Coastal Blue Carbon (BC) ecosystems, which include seagrass meadows, mangrove forests, and tidal marshes, can store carbon in their underlying sediments over centennial to millennial scales, and their protection and restoration could contribute to offsetting greenhouse gas emissions (Nellemann et al., 2009; Duarte et al., 2013; Macreadie et al., 2021). Concurrently, seagrass meadows are biodiversity hotspots, support fisheries, provide natural protection by attenuating wave energy and accreting sediments, so they offer several ecosystem services beyond carbon sequestration,. Therefore, the sustainable management of seagrass meadows and the assurance of their integrity has been recognized as a low-regret Nature-based Solution (NbS) towards mitigation and adaptation to climate change safeguarding coastal resilience (Hoegh-Guldberg et al., 2019; Hoegh-Guldberg et al., 2020; Gattuso et al., 2021).

However, seagrass sensitivity to stressors is acute, and global change impacts have resulted in extensive loss of meadows worldwide (Orth et al., 2006; Unsworth et al., 2022). The loss of at least 20% of the global seagrass area since 1880 (Dunic et al., 2021), with accelerating loss rates from 0.9% yr^-1 in the 1940s to 7% y^-1 in the 1990s and 2000s (Waycott et al., 2009), highlights a need for further conservation (Unsworth et al., 2022). The loss is attributed to multiple direct pressures and climate change acting synergistically, including coastal eutrophication, mechanical damage, warming and extreme events (de los Santos et al., 2019). Seagrass loss could compromise the climate change mitigation and adaptation potential of seagrass meadows. Yet, there is a significant lack of understanding of the extent to which global change stressors could affect carbon storage in these critical ecosystems. Seagrass loss usually decreases their carbon sequestration capacity (Marbà et al., 2015; Dahl et al., 2016; Thorhaug et al., 2017; Trevathan-Tackett et al., 2017; Githaiga et al., 2019; Salinas et al., 2020; Moksnes et al., 2021), and increases the risk of erosion and subsequent mineralization of historical carbon accumulated in sediments (Marbà et al., 2015; Arias-Ortiz et al., 2018; Roca et al., 2022; Egea et al., 2023), potentially transforming the meadows from sinks to sources of greenhouse gasses (Lovelock et al., 2017). However, the relationship between seagrass ecosystem integrity and carbon storage is not straightforward, and ecological transformations (even dramatic ones such as complete cover loss) may not necessarily threaten the carbon that is already stored in the sediments (Piñeiro-Juncal et al., 2021; Apostolaki et al., 2022). Despite the progress in blue carbon science over the past decade, we still lack an in-depth understanding of how seagrass deterioration and loss affect the functioning of seagrass ecosystems as carbon sinks, and it is not well-understood if there are thresholds to changes in carbon fluxes linked to ecological and/or biogeochemical processes.

Recent studies in seagrass ecosystems have suggested that biogeochemical and biophysical feedbacks (i.e., direct or indirect reciprocal interactions between system parts) can cause seagrass meadows to abruptly and irreversibly shift from a healthy growing state to a deteriorated barren state (Maxwell et al., 2017). Similar shifts due to competition by invasive species have also been modeled (Llabrés et al., 2023; Beca-Carretero et al., 2024). Abrupt shifts can occur at the crossing of thresholds with irreversible changes in ecological functions and services. Such regime shifts might be associated with the presence of tipping points that cause a radical shift to a usually contrasting state propelled by a strong positive feedback (van Nes et al., 2016). Regime shifts have been described for a range of ecosystems like clear lakes turning turbid (Scheffer et al., 1993), coral reefs getting overgrown by macroalgae (Knowlton, 1992), or drylands shifting to desertification (Kéfi et al., 2007), seaweed forests collapsing into urchin barrens (Ling et al., 2015) and seagrasses turning to bare or algae dominated meadows (Unsworth et al., 2015), or degraded dead matte (Rindi et al., 2024). But how likely it is that such nonlinear seagrass responses can also induce tipping responses in the dynamics of carbon storage in seagrass meadows remains unknown.

Although mechanisms and models for describing tipping responses in seagrass ecosystems have been proposed (van der Heide et al., 2007a; Carr et al., 2010; Christianen et al., 2014; Maxwell et al., 2017; Ruiz-Reynés et al., 2017; Adams et al., 2018; Mayol et al., 2022; Minguito-Frutos et al., 2023; Ruiz-Reynés et al., 2023), those have not been directly linked to the loss of carbon storage. This aspect is critical because these abrupt changes can directly influence carbon dynamics by modifying the biogeochemical processes and their potential feedbacks that ultimately will determine the fate of stored carbon. Perhaps the closest modeling approach is based on patch growth models simulating growth and colonization processes under different restoration scenarios and timeframes, by relating cover maps with carbon sequestration rates at local scales (Duarte et al., 2013). Another study modeled CO₂ emissions following the degradation of seagrass using decay rates of C_org in above- and below-ground tissues and underlying sediments (Lovelock et al., 2017). For both restoration projects and seagrass degradation scenarios, the most common model used is the Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST) model (Moritsch et al., 2021; González-García et al., 2022). The InVEST model estimates the carbon sequestration potential of seagrass ecosystems by calculating the amount of carbon that is stored in seagrass biomass, as well as the carbon that is stored in the underlying sediment. Yet, to the best of our knowledge, no mechanistic model has so far been developed to evaluate the potential cascading effects of tipping points in seagrass and associated carbon dynamics (see Supplementary Material S1). If abrupt seagrass shifts could also lead to cascading responses in carbon source-sink dynamics, it is important to understand under which conditions and at what scale such an effect takes place.

Here, we propose a minimal modelling perspective for studying the likelihood of tipping point responses in the carbon storage (i.e., from carbon sink to source of CO2) of seagrass ecosystems under global change stressors. We show how conceptual mechanistic models could be built based on expert knowledge by coupling a minimal set of biogeochemical interactions and ecological feedbacks. We then develop one such model to showcase that seagrass meadows can abruptly shift from sink to carbon source in response to anthropogenic induced physical damage. We use this type of anthropogenic induced mortality only as an example, as the mechanistic modelling framework we present could be used to study any other pressure by itself or in combination. We outline limitations and opportunities presented by such minimal modeling approaches towards improving the management of seagrasses to enhance their functioning as carbon storage systems.

We drew a simplified overview of the main processes that govern interactions in the carbon storage of seagrass ecosystems (Figure 1). We distinguished the three pools that make up the carbon stores in seagrass ecosystem: a) the seagrass, b) the upper sediment layer (i.e., the newly deposited organic matter (OM) within the rhizosphere that undergoes faster mineralization), and c) the lower sediment layer (i.e., the more stable OM pool). The origin of organic carbon in the sediment can be autochthonous (i.e., deriving from above- and below-ground tissues) or allochthonous (i.e., deriving from the suspended material trapped by seagrasses), often with similar contributions by the two sources (Kennedy et al., 2022). During the mineralization of OM in the sediment, nutrients are released to pore water and the water column (Figure 1). The factors affecting the different compartments were grouped according to whether they represent environmental settings (i.e., hydrodynamics, nutrient and light availability, and sediment origin, texture and composition), long-term stressors (e.g., warming, eutrophication, siltation, invasive species, herbivory), or acute short-term disturbances (e.g., mechanical damage, storms, heatwaves). Figure 1 summarizes how each of these factors potentially affects the processes that govern carbon dynamics across the three carbon pools in seagrass ecosystems.

To build our mechanistic model that couples ecological and biogeochemical dynamics in seagrass meadows we used the processes described above (Figure 1) that were derived through expert knowledge and previous theoretical work. Thus, we did not build nor parameterize a model based on collected ecological data but only developed a conceptual model based on processes. Furthermore, we did not include all described processes (Figure 1), but we focused on the impact of anthropogenic stress – related to physical damage like dredging and anchoring – that could lead to seagrass meadow degradation. There are two dynamical parts in the model: the ecological dynamics that describe seagrass growth (i.e., metabolic capture of carbon) and the biogeochemical dynamics that describe carbon storage in the sediment. Our aim is to show how such a mechanistic model can help understand the qualitative responses of seagrass carbon storage to stressors, even if we cannot fully describe and parameterize the model (Scheffer and Beets, 1994).

The bifurcation diagrams of Figure 2 show the asymptotic equilibria of seagrass and the three carbon pools within the sediment for a given level of anthropogenic induced mortality m _A . The greater the anthropogenic mortality (Figure 2A), the more likely it is that the meadow shifts from a seagrass-vegetated state to a bare state. At a certain threshold of anthropogenic induced mortality m _A , the seagrass meadow crosses a tipping point and shifts to a bare state with no standing seagrass biomass, thereby leaving the sediment exposed to faster mineralization and erosion. Looking at the carbon stored in each compartment in relation to the same anthropogenic mortality gradient (Figure 2B), we find that the carbon mass in each compartment responds in a similar way to the seagrass dynamics under different hydrodynamic regimes. Both allochthonous and autochthonous carbon mass decreases with increasing anthropogenic stress. In fact, the absence of standing seagrass biomass results in the loss of   C A u inputs and their erosion from the sediment, and in a greater hydrodynamic intensity that causes fewer particles to be trapped and less   C A l input. Consequently, the C _St follows the same decreasing pattern. Eventually, all carbon pools experience a tipping point to a low carbon storage state (that can be interpreted as carbon loss) as the tipping event in the seagrass ecological dynamics cascades on the carbon dynamics.

In this perspective paper, we suggest that there are thresholds beyond which seagrass meadows turn from net carbon sinks to net sources of CO₂ due to the crossing of tipping points in seagrass meadows. We described a set of positive and negative interactions at the level of seagrass meadows, as well as between the sediment and the seagrass that altogether affect carbon dynamics in seagrass meadows (Figure 1). We proposed that incorporating this knowledge into a coupled ecological-biogeochemical conceptual model can improve our understanding of the potential response of carbon storage to seagrass meadow degradation. We then built a simplified version of such model to showcase how tipping events in seagrass meadows driven by anthropogenic stress could cascade into the sedimentary carbon dynamics, leading to the eventual loss of stable carbon that implies a transition of the meadow from a net sink to a source of CO₂ (Figure 2).