The human population is expected to grow from 7.7 billion people at present to 9.1 billion in 2050 (Paillard et al., 2010). This increase will occur mainly in Africa and Asia. While some of humanity, particularly in industrialised countries, must make efforts to rebalance the plant:animal ratio of its diet, ensuring the global supply of animal products remains a major challenge. To meet the growing demand for food, combined with the increase in the purchasing power of densely populated countries, FAO recommendations call for a 70% increase in global meat production (all types of meat) and a 60% increase in milk and egg production (Steinfeld et al., 2006; Paillard et al., 2010), in a context of shrinking land area (e.g. desertification, rising sea level, urbanisation).

If animal production is to increase, it will be necessary to reconsider how these sectors are organised within territories and the issues at stake in order to make room for crops (e.g. cereals) and vegetables (e.g. pulses, other vegetables) for human consumption in the areas that are most favourable for them. How to reorganise the destinations of crops must also be considered. Indeed, a large percentage of crop production from arable land (34% of the world's surface area) is currently used to feed livestock, with a low valorisation of yield, as it takes 2.5-10 kg of plant protein to produce 1 kg of animal protein (Laisse et al., 2018; Mottet et al., 2018). However, it is important to put this last argument, which is often mentioned, into perspective. In fact, farm animals, particularly ruminants in areas with little or no crop production, consume mainly plant products that cannot be consumed directly by humans (e.g. crop residues, co-products from agri-food industries, fodder). Thus, for example, a grassland dairy system can produce up to twice as much human-consumable protein in milk and meat as it does in cereals and protein crops (Laisse et al., 2018). To limit food/feedcompetition, future livestock farming systems must therefore be designed to complement crop production for human consumption and give priority to food co-products and fodder produced in areas that are less or not suitable for crops. Moreover, this debate must be associated with the choices made for non-food use of agricultural land (e.g. energy production, urbanisation) and the need to reduce waste (Paillard et al., 2010; Couturier et al., 2016).

Grassed areas, particularly permanent grasslands, are usually located where crop production is not possible for reasons of accessibility (e.g. mountains) or where yields are too low (e.g. soils of low fertility, harsh climate). They contribute to the development of animal production, which makes plant products from these areas "edible" for humans. The high diversity of temporary and permanent grasslands has made it difficult to acquire reference data for their production characteristics and the factors that determine them. Since the early 2000s, many studies have characterized the performance of certain types of grassland and developed more generic approaches, drawing in particular on concepts of functional ecology (Diaz et al., 1998; Duru et al., 2007; Cruz et al., 2010) (Box 1).

Table

titre du tableau

Permanent, temporary and artificial grasslands Grasslands are agricultural areas whose vegetation is used to produce fodder for harvest and/or for grazing livestock. The term "grasslands" includes permanent, temporary, and artificial grasslands (Allen et al., 2011; Peeters et al., 2014). In mainland France, the distribution of grasslands varies by administrative department, and grassland area has decreased in the main grassland regions since 1950 (figure 1).Permanent grasslandsPermanent grasslands contain perennial or native species in an ecosystem managed over the long term (Allen et al., 2011; Couvreur et al., 2018). They are more complex to manage than other types of grasslands. Among permanent grasslands, semi-natural grasslands (the most diversified) that have been established for more than 10 years are distinguished from more recent grasslands, 5-10 years old or more intensively managed (Peeters et al., 2014). These grasslands contain grasses (Poaceae), legumes (Fabaceae) and other dicotyledons (i.e. "various species" in agronomy), proportions of grasses, legumes and other dicotyledons vary from one grassland to another. The floristic diversity of a permanent grassland in Europe ranges from 15-100 species (Plantureux et al., 1993; Tornambé et al., 2010) with averages of up to 30-60 species (Jeangros and Schmid, 1991).Temporary grasslandsTemporary grasslands contain annual, multiannual or perennial seeded species less than 6 years old (Allen et al., 2011), mainly grasses and legumes. Other dicotyledons are used in mixtures sown for specific purposes (e.g. drought resistance). The floristic diversity of temporary grasslands can reach 12 species in a mixture.Artificial grasslandsArtificial grasslands are areas less than 5 years old sown almost exclusively with fodder legumes. These grasslands are not detailed in this article.Diversified grasslandsDiversified grasslands generally contain several species. They may be temporary (with more than 3 species) or permanent (with more than 35 species). They can be used for harvest and/or for grazing in production systems and are thus considered in the fodder system, the grazing system or, if they also contain woody species, the pastoral system.Grassland dynamics depend on environmental conditions and management practicesThe floristic composition of temporary or permanent grasslands depends on their environmental conditions and management practices. Several studies have shown effects of the environment (e.g. elevation, soil composition) and management practices (e.g. number of cuts, stocking rate) on the botanical or functional composition of vegetation (Hopkins, 1986; Plantureux et al., 1993; Diaz et al., 1998; Klimek et al., 2007; Batary et al., 2010; Michaud et al., 2011; Pierik et al., 2017; Roukos et al., 2017). Assessing the environmental value of grasslands Based on their botanical or taxonomic composition Grassland species or the floristic composition of a grassland can be assessed based on their botanical or taxonomic composition (i.e. by considering vegetation as a set of species), with each species being an entity of its own (Maire, 2009). In this approach, the abundance and dominance of the species present is observed. Based on their functional compositionGrasslands can also be assessed based on the functional composition of their species, in which species are grouped according to their functions (e.g. plants that use the wind for pollination) (Grime, 1977). Assessing the agricultural value of grasslands (production, nutritional value) Increased knowledge about relations among the environment, management practices and vegetation has increased understanding of the main factors involved in predicting production or nutritive value. In particular, this research has been extended to permanent grasslands, whose high diversity (e.g. multiple phenologies) makes it difficult to estimate their agricultural value. To date, prediction models with differing degrees of complexity of the nutritive value and production of permanent grasslands have been developed based on the functional composition,especially the functional types developed by Cruz et al. (2010), temperature and other vegetation components (Duru et al., 2008; Michaud et al., 2014; Pierik et al., 2017). In temporary grasslands, this allows sown species to adapt better to the environment (Simon et al., 1997; Litrico et al., 2016).

Table

Future modes of production will have to remain productive but more respectful of the planet, as agricultural activities impact the environment strongly (Stassart et al., 2012). Indeed, the global livestock sector uses 30% of the land, 32% of the total agricultural water use and rainfall and contributes 18% of greenhouse gas emissions (Herrero etal., 2015). Livestock farming in its intensive form (i.e. a large use of inputs due to an imbalance between the potential of the environment and the stocking rate) also contributes to several other major environmental problems, such as eutrophication, soil degradation, deforestation and loss of biodiversity (Steinfeld etal., 2006). These elements fuel questions about the place of livestock farming in the global landscape of tomorrow and contribute to renewing societal expectations of agriculture (e.g. environment, health risks).

Faced with this situation, the alternatives are either to reduce negative impacts by improving existing systems or to design new systems based on agroecology, to which autonomous low-input production systems (e.g. Alard et al., 2002) and organic agriculture (Tichit and Dumont, 2016) are related. In addition to reducing environmental impacts, agroecology replaces chemical and energy inputs with natural processes, in particular by increasing the diversity of systems and improving the closing of cycles (e.g. minerals, energy, water). Permanent grasslands and, to a lesser extent, temporary grasslands, fit well into this agroecological option. While these alternative systems may have 5-30% lower productivity per hectare or per animal, depending on the situation (De Ponti et al., 2012; Ponisio et al., 2014), they have advantages, such as greater resilience to climate change (Chen and Chappell, 2009) or respect for the environment (Tuomisto et al., 2012). These production systems, which rely greatly on grass and a connection to the soil, are thus interesting avenues for addressing agricultural and environmental issues in a vision of increasing overall agricultural production while considering territorial conditions under which grassland systems would be less productive but would compensate for this with greater environmental value (Huguenin-Elie etal., 2018).

The concept of “ecosystem services”, although little used by the general public, provides a framework for reflection that can highlight benefits of including grasslands in livestock farming systems; these benefits were previously described in part as “multifunctionality” (Béranger and Bonnemaire, 2008; Amiaud and Carrère, 2012). This concept, which appeared in the 1970s, was first defined in the early 2000s as a set of benefits (or advantages) that humans derive from ecosystems (Millennium Ecosystem Assessment, 2005; Fisher and Turner, 2008). The French EFESE study on the assessment of ecosystem services called this definition into question by defining them no longer as benefits in themselves but instead as ecological processes or elements of ecosystem structure from which humans derive benefits (Therond et al., 2017).

Grassland ecosystems provide many benefits from which humans benefit (figure 2). Permanent and temporary grasslands have many and varied environmental benefits: limiting erosion; regulating water flows (e.g. preventing floods, storing water); filtering mineral and organic pollutants; preserving floristic, faunal and microbial biodiversity (especially under extensive management); reducing net greenhouse gas emissions in livestock farming bysequestering nearly as much carbon as forests; and providing landscape benefits, especially permanent grasslands (Mauchamp et al., 2013).

Any benefits to humans from agricultural ecosystems are related to two types of factors: the ecosystem services themselves and the use of “human capital” (e.g. inputs, energy, labour) to provide these benefits. To produce 1 t of fodder, a grassland generally uses less human capital (e.g. fertilisers, pesticides, machinery, working time) than a fodder crop (Couvreur et al., 2018). In this sense, the benefits that humans derive from grasslands are interesting because they are based largely on ecosystem services.

Thus, the concept of ecosystem services can highlight the many positive effects of grasslands. Several studies have attempted to list ecosystem services provided at the landscape scale (Lavorel et al., 2011; Lasseur etal., 2018) or by permanent grasslands (Baumont et al., 2012; Michaud et al., 2013; Galliot et al., 2019) in differing degrees of detail (Therond et al., 2017; Lemaire etal., 2019). Some of them even estimate levels of services in a qualitative or quantitative manner (Therond et al., 2017). Qualitative assessments are admittedly not precise as they are often calculated indirectly, such as for pollination or carbon sequestration, but they do provide benchmarks for society. Quantitative assessments are rarer because they are more complex to perform but are relevant for discussing the fair value of these services.

This knowledge is beginning to be transferred into grassland management tools to better express the potential to provide ecosystem services. Recent grassland typologies have included several ecosystem services based on the floristic composition of grasslands (Launay et al., 2011; Theau et al., 2017; Galliot et al., 2020). A tool at the scale of the production system (DIAM; Farruggia et al., 2012) includes agricultural aspects and the ecosystem services provided by grasslands. This type of tool can show environmental benefits of grassland systems to a public of advisers, farmers and citizens (figure 4).

Table

DIAM quantifies advantages of permanent or diversified grasslands for the environment or for the quality of cheese for farms that produce it. The assessment is based characterising a farm’s permanent grasslands and makes it possible to discuss the potential to manage them in light of environmental issues. PC: protein content; FC: fat content.

Current production methods and health inspection of products provide healthy food that is free from immediate and serious food risks (e.g. zoonoses, poisoning, contamination). Moreover, animal products provide an essential source of certain nutrients in both developed countries, for certain sections of the population (e.g. protein for the elderly), and developing countries, and also contribute to human well-being through the organoleptic and technological quality of the products, mainly in countries with high purchasing power, especially those with a strong culinary tradition. However, several events in recent decades (e.g. mad cow disease crisis, concerns about pesticide use) have led to public mistrust of more insidious, long-term health risks associated with the food consumed, particularly in countries with high purchasing power. Some citizens want to know more about what they eat and where and how their food is produced (Lacroix and Gifford, 2019; Sanchez-Sabate and Sabate, 2019; Wang and Basso, 2019). While the health of consumers must be guaranteed, the health and welfare of farm animals must also be considered. Indeed, these elements are the subject of recurring and influential questions from society, which is increasingly sensitive to animal production methods and their effects on animals (Saatkamp et al., 2019). The concept of “global health” (Koplan et al., 2009) or “single health” (Duru et al., 2016) connects human, animal and environmental health. This global and systemic approach aims to understand humans, animals, the environment and their interrelationships with a view to the health and well-being of these objects (Rockström et al., 2009). Grasslands can be an interesting response to global health issues, whether through their less environmentally damaging management practices, the ecosystem services they provide (Lasseur et al., 2018) or their effects on animal health (Likkesfeldt and Svendsen, 2007; Zeiler et al., 2010; Durand et al., 2013).

Finally, animal welfare and health are debated in Western societies. This even leads to wider reflections on what relationship humans should have with animals, which calls into question modes of agricultural production and the consumption of meat and animal products in general. Grassland systems make sense in such debates, but it is important to objectively assess positive effects of grasslands on these elements of the debate and to recognise them as such (Mee and Boyle, 2020).

Although dairy products have been criticised for their high saturated fat content, which is a risk factor for cardiovascular disease, this criticism should be placed into perspective, in particular by highlighting effects of the composition of plant species in animal rations on the fatty-acid composition of animal products (Martin et al., 2019) (figure 2). Grasslands are rich in omega-3 fatty acids, which are then found in dairy products (Chilliard et al., 2007) and have protective effects on cardiovascular disease risks. The floristic composition, growth conditions (e.g. climate, soil) and harvesting stage of grasslands and how their grass is stored strongly influence the type and quantity of secondary metabolites (Fraisse et al., 2007), some of which are of interest for human health (e.g. phenolic compounds, terpenes, carotenoids, alkaloids, quinones) (Mulligan and Doherty, 2008; Poutaraud et al., 2017). The content of carotenoids, which are precursors of vitamin A, or polyphenols in grass helps limit oxidative phenomena, which cause infectious and inflammatory diseases (Miller et al., 1993; Likkesfeldt and Svendsen, 2007; Farruggia et al., 2008; Durand etal., 2013). Much research is underway to clarify relations between the quality of grass consumed by ruminants, the quality of animal products (e.g. milk, meat, cheese) and human health (Martin et al., 2019).

Although focus lies on balanced nutrition and health of consumers, the health of ruminants is also a concern. Effects of antioxidants on problems related to oxidative stress are well known in several animal species (Miller et al., 1993; Celi, 2010; Niki, 2010), and some effects have been highlighted, such as the ability of vitamin E and selenium supplementation to reduce the risk of mastitis (Zeiler et al., 2010). Effects of other secondary metabolites on animal health have been highlighted, such as antiparasitic effects of condensed tannins (Minh et al., 2003; Hoste etal., 2005). More broadly, certain effects, whether medicinal or toxic, of grassland plant species on animals are fairly well known (Valnet, 1972; Bruneton, 2016), particularly in the veterinary field. Consequently, some farmers have established "medicinal grasslands" to test the health effects of a reportedly beneficial plant or mixture of plants. Species that influence health are offered to the animals during a given period, either alone or as a mixture in grasslands, or even in groves planted on the edges of grasslands. Finally, grazing on grasslands allows animals to express their natural behaviour more frequently by letting them choose their food and move freely.

Grasslands also influence the organoleptic quality of animal products. Relations between the grazing system or type of grassland and the creaminess, colour, flavour or taste of cheese have been established (Jeangros et al., 1997; Farruggia et al., 2008; Coppa etal. , 2012) but need to be better controlled and enhanced. Several mountain PDO (protected designation of origin) cheeses also emphasise this point in their marketing information to increase the recognition and promotion of their products. The Comté PDO has thus developed a rosette of the flavours of its cheeses, in line with the diversity of the grasslands. More anecdotally, hay from diverse grasslands can be used in cooking, as flavourings or in the form of herbal teas ().

Quantifying health effects of diversified grasslands is an important research avenue, as it would constitute additional arguments for using them in production systems. Agronomists, animal scientists, veterinarians, toxicologists, pharmacologists and doctors should be involved in research programmes, in line with the concept of "global or one health". Although the potential "health" role of grasslands (Bareille et al., 2019; Sulpice et al., 2019), particularly diversified grasslands, has been researched, avenues of research remain to be explored. As animal and human health are eminently multifactorial, isolating the specific influence of grasslands is a major experimental challenge. Moreover, knowing that a plant species found in grasslands may benefit health is not sufficient: the grassland must actually contain it, the animal must eat it, and effective doses, molecular forms that trigger an effect on health and potential synergies and antagonisms must be determined. Techniques for measuring molecules of interest to health have made great progress in recent years by lowering detection thresholds, reducing analysis costs and enabling "high-speed" analysis, which opens up interesting prospects. Although the secondary metabolite composition of the main forage species/varieties is known (Nozière et al., 2006; Reynaud et al., 2010; Graulet et al., 2012; Pickworth etal., 2012), that of plants less common in permanent or highly diversified grasslands remains to be determined. Moreover, in an agroforestry approach, the influence of woody plants in grasslands remains largely unknown, despite initial studies on their feed value (Emile et al., 2017). Effects of plants with secondary metabolites on animal or human health also remain to be clarified. More broadly, indicators for measuring these secondary metabolites in animals that have ingested the plants are rarely evaluated on commercial farms due to methodological difficulties or high implementation costs (Huang et al., 2005; Dudonné et al., 2009): it seems important to develop this type of indicator, particularly in a framework of agroecological transition.

The "medicinal" role of plants also remains to be clarified, particularly that of the more diversified plants of permanent mountain grasslands. Although medicinal characteristics of certain plants in humans and animals are well known (Valnet, 1972; Poutaraud et al., 2017), more needs to be known about relations between the ingestion of "medicinal" plants and animal health to specify the doses required and the frequency of intake (Valnet, 1972; Poutaraud et al., 2017).

Concerning the toxicity of certain plants, more detailed analysis is needed of effects of toxic doses of molecules with biological activity, toxic synergies (i.e. the "cocktail effect"), and thus acceptable proportions of these species in grasslands, as well as animals’ ability to avoid ingesting them, in order to prevent harmful effects on animal performance that can sometimes lead to death. Because of the problems they cause, toxic plant species are better known individually and are well documented in veterinary literature. However, the maximum proportions of these species in the biomass of grasslands are not well known.

In addition, it is also important to consider effects of consuming fodder, which helps filter pollutants, on animals, their health and that of humans. These effects may thus depend on grassland's location in a region. Research on contaminants is underway but focuses more on crops (Rychen et al., 2008; Chatelet et al., 2015).

Since 1980, the climate change observed in Europe has increased mean annual temperature (1.7°C higher in 2019 than in the pre-industrial era), the frequency of heat waves and drought periods (EEA, 2019) and the CO₂ content of the air (Soussana et al., 2002; IPCC, 2018). A continuation of these trends seems inevitable at the global scale, with a temperature increase of 1.5°C predicted by 2030 (IPCC, 2018). Europe would therefore be strongly impacted in coming years by global warming and its consequences, which would lead to large variations in crop yields, including those of fodder crops. This impact is unfavourable overall, with a decrease in yields (IPCC, 2018), although some regions may experience some beneficial effects due to a longer annual growing season (Chang et al., 2017). Effects of climate change on grasslands are already being measured, whether on their botanical composition, their yield (Mosimann et al., 2013), particularly without irrigation (Knapp et al., 2001), with decreases of up to 30% depending on the year (Picon-Cochard et al., 2013), or on the functional response of species (Volaire et al., 2009; Volaire etal., 2016). These effects can perturb the fodder balance of farms and the annual distribution of grassland production, often requiring farms to buy fodder, reduce the number of livestock and/or resort to irrigation. Resisting or adapting to climate change will require an evolution or even a revolution in the design of livestock farming systems.

Overall, climate change has a negative impact on the economic health of farms and is often compounded by economic uncertainties that are increasingly common in livestock farming: with the increase in international trade in raw materials and animal products, the global livestock market is sensitive to unpredictable fluctuations in the prices of raw materials and foodstuffs. This uncertainty influences prices of products (e.g. milk, meat) and inputs (e.g. cereals, fertilisers, petroleum), as shown by the fall in products prices on world markets and the milk crisis in 2009, as well as the increase in the prices of raw materials in 2008 (Institut de l'Élevage, 2019).

Farms must therefore become more resilient to better withstand and adapt to climatic hazards and these price fluctuations; one avenue for reflection is to decrease production costs of farms greatly. As feed costs represent a large percentage of production costs (ca. 25% in dairy cattle farming), working with low-input raw materials such as those from grasslands provides room to manoeuvre (Caillaud et al., 2013; Rubin etal., 2017; Devienne etal., 2018).

Climate change has led all stakeholders in the livestock sector (i.e. farmers, advisers, scientists, policy makers) to seek solutions to adapt to the climate's changing trend and increased variability. Some of these solutions have already been implemented, while others are avenues to be evaluated (Pottier et al., 2007). We believe that grasslands can play a decisive role in adaptation to climate change at several scales (figure 2):

(i) At the scale of the grassland, which unlike fodder crops such as maize, can be exploited throughout the year; thus, climatic stress in summer can be compensated by early growth in spring and later growth in autumn or even winter (Pottier et al., 2007), allowing it to take advantage of intra-year climatic opportunities. Indeed, climate change has caused milder autumns and winters in recent years and thus increased grass growth, despite less favourable summers. Moreover, grasslands seem to show plasticity and dynamics in the face of climatic hazards (Tuba and Kaligaric, 2008). The diversity of their plant species makes them less sensitive to climatic hazards than single-species grasslands and thus reduces the critical threshold of climatic stress. Indeed, relying on a mixture of species with different light, water and temperature requirements makes it easier to cope with periods of high climatic hazard (Durand et al., 2016; Hofer et al., 2016). For example, one species may not be able to withstand a drought period, but the other species present can ensure a minimum yield from the field. While interspecific diversity is important, intraspecific genetic diversity can also enhance adaptation potential (Durand, 2016; Meilhac et al., 2019). Finally, establishing species that are more resistant to extreme conditions such as intense droughts is a solution. Several studies are underway to test the forage value of species such as alfalfa, plantain and chicory (Gauly et al., 2013; Delagarde et al., 2014; Lee etal., 2015), in pure or mixed form, at the field scale.

(ii) At the scale of the group of fields, solutions exist to decrease effects of global warming. These solutions can apply to vegetation cover besides grasslands (e.g. planting hedgerows and trees). In addition to the benefits that hedgerows provide (e.g. lower wind or humidity; Liagre, 2007), trees provide shelter for animals and lower high temperatures (Béral et al., 2018). This fodder resource can also be used in ruminant rations: research is underway to quantify effects of trees on herbivore nutrition (Vandermeulen et al., 2018), the palatability and nutritional value of woody plants to animals (Habib et al., 2016; Bhatta et al., 2017) and effects on fodder crops grown under tree canopies (Lima et al., 2019).

iii) At the scale of the fodder system, room to manoeuvre is based on managing fields with diverse environmental conditions and agronomic potentials. Indeed, having fields that are more humid or, conversely, less hydromorphic makes it easier to traverse a period of uncertainty. Having fields of differing productivity during the year is also a factor of flexibility. Although grasslands that produce less biomass and begin growing later in the year are less useful for feeding herds, their flexibility is of interest in a context of uncertainty because their potential use can be delayed by a few weeks (Michaud et al., 2011). If a period of intense drought or rainfall occurs when a low-productivity grassland should have been mown, its agricultural value will remain more stable and decrease less over time than that of a more productive grassland. Finally, the grass supply can also be managed by carrying it over in the field (i.e. keeping a plot in which grass is not consumed, which serves as a reserve) or by using more flexible grazing methods.

Besides the room to manoeuvre for grassland area, the herd can also be managed differently. A one-time reduction in the stock or rearing fewer heifers or replacement ewe lambs would reduce the stock at the farm scale and thus increase flexibility in grassland management in the event of a hazard. In terms of herd organisation, moving from one calving season to two or more would also reduce risk on the farm (Pottier et al., 2007). However, these adjustments in herd management require predicting the balance between optimising grass management and climatic risks.

While grasslands show potential in the face of climate change, their economic advantages are undeniable. Many studies have shown that grassland systems have lower production costs than conventional systems, which gives grassland systems greater resilience (Rubin et al., 2017; Dieulot and Meyer, 2018). Grassland systems have greater economic resilience because forage production costs are lower than those of maize silage and because grazing costs less. In addition, these systems often depend less on inputs and thus on fluctuations in input prices (Delaby and Fiorelli, 2014; O'Donovan and Delaby, 2016). In this sense, a few local initiatives are emerging to try to recognise milk made from grass or hay; however, this type of initiative remains uncommon on the market.

Although grasslands with high floristic diversity have advantages in the face of climate change, many areas of research remain to be clarified. The sustainability and stability of their floristic composition needs to be understood better to guarantee long-term expression of the fitness of this composition. Several studies are underway to quantify effects of repeated climatic hazards on grassland production (Zwicke etal., 2013), responses of grassland species (Zwicke etal., 2015) and their potential adaptation (Volaire et al., 2018) to these hazards. This research should be pursued, particularly to adapt management practices as a function of these hazards.

For temporary grasslands composed of sown species, research remains to be performed on mixtures of species that have differing degrees of drought resistance, mixtures of many grassland species and mixtures of grassland species and cereals to improve resistance to climatic hazards. More generally, grassland systems will have to adapt to climatic hazards: many research avenues remain to be explored on possible adaptations of the functioning of production systems, fodder systems or both. Analysis of the place of grasslands in fodder systems (e.g. productive, dry and wet grasslands) and how grasslands fit into farm management depending on climatic conditions remains to be fine-tuned.

While the economic argument advanced to promote grassland systems should appeal to farmers and facilitate the development of grasslands and grazing (Caillaud et al., 2013), this is not always the case, particularly for dairy systems (figure 2). Indeed, while “grazing” know-how is transmitted from generation to generation in certain regions, grasslands have been replaced in other regions by more productive fodder crops that are more stable over time, which makes it easier to produce a stable amount of milk over time. In the areas where grazing is less common, besides the related technical obstacles (e.g. scattered fields, fields that are too small to produce the grass that the herd requires (Gomas et al., 2008)), grasslands are considered difficult to manage due to the high variability in milk production, which is related to the variability in grass growth and value: dairy farmers have difficulty accepting “sawtooth” milk production because milk is their main source of income (Michaud et al., 2008). They often prefer the better-known, reputedly simpler rations based on maize silage and complementary feed, which provide stable production over time and do not require transitioning. Grazing management considered to require observation and prediction to adapt to climatic conditions and grass growth (Michaud et al., 2008; Frappat et al., 2014).

This concern about including grasslands in dairy production systems is also highlighted by Couvreur et al (2019), who identified four categories of farmers who use grassland: “fulfilled grazers”, “moderates”, “flexible optimisers” and “undecided conservatives”. In these categories, grassland can represent an unquestioned heritage, be part of the system in a moderate way or even be a mechanism for sustainability. For farmers who use grasslands as the central element of the fodder system (i.e. “successful farmers” according to Couvreur et al. (2019)), other studies mention a degree of technicality in grassland management (Darré et al., 2004; Mathieu, 2004). Indeed, livestock farmers' predictions over time, estimates of effects of spring grazing on summer grazing and view of their grasslands shows the importance, technicality and transmission of this knowledge. Finally, although grazing can be a source of daily stress or an additional workload for certain farmers, for others (e.g. “fulfilled grazers”; Couvreur et al., 2019), the search for well-being at work can be a favourable development path toward a production system based on grazing (Lusson et al., 2014). For these farmers, their working and living conditions, as well as their new view of their herds, are arguments in favour of this transition.

Although much research has been performed to improve understanding of grasslands and develop accessible management tools, technical and sociological obstacles remain, with differences among types of farmers. Technical support for these system transitions is necessary (Lusson et al., 2014) and is developed in action-research projects (Devienne et al., 2018; Coquil et al., 2019). Development is also important for transmitting this knowledge and its image. More generally, providing farmers with the best possible support depending on whether or not they decide to enhance the value of grasslands in their systems is a major challenge. To do this, it seems essential to understand farmers' visions (i.e. how they perceive their surroundings and worldview). Cayre et al (2018) highlighted farmers' worldviews according to the production methods they chose. This initial study provides avenues to reflect on the type of technical support to provide given the sociological vision (e.g. favouring a type of grassland on the farm that is consistent with preserving biodiversity) to implement in response to farmers' perception of their environment and, more broadly, in work on agroecology.

Moreover, the image and technical nature of grasslands have yet to be updated to give them a place in current systems, which increasingly use digital and technology. There is a need to raise awareness about the multiple benefits of grasslands, including grasslands with high floristic diversity, which is less common. An interesting example is the flowering grassland competition launched in France in 2007, which brings together ca. 60 organising territories each year that represent 400-500 livestock farmers (De Sainte Marie et al., 2018). It recognises the farmers' technical ability to manage their grasslands by awarding a symbolic prize to those who use diversified permanent grasslands to increase economic and societal performance.

Obstacles to grazing are often mentioned for dairy systems, even though cow-calf systems use the most grassland. For dairy systems, the future trajectories of grassland and grass-based systems can be questioned. Should systems be geared primarily towards dairy production, with fodder such as maize and productive grasslands being the main ration and diversified grasslands being more of a supplement? Or should it be acceptable to adapt the system’s production to the area’s potential rather than vice-versa; in this framework, grasslands, especially diversified grasslands, can find their place at the heart of the feeding system.