Biosystems Diversity

Habitat classification of the Khortytsia Island floodplain after the Kahovka catastrophe using the European Nature Information System (EUNIS)

H. Tutova — 2026-03-12

This study aimed to analyse, categorise, and interpret the habitat diversity of Khortytsia Island's floodplain in Ukraine. It e m ployed the European Nature Information System H abitat C lassification alongside an integrated approach that incorporated geobotanical, geomorphological, bioindication, and remote sensing methods. A total of 876 vegetation relevés from 2024–2025 were examined. Plant communities were initially grouped through Two-Way Indicator Species Analysis, with diagnostic species identified based on fidelity values and indicator analysis. These floristic groups were then assigned to habitat types by comparing them with the system's key navigation, factsheets, formal definitions, and official crosswalks. To understand the main factors driving habitat differentiation, analyses included geomorphological variables, ecological indicator values, plant strategy analysis, M ultivariate A nalysis of V ariance, C anonical C orrespondence A nalysis, and P artial C anonical C orrespondence A nalysis. Habitat mapping using remote sensing was conducted to analyse spatial and temporal changes from 2022 to 2025. Ten terrestrial habitat types and two water-associated types were distinguished. The terrestrial habitats comprised pioneer sandy vegetation, drawdown-shore communities, riverine willow scrub, humid tall-herb assemblages, reed canary-grass beds, willow gallery forests, mixed poplar riparian forests, floodplain oak forests, and sandy steppes. Habitat differentiation was primarily influenced by geomo r phology ( especially elevation, topographic position, and wetness ) distinguishing low alluvial depressions, wet floodplain zones, and elevated sandy terraces. Ecological indicator values revealed that edaphic moisture, air humidity, soil pH, and nitrogen levels were the strongest predictors of habitat differences. These gradients mainly reflected hydrological, trophic, and substrate-chemical variations, with plant strategies indicating distinct traits related to competition, stress tolerance, and disturbance adapt a tion. Spatial analysis showed a shift from a relatively stable floodplain pattern in 2022 to a highly dynamic mosaic in 2024–2025, driven by drainage, substrate exposure, and faster succession. The findings demonstrate that the European Nature Information System offers a robust framework for interpreting habitat changes after disturbances, supports comparable biodiversity asses s ments globally, and provides vital information for conservation planning in the fo rmer Ka hovka floodplain. The study unde r scores the conservation potential of this area amidst ongoing ecological reorganisation.

Application of multilayer perceptron neural networks in classifying management effects based on morphometric traits of Carabus granulatus

V. Langraf — 2026-03-22

Morphometric traits are essential tools for understanding the variability of organisms and their responses to env i ronmental conditions. They provide valuable insights into ecological processes, as changes in body size and proportions often reflect adaptations to habitat characteristics and environmental pressures. During the years 2020 – 2023, we co n ducted research in 6 study areas falling under three habitats ( willow -poplar floodplain forest, poplar nursery, alluvial meadow) in the Danube Floodplains Protected Landscape Area (CHKO Dunajské Luhy ). This study examined the variability of nine morphometric characters (body length, body height, body width, head lenght , head width, pronotum lenght , pronotum width, right elytron , left elytron ) in the ground beetle Carabus granulatus Linnaeus, 1758 in relation to management, habitat, and sex. A total of 992 individuals (440 males, 552 females) were analyzed. Individuals from the managed sites showed slightly higher values in body width, head width, and pronotum width, whereas those from u n managed habitats exhibited greater body length, body height, and elytral length. Generalized linear models revealed that habitat and sex were the main factors influencing morphometric variability, while management had a weaker effect. Habitat significantly affected multiple traits, with lower values observed in more disturbed environments. Sex had a strong influence on most traits, with males consistently smaller than females, confirming sexual dimorphism. A Mult i layer Perceptron model demonstrated high classification performance (accuracy = 0.925), indicating that morphometric traits effectively distinguish individuals based on management. Overall, morphometric variability is primarily driven by habitat conditions and sex, with management playing a secondary role. Morphometric traits can be applied in enviro n mental monitoring as sensitive indicators of habitat quality and ecological changes. Their analysis enables more effe c tive assessment of land management practices and supports conservation strategies aimed at maintaining biodiversity and ecosystem stability.

From species traits to community gradients: Validation of Borhidi-like naturalness indicators in steppe vegetation

H. Tutova — 2026-04-26

Naturalness is a key ecological concept reflecting the extent to which plant communities correspond to their natural state under minimal anthropogenic influence. The Borhidi system provides a widely used framework for assigning species-level naturalness indicator values based on ecological behaviour; however, its application at the community level remains insufficiently validated. This study aimed to assess whether species-level naturalness indicators correspond to empirically observed patterns in vegetation and to develop a reproducible, community-calibrated naturalness scale for the steppe ecosystems of Ukraine. The analysis was based on a large vegetation dataset encompassing a broad gradient of naturalness and anthropogenic transformation, including natural, semi-natural, urban, agricultural, and technogenic habitats. A modified Borhidi-like classification was developed by integrating functional traits, ecological niche breadth, synanthropic status, and conservation information into a unified species-level index. Relationships between species traits, environmental gradients, and community composition were analysed using RLQ ordination. A curvilinear naturalness trajectory was derived from the ordination space and used to construct a continuous community-level indicator scale. The results show that species-level naturalness categories exhibit non-random patterns in community composition and are related to environmental gradients. They do not form a single linear gradient; instead, naturalness emerges as a multidimensional property structured by multiple ecological processes. The most consistent dimension corresponds to the contrast between stress-tolerant and ruderal strategies, which approximates a functional naturalness gradient. Functional traits, particularly stress tolerance and ecological niche breadth, significantly contribute to explaining variation along this gradient, although they do not fully determine species positions. The proposed approach allows the transformation of discrete Borhidi-like categories into a continuous, empirically calibrated naturalness scale. This scale reflects community-level organisation and provides a more robust basis for ecological interpretation than purely expert-based classifications. At the same time, the results highlight that the transfer of species-level indicators to the community level is only partially valid, as many categories exhibit context-dependent behaviour. The study demonstrates that naturalness can be operationalised as a continuous, community-derived indicator when species’ ecological properties are integrated with empirical vegetation data. The proposed framework offers a reproducible tool for assessing vegetation naturalness and analysing ecosystem transformation across heterogeneous landscapes.