Introduction

Sown temporary grasslands cover around 2.47 million hectares in France (Huyghe, Vliegher, Gils, & Peeters, 2014). They provide grazed or cut forage for livestock but also various other ecosystemic services such as soil fertility improvement, carbon sequestration and biodiversity shelter (Martin et al., 2020). Sown grasslands are seeded with improved cultivars from various perennial grass and legume species. Most of these species commonly grow as natural populations in permanent and natural grasslands covering 9.8 million hectares in France (Huyghe et al., 2014) and exhibit a diversity of ecotypes adapted to various conditions of soil, climate and use. Only few species, e.g. lucerne (Medicago sativa L.) and Italian ryegrass (Lolium multiflorum Lam.), were bred as landraces before the onset of modern plant breeding. Natural populations and landraces were the starting material to develop continuously improved cultivars adapted to various pedoclimatic conditions and forage systems, e.g. Sampoux et al. (2011) for perennial ryegrass (Lolium perenne L.). Grown as grass–legume binary associations, or as mixtures involving several species (up to ten or more), temporary grasslands can provide a high yield of good quality forage without artificial nitrogen and herbicide inputs (Surault et al., 2024). Furthermore, some grass species have also been bred for small size, high density and slow growth to create turf cultivars improved for either sport or lawn requirements (e.g. (Sampoux et al., 2012)). The breeding of turf-type genotypes of legume species, e.g. white clover (Trifolium repens L.), lucerne (M. sativa), birdsfoot trefoil (Lotus corniculatus L.), has also recently begun. Finally, all grassland and lawn areas contribute to providing pleasant landscapes for recreational activities.

The INRAE multidisciplinary research unit for grasslands and forage species (UR P3F) in Lusignan hosts the Prairies genebank targeted to the conservation of genetic resources of the main grass and legume species sown in France for either forage or turf usage (Figure 1).

Genetic materials maintained by the Prairies genebank

Collections of natural populations from grass and legume species of the Prairies genebank are the outcome of collection trips undertaken by scientists of INRA (Institut National de la Recherche Agronomique, former acronym of INRAE) in France and other countries. Some of these collection trips were undertaken by INRA staff alone and others together with academic research institutes from other countries (e.g. Spain, Portugal, Morocco, Algeria, USA) or breeding companies. Most of the seed-collecting trips were carried out from 1970 to 1990 to gather samples of genetic resources necessary to start or diversify breeding programmes. Since then, the long-term conservation and characterization of collected accessions have been progressively implemented (Prospéri & Sampoux, 2005). A more recent collecting campaign carried out in the southern part of France in 2014 gave the opportunity to collect new accessions from the natural diversity of grass and legume species that were afterwards included in the genebank besides already existing collections.

The genebank also maintains some lucerne landraces, as well as some grass and legume historical cultivars formerly bred by INRAE or other breeders, all removed from national lists. As a matter of fact, in France and other European countries, the commercialization of new cultivars is allowed only after their registration on national lists, which depends on sufficient value for cultivation and use (VCU) and distinctiveness, uniformity and stability (DUS) requirements. After a certain duration (usually ten years), cultivars are removed from national lists unless they have not been outpaced by more recent cultivars (see for examples rules in France at www.geves.fr/variety-seed-expertise/field-forage/registration-of-field-and-forage-varieties-in-the-french-catalogue).

Collection of orchard grass

Other genetic materials

Accessions publicly available from the genebank also include the D. glomerata subsp. glomerata cultivars ‘Floréal’ (1957), ‘Lully’ and ‘Lutetia’, formerly bred by INRA.

Table 1: Number of accessions per species made publicly available by the Prairies genebank in 2024. ^*, The whole set of 852 accessions proposed for notification by France as Contracting Party to the Multilateral System (MLS) of the International Treaty on Plant Genetic Resources for Food and Agriculture (ITPGRFA ) comprises three different subsets: 1) a subset of 531 accessions maintained by INRAE (Prairies genebank) and notified to the MLS in 2014, 2) a subset of 146 accessions maintained by INRAE (Prairies genebank) and 3) a subset of 175 accessions maintained by INRAE (Prairies genebank) and ACVF (society of companies breeding forage and turf species in France), 2) and 3) proposed in 2024 to the French Ministry of Agriculture for notification to the MLS. The subsets of 531 and 175 accessions can already be browsed on the public databases EURISCO, Florilège and Siregal. The subset of 146 accessions is currently in the process of addition to these databases. ^a, Accessions whose occurrence data were forwarded to the Global Biodiversity Information Facility (GBIF) database are a subset of the set of accessions proposed for notification to the MLS of the ITPGRFA. ^b, 52 perennial ryegrass cultivars, 27 Phalaris accessions and 3 x Festulolium cultivars are currently in the process of seed regeneration and will be afterwards added to the public databases Siregal, Florilège and EURISCO.

Species in collection		Number of accessions
Latin name	Vernacular name	Natural pop.	Landraces	Cultivars	EURISCO Florilège Siregal	Proposed to MLS of TIRPAA*	GBIFa
Dactylis glomerata	Orchard grass	211		3	175	214	172
Lolium perenne	Perennial ryegrass	223		54b	193	206	191
Lolium multiflorum	Italian ryegrass	7		2	2	9
Lolium hybridum				1	1	1
Festuca arundinacea	Tall fescue	169		3	128	164
Festuca pratensis	Meadow fescue	11				11
x Festulolium				3b
Festuca rubra aggr.	Red fescues	139		3	142	142	139
Festuca ovina aggr.	Sheep fescues	35			35	35	35
Phleum pratense	Timothy			2	2	2
Phalaris arundinacea	Phalaris	27b
Phalaris aquatica
Medicago sativa	Lucerne	12	17	25	41	48
Trifolium pratense	Red clover	2		3	3	5
Trifolium repens	White clover	6		4	4	10
Onobrychis viciifolia	Sainfoin			2	2	2
Vicia sativa	Vetch			3	3	3
Total		842	17	108	731	852	537

Collection of tall fescue

Other genetic materials

Several x Festulolium accessions created by the INRAE plant breeding unit of Lusignan are maintained. This includes three amphiploid cultivars created after hybridization between F. arundinacea var. glaucescens genotypes and Lolium multiflorum tetraploid genotypes, namely ‘Lueur’ (2007), ‘Luxane’ (2008) and ‘Lusilium’ (2008). This also includes several pools obtained from the backcross of amphiploid (F. arundinacea var. glaucescens × L. multiflorum) material into tetraploid L. multiflorum or Lolium perenne materials.

Accessions publicly available from the genebank also include the F. arundinacea subsp. arundinacea cultivars ‘Gloria’ (1976), ‘Lubrette’ (1981) and ‘Ludical’ (2002) bred by INRA.

Collection of ryegrasses

Other genetic materials

Accessions publicly available from the Prairies genebank also include the perennial ryegrass cultivars ‘Primevère’ (1957) and ‘Clerpin’ and the Italian ryegrass cultivars ‘Lutil’ (1972) and ‘Rina’ (early 1960s), all bred by INRA. Furthermore, a set of 52 perennial ryegrass cultivars from different breeding origins was gathered in 2005 to represent the history of modern breeding in this species for forage and turf usages. This set was used to experimentally assess the genetic improvement for these two usages (Sampoux et al., 2011; Sampoux et al., 2012) and it is currently in the process of seed regeneration in order to make it publicly available.

Collection of fine-leaved fescues

Other genetic materials

Accessions publicly available from the genebank also include the F. rubra subsp. fallax cultivars ‘Ludivine’ (1981) and ‘Milda’ (1975) and the F. rubra subsp. pruinosa cultivar ‘Luciole’ (2002), all formerly bred by INRA.

Collection of lucerne

Other genetic materials in collections

The seed-collecting campaign carried out in 2014 to collect natural populations of grass and legume species in southern France (see previous paragraphs) also enabled the collection of 11 populations of meadow fescue (Festuca pratensis), two populations of red clover (Trifolium pratense L.) and six populations of white clover (Trifolium repens). In 2017, another collecting trip enabled the collection of 27 Phalaris natural populations (Phalaris arundinacea L. and Phalaris aquatica L.) in southern France and Corsica, which are currently in the process of seed regeneration.

Finally, the genebank maintains a small number of cultivars of red and white clovers, thimothy (Phleum pratense L.), vetch (Vicia sativa L.) and sainfoin (Onybrychis viciifolia Scop.), all from INRA breeding.

Practical organization of the Prairies genebank

In 2008, the INRAE research unit UR P3F formally set up the Prairies genebank on its operational site in Lusignan in order to gather genetic resources of forage and turf species previously maintained in several INRAE sites across France. During the following years, substantial investments were achieved to equip the genebank with renewed facilities such as cold rooms for seed storage and a partitioned greenhouse to grow accessions in pollen isolation. Accessions were inventoried and a database was created to record all necessary information (passport data of accessions, quality and quantity of seed lots, characterization and evaluation data) and to trace the activity (seed quality control, seed regenerations, seed distributions, etc.).

For each accession, seeds harvested after a regeneration event are split into three seed lots: an active lot weighing from 200 to 800g which is stored in a cold room (+5°C and 30% hygrometry), a 20 to 30g safety lot stored at -20°C, and a 1 to 2g long-term conservation lot stored at -20°C in a different freezer than the safety lots. The genebank maintains active and safety seed lots only from the most recent regeneration event and long-term seed lots from all successive regeneration events and the initial seed batch introduced in the genebank. Germination tests are performed every three years on active lots, if they are not replaced by lots from a new regeneration event, as well as on safety seed lots if the germination percentage of active seed lots is below the established thresholds. Active and safety seed lots are regenerated when the germination percentage of active seed lots falls below 80%, i.e. every 10 to 15 years.

Species maintained in the genebank are allogamous and more or less self-incompatible. Because of this sexual reproduction biology, natural populations, landraces and cultivars (synthetic varieties) from these species are polygenotypic populations in principle at panmictic equilibrium. Therefore, the regeneration of accessions from these different kinds of genetic materials is performed by intercrossing a number of plants expected as sufficient to avoid genetic drift (i.e. 100 to 250 plants) in pollen isolation from any plant from the same species not belonging to the accession. Different means are implemented to ensure pollen isolation. In field conditions, accessions from grass species are grown in spaced spots within a triticale–rye crop which acts as a pollen barrier. Another way to achieve pollen isolation is the use of pollen-proof (for anemophilous grasses) or insect-proof (for entomophilous legumes) tents and a partitioned greenhouse with 12 confined compartments.

To date, 731 accessions made publicly available for sample delivery (Table 1) can be browsed online on the Siregal website of the Biology and Plant breeding department of INRAE (https://urgi.versailles.inra.fr/siregal), on the Florilege website (https://florilege.arcad-project.org/fr/collections) set up by the plant pillar (BRC4Plants) of the French National Research Infrastructure RARe (Bergheaud et al., 2025) and on the website of the European Search Catalogue for Plant Genetic Resources EURISCO (https://eurisco.ecpgr.org) with the holding institute field filled in as FRA001. These databases will be soon updated with the recently introduced accessions still missing in their records. All the species maintained in the Prairies genebank are listed in Annex I of the International Treaty on Plant Genetic Resources for Food and Agriculture (ITPGRFA). Therefore, 531 of the publicly available accessions were notified in 2014 as a contribution of France to the Multilateral System of the ITPGRFA. An additional set of 321 accessions has been proposed in 2024 for further contribution to the Multilateral System (Table 1). Furthermore, passport data of 537 accessions from the natural diversity of grass species have been uploaded to the Global Biodiversity Information Facility (GBIF) database (https://www.gbif.org/publisher/0b5846f7-20b5-410a-93c5-5de83b522deb).

The Prairies genebank was labelled as a Biological Resource Centre by the network IBiSA (French network of biology, health and agronomy infrastructures) in 2011 and approved as a genebank of the plant pillar of the French National Research Infrastructure RARe in 2021. It was furthermore officially recognized as a curator of plant genetic resources for food and agriculture by the French Ministry of Agriculture in 2020. The Prairies genebank manages the French network for conservation of forage and turf species genetic resources, which includes INRAE and private companies actively breeding these species in France. In 2023, the Prairies genebank as well as the biotechnology and chemistry facilities of UR P3F were acknowledged for ISO9001:2015 certification.

Over the five last years, the Prairies genebank has distributed around 1,300 seed samples to various recipients, which were, by decreasing order of magnitude, French and foreign academic research institutions, French and foreign plant breeding companies, farmers and hobby-growers.

Uniqueness of the collections

The collections of natural populations maintained by the Prairies genebank originate from areas not or little represented in other genebanks. Notably, most of the grass and lucerne populations from northern Africa kept by the genebank are not maintained in their country of origin, and sometimes do not grow anymore in their site of origin, because of changes in climate and land use, excessive grazing or replacement by cultivars from modern breeding. Recent collections of natural populations undertaken by the INRAE research unit UR P3F in France and other European countries have confirmed that the in situ conservation of these populations is often threatened by the continuous regression of natural and permanent grasslands, climate change, and the corruption of their genetic integrity by overseeding with unrelated cultivars. The ex situ collections of grass and legume natural populations maintained by the Prairies genebank thus make a unique, although non-exhaustive, contribution to the preservation of the genetic diversity of these species. Furthermore, the Prairies genebank is the only repository of INRAE cultivars removed from national lists. However, the polygenotypic nature and the allogamous sexual reproduction of forage and turf species make the seed regeneration of accessions costly, even though the seed market for these species, and consequently the means for genetic resources conservation, are relatively limited. International networks of genebanks, like the Forages Working Group of the European Cooperative Programme for Plant Genetic Resources (ECPGR), are thus essential to share conservation efforts in these species. In the coming years, the Prairies genebank will aim to achieve a better sampling of the French heritage for genetic resources of forage and turf species in its collections. Present collections of orchard grass and tall fescue natural populations only sample a limited part of the French territory, and it will be aimed to complete the collections through new sampling campaigns in regions not visited so far. It will also be considered to sample the French natural diversity of legume species, such as white and red clover. Another important goal will be to work towards a more representative collection of cultivars, bred by INRAE and also other breeders, that were grown in France since the 1950s for all forage and turf species that had significant use.

Valorization of the collections

Initially set up to have genetic resources available to start breeding programmes, collections of natural populations may provide relevant materials to study the environmental distribution of inter- and intra-specific plant diversity. In this respect, 167 scientific papers from various international institutions have to date used the occurrence data of grass species forwarded by the Prairies genebank to the GBIF database. The main asset of these collections is their capacity to supplement occurrence data with phenotypic and genomic information already existing or newly generated. Earlier in this paper, we reported the use of phenotypic data to document adaptive differentiation in fine-leaved fescue taxa (Sampoux & Huyghe, 2009) and adaptive variability within perennial ryegrass (Balfourier & Charmet, 1991). More recently, the FACCE-JPI ERA-NET+ (https://www.faccejpi.net/en/faccejpi/actions/core-theme-1/facce-era-net-plus-on-climate-smart-agriculture.htm) project GrassLandscape gathered INRAE, IPK (Leibniz Institut für Pflanzengenetik und Kulturpflanzenforschung - Institute of Plant Genetics and Crop Plant Research, Germany), IBERS (Institute of Biological, Environmental and Rural Sciences, Wales, United Kingdom) and ILVO (Instituut voor Landbouw-, Visserij- en Voedingsonderzoek - Flanders Research Institute for Agriculture, Fisheries and Food, Belgium) to implement phylogeographic and landscape genomics analyses on a set of 470 accessions from the natural diversity of perennial ryegrass provided by 15 genebanks from European countries and the USDA. High throughput genotyping of these accessions allowed for the reconstruction of the expansion history of perennial ryegrass across Europe (Blanco-Pastor et al., 2019). Then, a multivariate landscape genomics analysis retrieved 633 potentially adaptive loci associated with either winter cold or summer drought and heat stresses and pointed phenotypic traits putatively conferring adaptation to these stresses (Blanco-Pastor et al., 2020). The adaptive relevance of these traits was further confirmed by a functional ecology approach (Keep et al., 2021). Results of the project GrassLandscape were then used to set up a European core collection of perennial ryegrass natural populations within the frame of an ECPGR Activity (Keep et al., 2020; Sampoux & Willner, 2022). Adaptive phenotypic and genomic variations revealed by the project GrassLandscape have furthermore been used to assess the future of perennial ryegrass local populations exposed to climate change by implementing genomic offset analyses (Pégard, Fitzpatrick, & Sampoux, 2023). For other forage species than perennial ryegrass, collections maintained in European genebanks, and beyond, could likely provide sufficiently wide sets of natural populations to set up environmental niche models of species (Sampoux & Badeau, 2009) and reveal adaptive phenotypic and genomic diversities within species (see for example (Boller, Posselt, & Veronesi, 2010) for genetic resources and diversity in forage species). This information could be used to implement predictive analyses, such as genomic offset analyses, to foresee spatial shifts of adaptive diversity required for adaptation to ongoing changes in climate and other environmental factors (Aitken, Jordan, & Tumas, 2024; Fitzpatrick, Chhatre, Soolanayakanahally, & Keller, 2021; Gougherty, Keller, & Fitzpatrick, 2021; Hung et al., 2023; Zou, Yang, Zhang, & Xu, 2024). On this basis, strategies for in situ conservation could be developed, taking into account ongoing environmental changes on a European scale. For most forage and turf species that can spontaneously persist and evolve in long-duration grasslands, in situ conservation planned in this way would be the cheapest and most sustainable means to maintain valuable functional and adaptive genetic diversity in the long term. Incorporating adaption to future environmental changes across territories in in situ conservation programmes would contribute to maintaining an economically viable forage production on involved grasslands that would thus help the sustainability of these programmes.

Results of the project Grasslandscape also showed that the diversity of perennial ryegrass forage cultivars currently grown worldwide represents only a small part of the natural diversity of the species (Blanco-Pastor et al., 2019). Moreover, the H2020 European project EUCLEG (www.eucleg.eu) led by UR P3F studied a worldwide set of lucerne landraces and cultivars, which revealed clear phenotypic and genomic differentiation related to geographic origin and evidence that the cultivar diversity grown in Europe represents a limited part of the whole genetic diversity of the species (Pégard et al., 2023). These results suggest that natural diversity, landraces and old cultivars of forage and turf species may still provide potentially useful genetic variability to adapt grown cultivars to upcoming challenges. Such genetic resources may be useful sources to adapt species to new combinations of climate constraints (Blanco-Pastor et al., 2020) and to improve their tolerance or resistance to biotic stresses (Annicchiarico, Barrett, Brummer, Julier, & Marshall, 2015; Sampoux & Badeau, 2009). They could also be valuable sources to develop cultivars adapted to cultivation in species mixtures. During the last five decades, forage and turf species have intensively been bred for performances in pure stands and it may be assumed that phenotypes more adapted to interactions prevailing in species mixtures (Litrico & Violle, 2015; Sampoux, Giraud, & Litrico, 2020) could be recovered from natural diversity or genetic material that experienced only limited breeding (e.g. old cultivars). Furthermore, with the transition to agroecological practices, some forage species have emerged as service crops. They could be used as intermediary cover crops, e.g. Italian ryegrass, or as perennial living mulches, e.g. lucerne, red and white clovers, sainfoin (Onobrychis viciifolia Scop.) and birdsfoot trefoil (Lotus corniculatus). Cultivars bred for forage usage are probably not the best-adapted materials for these new uses (El-Ghazzal, Louarn, Julier, & Barillot, 2024) and genetic resources may help to breed more suitable phenotypes.

However, introgressing allele diversity from raw genetic resources to elite cultivars may have become more and more challenging as continuously improved cultivars have been released. This may especially be the case for the allogamous and partially auto-incompatible forage and turf species, for which it can be assumed that the alleviation of the genetic load has been a substantial part of the genetic improvement (Annicchiarico et al., 2015; Kimbeng & Bingham, 1998). Genotyping based on genome sequencing should be expected to help the efficient introgression of desirable multigenic features from genetic resources to elite germplasm within a reasonable timespan. While investigations such as the ones of the project GrassLandscape have proven that collections of natural populations can be used to reveal the signature of natural selection (loci whose allele frequency variation departs from neutral expectations), it could be expected that collections of cultivars punctuating the history of modern breeding may reveal the signature of artificial breeding (see for example (Welcker et al., 2022) for such approach in maize). Marker-assisted selection methods could then possibly be designed to optimize favourable allele frequencies at both loci involved in desirable natural adaptations and loci that determine agronomic performances in the cultivars of forage and turf species, essentially bred as synthetics. Efforts to broaden, phenotype and genotype the collections of the Prairies genebank are to be conceived in order to contribute in these directions.

Author contributions

The manuscript was written by JPS with contributions from BJ and MG. All authors reviewed and commented the manuscript and approved the final version.

Conflict of interest statement

The authors have no conflicts of interest to report.