Resilience attributes that were discussed in most case studies were “Infrastructure for innovation”, and “Production coupled with local and natural capital”. Resilience attributes related to diversity, policies or connection with actors outside the farming system were least discussed. In SE-Poultry and PL-Horticulture the “Functional diversity” and “Response diversity” was emphasized. In DE-Arable&Mixed, RO-Mixed and to a lesser extent in IT-Hazelnut, “Support rural life” relating to the embeddedness of the farming system in the rural society was discussed because of rural isolation and/or outmigration that is experienced . In ES-Sheep and IT-Hazelnut, the resilience attribute “Diverse policies” was discussed due to the pressure experienced from environmental regulations that reduce the competitive advantage because of higher production costs . Levels of most of the main function indicators and resilience attributes are currently perceived to be slightly decreasing. In the perceived moderately performing systems IT-Hazelnut, SE-Poultry and NL-Arable , overall moderately positive indicator developments were expected. In the perceived low performing systems ES-Sheep and PL-Horticulture , and also in UK-Arable, negative developments were expected. With reference to current performance and ongoing trends it is interesting to know between what levels the main system challenges, function indicators and resilience attributes need to stay in order to maintain the current system configuration. Critical thresholds were defined as levels beyond which performance of all other key system functions is expected to drop below acceptable levels. Although multiple types of critical thresholds can be distinguished, all types have in common that system change after exceeding them is large and that reversing that change is challenging and costly . To not over complicate the concept in a participatory setting, we therefore defined a critical threshold as a point beyond which large and permanent, system change is expected. This change can have a positive as well as a negative connotation. However, as challenges are the point of departure in this study, hydroponic nft system overall change has predominantly a negative connotation.

Workshop participants were asked to individually note down critical thresholds of the main system challenges, function indicators and resilience attributes. Participants were encouraged to provide quantitative assessments of critical thresholds. When asked for by participants, members of the research team could suggest units for expressing critical thresholds. Notes with the stakeholders’ assessment of critical thresholds were collected and posted on a wall and were left there for the remainder of the workshop. Notes were discussed in plenary sessions to explore possible critical thresholds and to reach consensus on critical thresholds. Stakeholders’ notes of enabling conditions that help avoiding the exceedance of critical thresholds, rather than estimations of values for critical thresholds, were included in the plenary discussions and are summarized in a separate paragraph in this paper. Closeness of challenges, function indicators and resilience attributes to critical thresholds was evaluated by the research team based on participants’ comments and literature, e.g. based on ongoing trends identified in the preparation phase before the workshop. The position relative to the threshold was considered to be either “not close”, “somewhat close” or “close” when it seemed respectively unlikely, somewhat likely or likely that the distance to critical thresholds would be trespassed in the coming ten years, based on knowledge on possible variation and/or trends. An indicator that is “close”, for instance, is likely to exceed a threshold within ten years, but exceedance can also happen after 30 years, which, however, is less likely. A fourth category of indicating the position relative to the threshold was “at or beyond”. Detailed argumentation about the evaluation of closeness to critical thresholds is provided in Supplementary Materials 2. After discussing critical thresholds, farming system performance was assessed in case critical thresholds of main challenges would be exceeded in the near future. For each identified challenge, sub-groups of a moderator and at least three participants were formed on a voluntary basis. In those subgroups, the impact of exceeding the critical threshold of a challenge on main indicators and resilience attributes was discussed.

A research team member functioned as moderator and used a poster to draw arrows between the challenges and main indicators and resilience attributes that were expected to be impacted. The strength of the expected impact was indicated by adding ++, +, -, –, representing a strong positive, moderate positive, moderate negative and strong negative expected impact. As the impacts of exceeding thresholds were determined for the current system, challenges and their impact were discussed in the context of other challenges that are already present in the system. In this paper, therefore, we present and consider the overall impact of exceeding challenge thresholds as the impact of simultaneous stresses that have a combined effect at system level . The possibility of interactions between critical thresholds of challenges, indicators and resilience attributes was discussed during the workshops. Based on this, and based on the information acquired in the previous step and from literature, research teams aimed to reveal interacting thresholds across domains and levels of integration that cause farming system dynamics. Interacting thresholds are thresholds that, when exceeded, lead to the exceedance of another threshold . Determining whether thresholds were interacting was based on qualitative argumentation by researchers using input from workshops. Detailed information on interacting thresholds per farming system is provided in Supplementary Materials 3.1 To be able to concisely compare results from 11 case studies, our focus in this paper is on reporting and discussing the perceived relative closeness to critical thresholds and their interactions. The actual thresholds as noted down and discussed by stakeholders during the workshop are often very case-specific. Moreover, the precise level of critical thresholds was in most cases challenging to assess as stakeholders differed in opinion, and used different metrics. The assessments of thresholds are therefore mainly used to illustrate the methodology and our findings. More than half of the identified challenges were perceived to be “close” or “at or beyond” critical thresholds . For extreme weather, closeness differed between farming systems: NL-Arable, ITHazelnut, PL-Horticulture, were perceived “somewhat close” to, DEArable&Mixed and BG-Arable seemed “close” to and RO-Mixed seems “at or beyond” the perceived critical thresholds. For the environmental challenge “pest & diseases”, NL-Arable, challenged by plant parasitic nematodes, and IT-Hazelnut, challenged by phytophatologies, were perceived to be “somewhat close” to critical thresholds.

For challenges in the social, economic and institutional domain, participants perceived more often that critical thresholds were reached than for the environmental domain. In ES-Sheep, participants indicated that for all challenges critical thresholds were reached, except for wildlife attacks . In DE-Arable&Mixed, the lack of infrastructure and low attractiveness of the area were perceived to be at or beyond a critical threshold. In SE-Poultry, the perceived mismatch between economic viability on the one hand and the high production standards and strict environmental regulations on the other hand made participants indicate that for both challenges critical thresholds were reached. Continuous change of laws and regulations was seen as a main challenge in NLArable, UK-Arable, PL-Horticulture as well as BG-Arable. Participants in these case studies, for instance, perceived a critical threshold in the case that certain crop protection products would be banned before replacements had become available. A policy implication here would be to study a reasonable time for phasing out/in of policies. In DE-Arable&Mixed, SE-Poultry and RO-Mixed, inadequate alignment of policies and regulations at national and EU level was mentioned: national production quality standards increase production costs, while abiding with EU trade regulations allows for cheaper imports from countries with lower production standards and constraints. Participants could define critical thresholds for most system function indicators ; for instance, critical thresholds for the yield per hectare, an indicator related to the function “Food production”, e.g. in BG-Arable, RO-Mixed and NL-Arable. Systems were perceived to be “close” to critical thresholds for “Food production” and “Economic viability” and “somewhat close” to those for “Natural resources” and “Attractiveness of the area”. In IT-Hazelnut, for instance, the threshold for “Gross margin” relating to the function “Economic viability” was assessed to be 5000 Euros per hectare, but was expected to differ from farm to farm. Based on current variability of markets and climate, it is likely that the value will someday drop below the indicated threshold, which makes that the system may be close to this critical threshold. For the seemingly low performing systems PL-Horticulture and ES-Sheep, some indicator levels were perceived to be at or beyond the threshold. In these systems, immediate action seems required, e.g. with regard to product prices and availability of labor in the area. Reaching critical thresholds for soil quality, an indicator representing “Natural Resources”, was a concern in UK-Arable and NL-Arable. In those systems,nft channel participants mentioned that continuous adaptation is needed to prevent further degradation. In NL-Arable, a participant from the regional water board indicated that in the long-term water availability would decline, thus the system would approach a threshold.

Most other participants took a more medium-term stance and therefore proximity to this threshold was considered somewhat close. Overall, there was rarely a disagreement between participants about threshold levels. In BE-Dairy, where a desk-study was performed, water quality and greenhouse gas emissions were perceived to be beyond acceptable levels set by European and regional policy makers. Farmers in BE-Dairy are likely to disagree with these externally determined thresholds. In SE-Poultry, DEArable&Mixed, ES-Sheep and NL-Arable, participants indicated that critical thresholds for economic viability differ from farm to farm. Hence, exceeding critical thresholds in these case studies may foremost imply the disappearance of economically less competitive farms from the farming system, rather than an immediate decline of the entire farming system performance. For resilience attributes, relatively fewer critical thresholds were defined than for function indicators . Thresholds of resilience attributes were mostly qualitatively determined. For instance, in DE-Arable& Mixed “Supports rural life” was assessed to be on the lower end of a 1 to 5 scale where 1 implied very low and 5 implied a very high support. Participants indicated that a further decline in support would imply crossing a critical threshold. Overall, when defined, resilience attributes seem less close to critical thresholds than function indicators. From a methodological point of view, resilience attributes might be harder to grasp, and therefore more difficult to define and also perceived to be less close to critical thresholds than function indicators. From a theoretical point of view, the distance to critical thresholds could suggest that under the current challenges, resilience capacities are still sufficient to, for instance, start an adaptation or transformation process that steers away from critical thresholds of system challenges and indicators. However, the presence of some attributes e.g. “Reasonably profitable”, when discussed and when a critical threshold was defined, was perceived to be close to a critical threshold, similar to the function “Economic viability” in most case studies . For the resilience attribute “Diverse policies”, i.e. policies that equally support robustness, adaptability and transformability , the systems in ES-Sheep and IT-Hazelnut were perceived to be at or beyond a critical threshold. In IT-Hazelnut the system was perceived to be close to a critical threshold regarding “Infrastructure for innovation”. In IT-Hazelnut, current innovation levels were perceived already high, but would benefit from more to ensure further adaptation and improvement. For most other resilience attributes the system was perceived to be close to critical thresholds. While noting down and discussing critical thresholds, participants often mentioned enabling conditions that help avoiding the exceedance of critical thresholds, rather than precise values for critical thresholds. Enabling conditions can be seen as general notions of how system specific problems can be solved for the current system. Enabling conditions in the agronomic domain were mentioned only in BG-Arable, NL-Arable and ES-Sheep; e.g. improving productivity levels and availability of geo-localization technologies . Enabling conditions in the economic domain were e.g. creating access to new markets , environmental payments and improving input/output price ratios . Enabling conditions in the environmental domain were e.g. low occurrence of extreme weather events , improved soil quality and ecological and resource management regulations . Specifically in UK-Arable, emphasis was put on enabling conditions in the environmental domain. Enabling conditions in the institutional domain included good governance practices of authorities and access to knowledge, finance and/or land .