Version classiqueVersion mobile

Dynamique et gestion des pelouses calcaires de Haute-Normandie

Thierry Dutoit

Troisième partie. Biodiversité des communautés (phytocénose, banque de graines, pédofaune)/Biodiversity of communities (phytocenosis, seed hanks, pedofauna)

Chapitre second. Permanent seed bank in chalk grasslands under various management regimes: their role in the restoration of species-rich plant communities

Banques de graines permanentes en pelouse calcicole sous divers régimes de gestion: leur rôle dans la restauration de communautés végétales riches en espèces.


Une étude a été réalisée sur quatre faciès d'une pelouse calcicole (ouverte, intensivement pâturée, haute et fourré) afin d'appréhender la banque de graines permanente sous des régimes de gestion responsables du déclin des intérêts écologiques de cet écosystème. Il n'y a pas de réelle ressemblance entre les banques de graines et la végétation associée, excepté dans la pelouse pâturée intensivement. Quand la végétation évolue, les banques de graines apparaissent plus stables et sont surtout dominées par trois espèces (Centaurium erythraea, Origanum vulgare, Hypericum perforatum). Les banques de graines semblent de peu d'utilité pour restaurer des communautées riches en espèces après intensification du pâturage ou abandon. Parce que les processus de recolonisation sont les facteurs critiques, la gestion de pelouses diversifiées doit donc être envisagée pour préserver une communauté et pour améliorer les processus de dispersion d'une pelouse “source” à des pelouses cibles.

Texte intégral


1The accumulation and persistence of seeds in the soil of arable field and pastures of most temperate habitats is a well-known phenomenon. The existence of this reservoir of buried viable seeds has been recognized for a long time and is described as the soil seed bank (Grime, 1979; Thompson & Grime, 1979).

2A striking feature of the seed bank is the general discordance between the species contained in the soil and the associated vegetation. Chippindale & Milton (1934) have shown that the presence of viable seeds of species of former successional stages or plant communities can reflect an earlier stage in vegetation succession history. The loss of the vegetative stage in a plant's life history, even if absent for several decades, does not necessarily show that the species is not present as dormant seed (Moore, 1983). This phenomenon can be applied in nature conservation to re-create former or endangered plant communities (Bakker, 1989).

3The maintenance of species-rich plant communities containing assemblages of species characteristic of mature chalk grassland is usually the aim of conservation management (Arlot & Hesse, 1981; Duffey et al,. 1974). The development of tall grassland and scrub communities, as a consequence of grazing abandonment, is considered undesirable due to both the reduction in floristic diversity and the loss of local or rare species associated with open calcareous grassland communities (Smith, 1980; Bobbink & Willems, 1987). Several studies have looked at the buried viable seeds in other community types in successional series of known age and vegetational composition (Livingstone & Allessio, 1968; Oosting & Humphreys, 1940). However, studies have shown that potential of seed banks for recolonizing large areas of tall grassland or scrub, which have been respectively mown or cleared (Grubb & Key, 1975), was quite weak. In France, there have been few studies on the seed bank in grassland focusing on the conservation of plant communities (Guyot & Massenot, 1950; Guyot, 1965; Delpech, 1969).

4The aim of this study is to investigate the size and the composition of the permanent seed bank beneath different stands of a chalk grassland under various management regimes. In a context of former abandonment, the results are used to assess the feasibility of restoring and managing chalk grassland vegetation from seeds already present in the seed bank.


5The study was conducted in Saint-Adrien nature reserve (Fig.1), 15 km south from Rouen (1°5'3"E, 49°19'22"N), an area of 32 ha of old chalk grassland. The vegetation consists of a mosaic of different age stands of chalk grassland succession on Cretaceous chalk.

Fig. 1: Map showing the study area in the Saint-Adrien nature reserve. Letters indicate the stands referred in the text, [A] open grassland, [B] grazed grassland, [C] tall grassland, [D] scrub. The rectangle indicates the localisation of the study area in France and in the Seine valley

6Four stands were chozen for further investigations on the vegetation and the seed bank (open [A], grazed [B], tall [C] grasslands and scrub [D]). These four stands are on the same slope and have had the same land-use history prior to 1950. They were used as vineyard before 1914 and sheep-grazed from then to approximately 1950 (Liger, 1952). These former agricultural practices on previous arable areas still have some effects on the composition of the chalk grassland (Wells et al, 1976), with the presence of typical weeds from old vineyards (e.g. Allium sphaerocephalum, Muscariracemosum, Thlaspi perfoliatum). Old aerial photographs (1947, 1955, 1963, 1973, 1991) show that management since 1950 is the main difference between the stands.

7The open grassland is a typical species-rich short-herb vegetation naturally grazed by rabbits. The most frequently recorded species are Sesleria albicans, Brachypodium pinnatum, Carex flacca, Festuca cf lemanii, Teucrium chamaedrys and Hieracium pilosella. Flints are locally abundant, and the ground surface is colonized by a dense cover of bryophytes and lichens, probably by selected grazing of the rabbits. Although this grassland is not managed, it can be considered as a good example of a natural extensively grazed species-rich area.

8The second stand, a fenced area of 0.7 ha, has been continuously grazed since 1984 with a mean stocking rate of 4 Texel sheep ha-1 year-1. Changes in the structure and botanical composition of the grassland are the increase of F. cf. lemanii and the decrease of B. pinnatum. Due to the intensive grazing, S. albicans was almost eliminated while species of mesotrophic calcicolous grassland appeared, e.g. Lotus corniculatus, Medicago lupulina and Trisetum flavescens (De Foucault, 1989).

9The vegetation of the third stand corresponds to a tall grassland which has been abandoned since 1950 and is dominated by tussock grasses, e.g. B. pinnatum and S. albicans, with encroachment by some scrub species. Grass litter is abundant, with seedlings of annuals and biennals e.g. Campanula rotundifolia, Galium pumilum and Digitalis lutea occurring on the ground surface.

10The fourth stand corresponds to the third stand except that the vegetation is now dominated by woody phanerophytes e.g. Cornus sanguinea, Crataegus monogyna and Viburnum lantana. The stand can be dated by historical events which changed the patterns of land use (Boullet 1989). Old aerial photographs show that scrub communities appeared after the cessation of grazing in 1950 and are now forty years old.

11More details of the vegetation of the site are given by Liger (1952) and De Foucault & Frileux (1988).


12The soil seed bank was sampled in January 1993 because the highest number of species germinating in chalk grassland is in the winter (Graham & Hutchings, 1988a). The soil cores were 7 cm diameter, 8.5 cm deep, having a volume of 327 cm3, though in practice a full core was sometimes not obtained due to the presence of stones and the shallow rendzina soil. Twenty sampled cores were randomly taken from each stand. This sampling intensity seems sufficient when compared with the recommended minimum soil volume of 1 200 cm3 for recovery of most of the species in grassland seed banks (Roberts, 1981).

13The soils cores were crumbled, spread out on plastic sheets and air-dried in the laboratory. When dry, the soil was sieved through a 1-cm-mesh sieve to remove the coarsest plant fragments and stones. Samples were aggregated into units of five soil cores to form four compound samples from each grassland. Each of these was placed over a 2 cm layer of sterile coarse sand (sterilised at 120° C for several periods of 2 hours) in a 45x30x8cm plastic seed tray. The sixteen trays (4 vegetation types x 4 replicates) were kept in a greenhouse, with a diurnal cycle of 8h/16h darkness/light and temperature regime of 15/20° C. Light was provided by fluorescent tubes (75/85 W). The soil was kept moist. Emerging seedlings were identified and removed or replanted for later identification. Muller (1978) was used when necessary as an identification aid. After two months, each sampling was sprayed with gibberellic acid (1 g/1), following the method of Barralis and Chadoeuf (1980, 1987). The soil was thoroughly stirred before the start of the second germination period which lasted two months, because mixing is known to cause more seeds to germinate (Roberts, 1981).

14A point quadrat method was used to investigate the vegetation in the four stands (Stampfli 1991). For each stand, two permanent 1 m≤ quadrats were marked for observation in homogeneous areas. For each of the eight plots, point frequency records of 100 regularly spaced points (2 cm), consisting of 2 rows at regular distances of 33 cm were made with a needle which was long enough to contact even the tallest plants possible. These quadrats were used to assess the relative contribution of dominant plant species. The flowering plants which occur outside the quadrats were also noted as potential seed sources in order to record the whole plant community. This was undertaken in June 1993 with complementary visits in order to monitor the plants which occured later in the year.


15The species identified in the seed bank and the number of recorded seedlings plus the associated vegetation in the quadrats is listed in Table 1.

16Viable seeds of 38 species were found in the soil samples. This number is quite small compared to the 69 taxa of above-ground species. 84% of the total seedlings recorded in all the samples derive from three species: Centaurium erythraea (46%), Hypericum perforatum (21%) and Origanum vulgare (17%). The first germination period (two months) provided a total of 204 seedlings identified to 23 different species. The second germination period revealed an additional 947 seedlings of 27 species in which 15 were not recorded in the previous germination period.

17Of the 38 identified taxa in the seed bank, 18% of the taxa identified emerged from plots where the species had not been recorded but no examples were found in which large numbers of germinated seeds coincided with the absence of the species from the vegetation. The greater part of these species consisted of short-lived weeds (Anagallis arvensis, Cerastium arvense, Conyza canadensis, Sonchus asper).

Tab. 1: Number of germinated seeds of each taxon recorded pet sample of soil cores of the four stands. (The pourcentage of the total number of each stand seedlings and the percent occurence of all species in quadrats are also given). Species absent from quadrats but observed in each stand of the succession are denoted by (+). Species are arranged Alphabetically; nomenclature follows Clapham et al. (1962)

18The majority of seedlings (82%) were restricted to plots where the species grew in the sampling season. The data about the seed banks from the four stands are summarized in Table 2.

Tab 2: Total number of germinated seeds and taxa recorded for each stand in the seed bank and in the above ground vegetation

19The open grassland has the smallest number of seeds recorded from any of the four stands but the dominance of the three species (C. erythraea, H. perforatum, O. vulgare) is less important in this sample (65%) although Centaurium has its maximum contribution here. In the grazed grassland, except for C. erythraea, H. perforatum and Reseda lutea, all the species present in the seed bank were observed in the quadrats. The seed bank of the tall grassland has the the most important number of seedlings and species. The majority of the characteristic species of this community are represented in the seed bank; Linum catharticum and Muscari sp. could be considered of nature conservation and ecological significance, whilst L. corniculatus and M. lupulina could be considered of agronomic interest. In the scrub communities, all the species (except A. arvensis) were present in the seed bank but absent from the quadrats. No seedlings of scrub species were found in the seed bank.

20The total numbers of seedlings of the three dominant species is quite different in the four stands. Moreover, there is also some disparity in the contribution of each of these species (Fig. 2). Whilst H. perforatum decreased in the scrub stand, C. erythraea and O. vulgare have their highest value in the tall grassland. The proportional contribution of the three dominant species is similar, with a maximum for the tall grassland (90%) and a minimum for the open grassland (65%).

21Sorensen's similarity indices (Cs) have been calculated to compare the composition of seed banks within and between seed banks and quadrats of each vegetation stand (Sorensen 1948 quoted by van der Valk & Davis, 1976; Jefferson & Usher, 1987), where Cs = 2j/(a+b), j is the number of species common to the two samples, a and b arc the total number of species in each sample. The only pattern revealed by Sorensen's index values (Table 3) is that seed banks from different stands are generally more similar than are seed banks and the above-ground vegetation. However, the comparison between the seed bank of the grazed grassland and the vegetation of the sample quadrats and seed bank of the tall grassland show the highest values (respectively 0.61 and 0.74).

Fig. 2: Seedlings and the frequency of the three predominant species for each stand of the study area

22This is due to the presence in both, of some species which persist in the seed bank of the tall grassland but disappear in the above-ground vegetation (e.g. Achillea millefolium, Campanula rotundifolia, Leucanthemum vulgare and Medicago lupulina) after grazing abandonment.

Source of species list Comparison





V egetation/V egetation
















Seed bank with each plant community





Seed bank/Seed bank















Tab. 3: Sorensen s indices comparing the floristic composition of seed banks and plant communities of each stand of chalk grassland ([A] open grassland, [B] grazed grassland, [C] tall grassland, [D] scrub)


23Many studies of seed banks have shown the disparity between the species composition of the seed bank in soil with that of the above-ground vegetation (Chippindale & Milton, 1934; Champness & Morris, 1948; Major & Pyott, 1966; Russi et al, 1992). In this study, there is also an absence of agricultural weeds which often contribute to the lack of correspondence between vegetation and seed bank (Graham & Hutchings, 1988a; Booth & Hutchings, 1990). This could be due to the long period of continuous grazing (Milberg, 1992; Poschlod & Jacket, 1993) since the end of wine cultivation in 1914 (Allorge, 1922).

24In the long term, species richness, species diversity, and density of seed banks generally decreased with successional age (Donelan & Thompson, 1980). Seed banks usually contain a high proportion of early successional (r-selected) species which rely on long-term viability as part of their opportunistic strategy. These species represent the predominant part of the persistent seed banks in our results (C. erythraea, H. perforatum, O. vulgare). Late successional (k-selected) species gain no advantage from such persistence of seeds but must in consequence re-invade sites by new propagules following disturbance (Moore, 1980).

25In an other way, plant richness in the seed bank and in the vegetation are not well correlated especially for ecological significant species. This may suggest the importance of vegetative processes in the maintenance and/or the consolidation of these species (Grime 1979). Similar conclusions are reached by Thompson & Grime (1979) and Graham & Hutchings (1988a).

26In conservation terms, it is then necessary to focus on the different roles of species featuring a transient seed bank which can play a role in the maintenance of settled populations and species featuring a persistent seed bank which can play a role in the regeneration of populations after unsuitable conditions (i.e. grazing intensification or abandonment) have caused species extinctions in the plant communities (Bakker et al., 1991).

27For instance, this investigation suggests that the re-establishment of species-rich open grassland from abandoned tall grassland may not occur only with the restoration of extensive management (Wells, 1980) or even a simple removal of vegetation cover (Hillier, 1990) if ecological significant species are not able to colonize (Graham & Hutchings, 1988b). Moreover, the results of the scrub stand show that there is a risk of increasing the germination of indesirable species after disturbance (e.g. Anagallis arvensis, Cirsium arvense, Turritis glabra), (Grubb & Key, 1975).

28The possibility to utilize the soil seed bank, originating from earlier vegetation stands, to restore species-rich grassland seems then to be quite small (Willems 1988, Berendse et al. 1992) and for successful restoration of a diversified flora, the seed rain must be more important than the soil seed bank (Jefferson & Usher, 1989). Since the remaining species-rich grasslands are few and isolated islands in an otherwise intensively farmed or semi-urban landscape (Green, 1990) and because many grassland plants have low seed dispersability (Verkaar et al., 1983b, Dessaint et al, 1991) the critical factor for successful restoration is seed dispersal and recolonisation processes.

29Moreover, it is known that sheep can transport large amounts of seeds, especially in their fleece (Hillegers, 1983), and that they influence the recolonisation process (Gibson et al, 1987; Mitchley, 1988c). Grazing systems have then to pay special attention to the restoration of an “ecological network”: for example, by the means of an itinerant flock moving between several areas. In this sense, chalk grassland management appears to be a multiscale problem, involving both ecological and agronomic approaches (Alard & Frileux, 1992).

30It must be emphasized that species-rich grasslands must be considered not only for the ecological significance, as regards the occurrence of some plant populations, but also, in a broader view, as a possible source for emigration and recolonization processes in the landscape (Alard et al, 1994). In fact, the maintenance of a species-rich plant community is related to an actual management compatible with the extensive requirement of this community, while the restoration of this kind of community is strongly related to internal or external regeneration potentialities. This leads to a double statute of natural reserves: on the one hand, species-rich and supplying reserves; on the other hand, potentially species-rich and target reserves.


31Due to both the absence of an appropriate seed bank and the fertility of scrub soils (Marrs, 1985), restoration of the formerly species-rich grasslands will probably be a long-lasting process. In this case it may be possible, with appropriate management, to establish a moderately rich grassland sward which in the long term may recruit the ecologically signficant species, given the presence of a nearby source pool and the long-term effects of the seed rain (Jefferson & Usher, 1989). On the other hand, the solution could be the deliberate reintroduction of desirable species by sowing a seed mixture containing the appropriate species in grasslands isolated in an “unsuitable landscape” (Wells 1990, 1991).

Table des illustrations

Légende Fig. 1: Map showing the study area in the Saint-Adrien nature reserve. Letters indicate the stands referred in the text, [A] open grassland, [B] grazed grassland, [C] tall grassland, [D] scrub. The rectangle indicates the localisation of the study area in France and in the Seine valley
Fichier image/jpeg, 44k
Légende Tab. 1: Number of germinated seeds of each taxon recorded pet sample of soil cores of the four stands. (The pourcentage of the total number of each stand seedlings and the percent occurence of all species in quadrats are also given). Species absent from quadrats but observed in each stand of the succession are denoted by (+). Species are arranged Alphabetically; nomenclature follows Clapham et al. (1962)
Fichier image/jpeg, 108k
Légende Tab 2: Total number of germinated seeds and taxa recorded for each stand in the seed bank and in the above ground vegetation
Fichier image/jpeg, 28k
Légende Fig. 2: Seedlings and the frequency of the three predominant species for each stand of the study area
Fichier image/jpeg, 114k

© Presses universitaires de Rouen et du Havre, 1996

Conditions d’utilisation :

Cette publication numérique est issue d’un traitement automatique par reconnaissance optique de caractères.
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search