OA OpenAgrar repository
Not a member yet
164310 research outputs found
Sort by
Development of a Modular 3D‐Printed Pollen Trap for Bumble Bee Monitoring
Full text linkAccurate pollen collection is essential for understanding bumble bee foraging dynamics, assessing environmental risks and monitoring colony health. Effective monitoring systems provide critical insights into pesticide exposure, floral resource availability and pollinator health. This study compares the efficiency of two pollen trap designs, the newly developed JKI trap and the USDA 3D-printed trap, in collecting pollen from Bombus terrestris colonies. Field tests using traps with two entrance diameters (6.5 and 7.2 mm) showed that the JKI trap collected significantly more pollen than the USDA trap, with the statistical model predicting approximately 24 times higher yields (p < 0.001); no significant effect of entrance diameter on pollen yield was observed. The JKI trap's effective performance, coupled with its design flexibility and potential for adaptation across different Bombus species and pollinators, makes it a valuable tool for long-term ecological monitoring, floral resource assessments, and pesticide risk studies
S2CIE: semantic, syntactic, and context-based information extraction for AOP development
No full textAdverse Outcome Pathways (AOPs) offer a structured framework for linking mechanistic events occurring at different levels of biological organization supporting chemical hazard identification and regulatory decision-making. However, the development of AOPs remains a labour-intensive process, requiring extensive literature search, and expert review to identify relevant mechanistic events and supporting evidence. Existing text mining tools are often limited in scope, operate on small datasets, and lack interactive, context-aware capabilities leading to missed relevant studies and inclusion of unrelated information.To address these challenges, we propose S2CIE (Semantic & Syntactic Context-driven Information Extraction), an interactive, real-time information extraction platform. S2CIE enables extraction of mechanistic information from bio-chemical domain. It annotates literatures at the token level with grammatical, entity-type and dependency graph annotations. These annotation helps to define custom extraction rules over surface the tokens and syntactic graphs for information extraction, followed by filtering and ranking based on biological context. The system supports real-time querying over large databases (currently PubMed abstracts). S2CIE also integrates entity recognition, biological concept scoring, and visual exploration tools to aid interpretation.To demonstrate its utility for AOPs, we present a case study focused on identifying chemical stressors linked to steatosis and cholestasis. The system identified 102 chemicals with direct evidence for steatosis/choleostasis with 98.2% precision. We also extract evidence of interactions between identified chemicals and PPAR, a key molecular initiating event. Furthermore, we demonstrate large-scale information extraction capability with posttranslational modification (PTM) statements. The S2CIE supports integration with proprietary data and fosters community-driven extension, ultimately contributing to more efficient and scienceinformed chemical risk assessment.S2CIE is available as open source through an API and as web application at
https://dev.s2cie.insilicohub.org
Erweiterte Sektorenverantwortung für eine nachhaltige Kreislaufwirtschaft im Tourismussektor
No full textFIT for purpose? Sampling Bees in Flowering Canopies with Flight Interception Traps (FITs)
No full textDepending on species and season, tree canopies can provide floral resources in abundance. However, researchers often do not consider these resources when sampling bee communities, because they are difficult to access. While observer-based methods, such as hand-netting along transects, are often used for collecting bees from near-ground resources like wildflower strips, flight interception traps (FITs), such as aerial Malaise traps (AMTs) and window interception traps (WITs), can be used for collecting bees from elevated resources like flowering canopies. We assessed the suitability of WITs and AMTs for sampling bees from flowering canopies. We sampled canopies of four different tree taxa (Salix, Malus, Robinia, Tilia) in Braunschweig, Germany, between March and June 2024. In total, we collected 395 bee individuals, including 247 honeybees. Sampled communities comprised ten genera and showed marked differences in the ratio of honeybee and wild bee individuals. In Salix trees, wild bees were more abundant than honeybees, whereas in Tilia trees and especially Robinia trees, honeybees outnumbered wild bees. The collectors above the interception surface of both WITs and AMTs collected no bees. This came as a surprise, because we expected bees to move upwards upon flight interception, as they do in ground-based Malaise traps. We conducted a systematic literature search in the Web of Science to compare our findings to previous studies using WITs and AMTs for sampling bees. To our knowledge, our study is the first to show that collectors above the interception surface are generally inefficient when sampling bees with WITs and AMTs. Our study provides methodological advice for future researchers seeking to sample pollinators in tree canopies
Recognizing stressed chicken signs: A comparison using the Happy Chicken Tool and the Stressed Chicken Scale
No full textThis study investigates the use of deep learning, specifically convolutional neural networks (CNNs), and transfer learning for detecting signs of discomfort in chickens through image analysis. We present a comprehensive framework that includes data preparation, model training, and evaluation using transfer learning with pre-trained CNN models such as EfficientNet and MobileNet. The methodology includes image extraction from video footage, followed by preprocessing, and augmentation to improve dataset diversity and robustness. Model performance was evaluated using cross-validation on the original dataset and validation on two separate datasets, with metrics such as accuracy, sensitivity, and specificity. Results of the CNNs were compared to human observers’ stress ratings on the same datasets (= images) of chickens using the Stressed Chicken Scale. We found that AI can detect discomfort in individual chickens in side-view images, comparable to humans. Our findings show that certain CNN models, in particular variants of EfficientNet, show high performance in identifying stress signs in chickens. These results highlight the potential of deep learning for automated animal welfare monitoring. To enhance model interpretability, we used a Grad-CAM, which provides valuable insights into the decisionmaking process of the models. We found that the AI “looks” at specific body parts of the chickens when making decisions. This research contributes to the development of innovative, non-invasive methods for monitoring chicken welfare, and may provide the foundation for a useful tool for early detection of stress and discomfort indicators in chickens at individual animal level
Divergent Target‐Site Substitutions at Pro197 Confer Variable Degrees of Resistance to Tribenuron‐Methyl and Florasulam in Tripleurospermum inodorum Populations Across Europe
No full textTripleurospermum inodorum (L.) Sch. Bip. is a widespread weed in cereal production systems across Europe and has evolved resistance to acetolactate synthase (ALS)-inhibiting herbicides in several Northern and Central European countries. This study identified and characterised resistance to the ALS-inhibiting herbicides tribenuron-methyl and florasulam in eight populations of T. inodorum from the Czech Republic, Germany, Norway and Sweden. The two Czech populations, with Pro-197-Gln + Pro-197-Ala substitutions in one population (CZ1) and a Pro-197-Thr substitution in the second population (CZ2), differed in their response to tribenuron-methyl: CZ1 showed low resistance (resistance factor, RF: 5.2), while CZ2 exhibited high resistance (RF: > 53). However, both showed similar and low resistance to florasulam (RF: 2.5 and 3.9, respectively). The two German populations also showed contrasting responses: one population, with a Pro-197-Leu substitution, exhibited low resistance to both ALS inhibitors (RF: 2.8 for tribenuron-methyl and 3.3 for florasulam), whereas the other population, with a Pro-197-Thr substitution, displayed high resistance to both herbicides (RF: > 53 and 12.9, respectively). Norwegian populations with a Pro-197-Tyr substitution and Swedish populations with Pro-197-Thr or Pro-197-Gln substitutions exhibited high resistance to tribenuron-methyl (RF: 15.2–> 53), but only low resistance to florasulam (RF: 2.5–4.8). Geographic patterns in substitution types were evident, with Nordic populations predominantly exhibiting polar substitutions and Central European populations showing a mix of polar and non-polar substitutions, suggesting divergent resistance evolution pathways. Notably, except for the Pro197Gln mutation, all other identified mutations have not been previously reported in T. inodorum. Overall, these results highlight the need for region-specific resistance management strategies
