Enabling conditions in the social domain were e.g. related to rural demographics and/or availability of labor and more horizontal and vertical cooperation and social self-organization . Specifically, in BGArable and RO-Mixed emphasis was put on enabling conditions in the institutional and social domain. In all case studies, interacting thresholds across level and/or domain were observed . More details on the interacting thresholds are presented in the Supplementary Materials 3. Common interactions between critical thresholds occur between field environmental and field-economic, from field-economic to farm economic, from farm-economic to farm-social, from farm-social to farming system-social, and from farming system-social to farm-social . Generally, an environmental issue at field level, for instance, decreasing soil quality , pest diseases , wildlife attacks , or drought is so much of a shock or stress that it leads to yields that are too low to sustain an adequate level of farm income . In a majority of the farming systems, high input prices and decreasing output prices and sales further diminish the farm income. Too low incomes at farm level were in all case studies resulting in reduced attractiveness of farming, farmers quitting or the lack of finding a successor for the farm. In UK-Arable, also reduced farmer happiness due to lack of recognition was mentioned as a reason for quitting a farm. Farmers quitting their farm without having a successor was in multiple farming systems also considered to contribute to a smaller rural population at farming system level . Interestingly,hydroponic nft although socially oriented function indicators and resilience attributes were less often formally included in the discussions, they eventually appeared when explaining how challenges impact the farming system. Having less farms in the farming system was also associated with a lower maintenance of natural resources and a less attractive countryside.

Interactions with critical thresholds in the environmental domain at farm and farming system level were mentioned in a few other case studies. In NL-Arable, at farm level in the environmental domain a narrow rotation in which starch potato is grown every second year was expected to lead to increased pressure of plant parasitic nematodes . In UK-Arable, low income at farm level was expected to lead to declining soil health at field level . In IT-Hazelnut and SE-Poultry, environmental regulations were expected to improve the maintenance of natural resources at farming system level, but also to push farm income levels below a threshold through increased costs . Overall we observed that environmental thresholds certainly feature, but differ in the level at which they play a role and in what direction they evolve. In farming systems for which access to land is an issue , quitting of farmers may also be an opportunity, provided land becomes available on the market for sale or to be leased. In ES-Sheep, quitting of farmers was experienced as a serious issue. In IT-Hazelnut, the retention of young people on the farms was specifically mentioned as something that could support the rural life and vice versa . Both low economic viability at farm level and low attractiveness of farming and a smaller rural population were considered to reduce the access to labor at farm level in BG-Arable, SE-Poultry, PLHorticulture, DE-Arable&Mixed, RO-Mixed, and ES-Sheep. Access to labor in BG-Arable, PL-Horticulture and RO-Mixed was important for the continuation of activities on farms, as lack of labor was expected to push yields below acceptable levels . In BG-Arable lack of labor could be overcome by implementing new technologies, but this would require a labor force with higher levels of education and qualification which is even harder to find. Lack of labor was also expected to push production costs beyond critical thresholds in SE-Poultry and RO-Mixed. Hence, in multiple systems, low economic viability, attractiveness of farming, rural depopulation and low level of services at farming system level, and low access to labor seem to be part of a vicious cycle. Following from Fig. 1, it can be made plausible that after exceeding critical thresholds of challenges, a decline in performance of system’s main function indicators and resilience attributes was expected by workshop participants in most case studies . Across farming systems, the functions “Food production”, “Economic viability”, and the “Natural resources” were in most cases expected to decline moderately or strongly .In ES-Sheep, ongoing decline of function performance was expected to be aggravated.

When discussed in case studies, “Biodiversity & habitat” and “Animal health & welfare” were on average expected to be less impacted compared to other functions. When exceeding critical thresholds of challenges, also a decline in resilience attributes was expected in most case studies, mainly because of a decline in profitability, production being less coupled with local and natural capital, a declining support of rural life and lower levels of self organization . By contrast, participants in BG-Arable and SE-Poultry generally expected improvements in resilience attributes after critical thresholds are exceeded . For instance, infrastructure for innovation was expected to develop positively in BG-Arable and SE-Poultry, while it was expected to develop negatively in other case studies . In the case of BG-Arable, participants expected increased collaboration, leading to innovation, in case the system would collapse. In the case of ES-Sheep, participants expected that the current low profitability of farmers will not allow investment in new infrastructures for innovation. All studied farming systems were perceived to be “close” or “at or beyond” at least one critical threshold for challenges, function indicators or resilience attributes . The actual state of the system may be more or less close to a threshold than the participant’s perception. Obviously, for case studies that are perceived to be “at or beyond” critical thresholds while still continuing business as usual, the actual state must be at a different position than perceived. Still, perceived closeness can be seen as a clear stress signal, indicating that change is needed, expected or even already experienced. An example refers to the ban of crop protection products before alternatives are available. This stress signal could instigate a study about a reasonable time to phase in/ out regulations regarding the use of crop protection products before actually implementing them. Perceptions of being close to or at critical thresholds also indicate that, from the perspective of farming system actors, immediate action is needed to preserve the farming system or guide it in its transition, thus avoiding a situation where sustainability is even lower. Looking at multiple challenges puts individual challenges into perspective.

To give an example, climate change may be a problem causing regime shifts in many socio-ecological systems , but for the studied farming systems this is not the only challenge and often also not perceived to be the most urgent, except for some arable systems . This supports the notion that climate change should be studied in the context of other drivers . At a global level, reducing anthropogenically induced climate change is, of course, urgent and agricultural systems’ contribution to it must be reduced. Some challenges experienced by FS actors, especially farmers, may also be implicitly caused by climate change; for instance changing legislation and high input costs. For most of the farming systems in our study, climate awareness of some stakeholders, such as conventional farmers, is however not likely triggered due to the impact of climate change on their system per se. When deliberated in an appropriate manner with those stakeholders, new legislation in the context of fighting climate change may however have considerably more effect regarding changing stakeholder perceptions. Function indicators for food production and economic viability were often perceived to be close to critical thresholds. This confirms the need to closely monitor economic indicators as is done in the CMEF of the CAP . When discussed, social function indicators were generally perceived to be “not close” or “somewhat close” to a critical threshold, except for ES-sheep where participants experienced that a critical threshold was exceeded . Environmental function indicators were in most cases perceived to be “not close” or “somewhat close” to critical thresholds . Only in arable systems,hydroponic channel environmental functions were experienced “close” or “at or beyond” critical thresholds. This was mainly related to the capacity of soils to deal with an excess or lack of water, often due to climate change. Participants in workshops of arable systems indicated that a lot of effort was already required to maintain rather than to improve the current soil quality. Arable systems, in need for soil improvement to avoid critical thresholds, would benefit from enabling conditions at national and EU level that foster the maintenance of natural resources.

Mitter et al. , based on a mechanistic scenario development approach for EU agriculture, expect improved attention for natural resources only in a scenario following a “sustainability pathway” out of five possible future scenarios. Current conditions and their future development hence do not seem to support a resilient future of arable systems. Overall, perceived closeness to critical economic thresholds could explain the perceived lower importance of social and environmental functions compared to economic and production functions . Defining critical thresholds seemed most difficult for resilience attributes . According to Walker and Salt it is actually impossible to determine critical thresholds for resilience attributes because they all interact. However, function indicators also interact, but were easier to assess for participants. We argue that difficulties in determining critical thresholds are probably more an indication of the perceived redundancy of resilience attributes for system functioning: presence and contribution to resilience was low to moderate according to stakeholders’ perceptions . This could be related to a control rationale , in which keeping a relatively stable environment and improving efficiency is more important than increasing the presence of resilience attributes. It should be noted, however, that participants often could indicate enabling conditions that improve the resilience attributes. This could be an indication that participants are aware of the importance of resilience attributes, but are in need for more concrete, locally adapted indicators that represent the resilience attributes. In any case, suggesting improvements for resilience attributes could be seen as an implicit acknowledgment by participants that building capacities for adaptation or transformation is required. Perceived thresholds may be different than the real threshold. For the systems that are perceived to be “at or beyond” critical thresholds, it is not necessarily too late to adapt in case the real threshold is actually at a different level than the perceived one. The extensive sheep system in Spain was judged to be close to a collapse, but alternative systems and strategies to reach those have been proposed . In IT-Hazelnut, introduction of new machinery in the past has made farming more attractive for the younger generation, thus avoiding depopulation . Further developments in IT-Hazelnut regarding local value chain activities at farming system level rather than farm scale enlargement, are aimed to further stimulate economic viability and the retention of young people in the area . In PL-Horticulture, the case study is relatively close to Poland’s capital where access to land is limited, system actors aim at increasing the economic viability via vertical and horizontal cooperation at farming system level, which keeps re-attracting seasonal laborers from nearby Ukraine, where wages are lower, to the region. The common factor in these examples of adaptation is that resources are needed to implement them. Be it financial, human, social or other forms of resources. The examples above also suggest that coming back to a desired state, even after exceeding a critical threshold, is possible, provided the disturbance causing the exceedance does not last too long , and adaptation strategies are available . The notion of a critical threshold being a combination of magnitude and duration was not discussed much in the workshops but could help to further define critical thresholds. For instance with regard to the number of years the farming system can deal with extreme weather events as was done in NL-Arable. It is worth noting that challenges are perceived to be more often “at or beyond” perceived critical thresholds than function indicators and resilience attributes.