Endemic freshwater fish from semiarid environments are among the most threated species in the world due to the water overexploitation and habitat fragmentation problems. Stepped or pool-type fishways are used worldwide to reestablish longitudinal connectivity and mitigate fish migration problems. Many of them are being installed or planned in rivers of semiarid environments, however, very few studies about fish passage performance through pool-type fishways has been carried out to date on these regions. The present work focuses on the passage performance of two potamodromous cyprinids endemic of these regions, with different ecological and swimming behavior: southern Iberian barbel (Luciobarbus sclateri) and Iberian straight-mouth nase (Pseudochondrostoma polylepis), in two of the most common types of stepped fishways: vertical slot and submerged notch with bottom orifice fishways. Experiments were carried out during the spawning season in the Segura River (South-Eastern Spain), using a PIT tag and antennas system. Ascent success was greater than 80%, with a median transit time lower than 17 minutes per meter of height in all trials and for both species and fishway types. Results show that both types of fishways, if correctly designed and built, provide interesting alternatives for the restoration of fish migration pathways on these regions.

Bayesian techniques for engineering problems, that rely on Gaussian process (GP) regression, are known for their ability to quantify epistemic and aleatory uncertainties and for being data efficient. The mathematical elegance of applying these methods usually comes at a high computational cost when compared to deterministic and empirical Bayesian methods. Furthermore, using these methods becomes practically infeasible in scenarios characterized by a large number of inputs and thousands of training data. The focus of this work is on enhancing Gaussian Process-based metamodeling and model calibration tasks, when the size of the training datasets is significantly large. To achieve this goal, we employ a stochastic variational inference algorithm that enables rapid statistical learning of the calibration parameters and hyperparameter tuning, while retaining the rigor of Bayesian inference. The numerical performance of the algorithm is demonstrated on multiple metamodeling and model calibration problems with thousands of training data.