Policy Interventions Promoting Sustainable Food- and Feed-Systems: A Delphi Study of Legume Production and Consumption

The foodand feed-value systems in the European Union are not protein self-sufficient. Despite the potential of legume-supported production systems to reduce the externalities caused by current cultivation practices (excessive use of N fertilizer) and improve the sustainability of the arable cropping systems and the quality of human diets, sufficient production of high-protein legume grains in Europe has not been achieved due to multiple barriers. Identifying the barriers to the production and consumption of legumes is the first step in realizing new pathways towards more sustainable food systems of which legumes are integral part. In this study, we engage stakeholders and decision-makers in a structured communication process, the Delphi method, to identify policy interventions leveraging barriers that hinder the production and consumption of legumes in the EU. This study is one of a kind and uses a systematic method to reach a common understanding of the policy incoherencies across sectors. Through this method we identify policy interventions that may promote the production of legumes and the creation of legume-based products in the EU. Policies that encourage reduced use of inorganic N fertilizer represent an important step toward a shift in the increased cultivation of legumes. Relatedly, investment in R&D, extension services, and knowledge transfer is necessary to support a smooth transition from the heavy use of synthetic N fertilizer in conventional agriculture. These policy interventions are discussed within current EU and national plant-protein strategies.


Introduction
Legume-supported value chains, from production to consumption, provide evidencebased advantages that include improved ecosystem functions and resource use efficiency, as well as farmed animal and human health provisions. Environmental co-benefits of legumes include reduced nitrate leaching, increased food sources for pollinators, a greater structural diversity of farmland, and improved soil fertility [1,2]. Despite the potential of legumes to improve the sustainability of arable cropping systems and enhance the quality of farmed animal and human diets, the production and consumption of legumes in Europe is low, and their demand as feed is high, which is the adaptation of a phenomenon termed "the legume paradox" [3].
Multiple barriers limit the production, processing, marketing, and consumption of legumes in Europe, which are caused by various forms of system lock-ins and capacity gaps that span multiple levels of the food system. These system lock-ins have been

The Delphi Method
Expert-based future methods systematically investigate and develop policies regarding complex problems via the stimulation and informing of public dialogues. Policymakers can benefit from such strategic foresight in three main ways, according to the Organisation for Economic Co-operation and Development (OECD): (1) better anticipation, to identify and prepare sooner for new opportunities and challenges that could emerge in the future; (2) policy innovation, to spur new thinking about the best policies to address these opportunities and challenges; and (3) futureproofing, to stress-test existing or proposed strategies against a range of future scenarios. It is helpful especially in times of uncertainty when multiple future possibilities emerge, and the predictive capacities are limited [15].
Several other foresight methods are used in food policy and governance studies to create pathways to sustainability transformations. Delphi studies are often implemented in multimethod settings, in combination with multiple other foresight methods, such as horizon scanning, road mapping, visioning, multi-criteria analysis, stakeholder mapping, scenarios, participatory planning, back-casting, and serious gaming [16]. Foresight for food system transformations often requires multi-actor settings. Based on insights from the four case studies of the TRANSMANGO project, Hebinck et al. [17] argued that such collaborative spaces are prefigurative. That is, such foresight initiatives not only conceptualize, but even initiate transformative change by conceptualizing the needed change, creating new actor networks, and generating high-chance implementation strategies. In another setting, several pathways towards a sustainable food system in Kyoto in Japan have been designed by applying the combination of visioning, back-casting, and simulation games that altogether stimulate learning about new food-system practices [18]. In addition, new modes of governance, components, and resources for degrowth (moving away from implementation models that place economic growth as a central tenant of sustainable development) for food systems have been explored, for example, by the Budapest City Lab via visioning workshops with the actors of the local food system [19].
Delphi as a semi-quantitative, interactive foresight method gained popularity over the last 20 years for deployment in multiple settings with aims that ranged from exploratory studies to policy recommendations, from anticipating possible futures to suggesting desirable ones [20]. Delphi is typically preferred when the skill and knowledge of individuals in a particular area or subject need to be mobilized via an expert panel around a specific policy problem(s). The "legume paradox" offered such a scenario and helped identify policy recommendations for legume-based food systems. The Delphi method, as developed by Rand Corporation futurologists Norman Dalkey and Olaf Helmer, is considered an expert (tacit and explicit) knowledge aggregation procedure that can help planners and decisionmakers inform pragmatic choices regarding possible futures [21]. The Delphi method was devised to "obtain the most reliable consensus of opinion from a group of experts by subjecting them to a series of in-depth questionnaires, interspersed with controlled-opinion feedback" (p. 458, [22]). Helmer further explained that, "Delphi inquiry is not an opinion poll relying on drawing a random sample from the 'population of experts'." Instead, once a set of experts has been selected (regardless of how), it provides a communication device for them, and uses the mechanism of the exercise as a filter to preserve the anonymity of responses (p. 19, [23]).
Through a structured future-oriented communication process, Delphi enables the integration of individual expert opinions into a shared worldview. It thus helps the exploration of a problem space without pushing for a quick compromise. The main benefit of Delphi, as Rikkonen et al. [24] contended, is that it enables relatively quick consensus for solution-seeking. Furthermore, it provides practical descriptions for future decisions, and it is in this sense that it can act as a decision-support tool. Delphi's structured process means that experts are invited into a communicative learning process that enables opinion formation via an anonymous space for feedback and dialogue. Thus, Delphi can reduce pressure on free interaction by dominant individuals or to conform to the majority view [25].
A particular type of Delphi explores and develops dialogues for policy development. The "policy Delphi," according to Rayens and Hahn [14], is a systematic method for obtaining, exchanging, and developing an informed opinion on a policy issue or any institutional problem-such as the legume paradox or puzzle [3], and the arguments around its possible solutions. Raynes and Hahn [14] further emphasized the consensus-seeking aspect of this process for or against policy issues, whereas de Loe [26] outlined that it creates opportunities for future policy decisions. Authors in the published literature differentiated four main types of policy Delphi. In an "Argumentative Delphi" [27], the outcomes are consensual, evidence-based arguments. Thus, Argumentative Delphi can help policymaking, as panel members are known to each other and researcher-facilitators directly generate debate about the conflicting points. In contrast, a "Disaggregative Delphi" [15] provides the clustering of diverse opinions, and researchers attempt to outline various scenarios. In a "Trend Modelling Delphi," experts assess current trends, whereas in a "Structural Modelling Delphi," they assess causal linkages to create a new conceptual models of the issues. In sum, the main advantages of the Delphi, according to Landeta [28], are as follows: Social dominance is avoided, as high-status advocates cannot shape opinions-therefore, a plurality of viewpoints emerge; feedback loops enable a learning journey for participants; and some mathematical-statistical methods can be used in aggregating opinions. The main challenges of a Delphi are subjectivity, as it often reflects the subjective views of the summarizing experts; confirmation bias, i.e., a tendency to select the arguments according to summarizing experts' preconceptions; and interest, since general researchers' interests might be overrepresented compared to other stakeholders [15,16,29].
We chose the eDelphi open-source software (Metodix Ltd, Helsinki, Finland) to send out invitations by email. To reach a balanced composition of the invited experts, beyond our previous professional contacts, stakeholders from regional Legume Innovation Network (LIN) workshops organized within the EU-funded H2020 project TRUE and corresponding authors from the relevant literature on legumes were invited to join the panel. Eighty experts representing various sectors such as research, policy, advocacy, food processing, seed supply/crop breeding, and agronomist/agricultural extension services agreed to participate. Thirty-three (Round 1) and 43 (Round 2) experts finally participated in the Delphi. We gathered basic demographics (gender, age, sector, and education) of the participant experts in Round 1 and 33 participants provided their data. Our panel participants were predominantly 35-65-year-old male researchers with a doctoral degree [30].
The primary goals of this Delphi study were to: 1. Enable contributions of opinions from a panel of geographically dispersed policy experts and stakeholders; 2.
Reach a common understanding of the legume policy incoherencies across sectors; and 3.
Help opinion formation to identify policy instruments to leverage barriers for legume production and consumption.
This Delphi was conducted in two rounds. Round 1 mainly focused on identifying the significant factors contributing to the relatively low production and consumption of Sustainability 2021, 13, 7597 5 of 43 legumes in Europe. Round 2 primarily focused on outlining the potential target areas of policy scenarios, where future actions could support legume production and consumption.

Round 1
In Round 1, participants were asked the following questions.
(a) What are the policy challenges to increase domestic legume production and consumption? (b) Why are legume cultivation and consumption rates relatively low in Europe? (c) Which policies would lead to more legumes in our food system? (d) What are the most effective policy mixes to forge pathways towards legume-supported food and feed systems?
After this set of questions, participants were asked to assess 10 statements (Table 1) outlining different policy interventions, which were identified based on literature review, document analysis, and case studies as reported by Balázs et al. [31]. The assessment looked at two aspects: the probability that a specific policy intervention leads to changes in legume production and consumption, and the impact that a specific policy intervention could have on legume production and consumption. Probability was ranked from 1 to 5, where 1 = very low probability, 2 = low probability, 3 = moderate probability, 4 = high probability, and 5 = very high probability. Impact was ranked from 1 to 5 on the same 1 = very low to 5 = very high impact scale. Participants had the option of including comments after each statement to explain the reasons for their assessment. After submitting their assessments and comments, other participants' responses were made visible so that the participant could choose to modify their answers if needed. It must be noted that various participants rated the impact only against legume production and not consumption, because for some participants (about 15%), these two aspects should have been assessed separately.

Statement Number and Code
Statement Description

(T)
Changing international trade agreements would reduce the EU's dependency on non-taxable soybean imports.

(CI)
Agricultural incentives within the CAP support cultivation of legumes.

(CG)
Green direct payments of the CAP foster the transition to sustainable food and feed systems.

(PS)
Policies supporting legume production and consumption increase industrialized livestock production as well.

(IRD)
Investments in agri-food and -feed research and knowledge transfer increase the competitiveness of protein crops and legume-supported food products.

(NFU)
Preventing the use of inorganic N fertilizers creates room for more legume production.

(CCP)
Climate change policies may influence the reduction of meat production and consumption; they also increase legume cropping and decrease the use of inorganic fertilizer.

(DH)
Nutrition, diet, and health policies and public campaigns that promote the inclusion of legumes in the human diet make legumes more visible and increase imports for consumption.

(PFP)
Public food procurement strategies that focus on sustainability offer healthier options in foodservice markets that trigger shifts towards legume-supported diets.

(TD)
Providing transparency of market data boosts legume value chains. In Round 2, participants were asked to assess the impact of seven policy scenarios (that included different measures and instruments derived from Round 1) that could impact (1) legume production and consumption, and (2) the sustainability of EU agriculture. The seven policy scenarios were extracted from Round 1 as the most relevant interventions, ranging from broad policy changes to more specific and targeted measures and instruments. Impact was assessed on a 7-point scale, where 1 = strongly not impactful, 2 = not impactful, 3 = somewhat not impactful, 4 = neutral, 5 = somewhat impactful, 6 = impactful, and 7 = strongly impactful. Participants were given the option to add comments after assessing each policy scenario. After the individual assessment, they were asked to select three of the seven scenarios they considered the most important to forge pathways toward legume-supported sustainable food and feed systems in Europe. The seven proposed policy scenarios are defined below.
(1) Legal measures to reduce the use of synthetic N fertilizer use (i.e., allowances for farmers). (2) Increased environmental, safety, and ethical standards for imported raw protein sources. (3) Funds for public-private extension plus research and development (R&D) services supporting legume-supported cropping systems. (4) Support for investments in technology (including breeding and agrotechnology), storage, and processing. (5) Climate measures related to food consumption e.g., tax on meat, ban on red meat. (6) Dietary guidelines and healthcare recommendations on why and how to shift to a (more) plant-based diet. (7) Elimination of the CAP.
In addition, participants were asked to answer the following questions.
(1) How fast do you expect the proposed policy instruments to be implemented? (2) Is it an incremental policy change, or rather a radical shift in policy that could lead to the implementation of the policy instruments you envisaged? (3) What could be the role of various institutions (public and state institutions, large and small business, and bottom-up civic initiatives) to induce policy change?

Data Analysis
Data were analyzed using Gretl [31] and QDA miner 4 lite (Provalis Research, Montreal, QC, Canada). Descriptive statistical analyses were undertaken that included measures of means, medians, and quartile and interquartile ranges. This provided a first indication of the consensus or disagreement amongst participants. The greater the interquartile range, the greater the degree of dispersion from the central point of the trend in the group's response (median). Narrative answers (the textual comments) were coded using 40 in vivo codes in the first round and 51 in vivo codes in the second round. In vivo codes in the first round were grouped into six categories and mapped into a basic mental model showing logical links between the different categories. The six categories were enriched, cross-checked, and refined in the second round (see Appendices A and B for a detailed explanation). The six categories that finally emerged were (1) economics, (2) social, (3) environmental, (4) policies, (5) technology, and (6) farming.
In the economics category we had one sub-category ("economic/business") that included two codes (trade and marketing). Other codes included in the economics category were incentives, competition, value chain, demand, investment, and costs. In "social" there were nine codes: cultural barriers, tradition, values, preferences, complexity, health, consumption, education, and activism.
In the environmental category the six codes were sustainability, carbon sequestration, biodiversity, externalities, integrated/diversified farming, and greening measures; it Sustainability 2021, 13, 7597 7 of 43 also included one sub-category ("environmental/externalities"), which consisted of the following codes: climate change, pollution, fertilizer, and pesticides.
In the policies category, there were six codes and one sub-category with three codes. The six codes were subsidies, policy reforms, bans, regulatory policies, inertia, and supportive. The three codes for the sub-category "policies/bottlenecks" were systems lock-ins, risks, and bottlenecks.
In the technology/innovation category the five codes were processing, agricultural extension services, research, breeding, innovation, and knowledge. In the farming category, there were eight codes: production, crops/arable area, meat/dairy, smallholder/small scale, nutrition, good agricultural practices, rotations, and pest resistance. The codes were used to categorize the text of the comments and to quantify the codes that occurred most often in the comments for each scenario in Round 2.

Round 1
Three policy interventions were ranked as most impactful and probable: (1) investments in research, development, and knowledge transfer (IRD); (2) restrictions on synthetic N fertilizer use (NFU); and (3) nutrition, diet, and health policies (DH). The results of the impact and probability assessment of all 10 statements are shown in Figures 1 and 2, and the qualitative analysis summary is shown in Figure 3.
The impact mean value of these policy interventions was around level 4, and IRD received lower values than NFU and DH ( Figure 1) because there was doubt whether R&D alone could increase the production and consumption of legumes, but the probability that this measure would occur received higher consensus ( Figure 2). Outliers represented extreme positive or negative visions that this (IRD) measure alone could either solve the problem, or that it would not be possible to modify the system at this point since it is locked in.
The probability that NFU would lead to change was less clear, as the score values were widespread and ranged from 1 to 5 (Figure 2), indicating that the participants had contrasting views. Some participants voiced the fear that such a measure could negatively affect the entire agricultural sector, in particular cereal production, thus triggering an abandonment of farming altogether. As reported in the comments by some of the participants, this policy intervention is likely to produce a drastic change towards higher legume production, but the success of it depends on how to make such a shift. Nutrition, diet, and health policies (DH) were rated higher. Still, such policies' impact is dependent on other policies that support innovation in food technology and incentivize short value chains. Participants pointed out that raising awareness of the benefits associated with legumes is a slow process, and the strategic design of such campaigns is key to the success of these interventions.
Policy interventions related to climate change (CCP), CAP greening (CG), CAP agricultural incentives (CI), and public food procurement (PFP) were considered of medium impact and probability, receiving values between 3 and 4 ( Figures 1 and 2). Participants were in consensus about the high probability ( Figure 2) that CAP-related policy interventions (CI and CG measures) would change legume consumption and production; however, the impact of these measures was less certain (Figure 1), as conventional farmers may need to be supported to meet the greening criteria and the market opportunities required to sustain these policies. Climate change policies received less consensus for impact (values ranged from 2 to 4, Figure 1) and probability (values ranged from 3 to 4, Figure 2) because of political inertia for implementing these measures and their limited scale of application. Public food procurement (PFP) had a medium impact, and the probability that this measure may change legume consumption and production is dependent on the innovative potential of food technology and the possible future change in consumer preferences and traditions.
The policy interventions that received the lowest values were changes in international trade agreements (T), transparency of data (TD) (Figures 1 and 2), and policies supporting legume production and consumption (PS). Policies in support of legume production and consumption may not be relevant because increased production of legume crops in the EU may not be possible/competitive and any of these policies must also address sustainability in general. Change in trade agreements received low support from participants who warned about complex interrelations within a globalized market (closing the gap between imported soybean sets the price level for protein crops in the EU). TD was considered to have a low impact, probably because the data are already available, and it showed no influence, according to the participants.
The policy interventions that received the lowest values were changes in i tional trade agreements (T), transparency of data (TD) (Figures 1 and 2), and polici porting legume production and consumption (PS). Policies in support of legume p tion and consumption may not be relevant because increased production of legum in the EU may not be possible/competitive and any of these policies must also a sustainability in general. Change in trade agreements received low support from pants who warned about complex interrelations within a globalized market (clos gap between imported soybean sets the price level for protein crops in the EU). T considered to have a low impact, probably because the data are already available showed no influence, according to the participants.  Table 1), where T = trade policies; CI = CA incentives; CG = CAP greening measures; PS = policy in support of legumes; IRD = investment in research and knowled transfer; NFU = reduction in N fertilizer use; CCP = climate change policies; DH = nutrition, diet, and health policies; P = public food procurement; and TD = transparency of data. Impact was assessed on a 5-point scale, where 1 = very lo impact, 2 = low impact, 3 = moderate impact, 4 = high impact, and 5 = very high impact. The box displays the second a third quartiles; the line shows the median value, the cross is the mean, the whiskers display the interquartile range, a dots depict outliers.  Table 1), where T = trade policies; CI = CAP incentives; CG = CAP greening measures; PS = policy in support of legumes; IRD = investment in research and knowledge transfer; NFU = reduction in N fertilizer use; CCP = climate change policies; DH = nutrition, diet, and health policies; PFP = public food procurement; and TD = transparency of data. Impact was assessed on a 5-point scale, where 1 = very low impact, 2 = low impact, 3 = moderate impact, 4 = high impact, and 5 = very high impact. The box displays the second and third quartiles; the line shows the median value, the cross is the mean, the whiskers display the interquartile range, and dots depict outliers.   Figures 1 and 2). High = the policy has high potential to improve legume production and consumption; medium = the policy has medium potential to improve legume production and consumption; and low = the policy has low potential to improve legume production and consumption. Figure 3. Summary description of participants' assessment (impact × probability, as shown in Figures 1 and 2). High = the policy has high potential to improve legume production and consumption; medium = the policy has medium potential to improve legume production and consumption; and low = the policy has low potential to improve legume production and consumption. Figure 4 shows the combined results of how respondents assessed the potential impact of the seven policy scenarios (Table 2) on the sustainability of EU agriculture and legume production and consumption. Participants rated the impact of eliminating the CAP as neutral or not impactful. Many commented that such a change is very drastic, and could result in a collapse in the EU agricultural sector. Few other participants (outliers) agreed that such an extreme measure is required to support the production and consumption of legumes, though the consequences may have negative social and economic repercussions if not properly implemented. The scenarios related to agricultural extension services, synthetic N fertilizer use, and R&D were considered to have a medium impact (between 5 and 6) both on EU agriculture sustainability and on legume production and consumption. The outliers in the R&D and agricultural extension services represent the vision that changes may require a structural reframing of the way agronomists and researchers are trained, because currently education in agricultural sciences is focused on cereals and oilseeds (and not on agroecological principles). Some respondents commented that reducing synthetic N fertilization may reduce yields to the extent that the whole agriculture sector would be impacted, causing an abandonment of farming altogether, as previously mentioned in Round 1. This impact-which has not occurred-was claimed before when the Water Framework Directive was implemented. Certainly, the impact of such a measure is stronger on legume consumption and production than on the sustainability of EU agriculture. Table 2. Seven relevant policy interventions (left column) were identified in Round 1, and corresponding scenarios (right column) were discussed and ranked in Round 2.

Measure "What If?" Scenarios
Climate measures There will be public consensus to launch radical climate mitigation measures (e.g., requiring a significant decrease in the carbon footprint caused by food production), and collective political decisions are made in this direction.
Eliminate the CAP A radical policy shift happens, and the EU stops all agricultural subsidies through the CAP.

Agricultural extension services
Farm advisory and agricultural extension services become fully capable (both in terms of knowledge and human and financial resources) of supporting farmers and other actors along the value chain to implement new research outcomes at the farm level.
Regulating synthetic nitrogen fertilizer use Stricter environmental and climate regulations are applied to crop and livestock production in Europe. The new measures combine legal restrictions on synthetic N use with a system of allowances for farmers.

Nutrition, diet, and health policies
Diet and health policies build on, and at the same time share with consumers, the best available knowledge on the nutritional value and the health and environmental impacts of different food sources. Nutritional guidance on calorie intake is provided to consumers through various means (e.g., doctors, public food procurement rules, etc.).

R&D investments
Public R&D investments are focused on new breeds, effective crop rotation schemes, new strategies to recycle N better, and novel options for the storage and processing of legumes.
Trade policy Environmental, safety, and ethical standards are increased for imported raw protein sources (used either for feed or food) to level the playing field for homegrown legumes.
Interestingly, reducing synthetic N fertilization was preferred over others because it was considered the one with the highest probability of increasing legume production and consumption ( Figure 5). Climate measures were rated positively on average (mean = 6, Figure 4) though their range was broad, and so consensus was not apparent. Similarly, trade policies were considered to have medium and high impact on legume production and consumption, and less impact on agricultural sustainability. Scenarios of nutrition, diet, and health received the highest consensus for the impact on EU agriculture sustainability, with some outliers, and less consensus for legume production and consumption. An indepth analysis of the text and the reasons given by the participants for the assessment are provided below.  Table 2, where NU = nitrogen fertilizer use; T = trade policies; RD = investment in research and development; ES = agricultural extension services; CM = climate measures; DH = nutrition, diet, and health; and EC = elimination of the CAP. Each policy scenario was evaluated for its impact on (L) legume production and consumption and (A) the sustainability of EU agriculture. Impact was based on a seven-point scale, where 1 = strongly not impactful, 2 = not impactful, 3 = somewhat not impactful, 4 = neutral, 5 = somewhat impactful, 6 = impactful, and 7 = strongly impactful. Each boxplot displays the second and third quartiles, the line within each box shows the median value, the whiskers display the full data range, and dots depict outliers.

Discussion of the Policy Scenarios
Delphi for food policy transformation is becoming increasingly popular. For example, in 2015 an international panel of food policy experts assessed the effectiveness of good food environment policies via the Delphi study by Mahesh et al. [32]. Food prices and promotion were found to be the most valued policy domains with regards to impact on improving population nutrition. Interestingly, trade received the lowest weighting. As for specific policies, taxing unhealthy foods and promoting healthy food provision in schools were the most highly valued, whereas nutrient declarations on packaged foods and healthy food policies in private-sector workplaces received the lowest weightings. Tiberius et al. [33] explored the potential development of cultured meat by 2027. The Delphi study participants doubted that the challenges of mass production, production costs, and consumer acceptance would be overcome by 2027.
Considering climate change and its impacts, consumer perceptions, and continued research and development, environmentally sustainable food systems are inevitable. In an expert Delphi, Antonelli et al. [34] explored the trends, challenges, and policy options in the agri-food sector of the Mediterranean region over the short (to 2020) and the long (to 2030) terms by developing "pessimistic" versus "optimistic" scenarios. Similarly, the assessment of country-of-origin labeling policy mobilized 19 food policy experts from 13 countries in a consensus-seeking Delphi [35]. Based on expert knowledge, the multiple-  Table 2, where NU = nitrogen fertilizer use; T = trade policies; RD = investment in research and development; ES = agricultural extension services; CM = climate measures; DH = nutrition, diet, and health; and EC = elimination of the CAP. Each policy scenario was evaluated for its impact on (L) legume production and consumption and (A) the sustainability of EU agriculture. Impact was based on a seven-point scale, where 1 = strongly not impactful, 2 = not impactful, 3 = somewhat not impactful, 4 = neutral, 5 = somewhat impactful, 6 = impactful, and 7 = strongly impactful. Each boxplot displays the second and third quartiles, the line within each box shows the median value, the whiskers display the full data range, and dots depict outliers.

Discussion of the Policy Scenarios
Delphi for food policy transformation is becoming increasingly popular. For example, in 2015 an international panel of food policy experts assessed the effectiveness of good food environment policies via the Delphi study by Mahesh et al. [32]. Food prices and promotion were found to be the most valued policy domains with regards to impact on improving population nutrition. Interestingly, trade received the lowest weighting. As for specific policies, taxing unhealthy foods and promoting healthy food provision in schools were the most highly valued, whereas nutrient declarations on packaged foods and healthy food policies in private-sector workplaces received the lowest weightings. Tiberius et al. [33] explored the potential development of cultured meat by 2027. The Delphi study participants doubted that the challenges of mass production, production costs, and consumer acceptance would be overcome by 2027.
Considering climate change and its impacts, consumer perceptions, and continued research and development, environmentally sustainable food systems are inevitable. In an expert Delphi, Antonelli et al. [34] explored the trends, challenges, and policy options in the agri-food sector of the Mediterranean region over the short (to 2020) and the long (to 2030) terms by developing "pessimistic" versus "optimistic" scenarios. Similarly, the assessment of country-of-origin labeling policy mobilized 19 food policy experts from 13 countries in a consensus-seeking Delphi [35]. Based on expert knowledge, the multiplecountries-of-origin labels can give accurate information about the origin of food produced by two or more countries, avoiding misinformation for consumers. In the TRANSMANGO EU project, 45 international food policy experts participated in a Delphi to identify global drivers of the food system affecting EU food and nutrition security. According to Moragues-Faus et al. [36], there seems to be a broad consensus around the FAO definition of food and nutrition security (identifying it as access, availability, and sustainability). However, much less consensus could be recorded about the food system, which is understood in more dynamic terms, either as actor oriented (a small number of dominant agents' structure value chains), or systemic, where a mixture of stakeholders, flows, material devices, institutions, norms, beliefs, and activities constitute mixed determining factors. Note that only a few respondents referred to "intangible" dimensions of the food system in their definitions, such as governance, culture, environmental externalities, and knowledge. Allen et al. [37] proposed sustainable food system metrics specifically for the Mediterranean area and involved expert-agreed consensus in a Delphi survey regarding the indicator selection process. Frewer et al. [16] suggested exploratory workshops to refine Delphi questions and showed that beyond the relevance of the issue to the invited participant, response rates could be increased by leveraging personal networks. They also emphasized that policy uptake of the outputs of Delphi merits further research. Boylan et al. [38] explored the perceptions and role of the Australian policy actors on a healthy, sustainable, and safe food system in a Delphi survey. They recorded a critical consensus on the definition and essential elements of a cross-sectoral food and nutrition policy to meet today's environmental, health, social, and economic challenges.
Agri-food policy development could therefore benefit from soliciting expert opinion and consensus solution-seeking by utilizing a Delphi. Although consumer engagement in food and health policy development is rare, public discourse regarding policy options for healthy eating, or against non-communicable diet-related illness, is dominated by industry and government stakeholders. Haynes et al. [39] explored the consensus on obesity policy priorities of underrepresented stakeholders in the Australian context. In a three-round online policy Delphi, consumers, public health practitioners, and policymakers prioritized options to reduced obesity and its impacts. In a final face-to-face discussion group, participants explored stakeholder perceptions of the intrusiveness (ethical acceptability) of obesity policy options. Hung et al. [40] investigated the challenge of improving consumer motivation and interest in healthy eating by using nutrition labels (health claims and symbols). In their policy recommendations and communication guidelines they concluded that health claims with shorter and less complicated messages and health symbols with a visible endorsement were the most highly valued and therefore more efficient.
In essence, seeking food policy transformation via foresight methods such as Delphi are becoming increasingly favored, especially when it comes to critical consensus-solution seeking or assessments of food policy instruments' legitimacy or acceptability. Agri-food policy development for sustainable food systems can benefit from a Delphi by identifying the internal incoherencies and creating new policy discourses about options and priorities that are usually dominated by industry stakeholders. Below we discuss the suite of policy options and our recommendations that may realize sustainable legume-supported food and feed systems more effectively across the value chain, from production to consumption.

Eliminate the CAP
This measure was considered too drastic because a certain level of incentives or subsidies are considered essential to support farmers' income and the economic sustainability of EU agriculture. There is a fear that in the absence of the CAP, market forces would overcome commercial competitiveness of EU farms, causing negative social, economic, and environmental consequences, including the abandonment of farming, land concentration in the hands of very few, and environmentally negative land-use changes. In addition, productivity may not change because a few large-holder farmers will acquire most of the land, and small and medium-sized landholders will cease to be commercially profitable and therefore forced out of farming. This, in turn, would have a serious impact on the economic viability of EU agriculture, and the effect on legume production would be secondary (Figure 4). A few participants commented that elimination of the CAP may be beneficial (whiskers spread to level 6, Figure 4) after careful analysis of costs and benefits and examination of the consequences of such changes in countries where this has been already implemented (i.e., New Zealand, Canada, Australia).
We argue that a reflection on CAP's future and what it can do to increase home-grown legume production must encompass the planning and coordination of the full range of CAP tools that each MS implements in its territory. Obligations on environmental and climate action are strictly dependent on local conditions and, thus, on the strategic planning and implementation within each MS. One of the proposed tools of the future CAP (post-2021) is to implement obligations and incentives for farmers to ensure that crop rotation occurs and that the temporal period of those rotations is increased, which goes beyond simply encouraging crop diversification (i.e., increasing the number of crops in the rotation) [10,41]. Encouraging lengthier crop rotations could induce an increase in legume production as an integral component of a truly holistic cropped system, and not simply for subsidy payment (i.e., greening obligations). A key lesson derived from the EU supranational Plant Protein Strategy is that optimized delivery of the plan must be targeted to accommodate regional-level considerations, such as pedoclimatic, biogeographical, and socio-economic differences [42].

Climate Measures
Many participants concentrated their answers on the suggestion of banning or limiting meat consumption and production, and therefore the impact in this policy scenario varied in a wider range (from point 2 to point 7) than in the other scenarios ( Figure 4). Their main argument against reducing meat production was that it may have a strong impact on the sustainability of farming systems through reduced availability of organic N (and carbon) sources, as well as the absence of grass-lays (and reduced pesticide use, amounts, and formulations) in rotations. However, the economy of many EU regions is dependent on meat production, particularly those with limited capacity for arable operations-since some participants argued that pastures are part of the landscape, and they provide ecological benefits (such as C sequestration, biodiversity). Implementation of climate measures may cause a positive impact on both indicators if integrated farming is considered, where legumes are produced as feed and food in a low-input farming system that includes livestock as an integral part of the production unit [43]. Most participants thought that any reduction in livestock and meat production would cause a decrease in biodiversity and sustainability (Appendix C, Table A1). Potential arguments behind this opinion might be the negative biodiversity outcomes associated with the abandonment of traditional pastures, or the expected increase in land-use pressure associated with the production of meat-replacement options. According to several respondents, the decrease in biodiversity would be visible beyond the farmgate, too, as processing facilities for meat and dairy would need to be replaced, and innovation and investments would be required in new types of processing technologies for the development of plant-based products. The scenario of reducing meat production was favored only by some participants, whereas others assumed this to be currently too extreme due to political inertia and public uncertainty regarding the immediacy of the climate change risk. Reducing the dependency on plant protein imports (i.e., soybean) and/or decreasing the risks of potential protein insufficiency, mainly with respect to the animal feed requirement, is a main goal of the EC report on the development of plant proteins in the European Union, a report that does not discriminate legumes from other protein crops.

Regulating the Use of Inorganic N Fertilizers
Overall, restrictions on mineral N fertilizer use can create a positive impact on the production of legumes, whereas the impact on consumption is less certain. Increased production of legumes requires the development of a set of best practices and crop rotation schemes that are more efficient (i.e., adapted to various pedoclimates to minimize crop inputs) (Appendix C, Table A1). This measure was considered too drastic by some participants, who feared that it may push many farmers out of business, especially conventional farmers.
The inclusion of N-fixing plants such as legumes in the crop rotation or production system (grass mixes) is another management practice that replaces N fertilizer by using biological atmospheric N (N 2 ) as a source and making it available to the plant. Environmental co-benefits include reduced nitrate leaching, increased food sources for pollinators, greater structural diversity of farmland, and improved soil fertility [44]. However, simply regulating N fertilizer use is not enough to make the shift towards increased legume production. Closing the nutrient cycle, i.e., encouraging a circular organic-N economy, at the local, regional, and national levels, including linking urban and rural nutrient cycles, would lead to a more sustainable use of natural resources. This type of approach is partly implemented in organic farming and food systems based on agroecological principles. As highlighted in the development process of the German National Protein Strategy, this return on investment, or risk, is delivered over the longer term, such as 5-8 years compared to a single year, for the immediate commercial rewards offered by market opportunities. As such, farmers need stronger incentives to de-risk this long-term undertaking and bridge any financial shortfall [42]. It is also necessary to integrate N-fertilizer reductions with other interventions, including the breeding of new varieties (e.g., for pest resistance and higher yield), provision of independent agricultural extension services, more research and innovation, and investment in the processing capacities for legume-derived carbohydrates (i.e., fiber/starches and oils), as well as proteins to help add value and boost demand and consumption.

Agricultural Extension Services
Agricultural extension services, including farmer training and cooperative research, can have a strong and positive impact on the sustainability of EU agriculture and the potential increase in legume production and consumption ( Figure 4). It was perceived that this provision would have a stronger impact on production than on consumption. Nevertheless, incentives and policies that support farmers, and other actors along the value chain, to increase legume cultivation may trigger increased demand for agricultural extension services by farmers. Currently, the EU area cultivated with legumes is very small (relative to cereals) and extension services and research agencies are not interested in developing training for agronomic support for a crop such as legumes, which can be considered marginal or underutilized (Appendix C, Table A1). Within the EU, the development of supranational protein strategies and their implementation via effective knowledge transfer is considered a pillar to realize rural areas' economic development. However, most of these strategies value legumes in a rather narrow way, mostly for their protein content, and consequently, they do not help actors along the value chain to capitalize on the multitude of benefits legumes provide for agri-food and -feed systems. Setting up producer organizations and cooperatives, and cooperation along the value chain, are additional strategies that may be adopted to decrease dependence on plant protein imports. The Farm Advisory System (FAS) is currently available to the MS, and this agency also works to ensure farmers' awareness of environmental co-benefits of long-term legume-supported crop rotations for plant protein provision.
Finally, increased funds for training, advisory, cooperation, and innovation in support of greater "fiscal literacy" were also seen as highly important: Such knowledge is very often the key to achieving both commercial profitability and environmental sustainability.

Nutrition, Diet, and Health Policies
Knowledge regarding the health and nutrition benefits of legumes may be insufficient to trigger an increase in consumption. Existing preferences and cultural barriers (Appendix C, Table A1) may be a strong deterrent to the consumption of legumes, even if nutritional, environmental, and health benefits are high, but the effective marketing of novel, easy-to-cook, tasty, and environmentally beneficial plant-based products may effectively increase legume consumption [45].
Therefore, it is important to increase investment in R&D and processing technology to create a market for high-quality plant-based products. Participants believed that impact on production from the perspectives of nutrition, diet, and health would be minimal. Nevertheless, the impact on the sustainability of EU agriculture is expected to be moderate, and the consensus amongst participants on this was high, with only a low number of outliers ( Figure 4).
The market for plant-based "meat" and "dairy" products is rapidly growing, and it is predicted to reach a global net of EUR 4.2 bn by 2020. Currently, Europe is the largest market for meat substitutes, with a 39% global market share [46].
The EU plan on the development of plant proteins suggests that using demand-side policy tools, such as standardization, labeling, and public procurement, may create the necessary market pull for legume-based products and ultimately lead to lower dependence on feed protein imports.

Investing in Research and Development (R&D)
R&D can have a strong impact on legume production and the sustainability of EU agriculture (Figure 4). Legume cultivation has many technical challenges-as reported by the participants-that need to be tackled through applied research. For example, developing more effective crop types such as new varieties with higher and more-stable yields, better pest resistance, short time to harvest, and that may also minimize N loss and optimize N-cycling in-field [19]. These measures could be coupled with other types of incentives (policy reforms) to stimulate legume cultivation (Appendix C Table A1).
An opportunity to encourage the breeding of grain legumes is to ensure farmers with a minimum level of profitability [47]. Besides financial incentives within a national protein strategy, adjustments could be directed towards improving support for decentralized capacities for breeding, storage/aggregation, processing (dehulling, milling, fractionation), and marketing of local and regional legume varieties.

Reforming Trade Policies
Some participants thought that imports would become very costly because of the barriers imposed with the predicted scenario presented in this study (Appendix C Table A1). There was a fear that this policy scenario would increase food prices, with social and economic consequences; hence, the impact on the sustainability of EU agriculture spanned from low to high (Figure 4). Some participants commented that EU homegrown legumes may not be necessarily more sustainable (from "cradle to grave" life cycle assessment) than those produced outside the EU.
This integrative policy scenario may increase the competitiveness of homegrown legumes, but success demands that this approach be complemented with other reforms in R&D and innovations in production and processing, etc. To many participants, it was not clear that the EU would be able to replace imported legumes with homegrown production, as the current area cultivated with legumes and the scale of current imports do not support this vision.

Summary: Forging Pathways towards a More Sustainable, Legume-Based Agriculture
When participants were asked to rank the most impactful policy scenarios, regulating the use of synthetic N was the most highly ranked option ( Figure 5). Elimination of the CAP was considered a very radical change, and most participants did not support it for reasons already explained above (Appendix C Table A1). A few participants commented that most of the scenarios may act negatively against other parts of the sector. Instead, they proposed to have a mix of "positive discrimination" measures in favor of legumes, such as "carbon credits" or an "environmental credit" system. The CAP should not be eliminated; instead, it can be reformed using a system of positive rewards for legume production. Currently, the EFA is the only positive support system in favor of legumes, and it has been ineffective, largely due to the associated ban on pesticide (and N fertilizer) use. Overall, participants preferred incremental changes because adaptation and preparation are needed when reforms are put in place. A radical change was deemed necessary by a small fraction of respondents (<5%), whereas it was considered a potentially costly disruption by the majority. CAP was considered a very radical change, and most participants did not support it for reasons already explained above (Appendix C Table A1). A few participants commented that most of the scenarios may act negatively against other parts of the sector. Instead, they proposed to have a mix of "positive discrimination" measures in favor of legumes, such as "carbon credits" or an "environmental credit" system. The CAP should not be eliminated; instead, it can be reformed using a system of positive rewards for legume production. Currently, the EFA is the only positive support system in favor of legumes, and it has been ineffective, largely due to the associated ban on pesticide (and N fertilizer) use.
Overall, participants preferred incremental changes because adaptation and preparation are needed when reforms are put in place. A radical change was deemed necessary by a small fraction of respondents (<5%), whereas it was considered a potentially costly disruption by the majority. The role of the institutions capable of affecting change were identified as businesses, public and state institutions, and bottom-up civic initiatives. Innovation, processing capacities, and breeding were associated with high responsibility for the private sector (business), whereas knowledge, regulations, and campaigns for awareness-raising were considered to pertain to public and state institutions. Bottom-up civic initiatives should be more concerned with activism, education, and awareness-raising.

Conclusions
The Delphi method aimed to assess policy options by extracting the most controversial elements. That is, although our aim was not to achieve consensus amongst stakeholders, we could identify an agreement on most of the most important questions.
• Implementing policies that encourage reduced use of inorganic N fertilizer is an important step towards a shift in the increased cultivation of legumes. It is not clear to what extent this would create an impact on the consumption of legumes. • Investment in R&D, agricultural extension services, and knowledge transfer is necessary to support the point above and allow for a smooth transition from high use of synthetic N fertilizer in conventional agriculture to precision farming and agroecological farming.  The role of the institutions capable of affecting change were identified as businesses, public and state institutions, and bottom-up civic initiatives. Innovation, processing capacities, and breeding were associated with high responsibility for the private sector (business), whereas knowledge, regulations, and campaigns for awareness-raising were considered to pertain to public and state institutions. Bottom-up civic initiatives should be more concerned with activism, education, and awareness-raising.

Conclusions
The Delphi method aimed to assess policy options by extracting the most controversial elements. That is, although our aim was not to achieve consensus amongst stakeholders, we could identify an agreement on most of the most important questions.

•
Implementing policies that encourage reduced use of inorganic N fertilizer is an important step towards a shift in the increased cultivation of legumes. It is not clear to what extent this would create an impact on the consumption of legumes. • Investment in R&D, agricultural extension services, and knowledge transfer is necessary to support the point above and allow for a smooth transition from high use of synthetic N fertilizer in conventional agriculture to precision farming and agroecological farming. • Mitigation and adaptation strategies to combat climate change can have an indirect positive effect on legume production and consumption if these policies are implemented on a large scale and effectively.
• Policies that tackle nutrition, health, and diet are relevant for the increase in legume consumption and, indirectly, legume production. However, preferences, culinary traditions, and cultural habits are difficult to change with top-down approaches or via promotional and information campaigns. • Citizen-led initiatives that inform and educate the public on the environmental and health benefits of legume consumption should accompany policies that tackle production and farming strategies (i.e., greening payments with other agroecological and rural development incentives, etc.).

•
The CAP is an important tool to support food production and protection of the natural environment, and of which the farm is a key component for the creation of any positive externalities. Hence, the CAP needs to be reoriented.

•
Trade agreements are part of a complex web of interconnections between economies around the world, and any intervention will have repercussions along the food chain in general; therefore, this possibility is not advocated for.
A small number (<5%) of participants highlighted the importance of framing incentives instead of "punishments" (taxes, penalties, etc.) more positively for the implementation of farming practices that include legumes in the farming plan (rotations, crop mixtures, intercropping). The EC "Green Deal" aims to stimulate such a policy framework, promoting positive incentives that may work in support of legume production and consumption. By 2050, European agriculture aims to be "carbon neutral," with goals set in the Climate Act, and Directives of Renewable Energy, Energy Efficiency, Land-use Change and Emission Trading. At least 40% of the overall CAP budget for 2021-2027 would contribute to "climate action." A reduction in carbon emissions may also be coupled with reduced use of synthetic fertilizers, especially N, which need to be substituted by more environmentally friendly and diversification-focused practices, including increased legumes in the rotation, and more mixed (arable plus livestock) farmed units, which are largely self-sufficient in animal feed provision.
The most important sections of the Green Deal for the increased production and consumption of legumes are within the "Farm to Fork Strategy." The strategy aims at a "fair, healthy and environmentally-friendly food system" via the introduction of a new business model focusing on performance rather than compliance. Within this framework, ecoschemes reward farmers for improved environment and climate performance, including carbon capture and retention and improving nutrient management to enhance the quality of water and reduce GHG emissions. These practices shall be financially stimulated within the CAP and other public or private initiatives. The Loss of Nutrient Strategy entails a reduction in nutrient losses by 50% and a reduction in fertilizer use by 20% by 2030. Reducing pesticide use (50% by 2050) is another strategy within the Green Deal that may favor R&D investment in leguminous crops.
The Farm to Fork Strategy also aims to reduce food processing's environmental impact, launching a process to identify new innovative food and feed processes and products such as plant-protein-rich food. Lastly, the strategy will strive to stimulate sustainable food consumption and to give consumers better information on the provenience, nutritional value, and environmental footprint of food.
Hence, the Green Deal in general, and the Farm to Fork Strategy in particular, represent a substantial policy innovation that may positively impact legume production and consumption in the EU. The question remains: Will the strategy be well implemented in the various member states, and how will the barriers and system lock-ins be overcome? For example, in this study, the provision of R&D and agricultural extension services was been as one of the most important policy instruments to increase legume production (and consumption) in Europe. However, this factor is not directly addressed in the Green Deal, and it may be a necessary factor for implementing such policy innovation. Oliver et al. (2018) [48] highlighted knowledge constraints as one of the mechanisms locking the food system into its current unsustainable state. Knowledge constraints include lack of access to specific skills and uncertainty regarding the benefits of alternative approaches-both of which are crucial aspects for the adoption of the Farm to Forks Strategy's ambitious goals by European farmers. Regional social "knowledge" or "learning clusters" and multistakeholder collaboration are considered necessary triggers of synergies in the bioeconomy to reduce financial barriers for innovators and SMEs (small and medium enterprises). Hassink (2005) [49] proposed learning clusters as focal points within which dependencies or lock-ins may be overcome. To implement effective regionalized social networking and education strategies, there is a need first to understand and manage the different processes of learning among the various value-chain clusters. Then, innovation-related regional actors (politicians, policymakers, chambers of commerce, trade unions, higher education institutes, public research establishments, and companies) can cooperatively and carefully identify resources in the region that could help foster development (anti-lock-ins). We support this view of empowering localized citizen-led regional clusters to implement policy innovations such as the Farm to Fork Strategy. This study identified policy interventions that can determine the future of legumes within the EU food system, and they concur with innovation policies proposed at the EU level, such as the Farm to Fork Strategy within the EU Green Deal. Nevertheless, future efforts should focus on discerning the details regarding the effective implementation of overarching policies. Systems lock-ins and path dependencies stem from well-established routines and capacities that are engrained and embedded in (often successful) local economies. Breaking lock-ins is key to the implementation of innovative policy instruments for the required rapid change toward more sustainable climate-positive cropped systems and healthier diets. Acknowledgments: For successfully deploying the eDelphi (https://www.edelphi.org/), the support from Hannu Linturi and Antti Kauppi at metodix.fi is greatly appreciated.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix A. Diagram of Codes Used for Text Analysis
Explanation of Appendix A Forty in vivo codes in the first round and 51 in vivo codes in the second round were grouped in six categories: (1) economics, (2) social, (3) environmental, (4) policies, (5) technology, and (6) farming. The mental model above shows logical links between the different categories.
In the economics category, we had one sub-category ("economic/business"), which included two codes (trade and marketing). Other codes included in the economics category were incentives, competition, value chain, demand, investment, and costs.
In the Social category there were nine codes: cultural barriers, tradition, values, preferences, complexity, health, consumption, education, and activism.
In the environmental category the six codes were sustainability, carbon sequestration, biodiversity, externalities, integrated/diversified farming, and greening measures; it also included one sub-category ("environmental/externalities") which consisted of the following codes: climate change, pollution, fertilizer, and pesticides.
In the policies category there were six codes and one sub-category with three codes. The six codes were subsidies, policy reforms, bans, regulatory policies, inertia, and sup-portive. The three codes for the sub-category "policies/bottlenecks" were systems lock-ins, risks, and bottlenecks.
In the technology/innovation category the five codes were processing, agricultural extension services, research, breeding, innovation, and knowledge.
In the farming category, there were eight codes: production, crops/arable area, meat/dairy, smallholder/small scale, nutrition, good agricultural practices, rotations, and pest resistance.
These codes were used to analyze the text of the comments and to quantify the codes that occurred most often (see Figure 4) in the comments for each scenario in Round 2. Consumers ask already today for healthy legumes; a more vegetarian diet will increase and with that the consumption of legumes 20 1.20%

Consumption
Regulating synthetic N use Improvement/increases in consumption is likely, in that production can be a driver for consumption in the case of legumes, i.e., increased production has the potential to increase consumption 30 1.80%

Preferences Nutrition, diet, and health
If calories-intake is to be reduced this could mean less meat not necessarily more legumes. Humans like and eat food not only because of its nutritional value but because they enjoy the taste, flavour, and texture etc. I think people are more willing to change their diet slightly for a greater environmental benefit than for personal health. 58 6.10%

Preferences Nutrition, diet, and health
The question is how to get it into consumer hands and then how to get them to use it. Mild positives in both axes as the outcome would be reduced consumption of animal products, with all of its on-costs, and mild increases in plant-protein production.  Under confinement in Spain the increase in the use of grain legumes in the diets has increased significantly: the reason that was given was mainly parents or families had more time for cooking. so the demand is there built in the traditions and also nowadays the gourmet approaches 48 5.00% Heatlh Nutrition, diet, and health Proteins especially legumes in human diets will increase significantly over the next 10 years. They will not only be a source of protein but provide healthier forms of carbohydrate, fibre, mineral and vitamins I am very uncertain about my answer, but my feeling is there is going to be a negative effect on legumes and overall sustainability by such a radical move. Sustainability will decrease because biodiversity associated with temporary and permanent grasslands will drastically decrease. 44 4.60%

Fertilizer Climate measures
It may be advantageous to apply cheap mineral fertilizer (produced with cheap excess wind power) rather than introducing unreliable legumes in the crop rotation. If legumes have become a reliable source of income by then, more legumes would be used (maybe also due to a high demand for products replacing meat and dairy The massive cost of Covd19 will have a major impact on EU economies over the next 10 years. Political pressure will be to reduce public expenditure and farm subsidies will be questioned. There will not be the funds available to support agriculture as in the past. The impact on EU legume production and farm sustainability may not be great but the impact on farming will be enormous. The temptation is to believe that farmers would embrace grain legumes wholeheartedly, but they need also to be able to make a profit form their sale and to be reasonably confident in the performance of legumes as a crop. Frequently this is not the case. The inability to combat pests and diseases is a major reason why many organic growers do not produce grain legumes.

Policy reforms R&D
Is it profitable for the farmer? that's the question. and is the aim to produce for the meat farms or looking to change consumer diets to more plant based. as it won't work on standalone basis. long term it could be positive for sustainability. and improved ground conditions. but farmers will need cash crops to continue farming. protein crops for animal feed are price driven. globally. with many European countries not offering the most competitive advantages for a farmer in the global playfield. ask the question where in Europe can you start a commercial farm today and make a living as a farmer from scratch. today this is impossible investment is too high to carry. so, in addition to promote these changes. EU and EU countries will need to review the position of the farmer. a jungle of regulations will not help. More strict and really able to limit the import of soy produced by destroying rain forests in Amazon, there could have a positive impact in legume production in Europe and on sustainability as well. 34 2.50%

Regulatory policies Trade policy
Other rules and regulations (e.g., DG SAN) Also in the meat consumption, the farm-to-fork principle is increasing. More people will be sensible for the kind of animal production. Better informed consumption choices, and the related nutrition, diet and health policies, will always have a significant positive impact. 19 2.00% Knowledge Nutrition, diet, and health Support consumption together with a deeper education (that is: not just information available but a whole support-program, such as directions for cooking, taste lab, taste experiences, etc.), not sure how much it will affect production. 36 3.80% Research Nutrition, diet, and health However, the need is to undertake research to establish the functionality of pulses for use by the food industry as a food ingredient. 23 2.40%

Processing Nutrition, diet, and health
It is feasible to replace large amounts of wheat and maize with pulse flours as more healthy food ingredients if the food processors know how they can utilise them. 29 3.00% Processing Nutrition, diet, and health Alternative foods must be promoted. Primarily however they must be tasty and attractively priced, i.e desirable. This largely means improvements in processed foods with legumes as increasing proportion of the ingredients and a price that moves peoples' attention away from animal protein sources. 43 4.50% Processing R&D Improved processing for feed and food the demand may also be growing, 12 1.10%

Breeding R&D
It is also important that the new types are also bred to optimise system functions, 14 1.30% Breeding R&D New cultivars resistant to diseases (Conventional and genetic editing) and new cropping systems can facilitate introduction of (grain) legumes in rotations because farmers will have an easier task, 28 2

Research R&D
Legume research will only be playing catch up in this respect and will need very considerably more input to make rapid progress, 22 2.00%

Breeding R&D
Progression of varieties is the natural territory of the breeder BUT they are limited in their interests (as is the rest of the value chain) by the size of the market. Production of more is possible using the varieties that already exist. Regional production of species that are not already produced is one potential area that could benefit e.g., chickpeas and lentil in NW Europe, though in reality there is probably material in existence that can already be made to work in many situations IF growers see an opportunity to profit and are prepared to experiment, learn and accept some variability in performance year on year. It is also not realistic either, so I am not in favour of this. We need to consider the whole food system and also population health, how to best fit animal production to plant production and also take into account non-food side-streams that be utilised in animal husbandry to produce high-value protein for humans. 57 6.00%

Nutrition Climate measures
Decreasing the consumption of red meat from the current level may do us good but whole nations to move to vegan diets may expose to some serious health risks, particularly in vulnerable groups such as growing children and aged people. Be careful with the huge increase of biomass production and the requirement to find and to develop new end uses (non-food?). 80% of the land is used to feed animals today, so the livestock reduction will offer the opportunity to grow something else: the question is what. As regards the implications for the sustainability of EU agriculture, clearly the potential expansion of legumes needs to be part of a new crop rotation regime in which beans & pulses co-exist with grains, roots, and tubers. 37 2.20%

Rotations
Regulating synthetic N use Any restriction in use of N will certainly lead to a modification of the cropping systems. Rotations will be adapted to include N-fixing crops. 25 1.50%

Production
Regulating synthetic N use Restrictions on synthetic N use will have a negative impact on crop productivity. 13 0.80%

Regulating synthetic N use
To counter this the need will be to create a different balance in the rotations which are likely to lead to the use of more legume crops. 27 1.60% Production Regulating synthetic N use Increased availability will increase consumption in the animal feed sector, perhaps at the expense of imported soya, BUT this sector may also decline if the costs of production rise and meat consumption is to be discouraged with higher prices and negative environmental messages. Consistency of availability of product is an often-declared reason for the current low level of use, therefore more availability would potentially drive consumption.

Rotations R&D
The only one of these measures to affect legume production is the crop rotation schemes. 15 1.30%

Production R&D
Technical shortcomings of legume production, this would substantially improve their production and sustainability in EU agriculture. 16 1.40%

Pest resistance R&D
New cultivars resistant to diseases (Conventional and genetic editing) and new cropping systems can facilitate introduction of (grain) legumes in rotations because farmers will have an easier task.  Table A1 shows the frequency of keywords (codes) present in the participants' comments for the seven proposed scenarios. It was perceived that abandonment of the CAP would have a strong negative impact on the competitiveness and sustainability of EU agriculture. Though there was an acknowledgement that further CAP reform might be necessary, respondents advised that the CAP should not be eliminated. Table A1. Frequencies (%) of keywords (codes) mentioned in the participants' comments in relation to the seven policy scenarios described in Table 2. Absence of data refers to frequencies <4%.