Vegetation engineers marsh morphology through

multiple competing stable states

Marco Marania,b,1, Cristina Da Liob, and Andrea D’Alpaosc
a
Division of Earth and Ocean Sciences, Nicholas School of the Environment and Department of Civil and Environmental Engineering, Pratt School of
Engineering, Duke University, Durham, NC 27708; and Departments of bCivil, Environmental, and Architectural Engineering and cGeosciences, University of
Padova, I-35131 Padova, Italy

Edited by Ignacio Rodriguez-Iturbe, Princeton University, Princeton, NJ, and approved January 2, 2013 (received for review October 21, 2012)

Marshes display impressive biogeomorphic features, such as ∂z

zonation, a mosaic of extensive vegetation patches of rather ðx; tÞ = Qs ðx; tÞ + Qo ðx; tÞ − R; [1]
∂t
uniform composition, exhibiting sharp transitions in the pres-
ence of extremely small topographic gradients. Although gen- which expresses variations in soil elevation, referenced to mean
erally associated with the accretion processes necessary for sea level (MSL), as the net result of (i) the rate of inorganic soil
marshes to keep up with relative sea level rise, competing deposition, Qs, determined by the hydrodynamic circulation/
environmental constraints, and ecologic controls, zonation is still sediment transport processes; (ii) the rate of organic soil pro-
poorly understood in terms of the underlying biogeomorphic duction by vegetation, Qo = γB0fi(z), modulated by compaction/
mechanisms. Here we find, through observations and modeling decomposition processes (encapsulated by the factor γ) and by
interpretation, that zonation is the result of coupled geomor- a fitness function (0 ≤ fi(z) ≤ 1), describing how biomass pro-
phological–biological dynamics and that it stems from the ability duction and competitive abilities of species i vary as a function of
of vegetation to actively engineer the landscape by tuning soil elevation z (summarizing local environmental stressors, such as
elevation within preferential ranges of optimal adaptation. We salinity, sediment aeration, etc.) (3, 18, 24, 25); and (iii) the rate
find multiple peaks in the frequency distribution of observed of relative sea level rise, R (SI Text).
topographic elevation and identify them as the signature of Field manipulations of marsh species distributions, when in-
biologic controls on geomorphodynamics through competing terspecific competition is allowed to take place or is artificially
stable states modulated by the interplay of inorganic and organic suppressed, show the important role of edaphic constraints on
deposition. Interestingly, the stable biogeomorphic equilibria biomass production (26). Detailed biomass determinations in
correspond to suboptimal rates of biomass production, a result transplant experiments with Spartina spp. further indicate that
coherent with recent observations. The emerging biogeomor- fi(z) takes on maximum values at elevations that are character-
phic structures may display varying degrees of robustness to istic for each species and that it decreases as elevation departs
changes in the rate of sea level rise and sediment availability, from this optimal range (6, 27). This qualitative feature has
with implications for the overall resilience of marsh ecosystems important geomorphic consequences, and we incorporate it by
to climatic changes. adopting the following expression for the fitness function: fi(ζ) =
2 · [exp(λ(ζ − ζ i0)) + exp(−λ(ζ − ζ i0))]−1, where ζ = z/H. fi(ζ i0) = 1
biogeomorphology | ecohydrology | ecotone defines the elevation of maximum biomass production (and

AND PLANETARY SCIENCES

highest competitive abilities), whereas λ controls the rate at

EARTH, ATMOSPHERIC,
which fitness tends to zero away from its maximum: higher values
M arsh vegetation zonation patterns occur widely in tidal
environments worldwide (1) (SI Text). Although we know
that vegetation plays an important role in offsetting the local rate
of λ corresponding to more “specialized” vegetation species (we
first consider the case of relatively “specialized” species, λ = 5,
of relative sea level rise through organic soil production and and later compare with the case of less “specialized” plants, with
inorganic sediment trapping (2–10), zonation patterns tradi- λ = 2). Hence, all species have an equal maximum fitness and an
tionally are explained as the result of interspecific competition equal rate of fitness decrease away from their respective optima
and of the “passive” adaptation of marsh vegetation to spatially (Fig. 1B). This is a convenient and flexible way of expressing the
general known properties of marsh vegetation species (18, 28),

ENVIRONMENTAL
varying soil conditions (11–16). In fact, this interpretational
but other expressions displaying similar general behaviors have

SCIENCES
framework, as well as prevailing explanations of ecotones in
general (abrupt edges in vegetation distributions, e.g., see ref. been used as well, leading to very similar results.
17), view vegetation distributions chiefly as a response to envi- We note that Eq. 1 describes elevation changes taking place on
biogeomorphic time scales of 1 y or greater, whereas inorganic
ronmental drivers.
deposition is governed by hydrodynamic transport regulated by
On the basis of detailed observations and modeling, we
tidal flooding on subhourly time scales. This clear separation of
propose here an interpretation, which couples geomorphic dy-
time scales allows us to decouple the numerical integration of
namics and species competition in a spatially extended setting,
Eq. 1, performed with yearly time steps, and the solution of the
previous models being incapable of insight into zonation-gener- sediment transport problem from which the inorganic deposition
ating mechanisms because they either are lumped in space (18) or rate is obtained (5, 24). At the scale of the main, half-daily tidal
are not inclusive of interspecific competition (19–23). We find that period, suspended sediment transport from the channel to the
marsh landscapes are actively engineered by competing vegetation
species through a set of vegetation-controlled equilibrium states,
of which zonation patterns are the observable biogeomorphic Author contributions: M.M. designed research; M.M. and A.D. lead model development;
signatures. C.D.L. coded the model and ran the simulations; C.D.L. and A.D. collected the data; M.M.,
C.D.L., and A.D. analyzed data; and M.M. and A.D. wrote the paper.
Spatial Biogeomorphodynamics with Competition The authors declare no conflict of interest.
We describe the time evolution of a marsh transect oriented in This article is a PNAS Direct Submission.
a direction perpendicular to the nearest channel feeding the 1
To whom correspondence should be addressed. E-mail: marco.marani@duke.edu.
marsh with inorganic sediment. Changes in soil elevation are This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
everywhere dictated by Exner’s equation, 1073/pnas.1218327110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1218327110 PNAS | February 26, 2013 | vol. 110 | no. 9 | 3259–3263

 A B C

Fig. 1. Zonation patterns generated by the model. (A) The time evolution of transect topography was started here from a linear initial condition, but several
other initial conditions were explored with analogous results. Monospecific vegetation patches, very similar to observed zonation patterns (Inset), and
terrace-like topographic structures emerge as a result of multiple stable states defined by ∂z/∂t = 0 and ∂/∂z(∂z/∂t) < 0. (B) Fitness functions of the species
populating the marsh, which define the rate of organic soil production as Qo = γ · B0 · fi(z), as well as the species competitive abilities (γ incorporates typical
vegetation characteristics and the density of the organic soil produced; B0 is the biomass density of a fully vegetated marsh). (C) The superscripts “(s)” and
ðuÞ ðsÞ
“(u)” denote stable and unstable equilibria, respectively. If the initial elevation of the site at x^2 is comprised between z2 and z2 , the elevation will tend
ðsÞ ðuÞ ðsÞ
toward z2 . If the initial elevation of the site at x^2 is located below z2 , the elevation will tend toward z3 .

marsh and the associated deposition of inorganic sediment on noise,” thus providing interesting insights into the relative role of
the marsh surface are described here through the advection– these factors in determining observed biogeomorphic patterns.
dispersion equation (e.g., see ref. 20): We note that, in both cases, the fitness function not only regu-

 lates biomass production but also species competitive abilities,
∂yC ∂ ∂C thus incorporating a competitive displacement mechanism.
+ uCy − kd y = − ws C; [2]
∂t ∂x ∂x In summary, the system evolves in time through the following
steps: (i) Computation over a tidal cycle of flow velocity, sus-
where y(x, t) = zw(t) − z(x, t) is the water depth, zw(t) is the el- pended sediment concentration, and inorganic deposition rates
evation of the free surface, kd is the dispersion coefficient [as- for the current topographic profile. The yearly average R deposition
sumed to be a constant value, kd = 1.5 m2/s (29), characteristic of rate at year tj is evaluated as Qs ðxk ; tj Þ = nT =T · T ws · Cðxk ; tÞdt
the small flow velocities typical in marshes], and u(x, t) is the (24) [nT being the number of tidal cycles in 1 year, C(xk, t) the
fluid advective velocity. The latter is computed from the water local instantaneous suspended sediment concentration, and ws =
continuity equation (in a quasi-static propagation of tidal levels 0.2 mm/s a typical settling velocity of fine intertidal sediments
assumption) by assuming a sinusoidal fluctuation of the tidal (32)]. (ii) Computation of Qo(xk, tj) = γ · B0 · fi(z(xk, tj)), where γ ·
level zw(t) = H · sin(2πt/T) (with T = 12 h and H = 0.5 m typical B0 = 2.5 mm·y−1 (24), i being the species currently occupying
of microtidal settings). Sediment erosion is neglected because of coordinate xk. (iii) Computation of ∂z/∂t = Qs(xk, tj) + Qo(xk, tj) −
the effective wave dissipation by above-ground biomass on the R and update of the topography at each site as z(xk, tj + Δt) = z(xk,
marsh surface (30, 31). Eq. 2 is solved numerically over one tidal tj) + ∂z/∂t(xk, tj) · Δt (Δt = 1 y). (iv) Update of the species dis-
cycle, and the resulting settling term, ws C(x, t), is integrated over tribution throughout the transect using either the fittest-takes-all
time to yield the Qs(x, t) to be used in Eq. 1 (see SI Text for or the stochastic competition mechanism.
further details). It is useful to note here that although Exner’s
Eq. 1 describes the local balance between the total deposition Results and Discussion
rate and the rate of relative sea level rise R, it actually embeds The fittest-takes-all colonization mechanism produces stable
nonlocal dynamics due to the dependence of Qs(x, t) on space states characterized by sharp transitions between neighboring
and on the topographic configuration of the entire transect, biogeomorphic terrace-like structures (Fig. 1A). The marsh
rather than just on the local topography. profile is characterized by gently sloping areas, colonized by
Qo(x, t) also plays a fundamental role in determining accretion a single vegetation species, very much reminiscent of observed
rates, as changes in the distribution of topographic elevation marsh zonation structures (13, 15, 16) (SI Text). It is interesting
along the marsh transect are strongly affected by changes in to see that the sharp boundaries displayed by the emerging
species distribution as a consequence of interspecific com- morphologies essentially are the result of vegetation pinning
petition. We consider two possible descriptions of the latter, topography within near-optimal elevation ranges. In fact, the
which are based on either (i) selecting, at each site with co- analysis of the governing equations shows that zonation struc-
ordinate xk, the species i for which fi(zk) is maximum (“fittest tures emerge from feedbacks involving biomass production, in-
takes all”), or (ii) P randomly selecting species i with a probability organic deposition, and soil accretion, leading to pairs of stable
pði; xk Þ = fi ðzk Þ= j fj ðzk Þ (“stochastic competition” mechanism), and unstable equilibrium states. This can be seen by considering,
to account for the fact that the current species may not be dis- as an illustrative example, the lowest site in the middle patch
placed instantaneously by the fittest species, that soil properties (species i = 2, in green in Fig. 1). The values of inorganic and
and sediment fluxes may be stochastically heterogeneous in organic deposition rates at this site determine two solutions to
space and time, and that stochastic dispersal may locally affect the equilibrium condition ∂z/∂t = 0: a stable equilibrium with
ðsÞ
species distribution. Although the second criterion is more re- elevation, say, z2 (located above the maximum of f2(z), solid
ðuÞ
alistic, the first allows us to analyze an ideal case in which the green circle in Fig. 1 B and C), and an unstable equilibrium z2
effect of competition can be isolated from “environmental (located below the maximum of f2(z), open circle). To determine

3260 | www.pnas.org/cgi/doi/10.1073/pnas.1218327110 Marani et al.

the stability properties of the equilibrium state in ^x2 (Fig. 1A), we A
numerically find the value of ∂z/∂t from Eq. 1 for a set of transect
configurations obtained by perturbing the local elevation in ^x2
ðuÞ ðsÞ
within a neighborhood of z2 and z2 (Fig. 1C). This analysis
ðsÞ
shows that, indeed, z2 is a stable equilibrium: a perturbation
ðsÞ
that increases the local elevation with respect to z2 generates
a decrease in biomass production (Fig. 1B) as well as a decrease
in Qs (due to an increased local velocity, not shown), which
makes ∂z/∂t < 0 (Fig. 1C), thus bringing the system back to the
original equilibrium. Similarly, a perturbation that decreased the
ðsÞ
local soil elevation with respect to z2 would induce an increase
in the biomass production (Fig. 1B) and in Qs, and hence would
make ∂z/∂t > 0 (Fig. 1C), again bringing the system back to the
stable equilibrium. This analysis may be summarized by noting
ðsÞ
that ∂/∂z(∂z/∂t) < 0 at z2 , which defines the condition for a stable
equilibrium. Note that Qs(x, t) varies in space, such that stable
equilibria for sites within the same vegetation patch located
closer to the source of inorganic sediment are higher than z2 ,
ðsÞ B
generating a mildly sloping geomorphic structure, consistent with
ðuÞ
observations. The second equilibrium solution, z2 , is an un-
stable equilibrium (open green circle in Fig. 1C). A positive
perturbation with respect to this elevation enhances the organic
deposition rate (Fig. 1B) and reduces Qs(x, t). However, the in-
crease in Qo(x, t) outweighs the decrease in Qs, thereby making
∂z/∂t > 0 (Fig. 1C) and driving the elevation of the site away from
ðuÞ ðsÞ
z2 toward z2 . If, on the contrary, a perturbation acts to de-
crease the elevation of the site, the organic deposition rate
decreases faster than Qs increases, making ∂z/∂t < 0 (Fig. 1C),
thus pushing the local elevation to even lower values, toward the
ðsÞ
stable equilibrium, z3 , of the vegetation species i = 3 (blue solid
circle in Fig. 1C).
We note that in the case of the “green” species (i = 2) (as well
as for the lower “blue” species), ∂Qs/∂z  ∂Qo/∂z; therefore, the C
stability of the equilibria is not influenced by variations of Qs(x,
t), because ∂/∂z(∂z/∂t) ≅ ∂Qo/∂z. This equilibrium state is com-
pletely determined by organic soil production, and the stability of
these equilibria may be established by studying fi(z) alone, the

AND PLANETARY SCIENCES

condition dfi/dz < 0 denoting a stable equilibrium.

EARTH, ATMOSPHERIC,
This is not the case for the portion of the marsh nearest the
margin (red in Fig. 1A), where the inorganic deposition rate is
very sensitive to changes in the topographic elevation because of
the large local availability of inorganic sediment. The equilib-
rium state in the upper marsh zone (red in Fig. 1) is, in fact,
ðsÞ
stable as ∂/∂z(∂z/∂t) < 0 even though z1 lies on the branch of the
fitness function located below the maximum (Fig. 1B). The sta-
bility of the equilibrium near the tidal creek thus is jointly con-

ENVIRONMENTAL
trolled by inorganic deposition and organic soil production.

SCIENCES
We note that, in all cases, the terrace-like structures lie, for the
most part, above the elevation at which maximum biomass pro-
duction occurs (Fig. 1B), suggesting that zonation patterns may Fig. 2. (A) Stochastic interspecific competition. Species i is selected as the
P
be associated with a higher morphological stability at the expense successful competitor with a probability pði; xk Þ = fi ðzk Þ= j fj ðzk Þ (for each
site xk and each time step), generating relatively noisy patterns. However,
of a reduced productivity.
the multimodal frequency distribution of topographic elevation, the signa-
Biogeomorphodynamics in the real world are affected by ture of the underlying biogeomorphic coupling, remains detectable. Each
stochastic forcings, stochasticity in competition mechanisms, peak (color coded according to the species that is most abundant within
heterogenous edaphic conditions, etc. The stochastic competition each elevation interval) clearly is associated with the unique species that
mechanism thus may be considered to generate more realistic generates it. (B) In the absence of an organic soil contribution to the ac-
dynamics and nondeterministic patterns (Fig. 2A), which, how- cretion rate, the resulting smooth topography is determined entirely by in-
ever, are more difficult to interpret. Also, in this case, one still organic deposition. Vegetation species colonize the transect in banded
may identify the role of the underlying biogeomorphic feedbacks patterns according to their respective fitnesses. (C) The frequency distribu-
by studying the probability distribution of elevations. A multimodal tion of topographic elevation shows uncertain symptoms of multiple peaks
when less specialized vegetation species (λ = 2) are considered, a sign that
elevation distribution (Fig. 2A), by highlighting the presence of
a decreased vegetation specialization produces less easily detectable multi-
engineered preferential elevation ranges, is a clear signature of ple peaks in the topographic elevation frequency distribution.
the governing feedback between biomass production and elevation,
even when vegetational and topographic patterns are significantly
less visually evident (Fig. 2A, Inset). colonize the transect, without having the possibility to affect its
In fact, when the organic contribution to accretion is turned morphodynamic evolution, Fig. 2B), we obtain a smooth topo-
off (but vegetation species are still allowed to compete and graphic profile, in which banded vegetation is present, but

Marani et al. PNAS | February 26, 2013 | vol. 110 | no. 9 | 3261
multiple peaks in the frequency distribution of topographic ele- of environmental noise, we now seek the signature of this bio-
vation are absent. Hence, banded vegetation patterns do not geomorphic coupling in real marshes. To this end, we performed
automatically imply biogeomorphic feedbacks, but, by them- detailed marsh topographic surveys in the Venice lagoon, with
selves, are just the symptom of a passive adaptation of vegetation a total station (i.e., an electronic theodolite) allowing for a final
species to a topographic profile that cannot be affected by veg- accuracy better than 1 mm in elevation. The analysis of these
etation dynamics. If less specialized vegetation is considered (e.g., data shows the presence of multiple peaks in the frequency
λ = 2, Fig. 2C), biologic controls on elevation may be too weak to distributions of observed topographic elevation (Fig. 3; see
emerge above environmental noise and multiple peaks in the also SI Text for an analysis of the whole dataset). Each peak in
topographic elevation frequency distribution may be difficult to the elevation distribution is associated with a different and
detect. Thus, we must conclude that multiple peaks in the fre- characteristic vegetation species. This correspondence be-
quency distribution of topographic elevation, which are univo- tween vegetation and topographic frequency distributions is
cally associated with single vegetation species, are the distinctive found consistently in all the data examined (see SI Text for
signature of a strong feedback between soil accretion processes more data analyses).
and specialized vegetation species.
As our analyses show that zonation structures are de- Conclusions
termined largely by the degree of species adaptation to char- Our observations and model results show that ubiquitous zonation
acteristic ranges of topographic elevation even in the presence patterns are largely the product of landscape construction by marsh
vegetation species through the biomass-elevation feedback. We
note that this “active” biogeomorphic mechanism is very different
in nature from previously studied “passive” biological controls on
A the physical environment occurring through biostabilization/bio-
turbation processes (33). The coupled vegetation–topography dy-
namics partition marsh topographies into an almost discrete set of
stable equilibria, and the ecotones characteristic of salt marsh zo-
nation emerge because of a two-way feedback between biomass
production and elevation, rather than as a result of sharp gradients
in environmental forcings (17).
Because stability in the marsh zone controlled by organic
deposition requires dfi/dz < 0, our findings imply that marsh
vegetation species tend not to operate at the maximum biomass
production rate, gaining, however, added environmental stabil-
ity in return. In fact, our results show that entire sections of
a vegetation patch, particularly where the accretion process is
dominated by plant organic soil production (Fig. 1B), are lo-
cated well above the elevation corresponding to maximum
productivity. Interestingly, root growth by common marsh veg-
etation species recently was observed to be suboptimal at several
study sites (34).
Numerical experiments in which the organic soil contribution
to accretion is artificially turned off leads to frequency dis-
B tributions of topographic elevation with a single maximum.
Furthermore, when the degree of specialization of vegetation
species is reduced, the tendency toward a multimodal distribu-
tion of topographic elevation may be overwhelmed by environ-
mental noise, and may not be detectable. We thus conclude that
the detection, in the frequency distribution of topographic ele-
vation, of multiple peaks associated with a single characteristic
vegetation species is the distinctive sign of the active tuning of
topography by specialized plant species. As a consequence, the
observed multimodal elevation frequency distributions show that
actual biogeomorphic zonation structures are not compatible
with the simplistic picture of a passive adaptation by vegetation
to a given distribution of topographic elevations.
The spatial dependence of inorganic sediment deposition,
jointly induced by sediment transport processes and the evolving
topography, defines the spatial extent of the emerging bio-
geomorphic structures. The resilience of marsh patterns and
ecosystem properties (e.g., as represented by the elevation dif-
ference between the stable and the unstable equilibria; Fig. 1C)
thus depends on the species-specific landscape-building abilities of
Fig. 3. (A and B) Observed zonation patterns. An accurate topographic intertidal vegetation. Changes in the rate of relative sea level rise
survey (uncertainty smaller than 1 mm) reveals a multimodal frequency may result in the selective disappearance of some or all of the
distribution of soil elevation, highly suggestive of the major role played by stable equilibria associated with marsh morphological/biological
the biomass-elevation feedback in tuning marsh topography. Each bar is patterns, with consequent reductions in the associated biodiversity.
color coded according to the vegetation species that is most abundant
within the pertinent elevation interval, showing that, indeed, elevation ACKNOWLEDGMENTS. We thank Massimiliano Ignaccolo for discussions on
ranges are characteristic of the vegetation species (or of a typical mix of the mathematical representation of vegetation fitness functions as well as
species at high elevations) that maintain them. two anonymous reviewers whose contributions significantly improved the

3262 | www.pnas.org/cgi/doi/10.1073/pnas.1218327110 Marani et al.

manuscript. We acknowledge support by Duke University, the PhD School of and the Italian National Project “Eco-Morfodinamica di Ambienti a Marea e
Civil and Environmental Engineering Sciences at the University of Padova, Cambiamenti Climatici.”

1. Adam P (1990) Saltmarsh Ecology (Cambridge Univ Press, Cambridge, UK). 19. D’Alpaos A, Lanzoni S, Marani M, Rinaldo A (2007) Landscape evolution in tidal
2. Allen JRL (2000) Morphodynamics of Holocene salt marshes: A review sketch embayments: Modeling the interplay of erosion, sedimentation, and vegetation dy-
from the Atlantic and Southern North Sea coasts of Europe. Quat Sci Rev 19: namics. J Geophys Res 112:F01008.
1155–1231. 20. Fagherazzi S, et al. (2012) Numerical models of salt marsh evolution: Ecological,
3. Blum LK, Christian RR (2004) Belowground production and decomposition along geomorphic, and climatic factors. Rev Geophys 50:RG1002.
a tidal gradient in a Virginia salt marsh. The Ecogeomorphology of Salt Marshes, 21. Kirwan ML, Murray AB (2007) A coupled geomorphic and ecological model of tidal
Coastal and Estuarine Studies (American Geophysical Union, Washington, DC), Vol 59, marsh evolution. Proc Natl Acad Sci USA 104(15):6118–6122.
pp 47–74. 22. Mudd SM, Fagherazzi S, Morris JT, Furbish DJ (2004) Flow, sedimentation, and bio-
4. D’Alpaos A, Mudd SM, Carniello L (2011) Dynamic response of marshes to perturba- mass production on a vegetated salt marsh in South Carolina: Toward a predictive
tions in suspended sediment concentrations and rates of relative sea level rise. J model of marsh morphologic and ecologic evolution. The Ecogeomorphology of Salt
Geophys Res 116:F04020. Marshes, Coastal and Estuarine Studies (American Geophysical Union, Washington,
5. Marani M, D’Alpaos A, Lanzoni S, Carniello L, Rinaldo A (2007) Biologically-controlled DC) Vol 59, pp 165–188.
multiple equilibria of tidal landforms and the fate of the Venice lagoon. Geophys Res 23. Temmerman S, et al. (2007) Vegetation causes channel erosion in a tidal landscape.
Lett 34(11):L11402. Geology 35:631–634.
6. Morris JT, Sundareshwa PV, Nietch CT, Kjerfve B, Cahoon DR (2002) Responses of 24. Marani M, D’Alpaos A, Lanzoni S, Carniello L, Rinaldo A (2010) The importance of
coastal wetlands to rising sea level. Ecology 83:2869–2877. being coupled: Stable states and catastrophic shifts in tidal biomorphodynamics. J
7. Nyman JA, Walters RJ, Delaune RD, Patrick WH (2006) Marsh vertical accretion via
Geophys Res 115:F04004.
vegetative growth. Estuar Coast Shelf Sci 69:370–380.
25. Mudd S, Howell S, Morris J (2009) Impact of dynamic feedbacks between sedimen-
8. Neubauer SC (2008) Contributions of mineral and organic components to tidal
tation, sea level rise, and biomass production on near surface marsh stratigraphy and
freshwater marsh accretion. Estuar Coast Shelf Sci 78:78–88.
carbon accumulation. Estuar Coast Shelf Sci 82:377–389.
9. Perillo GME, Wolanski E, Cahoon DR, Brinson MM (2009) Coastal Wetlands: An In-
26. Pennings SC, Grant M, Bertness MD (2005) Plant zonation in low-latitudes salt
tegrated Ecosystem Approach (Elsevier, New York).
marshes: Disentangling the roles of flooding, salinity and competition. J Ecol 93:
10. van de Koppel J, Bouma TJ, Herman PMJ (2012) The influence of local- and landscape-
159–167.
scale processes on spatial self-organization in estuarine ecosystems. J Exp Biol 215(Pt
27. Kirwan ML, Christian RR, Blum LK, Brinson MM (2012) On the relationship between
6):962–967.
sea level and Spartina alterniflora production. Ecosystems (N Y) 15:140–147.
11. Bertness MD, Ellison AM (1987) Determinants of pattern in a New England salt marsh
28. Morris JT (2007) Ecological engineering in intertidial saltmarshes. Hydrobiologia 577:
plant community. Ecol Monogr 57(2):129–147.
12. Bockelmann AC, Bakker JP, Neuhaus R, Lage J (2002) The relation between vegetation 161–168.
zonation, elevation and inundation frequency in a Wadden Sea salt marsh. Aquat Bot 29. Elder JW (1959) The dispersion of marked fluid in turbulent shear flow. J Fluid Mech
73:211–221. 5:544–560.
13. Marani M, Zillio T, Belluco E, Silvestri S, Maritan A (2006) Non-neutral vegetation 30. Gedan KB, Kirwan ML, Wolanski E, Barbier EB, Silliman BR (2011) The present and
dynamics. PLoS ONE 1(1):e78. future role of coastal wetland vegetation in protecting shorelines: Answering recent
14. Moffett KB, Robinson DA, Gorelick SM (2010) Relationship of salt marsh vegetation challenges to the paradigm. Clim Change 106:7–29.
zonation to spatial patterns in soil moisture, salinity, and topography. Ecosystems (N 31. Möller I, Spencer T, French JR, Leggett D, Dixon M (1999) Wave transformation over
Y) 13:1287–1302. salt marshes: A field and numerical modelling study from North Norfolk, England.
15. Pennings SC, Callaway RM (1992) Salt marsh plant zonation: The relative importance Estuar Coast Shelf Sci 49:411–426.
of competition and physical factors. Ecology 73(2):681–690. 32. Gibbs RJ (1985) Estuarine flocs: Their size, settling velocity and density. J Geophys Res
16. Silvestri S, Defina A, Marani M (2005) Tidal regime, salinity and salt marsh plant zo- 90(C2):3249–3251.
nation. Estuar Coast Shelf Sci 62:119–130. 33. Murray AB, Knaapen MAF, Tal M, Kirwan ML (2008) Biomorphodynamics: Physical-
17. Walker S, et al. (2003) Properties of ecotones: Evidence from five ecotones objectively biological feedbacks that shape landscapes. Water Resour Res 44(11), 10.1029/
determined from a coastal vegetation gradient. J Veg Sci 14:579–590. 2007WR006410.
18. Morris JT (2006) Competition among marsh macrophytes by means of geo- 34. Kirwan ML, Guntenspergen GR (2012) Feedbacks between inundation, root pro-
morphological displacement in the intertidal zone. Estuar Coast Shelf Sci 69(3–4): duction, and shoot growth in a rapidly submerging brackish marsh. J Ecol 100:
764–770.

AND PLANETARY SCIENCES

395–402.

EARTH, ATMOSPHERIC,
ENVIRONMENTAL
SCIENCES

Marani et al. PNAS | February 26, 2013 | vol. 110 | no. 9 | 3263