This perspective paper draws on insights from a 2024 symposium entitled ‘Exploring methods for researching shifts in knowledge production for agroecology transition’. The symposium critically examined emerging conceptual and methodological challenges arising from combining agroecology with living labs and research infrastructures as key instruments promoted within EU policy to strengthen Agricultural Knowledge and Innovation Systems (AKIS). Through presentations, group discussions, and iterative reflections, we identified four key tensions: structural constraints limiting farmers’ agency within living lab approaches, the problematic nature of AKIS as supposedly neutral frameworks, the oversimplification of transition frameworks as linear rather than overlapping categories, and risks of definitional dilution or cooptation. We then demonstrate that these tensions are not unique to agroecology, bridging the concepts and methods within agroecology research with those used in other fields of sustainable food system transition research, such as transdisciplinary research and sustainable transitions. This conceptual mapping of shared tensions reveals opportunities for mutual learning. Bridging these fields would help create clarity at the conceptual and methodological levels, ultimately strengthening the theoretical foundations and enabling more nuanced approaches to food system transition research.
Introduction. The use of herbicides has been called into question due to their impact on health and the environment. Physical weeding solutions are an alternative which are still rarely used in Belgium. Agriculture and the regional context in which it develops has an impact on farmers' ability to integrate physical weeding tools into their crop management practices. Literature. External or internal factors related to farms can help explain the limited use of alternatives to chemical weeding. Soil characteristics, particularly clay content, surface stone load and the presence of slopes, all have an impact on the feasibility of using mechanical weeding tools that interact with the soil. Climate is also a limiting factor, as it reduces the time windows available for successful weeding operations. Intrinsic characteristics of farms, such as crop rotation, farm size, labour structure, and profitability, also help explain the potential for introducing physical weeding in Wallonia. Conclusion. Wallonia presents opportunities for the deployment of physical weed control solutions. Its potential expressed in terms of suitable agricultural area is estimated at 42% of the usable agricultural land, mainly concentrated in the Limoneuse and Sablo-limoneuse agricultural regions. The climate is an obstacle to the widespread use of physical weed control in Wallonia. Regardless of the agricultural region, it can be observed that the number of available days for successful physical weeding is zero in some years. A more specific analysis of the structure of specialized field-crop-farms shows that their diversified crop rotation is favourable for the introduction of physical weed control and they generally have sufficient resources, both human and financial, to support the integration of these new techniques and tools into crop management practices.
The conversion of agro-industrial co-products and unsold organic plant-based residues into black soldier fly (BSF; Hermetia illucens (L. 1758)) proteins was assessed for use in organic post-weaning piglet diets in Belgium. A total of 72 crossbred female piglets (Landrace × Pietrain) were enrolled in a 5-week feeding trial. Experimental diets consisted of a common energy core (81.2% of the feed) and a protein core (18.8%) composed of organic soybean meal, pea meal, and potato protein, partially replaced by defatted BSF meal at inclusion levels of 15%, 25%, and 35%. All diets were formulated to be isoenergetic and isonitrogenous, with standardized ileal digestibility values for lysine, methionine, threonine, and tryptophan held constant. Incorporating 15% defatted BSF meal can substitute conventional organic protein sources without compromising growth performance in post-weaning piglets. However, economic modelling based on a cumulative feed conversion ratio expressed on a dry matter (DM) basis showed that break-even prices for organic BSF meal remained well below the price of the control protein nucleus (€1039·t−1, excluding VAT), indicating that economic parity could not be achieved at typical market prices under the observed feed efficiency.
The introduction of terrestrial invertebrate processed animal proteins (PAPs) from insects in animal feed has introduced challenges for official controls in the European Union. Currently, their detection relies only on light microscopy and is entirely dependent on the expertise of trained operators. Recent studies have highlighted two primary problems regarding insect PAP detection: specificity issues due to the misidentification of non-insect structures as insect derived ones, and the difficulty in categorizing a wide morphological diversity of particles resulting from the grinding of whole insect larvae. To address these limitations, a proof of concept was developed involving automated microscopic image analysis and classification by deep learning. A classification pipeline integrating a YOLO object detection model with an improved ConvNeXt architecture coupled with transformer blocks was tested. After supervised learning, training and validation of the model, obtained results allowed to successfully distinguish insect particles from other non-insect materials, and to discriminate PAPs particles derived from Tenebrio molitor and Hermetia illucens achieving average Matthews Correlation Coefficient of 0.89 and accuracies exceeding 95% for both tasks. The robust performance and the rapid processing speed of the model enabled to couple it to a digital microscope for real-time automated particle classification, including confidence scores for individual particle predictions. This artificial intelligence model for microscopic identification of complex structures opens new perspectives for official controls achieving performance scores never reached before. It provides to microscopists an optimisation of image analysis and a valuable decision-support tool in complex feed matrices where human expertise is frequently challenged.
Metabarcoding is a powerful tool for biodiversity comparisons, where standard-size DNA barcodes (> 500 bases) offer better taxonomic resolution than shorter ones. Still, the choice of sequencing platforms and bioinformatics pipelines may strongly affect inferred diversity due to various technical biases. We assessed the relative performance of Illumina MiSeq i100 (2 × 500 paired-end), PacBio Revio and Oxford Nanopore MinION sequencing and bioinformatics pipelines, using full-length ITS amplicon sequencing datasets from a 103-species mock community and 45 composite soil samples. Despite numerous low-quality reads, PacBio yielded the lowest overall error rate and highest number of taxa. Illumina revealed the highest proportion of chimeric and index-switched reads, along with a strong bias towards shorter amplicons. MinION data analysed using PRONAME and Minovar-a bioinformatics pipeline presented here-had the largest proportion of low-quality data, and rare taxa were lost during data filtering and read polishing steps. Although Minovar enabled amplicon sequence variant (ASV) level precision for common taxa, we recommend clustering ASVs into OTUs. For PacBio, standard filtering approaches outperformed the ASV approach because they retained rare taxa. For Illumina, a stringent ASV approach or removal of rare OTUs would limit artefacts. Across all platforms, excess PCR cycles promoted chimeric and low-quality reads and lost quantitativity in biodiversity assessments. With moderate differences in effect sizes, all analytical approaches supported the conclusion that sampling design determines how we see soil biodiversity responses to land use. For biodiversity surveys based on the full-length ITS metabarcoding, we recommend using PacBio sequencing with standard, non-ASV pipelines.
