For a competition system with an arbitrary number of species, we can assume a standard set of dynamical equations given bydNi dt ¼ Niðai − ∑jbijNj Þ,
where ai > 0 is the intrinsic growth rate, bij > 0
is the competition interaction strength, and Ni
is the abundance of species i. Recall that the
Lyapunov diagonal stability of the interaction
matrix b would imply the global stability of any
feasible equilibrium point. However, in nonsym-metric competition matrices Lyapunov stability
does not always imply Lyapunov diagonal stability (53). This establishes that we should work
with a restricted class of competition matrices
such as the ones derived from the niche space of
(54). Indeed, it has been demonstrated that this
class of competition matrices are Lyapunov diagonally stable and that this stability is independent of the number of species (55). For a
competition system with a symmetric interaction-strength matrix, the structural vector is equal to
its leading eigenvector. For other appropriate
classes of matrices, we can compute the structural vectors in the same way as we did with the
effective competition matrices of our mutualistic
model and numerically simulate the feasibility
domain of the competition system. In general,
following this approach we can verify that the
lower the average interspecific competition, the
higher is the feasibility domain, and in turn,
the higher is the structural stability of the competition system.
In the case of predator-prey interactions in
food webs, so far there is no analytical work
demonstrating the conditions for a Lyapunov–
diagonally stable system and how this is linked
to its Lyapunov stability. Moreover, the computation of the structural vector of an antagonistic
system is not a straightforward task. However,
we may have a first insight about how the network architecture of antagonistic systems modulates their structural stability by transforming a
two-trophic–level food web into a competition
system among predators. Using this transformation, we are able to verify that the higher the
compartmentalization of a food web, then the
higher is its structural stability. There is no universal rule to study the structural stability of
complex ecological systems. Each type of interaction poses their own challenges as a function
of their specific population dynamics.
We have investigated the extent to which differ-
ent network architectures of mutualistic systems
can provide a wider range of conditions under
which species coexist. This research question is
completely different from the question of which
network architectures are aligned to a fixed set of
conditions. Previous numerical analyses based on
arbitrary parameterizations were indirectly ask-
ing the latter, and previous studies based on local
stability were not rigorously verifying the actual
coexistence of species. Of course, if there is a
good empirical or scientific reason to use a spe-
cific parameterization, then we should take ad-
vantage of this. However, because the set of
conditions present in a community can be constant-
ly changing because of stochasticity, adaptive
mechanisms, or global environmental change,
we believe that understanding which network
architectures can increase the structural stability
of a community becomes a relevant question.
Indeed, this is a question much more aligned
with the challenge of assessing the consequences
of global environmental change—by definition,
directional and large—than with the alternative
framework of linear stability, which focuses on
the responses of a steady state to infinitesimally
We advocate structural stability as an integra-
tive approach to provide a general assessment of
the implications of network architecture across
ecological systems. Our findings show that many
of the observed mutualistic network architec-
tures tend to maximize the domain of parameter
space under which species coexist. This means
that in mutualistic systems, having both a nested
network architecture and a small mutualistic
trade-off is one of the most favorable structures
for community persistence. Our predictions could
be tested experimentally by exploring whether
communities with an observed network archi-
tecture that maximizes structural stability stand
higher values of perturbation. Similarly, our re-
sults open up new questions, such as what the
reported associations between network architec-
ture and structural stability tell us about the
evolutionary processes and pressures occurring
in ecological systems.
Although the framework of structural stability
has not been as dominant in theoretical ecology
as has the concept of local stability, it has a long
tradition in other fields of research (29). For
example, structural stability has been key in evo-
lutionary developmental biology to articulate the
view of evolution as the modification of a con-
served developmental program (27, 28). Thus,
some morphological structures are much more
common than others because they are compatible
with a wider range of developmental conditions.
This provided a more mechanistic understand-
ing of the generation of form and shape through
evolution (56) than that provided by a historical,
functionalist view. We believe ecology can also
benefit from this structuralist view. The analo-
gous question here would assess whether the
invariance of network architecture across diverse
environmental and biotic conditions is due to
the fact that such a network structure is the one
increasing the likelihood of species coexistence
in an ever-changing world.
REFERENCES AND NOTES
1. R. M. May, Will a large complex system be stable? Nature 238,
413–414 (1972). doi: 10.1038/238413a0; pmid: 4559589
2. A. R. Ives, S. R. Carpenter, Stability and diversity of
ecosystems. Science 317, 58–62 (2007). doi: 10.1126/
science.1133258; pmid: 17615333
3. T. J. Case, An Illustrated Guide to Theoretical Ecology (Oxford
Univ. Press, Oxford, UK, 2000).
4. S. L. Pimm, Food Webs (Univ. Chicago Press, Chicago, 2002).
5. A. Roberts, The stability of a feasible random ecosystem.
Nature 251, 607–608 (1974). doi: 10.1038/251607a0
6. U. Bastolla et al., The architecture of mutualistic networks
minimizes competition and increases biodiversity. Nature
1018–1020 (2009). doi: 10.1038/nature07950; pmid: 19396144
7. T. Okuyama, J. N. Holland, Network structural properties
mediate the stability of mutualistic communities. Ecol. Lett. 11,
208–216 (2008). doi: 10.1111/j.1461-0248.2007.01137.x;
8. E. Thébault, C. Fontaine, Stability of ecological communities
and the architecture of mutualistic and trophic networks.
Science 329, 853–856 (2010). doi: 10.1126/science.1188321;
9. S. Allesina, S. Tang, Stability criteria for complex ecosystems.
Nature 483, 205–208 (2012). doi: 10.1038/nature10832;
10. A. James, J. W. Pitchford, M. J. Plank, Disentangling
nestedness from models of ecological complexity. Nature 487,
227–230 (2012). doi: 10.1038/nature11214; pmid: 22722863
11. S. Saavedra, D. B. Stouffer, “Disentangling nestedness”
disentangled. Nature 500, E1–E2 (2013). doi: 10.1038/
nature12380; pmid: 23969464
12. S. Suweis, F. Simini, J. R. Banavar, A. Maritan, Emergence of
structural and dynamical properties of ecological mutualistic
networks. Nature 500, 449–452 (2013). doi: 10.1038/
nature12438; pmid: 23969462
13. A. M. Neutel, J. A. Heesterbeek, P. C. De Ruiter, Stability in real
food webs: Weak links in long loops. Science 296, 1120–1123
(2002). doi: 10.1126/science.1068326; pmid: 12004131
14. B. S. Goh, Global stability in many-species systems. Am. Nat.
111, 135–143 (1977). doi: 10.1086/283144
15. D. O. Logofet, Stronger-than-lyapunov notions of matrix
stability, or how flowers help solve problems in mathematical
ecology. Linear Algebra Appl.
398, 75–100 (2005).
doi: 10.1016/ j.laa.2003.04.001
16. D. O. Logofet, Matrices and Graphs: Stability Problems in
Mathematical Ecology (CRC Press, Boca Raton, FL, 1992).
17. Y. M. Svirezhev, D. O. Logofet, Stability of Biological
Communities (Mir Publishers, Moscow, Russia, 1982).
18. Any matrix B is called Lyapunov-stable if the real parts of
all its eigenvalues are positive, meaning that a feasible
equilibrium at which all species have the same abundance is
at least locally stable. A matrix B is called D-stable if DB is a
Lyapunov-stable matrix for any strictly positive diagonal
matrix D. D-stability is a stronger condition than Lyapunov
stability in the sense that it grants the local stability of any
feasible equilibrium (15). In addition, a matrix B is called
Lyapunov diagonally stable if there exists a strictly positive
diagonal matrix D so that DB + BtD is a Lyapunov-stable
matrix. This notion of stability is even stronger than D-stability
in the sense that it grants not only the local stability of any
feasible equilibrium point, but also its global stability (14).
A Lyapunov–diagonally stable matrix has all of its principal
minors positive. Also, its stable equilibrium (which may be only
partially feasible; some species may have an abundance of
zero) is specific (57). This means that there is no alternative
stable state for a given parameterization of intrinsic growth
rates (55). We have chosen the convention of having a
minus sign in front of the interaction-strength matrix B. This
implies that B has to be positive Lyapunov-stable (the real
parts of all eigenvalues have to be strictly positive) so that the
equilibrium point of species abundance is locally stable. This
same convention applies for D-stability and Lyapunov–diagonal
19. J. H. Vandermeer, Interspecific competition: A new approach
to the classical theory. Science 188, 253–255 (1975).
doi: 10.1126/science.1118725; pmid: 1118725
20. T. J. Case, Invasion resistance arises in strongly interacting
species-rich model competition communities. Proc. Natl. Acad.
Sci. U.S.A. 87, 9610–9614 (1990). doi: 10.1073/
pnas.87.24.9610; pmid: 11607132
21. All simulations were performed by integrating the system of
ordinary differential equations by using the function ode45 of
Matlab. Species are considered to have gone extinct when their
abundance is lower than 100 times the machine precision.
Although we present the results of our simulations with a fixed
value of r = 0.2 and h = 0.1, our results are robust to this
choice as long as r < 1 (45). In our dynamical system, the units
of the parameters are abundance for N, 1/time for a and h,
and 1/(time · biomass) for b and g. Abundance and time can
be any unit (such as biomass and years), as long as they are
constant for all species.
22. V. I. Arnold, Geometrical Methods in the Theory of Ordinary
Differential Equations (Springer, New York, ed. 2, 1988).
23. R. V. Solé, J. Valls, On structural stability and chaos in
biological systems. J. Theor. Biol. 155, 87–102 (1992).
24. A. N. Kolmogorov, Sulla teoria di volterra della lotta per
l’esistenza. Giornale Istituto Ital. Attuari 7, 74–80 (1936).
1253497-8 25 JULY 2014 • VOL 345 ISSUE 6195
RESEARCH | RESEARCH ARTICLE