Unlocking flow–habitat relationships in mountain rivers of Epirus, Greece using object detection and hydrodynamic simulation

Human activities impact aquatic ecosystems by altering abiotic and biotic factors, which in turn affect habitat structure and biodiversity. Environmental flows, or the necessary water flow levels to sustain ecosystems, influence fish habitats, with flow variations affecting fish distribution, migration, and behavior. This study integrates machine learning (ML) algorithms, specifically the Faster Region-Based Convolutional Neural Network (Faster R-CNN) and the You Only Look Once (YOLO) model, with ecohydrodynamic modeling to assess fish microhabitat suitability under varying flow conditions. Microhabitat characteristics of individual West balkan trout (Salmo farioides) were studied in transitional zones between pool and riffles across three representative reaches of the Voidomatis River, with 16 locations sampled at a discharge of 4.1 m<sup>3</sup>/s. Hydrodynamic simulations for 18 discharge scenarios indicated that habitat suitability peaked at a discharge of 14.8 m<sup>3</sup>/s, corresponding to optimal depth (~1.0 m) and velocity (~0.6 m/s) conditions. The resulting maximum Weighted Usable Area (WUA) reflected the best combination of hydraulic parameters for sustaining fish habitat. Both Faster R-CNN and YOLO effectively detect fish in visually noisy, turbid environments, achieving F1-scores above 90 % in their best configurations. Notably, Faster R-CNN outperforms YOLO across the primary performance metric (mAP50-95). While underwater video is limited under high-flow conditions and may miss some fish behaviors, combining it with habitat-hydrodynamic modeling provides valuable insights into microhabitat use. By integrating ML-based detection, hydrodynamic models, and habitat suitability curves, this research offers a robust framework for assessing fish habitats and understanding how flow variability may impact habitat quality. These insights are vital for effective conservation and river management strategies, ensuring the sustainability of aquatic ecosystems.

Authors

Institutions

Publication Details

Journal
The Science of The Total Environment
Published
2025-10-15
DOI
https://doi.org/10.1016/j.scitotenv.2025.180703
Primary Topic
Fish Ecology and Management Studies
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Unlocking flow–habitat relationships in mountain rivers of Epirus, Greece using object detection and hydrodynamic simulation

Christina Papadaki, Sergios Lagogiannis, Elias Dimitriou, Antonis Kavvadias et al.
The Science of The Total Environment
Fish Ecology and Management Studies
article

Unlocking flow–habitat relationships in mountain rivers of Epirus, Greece using object detection and hydrodynamic simulation

Christina Papadaki, Sergios Lagogiannis, Elias Dimitriou, Antonis Kavvadias, Stamatis Zogaris, Dimitris N. Makropoulos
article en

Abstract

Human activities impact aquatic ecosystems by altering abiotic and biotic factors, which in turn affect habitat structure and biodiversity. Environmental flows, or the necessary water flow levels to sustain ecosystems, influence fish habitats, with flow variations affecting fish distribution, migration, and behavior. This study integrates machine learning (ML) algorithms, specifically the Faster Region-Based Convolutional Neural Network (Faster R-CNN) and the You Only Look Once (YOLO) model, with ecohydrodynamic modeling to assess fish microhabitat suitability under varying flow conditions. Microhabitat characteristics of individual West balkan trout (Salmo farioides) were studied in transitional zones between pool and riffles across three representative reaches of the Voidomatis River, with 16 locations sampled at a discharge of 4.1 m<sup>3</sup>/s. Hydrodynamic simulations for 18 discharge scenarios indicated that habitat suitability peaked at a discharge of 14.8 m<sup>3</sup>/s, corresponding to optimal depth (~1.0 m) and velocity (~0.6 m/s) conditions. The resulting maximum Weighted Usable Area (WUA) reflected the best combination of hydraulic parameters for sustaining fish habitat. Both Faster R-CNN and YOLO effectively detect fish in visually noisy, turbid environments, achieving F1-scores above 90 % in their best configurations. Notably, Faster R-CNN outperforms YOLO across the primary performance metric (mAP50-95). While underwater video is limited under high-flow conditions and may miss some fish behaviors, combining it with habitat-hydrodynamic modeling provides valuable insights into microhabitat use. By integrating ML-based detection, hydrodynamic models, and habitat suitability curves, this research offers a robust framework for assessing fish habitats and understanding how flow variability may impact habitat quality. These insights are vital for effective conservation and river management strategies, ensuring the sustainability of aquatic ecosystems.

The Science of The Total EnvironmentVol. 1003
Hellenic Centre for Marine Research (GR)
Openalex Percentile: Top 71%
Fish Ecology and Management Studies
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.