Functional tradeoffs determine species coexistence
via the storage effect
Amy L. Angerta,b,1, Travis E. Huxmanb,c, Peter Chessonb, and D. Lawrence Venableb
aDepartment of Biology, Colorado State University, Fort Collins, CO, 80523; andbDepartment of Ecology and Evolutionary Biology andcB2 Earthscience,
University of Arizona, Tucson, AZ 85721
Communicated by Jose ´ Sarukha ´n, Universidad Nacional Auto ´noma de Me ´xico, Mexico D.F., Mexico, April 28, 2009 (received for review March 23, 2009)
How biological diversity is generated and maintained is a funda-
mental question in ecology. Ecologists have delineated many
mechanisms that can, in principle, favor species coexistence and
hence maintain biodiversity. Most such coexistence mechanisms
require or imply tradeoffs between different aspects of species
performance. However, it remains unknown whether simple func-
tional tradeoffs underlie coexistence mechanisms in diverse natu-
ral systems. We show that functional tradeoffs explain species
differences in long-term population dynamics that are associated
with recovery from low density (and hence coexistence) for a
community of winter annual plants in the Sonoran Desert. We
develop a new general framework for quantifying the magnitude
of coexistence via the storage effect and use this framework to
assess the strength of the storage effect in the winter annual
community. We then combine a 25-year record of vital rates with
morphological and physiological measurements to identify func-
tional differences between species in the growth and reproductive
phase of the life cycle that promote storage-effect coexistence.
Separation of species along a tradeoff between growth capacity
and low-resource tolerance corresponds to differences in demo-
graphic responses to environmental variation across years. Grow-
ing season precipitation is one critical environmental variable
underlying the demographic decoupling of species. These results
demonstrate how partially decoupled population dynamics that
promote local biodiversity are associated with physiological dif-
ferences in resource uptake and allocation between species. These
results for a relatively simple system demonstrate how long-term
community dynamics relate to functional biology, a linkage scien-
tists have long sought for more complex systems.
biodiversity ? coexistence mechanism ? functional trait ? population
dynamics ? specific leaf area
istence rely on ecological differences that enable each species to
recover when perturbed to low density and thus remain in the
community. Some of these coexistence mechanisms, such as dif-
ferential exploitation of multiple limiting resources (1, 2) and
frequency-dependent predation (3), operate independently of fluc-
tuations in the environment. Other stable coexistence mechanisms
depend critically upon environmental fluctuations that allow spe-
cies to recover from low density (4). These include competition/
colonization tradeoffs in a disturbance matrix (5, 6), relative
nonlinearity of competition (7) and the storage effect (8, 9). The
storage effect combines species-specific responses to the environ-
ment and population-dynamic buffering by persistent life history
stages in a way that results in a positive average low-density growth
rate for each species. It is perhaps the dominant fluctuation-
dependent mechanism for organisms in variable environments. Its
role has been explored for diverse groups ranging from freshwater
zooplankton (10) and coral reef fishes (8) to desert annuals (11),
prairie grasses (12) and tropical trees (13), but in no case is the
mechanism underlying species-specific responses to the environ-
ment well-understood. Specifically, temporal environmental varia-
tion increases coexistence through the storage effect when (i)
ow competing species stably coexist is a long-standing eco-
logical problem. All niche-based mechanisms for stable coex-
demographic decoupling of species arises from partially uncorre-
lated responses to environmental variation, (ii) the strength of
competition covaries with environmental conditions, and (iii) cer-
tain life history traits, such as seed banks or long-lived adults, limit
we address the critical challenges of identifying the functional
differences between species that create demographic decoupling
and quantifying their relationship to species coexistence (14).
The winter annual species of the Sonoran Desert have played an
important role in the development and testing of general concepts
they form a mature, persistent community where unpredictable
weather creates substantial demographic variability. Desert annual
germination is controlled by temperature and rainfall, and winter
and summer rains in the Sonoran Desert give rise to distinct winter
and summer annual plant communities. Annuals comprise up to
50% of the desert flora and have been the subject of classic
investigations on physiology and population dynamics (16, 17).
Winter annuals complete their life cycles within a few weeks to
months. Their short life cycles, small size and sessile habit permit
and the observation of multiple generations during the course of a
single long-term project. These qualities enable quantitative esti-
dependent theories. Desert annuals meet a key requirement for
storage-effect coexistence: long-lived seed banks buffer popula-
favor high germination directly result in greater density, creating
positive covariance between an environmental parameter and
competition. A similar environment-competition covariance arises
from varying environmental factors affecting seedling survival and
individual growth, because higher survival and larger survivors
increase demand for resources (19). Although these effects reduce
the ability of a high density species to take advantage of favorable
environment conditions, it does not alter our ability to measure the
relative responses of different species to the environmental condi-
Coexistence is promoted if a set of species is buffered by
persistent seed banks and if species’ environment-competition
covariances decline when their densities decline. These effects
create low-density advantages, and occur when there is sufficient
demographic decoupling between species driven by differences in
their responses to temporally varying physical environmental con-
ditions (15, 19). In desert annuals, such demographic responses can
be suitably divided into germination and fecundity. Decoupling
through germination is a well-known and well-studied scenario by
which low-density advantages are created (9, 15, 20). Decoupling
Author contributions: A.L.A., T.E.H., and D.L.V. designed research; A.L.A., T.E.H., P.C., and
A.L.A., T.E.H., P.C., and D.L.V. analyzed data; and A.L.A., T.E.H., P.C., and D.L.V. wrote the
The authors declare no conflict of interest.
1To whom correspondence should be addressed. E-mail: email@example.com.
This article contains supporting information online at www.pnas.org/cgi/content/full/
July 14, 2009 ?
vol. 106 ?
no. 28 ?
through reproduction, which is the focus of this article, reflects the
combined effects of survival and growth (19), and provides strong
low-density advantages, which we show here (SI Appendix). The
decoupling of reproductive success between species that promotes
species coexistence can be measured as the statistical interaction
between species and time for per germinant fecundity.
magnitude of the storage effect, assess the strength of the storage
effect in a community of Sonoran Desert winter annuals, and
identify functional differences between species in the growth and
reproductive phase of the life cycle that promote storage-effect
coexistence. We have examined interspecific variation in traits
associated with resource uptake and resource allocation for 9
fundamental tradeoffs in plant function create the differential
demographic responses to environmental variation that help main-
tain diverse species assemblages in plant communities. Specifically,
species-by-time interaction for reproduction that contributes quan-
titatively to recovery from low density. Elucidating such functional
links in annual plants should provide insights relevant to less
Results and Discussion
Using a dataset collected annually from 1982 to 2007 from 72
permanently marked quadrats, we obtained estimates of germina-
tion, survival and fecundity, and used these data to estimate
variance components by standard ANOVA methods. Sonoran
Desert winter annual species exhibit striking demographic fluctu-
ations (11) and respond in partially dissimilar ways to yearly
variation, as evidenced by species-by-year interactions for ln-
per germinant fecundity (F165,2926? 7.99, P ? 0.0001) (Table S1).
For our system, the low-density advantage due to germination
variation alone adds 0.025 to r ?, where r ? is the species average
recovery rate from low density (the low-density long-term per
capita population growth rate) (Table S2 and SI Appendix). An
additional increase to r ? of 0.027 comes from the covariance of
germination fraction and reproductive variation, giving a total
storage effect of 0.103 (Table S2). For a system as large as this one
(millions of individuals of most species at one field station),
The observed storage effect is a substantial population growth rate
for a species bouncing back from low density in competition with
other species (equivalent to a doubling time of 7 years).
To understand how differences in functional traits create
population dynamic decoupling, we examined interspecific
variation in traits related to growth, allocation and low-
resource tolerance and related these differences to the species-
by-year interaction for per germinant fecundity. A fundamen-
tal tradeoff thought to be important in plants is that between
the ability to photosynthesize and grow rapidly versus the
ability to withstand the stresses inherent in low resource
environments (21, 22). Such a tradeoff has been described
across life forms, but we find the tradeoff within one functional
guild (Fig. 1) (23). Sonoran Desert winter annuals are arrayed
along a tradeoff between relative growth rate (RGR) and a
measure of intrinsic water-use efficiency (WUE), carbon
isotope discrimination [?, where lower ? indicate higher
intrinsic WUE (24)] (Fig. 1). Species with high RGR exhibit
low WUE, whereas species with high WUE have low RGR.
Our prior work has identified the key morphological and
physiological traits that underlie growth capacity and low-
water tolerance in these species (23, 25). Species that display
high growth capacity allocate a large fraction of biomass to
photosynthetic surfaces and have the ability to rapidly deploy
large leaf area displays to maximize growth after infrequent,
large rainfall events (23). Conversely, species that display high
intrinsic WUE invest a large fraction of leaf nitrogen in the
photosynthetic processes that become limiting at low temper-
atures characteristic of postrainfall periods, which optimizes
carbon assimilation for the short windows of time after small
but relatively frequent rain events (25). To capture this
complexity, we conducted a principal components analysis of
these traits that reflect physiological and morphological ca-
pacities underlying the growth/low-resource tolerance
tradeoff: specific leaf area (SLA), leaf mass ratio (LMR),
relative growth rate plasticity, the ratio of maximum electron
transport to maximum carboxylation velocity (Jmax:VCmax), and
leaf nitrogen content (Nleaf). Scores on the first principal
component contrast species with high leaf mass allocation, leaf
N, and electron transport capacity (which optimizes carbon
assimilation at low temperatures after small rainfall events)
versus species that attain high growth via high leaf area
investment and morphological plasticity after large rainfall
events (Table 1). Species’ pairwise differences in PC I scores
quantify composite differences in the key physiological traits
that underlie the growth rate/low-resource tolerance tradeoff.
To calculate species’ differences in response to year, we first
decomposed the species-by-year interaction for per germinant
fecundity into a residual effect for each species in each year
after removing main effects and sampling error (Fig. 2). We
then squared species’ pairwise differences in residual interac-
tion terms. The average over pairs of species and time of these
squared pairwise differences estimates the species-by-year
interaction component of variance that is used to calculate the
magnitude of the storage effect due to fecundity (SI Appendix).
The average squared differences between species in these
interaction effects were placed in a 9 ? 9 matrix that describes
species’ pairwise contributions to the species-by-year interac-
tion term, and hence, the magnitude of their contributions to
the demographic differences that promote coexistence (Table
S3). Species that respond similarly to yearly variation will tend
to have low differences, whereas species that respond dissim-
ilarly to yearly variation will tend to have high differences.
Using a matrix correlation approach, we find that the matrix
of species differences in PC I scores (Table S4) is highly
correlated with the matrix of contributions to the species-by-
year interaction (Mantel test, P ? 0.0003) (Fig. 3A). The
RGR, in g?g?1?day?1) and low-resource tolerance (intrinsic water-use effi-
ciency, assayed by leaf carbon isotope discrimination, ?, ‰). Species abbrevi-
ations are the first two letters of the genus and specific epithet given in
Materials and Methods. Asterisks (*) denote 2 naturalized species.
www.pnas.org?cgi?doi?10.1073?pnas.0904512106Angert et al.
relationship remains highly significant using a partial Mantel
test with a third matrix of phylogenetic distances (Table S5).
This finding provides a mechanistic explanation of how func-
tional traits underlie species differences in population dynamic
responses to the environment of different years. To see if this
relationship can be captured with simpler integrative mea-
sures, we recalculated the matrix correlations using the PC I
score based on the emergent growth rate/low-resource toler-
ance tradeoff alone. The matrix of species differences along
the RGR-? tradeoff (Table S4) is also significantly related to
the matrix of species contributions to the species-by-year
interaction, but with a lower P value (Mantel test, P ? 0.0410)
(Fig. 3B). Thus, while a simple functional tradeoff is involved
in the demographic decoupling that contributes to coexistence,
explicit recognition of variation in the underlying parameters
suggests a more intricate relationship between physiology and
Temporal demographic decoupling captured by the species-by-
year interaction reflects differences in demographic response to
environmental variation. We investigated the relationship between
demography, traits, and climate variables to determine how the
traits, result in differential performance. Because precipitation
controls the rate and timing of most biological processes in arid
ecosystems (26), we hypothesized that differential response to
precipitation is a major contributor to the species-by-year interac-
tion. Species differ strongly in their demographic responses to
precipitation (species by precipitation, F8,3104? 11.77, P ? 0.0001).
as the slope of the relationship between per germinant fecundity
and growing season precipitation. The matrix of species squared
differences in demographic sensitivity to precipitation (Table S3)
correlates with the matrix of contributions to the species-by-year
interaction (Mantel test, P ? 0.0075). Differences in demographic
sensitivity to other climate parameters, such as growing season
contributions to the species-by-year interaction (Mantel tests: sea-
son length, P ? 0.30; average maximum temperature, P ? 0.50;
average minimum temperature, P ? 0.72). Furthermore, the ma-
trices of species differences in functional traits are significantly
related to the matrix of differential demographic sensitivity to
precipitation (Mantel tests: 5-trait matrix, P ? 0.0068; RGR-?
matrix, P ? 0.0453). Thus, tradeoffs in key physiological processes
that result in different utilization of soil moisture explain demo-
graphic decoupling that is driven by inter-annual variation in
We have shown that the same fundamental tradeoff between
growth capacity and low-resource tolerance that separates life
forms (21, 22) is found within what is commonly considered to be
1 plant functional type. The degree of separation between species
on this tradeoff axis is quantitatively related to the magnitude of
between years, specifically variation in the amount of growing
season precipitation. Incorporation of lower-level functional traits
that describe resource uptake and allocation behavior produces a
more exact description of how physiological differences between
species explain the temporal population dynamics that promote
local biodiversity via the storage effect. These tests have relied on
an extension of storage-effect theory to consider different degrees
of decoupling between different pairs of species (SI Appendix).
Although not required in theory for the storage effect (9), it stands
to reason that more physiologically similar species are demograph-
approaches for understanding the factors driving demographic
Table 1. Trait loadings, species scores, and percent variation
explained by the first principal component of variation (PC I)
in functional traits
PC I results Species or traitAnalysis 1 Analysis 2
54% Variation explained
Analysis 1: leaf mass ratio (LMR), maximum electron transport capacity
(Jmax:VCmax), leaf nitrogen content (Nleaf), specific leaf area (SLA), and growth
plasticity (RGR plasticity). Analysis 2: relative growth rate (RGR) and carbon
isotope discrimination (?; inversely related to intrinsic water-use efficiency).
times fecundity, b) for each species (left axis, lines) and decomposition of the
species-by-year interaction for per germinant fecundity into effects for each
species and year after subtracting species and year main effects and sampling
error (right axes, black bars). If the species-by-year interaction effects are
positive, then lb was greater than expected based on the main effects of
species and year alone. (Lower) Interannual variation in growing season
precipitation (total precipitation from the first germination-inducing rain
until the final reproductive census each year).
Annual per germinant fecundity (survivorship to reproduction, l,
Angert et al.PNAS ?
July 14, 2009 ?
vol. 106 ?
no. 28 ?
decoupling that we have demonstrated here. The results from this
investigation of short-lived plants in extreme environments should
ing of more complex communities.
Materials and Methods
From 1982 to 2007, permanent plots at the University of Arizona Desert
Laboratory (Tucson, AZ) were censused after each rainfall to document
germination, survivorship to reproduction (l), and fecundity (b) of all
winter annual species (18). Species with at least 5 individuals in each of at
of lb: Pectocarya recurvata (Boraginaceae), Erodium cicutarium—
naturalized, Erodium texanum (Geraniaceae), Eriophyllum lanosum, Evax
multicaulis, Stylocline micropoides (Asteraceae), Plantago insularis, Plan-
tago patagonica (Plantaginaceae) and Schismus barbatus—naturalized
During 2004–2005, we measured functional traits of these species plus
Lotus humistratus (Fabaceae) and Pectocarya heterocarpa, which were
abundant that year. We determined relative growth rate (RGR), specific
leaf area (SLA) and leaf mass ratio (LMR) by harvesting up to 30 plants per
species biweekly throughout the season (see ref. 24). We calculated VCmax
(maximum carboxylation rate by Rubisco) and Jmax (maximum light-
saturated electron transport rate) from assimilation versus internal CO2
concentration curves on fully expanded leaves of 3 to 5 individuals per
species (see ref. 26). Leaf nitrogen (Nleaf) and carbon isotope composition
were analyzed at the University of Arizona Geosciences Stable Iso-
tope Facility. Carbon isotope ratios were converted to discrimination val-
ues (?) (27).
We used ANOVA to assess the effects of species, year (or ln-seasonal
precipitation) and their interaction on ln-transformed plot-level per ger-
minant fecundity (lb ? 0.5), weighted by the number of seedlings per
species per plot (PROC GLM, SAS 9.1; SAS Institute). Significance of effects
was tested with Type III sums of squares. We added 0.5 to lb because ln(0)
is undefined, and lb occasionally equalled 0 if no seedlings of a given
germinated but all died before reproduction. Results were qualitatively
identical when different constants were added to per germinant fecundity
(e.g., lb ? 0.17 or lb ? 1). Results also were qualitatively identical when
untransformed data were analyzed with generalized linear models
(gamma distribution, log link function; PROC GLIMMIX). We decomposed
the species-by-year interaction into an interaction effect for each species in
each year based on the following equation: Interaction(Si,Yj) ? LS(Si,Yj) ?
LS(Si) ? LS(Yj) ? LS(grand mean), where LS denotes the least-squares mean
estimate for a particular species, Si, year, Yj, or Si,Yjcombination. We used
the ‘‘LSMEANS’’ statement of PROC GLM to obtain least-squares means for
each Si,Yjand Si,Yjcombination. The interaction(Si,Yj), when squared and
summed over years, divided by the degrees of freedom, and corrected for
sampling error, gives us the species x year interaction component of
) for log per germinant fecundity, ln(lb), used to calculate
the magnitude of the storage effect (Tables S1 and S2 and SI Appendix).
To see how this population dynamic interaction relates to species func-
tional biology, we first expressed it as a species difference matrix. For each
pair of species, we calculated the average of squared differences in their
(excluding any years in which one of the species remained dormant) and
then dividing by the number of years in which both species were observed
in the vegetative phase (n ? 19–23 years). These differences between
species were placed in a 9 ? 9 matrix describing species’ pairwise contri-
butions to the species-by-year interaction term (Table S3, upper diagonal).
Species differences in demographic sensitivity to climate variables were
estimated as squared differences in slopes of individual regressions of ln(lb
? 0.5) versus growing season precipitation (Table S3, lower diagonal),
season length, average maximum temperature or average minimum tem-
perature. Daily precipitation was recorded at the University of Arizona
Desert Laboratory. Daily temperature data were obtained from the Uni-
versity of Arizona weather station ?5 km from the Desert Laboratory
(National Climatic Data Center, National Oceanic & Atmospheric Adminis-
tration, Asheville, NC)
To summarize interspecific variation in functional traits, we conducted
principal component analysis on trait correlation matrices using SAS IN-
SIGHT. The first analysis described species differences in 5 key traits that
underlie the growth capacity/low-resource tolerance tradeoff: SLA, LMR,
RGR plasticity, Jmax:VCmax, and Nleaf.The second analysis described species
differences in position along the emergent tradeoff between RGR and ?.
We ran the principal component analysis using data from all native species
and then manually calculated the 2 naturalized species’ scores on the first
principal component (PC I) by using the standardized regression coeffi-
cients relating each trait to PC I. However, our results do not change when
all species are included in the principal component analysis. For each
analysis, squared differences between species in PC I scores were placed in
a 9 ? 9 matrix of functional trait differences (Table S4).
Associations between trait and demographic difference matrices were
examined using Mantel tests (28) [program supplied to D.L.V. by E. J. Dietz
(Department of Mathematics and Computer Science, Meredith College,
Raleigh, NC) and D. E. Cowley (Department of Fish, Wildlife and Consera-
tion Ecology, New Mexico State University, Las Cruses, NM)]. Correlations
of corresponding cells of each pair of matrices were calculated with
Mantel’s Z. Permutations preserving the dependencies between matrix
elements were performed and the Z statistic was recalculated 4,000 times,
generating a null distribution against which the observed statistic was
We calculated phylogenetic distance matrices to assess the effects of
phylogenetic history on our results. To estimate phylogenetic distances, we
first used the online tool Phylomatic (29) to create a hypothesis of the
relationships among species based on the conservative seed plant tree
available at the Angiosperm Phylogeny Website (30). We created 2 trees,
one with equal branch lengths and one with pseudo branch lengths based
on ages given by Wikstrom et al. (2001) (31). We calculated matrices of
pairwise phylogenetic distances using the phydist function in Phylocom
(32). We then conducted partial Mantel tests to assess the relationship
between the residuals of the demographic and trait difference matrices
after removing phylogenetic distance from each. Partial Mantel tests were
conducted, using the R platform (compilation 2.6.2) and using the package
vegan. With all analyses, the results from partial Mantel tests controlling
for phylogenetic distance were qualitatively identical to results from Man-
tel tests without phylogenetic matrices (Table S5).
ACKNOWLEDGMENTS. We thank S. Adondakis, G. Barron-Gafford, A. Bell,
H. Bruce, T. Caprio, M. Clauss, B. Collins, C. Contreras, J. Cox, M. Davis, N.
Douglas, J. Duke, C. Enquist, K. Gerst, K. Gilliam, C. Golightly, A. Halloran,
A. Hazard, J. Horst, T. Hubbard, R. Janaway, A. Jaksha, G. Ketner, O.
Kougot, S. Kunkel, H. Lawson, K. McCoy, C. McDonald, K. Moriuchi, C. Pake,
M. Pantastico, C. Pearson, J. Pearson, S. Roberts, P. Sanchez, M. Schneider,
S. Stebens, M. Stubbs, A. Tyler, M. Wagenheim, and B. Weeks for assistance
with data collection; B. Igic, S. Kimball, S. Stark, and J. Tewksbury for
providing valuable comments on the manuscript; and N. Swenson and A.
Zanne provided advice on statistical analyses. This work was supported by
the Philecology Foundation and National Science Foundation Grants BSR
9107324, DEB 9419905 (Long Term Research in Environmental Biology),
(Long Term Research in Environmental Biology), DEB 0453781, DEB
0542991, and DEB 0717380.
functional traits that underlie a growth capacity/low-resource tolerance
tradeoff (leaf mass ratio, maximum electron transport capacity, specific leaf
area, relative growth plasticity and leaf nitrogen content) (A) and differences
in position along the first principal component constructed using relative
the pairwise squared difference between 2 species. Significance was tested
with Mantel permutation tests to account for non-independence of data
www.pnas.org?cgi?doi?10.1073?pnas.0904512106Angert et al.
1. Schoener TW (1974) Resource partitioning in ecological communities. Science 185:27–39. Download full-text
(Princeton Univ Press, Princeton).
3. Gendron RP (1987) Models and mechanisms of frequency-dependent predation. Am
4. Chesson P (2000) Mechanisms of maintenance of species diversity. Annu Rev Ecol Syst
species. Proc Natl Acad Sci USA 68:1246–1248.
6. Hastings A (1980) Disturbance, coexistence, history, and competition for space. Theor
Popul Biol 18:363–373.
7. Armstrong RA, McGehee R (1980) Competitive exclusion. Am Nat 115:151–170.
8. Chesson PL, Warner RR (1981) Environmental variability promotes coexistence in
lottery competitive systems. Am Nat 117:923–943.
9. Chesson P (1994) Multispecies competition in variable environments. Theor Popul Biol
10. Caceres CE (1997) Temporal variation, dormancy, and coexistence: A field test of the
storage effect. Proc Natl Acad Sci USA 94:9171–9175.
11. Pake CE, Venable DL (1995) Is coexistence of Sonoran Desert annuals mediated by
temporal variability in reproductive success? Ecology 76:246–261.
has a stabilizing effect on the coexistence of prairie grasses. Proc Natl Acad Sci USA
13. Runkle JR (1989) Synchrony of regeneration, gaps, and latitudinal differences in tree
species diversity. Ecology 70:546–547.
functional traits. Trends Ecol Evol 21:178–185.
in arid and semi-arid environments. Oecologia 141:236–253.
16. Forseth IN, Ehleringer JR, Werk KS, Cook CS (1984) Field water relations of Sonoran
Desert annuals. Ecology 65:1436–1444.
17. Tevis L Jr (1958) A population of desert ephemerals germinated by less than one inch
of rain. Ecology 39:688–695.
18. Venable DL (2007) Bet hedging in a guild of desert annuals. Ecology 88:1086–1090.
of Chicago Press, Chicago), pp 279–306.
20. Facelli JM, Chesson P, Barnes N (2005) Differences in seed biology of annual plants in
arid lands: A key ingredient of the storage effect. Ecology 86:2998–3006.
21. Grime JP (2001) Plant Strategies, Vegetation Processes, and Ecosystem Properties
(Wiley, Chichester, UK), 2nd Ed.
22. Chapin FS, Autumn K, Pugnaire F (1993) Evolution of suites of traits in response to
environmental stress. Am Nat 142:S78–S92.
23. Angert AL, Huxman TE, Barron-Gafford GA, Gerst KL, Venable DL (2007) Linking
Sense and sensitivity. Oecologia 155:441–454.
25. Huxman TE, et al. (2008) Photosynthetic resource-use efficiency and demographic
variability in desert annual plants. Ecology 89:1554–1563.
26. Noy-Meir I (1973) Desert ecosystems: Environment and producers. Annu Rev Ecol Syst
27. Farquhar GD, Ehleringer JR, Hubick KT (1989) Carbon isotope discrimination and
photosynthesis. Annu Rev Plant Physiol Plant Mol Biol 40:503–537.
28. Dietz EJ (1983) Permutation tests for association between two distance matrices. Syst
29. Webb CO, Donoghue MJ (2005) Phylomatic: Tree assembly for applied phylogenetics.
Mol Ecol Notes 5:181–183. Available at www.phylodiversity.net/phylomatic/
30. Stevens PF (2001) Angiosperm Phylogeny web site. Version 9. Available at www.
family tree. Proc R Soc London Ser B 268:2211–2220.
32. Webb CO, Ackerly DD, Kembel SW (2008) Phylocom: Software for the analysis of
phylogenetic community structure and trait evolution. Bioinformatics 24:2098–
2100. Available at www.phylodiversity.net/phylocom.
Angert et al.PNAS ?
July 14, 2009 ?
vol. 106 ?
no. 28 ?