This PhD thesis developed a multiscale framework to understand how chemical pollution shapes ecological dynamics in the Albufera Natural Park (ANP), a highly connected Mediterranean coastal wetland located in Valencia (Spain) and strongly influenced by rice cultivation. The thesis begins with a long-term analysis of multiple stressors in the Albufera over the last 50 years, based on time-series modelling using Generalized Additive Models. From this lake-scale perspective, the ecological dynamics of the ANP were shown to reflect the interaction of multiple pressures, with predominantly non-additive effects. Within this context, chemical pollution played a secondary role relative to hydrology and nutrient availability. In contrast, analyses conducted at the landscape scale using spatially explicit modelling approaches revealed chemical pollution as ecologically relevant and likely to affect ecosystem status and biodiversity within the ANP. Results showed that ecological risks are dominated by chronic exposure and complex mixtures, with several pesticides consistently exceeding accepted risk thresholds and generating distinct hotspots of exposure and risk across the ANP. Under chronic exposure scenarios, potentially affected fractions of species exceeded 50% in rice paddies and reached 30–45% in drainage ditches and the lake. Based on the results of the spatial risk analysis, a mesocosm experiment was conducted focusing on the compound that emerged as the most ecotoxicologically relevant in the system, the fungicide azoxystrobin. The experiment showed that this compound can cause severe effects on both ecosystem structure and functioning. For both experimental treatments, representative of spring habitats and eutrophic lake waters, no-observed-effect concentrations (NOECs) were ≤ 2 μg L⁻¹, comparable to concentrations recorded in the lake. The observed responses were dependent on ecological context, with clear differences between habitats. In addition, part of the impacts was driven by indirect effects, which amplified and propagated chemical pressure beyond the most sensitive taxa. The results of this thesis provide relevant information for understanding and managing chemical pollution in the ANP, clarifying how its ecological role varies across spatial and temporal scales. They also reveal important shortcomings of traditional ecological risk assessment approaches, particularly their limited ability to capture spatial heterogeneity in exposure as well as indirect and mixture-related effects in complex agricultural landscapes. More broadly, the thesis illustrates how integrating experimental results, field observations, and spatially explicit modelling across scales can support more realistic and ecologically meaningful risk assessments.
