Content uploaded by Mahesh Sankaran
Author content
All content in this area was uploaded by Mahesh Sankaran on Feb 26, 2014
Content may be subject to copyright.
effects have the potential to be influential agents
of natural selection (25). Imbalances of expec-
tation and reward may therefore have broad
effects on health and physiology in humans and
may represent a powerful evolutionary force in
nature.
References and Notes
1. J. Apfeld, C. Kenyon, Nature 402, 804–809
(1999).
2. S. Lib ert et al., Science 315, 1133–1137 (2007).
3. N. J. Linford, T. H. Kuo, T. P. Chan, S. D. Pletcher,
Annu. Rev. Cell Dev. Biol. 27, 759–785 (2011).
4. P. C. Poon, T. H. Kuo, N. J. Linford, G. Roman,
S. D. Pletcher, PLOS Biol. 8, e1000356 (2010).
5. E. D. Sm ith et al., BMC Dev. Biol. 8, 49 (2008).
6. J. Alc edo, C. Kenyon, Neuron 41,45–55 (2004).
7. S. J. Lee, C. Kenyon, Curr. Biol. 19, 715–722
(2009).
8. R. Xiao et al., Cell 152, 806–817 (2013).
9. R. M. Sa polsky, Science 308, 648–652 (2005).
10. L. Partridge, N. H. Barton, Nature 362, 305–311
(1993).
11. L. Partridge, D. Gems, D. J. Withers, Cell 120, 461–472
(2005).
12. J. C. Billeter, J. Atallah, J. J. Krupp, J. G. Millar,
J. D. Levine, Nature 461, 987–991 (2009).
13. J. F. Ferveur, Behav. Genet. 35, 279–295 (2005).
14. J. F. Ferveur et al., Science 276, 1555–1558 (1997).
15. M. P. Fer nández et al., PLOS Biol. 8, e1000541
(2010).
16. M. C. Larsson et al., Neuron 43, 703–714 (2004).
17. W. Boll, M. Noll, Development 129, 5667–5681
(2002).
18. R. Thistle, P. Cameron, A. Ghorayshi, L. Dennison,
K. Scott, Cell 149, 1140–1151 (2012).
19. G. Shohat-Ophir, K. R. Kaun, R. Azanchi,
H. Mohammed, U. Heberlein, Science 335,
1351–1355 (2012).
20. S. P. Kalra, J. T. Clark, A. Sahu, M. G. Dube, P. S. Kalra,
Synapse 2, 254–257 (1988).
21. M. Heil ig, Neuropeptides 38, 213–224 (2004).
22. M. Ashburner, K. Golic, R. S. Hawley, Drosophila:
A Laboratory Handbook (Cold Spring Harbor Laboratory
Press, Cold Spring Harbor, NY, ed. 2, 2004).
23. G. Landis, J. Shen, J. Tower, Aging 4, 768–789
(2012).
24. T. H. Kuo et al., PLOS Genet. 8, e1002684 (2012).
25. J. W. McGlothlin, A. J. Moore, J. B. Wolf, E. D. Brodie 3rd,
Evolution 64, 2558–2574 (2010).
Acknowledgments: We thank the members of the Pletcher
laboratory for Drosophila husbandry, N. Linford for comments
on the revision, P. J. Lee for figure illustration, and members
of the Dierick and Pletcher laboratories for suggestions on
experiments and comments on the manuscript. Supported
by NIH grants R01AG030593, TR01AG043972, and
R01AG023166, the Glenn Foundation, the American
Federation for Aging Research, and the Ellison Medical
Foundation (S.D.P.); Ruth L. Kirschstein National Research
Service Award F32AG042253 from the National Institute on
Aging (B.Y.C.); NIH grant T32AG000114 (B.Y.C.); NIH grants
T32GM007863 and T32GM008322 (Z.M.H.), a Glenn/AFAR
Scholarship for Research in the Biology of Aging (Z.M.H.); NSF
grant IOS-1119473 (H.A.D.); and the Alexander von Humboldt
Foundation and Singapore National Research Foundation grant
RF001-363 (J.Y.Y.). This work made use of the Drosophila
Aging Core of the Nathan Shock Center of Excellence in the
Biology of Aging, funded by National Institute on Aging grant
P30-AG-013283. RNA-seq expression data are provided in
table S1. The funders had no role in study design, data
collection and analysis, decision to publish, or preparation
of the manuscript. The authors declare that they have no
competing interests. C.M.G., T.-H.K., Z.M.H., and S.D.P.
conceived and designed the experiments; C.M.G., T.-H.K.,
Z.M.H., B.Y.C., J.Y.Y., H.A.D., and S.D.P. performed the
experiments; C.M.G., T.-H.K., Z.M.H., B.Y.C., J.Y.Y., and
S.D.P. analyzed the data; and C.M.G., T.-H.K., J.Y.Y., H.A.D.,
and S.D.P. wrote the paper.
Supplementary Materials
www.sciencemag.org/content/343/6170/544/suppl/DC1
Materials and Methods
Figs. S1 to S14
Table S1
References (26–28)
16 July 2013; accepted 31 October 2013
Published online 28 November 2013;
10.1126/science.1243339
Savanna Vegetation-Fire-Climate
Relationships Differ Among Continents
CarolineE.R.Lehmann,
1,2
*T. Michael Anderson,
3
Mahesh Sankaran,
4,5
Steven I. Higgins,
6,7
Sally Archibald,
8,9
William A. Hoffmann,
10
Niall P. Hanan,
11
Richard J. Williams,
12
Roderick J. Fensham,
13
Jeanine Felfili,
14
Lindsay B. Hutley,
15
Jayashree Ratnam,
4
Jose San Jose,
16
Ruben Montes,
17
Don Franklin,
15
Jeremy Russell-Smith,
15
Casey M. Ryan,
2
Giselda Durigan,
18
Pierre Hiernaux,
19
Ricardo Haidar,
14
DavidM.J.S.Bowman,
20
William J. Bond
21
Ecologists have long sought to understand the factors controlling the structure of savanna
vegetation. Using data from 2154 sites in savannas across Africa, Australia, and South America,
we found that increasing moisture availability drives increases in fire and tree basal area, whereas
fire reduces tree basal area. However, among continents, the magnitude of these effects varied
substantially, so that a single model cannot adequately represent savanna woody biomass
across these regions. Historical and environmental differences drive the regional variation in
the functional relationships between woody vegetation, fire, and climate. These same differences
will determine the regional responses of vegetation to future climates, with implications for
global carbon stocks.
Savannas cover 20% of the global land sur-
face and account for 30% of terrestrial net
primary production (NPP) and the vast ma-
jority of annual global burned area (1–3). Savanna
ecosystem services sustain an estimated one-fifth
of humans, and savannas are also home to most
of the remaining megafauna (1). Tropical savanna
is characterized by the codominance of C
3
trees
and C
4
grasses that have distinct life forms and
photosynthetic mechanisms that respond differ-
ently to environmental controls (4). Examples
include the differing responses of these func-
tional types to temperature and atmospheric CO
2
concentrations, predisposing savannas to altera-
tions in structure and extent in the coming cen-
tury (4–6).
Tropical savannas are defined by a contin-
uous C
4
herbaceous layer, with a discontinuous
stratum of disturbance-tolerant woody species
(7). Although savanna tree cover varies greatly in
space and time (8,9), the similarities in structure
among the major savanna regions of Africa,
Australia,andSouthAmericahaveledtoan
assumption that the processes regulating veg-
etation structure within the biome are equiva-
lent (10,11). Current vegetation models treat
savannas as a homogenous entity (12,13). Recent
studies, however, have highlighted differences
in savanna extent across continents (14,15), and
it remains unknown how environmental drivers
interact to determine the vegetation dynamics
and limits of the biome (10,14,15).
We sought universal relationships between
savanna tree basal area (TBA, m
2
ha
−1
), a key
metric of woody biomass within an ecosystem,
and the constraints imposed by resource availa-
bility (moisture and nutrients), growing condi-
tions (temperature), and disturbances (fire).
Ecologists have devoted considerable effort to
the identification of universal relationships to de-
scribe the structure and function of biomes (16).
However, it has not been clear whether such re-
lationships exist. Any such relationships may
be confounded by the unique evolutionary and
environmental histories of each ecological set-
ting (11).
Across Africa and Australia, TBA scales sim-
ilarly with rainfall, but the intercepts and the 95th
quantile differ substantially (Fig. 1, A to C). On
average, at a given level of moisture availability,
TBA is higher in Africa and lower in Australia.
However, in South America there is almost no
relationship between rainfall and TBA, which is
probably in part attributable to the narrow range
of rainfall that savanna occupies on this continent
(Fig. 2). Further, across the observed range of
rainfall, the upper limits of TBA increase linearly
with effective rainfall for Australian savannas
(Fig. 1B) but show a saturating response in
African and South American savannas (Fig. 1, A
and C). When TBA is used to estimate above-
ground woody biomass (AWB) (17), the large
differences in intercepts between Africa and
Australia are reduced but substantial differences
in the limits remain (fig. S1, A to C). By con-
31 JANUARY 2014 VOL 343 SCIENCE www.sciencemag.org
548
REPORTS
verting TBA to AWB, we attempted to quantify
how variation in biomass allometry, modal tree
size, maximum tree size, and the mean number of
stems per hectare affects our interpretation of
the functional relationships between savanna
woody vegetation, climate, and fire. These re-
gional differences imply that savanna vegetation
dynamics are region-specific and are influenced
by both regional climates and the allometric traits
specific to the woody species of each region
(17). We interpret the fact that so few sites reach
the maximum values as being partially a result of
variation in soil properties and disturbance pro-
cesses. Fire is a prevalent agent of vegetation
change, as shown by experimental, landscape,
and modeling studies (8,15,18).
To investigate the drivers underlying the ob-
served continental differences in TBA, we con-
structed a conceptual model of the determinants
of woody vegetation structure based on a priori
hypotheses about the direct and indirect effects
of climate, soils, and fire on TBA (Fig. 3A)
(1, 8, 10, 14). The model estimates the effects of
resource availability (moisture availability and soil
properties), growing conditions (temperature), and
disturbance (fire) on TBA. Globally, data avail-
ability on herbivore abundance is sparse and un-
reliable and, as a result, we could not include
herbivore abundance as a predictor (17). We used
our conceptual model to develop a series of struc-
tural equation models (SEMs) to quantify the re-
sponse of TBA to functionally related composite
variables (17). Composite variables in our analy-
sis were (i) moisture availability, composed of
effective rainfall, rainfall seasonality, and Foley’s
drought index; (ii) soil properties, composed of
percent of organic carbon and percent of sand;
and (iii) temperature, composed of mean annual
temperature and annual temperature range [de-
scribedindetailin(17)]. Specifically, our model
allowed us to test the extent to which TBA is
directly determined by climate and edaphic fac-
tors, versus the extent to which TBA is indirect-
ly effected by these factors through their effects
on fire.
Our results highlight that interactions among
moisture availability, fire, and TBA are a defining
characteristic of savannas. Increasing moisture
availability simultaneously promotes increases
in both TBA and grass-fueled fire frequencies
(Fig. 3, B to D). As moisture availability increases,
mean TBA can approach a maximum value,
which is different in each region (Fig. 1, A to C).
Fire, by preventing the accumulation of TBA,
generally maintains TBA below a maximum value.
Therefore, on a qualitative level there is universality
in savanna vegetation dynamics, evidenced by
our analysis of each region identifying the same
network of factors influencing TBA. The excep-
tion was that soil properties influenced TBA in
South America and did not influence fire fre-
quency, in contrast to Africa and Australia.
Interactions between moisture, fire, and TBA
are unequal across continents. Moisture availa-
bility explains approximately two to three times
more of the variation in both TBA and fire fre-
quency in Africa and Australia as compared to
South America (Fig. 3). Similarly, the negative
effect of fire explains 1.5 to 2.5 times more of
the variation in TBA in Africa as compared with
South America or Australia, with only a very weak
effect of fire on TBA is in Australia (Fig. 3). Our
findings are consistent with studies that have
shown that the importance of the effect of fire
on TBA is conditional on seedling and sapling
growth rates, fire resilience traits, and fire in-
tensity (10,18,19). Woody plant growth rates de-
termine the post-fire rates of TBA accumulation,
Fig. 1. Change in TBA of
savannas relative to ef-
fective rainfall. The rela-
tionships between TBA and
effective rainfall (in mil-
limeters per year) across
(A)Africa[coefficientofde-
termination (r
2
) = 0.203,
F(1, 363) = 92.4, Pvalue =
<0.001]; (B) Australia [r
2
=
0.385, F(1, 1485) = 930.9,
Pvalue = < 0.001]; and
(C) South America [r
2
=
0.008, F(1, 300) = 2.6, P
value = 0.111] are shown.
Also depicted are the piece-
wise quantile fits of the 5th
and 95th quantiles.
TBA (m2 per ha)TBA (m2 per ha)TBA (m2 per ha)
Africa
Australia
Effective Rainfall (mm)
South America
A
B
C
1
Department of Biological Sciences, Macquarie University, New
South Wales 2109, Australia.
2
School of GeoSciences, Univer-
sity of Edinburgh, Edinburgh EH9 3JN, UK.
3
Department of
Biology, Wake Forest University, 226 Winston Hall, Box 7325
Reynolda Station, Winston-Salem, NC 27109, USA.
4
National
Centre for Biological Sciences, Tata Institute of Fundamental
Research, Gandhi Krishi Vignana Kendra, Bellary Road, Bangalore
560 065, India.
5
School of Biology, University of Leeds, Leeds LS2
9JT, UK.
6
Institute for Physical Geography, J. W. Goethe Uni-
versity Frankfurt am Main, Altenhöferallee 1, 60438 Frankfurt,
Germany.
7
Department of Botany, University of Otago, Post Of-
fice Box 56, Dunedin 9054, New Zealand.
8
Natural Resources
and the Environment, Council for Scientific and Industrial
Research, Post Office Box 395, Pretoria, South Africa.
9
School
of Animal, Plant and Environmental Sciences, University of
the Witwatersrand, Post Office WITS, 2050 Johannesburg,
South Africa.
10
Department of Plant Biology, North Carolina
State University, Raleigh, NC 27695, USA.
11
Geographic In-
formation Science Centre of Excellence, South Dakota State
University, Brookings, SD 57007, USA.
12
Commonwealth Sci-
entific and Industrial Research Organisation Ecosystem Sciences,
Tropical Ecosystems Research Centre, PMB 44 Winnellie, North-
ern Territory 0822, Australia.
13
School of Biological Sciences,
University of Queensland, Brisbane, Queensland 4072, Australia.
14
Forestry Department, University of Brasilia, Brasilia 70919-970,
DF–Brazil.
15
Research Institute for Environment and Livelihoods,
Charles Darwin University, Casuarina, Northern Territory 0810,
Australia.
16
Centro de Ecología, Instituto Venezolano de In-
vestigaciones Científicas, Apartado 21827, Caracas 1020-A,
Venezuela.
17
Departamento de Estudios Ambientales, Uni-
versidad Simón Bolívar, Apartado 89000, Caracas 1080-A,
Venezuela.
18
Laboratório de Ecologia e Hidrologia Florestal,
Floresta Estadual de Assis, Instituto Florestal, 19802-970 -
Assis - SP Brazil.
19
Géosciences Environnement Toulouse
Observatoire Midi-Pyrénées, Université de Toulouse, CNRS
31401, Toulouse, France.
20
School of Plant Science, Univer sity of
Tasmania, Hobart, Tasmania 7001, Australia.
21
Department of
Botany, University of Cape Town, Rondebosch 7700, South Africa.
*Corresponding author. E-mail: c.e.r.lehmann@gmail.com
www.sciencemag.org SCIENCE VOL 343 31 JANUARY 2014 549
REPORTS
whereas fire frequency and intensity are a product
of grassy fuels (19). Differences among con-
tinents in the effect of fire on TBA probably
reflect differences in woody plant traits and the
fuel loads and flammability of C
4
grasses.
In Africa, moisture availability has a strong-
ly positive relationship with fire, implying that
fire and the accumulation of C
4
grasses in
African savanna are more tightly controlled by
yearly variation in the timing and amount of
rainfall than in either Australia or South Amer-
ica. These cascading relationships appear weaker
in Australia, and less so in South America, and
can be partially explained by the differences in
the climatic domain occupied by the savanna of
each region (Fig. 2). Thus, continental differ-
ences in TBA and AWB relate to a combination
of differences in the climatic drivers of fire fre-
quency and intensity, as well as to differences
in the growth and fire resilience traits of woody
plants.
In Africa and Australia, temperature has a
strong effect on fire (Fig. 3, B to C, and tables
S1 and S2), probably determined by a composite
of two factors. First, at warmer sites, fuels are
more likely to cure, facilitating more frequent fire
(3). Second, the physiology of C
4
grasses means
Fig. 2. Climate domain of savannas in Africa,
Australia, and South America. The savanna cli-
mate domain relative to (A) mean annual rainfall
versus mean annual temperature and (B) effective
rainfall versus annual temperature range. Black points
represent all vegetated 0.5° grid cells within 30°
of the equator across Africa, Australia, and South
America. Gray points represent all 0.5° grid cells
where savanna is present as in (14). Lines represent
the 95th quantile of the density of these points for
savanna on each continent.
10 15 20 25 30
0 500 1000 2000 3000
Mean Annual Temperature (degrees)
Mean Annual Rainfall (mm)
A
10 15 20 25 30 35
−2000 −1000 0 1000 2000
Annual Temperature Range (degrees)
Effective Rainfall (mm)
Savanna extent
Africa
Australia
South America
B
Temperature
Soil
Moisture
Fire
Tree
basal area
Rainfall - PET
Drought
index
Rainfall
seasonality
Percent
sand
Organic
carbon
Average
temperature
Temperature
range
Tree
basal area
Fire
frequency
Moisture availability
Disturbance
Soil fertility
Plant growing conditions
Woody biomass
A Conceptual model
C Australia
R2 = 0.42
R2 = 0.28
-0.07
0.69
0.44
0.14
0.77
0.26
B Africa
D South America
R2 = 0.21
R2 = 0.11
-0.12
0.34
0.28
ns
ns
0.21
R2 = 0.40
R2 = 0.29
-0.18
0.89
0.73
0.23
0.74
0.32
0.16
ns
ns
Moisture
Fire
frequency
Tree
basal area
Soil
Temperature
Moisture
Fire
frequency
Tree
basal area
Soil
Temperature
Moisture
Fire
frequency
Tree
basal area
Soil
Temperature
Fig. 3. Structural equation modeling of TBA for Africa, Australia, and South
America. Structural equation modeling of TBA for Africa, Australia, and South America. (A)
Conceptual model depicting theoretical relationships among moisture availability, soil
fertility, plant growing conditions (temperature), and disturbance (fire frequency), and their
effects on TBA either directly or indirectly as mediated by fire frequency. (Bto D) The final
model for each continent. Values associated with arrows are absolute path strengths, which
combine positive and negative effects of indicators into a composite effect (17); the arrow
thickness is proportional to the absolute path strength. The arrows from fire to TBA represent
standardized path coefficients and are depicted in gray to express their negative impacts. Full
models results are presented in (17).
31 JANUARY 2014 VOL 343 SCIENCE www.sciencemag.org
550
REPORTS
that they have a higher temperature optima for
photosynthesis and growth (relative to C
3
plants),
facilitating the potential for rapid biomass ac-
cumulation in hot seasonally dry environments
(4,20), conditions that are more extensively found
in the savannas of Africa and Australia (Fig. 2).
In South America, we found a limited ex-
planation of TBA (Fig. 3D). This is consistent
with previous studies that have examined sa-
vanna extent and found that the limited explan-
atory power was not simply a product of South
American savanna being wetter (14,15). A pre-
vious study found that if the climatic range of
South American savanna were projected to Africa
or Australia, the global extent of savanna would
diminish (14). One potential explanation is that
acid and infertile soils may act as constraints on
both the distribution and vegetation structure of
South American savanna, as discussed in numer-
ous studies (14,15,21), although the quality of
global soils data limits any ability to detect the
regional influence of soils.
Taken together, our findings illustrate how a
common set of environmental drivers shape
savannas across the globe in qualitatively similar
ways. However, the quantitative details of how
these factors interact (Fig. 3) and the climatic
domains occupied (Fig. 2) differ substantially
among continents, so that for practical purposes
we must dismiss the use of a single global model
relating savanna TBA and AWB to environ-
mental drivers. Instead, we make a case for
regionally calibrated models to investigate the
response of savanna vegetation to climate change.
For example, we show that our global analysis,
in which the role of continent is ignored, fails
to capture regional differences in the predicted
response to a hypothetical 4°C increase in mean
annual temperature (Fig. 4). In particular, for
Africa our global analysis predicts a net decrease
in woody biomass, whereas the regional model
predicts a net increase (Fig. 4) due to fire-
temperature interactions within our model. Our
regional analyses show that changing climates
could set these three savanna regions on differ-
ent paths of vegetation change.
Why are these structurally similar ecosystems
in different geographic regions regulated in dif-
ferent ways by the same environmental drivers?
The answer may lie in the evolutionary history
of this biome. Tropical savanna is relatively new,
originating with the global expansion of C
4
grasses
3 to 8 million years before the present (22). When
savannas arose, the southern continents had been
separated for >40 million years. C
4
grasses and
the coincident increase in fire frequency (and also
megaherbivory) exerted novel selective pressures
on regional woody floras, while the phylogenetic
and geographic distance between savanna regions
led to the development of analogous but not
identical solutions in woody plants to these new
selective pressures. Today, savanna tree canopies
are dominated by Myrtaceae in Australia and
in Africa by Mimosaceae, Combretaceae, and
Caesalpiniaceae (23). In South America, there is
a mix of dominance, with savanna taxa derived
from forests in the past 10 million years (24). These
distantly related woody taxa are disturbance-
tolerant but differ in their phenology (23), growth
rates (19), resilience to fire (19), canopy archi-
tecture, and biomass allometry (17).
The global ensemble of regions constituting
modern savannas has, over millennia, converged
on a similar open-canopy vegetation structure
due to the evolution and invasion of C
4
grasses
(22) and the resulting ubiquity of disturbance
(3,14). The environmental space occupied by
modern savanna allows for the multiple interac-
tions among moisture availability, temperature,
fire, and vegetation. However, the functional and
architectural traits of the woody species domi-
nating each region determine the form and strength
of the functional relationships to environmental
drivers. Our data indicate that each savanna re-
gion may respond differently to changes in cli-
mate. Currently, remote sensing evidence suggests
differing trajectories of change in Australian and
southern African savannas (9,25). The one
climate–one vegetation paradigm is an under-
pinning of many global vegetation models (12,13),
and these models are a primary tool for antici-
pating the response of vegetation to future cli-
mates (5,26), but are based on a notion that the
same environmental controls will produce the
same vegetation structure irrespective of envi-
ronmental and evolutionary history. We show
Frequency
Africa only
Frequency
Africa − global
Frequency
Australia only
Frequency
Australia − global
AWB anomaly (tonnes per ha) AWB anomaly (tonnes per ha) AWB anomaly (tonnes per ha)
Frequency
S. America only
Frequency
S. America − global
Fig. 4. Hypothetical shifts in AWB on three continents relative to a 4°C
increase in mean annual temperature. Frequency distributions of the pre-
dicted anomalies in AWB (metric tons per hectare) with a 4°C increase in mean
annual temperature, where a region-specific model and a global model are
compared. Distributions are calculated at a 0.5° resolution. The global model
shows the results of an analysis where “continent”is ignored (table S4).
www.sciencemag.org SCIENCE VOL 343 31 JANUARY 2014 551
REPORTS
that the convergence of structure in savanna con-
ceals substantial differences in the relationships
between savanna woody vegetation, climate, and
fire. Just as the regional evolutionary and envi-
ronmental histories underpin differences in these
relationships, these same differences will deter-
mine the contemporary vegetation response of
each region to future climates.
References and Notes
1. R. J. Scholes, S. R. Archer, Annu. Rev. Ecol. Syst. 28,
517–544 (1997).
2. C. B. Field, M. J. Behrenfeld, J. T. Randerson, P. Falkowski,
Science 281, 237–240 (1998).
3. S. Archibald, C. E. R. Lehmann, J. L. Gómez-Dans,
R. A. Bradstock, Proc. Natl. Acad. Sci. U.S.A. 110,
6442–6447 (2013).
4. W. J. Bond, G. F. Midgley, Glob. Change Biol. 6,
865–869 (2000).
5. S. I. Higgins, S. Scheiter, Nature 488, 209–212
(2012).
6. I. C. Prentice, S. P. Harrison, P. J. Bartlein, New Phytol.
189, 988–998 (2011).
7. J. Ratnam et al., Glob. Ecol. Biogeogr. 20, 653–660
(2011).
8. M. Sankaran et al., Nature 438, 846–849 (2005).
9. B. P.Murphy,C.E.R.Lehmann,J.Russell‐Smith,
M. J. Lawes, J. Biogeogr. 41,133–144 (2014).
10. W. J. Bond, Annu. Rev. Ecol. Evol. Syst. 39, 641–659
(2008).
11. A. K. Knapp et al., Front. Ecol. Environ 2, 483–491
(2004).
12. S. Sitch et al., Glob. Change Biol. 9, 161–185
(2003).
13. R. Fis her et al., New Phytol. 187,666–681 (2010).
14. C. E. R. Lehmann, S. A. Archibald, W. A. Hoffmann,
W. J. Bond, New Phytol. 191, 197–209 (2011).
15. A. C. Staver, S. Archibald, S. A. Levin, Science 334,
230–232 (2011).
16. J. H. Lawton, Oikos 84, 177 (1999).
17. Materials and methods and other information are
available in the supplementary materials on Science
Online.
18. S. I. Higgins, W. J. Bond, W. S. W. Trollope, J. Ecol. 88,
213–229 (2000).
19. W. A. Hoffmann et al., Ecol. Lett. 15, 759–768
(2012).
20. B. Ripley, G. Donald, C. P. Osborne, T. Abraham, T. Martin,
J. Ecol. 98, 1196–1203 (2010).
21. M. Haridasan, in Nature and Dynamics of
Forest-Savanna Boundaries, P. A. Furley, J. Proctor,
J. A. Ratter, Eds. (Chapman & Hall, London, 1992),
pp. 171–184.
22. E. J. Edwards et al., Science 328, 587–591 (2010).
23. D. M. J. S. Bowman, L. Prior, Aust. J. Bot. 53, 379
(2005).
24. M. F. Simon et al., Proc. Natl. Acad. Sci. U.S.A. 106,
20359–20364 (2009).
25. B. J. Wigley, W. J. Bond, M. T. Hoffman, Glob. Change
Biol. 16, 964–976 (2010).
26. G. B. Bonan, Science 320, 1444–1449 (2008).
Acknowledgments: C.L. conceived the project and led the
writing; C.L., T.M.A., M.S., S.I.H., W.A.H., N.H., J.F., G.D., and
S.A. compiled the data; and C.L., T.M.A., M.S., and S.I.H.
analyzed the data. All authors provided new data and
contributed to the writing and/or intellectual development of
the manuscript. M. Crisp, B. Medlyn, and R. Gallagher
provided manuscript feedback. Data used in this study are
available in the supplementary materials and at http://modis.
gsfc.nasa.gov/, http://www.worldclim.org/, and www.fao.org/nr/
land/soils/harmonized-world-soil-database/en/.
Supplementary Materials
www.sciencemag.org/content/343/6170/548/suppl/DC1
Materials and Methods
Figs. S1 to S4
Tables S1 to S5
References (27–188)
Data Sets S1 and S2
18 October 2013; accepted 18 December 2013
10.1126/science.1247355
Effector Specialization in a Lineage of
the Irish Potato Famine Pathogen
Suomeng Dong,
1
Remco Stam,
1
*Liliana M. Cano,
1
Jing Song,
2
†Jan Sklenar,
1
Kentaro Yoshida,
1
Tolga O. Bozkurt,
1
Ricardo Oliva,
1
‡Zhenyu Liu,
2
Miaoying Tian,
2
§Joe Win,
1
Mark J. Banfield,
3
Alexandra M. E. Jones,
1
||
Renier A. L. van der Hoorn,
4,5
Sophien Kamoun
1
¶
Accelerated gene evolution is a hallmark of pathogen adaptation following a host jump. Here,
we describe the biochemical basis of adaptation and specialization of a plant pathogen effector
after its colonization of a new host. Orthologous protease inhibitor effectors from the Irish potato
famine pathogen, Phytophthora infestans, and its sister species, Phytophthora mirabilis, which
is responsible for infection of Mirabilis jalapa, are adapted to protease targets unique to their
respective host plants. Amino acid polymorphisms in both the inhibitors and their target proteases
underpin this biochemical specialization. Our results link effector specialization to diversification
and speciation of this plant pathogen.
The potato blight pathogen, Phytophthora
infestans, is a recurring threat to world ag-
riculture and food security. This funguslike
oomycete traces its origins to Toluca Valley,
Mexico, where it naturally infects wild Solanum
plants (1). In central Mexico, P. infestans co-
occurs with closely related species in a tight
phylogenetic clade known as clade 1c. These
species evolved through host jumps followed by
adaptive specialization on plants belonging to
different botanical families (2,3) (fig. S1). One
species, Phytophthora mirabilis, is a pathogen
of four-o’clock (Mirabilis jalapa). It split from
P. infestans about 1300 years ago (1), and the two
species have since specialized on their Solanum
and Mirabilis hosts. Adaptive evolution after
the host jump has left marks on the genomes
of P. infestans and P. mirabilis (3). Comparative
genomics analyses revealed signatures of accel-
erated evolution, structural polymorphisms, and
positive selection in genes occurring in repeat-
rich genome compartments (3). In total, 345 genes
induced within plants show signatures of posi-
tive selection between the two sister species (3).
These include 82 disease effector genes, rapidly
evolving determinants of virulence that act on
host target molecules. We lack a molecular frame-
work to explain how plant pathogen effectors
adapt and specialize on new hosts, even though
this process affects pathogen evolution and
diversification (4–6).
To gain insight into the molecular patterns
of host adaptation after host jumps, we selected
the cystatinlike protease inhibitor EPIC1, an ef-
fector protein of P. infestans that targets extra-
cellular (apoplastic) defense proteases of the
Solanum hosts (7,8). The epiC1 gene and its
paralogs epiC2A and epiC2B evolved relative-
ly recently in the P. i n f e s t a n s lineage, most likely
as a duplication of the conserved Phytophthora
gene epiC3 (7) (Fig. 1). To reconstruct the evo-
lution of these effectors in the clade 1c species,
we aligned the epiC gene cluster sequences, per-
formed phylogenetic analyses, and calculated var-
iation in selective pressure across the phylogeny
(Fig.1,fig.S2,andtableS1)(9). We detected a
signature of positive selection in the branch of
PmepiC1,theP. mirabilis ortholog of P. infestans
epiC1 [nonsynonymous to synonymous ratio
(w) = 2.52] (Fig. 1B). This is consistent with our
hypothesis that PmEPIC1 evolved to adapt to a
M. jalapa protease after P. mirabilis diverged from
P. infestans.
To test our hypothesis, we first determined
the inhibition spectra of the EPIC effectors using
DCG-04 protease profiling, a method based on
the use of a biotinylated, irreversible protease
inhibitor that reacts with the active site cysteine
of papainlike proteases in an activity-dependent
1
The Sainsbury Laboratory, Norwich Research Park, Norwich
NR4 7UH, UK.
2
Department of Plant Pathology, Ohio Agri-
cultural Research and Development Center, The Ohio State
University, Wooster, OH 44691, USA.
3
Department of Biolog-
ical Chemistry, John Innes Centre, Norwich Research Park,
Norwich NR4 7UH, UK.
4
The Plant Chemetics Laboratory, De-
partment of Plant Sciences, University of Oxford, Oxford OX1
3RB, UK.
5
Plant Chemetics Laboratory, Max Planck Institute for
Plant Breeding Research, 50829 Cologne, Germany.
*Present address: Division of Plant Sciences, University of
Dundee, Invergowrie, Dundee DD2 5DA, UK.
†Present address: Center for Proteomics and Bioinformatics,
School of Medicine, Case Western Reserve University, Cleveland,
OH 44106 USA.
‡Present address: Plant Breeding, Genetics, and Biotechnology,
International Rice Research Institute (IRRI), Los Baños, Laguna,
Philippines.
§Present address: Department of Plant and Environmental
Protection Sciences, University of Hawaii, Honolulu, HI 96822,
USA.
||Present address: School of Life Sciences, Gibbet Hill Campus,
The University of Warwick, Coventry, CV4 7AL, UK.
¶Corresponding author. E-mail: sophien.kamoun@tsl.ac.uk
31 JANUARY 2014 VOL 343 SCIENCE www.sciencemag.org552
REPORTS