Martin Marinov

Agroecosystems provide a wide range of ecosystem services, including food and raw materials, while also supporting biodiversity. However, increasing agricultural intensification has reduced the biodiversity value of many agricultural landscapes. Some perennial crops, such as Damask rose ( Rosa damascena Mill.), are of particular interest because of their long-term management and sensitivity to environmental change, and provide a suitable system for assessing the effects of human activities and climate change on biodiversity. Traditionally cultivated in Bulgaria's Valley of Roses, rose plantations have a structure similar to that of shrubland ecosystems, and together with surrounding grasslands, woodlands, and other crops, create heterogeneous landscapes that may support diverse organisms. In this context, birds are widely recognised as key bioindicators of environmental change. Recent studies have shown that farmland bird populations have undergone long-term declines across Europe, including Bulgaria. Despite the ongoing national monitoring initiatives, there are no data on bird community composition and diversity in rose plantations. Addressing this knowledge gap is essential for understanding the ecological role of rose plantations and supporting biodiversity conservation efforts. The aim of our study was to assess bird species diversity, abundance, and community composition in rose plantations in Bulgaria's Valley of Roses. Specifically, we studied how landscape compositional and structural heterogeneity influence bird diversity and communities. We also identified the landscape elements most strongly associated with species richness and indicator bird species characterising these agroecosystems. Birds were surveyed using the point-count method at 100 sampling plots across 20 sites in the municipalities of Pavel Banya, Kazanlak, and Maglizh during the breeding season of 2023 and 2024 (with two surveys conducted each season). Vegetation metrics were collected using LiDAR technology mounted on an unmanned aerial vehicle (UAV). Maximum height, mean NDVI, and the total area of twelve vegetation categories were included in the statistical analyses. K-means clustering analysis identified three groups of sampling plots based on bird species composition and abundance, corresponding to open, woodland, and shrub-associated bird communities. Environmental fitting (envfit) revealed that the variables most strongly associated with the clusters identified in the ordination were the maximum heights of woodlands, lavender fields, first-year roses, and grasslands; woodland NDVI; the total areas of annual fields, hedgerows, lavender fields, rose plantations, first-year roses, nut trees, and grasslands; as well as distance to buildings. In addition, PERMANOVA identified the total areas of annual crops, rose plantations, first-year roses, hedgerows, and grasslands, together with woodland maximum height, as the most important factors explaining differences in community composition. The indicator species analysis identified the characteristic bird species for each cluster. Cluster 1, representing open habitats, was characterised by Eurasian Skylark, Western Yellow Wagtail, and Crested Lark; Cluster 2, representing woodland-associated bird communities dominated by Common Blackbird, Great Tit, Eurasian Chaffinch and Eurasian Golden Oriole, while in cluster 3, which primarily includes birds with a preference for shrubland habitats, including rose plantations, the indicator species were the Common Linnet and Ortolan Bunting. Our findings indicate that maintaining a heterogeneous landscape mosaic around rose plantations can contribute to supporting diverse bird communities.

Background When working with stakeholders, context and presentation are of utmost importance since users seldom have the knowledge and motivation to comprehend and critically analyse data. This is especially so for ecologists whose knowledge is not immediately related to many of the stakeholders' daily life or economic activities and hence is often beyond - or in some cases in direct conflict with - their everyday activities. Reliance on the scientist's interpretation based on scientific analysis and its subsequent visualisation in the form of graphs, tables etc. has a number of limitations: Data is only one form of knowledge for characterising socio-ecological systems. Non-data knowledge includes both important persistent descriptive characteristics of an ecosystem ("Srebarna is a hyper-eutrophic lake"), and shared socio-cultural context implicitly acknowledged by both sides ("Bai Ganyo is an exporter of essential oil from Rosa damascena in the classic 19th century satiric fiction by Aleko Konstantinov, a part of Bulgaria's literary heritage.") Discrepancies in implicit knowledge: common scientific knowledge not explicitly mentioned in research papers whose omission deprives non-specialist readers of important context. This can cause numerous biases in verbal and written communication (Kellerman 2007, Ahlert et al. 2016). Specialisation limits the knowledge of individual scientists so their work does not contain all information about the model ecosystem that the stakeholders need for informed decision-making; larger scientific teams, on the other hand, require careful wording for clarity and brevity when communicating their finding about complex ecosystem processes. Data is only one form of knowledge for characterising socio-ecological systems. Non-data knowledge includes both important persistent descriptive characteristics of an ecosystem ("Srebarna is a hyper-eutrophic lake"), and shared socio-cultural context implicitly acknowledged by both sides ("Bai Ganyo is an exporter of essential oil from Rosa damascena in the classic 19th century satiric fiction by Aleko Konstantinov, a part of Bulgaria's literary heritage.") Discrepancies in implicit knowledge: common scientific knowledge not explicitly mentioned in research papers whose omission deprives non-specialist readers of important context. This can cause numerous biases in verbal and written communication (Kellerman 2007, Ahlert et al. 2016). Specialisation limits the knowledge of individual scientists so their work does not contain all information about the model ecosystem that the stakeholders need for informed decision-making; larger scientific teams, on the other hand, require careful wording for clarity and brevity when communicating their finding about complex ecosystem processes. Ontology-based approach to communication In this work, we present the use of machine-readable formalised ontologies that contain known and inferred facts about relations within the ecosystem in the form Subject-Predicate-Object. This form of presenting relations is significantly more expressive than the fuzzy knowledge graphs used in socio-ecological research. For example, "Pelecanus crispus had X nesting pairs in year Y" is a relationship that cannot be denoted as positive or negative in the typical fuzzy cognitive network out of the context of a dataset with Pelecanus population trends. Furthermore, ontologies allow for different relations between the same objects such as food competition between two species and expressing their taxonomic relation, annotating hierarchies such as the taxonomic relationships or the elements of a standard observation, authorship and sources, etc. in a machine readable and unambiguous manner. Our ontology production workflow uses an AI-in-the-loop extraction of information from publications, its linking to historical data and standard observation datasets (including the ones published in the eLTER data repository as open datasets), and reasoning across these sources of heterogeneous information to create a flexible representation of the ecosystem complexities. We use the Ontotext graph tools dataset - GraphDB and OntoRefine. A sufficiently comprehensive ontology can contain tens of thousands of factual and inferred relationships and by itself is not a suitable tool for stakeholder communication. However, querying the complex ontology to retrieve a subset of the knowledge therein (referred to as ontoset), allows for extracting the information pertinent to a given communication topic and its reuse for simplified but factual communication with one or more target groups. We present two cases of stakeholder communication basing on ontologies about two model ecosystems - the lake ecosystem of biosphere reserve Srebarna and the agroecosystems of the Valley of Roses. For both model ecosystems, we create ontologies from heterogeneous sources. We then proceed to extract thematic ontosets for the target audience and use the ontoset for specific communication purposes (Fig. 1). In the case of lake Srebarna, we present a set of games developed for the Science day (September, 2026) basing on a subset of the ontology about Srebarna that includes trophic and habitat interactions, historic and fun facts about Srebarna from Michev and Iliev (1982) and the experience of the Srebarna research team, as well as information about the method and significance of standard observations performed in the site. The games are intended to support system thinking and engage the most interested citizen scientists (school and university students and interested adults - teachers, parents). They can be given opportunity for job shadowing the researchers during actual fieldwork and performing simple observations. In the case of the future socio-ecological platform Valley of Roses, we present the transition from an earlier knowledge graph to an ontology containing information about the characteristic crops in the region - essential oils from Rosa damascena and Lavandula angustifolia. The ontology links the ecosystem data collected in an ongoing seed funding project to create the platform, existing data from research and other projects, as well as ecosystem information extracted from textual publications, with social and economic data derived from interviews and a survey. We present the way businesses - farmers, distilleries, traders, local tourism and attractions, can interact with rose oil ontoset, including the yield parameters for rose oil, the influence of specific pests, and links to national and international legislation, market and customer requirements and the rose as an iconic tourist attraction. The presented work demonstates a workflow for processing and semantically linking heterogeneous information from transdisciplinary research teams with AI-in-the-loop. We use a structured approach to present facts free of hallucination and context deviations - the main shortcomings of current AI models. This approach supports the formalisation and reproducibility of socio-ecologic research and stakeholder involvement using the new tools for extracting actionable insights from big data that became available in the last years.

Long-term population studies are of great importance in ecology and conservation. We present a comprehensive analysis of long-term data (1955–2024) on numbers of nests and breeding success of the colony of the Dalmatian pelican (Pelecanus crispus) in the Srebarna Nature Reserve, Bulgaria. The variation in breeding success was estimated by the number of chicks successfully raised per occupied nest, indicating a significant overall increase. The number of nests ranged from 7 to 150 over the years and had a noticeable downward trend in the last five years. We evaluated various conservation measures implemented in the reserve, such as the construction of raised wooden platforms and the establishment of protective fences, highlighting their effectiveness in maintaining the breeding colony. Our findings indicated that although breeding success had improved over the decades, recent declines in nest numbers were linked to external factors, including human disturbance, predation by invasive species and the availability of alternative nesting sites in other protected areas. We call for continuation of the conservation efforts to enhance the long-term viability of the Dalmatian pelican population in the reserve.

Wader populations have been declining globally due to widespread anthropogenic habitat degradation and loss. In this context, food quality and availability at migration stop-over sites is crucial for wader migration success. We studied the diet composition and preferences of the Grey plover (Pluvialis squatarola) at a stop-over site on the Black Sea-Mediterranean Flyway - Pomorie Lake, SW Black Sea coast, Bulgaria - through DNA metabarcoding of faeces. The study aimed to determine the bird's feeding strategy, coastal habitat use and possible seasonal variations as a result of local prey availability. The dataset includes new data on Grey plover (Pluvialis squatarola) prey items on the south-western Black Sea coast, Bulgaria. The diet composition data provides valuable insights on coastal biodiversity in the area as "sampled" by this generalist invertivore and contributes to the knowledge of Grey plover prey choice and habitat use in a little-studied microtidal marine region.

Food quality and availability at migration stopover sites is crucial for wader migration success, particularly nowadays with their populations in decline due to widespread anthropogenic habitat degradation and loss. We studied the diet composition and preferences of the Grey plover (Pluvialis squatarola) at a stopover site on the Black Sea-Meditteranean Flyway - Pomorie Lake, SW Black Sea coast, Bulgaria, - through DNA metabarcoding of faeces. The study aimed to determine the bird's feeding strategy, coastal habitat use, and possible seasonal variations as a result of local prey availability. The dataset includes new data on Grey plover (Pluvialis squatarola) prey items on the south-western Black Sea coast, Bulgaria. The diet composition data provides valuable insights on coastal biodiversity in the area as "sampled" by this generalist invertivore, and contributes to the knowledge of Grey plover prey choice and habitat use in a little-studied microtidal marine region.

We studied the diet composition and feeding preferences of the Grey Plover (Pluvialis squatarola) at its migration stopover at Pomorie Lake on the South-Western Black Sea coast, Bulgaria, through DNA metabarcoding of faeces. Faecal samples were collected in autumn 2020 and spring 2021, preserved in 95% ethanol, and stored at -20ºC until DNA extraction and metabarcoding. Two molecular markers were initially targeted - ITS2 for plant identification, and CO1 for animal identification; however, the subsequent analyses focused on the dominant animal prey identified by CO1. Concurrently, coastal habitats where the birds were observed to feed, such as the surf zone and the sandy supralittoral shore, were also sampled to determine prey availability. The Grey Plover had a broad diet spectrum, with 332 prey taxa in total identified by metabarcoding, belonging to 7 phyla, 14 classes, 24 orders, 78 families and 119 genera. The birds consumed both terrestrial and marine taxa, suggesting that multiple coastal habitats are used for foraging. Terrestrial arthropods, particularly insects, predominated in the diet, while mytilid bivalves were the most represented marine taxon. The results suggest that at the studied migration stopover site at the SW Black Sea, the Grey Plover was a generalist feeder, which agrees with previous studies from different areas. Most birds fed on insects (Diptera, Lepidoptera, Hymenoptera); a few individuals also specialised on marine molluscs, arachnids and polychaetes. The most abundant potential prey in the environ­ment were not the most frequently consumed, possibly because the sampling methods were not suitable for the full spectrum of available prey (particularly the flying insects). We found limited seasonal variation in Grey Plover diet composition and preferences, with the proportions of some prey taxa differing slightly between seasons, probably due to seasonal variations in their abundance and activity in coastal habitats. Our results contribute to the knowledge of Grey Plover diet, prey choice and habitat use in a little-studied region on the Black Sea-Mediterranean Flyway. The new information could contribute towards designing effective conservation measures to preserve habitat quality at stopover sites for this wader and others with similar ecology, ensuring migration success at a time when migratory bird populations are increasingly threatened by habitat loss and degradation. The results are presented and analysed in detail in Vassileva et al. (2024). The full CO1 dataset is available in BOLD (https://doi.org/10.5883/DS-PLUVS).

This study aimed to assess the present distribution of red wood ant species and changes in their nest density over a period of eight years in the area of Rila Mt, Bulgaria. One of the sampling areas was the Parangalitza Biosphere Reserve, the nation’s second oldest protected area. The other two sites are Rila Monastery region and Samokov region. Field monitoring via transect sampling was conducted in 2014 and 2022. We marked 35 nests along 29 transects in 2014 and 76 nests along 27 transects in 2022 of the red wood ants Formica rufa Linnaeus, 1761, F. lugubris Zetterstedt, 1838, F. pratensis Retzius, 1783, and F. exsecta Nylander, 1846, the latter of which is not a member of the rufa group. The most abundant species was F. rufa, followed by F. lugubris, F. pratensis, and F. exsecta. The nest density of F. rufa and F. pratensis decreased in some of the revisited transects after 8 years, which can be explained by increased shading of their habitats. Monitoring is recommended at least every 5 years to detect trends in red wood ant populations and evaluate the effects of protection measures.

The only National Inventory of red wood ants in Bulgaria was carried out about 50 years ago (1970–1973). Formica rufa Linnaeus, 1761, F. pratensis Retzius, 1783, F. lugubris Zetterstedt, 1838 and F. polyctena (as F. polyctena x rufa hybrid) were found in a current monitoring programme. This study presents data on their current distribution and nest density, and provides more details about the habitat requirements for conservation purposes. Field studies were carried out by the transect method along the main mountainous areas in Bulgaria. We found 256 nests of red wood ants along 172 transects. The most abundant species was F. lugubris, followed by F. rufa and F. pratensis. Among the environmental variables, the elevation, exposure, ecological groups of plants, stone cover, grass cover, canopy cover and forest age appeared as significantly related to the presence and nest density of red wood ants.

Identifying patterns in roots spatial distribution and dynamics, and quantifying the root stocks, annual production and turnover rates at species level is essential for understanding plant ecological responses to local environmental factors and climate change. We studied selected root traits in four different stands, two European beech (Fagus sylvatica L.) forests and two Douglas fir (Pseudotsuga menziezii Mirb. Franco) plantations. Root system vertical distribution and dynamics were studied using sequential coring method and characterised into three root diameter size classes (0–2, 2–5 and 5–10 mm) sampled at three different soil depths (0–15, 15–30, 30–45 cm). Root annual production and turnover rates were analysed and quantified using Decision Matrix and Maximum-Minimum estimation approaches. The overall root mass (−1 and 0.76 yr⁻¹ based on mean and maximum biomass data, respectively. They were similar to the averaged values suggested for Central and Northern European forests but higher compared to those reported from Southern Europe.

Immune response to pathogens such as avian malaria (Plasmodium spp.) and related haemosporidian parasites (Haemoproteus spp. and Leucocytozoon spp.) can lead to increases in energy demands of birds. Migrations are energetically costly for birds and require resources that could be otherwise invested in immune defence against pathogens. Here, we examined the prevalence and intensity of haemosporidian parasites (Apicomplexa, Haemosporida) in rosy starlings (Pastor roseus), an irruptive migrant on the Balkans, during their pre-migratory period. We further evaluated if these infections affected the stress levels of birds brought into overnight captivity, using leukocyte profiles (ratios of heterophils to lymphocytes, or H/L ratios). Haemosporidian parasites were detected based on microscopic examination, PCR and sequencing. We found high prevalence of haemosporidian parasites (83.9%) in rosy starlings (n=31), with predominant infections of Haemoproteus pastoris (cytochrome b lineage LAMPUR01) (80.1%). All birds had high heterophil/lymphocyte ratios (mean = 3.86), indicating the transition to captivity was stressful. However, there was no added stress from blood parasite infection during captivity, since infected birds did not have significantly higher heterophil/lymphocyte ratios than uninfected birds, despite all birds facing the same stress conditions. There was a negative correlation between intensity of H. pastoris and monocyte proportions, which may indicate an association with infection, or an artefact of the stress-induced shift in leukocyte profile following captivity. Our findings suggest that blood parasites have minimal additional impact on rosy starlings during an acute stress scenario (overnight captivity), arguing that they carry little cost to the energetically challenging migratory flight.