This review presents the main results of a 10‐year research study conducted in a Mediterranean intermittent basin (Evrotas River). By assembling the main outcomes of past and ongoing research projects, this study provides an overview of multiple stressor effects, with emphasis on water scarcity, focusing on hydro‐biogeochemical processes, as well as on spatial and temporal variations in benthic macroinvertebrates and fish fauna. The major impact in the basin has been the over‐exploitation of surface and groundwater resources, which, in combination with droughts, has resulted in the recurrent artificial desiccation of large parts of the hydrological network. The response to intermittency of the macroinvertebrate fauna is characterized by high resilience through various drought‐resistant evolutionary mechanisms, with assemblages recovering successfully after recurrent droughts. However, when pollution is evident in combination with drought, effects on benthic species richness, abundance, and assemblage structure can be severe. Similarly, pollution and water stress may result in massive fish mortalities due to hypoxic conditions, with fish populations requiring long periods to recover. However, the fish fauna appears to be relatively resilient to drought‐driven reach‐scale desiccation, and ultimately recovers, provided that aquatic refugia are available to supply colonists and that there are no physical barriers impeding recolonisation. Appropriate conservation measures are urgently required to address the effects of recurrent bouts of water stress, as well as of other stressors on the freshwater communities of the Evrotas River, both at the level of water management and of water policy and at the local and the national level.

Although various methods are currently available for modelling the habitat preferences of aquatic biota, studies comparing the performance of data-driven habitat models are limited. In this study, we assembled a benthic-macroinvertebrate microhabitat-preference dataset and used it to evaluate the predictive accuracy of regression-based univariate Habitat Suitability Curves (HSC), Boosted Regression Trees (BRT), Random Forests (RF), fuzzy-logic-based models using the weighted average (FLWA), maximum membership (FLMM), mean of maximum (FLM) and centroid (FLC) defuzzification algorithms and fuzzy rule-based Bayesian inference (FRB). The results show that the BRT model was the most accurate, closely followed by RF, FRB, FLM and FLMM while the FLC and FLWA algorithms had the lowest performance. However, due to the imbalanced nature of the dataset and in contrast to the fuzzy rule-based algorithms, the HSC, BRT and RF models failed to accurately predict the habitat suitability in low-scored microhabitats. We conclude that, given balanced datasets, BRT and RF can be effectively used in habitat suitability modelling. For imbalanced datasets, a properly adjusted RF model can be applied but when the input dataset is large enough to provide sufficient data-driven IF–THEN rules to train an FRB, FLMM or FLM algorithm, these models will produce the most accurate predictions.