Ecological Modelling

Non-compliance with regulations threatens the sustainability of fisheries worldwide. Understanding the interconnected feedbacks of this complex social-ecological problem is key for sustainability but rarely integrated into fisheries management. We provide an adaptive stochastic modelling framework that integrates economic, social behavior, and ecological aspects of the Chilean kelp fishery, which plays a critical economic and ecological role in coastal social-ecological ecosystem. High levels of non-compliance is threatening sustainability, fishers well-being, and ecosystem health. Our model considers inherent environmental uncertainties and enables the assessment of different harvesting and compliance scenarios and the role of market-based economic incentives in reducing non-compliance. Results show that, unlike the sustainability obtained under an idealized full-compliance scenario, under dynamic compliance the social, economic, and ecological feedbacks leads to system collapse. Importantly, price premiums can promote compliance and sustainability, but the probability of collapse, albeit small, still exist. Our generalizable stochastic modeling framework evidenced that accounting for inherent uncertainty in natural resource management is key to designing interventions for sustainability.

Guinea worm disease eradication efforts may benefit from environmental surveillance methods capable of detecting infected copepod intermediate hosts in aquatic habitats. We developed a three-dimensional, spatially explicit agent-based model to examine how ecological processes influence detection probability for a hypothetical water sampling method. The results show that surveillance sensitivity is shaped by the combined effects of larval diffusion, copepod density, and pond size, with interactions among these factors producing nonlinear relationships. Detection, in our model, was concentrated within a relatively restricted period after larvae matured to the infective stage and before dispersal and mortality reduced presence, indicating a limited spatiotemporal window for effective sampling. Surveillance performance peaked under intermediate dispersal regimes that generated sufficient spatial overlap between larvae and intermediate hosts, while both limited dispersal and excessive diffusion reduced detection by constraining encounters or diluting larval concentrations. Increasing habitat size reduced detection by diluting larval concentrations, but the magnitude of this effect depended on copepod density and dispersal dynamics, producing nonlinear and threshold responses rather than simple scaling with pond volume. Spatial and temporal patterns of detection shifted as larvae dispersed, with the most favorable detection periods occurring when both larval abundance and intermediate host encounters were elevated. These findings indicate that surveillance can be guided by local ecological conditions. When the timing of larval introduction is uncertain, effective surveillance requires repeated sampling over time to capture transient windows of detectability and the sampling will be less effective in very stagnant and highly mixed waterbodies. Overall, this study demonstrates how mechanistic modeling can support the design and interpretation of environmental surveillance strategies for Guinea worm eradication programs. Author summaryGuinea worm disease is close to eradication but confirming that transmission has fully stopped remains difficult because detecting infectious larvae in water is challenging. Transmission depends on freshwater copepods that become infected after ingesting Guinea worm larvae. These copepods are short-lived and unevenly distributed within ponds, and infected individuals may die before larvae reach the infective stage. As a result, environmental detection is inherently uncertain. We developed a three-dimensional agent-based model to simulate larval dispersal, copepod infection, and water sampling in a pond environment. The model shows that detection is constrained to a brief period when mature larvae and copepods overlap in space and time, and that this window depends strongly on local ecological conditions such as larval dispersal, copepod density, and pond size. Because infected copepods can be present outside these narrow detection windows, negative water samples do not necessarily indicate absence of transmission, highlighting the need for repeated, spatially targeted surveillance during the final stages of eradication.

Multiple global-change forces, from habitat alterations to warming, are altering food webs and trophic interaction strengths. Such changes in trophic interactions have important implications, as it is a tenet of ecology that trophic interactions are linked to the functioning and stability of ecosystems. For example, changes in the presence or strength of intraguild predation (IGP), the consumption of a predator by another predator that competes for shared prey, can have cascading effects on the biomasses of species and trophic levels. For this reason, IGP can affect key ecosystem functions at the base of the food web and is of special interest to practitioners of biological pest control. However, the relationship between IGP and ecosystem stability is not yet well understood, especially whether and how IGP might affect the stability of non-adjacent lower trophic levels including primary producers. In this study we simulate the dynamics of a six-species, four-trophic-level food web plus a limiting nutrient to explore the relationship between IGP strength and the temporal variability of species- and trophic group-biomass. By varying the IGP rate given the abundance of the eaten predator, we find that the model food web abruptly shifts between equilibria in which all species maintain either constant biomass or stable limit cycles where all trophic levels exhibit sustained and significant oscillations. While complex feedback in the model creates a divergence between the IGP functional response and the resulting realized IGP strength, both stronger IGP functional responses and stronger realized IGP are associated with a higher likelihood of oscillations. Furthermore, analyses indicate that the strongest consumptive interaction induces the oscillating behavior in an indirect effect initiated by the change in IGP. Overall, these results suggest that as food web structure changes in ecosystems worldwide, strengthening IGP runs the risk of inducing destabilizing effects that extend to the base of food webs, while weakening IGP could confer stability to ecosystem functions such as primary production. Finally, we discuss relevance to management, including the implication that IGP among biological control agents should be minimized to maintain stable crop production.

Wildfire regimes in Mediterranean landscapes are undergoing significant changes due to the combined effects of land-use transitions and climate change. In particular, land abandonment increased fuel availability, the expansion of the wildland-urban interface increased ignition frequency, while climate change increases the chances of fire-weather conditions and reduces vegetation recovery capacity. This study presents a modelling framework to investigate the coupled dynamics of fire and vegetation under different fire regimes scenarios, using a case study in central Italy (Monte Pisano). The approach integrates two cellular automata models for vegetation dynamics (Batllori et al., 2017) and for fire-spread (PROPAGATOR; Trucchia et al., 2020). The vegetation model represents succession among six functional classes, including grasslands, shrubs, and trees with different fire-response strategies (seeders and resprouters), while explicitly accounting for post-fire recovery processes. The model was calibrated for the area using historical fire perimeters and vegetation maps over 40 years. Fire spread is simulated probabilistically using PROPAGATOR, driven by fuel types, topography, and weather conditions. A stochastic coupling was implemented by sampling fuel classes from vegetation composition, and by feeding simulated burned areas back into the vegetation model, thus enabling dynamic fire-vegetation feedback. Future wildfire scenarios are constructed by linking ignition probability to fire-weather conditions derived from historical reanalysis data (1981-2023). Extreme fire events are defined based on thresholds of wind speed and fuel moisture, and their probability of occurrence is varied across scenarios to represent increasing climate-driven risk. Simulations are performed over a 100-year horizon starting from current vegetation conditions. Results show that, in the absence of fire, vegetation dynamics lead to dominance of late-successional, fire-resilient species (resprouters). This is particularly evident for low probabilities of extreme fire events, with fire impacts diminishing over time as landscapes become less flammable. However, increasing the frequency of extreme fire conditions resulted in persistent disturbance, maintaining higher proportions of shrubs and early successional vegetation, and sustaining elevated burned areas over time. Overall, the study shows that coupling fire spread and vegetation dynamics provides a useful framework for exploring long-term ecosystem trajectories under climate change. The results highlight the critical role of extreme fire events in shaping landscape resilience and suggest that future management strategies should account for fire-vegetation feedbacks to support more stable and less fire-prone ecosystems.

Humans are efficient and deadly predators, yet they may also interact with wildlife in non-lethal ways. This study explores how interactions with lethal and non-lethal human "superpredators" alter predator-prey dynamics using an agent-based modelling approach. Our model incorporates both the consumptive (lethal) and non-consumptive (behavioural) effects of humans, as well as of predators on prey. We explored how the replacement of apex predators by humans affects mesopredator-prey dynamics, with particular emphasis on trophic targeting and differences between lethal and non-lethal interactions. We found that human superpredators have a greater effect on model outcomes than apex predators. When superpredators consume mesopredators alone or with prey, the probability of mesopredator-prey coexistence increases to a greater extent than when apex predators consume mesopredators. In contrast, superpredators consuming only prey slightly increases overall extinction risks and reduces coexistence. Non-lethal superpredators, despite eliciting anti-predator responses in mesopredators and prey, had a negligible effect on population dynamics. Our findings demonstrate that human superpredators may functionally replace apex predators when they are lethal. However, non-lethal interactions with humans may not be as ecologically significant as lethal interactions, even when humans induce anti-predator responses.

Larval dispersal models are central to mapping and predicting ichthyoplankton dynamics in the ocean, yet despite decades of refinement they remain fundamentally limited by their ability to represent adaptive behaviors, relying instead on static trait parameterizations. This deficiency constrains our capacity to design effective restoration and mitigation strategies in an increasingly stressed ocean. SWARM (Simulating Waterborne Agent Routes for Marine connectivity) overcomes this barrier by integrating Large Language Model (LLM)-based behavioral agents with a standard biophysical model to simulate active decision-making during the pelagic larval stage. In both idealized and realistic conditions focusing on Red Snapper larvae in the Gulf of Mexico, agents develop adaptive behaviors that improve settlement and generate explainable vertical distribution patterns. SWARM demonstrates that LLMs can overcome long-standing limitations in dispersal modelling by explicitly representing behavioral drivers of movement, opening new pathways for predicting connectivity and designing effective marine-ecosystem restoration.

Agroecosystems are dynamic ecosystems, constituted by patches of vegetation and agricultural use, where biodiversity is shaped by spatial and temporal variability. While most studies have focused on spatial composition and configuration, the role of temporal variability remains poorly understood. Yet, temporal dynamics can strongly modify species composition, abundance, and persistence in ecological communities. Temporal variability is particularly relevant in agroecosystems with rainfed agriculture where environmental conditions shift dramatically between rainy and dry seasons. In this paper, we assess the role of temporal variability on biodiversity maintenance in an agricultural matrix using a metacommunity model that simulates an agricultural landscape under rainfed conditions, that is, with abrupt seasonal changes in the agricultural patches. This model couples a local community network dynamic with a migration dynamic and is based on empirically documented features of rainfed agricultural matrices. Our results show that temporal variability provides new opportunities for species to recover from low densities. However, the effect of temporal variability is not straightforward. It depends on the initial and final conditions, the migration and mortality rates and the intensity of temporal variability. Overall, our findings highlight the need to further investigate temporal variability to better understand its role in shaping biodiversity in agricultural landscapes.

Coexistence of multiple predators in ecological communities and their combined effects on the abundance and diversity of shared prey are often difficult to predict. In some theoretical models, predator coexistence is limited by antagonistic interactions, especially in the form of intraguild predation (IGP) that typically leads to out-competition between predators and high prey densities. However, empirical studies show that predator coexistence is common even in the presence of IGP. This discrepancy between theoretical expectations and empirical observations can be highly relevant for practical applications like using multiple natural enemies for pest suppression in agriculture. It is proposed that greater functional differences between natural enemies (i.e. predators) could reduce competition and overcome the negative effects of IGP, thereby promoting their coexistence and enhancing herbivore (i.e. prey) suppression. In this study, we theoretically explore this proposition. We develop a theoretical model based on the types of natural enemies of aphids to identify how functional differences between predators in IGP modules affect predator coexistence and herbivore suppression. We show that pairwise combinations of four functional predator types (ladybird, predatory bug, hoverfly, and parasitoid) can increase the coexistence range for different intraguild predation and competition strengths between predators (IGP symmetry), along a productivity gradient. This outcome depends on the external food input rate for the predatory bug and hoverfly types, and on their position as IG predator or IG prey. Herbivore suppression was primarily driven by IGP symmetry (i.e. the relative intraguild predation and exploitative competition strength between predators) and was especially pronounced in competitive-like modules where the IG predator was excluded for most scenarios. However, for some competitive-like IGP modules with predatory bug and hoverfly types, both predators can persist and provide a high herbivore suppression across increasing productivity. Our results can help explain experimental findings in conservation biocontrol, where coexistence between natural enemies is joined with effective herbivore suppression, and offer additional support for the role of functional diversity in reconciling theoretical predictions with experimental observations in multiple-predator communities.

In order to feed a growing global population without silencing nature, conceiving agricultural management strategies reconciling yield and conservation goals is key. Using numerical simulations of a metacommunity model, we explore the possibilities for compromise offered by spatial management strategies of farmed areas. Each strategy is characterized by its farming intensity, the proportion of farmed lands and their spatial aggregation. We show that achieving equitable yield-biodiversity compromise is difficult. While conciliatory strategies offering top yield and biodiversity are typically not possible, accepting slightly lower yields (ie, "Pretty Good Yield strategies") allows to recover substantial biodiversity. Such reconciliation possibilities are limited for species with small dispersal. Yield increases mainly through farmland expansion, whereas farming intensity strongly influences biodiversity, increasing it at low intensity before decreasing with further intensification. Finally, we demonstrate that reconciliation is easier if agricultural production relies on biodiversity through ecosystem services.

O_LIAnimal-mediated seed dispersal is key for the maintenance and functioning of tropical ecosystems. Specifically, in the Cerrado, the largest Neotropical savanna and a global biodiversity hotspot, nearly 60% of plant species rely on animals for dispersal. C_LIO_LIClimate change threatens these interactions by affecting species distributions, reshaping communities, and potentially decoupling plants from their dispersers. Anticipating how such disruptions may alter seed dispersal networks is particularly relevant for understanding the resilience of future tropical ecosystems. C_LIO_LIHere, we combined empirical data on 139 pairwise plant-frugivore interactions with species distribution forecasts to build probabilistic interaction matrices under present and future climate scenarios, which were then used to construct 6,221 local seed dispersal networks. Using ecological niche modelling, we tested how climate change influences species range size and centroid displacement. Then, we evaluated whether such changes translate into losses of pairwise plant-frugivore co-occurrence. Finally, we investigated how these changes in occurrence overlap may affect key structural properties of future local seed dispersal networks. C_LIO_LIWe forecast that by the 2070s, under a business-as-usual climate scenario, species are likely to contract their ranges by 56 {+/-} 33% and shift their distribution centroids by 88 {+/-} 57 km within the Cerrado, leading to a 27 {+/-} 29% loss in plant-frugivore co-occurrence mainly driven by reductions in plant species distributions. At the community level, these losses will lead to smaller and more nested networks and specialized, indicating a structural simplification of seed dispersal systems in the Cerrado. C_LIO_LISynthesis: By combining empirical data on animal-mediated seed dispersal with forecasts of species distributions, we found that climate change may simplify frugivore-plant interaction networks in the Cerrado by decreasing species ranges and co-occurrence of partners. Our study demonstrates that future climate may pose a threat not only to species distributions but also to ecological interactions, such as seed dispersal, that are key to enabling climate-tracking by plants. Thus, preventing the simplification of interaction networks will be essential to conserve biodiversity in species-rich regions. C_LI

Spatial population genetic and ecological modeling is often necessary to predict outcomes accurately. One example is gene drive, a rapid process involving spread of gene drive alleles through a population, usually to suppress pests or reduce transmission of vector-borne disease. Several existing models have been used to assess gene drive and other spatial processes. However, each of these has limitations, such as high computational cost and limited scalability, difficulty in incorporating environmental factors and complex lifecycles, or potentially simplified spatial structure. To overcome these challenges, we propose a hexagon-based computational framework that is designed to mimic continuous space for rapid genetic wave advances. This allows us to accurately simulate a larger spatial domain with lower computational investment. We implemented this model and compared the wave speeds of different gene drives with those obtained from other models. The results showed good agreement when hexagon width and dispersal were properly calibrated. We then determined optimal circular and linear (along roads) release patterns for a variety of gene drives and Wolbachia bacteria. To demonstrate the application of our framework to a hypothetical scenario, we constructed a model Culex quinquefasciatus mosquitoes on Hainan Island. We then evaluated the outcome of different gene drive release strategies, showing the transgenic insect release level necessary to achieve high gene drive coverage and how this could be further optimized based on mosquito and human distribution. Overall, our hex-based population genetic framework provides a flexible platform for realistic and large-scale models for gene drive and related applications.

The mosquito Aedes albopictus is a competent vector for the transmission of several arboviruses and is currently spreading across many continents. Since conventional control methods, like insecticides, often lead to environmental problems and the emergence of resistance, scientists developed alternative mosquito control strategies. One of the most used is the Sterile Insect Technique (SIT), which involves the mass release of males sterilized through irradiation. The Toxic Male Technique (TMT) is instead based on the release of genetically modified males expressing toxic proteins that kill females when they mate. Control strategies are often intended as methods to eradicate mosquito populations, yet a less ambitious and more cost-effective task is to reduce them such that the probability of transmission of viruses to humans becomes negligible. To compare the efficacy of these control strategies, we develop a mathematical model with two communicating compartments: a mosquito population and epidemiological model coupled with a human epidemiological model. As a proof-of-concept, we test the model using meteorological and entomological data for the Emilia-Romagna region. Our results indicate that the TMT strategy is more effective in lowering the probability of transmission and provides indication for the deployment of control strategies.

ContextThe movement of vertebrate hosts across urbanized landscapes can play a key role in the transmission of direct-contact diseases. Understanding how wildlife hosts move in urban landscapes, and how transmission is affected by their landscape-constrained and disease-altered movements, is imperative for better predicting the spread of disease. ObjectiveWe assess how the movement of red foxes (Vulpes vulpes) according to landcover type, and their infection status, affect the spread of mange (caused by Sarcoptes scabiei) in an urbanized landscape. MethodsWe developed a mange transmission model (MTM) using an agent-based model to compare two movement behaviours of foxes in Scarborough (Ontario, Canada): random and landcover-based. We further assessed the effects of movement on disease transmission by considering the foxs infection status and comparing a range of movement probability scenarios. We quantified the number of effective contact events and the effective reproduction number (Re) according to each scenario. ResultsWe found that both landcover-dependent movement and infection status influenced the spread of mange within fox populations. The number of effective contact events and effective reproduction number Re was greatest when landscape heterogeneity was included in the model and foxes moved through paths of least resistance to movement, and when susceptible and infected foxes had an equal probability of leaving a fragmented habitat patch. ConclusionsOur findings suggest that mange spread may be accelerated along movement corridors in fragmented, heterogenous landscapes. As urban areas expand and remnant habitat within these is further lost and animals are relegated to fewer movement pathways, disease transmission may increase.

Successful ecosystem restoration is intimately linked to the persistence of species in local communities and across landscapes. As such quantitative approaches to ecological restoration require the integration of community and metapopulation ecology. Together these disciplines demonstrate that local colonisation, via habitat connectivity and size, and species interactions, both modulate the process of community assembly. However, thus far restoration ecology still remains disconnected from network ecology this preventing a holistic, community-wide perspective to restoration. We aim to inform ecological restoration using a multi-layer modelling framework integrating ecological interactions and species dispersal dynamics. We explore the drivers that modulate recolonisation dynamics of species across restored landscapes. We further investigate how restoration improves the structural properties of food webs, the number of successful recolonisations and the role of configuration of restored patches in restoration outcomes. We find that recolonisation is the result of a trade-off between dispersal ability and energy requirements. 97% of plant recolonisation and 88% of herbivore recolonisations happened within close proximity to the source patches. Better dispersers - intermediate and top species in the food webs - were able to recolonise habitat by benefitting from the increased biomass influx from restoration. When only a small proportion of the landscape could be restored, the location and connectivity of restored areas strongly influenced the outcome of restoration: more connected patches enabled on average the recolonisation of about 1 additional intermediate species compared to that of isolated patches. However, this difference faded as soon as more patches were restored, and improving larger portions of the landscape always resulted in better outcomes. Restoring 1/3 of the landscape enabled on average the recolonisation of ~4 additional species. Our findings suggest that quantitative models can inform restoration efforts necessary to bring native species back to restored areas. They also suggest that attention should be given to the requirements of the recolonisers, the distance of their introduction from restored areas and their trophic and ecological niche. These aspects are crucial to assess their energy and habitat requirements for successful establishment.

A major goal of community ecology lies in the deciphering of the processes underlying species distribution. A widespread approach to this question is to identify patterns in species community data and relate them to possible processes. Joint Species Distribution Models (JS-DMs) offer one way to do so through the infernece of association networks that describe patterns of statistical correlations and dependencies between species, but it is unclear what processes can explain the presence of such correlations. While it has now been established that there is no equivalence between JSDM-inferred associations and biotic interactions, the later remain one possible explanation, among others, for the former. However, to our knowledge, there is no specific study of the statistical patterns induced by different types of interactions or of the conditions under which they may or may not appear as statistical correlations / dependencies in species communities. To explore these questions, we propose a "virtual ecologist" approach that consists in simulating community data based on abiotic and biotic processes with the VirtualCom model that emulates the effects of environmental processes and of competition and facilitation interactions. Then, we study to what extent JSDMs retrieve correlations between species that match the simulated interactions. We show that these interactions are better identified when using JSDMs that model partial correlations between species rather than marginal ones. We further demonstrate how critical it is to correctly model abiotic effects in order to identify biotic ones and that the "correct modelling" of these effects depend on the type of interactions at stake.

Phytoplankton encounter a range of light and nutrient conditions in nature and must adjust their internal carbon and nitrogen allocations to grow across different resource environments. Current phytoplankton carbon budget models treat respiration simply as a carbon loss. In reality, respiration is a critical cellular process that produces energy for nutrient uptake and cellular maintenance. Drawing on empirical evidence, we developed an eco-physiological model that incorporates a more realistic role of respiration. In our model, photosynthetic carbon is partitioned into: (i) the Pentose Phosphate Pathway (PPP) for assimilation and (ii) respiration for energy production that is then used in nutrient uptake. Stored nitrogen is partitioned between three pools: cellular structure, photosynthesis and nutrient uptake. Using an optimality-based approach, we identify strategies that maximize either exponential growth rate or competitive ability. We find that optimal internal allocations follow a growth-maintenance tradeoff, favoring population growth through carbon acquisition in nitrogen-replete conditions and population maintenance through nitrogen acquisition in nitrogen-limited conditions. The optimal allocations match empirically observed shifts in carbon partitioning at different dilution rates. Our model also generates an interactive growth response surface with an asymmetry, where light is the dominant limiting factor at low light intensities and co-limitation by light and nitrogen only occurs at high light levels. Furthermore, the model recovers the widely accepted Droop function for growth vs nitrogen quota and predicts a hyperbolic decline in growth vs energy quotas. Through a simple growth-maintenance tradeoff, our model provides a mechanistic foundation for predicting phytoplankton productivity in biogeochemical models.

We describe a statistical agent-based model (SABM) for binary spatio-temporal data in which the occupancy of each cell evolves as a Bernoulli mixture of three mechanistically distinct processes: local persistence, anisotropic neighborhood dispersal, and long-distance dispersal. The model is embedded in a hierarchical Bayesian framework with conjugate Beta full-conditionals for the persistence and long-distance parameters and a Dirichlet prior on the directional dispersal kernel. A nonstationary extension links the dispersal kernel to a latent habitat-suitability surface through directional gradients of a Gaussian process. We show that, in the small-step regime, the Lagrangian recurrence for the dispersal kernel scales to a classical two-dimensional advection-diffusion partial differential equation whose drift and dispersion coefficients are the first and second moments of the dispersal probabilities. We provide an MCMC algorithm exploiting the exact full-conditionals and demonstrate parameter recovery and PDE-scaling agreement in a simulated example.

AO_SCPLOWBSTRACTC_SCPLOWO_LICommunity assembly graphs (CAGs) summarize which species combinations can coexist and how single-species invasions drive transitions between them, encoding the pathways, alternative endpoints, and cycles that make up a communitys assembly history. Constructing CAGs from dynamical models requires methods that are both computationally tractable and faithful to the underlying ecological dynamics. However, existing methods rely on restrictive assumptions, such as global stability, that exclude alternative stable states and non-equilibrium dynamics known to occur in empirical systems. C_LIO_LIWe develop a computational pipeline that constructs CAGs from any generalized Lotka-Volterra model. Building on the invasion graph framework and its connection to permanence, the pipeline verifies that community dynamics are bounded, identifies which subsets of species coexist in the sense of permanence, determines which single-species invasions are dynamically realized, and assigns each community a topographic height equal to the length of the longest assembly path leading to it. We also provide a numerical algorithm to simulate the dynamics of community assembly. C_LIO_LIWe prove several general properties of the resulting graphs, including that a successful invader is never subsequently excluded and that, in the absence of assembly cycles, permanent communities can be reassembled by introducing their species one at a time in the right order. We prove that the CAG faithfully reproduces the compositional shifts seen in the numerically simulated dynamics of assembly. Applying the pipeline to three empirically based models (a New Zealand grassland, a European pasture, and a Puerto Rican ant community), we show how competition strength and mutualistic feedbacks reshape the assembly landscape and how intransitive competition generates assembly cycles. C_LIO_LIOur approach accommodates alternative stable states and non-equilibrium dynamics without requiring global stability, and it turns the long-standing landscape metaphor into a quantitative, mechanistically grounded object by resolving what "height" means. More broadly, it makes the topography of the assembly pathways measurable, providing a way to compare the historical contingency and predictability of the assembly in ecological systems. C_LI

As climate changes, habitat suitability for multiple taxa are also expected to change. In recent years, mangrove latitudinal range expansion has been linked to increasing temperatures and reduced freezing in temperate regions, happening mainly through encroachment into saltmarshes. The range limit of mangrove forests in Eastern South America has not seen drastic changes in the last four decades, despite trends of increased temperature and the seemingly favorable direction of the Brazilian Current. Here, we investigate if and how the distribution of South-Atlantic mangrove forests may respond to different scenarios of climate change. To do this, we combine ecological niche modelling with propagule dispersal simulations to understand the roles of climate and ocean currents in defining the austral limits of South-Atlantic American mangroves. Our results indicate that minimum sea surface temperature strongly constrains habitat suitability beyond the current distribution of mangroves (28{degrees}2868" S), while dispersal processes heavily limit propagule stranding beyond 35{degrees} S. The Brazil-Malvinas currents confluence zone creates steep temperature gradients and an oceanographic barrier that makes the latitudinal expansion of mangroves unlikely in this region, even in future scenarios of heating climate. We found no evidence of current nor future poleward expansion of mangroves, but total mangrove area has increased in Brazil over the last decades, likely due to landward migration, but anthropogenic interference and urban expansion may restrict this process, leading to coastal squeeze. Under scenarios where both landward and poleward migration are limited, South American mangroves may face increasing vulnerability, with potential impacts on the several ecological, biogeochemical and social cycles they support. Our results contribute to leading hypotheses of climate restriction and shed light on the role of ocean currents in South America, helping to explain why the poleward expansion reported in other regions has not yet been observed in the South-Atlantic mangrove range limit.

Environmental variability can strongly alter coexistence among competing species and their extinction risk, particularly when population dynamics are shaped by behavioral interactions, such as fear. In this work, we develop a novel stochastic differential equation competition model that incorporates both non-consumptive fear effects and environmental variability to investigate how behavioral interactions influence species coexistence under random fluctuations. Our result reveals that environmental stochasticity can drive species to extinction even when the corresponding deterministic system admits coexistence. In particular, under an explicit stability condition on the fear and competition parameters and sufficiently strong averaged noise intensities, we prove that both competing species become extinct exponentially almost surely. Conversely, we derive a stochastic persistence criterion in terms of fear, competition, and noise-induced suppression parameters for the fearful species. We further demonstrate that environmental noise may reverse classical competition-exclusion outcomes, leading to qualitatively different long-term dynamics from those predicted deterministically. These results provide rigorous thresholds separating stochastic extinction from persistence and highlight the critical role of environmental variability in fear-mediated competitive ecosystems. From an applied perspective, these results provide insight into how behavioral interactions and environmental variability influence species survival, with potential applications in ecological management and conservation.