Freshwater fisheries ecology encompasses the intricate interactions among fish species, their aquatic habitats, and human activities

Transitional water ecosystems exhibit pronounced spatio-temporal variability and increasing anthropogenic pressures, posing substantial challenges for ecological monitoring and management. Artificial Intelligence (AI), particularly Machine Learning (ML) and Deep Learning (DL), has emerged as a powerful framework for addressing the structural complexity of these systems. This systematic review synthesizes peer-reviewed studies applying ML and DL to ecological research and monitoring in transitional waters. A structured search of the Scopus^® database was conducted up to 31 December 2024, and studies were screened according to predefined eligibility criteria and PRISMA 2020 guidance; methodological quality was appraised using a structured assessment framework. Ninety-six studies met the inclusion criteria. Regression was the most frequent analytical task (44.1%), followed by classification (36.2%) and clustering (19.7%), with water quality monitoring representing the dominant thematic domain. Tree-based and kernel-based ML models prevailed overall, whereas DL architectures increased markedly after 2020, particularly in remote sensing and high-dimensional applications. Despite methodological heterogeneity and variable validation practices, the evidence indicates that ML and DL approaches effectively accommodate non-linearity, data heterogeneity, and scale mismatches typical of transitional waters. Standardized validation strategies and improved model interpretability remain essential for robust ecological inference and operational implementation.

Three ageing reservoirs in Ekiti State, Nigeria (Ureje constructed in 1958, Egbe in 1982, and Ero in 1989), were comparatively assessed to evaluate fish assemblages and their conservation relevance. Despite the absence of formal fisheries governance, all three reservoirs supported temporally stable fish communities with low overall diversity. A core assemblage of six species dominated across sites, while species richness increased from seven species in the small urban Ureje reservoir to nine species in the larger and more rural Ero reservoir. Four native species that have become locally scarce in surrounding river systems (Heterotis niloticus, Parachanna obscura, Hepsetus odoe, and Hyperopisus bebe) persisted at low but consistent abundance. Aquatic environmental variables remained within suitable limits for freshwater fishes, and trophic structure appeared intact across the reservoirs. Catch density was substantially higher in the urban reservoir compared to the rural systems, reflecting spatial differences in fishing intensity. Overall, the findings demonstrate that small tropical reservoirs can function as important freshwater habitats that sustain fish biodiversity and fisheries production in modified landscapes.

Understanding how environmental variability structures essential fish habitat (EFH) is central to managing keystone populations in dynamic marine ecosystems. Using nearly three decades of standardized survey data (1994–2021) from the Adriatic Sea, we present the first basin‐scale, process‐based framework linking estimated habitat suitability and abundance for the European anchovy Engraulis encrasicolus . Maximum entropy (MaxEnt) species distribution models were coupled with quantile generalized additive models (QGAMs) to quantify spatiotemporal dynamics of EFH and identify environmental thresholds that regulate population expansion and contraction. Predicted habitat suitability was primarily driven by variation in bottom temperature and biotic interactions, with persistent core areas along the northern and central shelves. Quantile analyses revealed asymmetric abundance responses to environmental gradients, indicating stronger constraints under unfavorable hydrographic regimes. This integrative approach bridges potential habitat and population dynamics, offering transferable tools for climate‐ready, ecosystem‐based fisheries management.