Thèse-EliseBuisson-Chapitre4

Elise Buisson

Informations

• Publié par: Shuwyor
• Nombre de lectures: 29
• Langue: English

Extrait

Interchapter 3-4 Seeding experiment in La Crau In the previous chapters, we described the impact of strong exogenous disturbances on Mediterranean herbaceous plant communities that have evolved with regular endogenous disturbances for centuries. We show that these communities are degraded and need to be restored. In order to plant restoration adequately, it is necessary to find out what irreversibility thresholds have been passed. We know that reference grassland seeds hardly disperse into degraded grasslands, but when they do get there, do they find appropriate environmental conditions to germinate? In other words, we need to know if grassland species can be re-introduced by sowing.    In order to answer this question, we carried out some experiments on the germination potential of our two species,Thymus vulgaris and Brachypodium retusum, in the lab and on their emergence potential in the field. I now briefly discuss our experiments and results.

First, we carried out standard germination tests in growth chamber (20°C, 16 hours of day light) as well as germination test with various scarification treatments (cold, heat). As the two species (Thymus vulgaris andBrachypodium retusum) germinated relatively well (>50%), we carried out emergence tests in the field. Sowing seeds to restore grasslands is a technique widely used to enhance species richness, accelerate succession or fully restore grasslands (Wells 1990; Stevenson et al. 1995; Hutchings & Booth 1996b; Jones & Hayes 1999; Kailova 2000; van der Putten et al. 2000; Warren et al. 2001; Pakeman et al. 2002; Pywell et al. 2002; Hitchmough et al. 2003; Walker et al. 2004). Sowing species-rich mixtures can give good results in northern Europe (Wells 1990; Stevenson et al. 1995; Pakeman et al. 2002) and/or in presence of arable weeds (van der Putten et al. 2000). Sowing success depends on the seeds sown (Hutchings & Booth 1996b), and on concurrent vegetation management (Jones & Hayes 1999; Warren et al. 2001; Pywell et al. 2002). We thus carried out our sowing experiment with the same combined treatments than for transplanted seedlings. We sowedThymus vulgaris   andBrachypodium retusum the abandoned melon field in Oct. 2002 on and 2003. We show that emergence of both species is extremely low. Contrary to our experiment, another experiment in a Mediterranean

climate (Spain) sowing seed mixtures within arable weeds have succeeded in changing the initial vegetation succession (Van der Putten et al. 2000). To try to understand low field emergence, we then carried out more germination and emergence experiments. The seeds of the two species were subjected to various germination tests in controlled-temperature chamber: germination tests in solutions of soil from the study sites to introduce mycorrhizae and other micro-organisms; and a germination test with addition of phosphorus. The seeds ofThymus vulgaris and Brachypodium retusum were also subjected to various greenhouse emergence tests to further determine the role of phosphorus and mycorrhizae. A factorial experiment was set up with three combined treatments on four soil types: application of iron oxide to limit phosphate ion availability (Bastin et al. 1999); of phosphorus; and of fungicide. Unfortunately, results of both experiments do not show any clear patterns.

The reasons of low field emergence of our species in La Crau remain unclear.Thymus vulgarismay not be able to germinate on soils with high concentrations of phosphorus as suggested by some germination tests. As field emergence is low and because we do not yet understand the irreversibility thresholds that have been passed and how to fix them, we need to find out grassland species can be re-introduced by transplanting, if they can survive and/or establish. We thus carried out experiments described in the following chapter.   

 

Chapter 4

Restoration of La Crau Dry grassland recovery: determining the limiting processes by assessing the effects of grazing, stone cover and plant neighbour interactions on two planted perennial plants   Elise Buisson 1 , Emmanuel Corcket 2 , Alain Peeters 3 , Thierry Dutoit 4  

1 Institute of Mediterranean Ecology and Paleoecology, UMR/CNRS 6116 University P. Cézanne, FST Saint Jérôme, case 462 13397 Marseille Cedex 20 France

2 UMR INRA BIOdiversité, Gènes, ECOsystèmes, Université Bordeaux 1, Avenue des Facultés, 33405 Talence Cedex - France

3 Laboratoire d’Ecologie des Prairies,    Université Catholique de Louvain, Place Croix du Sud, 2 bld 11 1348 Louvain-La-Neuve, Belgium

4 UMR INRA-UAPV 406, Écologie des Invertébrés Site Agroparc 84914 Avignon France

Article in prep.

Abstract The ecological restoration of dry herbaceous ecosystems can be initiated by improving habitat quality and by re-introducing perennial species. To improve habitat quality, various components of the ecosystem can be manipulated, and their choice depends on our understanding of the functioning of both intact and degraded ecosystems. In order to better understand intact grassland functioning and to diagnose degraded ecosystem dysfunctions, we set up an experiment in the steppe of La Crau (southern France). To discriminate the importance of various biotic and abiotic factors in the establishment of plant species, we planted the two perennial steppe speciesThymus vulgarisand   Brachypodium retusum  atthree sites, one patch of undisturbed steppe and two abandoned fields with different soil fertilities, in a multi-factorial experiment: we manipulated stone coverage (low or 50% cover), grazing (grazed or ungrazed) and plant neighbour interactions (neighbour removal or intact neighbours). The steppe is a good environment for the early survival (Thymus: 94%; Brachypodium93%) of stress-tolerant species adapted to low soil: nutrient status, grazing and drought. However, slow growth resulting from low soil nutrients does not allow plants to establish well enough to cope with a severe summer drought (late survival:Thymus: 9%; Brachypodium: 8%). On the fields, target species survive well and make the most of the available nutrients to grow well (Thymus: 93%;Brachypodium: 80%). However, the combination of dense weed competition and heavy grazing exacerbated by the absence of large stones, does not allow plants to establish well enough to cope with a severe summer drought. Stones provide some protection against grazing and are potentially very valuable in protecting target species from average drought.   This study shows that restoring the steppe will be labour-intensive. Germination in the field being extremely low, we suggest transplanting Thymus vulgarisandBrachypodium retusumas a restoration technique. The ideal combination of treatments would be to exclude sheep grazing during the first spring to allow seedlings to establish well, as well as to restore the original 50% stone cover and to reduce arable weed competition.  

Introduction  Grasslands were once widespread species-rich ecosystems, representing ~25% of the world vegetation cover (Shantz 1954;

Henwood 1998) and were exceptionally diverse in flora and fauna (Henwood 1998; Taylor 1998). They have evolved with a number of disturbances: grazing, drought and often human management (Henwood 1998). In Europe, as in many other parts of the world, grasslands were established and maintained under human management and thus result from many centuries of interaction between nature and traditional land- uses (Hillier et al. 1990; Pärtel t e h ig Bl naMc& acCr nek0002 .)ta .l1 c    n 9 e 9 u ; t 9 Since the beginning of the 20 ry, grasslands have decreased drastically world-wide (Jacobs et al. 1999) due to land-use intensification (development, agriculture) and to altered or abandoned disturbance regimes and traditional land-uses (Wilmanns 1997); the habitat quality of those remaining is reduced by degradation and fragmentation (Saunders et al. 1991; Harrison & Bruna 1999). Habitat quality on degraded grasslands can sometimes be improved simply by removing the causes of degradation or re-introducing disturbance regimes, particularly if the disturbance intensity is moderate, e.g. traditional land-use abandonment (Muller et al. 1998). The re- introduction of mowing (Hansson & Hakan 2000; Moog et al. 2002), grazing (Gibson & Brown 1992; Miller et al. 1999; Sternberg et al. 2000; Moog et al. 2002), itinerant grazing (Fisher et al. 1996; Poschlod & WallisDeVries 2002), tree cutting (Pärtel, et al. 1999), soil disturbance (Dolman & Sutherland 1992) or mulching (Moog et al. 2002) have helped trigger resilience after disturbance. However, when disturbance intensity is high, recovery through natural processes appears to be delayed indefinitely (SER 2004) because irreversibility thresholds have been crossed (Aronson et al. 1993). Grasslands disturbed by long-term abandonment or by agricultural intensification (ploughing, fertilisation, drainage, liming, over grazing) can be so impacted that their recovery is extremely slow or even unlikely (Gibson & Brown 1991). This may be due to the lack of propagules in the seed bank (Thompson & Grime 1979; Graham & Hutchings1988a,b; Dutoit & Alard, 1995; Bakker et al. 1996a b; Hutchings & Stewart 2002; Wilson 2002; Römermann et al. 2005) and in the landscape (Poschlod et al. 1998) and to low long-distance seed dispersal of herbaceous species (Verkaar et al. 1983; Wilson 2002; Buisson & Dutoit 2004). This may also be due to degraded habitat quality, such as increased soil fertility (Gough & Marrs 1990; Marrs 2002) and/or the presence of unwanted species like arable weeds or exotic species (Davy 2002; Hutchings & Stewart 2002; Wilson 2002). These degraded grasslands with low species richness, few characteristic species and altered ecosystem functions, need to be restored, especially in arid or semiarid environment where ecological processes are slow (Blondel & Aronson 1999). Restoration thus consists

in correcting multiple changes in various components of the ecosystem (SER 2004). We selected an 11 500 ha semiarid grassland in south-eastern France that was representative of many other grasslands in the Mediterranean Basin (Malo & Peco 1995) and that was in need of restoration (Buisson et al. 2004). Our site, the steppe of La Crau, offers an original association of plant species (Devaux et al. 1983) that has evolved with grazing and both edaphic and climatic aridity, and which is haven to several interesting animal species, such as the rare endemic grasshopperPrionotropis hystrix rhodanica, and the birdsPterocles alchata(pin-tailed sandgrouse) andFalco naumanni(lesser kestrel). Part of this steppe has been degraded by cultivation, including ploughing, stone removal and fertilisation between 1960 and 1985. Twenty to forty years after abandonment, the abiotic conditions of the abandoned fields are still altered and the plant community is composed of arable weeds instead of steppe species although the disturbance regime, grazing, was re-established on all fields just after abandonment (Römermann et al. 2005). Remnant patches of steppe have a limited role in the colonisation of adjacent abandoned fields as seed production on the steppe is low due to overgrazing, and seed dispersal limited (Buisson & Dutoit 2004; Chapter 4). Also, seed germination of steppe species on abandoned fields seems extremely limited (pers. obs.; interchapter 3-4). In dry environments, ecological restoration can be initiated i) by enhancing soil and micro-environmental conditions and ii) by re- introducing some perennial species that will improve habitats by capturing soil particles, nutrients and microorganisms, thus facilitating the re-establishment of other species   (Whisenant et al. 1995; Le Floc'h et al. 1999). We propose to reintroduce the two dominant perennial species of the steppe,Thymus vulgarisand retusum Brachypodium(Devaux et al. 1983), to degraded abandoned fields as a first step towards restoration, as they may improve habitat quality (Bergkamp 1998; Buisson & Dutoit 2004). Target species need to be transplanted under a combination of treatments because although sowing may result in high establishment in northern Europe (Wells 1990; Stevenson et al. 1995; Pakeman et al. 2002), simply sowing target species, with or without re- introducing disturbance regimes has not always proved successful in temperate climates (Hutchings & Booth 1996b; Coulson et al. 2001; Warren et al. 2001; Marriott et al. 2002) and even more rarely in drier climates (Jusaitis et al. 2004, pers. obs. see interchapter 4-5). We thus conducted a multi-factorial experiment to test a combination of treatments in order to better understand intact steppe functioning and to diagnose degraded ecosystem dysfunctions by discriminating the importance of various biotic and abiotic factors in the establishment of

plant species. We tested 1) resource availability by conducting the experiment at three sites on a soil fertility gradient and by manipulating the micro-climate through various stone coverage; and 2) biotic interactions by manipulating two grazing intensities (grazed/ungrazed) and by removing plant neighbours that define competition/facilitation relationships.  We tested whether an elevated nutrient status benefits or inhibits restoration. Soil nutrient status which is elevated as a result of cultivation reduces species diversity and influences plant growth (Marrs 2002). In order to restore plant communities, soil nutrients sometimes have to be reduced by grazing, cropping, burning or topsoil removal to remove nutrients, or by reducing nutrient availability with ferric sulphate amendments (Janssens et al. 1998; Marrs 2002). Most of these treatments are intense and expensive restoration techniques that need careful, judicious consideration. We investigated the role of stones and the potential role of stone cover restoration in creating safe-microsites for seedlings to establish. Large stones, that cover ~50% of the ground in the steppe and that were removed for cultivation, may create a micro-climate and allow root biomass development (Bourrelly 1984). Stones have been shown to influence positively the surroundings of seedlings in dry environments i) by increasing shade and thus reducing evaporation (Fowler 1988); ii) by allowing water condensation (especially large stones) thus increasing soil moisture and microbial activity under stones (Lavah & Steinberger, 2001); iii) by enhancing soil moisture and protecting seedlings from grazing (Noy-Meir 2001); iv) and by favouring perennials over annuals (Blumler 1992). We used two additional combined treatments to help target species establish: grazing and plant neighbour removal. In areas invaded by competitive plant species, restoration of ecosystems implies reducing their vigour because they may have a competitive advantage over the transplanted target species (Davy 2002; Hutchings & Stewart 2002; Wilson 2002). This can be done by reducing the seed bank (tilling, topsoil removal) before transplantation, or/and by reducing plant neighbour cover (herbicide, removal of above-ground biomass) (Davy 2002; Hutchings & Stewart 2002; Wilson 2002). This experiment, carried out on the steppe and on fields, early successional stages, is also an opportunity to test the theoretical models describing mechanisms of species coexistence in grassland succession (Alard 2002; Alard & Poudevigne 2002; Chabrerie 2004). They state that community structure evolves from overlapping niches with competition processes to partitioned niches with facilitation processes (Alard 2002; Alard & Poudevigne 2002; Chabrerie 2004). Grazing has also

succeeded in providing desirable species with suitable safe sites to germinate and establish (Fisher et al. 1996; Miller et al. 1999; Sternberg et al. 2000). The goals of this experiment were 1) to assess the importance of the various factors in the establishment of two perennial steppe species Thymus vulgarisand Brachypodium retusumin the steppe; 2) to use these results in order to advise a combination of treatments that will maximise the survival and growth of planted-target-species seedlings on degraded abandoned fields with an objective of ecological restoration.

Methods Site description We conducted the experiment at three sites on the sheepfold land of Peau de Meau (43°33’E 4°50’N) in the steppe of La Crau in south- eastern France (Fig. 7). Average annual rainfall is 500 mm with high inter-annual variability. Elevation is 10 m, there is no slope, and the soil is calcareous < 40 cm deep over an impermeable, conglomerate bedrock. The first site (control site) is a remnant patch of steppe (35 ha) that has not been cultivated. The dominant vegetation isThymus vulgaris and Brachypodium retusum, and other common species areAira cupaniana, Euphorbia exigua, Vulpia bromoides, Sideretis romana, Linum trigynum and Galium parisiense.The second site is an abandoned cereal field (12 ha) that was   cultivated with cereals and alfalfa between 1960 and 1966. The disturbance intensity on this site was intermediate and involved surface ploughing, fertilising (phosphorus and potassium) and some stone removal. The dominant vegetation now is Aegilops ovata, Trifolium stellatum, T. scabrum andT. subterraneum,Medicago minima, Crepis sancta and Plantago bellardii. The third site is an abandoned melon field (5 ha) that was cultivated with melons in 1971 and with cereals and alfalfa in 1972. The disturbance intensity on this site was great: it was ploughed, watered, fertilised, treated and stones were removed. The dominant vegetation now is Lobularia maritima, Diplotaxis tenuifolia, Bromus rubens, B. hordeaceusandGalactites tomentosa. All sites were and are grazed by itinerant sheep flocks before and after cultivation, continuously from February to June. Stocking rates are 3.6 animals/ha on the steppe and 5.6 animals/ha on the fields because of different land ownership. The abandoned melon field is also more used by rabbits. True field replicates could not be found as cultivation periods, dates of abandonment and field locations are particular to each field (Römermann et al. 2005); however, melon cultivation is representative of disturbances in the area (20% of the original steppe area) and in the Mediterranean

Basin. The abandoned cereal field, although not representative of what happens at a larger scale, is an interesting intermediate which, we hope, will help to understand the functioning of the degraded ecosystem.

3 m

Neighbour Neighbour removal intact

Main experimental design At each site, we randomly installed twelve experimental units in autumn 2002, composed of two 3× 1.5 m plots a few meters away from each other to avoid edge effects: one fenced plot (ungrazed treatment) and one non-fenced plot (grazed treatment) (Fig. 20). In each plot, we experimentally manipulated stone cover and plant interactions in a split- s   Grazed Un ed graz     ~50% stone   cover     Neighbour Neighbour   intact removal                               Fig. 20. Experimental design layout. At each site, we randomly set 12 paired 3× 1.5 m plots (experimental units or replicates), one fenced, one not. Each plot was split in two 1.5 ×1.5 m split plots: low stone cover (white) and ~50% stone cover (shaded). Each subplot was split in two 0.75×split split plots (separated by dashed line) and was allocated1.5 m to a neighbour treatment. Transplanted plants are indicated by triangles (Thymus vulgaris) and stars (Brachypodium retusum). Dashed squares to correspond to plant neighbour removal. To test the effects of stones in the steppe, we removed the large stones from one of each 1.5×1.5 m subplots. To test the effects of stone cover

low stone cover 1.5 m 3 m

 

restoration on the abandoned fields, we restored the 50% stone cover in one of each 1.5×1.5 m subplots in autumn 2002. To test the effect of plant neighbour interactions on planted seedlings, we hand-pulled all small seedlings and clipped all larger plants to the ground within a 25-cm diameter area surrounding oneT. vulgaris and oneB. retusum; we chose a 25-cm diameter as Davies et al. (1999) showed that some grassland plants responded to neighbour removal in areas >15-cm diameter. There were also oneT. vulgaris and oneB. retusumintact neighbours on each subplot. We removed neighbourswith in fall 2002, once in spring 2003 and once in winter 2003-2004. Transplanting  In October 2002, we transplanted oneT. vulgaris and oneB. retusum   seedling into each grazing× stone× neighbour removal treatment on each of the 12 experimental units at each site (Fig. 20). These seedlings were grown outdoors in individual containers for nine months, in 2/3 sand - 1/3 soil from each site, watered as needed and not fertilised. The seedlings were watered once when outplanted. Seedlings which died within the first month were replaced.

Data collection We monitored plantedT. vulgaris andB. retusum survival and growth over 1.5 years, which included one complete growth season and two grazing cycles. In June 2004, we collected the above and below-ground biomass of all the plants. Above and below-ground biomass were washed, dried (70°C) until weight was constant and weighed. Above- ground biomass were then ground and pooled when necessary to obtain for each combination of the grazing×stone treatments at least 3 batches substantial enough to be analysed for K 2 O, P 2 O 5 , NaO, MgO, CaO at the Laboratory of Ecology of Louvain, Belgium (Baize 2000).

Community and habitat description The above-ground biomass of the plant community was assessed on each field by cutting all plants to the ground in four 50×50 cm quadrats in the vicinity of each experimental unit, in April 2004. Before cutting, mean plant height was estimated in each quadrat. Biomass was dried (70°C) until weight was constant and weighed. The composition of the plant community was assessed by visually estimating all species % cover in one 40× cm quadrat on each combination of the grazing 40×  

stone treatments for each experimental units and species richness was calculated. In June 2004, we collected soil samples inside (12) and outside (12) the exclosure at each site and analysed them for pH, Kjeldahl nitrogen, C, C/N, O.M., K 2 O, P 2 O 5 , NaO, MgO and CaO at the Laboratory of Ecology of Louvain, Belgium (Baize 2000). Soil moisture was measured after a dry period and after a rainy period: we took twelve soil samples at each site, sieved them on 2 mm mesh, and dried known weights (Weight humid ) at 105°C before re-weighing them (Weight dry ). Residual humidity (RH) was calculated using the following formula: RH = [(Weight humid   - Weight dry )*100]/ Weight humid (Baize 2000). We compiled weather data (precipitation, air temperature) from the closest weather station (Istre 13047001). The site had slightly higher temperatures in 2003 than during the two previous years (15.7 ± 0.4°C in 2003 vs. 15.3 ± 0.2°C in 2001-02). The site received 683 mm of rainfall, which is less than in 2002 (763 mm) and more than in 2001 (431 mm, low autumn rainfall). June temperature were much higher in 2003 (T max: 31.6 ± 0.6°C in 2003 vs. 27.9 ± 0.7°C in 2001-02; T mean: 25.5 ± 0.4°C in 2003 vs. 22.1 ± 0.5°C in 2001-02) while spring rainfall remain about the same from 2001 to 2003. Statistical analyses In order to characterise each site, we compared community and habitat description data using one-way ANOVA with site as a categorical factor. Response variables included soil and plant mineral analyses (%N, C, K 2 O, P 2 O 5 , NaO, CaO and MgO), soil residual humidity, and plant community biomass, mean plant height and species richness. In order to analyse survival and biomass data on target plant, we carried out analyses on six separate models, one for each site and for each target species. Only results on survival and total biomass 1.5 years after planting (June 2004) are presented here as analyses from other sampling dates were relatively similar. The few differences in survival in June 2003 are discussed. We analysed survival data using ANOVA for split-split plot design with R Statistical Computing version 2.0.1. Many plants did not survive, so some levels of some factors were missing, and not all combinations of treatments could be tested in each model. We thus analysed biomass data using nested ANOVA with Statistica software version 6.0 (Statsoft 2004). Percentages (survival) were arcsine-transformed and measurements (biomass) log-transformed when normality and homogeneity assumptions were not met (Sokal & Rohlf 1998).