Looking across a tidal landscape, can one foresee the signs of impending shifts among different geomorphological structures? This is a question of paramount importance considering the ecological, cultural and socio-economic relevance of tidal environments and their worldwide decline. In this Letter we argue affirmatively by introducing a model of the coupled tidal physical and biological processes. Multiple equilibria, and transitions among them, appear in the evolutionary dynamics of tidal landforms. Vegetation type, disturbances of the benthic biofilm, sediment availability and marine transgressions or regressions drive the bio-geomorphic evolution of the system. Our approach provides general quantitative routes to model the fate of tidal landforms, which we illustrate in the case of the Venice lagoon (Italy), for which a large body of empirical observations exists spanning at least five centuries. Such observations are reproduced by the model, which also predicts that salt marshes in the Venice lagoon may not survive climatic changes in the next century if IPCC's scenarios of high relative sea level rise occur.
 The time evolution of the spatially-averaged elevation of a tidal platform, z(t) (computed with respect to the local mean sea level), is described by mass balance:
B is the annually-averaged above-ground halophytic vegetation biomass. MPB(z) indicates the functional dependence on microphytobenthos, controlled by elevation. R is the rate of relative sea level (RSL) change, i.e. sea level variations plus local subsidence. QS(z,B) = 1/TC(z,B,t)ws/ρsdt is the average sediment settling flux over a tidal cycle, with period T, chiefly depending on the instantaneous sediment concentration, C(z, B, t), in turn determined by a sediment balance equation for the water column forced by the average sediment concentration, C0, resulting from past re-suspension events and possible (e.g., fluvial) sediment inputs [Krone, 1987] (see the auxiliary material for details). ws is the settling velocity, ρs is the sediment density. QT(z,B) = 1/TC(z,B,t)αBβ/ρsdt, is the average over a tidal cycle of the deposition rate due to trapping of suspended sediment by the canopy (α and β are parameters accounting for vegetation and flow characteristics [Mudd et al., 2004]). QO(B) = γB is the production of organic soil due to vegetation (combining above- and below-ground biomass production) [Randerson, 1979]. E(z, B, MPB) = 1/ρsν/τc is the tidally-averaged erosion rate due to wind-induced waves. The erosion rate depends, through the erosion coefficient ν, on sediment characteristics, an effective bottom shear stress τ (a complex function of water depth, wind velocity, fetch [Carniello et al., 2005], and vegetation presence, which efficiently dissipates wind waves [Möller et al., 1999]), and a threshold shear stress for erosion, τc, strongly dependent on stabilizing polymeric biofilms produced by benthic microbes [e.g., Paterson, 1989; Amos et al., 1998]. Because microphytobenthos growth is light-limited, we assume a sharp increase in erosion thresholds when the platform elevation yields sufficient incoming solar irradiance for microbial photosynthetic activity to occur [MacIntyre et al., 1996] (see the auxiliary material for details).
 Vegetation dynamics is described through a logistic model [Levins, 1969], which expresses biomass as the product of vegetation fractional cover, p, and the carrying capacity of the system, d (maximum biomass per unit area), i.e., B = p d. Rates of biomass change are given by:
r(z) and m(z) are elevation-dependent reproduction and mortality rates, respectively, reflecting the physiological responses of halophytic species to the controlling environmental conditions, chiefly soil water saturation, locally surrogated by elevation [Silvestri et al., 2005; Marani et al., 2006c].
 We compare two typical and contrasting situations: (1) a Spartina alterniflora-dominated case (characteristic of many North-American and U.K. sites, indicated as Spartina-dominated scenario in the following), in which biomass is a decreasing function of elevation between mean sea level (z = 0) and mean high water level (z = H), reflecting the adaptation of Spartina spp. to hypoxic conditions [Morris et al., 2002]; and (2) a case in which biomass increases with soil elevation, e.g., because of the competition among species adapted to progressively more aerated conditions [Marani et al., 2004; Silvestri et al., 2005], typical of Mediterranean tidal environments, such as the Venice lagoon (Italy) [e.g., Belluco et al., 2006], or of sites in northern continental Europe (indicated as multiple-species case in the following, even though a biomass increase with elevation is also observed at sites dominated by Spartina anglica). The physiological adaptation of Spartina alterniflora to waterlogged conditions is described using a reproduction rate which linearly decreases with elevation, while the mortality rate increases linearly with z. We also assume an isolated plant to produce at most one daughter plant per year in the most favourable conditions, i.e. r(0) = 1 year−1, while m(0) = 0 year−1. In order for the steady-state biomass at z = H to be equal to zero, as observed [Morris et al., 2002], we take r(H) = m(H) = 0.5 year−1. Similarly, the multiple-species case is modelled by assuming r(H) = 1 year−1 and m(H) = 0 year−1, whereas r(0) = m(0) = 0.5 year−1 in order for the steady-state biomass to be zero at z = 0, according to observations [Marani et al., 2004; Silvestri et al., 2005].
3. Results and Discussion
 As an illustration of the approach, which is of a general nature, we first analyze the case of landforms within the Venice lagoon, whose dynamics have been documented for several centuries and whose fate is of great concern. The tidal excursion is 2H = 1.48 m and the tidal period is T = 12 hours. We also take C0 = 20 g/m3 as a characteristic suspended sediment concentration, based on long series of water turbidity observations (see the auxiliary material). The settling velocity is ws = 0.2 mm/s (computed for a typical d50 = 50 μm), the erosion coefficient is ν = 10−4 kgm−2s−1, while the sediment density is ρs = 2650 kg/m3. The values of the vegetation parameters are: α = 1.0 · 106 m1+2βs−1 kg−β; β = 0.38; γ = 2.5 · 10−3 m3 kg−1 year−1. We first consider a 20th-century scenario, which assumes the characteristic rate of sea-level rise of 2 mm/year [Intergovernmental Panel on Climate Change (IPCC), 2001; Carbognin et al., 2004], and a local subsidence of 1.5 mm/year [Carbognin et al., 2004], for a total R = 3.5 mm/year. The dynamics of the system may be represented in phase space (Figure 1a), in which two stable equilibria are present: A sub-tidal (i.e. permanently submerged) platform and a vegetated marsh. The arrows in phase space represent the time evolution of the system out of equilibrium and highlight the stable nature of the equilibria identified (solid circles in Figure 1a). The stability of the equilibrium for B > 0 is controlled by the eigenvalues of the Jacobian matrix associated to equations (1) and (2) [Strogatz, 1994]. In this case the values of the determinant and of the trace of the Jacobian (inset in Figure 1a) are such that both eigenvalues are real and negative, and thus the marsh equilibrium state is a stable node. The nearly vertical trajectories for z > 0 (Figure 1a) show that biomass adjustments are quasi-instantaneous with respect to elevation changes. The bifurcation diagram (Figure 1b) shows the alternative system equilibria as a function of the rate of RSL change. We note that: (1) no equilibrium state exists for R < −1.4 mm/year. A relatively fast sea regression transforms the tidal environment into a terrestrial one; (2) for R ≥ −1.4 mm/year a sub-tidal platform stable equilibrium appears; (3) a second equilibrium appears for 2 ≤ R < 3.9 mm/year, corresponding to a vegetated marsh (as e.g., for the 20th-century value R = 3.5 mm/year, Figure 1b); (4) values R ≥ 3.9 mm/year lead to a transition from a marsh to a tidal-flat equilibrium; (5) for R ≥ 7.2 mm/year the sub-tidal equilibrium disappears; whereas (6) for R ≥ 7.7 mm/year all intertidal equilibria vanish.
 Because of the instantaneous adjustment of biomass to elevation changes equilibrium states can also be identified by posing dz/dt = 0 in equation (1) upon assuming B = B(z), the steady-state biomass, defined by the condition dB/dt = 0 in equation (2). This type of analysis for a Spartina-dominated system retrieves the same unvegetated equilibria as in the multiple-species case. The vegetated marsh stable equilibrium exists for 0 ≤ R < 5.9 mm/year (Figure 2a, where the arrows show that small perturbations in z near equilibria are dissipated by the system dynamics, thus marking their stable nature). Three stable equilibria coexist for 3.9 ≤ R < 5.9 mm/year (Figure 2b), as a stable tidal flat equilibrium makes its appearance. For 5.9 ≤ R < 7.2 mm/year the marsh equilibrium is no longer possible and only the sub-tidal-platform and the tidal-flat equilibria coexist (not shown). In the multiple-species case one retrieves the results of Figure 1b: Figure 2c represents the 20th century scenario characterized by the coexisting sub-tidal and marsh equilibria. Figure 2d shows that salt-marsh accretion is unable to balance RSL rise for R ≥ 3.9 mm/year and that, as a consequence, the system evolves towards a stable tidal flat. Hence, coastal marshes may not survive climatic changes in the next century as global rates of sea level rise are predicted in the range 0.8–8.5 mm/year [IPCC, 2001].
 The geomorphic role of biological processes is best appreciated by considering the hypothetical situation in which microphytobenthos and vegetation are absent (Figure 2, dashed lines). In this case a tidal flat equilibrium is possible only for narrow ranges of rates of RSL rise (−1.5 < R < −1.1 mm/year and 0 < R < 1.8 mm/year) and the basin of attraction of stable equilibria lying within the tidal range (tidal flats or salt marshes) is much narrower (0.27 < z < 0.74 m a.m.s.l) than in the presence of bio-stabilization (−0.74 < z < 0.74 m a.m.s.l). The abiotic scenario also provides insight into the dynamics of the system in the case of a sediment poor in polimeric biofilms owing to external disturbances (such as clam harvesting mechanically disrupting the surficial biofilm [Paterson, 1989]), or bioturbation, e.g. due to grazing or invertebrates [Daborn et al., 1993]. The disruption of the microbial biofilm radically changes the direction of the system evolution (compare the corresponding dashed and solid lines in Figure 2), leading to the demolition of tidal flats, which would be accreting in the presence of microphytobenthos. A sub-tidal platform is the only accessible stable equilibrium under the new conditions. Biological controls thus largely determine the existence of stable intertidal structures and the transition among them.
 In the 20th-century scenario for the Venice lagoon the sub-tidal platform stable elevation is z ∼ −2.14 m a.m.s.l., whereas vegetated marshes stabilize at z ∼ 0.30 m a.m.s.l. (Figure 2c), coherently with observations [Day et al., 1999; Marani et al., 2004; Silvestri et al., 2005]. Model predictions are also confirmed by the observed erosional trend that caused a major reduction of salt-marsh areas during the 20th century in response to the diversion of the main rivers directly out to sea and to the construction of jetties at the lagoon inlets carried out in the 16th–19th centuries [e.g., see Dorigo, 1983]. In fact, the study of the lagoon geomorphology over the last century shows that, in response to these changes, the typical elevation of unvegetated platforms (approximately equal to −0.5 m a.m.s.l. in the early ‘900s) has been steadily decreasing. However, no increase in depth beyond z = −2.4 m a.m.s.l. has been observed [Defina et al., 2007]. This suggests that indeed the average lagoon depth is increasing but that the maximum depth of a sub-tidal platform is bounded from below by a stable equilibrium for z ≅ −2.4 m a.m.s.l., quite close to the value predicted by the model.
 The analysis of different scenarios of sediment availability can elucidate trends and mechanisms characterizing different phases in the life of a tidal system. The largely positive sediment balance of the Venice lagoon typical of pre-16th century conditions is represented here by assuming C0 = 40 g/m3, compared to C0 = 20 g/m3 characteristic of the 20th century case. The model shows that in the pre-16th century conditions deposition dominated over erosion and the lagoon tended toward a configuration dominated by high marshes, in which tidal flats and sub-tidal platforms were disappearing (Figure 3a). This picture is in agreement with various accounts [Dorigo, 1983] and 19th century bathymetries [D'Alpaos, 2004]: Marsh area amounted to about 150 km2 in 1811, compared to a total lagoon area of 580 km2 (see Figure 3b, top). In 2000, marshes extended for about 50 km2 while the total lagoon area was 480 km2 (reduced mainly because of land reclamation. See Figure 3b, bottom). Marsh surfaces were thus reduced from about 26% of the total lagoon area to just 10%.
 The model presented provides a concise description of the dynamics of tidal landforms which, in spite of its structural simplicity, yields a surprisingly rich variety of system responses to changes in forcings. We suggest that the complexity observed in tidal geomorphological patterns may indeed arise from the mutual influence of biotic and abiotic components, and that the fate of landforms and of their possible geomorphological restoration can be predicted, thus pointing at the importance of eco-morphodynamic approaches for conservation studies.
 This work was supported by the PRIN 2006 projects ‘Modelli dell'evoluzione eco-morfologica di bassifondi e barene lagunari’ and ‘Fenomeni di trasporto nel ciclo idrologico’, the University of Padova project ‘Telerilevamento della zonazione e della biodiversità della vegetazione sulle barene della laguna di Venezia’, and the VECTOR-FISR 2002 project, CLIVEN research line. The Authors would like to thank Gianluigi Bugno for artwork production.