical niche studies (Brown etal. 2018). These community-
led initiatives are important because the approach towards
processing model outputs can be application specific, and
scientists in the disciplines requiring climate outputs may
not have the required programming skills and experience
to successfully deal with unprocessed palaeoclimate model
data. Initiatives such as PaleoClim provide a template of
how communities can come together to discuss the removal
of barriers and enhance awareness of, and access to, palaeo-
climate modelling data.
Whilst the initiatives described above can improve access
to palaeoclimate model output by other communities, they
will not fully resolve the issue of rigorously embedding pal-
aeoclimate model outputs into other disciplines. There is
an underlying concern as to whether model outputs used
in specific applications contain an adequate appreciation
and expression of uncertainty. The obvious solution is to
embed palaeoclimate modellers within multidisciplinary
teams. However, the global pool of available palaeoclimate
modellers is small, and thus collaborative capacity is lim-
ited. A complementary solution promoted more generally
in efforts to foster multi-disciplinarity is the development
of T-shaped researchers (e.g., Palmer 1990). The develop-
ment of T-shaped skills is a concept promoted for more
than 20years, with the T representing the range of research
expertise an individual develops, and the foundation/depth
of individual understanding represented by the vertical bar
(Palmer 1990). Using this philosophy a researcher first
develops an expertise in their own discipline before subse-
quently developing their skills base in a way that facilitates
the deployment of their knowledge in a wider array of sci-
entific disciplines. The development of T-shaped researchers
is essential to the success of multi-disciplinarity, but it is
unclear how conducive academic environments currently are
to those wishing to adopt a T-shaped research skills base.
The required investment of time versus immediate scientific
return is of paramount consideration for early career scien-
tists, with the requirement to demonstrate sustained levels
of high research performance very clear.
9.4 Developing anEnhanced Focus onPast
forFuture Relevant Science
The potential contribution of different types of palaeocli-
mate modelling to the generation of science underpinning
UN SDGs or global scientific grand challenges is not equal.
This is also the case for the support palaeoclimate modelling
can provide to multidisciplinary research. For the science
to grow its influence in these regards, more targeted and
co-ordinated approaches will be required that maximise the
utility of palaeoclimate modelling. We highlight this need
and opportunity by reference to specific examples.
Mitigation and adaptation strategies for global climate
change are informed by studies that seek to better constrain
the extremes of natural variability in climate and weather
phenomena from the past (beyond the observed climate
period). However, a weakness of approaches that consider
only the very recent past, for example the last millennium,
is that the effect of the warming trend since the onset of
the industrial revolution is omitted from the assessment of
the behaviour of weather and climate extremes. Given the
current and projected rates of GHG emission, and the asso-
ciated rapid warming trend, palaeoclimate modelling is pay-
ing increasing attention to warmer (than the pre-industrial)
intervals, and also to intervals when CO2 concentrations in
the atmosphere were analogous to current and near future
concentrations (e.g., Haywood etal. 2013). Given the cur-
rent rapid rate of temperature increase, it is necessary to go
back in time as far as the Pliocene epoch (~ 3 million years
ago) to find the estimated 3°C global annual mean surface
temperature change that we are on track to achieve by the
end of this century (Haywood etal. 2013). This rapid pro-
gress towards analogous past warm climates was recently
highlighted by Burke etal. (2018) who used different cli-
mate model simulations for future GHG scenarios, and then
compared these to different simulated climates of the past
including the Last Glacial Maximum, the Mid-Holocene,
the Last Interglacial, the Pliocene and the early Eocene. Sta-
tistically, the climate state that they considered to be most
similar to what models predict will be reality by 2030–2050
AD was the mid-Pliocene Warm Period (~ 3 million years
ago), and by ~ 2200 AD the early Eocene (Burke etal. 2018).
This provides a sobering assessment of the rate of climate
change currently occurring, and underlines the importance
of an increasing focus on past warm intervals.

A.M.Haywood et al.
1 3
However, even a general concentration of community
efforts on warm epochs may not be sufficient to guarantee
the maximum utility of palaeo science in informing UN
SDGs and global scientific grand challenges. Climate vari-
ability is driven by variations in Earth’s orbit around the
sun with predictable periodicities (the Milankovitch cycles).
Orbital forcing has acted as a natural pacemaker for insola-
tion since our planets formation. Therefore, while epochs in
the past may be analogous in the sense of different conceiv-
able CO2 stabilisation scenarios for the future, at any point
in time within these epochs the surface expression of climate
(i.e., difference compared to the pre-industrial baseline) will
not solely be a response to GHG forcing (Haywood etal.
2016). While this does not matter greatly in terms of the
global annual mean temperature response, it is important
for the time-specific expression of climate change locally,
regionally and seasonally. Orbital parameters, and the result-
ing insolation pattern at the top of the atmosphere, are reli-
ably calculable for the Cenozoic (Laskar etal. 2011). It is
possible to isolate specific intervals within a warmer than
pre-industrial epoch with the same, or very similar, insola-
tion forcing (e.g., Haywood etal. 2016). Such an approach
provides an obvious benefit of studying a mean state climate
that is more influenced by a carbon cycle perturbation and
less influenced by other forcing agents. This approach has
been adopted within the scientific strategy underpinning
the second phase of the Pliocene Model Intercomparison
Project36, whereby a specific interglacial within the Pliocene
has been identified for study. Such a methodology differs
from more classical approaches in palaeoclimatology where
the most concentrated effort tends to focus on the most rapid
and/or largest transitions in Earth system behaviour, but that
does not necessarily mean those intervals are the most rele-
vant in the context of the future. The judgement is dependent
upon the scientific question which is asked. Nevertheless, it
is important to recognise that warm (and warming) intervals
in the past characterised by very different orbital forcing
compared to present-day (e.g., the Last Interglacial and the
Last Deglaciation) will remain very important to study. For
example, they are important for the assessment of regional
and seasonal variations in climate (past and future), and for
understanding how the Greenland and Antarctic Ice Sheets
respond to a warmer (and warming) atmosphere and oceans
(Otto-Bliesner etal. 2006).
Palaeoclimate modelling studies have focussed a great
deal on large-scale mean state climate changes. Within such
a context, numerous studies have examined modes of natural
climate variability, but very few have examined the nature of
extreme weather and climate events during warm episodes of
the deeper past. Given that society is likely to experience the
worst initial effects of anthropogenic climate change through
a change in the frequency and/or magnitude of extreme
events (IPCC 2013), a more concerted effort in this regard
is required. Whilst geological data may not always be avail-
able to assess the quality of model results in this regard, data
are available for the climatic mean state. This mean state is
a product of the average weather and climate variability at
any given time and place. Therefore, if models demonstrably
simulate the mean climate faithfully, this may add credence
to their simulation of higher order climate and weather vari-
ability, even if geological data to assess the model predic-
tions of extreme weather and climate events are absent.
10 Conclusion
In conclusion, palaeoclimate modelling over the last 4 dec-
ades has provided a broad and deep contribution to multi-
disciplinary science, and to the science underpinning global
grand challenges and SDGs. First-order questions about the
operation of climate and environmentally relevant processes,
and our planet’s limits in terms of sustaining life during peri-
ods of rapid change remain unanswered. This includes the
fundamental understanding of the carbon cycle, identifica-
tion of critical environmental thresholds for species distribu-
tion and life, and what are the longer term implications of
climate and ecosystem change on human adaptability and
vulnerability. The great strides made in the development
of more and more compete and capable models provide a
wealth of opportunity for further discovery, but only if the
unique challenges associated with simulating climate of the
past are properly appreciated and understood.
Acknowledgements This work was carried out as a part of, and finan-
cially supported by, the Past Earth Network (PEN: https ://www.paste
arth.net/). PEN was one of the UK EPSRC Forecasting Environmental
Change Networks led by J.V. A.M.H and A.M.D.
Author Contributions AMH and PJV conceived and led the overall
effort and contributed directly to the writing of the manuscript. All
authors shared joint responsibility for writing the manuscript and for
the preparation of the figures.
Compliance with Ethical Standards
Conflict of Interest The authors declare no competing interests.
Open Access This article is distributed under the terms of the Crea-
tive Commons Attribution 4.0 International License (http://creat iveco
mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribu-
tion, and reproduction in any medium, provided you give appropriate
credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made.

What can Palaeoclimate Modelling doforyou?
1 3
References
Adloff M, Reick CH, Claussen M (2018) Earth system model simula-
tions show different feedback strengths of the terrestrial carbon
cycle under glacial and interglacial conditions. Earth Syst Dyn
9:413–425
Baatsen M etal (2018) Equilibrium state and sensitivity of the simu-
lated middle-to-late Eocene climate. Clim Past Discuss. https ://
doi.org/10.5194/cp-2018-43
Banks WE, d’Errico F, Zilhão J (2013) Human–climate interaction
during the Early Upper Paleolithic: testing the hypothesis of
an adaptive shift between the Proto-Aurignacian and the Early
Aurignacian. J Hum Evol 64(1):39–55
Barnosky AD, Koch PL, Feranec RS, Wing SL, Shabel AB (2004)
Assessing the causes of late Pleistocene extinctions on the con-
tinents. Science 306:70–75
Barnosky AD etal (2011) Has the Earth’s sixth mass extinction
already arrived? Nature 471:51–57
Barron EJ (1985) Numerical climate modeling, a frontier in petro-
leum source rock prediction: results based on Cretaceous simu-
lations. AAPG Bull 69:448–459
Berner RA (1998) The carbon cycle and carbon dioxide over Phan-
erozoic time: the role of land plants. Philos Trans R Soc B
Biol Sci 353:75–82
Bitz CM, Shell KM, Gent PR, Bailey D, Danabasoglu G, Armour
KC, Holland MM, Kiehl JT (2012) Climate sensitivity in
the community climate system model version 4. J Clim
25:3053–3070
Braconnot P, Harrison SP, Kageyama M, Bartlein PJ, Masson-Delmotte
V, Abe-Ouchi A, Otto-Bliesner B, Zhao Y (2012) Evaluation
of climate models using palaeoclimatic data. Nat Clim Change
2:417
Brady EC, Otto-Bliesner BL, Kay JE, Rosenbloom N (2013) Sensitivity
to glacial forcing in the CCSM4. J Clim 26:1901–1924
Brigham-Grette J etal (2013) Pliocene warmth, polar amplification,
and stepped Pleistocene cooling recorded in NE Arctic Russia.
Science 340:1421–1427
Brown JL etal (2018) PaleoClim, high spatial resolution paleoclimate
surfaces for global land areas. Nat Sci Data 5:180254
Buchanan PJ etal (2016) The simulated climate of the Last Glacial
Maximum and insights into the global marine carbon cycle. Clim
Past 12:2271–2295
Burke A etal (2017) Risky business: the impact of climate and climate
variability on human population dynamics in Western Europe
during the Last Glacial Maximum. Quat Sci Rev 164:217–229
Burke KD etal (2018) Pliocene and Eocene provide best analogs for
near-future climates. Proc Natl Acad Sci 115:13288–13293
Capron E etal (2014) Temporal and spatial structure of multi-millen-
nial temperature changes at high latitudes during the Last Inter-
glacial. Quat Sci Rev 103:116–133
Charney JG etal (1979) Carbon dioxide and climate: a scientific assess-
ment. National Academy of Science, Washington, DC, p 22
Chen I-C, Hill JK, Ohlemüller R, Roy DB, Thomas CD (2011) Rapid
range shifts of species associated with high levels of climate
warming. Science 333:1024–1026
Church JA, Clark PU, Cazenave A, Gregory JM, Jevrejeva S, Lever-
mann A, Merrifield MA, Milne GA, Nerem RS, Nunn PD, Payne
AJ, Pfeffer WT, Stammer D, Unnikrishnan AS (2013) Sea level
change. In: Stocker TF, Qin D, Plattner G-K, Tignor M, Allen
SK, Boschung J, Nauels A, Xia Y, Bex V, Midgley PM (eds)
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. Cambridge University
Press, Cambridge
Clark PU etal (2016) Consequences of twenty-first-century policy for
multi-millennial climate and sea-level change. Nat Clim Change
6:360–369
Claussen M, Kubatzki C, Brovkin V, Ganopolski A, Hoelzmann P,
Pachur HJ (1999) Simulation of an abrupt change in Saha-
ran vegetation in the mid-Holocene. Geophys Res Lett
26(14):2037–2040
Contoux C etal (2013) Megalake Chad impact on climate and vegeta-
tion during the late Pliocene and the mid-Holocene. Clim Past
9:1417–1430
Contoux C, Dumas C, Ramstein G, Jost A, Dolan AM (2015) Model-
ling Greenland ice sheet inception and sustainability during the
Late Pliocene. Earth Planet Sci Lett 424:295–305
Cox PM, Betts RA, Jones CD, Spall SA, Totterdell IJ (2000) Erratum:
Acceleration of global warming due to carbon-cycle feedbacks
in a coupled climate model. Nature 408:184–187
D’Hondt S, Arthur MA (1996) Late Cretaceous oceans and the cool
tropic paradox. Science 271(5257):1838–1841
Davis MB, Shaw RG (2001) Range shifts and adaptive responses to
Quaternary climate change. Science 292:673–679
Dawson TP, Jackson ST, House JI, Prentice IC, Mace GM (2011)
Beyond predictions: biodiversity conservation in a changing cli-
mate. Science 332:53–58
DeConto RM, Pollard D (2003) Rapid Cenozoic glaciation of Antarc-
tica induced by declining atmospheric CO2. Nature 421:245–249
DeConto RM, Pollard D (2016) Contribution of Antarctica to past and
future sea-level rise. Nature 531:591–597
Dolan AM etal (2015) Using results from the PlioMIP ensemble to
investigate the Greenland Ice Sheet during the mid-Pliocene
Warm Period. Clim Past 11(3):403–424
Dowsett HJ etal (2013) Sea surface temperature of the mid-Piacen-
zian ocean: a data-model comparison. Sci Rep. https ://doi.
org/10.1038/srep0 2013
Dowsett H etal (2016) The PRISM4 (mid-Piacenzian) paleoenviron-
mental reconstruction. Clim Past 12:1519–1538
Dutton A etal (2015) Sea-level rise due to polar ice-sheet mass loss
during past warm periods. Science 349(6244):aaa4019. https ://
doi.org/10.1126/scien ce.aaa40 19
Eriksson A etal (2012) Late Pleistocene climate change and the global
expansion of anatomically modern humans. Proc Natl Acad Sci
109(40):16089–16094
Feng R, Otto-Bliesner BL, Fletcher TL, Tabor CR, Ballantyne AP,
Brady EC (2017) Amplified Late Pliocene terrestrial warmth in
northern high latitudes from greater radiative forcing and closed
Arctic Ocean gateways. Earth Planet Sci Lett 466:129–138
Finnegan S etal (2015) Extinctions. Paleontological baselines for
evaluating extinction risk in the modern oceans. Science
348(6234):567–570
Gates WL (1976) The numerical simulation of ice-age climate with
a global general circulation model. J Atmos Sci 33:1844–1873
Gavin DG etal (2014) Climate refugia: joint inference from fossil
records, species distribution models and phylogeography. New
Phytol 204:37–54
Goelzer H, Huybrechts P, Loutre MF, Fichefet T (2016) Last Intergla-
cial climate and sea-level evolution from a coupled ice sheet–cli-
mate model. Clim Past 12:2195–2213
Goodess CM, Palutikof JP, Davies TD (1990) A first approach to
assessing future climate states in the UK over very long time-
scales: input to studies of the integrity of radioactive waste
repositories. Clim Change 16:115–140
Gregoire LJ, Payne AJ, Valdes PJ (2012) Deglacial rapid sea level
rises caused by ice-sheet saddle collapses. Nature 487:219–222
Gregoire LJ, Otto-Bliesner B, Valdes PJ, Ivanovic R (2016) Abrupt
Bølling warming and ice saddle collapse contributions to the
Meltwater Pulse 1a rapid sea level rise. Geophys Res Lett
43:9130–9137

A.M.Haywood et al.
1 3
Grove M (2011) Change and variability in Plio-Pleistocene cli-
mates: modelling the hominin response. J Archaeol Sci
38(11):3038–3047
Guillot D, Rajaratnam B, Emile-Geay J (2015) Statistical paleocli-
mate reconstructions via Markov random fields. Ann Appl Stat
9(1):324–352
Hansen J etal (2008) Target atmospheric CO2 where should humanity
aim. Open Atmos Sci J 2:217–231.s
Harris J etal (2017) Paleogeography and paleo-earth systems in the
modeling of marine paleoproductivity: a prerequisite for the
prediction of petroleum source rocks. In: AbuAli MA, Moretti
I, Bolås HMN (eds) Petroleum systems analysis—case Studies,
vol 114. AAPG Memoir, Tulsa, pp 37–60
Harrison SP, Prentice CI (2003) Climate and CO2 controls on global
vegetation distribution at the last glacial maximum: analysis
based on palaeovegetation data, biome modelling and palaeo-
climate simulations. Glob Change Biol 9:983–1004
Harrison SP, Bartlein PJ, Prentice IC (2016) What have we learnt
from palaeoclimate simulations? J Quat Sci 31:363–385
Haywood AM, Valdes PJ, Markwick PJ (2004) Cretaceous (Wealden)
climates: a modelling perspective. Cretac Res 25(3):303–311
Haywood AM etal (2013) Large-scale features of Pliocene climate:
results from the Pliocene Model Intercomparison Project. Clim
Past 9:191–209
Haywood AM etal (2016) Pliocene model intercomparison (Pli-
oMIP) Phase 2: scientific objectives and experimental design.
Clim Past 12:663–675
Herbert TD, Ng G, Cleaveland Peterson L (2015) Evolution of Medi-
terranean sea surface temperatures 3.5–1.5Ma: regional and
hemispheric influences. Earth Planet Sci Lett 409:307–318
Hoeting JA etal (1999) Bayesian model averaging: a tutorial. Stat
Sci 14(4):382–401
Horton R, Little C, Gornitz V, Bader D, Oppenheimer M (2015)
New York City panel on climate change report chapter2: sea
level rise and coastal storms. Ann N Y Acad Sci 1336:36–44
Huber M, Caballero R (2011) The early Eocene equable climate
problem revisited. Clim Past 7:603–633
Hughes JK etal (2007) Investigating early hominin dispersal pat-
terns: developing a framework for climate data integration. J
Hum Evol 53(5):465–474
IPCC (2013) Climate Change 2013: the physical science basis. In:
Stocker TF, Qin D, Plattner G-K, Tignor M, Allen SK, Bos-
chung J, Nauels A, Xia Y, Bex V, Midgley PM (eds) Contribu-
tion of Working Group 1 to the Fifth Assessment Report of
the Intergovernmental Panel on Climate Change. Cambridge
University Press, Cambridge, p 1535
Jackson ST, Blois JL (2015) Community ecology in a changing envi-
ronment: Perspectives from the Quaternary. Proc Natl Acad
Sci 112:4915–4921
Janson L, Rajaratnam B (2014) A methodology for robust multiproxy
paleoclimate reconstructions and modeling of temperature con-
ditional quantiles. J Am Stat Assoc 109(505):63–77
Jouzel J etal (2007) Orbital and millennial Antarctic climate vari-
ability over the last 800,000 years. Science 317:793–796
Kageyama M etal (2012) Mid-Holocene and Last Glacial Maximum
climate simulations with the IPSL model part II: model-data
comparisons. Clim Dyn 40:1–27
Kageyama M etal (2018) The PMIP4 contribution to CMIP6—Part
1: overview and over-arching analysis plan. Geosci Model Dev
11:1033–1057
Kennett JP (1977) Cenozoic evolution of Antarctic glaciation, the
circum-Antarctic oceans and their impact on global paleocean-
ography. J Geophys Res 82:3843–3859
Klink HJ (2008) Vegetationsgeographie, 4th edn. Westermann,
Braunschweig
Kopp RE etal (2010) The impact of Greenland melt on regional sea
level: a partially coupled analysis of dynamic and static equi-
librium effects in idealized water-hosing experiments. Clim
Change 103:619–625
Kozar ME, Mann ME, Emanuel KA, Evans JL (2013) Long-term
variations of North Atlantic tropical cyclone activity down-
scaled from a coupled model simulation of the last millennium.
J Geophys Res 118:13383–13392
Kutzbach JE (1980) Estimates of past climate at Paleolake Chad,
North Africa, based on a hydrological and energy-balance
model. Quat Res 14(2):210–223
Kutzbach JE, Street-Perrott FA (1985) Milankovitch forcing of fluc-
tuations in the level of tropical lakes from 18 to 0 kyr BP.
Nature 317:130–134
Larrasoaña JC, Roberts AP, Rohling EJ (2013) Dynamics of Green
Sahara Periods and their role in Hominin evolution. PLoS One
8(10):e76514
Laskar J, Fienga A, Gastineau M, Manche H (2011) La2010: a new
orbital solution for the long-term motion of the Earth. Astron
Astrophys 532:A89
Lawing AM, Polly PD (2011) Pleistocene climate, phylogeny, and
climate envelope models: an integrative approach to better
understand species’ response to climate change. PLoS One
6:e28554
Le Quéré C etal (2009) Trends in the sources and sinks of carbon
dioxide. Nat Geosci 2:831–836
Lemoalle J, Bader JC, Leblanc M, Sedick A (2012) Recent changes in
Lake Chad: Observations, simulations and management options
(1973–2011). Glob Planet Change 80–81:247–254
Lenton TM, Daines SJ (2017) Biogeochemical transformations in the
history of the ocean. Ann Rev Mar Sci 9:31–58
Lindborg T etal (2005) Climate change and landscape development
in post-closure safety assessment of solid radioactive waste dis-
posal: results of an initiative of the IAEA. J Environ Radioact
183:41–53
Lisiecki LE, Raymo ME (2005) A Pliocene-Pleistocene stack of 57
globally distributed benthic δ18O records. Paleoceanography
20:PA1003
Lunt DJ, Foster GL, Haywood AM, Stone EJ (2008) Late Pliocene
Greenland glaciation controlled by a decline in atmospheric CO2
levels. Nature 454:1102–1105
Lunt DJ, Haywood AM, Schmidt GA, Salzmann U, Valdes PJ, Dowsett
HJ (2010a) Earth system sensitivity inferred from Pliocene mod-
elling and data. Nat Geosci 3(1):60–64
Lunt DJ, Flecker RM, Clift PD (2010) The impacts of Tibetan uplift
on palaeoclimate proxies in C.D.P.R Tada and H. Zheng (eds).
Monsoon Evolution and Tectonic–Climate Linkage in Asia, vol
342, pp 279–291
Lunt DJ etal (2017) DeepMIP: experimental design for model simula-
tions of the EECO, PETM, and pre-PETM. Geosci Model Dev
10:889–901
Manabe S, Terpstra TB (1974) Effects of mountains on general circu-
lation of atmosphere as identified by numerical experiments. J
Atmos Sci 31(1):3–42
Mannshardt E, Craigmile PF, Tingley MP (2013) Statistical modeling
of extreme value behavior in North American tree-ring density
series. Clim Change 117(4):843–858
Martinez-Meyer E, Townsend Peterson A, Hargrove WW (2004) Eco-
logical niches as stable distributional constraints on mammal
species, with implications for Pleistocene extinctions and cli-
mate change projections for biodiversity. Glob Ecol Biogeogr
13:305–314
Maslin MA, Christensen B (2007) Tectonics, orbital forcing, global cli-
mate change, and human evolution in Africa: introduction to the
African paleoclimate special volume. J Hum Evol 53(5):443–464

What can Palaeoclimate Modelling doforyou?
1 3
Masson-Delmotte V etal (2006) Past and future polar amplification
of climate change: climate model intercomparisons and ice-core
constraints. Clim Dyn 26:513
McClymont E, Dekens P, Dowsett H, Dupont L, Haywood A, Rosell-
Melé A, Salzmann U (2015) Pliocene climate variability over
glacial-interglacial timescales—(PlioVAR) working group. Past
Glob Changes Mag 23(2):82
McKinney ML (1997) Extinction vulnerability and selectivity: combin-
ing ecological and paleontological views. Annu Rev Ecol Syst
28:495–516
Montanez IP etal (2016) Climate, pCO2 and terrestrial carbon cycle
linkages during late Palaeozoic glacial-interglacial cycles. Nat
Geosci 9:824
Moritz C etal (2009) Identification and dynamics of a cryptic suture
zone in tropical rainforest. Proc Biol Sci 276(1660):1235–1244
Mudelsee M, Raymo ME (2005) Slow dynamics of the Northern
Hemisphere glaciation. Paleoceanogr Paleoclimatol 20:4022
Myers CE, Stigall AL, Lieberman BS (2015) PaleoENM: applying
ecological niche modeling to the fossil record. Paleobiology
41:226–244
Nogués-Bravo D, Rodríguez J, Hortal J, Batra P, Araújo MB (2008)
Climate change, humans, and the extinction of the woolly
mammoth. PLoS Biol 6:e79
North Greenland Ice Core Project Members (2004) High-resolution
record of Northern Hemisphere climate extending into the last
interglacial period. Nature 431:147–151
O’Loughlin J, Linke AM, Witmer FDW (2014) Effects of tem-
perature and precipitation variability on the risk of violence
in sub-Saharan Africa, 1980–2012. Proc Natl Acad Sci
111(47):16712–16717
Otto-Bliesner BL etal (2006) Simulating Arctic climate warmth
and icefield retreat in the Last Interglaction. Science
311(5768):1751–1753
Otto-Bliesner BL etal (2014) Coherent changes of southeastern
equatorial and northern African rainfall during the last degla-
ciation. Science 346:1223–1227
Palmer C (1990) Hybrids—a critical force in the application of infor-
mation technology in the nineties. J Inf Technol 5:232–235
Panitz S, Salzmann U, Risebrobakken B, De Schepper S, Pound MJ,
Haywood AM, Lunt DJ, Dolan AM (2018) Orbital, tectonic
and oceanographic control of Pliocene climate and atmos-
pheric circulation in Arctic Norway. Glob Planet Change
161:183–193
Parmesan C, Yohe GA (2003) globally coherent fingerprint of climate
change impacts across natural systems. Nature 421:37–42
Parrish JT, Curtis RL (1982) Atmospheric circulation, upwelling, and
organic rich rocks in the Mesozoic and Cenozoic. Palaeogeogr
Palaeoclimatol Palaeoecol 40:31–66
Pearson PN, Ditchfield PW, Singano J, Harcourt-Brown KG, Nicholas
CJ, Olsson RK, Shackleton NJ, Hall MA (2001) Warm tropi-
cal sea surface temperatures in the Late Cretaceous and Eocene
epochs. Nature 413:481
Peng Y, Shen C, Cheng H, Xu Y (2014) Modelling of severe persistent
droughts over eastern China during the last millennium. Clim
Past 10:1079–1091
Peterson AT, Soberón J, Pearson RG, Anderson RP, Martínez-Meyer
E, Nakamura M, Araújo MB (2011) Ecological niches and geo-
graphic distributions (MPB-49). Princeton University Press,
Princeton, USA
Potts R, Faith T (2015) Alternating high and low climate variability:
the context of natural selection and speciation in Plio-Pleistocene
hominin evolution. J Hum Evol 87:5–20
Pound MJ, Haywood AM, Salzmann U, Riding JB, Lunt DJ, Hunter S
(2011) A Tortonian (Late Miocene, 11.61–7.25Ma) global veg-
etation reconstruction. Palaeogeogr Palaeoclimatol Palaeoecol
300(1–4):29–45
Pound MJ, Haywood AM, Salzmann Riding JB (2012) Global veg-
etation dynamics and latitudinal temperature gradients dur-
ing the mid to late Miocene (15.97–5.33Ma). Earth-Sci Rev
112(1–2):1–22
Prescott GW etal (2012) Quantitative global analysis of the role of
climate and people in explaining late Quaternary megafaunal
extinctions. Proc Natl Acad Sci 109(12):4527–4531
Prinn RG (2013) Development and application of earth system models.
Proc Natl Acad Sci 110(1):3673–3680
Ramstein G, Fluteau F, Besse J, Joussaume S (1997) Effect of orog-
eny, plate motion and land-sea distribution on Eurasian climate
change over the past 30 million years. Nature 386:788–795
Ritz C et al (2015) Potential sea-level rise from Antarctic ice-
sheet instability constrained by observations. Nature
528(7580):115–118
Rohling EJ etal (2012) Making sense of palaeoclimate sensitivity.
Nature 491:683–691
Royer D etal (2004) CO2 as a primary driver of Phanerozoic climate.
GSA Today 14:3–7
Salzmann U, Haywood AM, Lunt DJ, Valdes PJ, Hill DJ (2008) A new
global biome reconstruction and data-model comparison for the
middle Pliocene. Glob Ecol Biogeogr 17:432–447
Sarmiento JL etal (2004) Response of ocean ecosystems to climate
warming. Global Biogeochem Cycles 18:GB3003
Saupe EE etal (2014) Macroevolutionary consequences of profound
climate change on niche evolution in marine molluscs over the
past three million years. Proc R Soc B Biol Sci 281:20141995
Saupe EE etal (2015) Niche breadth and geographic range size as
determinants of species survival on geological time scales. Glob
Ecol Biogeogr 24:1159–1169
Scotese CR, Summerhayes CP (1986) Computer model of palaeocli-
mate predicts coastal upwelling in the Mesozoic and Cenozoic.
Geobyte 1(3):28–44
Sepulchre P, Schuster M, Ramstein G, Krinnezr G, Girard J-F, Vignaud
P, Brunet M (2008) Evolution of Lake Chad Basin hydrology
during the mid-Holocene: a preliminary approach from lake to
climate modelling. Glob Planet Change 61:41–44
Sexton PF, Wilson PA, Pearson PN (2006) Microstructural and geo-
chemical perspectives on planktic foraminiferal preservation:
“Glassy” versus “Frosty”. Geochem Geophys Geosyst 7:Q12P19
Solomon S (ed) (2007) Contribution of Working Group I to the Fourth
Assessment Report of the Intergovernmental Panel on Climate
Change. Cambridge University Press, Cambridge
Stigall AL (2014) When and how do species achieve niche stability
over long time scales? Ecography (Cop.) 37:1123–1132
Sutter J, Gierz P, Grosfeld K, Thoma M, Lohmann G (2016) Ocean
temperature thresholds for Last Interglacial West Antarctic Ice
Sheet collapse. Geophys Res Lett 43:2675–2682
Svenning J-C, Fløjgaard C, Marske KA, Nógues-Bravo D, Normand S
(2011) Applications of species distribution modeling to paleo-
biology. Quat Sci Rev 30:2930–2947
Svenning J-C, Eiserhardt WL, Normand S, Ordonez A, Sandel B
(2015) The influence of paleoclimate on present-day patterns
in biodiversity and ecosystems. Annu Rev Ecol Evol Syst
46:551–572
Tabor CR, Poulsen CJ, Lunt DJ, Rosenbloom NA, Otto-Bliesner BL,
Markwick PJ, Brady EC, Farnsworth A, Feng R (2016) The cause
of late Cretaceous cooling: a multi-model/proxy comparison.
Geology 44:963–966
Tallavaara M etal (2015) Human population dynamics in Europe
over the Last Glacial Maximum. Proc Natl Acad Sci
112(27):8232–8237
Tarasov PE etal (2000) Last glacial maximum biomes reconstructed
from pollen and plant macrofossil data from northern Eurasia. J
Biogeogr 27:609–620

A.M.Haywood et al.
1 3
Taylor etal (2016) Enhanced weathering strategies for stabilizing
climate and averting ocean acidification. Nat Clim Change
6(4):402–406
Thomas CD etal (2004) Extinction risk from climate change. Nature
427:145–148
Tierney JE, Pausata FSR, deMenocal PB (2017) Rainfall regimes of
the Green Sahara. Sci Adv 3:1
Timmermann A, Friedrich T (2016) Late Pleistocene climate drivers
of early human migration. Nature 538(7623):92–95
Tjallingii R, Claussen M, Stuut J-BW, Fohlmeister J, Jahn A, Bickert
T, Lamy F, Röhl U (2008) Coherent high and low-latitude control
of the northwest African hydrological balance. Nat Geosci 1:670
Troll C (1948) Der asymmetrische Aufbau der Vegetationszonen und
Vegetationsstufen auf der Nord- und Südhalbkugel. Ber Geobot
Inst Rübel, Zürich, pp 46–83
Urban MC (2015) Climate change. Accelerating extinction risk from
climate change. Science 348:571–573
Van Andel TH, Davies W (eds) (2003) Neanderthals and Modern
Humans in the European Landscape of the Last Glaciation:
Archaeological results of the stage 3 project. McDonald Institute
for Archaeological Research Monographs, Cambridge
Varela S, Lobo JM, Rodríguez J, Batra P (2010) Were the Late Pleis-
tocene climatic changes responsible for the disappearance of
the European spotted hyena populations? Hindcasting a species
geographic distribution across time. Quat Sci Rev 29:2027–2035
Veloz SD etal (2012) No-analog climates and shifting realized niches
during the late quaternary: implications for 21st-century pre-
dictions by species distribution models. Glob Change Biol
18:1698–1713
Von der Heydt AS etal (2016) Lessons on climate sensitivity from past
climate changes. Curr Clim Change Rep 2(4):148–158
Vrba E (1995) On the connections between paleoclimate and evolution.
In: Vrba GHDES, Partridge TC, Burckle LH (eds) Paleoclimate
and evolution; with emphasis on human origins. Yale University
Press, New Haven, pp 24–48
Williams JW, Jackson ST (2007) Novel climates, no-analog commu-
nities, and ecological surprises. Front Ecol Environ 5:475–482
Williams JW, Jackson ST, Kutzbach JE (2007) Projected distributions
of novel and disappearing climates by 2100 AD. Proc Natl Acad
Sci USA 104:5738–5742
Yau AM etal (2016) Reconstructing the last interglacial at Sum-
mit, Greenland: insights from GISP2. Proc Natl Acad Sci
113(35):9710–9715
Zachos JC, Dickens GR, Zeebe RE (2008) An early Cenozoic perspec-
tive on greenhouse warming and carbon-cycle dynamics. Nature
451:279–283