31 OCTOBER 2008VOL 322SCIENCEwww.sciencemag.org
is pressed into a polymer film. The films are
monodispersed polystyrenes with average
molecular masses of 9000 kD, 900 kD, and 44
kD, with corresponding Rgvalues of approxi-
mately 84 nm, 26 nm, and 6 nm, respectively,
in the bulk melt state (Rgscales with the
square root of molecular mass). The film
thickness, h, is varied from 170 nm to 36 nm,
becoming thinner than the Rgof the highest-
molecular-mass polystyrenes. The authors
argue that the rheological response where the
thickness of the film is strongly confining rel-
ative to the diameter of the molecule is rele-
vant to an NIL imprint where the mold cavity
is smaller than the Rgof the polymer.
The results are striking. For thick films
(h >> Rg), the resistance to the large-strain
deformation of the polymer melt increases
substantially with the molecular mass of the
polystyrene, consistent with the bulk viscos-
ity. However, when the film thickness is
smaller than the radius of gyration, both the
contact modulus (the resistance to small-
scale elastic deformation) and the forming
stress (the load required to induce large-scale
plastic deformation) are strongly reduced.
For the polystyrene with the highest molecu-
lar mass (9000 kD) in the 36-nm film, which
is approximately one-half the bulk Rg, both
the forming stress and large-strain deforma-
tion resistance are smaller than for the lowest-
molecular-mass polystyrene (44 kD) of the
same thickness. This thickness is still about 6
times the bulk Rgfor the 44-kD polystyrene
and is therefore presumably less confined.
Why such a dramatic reduction of the
forming stress and flow resistance in high-
molecular-mass polymers relative to the bulk
viscosity? The large-strain properties of poly-
mers are dominated by the topological entan-
glements of the transient network established
by the interpenetrating polymer coils (6). For
chains at surfaces, at interfaces, and in thin
films, it has been suggested that the interface
acts as a reflecting plane. The polymer coil is
not allowed to cross the boundary, so it must
“reflect” and remain within the confines of
the interface (7–9). Small-angle neutron scat-
tering measurements on thin polymer films
have shown that the Rgin the plane of the film
is unaffected by thin-film confinement (10).
This means that when the film thickness
decreases and starts to compress the coil in the
vertical direction, the polymer does not
respond by spreading laterally in-plane (see
the figure). Rather, the chain folds back on
itself at the film interface, resulting in the
chain segment’s nearest neighbors belonging
to the same chain, thus decreasing the degree
of coil-coil interpenetration (11).
These arguments are provocative given the
strong correlation between entanglement and
melt rheology. A loss of entanglement would
seem to facilitate flow in polymer thin films.
Although this has been very difficult to prove,
the experimental results of Rowland et al.pro-
vide some of the strongest evidence to date to
support this argument. Si and co-workers (12)
used tensile deformation measurement of
glassy polystyrene to deduce a loss of entan-
glement in thin polymer films, which seems to
support the reports of facilitated flow here.
However, there are also compelling reports
from bubble inflation (13) and surface force
(14) measurements of polymer melts “stiffen-
ing” in very thin films. How this problem
unravels is not only a scientifically intriguing
question, but is also of technical relevance as
manufacturing processes such as NIL evolve
to fabricate nanoscale features from relatively
1. J. D. Ferry, Viscoelastic Properties of Polymers (Wiley,
New York, ed. 3, 1980).
2. H. D. Rowland, W. P. King, J. B. Pethica, G. L. W. Cross,
Science 322, 720 (2008); published online 2 October
3. S. Y. Chou, P. R. Krauss, P. J. Renstrom, Science 272, 85
4. S. H. Ahn, L. J. Guo, Adv. Mater. 20, 2044 (2008).
5. Y. F. Ding et al., Adv. Mater. 19, 1377 (2007).
6. M. S. Green, A. V. Tobolsky, J. Chem. Phys. 14, 80 (1946).
7. E. A. DiMarzio, J. Chem. Phys. 42, 2101 (1965).
8. H. R. Brown, T. P. Russell, Macromolecules 29, 798 (1996).
9. L. J. Fetters, D. J. Lohse, D. Richter, T. A. Witten, A. Zirkel,
Macromolecules 27, 4639 (1994).
10. R. L. Jones, S. K. Kumar, D. L. Ho, R. M. Briber, T. P.
Russell, Nature 400, 146 (1999).
11. P. G. de Gennes, Scaling Concepts in Polymer Physics
(Cornell Univ. Press, Ithaca, NY, 1979).
12. L. Si, M. V. Massa, K. Dalnoki-Veress, H. R. Brown, R. A. L.
Jones, Phys. Rev. Lett. 94, 4 (2005).
13. P. A. O’Connell, G. B. McKenna, Science 307, 1760
14. H. W. Hu, S. Granick, Science 258, 1339 (1992).
graphical distribution (1–3), population col-
lapses or local extinctions (4), failure of large-
scale animal migrations (5), changes in the
seasonal timing of biological events (6), and
changes in food availability and food web
structure. These changes are largely driven
by environmental temperature (1, 7). Exam-
ples from aquatic animal communities show
that study of physiological mechanisms
can help to elucidate these ecosystem chan-
ngoing ecosystem changes in re-
sponse to climate change include
poleward or altitudinal shifts in geo-
gesand to project future ecological trends.
All organisms live within a limited range of
body temperatures, due to optimized structural
and kinetic coordination of molecular, cellular,
and systemic processes. Functional constraints
result at temperature extremes. Increasing
complexity causes narrower thermal windows
for whole-organism functions than for cells
and molecules, and for animals and plants than
for unicellular organisms (8). Direct effects of
climatic warming can be understood through
fatal decrements in an organism’s performance
in growth, reproduction, foraging, immune
competence, behaviors and competitiveness.
Performance in animals is supported by aero-
bic scope, the increase in oxygen consumption
rate from resting to maximal (9). Performance
falls below its optimum during cooling and
warming. At both upper and lower pejus tem-
peratures, performance decrements result as
the limiting capacity for oxygen supply causes
hypoxemia (4, 8) (see the figure, left). Beyond
low and high critical temperatures, only a pas-
sive, anaerobic existence is possible. Fish
rarely exploit this anaerobic range, but inverte-
brates inhabiting the highly variable intertidal
environment use metabolic depression, anaer-
obic energy production, and stress protection
mechanisms to provide short- to medium-term
tolerance of extreme temperatures.
Thermal windows likely evolved to be as
narrow as possible to minimize maintenance
costs, resulting in functional differences,
between species and subspecies in various
climate zones (10–12) and even between pop-
ulations of a species (13); for example, the
Physiology and Climate Change
Hans O. Pörtner1and Anthony P. Farrell2
Studies of physiological mechanisms
are needed to predict climate effects on
ecosystems at species and community levels.
1Animal Ecophysiology, Alfred-Wegener-Institute for Polar
and Marine Research, 27515 Bremerhaven, Germany.
E-mail: firstname.lastname@example.org 2Department of Zoology
and Faculty of Land and Food Systems, University of British
Columbia, Vancouver, V6T 1Z4 Canada.
Published by AAAS
on October 31, 2008
optimal and critical temperatures differ by
2° to 3°C between two sockeye salmon pop-
ulations from the Fraser River in British
Columbia, Canada (5).
Long-term fisheries data revealing climate
impacts on fish stocks have often been related
to food web effects. However, they can also
involve direct warming impacts on individual
species, linked to thermal windows. For ex-
ample, in the German Wadden Sea, growth
and abundance of a nonmigratory eelpout
decreased when summer maximum tempera-
tures surpassed the upper pejus temperature,
with larger individuals affected first (4). In
the Japan Sea, different thermal windows
between sardines and anchovies for individual
growth, gamete production and quality, and
spawning activity caused a regime shift to
anchovies in the late 1990s (14, 15). In the
Fraser and Columbia River systems, warming
has often delayed spawning migrations of
nonfeeding Pacific salmon, potentially caus-
ing loss of fitness (16). Cardiac collapse
above the critical temperature likely brought
on swimming failure and mortality among
Fraser River sockeye in 2004 (5).
The ongoing northward shifts of North
Sea Atlantic cod stocks likely involve both
direct effects on cod and indirect food web
effects. Clear correlation of these shifts with
winter warming indicates greatest sensitiv-
ity of the fishes during their winter repro-
ductive period (1). One reason may be that
the oxygen demand of a 20% gonadal mass
(17) disadvantages mature females by nar-
rowing their thermal window (see the fig-
ure, middle). Also, the enhanced reproduc-
tive capacity of large body size reduces opti-
mal temperatures for growth and increases
heat sensitivity (13). Furthermore, thermal
windows for growing larval fish, which might
be as narrow as those of reproducing adults,
may also reflect limited oxygen supply, when
the developing ventilation and circulatory sys-
tems take over from simple diffusion across
the body surface.
An indirect effect of warming is implied in
the shifted community composition in the
Southern North Sea from larger to smaller
zooplankton prey (18), reducing the food
available to juvenile cod. This shift likely re-
flects different thermal windows for these cope-
pod species as well as for cod and their prey,
given that oxygen-limited thermal tolerance was
recently confirmed for small zooplankter (19).
Such differences between windows may, in gen-
eral, underpin changes in species interactions
and cause shifts in spatial or temporal overlap
(see the figure, right).
Further ecosystem-level responses to cli-
mate change include shifts in the seasonal
timing of recurring processes (20). Earlier
seasonal development of zooplankton or its
grazing later in the year may no longer match
the timing of phytoplankton blooms (6).
Climate could elicit such shifts when warming
cues enter or leave thermal windows earlier in
the year (see the figure, right). As other cues
like seasonal light conditions remain constant,
this may cause previously matched species
interactions to go out of phase; food availabil-
ity may change.
Extending the principle of specialization
on differing thermal windows to interacting
species can help explain changing biogeo-
graphies, community composition, and food
web structures. These changes mostly set in
at the borders of current distributions, where
species operate at the limits of their thermal
windows; acclimatization mechanisms fail
to maintain performance and shift thermal
limits further. Such trends can be compen-
sated for by evolutionary selection for ade-
quate genotypes. However, such adaptation
may be too slow for long-lived species.
Climate change will thus differentially favor
species with wide thermal windows, short
generation times, and a range of genotypes
among its populations.
Carbon dioxide, hypoxia, salinity change,
and eutrophication contribute to ecosystem
responses to climate change (21). Key to set-
ting sensitivity to ocean acidification are the
mechanisms and efficiency of systemic acid-
base regulation (22). Such specific effects of
each stressor will reduce whole-organism per-
formance, especially at extreme temperatures,
thereby narrowing thermal windows and
reducing biogeographical ranges. Studies of
ecosystem consequences of stressors like
ocean acidification through carbon dioxide
should thus consider effects on thermally lim-
ited oxygen supply. The principles elaborated
here may also be applicable to organisms
other than animals and to both aquatic and ter-
restrial ecosystems (23).
1. A. L. Perry, P. J. Low, J. R. Ellis, J. D. Reynolds, Science
308, 1912 (2005).
2. K. Brander et al., ICES Mar. Sci. Symp. 219, 261 (2003).
3. J. M. Grebmeier et al., Science 311, 1461 (2006).
4. H. O. Pörtner, R. Knust, Science 315, 95 (2007).
www.sciencemag.org SCIENCE VOL 32231 OCTOBER 2008
Thermal windows for animals
(may include time dependent shifts through acclimatization)
Onset of anaerobiosis
Onset of denaturation
Onset of loss of performance and abundance
T (°C)Aerobic thermal window
T ranges in climate zones,
T ranges throughout seasons
Scope for aerobic performance
Sequence of life stages
Thermal window widths
across life stages (fishes)
Competition, food web interactions, phenologies
Spring warming cue
Eggs, early larvae
Temperature effects on aquatic animals. The thermal windows of aerobic per-
formance (left) display optima and limitations by pejus (pejus means “turning
worse”), critical, and denaturation temperatures, when tolerance becomes increas-
ingly passive and time-limited. Seasonal acclimatization involves a limited shift or
reshaping of the window by mechanisms that adjust functional capacity, endurance,
or protection (4). Positions and widths of windows on the temperature scale shift
with life stage (middle). Acclimatized windows are narrow in stenothermal species,
or wide in eurytherms, reflecting adaptation to climate zones. Windows still differ
for species whose biogeographies overlap in the same ecosystem (right, examples
arbitrary). Warming cues start seasonal processes earlier (shifting phenology), caus-
ing potential mismatch with processes timed according to constant cues (light).
Synergistic stressors like ocean acidification (by CO2) and hypoxia narrow thermal
windows according to species-specific sensitivities (broken lines), modulating bio-
geographies, coexistence ranges, and other interactions further.
Published by AAAS
on October 31, 2008
31 OCTOBER 2008VOL 322 SCIENCE www.sciencemag.org Download full-text
5. A. P. Farrell et al., Physiol. Biochem. Zool. 81, 697 (2008).
6. K. H. Wiltshire, B. F. J. Manly, Helgol. Mar. Res. 58, 269
7. C. Rosenzweig et al., Nature 453, 353 (2008).
8. H. O. Pörtner, Comp. Biochem. Physiol. 132 A, 739
9. J. R. Brett, Am. Zool. 11, 99 (1971).
10. H. O. Pörtner, Deep Sea Res. II 53, 1071 (2006).
11. N. A. Fangue, M. Hofmeister, P. M. Schulte., J. Exp. Biol.
209, 2859 (2006).
12. J. H. Stillman, Science 301, 65 (2003).
13. H. O. Pörtner et al., Climate Res. 37, 253 (2008).
14. A. Takasuka, Y. Oozeki, I. Aoki, Can. J. Fish. Aquat. Sci.
64, 768 (2007).
15. A. Takasuka, Y. Oozeki, H. Kubota, Mar. Ecol. Prog. Ser.
360, 211 (2008).
16. T. M. Goniea et al., Trans. Am. Fish. Soc. 135, 408
17. R. Dahle, G. L. Taranger , Ø. Karlsen, O. S. Kjesbu, B.
Norberg, Comp. Biochem. Physiol. 136 A, 641 (2003).
18. P. Helaouët, G. Beaugrand, Mar. Ecol. Progr. Ser. 345,
19. O. Pinkhaus et al., Freshw. Biol. 52, 1537 (2007).
20. M. Winder, D. B. Schindler, Ecology 85, 2100
21. J. M Guinotte, V. J. Fabry, Ann. N.Y. Acad. Sci. 1134, 320
22. H. O. Pörtner, M. Langenbuch, B. Michaelidis, J. Geophys.
Res. 110, C09S10, 10.1029/2004JC002561 (2005).
23. J. J. Tewksbury, R. B. Huey, C. A. Deutsch, Science 320,
ter of debate for almost a century. Although
aneuploidy is a hallmark of most tumor cells,
whether it is a cause or a consequence of the
malignant transformation (oncogenesis) has
not been clear. In 1914, German biologist
Theodor Boveri postulated that aneuploidy
arising from altered cell division (mitosis)
might lead to oncogenesis. However, recent
studies with genetically modified organisms
have kept the issue open to argument. Certain
defects in chromosome segregation during
mitosis that lead to aneuploidy can either pro-
mote (1, 2) or inhibit tumor formation (2, 3),
or even have no effect at all (4). On page 703
in this issue, Williams et al. (5) provide an
interesting twist, by showing that harboring an
extra chromosome may or may not drive a
mammalian cell into oncogenesis, depending
on the chromosome itself and on the state
of the cell.
Earlier studies reported the deleterious
he role of aneuploidy—the presence
of an abnormal number of chromo-
somes—in cancer has been at the cen-
effects of aneuploidy during human develop-
ment (causing miscarriages) and in adulthood
(underlying mental retardation). The findings
of Williams et al. are compatible with this
view, showing that having an abnormal num-
ber of chromosomes is initially disadvanta-
geous for mammalian cells. The authors cul-
tured mouse cells that were engineered to
express a specific additional chromosome
(trisomy). These cell lines had decreased rates
of proliferation, and increased cell size and
metabolic rates, all conditions that reduce cell
fitness. However, in some cases, these limita-
tions could be overcome. The ability of a cell
line to proliferate indefinitely in culture
(immortalization) depended on the identity of
the extra chromosome. Certain chromosome
gains accelerated the attainment of immortal-
ization, whereas others delayed or impaired it.
To what extent do these in vitro results
reproduce the survival pressure that somatic
cells undergo in vivo, and their capacity to
adapt to stressful conditions? The mouse
embryonic fibroblasts used by Williams et al.
have higher spontaneous immortalization
rates than other primary mouse or human cells
in culture. Do the effects of aneuploidy in
these fibroblasts occur in other cell types from
which most common tumors arise? Also, the
elegant strategy of chromosomal transloca-
tion used by the authors to simulate increased
chromosome numbers may not strictly repre-
sent all forms of aneuploidy, nor fully recapit-
ulate, from a structural standpoint, the gain or
loss of individual chromosomes.
In any case, Williams et al. propose that
certain gains or losses of specific chromo-
somes are more compatible with cell viability
than others, thus explaining the variable
effects of chromosome gains observed in the
mouse cells. Thus, in a normal cellular con-
text—that is, in the absence of mutations that
predispose a cell for transformation—aneu-
ploidy alone seems an unlikely driver of onco-
genesis. But in a procancerous context, aneu-
ploidy could promote malignant cell transfor-
mation. This hypothesis could be tested by
introducing aneuploidy in immortalized (not
yet transformed) cell lines.
What are the advantages conferred by
aneuploidy in a permissive context? Although
initially less proliferative, aneuploid cells are
inherently unstable, and thus endowed with
increased genomic instability and mutational
rate. This may lead them to acquire the hall-
marks of cancer, such as resistance to cell
The gain or loss of specific chromosomes can
determine whether a cell becomes tumorigenic.
Department of Pathology, New York University School
of Medicine, New York, NY 10016, USA. E-mail: eva.
Normal cell Aneuploidy
Cell metabolism increases
Genomic stability and fitness decrease
Aneuploidy-tolerating mutationProliferation increases
Gains and losses. According to the aneuploidy model of Williams et al., an
abnormal chromosome number may be costly to cell fitness. However, if
mutations arise that allow the cell to adapt to cellular imbalances caused by the
abnormal chromosome content, cells may eventually form tumors.
Published by AAAS
on October 31, 2008