Deep learning models, known as convolutional neural networks (CNNs), have paved the way for reliable automated image recognition. These models are increasingly being applied in research on freshwater biodiversity, aiming to enhance efficiency and taxonomic resolution in biomonitoring. However, insufficient or imbalanced datasets remain a significant bottleneck for creating high-precision classifiers. The highly imbalanced data, where some species are rare and others are common, are typical of the composition of most benthic communities. In this study, a series of CNN models was built using 33 species of aquatic insects, with datasets ranging from 10 to 80 individuals, to determine the optimal number of individuals each class should have to build a high-precision classifier. We also consider the effect of class imbalance in the training dataset and the use of oversampling technique. The results showed that a robust model with acceptable accuracy (99.45%) was achieved with at least 30 individuals per class. A strongly imbalanced dataset caused an approximately 2% decrease in classification accuracy, while a moderately imbalanced dataset had no significant effect. The application of the oversampling technique enhanced in 1.88% the accuracy of strongly imbalanced models. These findings can help effectively tailor future aquatic macroinvertebrate training datasets.

Species Taeniopteryx hubaulti Aubert, 1946 and T. schoenemundi (Mertense, 1923) have been emphasized as endangered or vulnerable species in several European countries due to their local and limited distribution. More intensive research of these species on the territory of Serbia was conducted from 2011 to 2022. Taeniopteryx hubaulti was registered in seven localities. Five of these localities (Brusnička and Jerma rivers, the spring of the Raška River, and two localities on Grza River, spring and tufa barriers) were recorded as new sites of this species in Serbia. Two localities (Barska River and Lisinski Stream) had already been known from previous studies. The species T. schoenemundi was recorded in four new localities (Dulenska, Visočica, Temštica, and Trnavska rivers). Both species were found in waters with high oxygen concentration, relatively uniform low water temperature, and variable values of hardness, electrical conductivity, and pH values, with a small amount of inorganic nutrients. Considering the rare occurrence and prominent importance of these species as biological indicators of water quality and climate change, their findings have a great deductive value. This research represents an important contribution to the insight into the distribution of T. hubaulti and T. schoenemundi species.

The Lepenica River is the most important watercourse in the city of Kragujevac. Regardless of its importance, this river and its tributaries are extremely poorly investigated from a biological aspect. Only a few investigations were conducted from the end of the 20th century. After that, only physicochemical and microbiological parameters were monitored by scientists and Public Institutions. Our research was performed in 2021 at five localities at the Lepenica River and included an analysis of phytobenthos, macroinvertebrate, and fish communities, with the aim to assess ecological status/potential according to National Regulative. Our results indicate that the ecological status of this river was good (II class) only at one locality before the industrial and urbanized zone of the city of Kragujevac. From the entrance into the city of Kragujevac, the Lepenica River becomes highly polluted by multiple pollutants and belongs to the V class of ecological potential.

The region of South-East Europe, located in geologically and climatically diverse areas, hosts a wide range of freshwater habitats. However, comprehensive studies of macroinvertebrate communities are limited, and research on Chironomidae (Diptera) is particularly scarce. We present data on the diversity and structure of chironomid assemblages in hilly and mountainous streams across three ecoregions: the Pannonian Lowland (Ecoregion 11), the Dinaric Western Balkans (Ecoregion 5), and the Eastern Balkans (Ecoregion 7) and provide a comparative overview of their community patterns based on 130 samples. According to the CCA results and Monte Carlo permutation tests, water temperature, dissolved oxygen, conductivity, pH, and altitude were identified as statistically significant parameters influencing Chironomidae assemblages across the ecoregions, collectively explaining 72.20% of the variation. The higher diversity indices were recorded in each season in the Pannonian Lowland and the highest within-ecoregion similarity. Dissimilarity was highest between ER11 and ER7 and lowest between ER5 and ER7. These results demonstrate that the ecoregion was the strongest influence of the studied environmental variables on Chironomidae assemblages, with community patterns closely reflecting their spatial distribution across distinct ecoregional settings.