Phytodiversity of temperate permanent grasslands: ecosystem services for agriculture and livestock management for diversity conservation

What can biodiversity of pastures and meadows mean for the farmer who needs biomass for his livestock? Table 1 summarizes results of studies on biodiversity effects on productivity or other ecosystem services. Due to the difficulties involved in transferring results from experimental grassland plots to agricultural situations (Caliman et al. 2010; Isselstein 2005), we will concentrate here mostly on the few studies carried out in agriculturally managed swards.

Results from these studies are conflicting: while some experimental studies found no consistent effect of biodiversity on primary production (de Lafontaine and Houle 2007; Deak et al. 2009; Kahmen et al. 2005; Soder et al. 2006; Tracy and Faulkner 2006), others, both observational (Bai et al. 2007) and experimental (Caldeira et al. 2001; Tracy and Sanderson 2004; van Peer et al. 2004; Weigelt et al. 2009), found a positive effect (Table 1). Despite initially positive impacts on plant production, Tracy and Faulkner (2006) did not measure increased daily liveweight gains of cattle nor could they increase stocking rates in more diverse pastures. Also Soder et al. (2006) found no effects on herbage intake or milk production of dairy cattle with increased plant diversity. In a survey of 854 meadows and pastures in Inner Mongolia, Bai et al. (2007) observed increased primary production with increased plant diversity. However, the authors pointed out that this coincided with patterns of annual rainfall and soil nitrogen. Furthermore, conditions in this area were representative of those in the Eurasian steppe, but not necessarily directly comparable with managed temperate grassland. The voluntary daily dry matter intake of sheep has been found to increase with species richness up to eight species out of 11 in an indoor cafeteria trial (Wang et al. 2010). This should translate into weight gains of the animal, which were however not determined. In a field experiment, no difference in intake was observed between fields with four to six and with more than eight plant species. The authors discuss that this might be due to supplementary corn offered in the field (Wang et al. 2010). Interestingly, the studies finding positive effects were mainly carried out in experimental plots, not in agricultural grassland (Caldeira et al. 2001; Tracy and Sanderson 2004; van Peer et al. 2004; Weigelt et al. 2009). In other studies of experimental plots, positive effects on production were found when the number of sown species was considered. However, based on the total number of species present (i.e. including weeds), no consistent effects were found (Bezemer and van der Putten 2007; Dodd et al. 2004).

It has been a principle of ecological theory that the assembly of species in a given habitat depends on the niches present. Therefore, within the limits of historical influences and site accessibility for propagules, the available resources determine phytodiversity in the first place. Here, diversity has been found to be maximal at intermediate resource availability (Critchley et al. 2002; Janssens et al. 1998; Schmid 2002). Hautier et al. (2009) could show that a negative effect of fertilisation on phytodiversity of fertilised grassland communities was mainly due to increased competition for light and restriction of light reaching the lower layers of vegetation. In contrast to this, Rajaniemi (2002) did not find an effect of shading on species richness or diversity in an unproductive former field and concluded that the observed significant effects of fertilisation were due to increased total above- and belowground competition. The importance of belowground competition in such a system where light is not limiting could later be confirmed (Rajaniemi et al. 2003).

In agricultural grassland, this initial diversity determined by the available niches is manipulated by management. A new situation develops where species richness is in dynamic equilibrium with the management, if this is constant. In contrast to this, the experimental grassland plots used for biodiversity–productivity research have usually been weeded intensively, inhibiting the establishment of such a dynamic equilibrium. If weeding was terminated, similar species richness developed within 2 years in all plots of initially different richness (Pfisterer et al. 2004).

Taking a closer look at the results from experimental grassland studies, it becomes obvious that observed diversity effects were most pronounced with species numbers increasing from one to two or four. Many studies found that 90% of the productivity effect was reached with five plant species (Roy 2001). In permanent grassland, the plant diversity is usually larger. For example, Sanderson et al. (2004) summarized that American grazing lands comprised between nine and 50 species per 1000 m² and European grasslands between 10 and 60 species per 100 m², depending on management intensity. Thus, species richness may usually be too large in permanent grassland to find effects of diversity on productivity.

Several studies have pointed out the larger impact of species identity (Hooper and Vitousek 1997) or functional diversity (Díaz and Cabido 2001) than species number on primary production. Here, functional diversity is not necessarily only the presence or absence of legumes, but can encompass the range of traits like leaf sizes, canopy heights, or rooting depths (Díaz and Cabido 2001). These findings should have implications for the assembly of seed mixtures for grassland renovation, where the species number is furthermore usually in the range where species richness-productivity effects have been found. In practice, this principle has already been used and the long-term experience of seed companies and farmers has been found to deliver a superior product to experimental mixtures in Switzerland (Suter et al. 2010).

To sum up, a clear effect of species number on primary or secondary production of grassland under agricultural conditions could not yet be demonstrated. This may be due to primary effects not translating into animal production, vegetation composition developing a dynamic equilibrium with management conditions or higher species richness in permanent pastures than found effective in experimental grassland. If fertilisation was also manipulated in permanent grassland experiments, its effect on biomass production outreached that of diversity [Crawley et al. (2005); Silvertown et al. 2006; but see also Weigelt et al. (2009) for results in weeded experimental grassland]. Thus, a potential production benefit may not convince farmers to protect diversity in their grasslands.

Despite an unclear productivity effect, increased diversity can still have benefits for livestock farming. First of all, the production stability has been found to increase, granting good harvests also in years with adverse weather conditions (Deak et al. 2009; Silvertown et al. 2006; Tilman et al. 2006). However, in a comparison of stability of biomass production of plots sown with 0, 4 or 15 different species and not weeded, Bezemer and van der Putten (2007) found a positive relation with sown species number, but not with actual species richness and concluded that the relationship is context-dependent.

Nutrient losses may be smaller under diverse grassland (Mulder et al. 2002; Niklaus et al. 2006), probably due to resource complementarity and a better use of the soil space (Harrison et al. 2007; Weigelt et al. 2005). This can also cause a better water use efficiency of more diverse systems (Caldeira et al. 2001; van Peer et al. 2004). So far, most studies looking at these relationships have been carried out in experimental grassland plots. Research on long-term grassland, where root structures have developed over long time periods, is needed.

Important effects of phytodiversity on product quality and animal health have been found, which will now be discussed in more detail. Grazing, as compared to indoor fattening, results in a different fatty acid composition (higher proportions of linoleic and linolenic acid), darker and redder meat with better sensory qualities and an increased shelf-life (Dieguez et al. 2006; Farruggia et al. 2008; Fraser et al. 2009; Hocquette et al. 2007). Fraser et al. (2009) conducted grazing experiments with different breeds on improved permanent pasture (ryegrass/clover) and semi-natural rough grazing on Molinia caerulea dominated swards. Their results indicated a greater influence of the sward type on animal performance, grazing behaviour and meat quality than the breed when beef cattle are produced in less favoured areas. Under rough grazing, loin steaks contained more vitamin E and had a lower lipid oxidation (Fraser et al. 2009).

Some recent studies have demonstrated that dairy products from grazing ruminants have a composition thought to be beneficial to human health, compared to that from animals fed concentrate diets; the content of unsaturated fatty acids in milk, for example, increases with grazing (Cuchillo et al. 2010b; Elgersma et al. 2006). Milk yields and animal productivity are limited by genetic potential, botanical composition and trophic status of the pasture, which needs to meet basic requirements to ensure a sustainable system (Osoro et al. 2007). Extensive grazing on bio-diverse swards for milk production is often characterized by smaller milk yields but more solid contents (Farruggia et al. 2008). Moloney et al. (2008) concluded from a review of several experiments that more phytodiverse pastures produced milk with increased C18:3n-3 and polyunsaturated fatty acid concentrations whereas the saturated fatty acid concentrations were in most cases reduced. Leiber et al. (2005) discussed that changes in the ruminal ecosystem due to energy shortage or specific secondary plant metabolites may be possible causes for the high C18:3n-3 concentrations in alpine milk.

Animals mix plant and biochemical diversity to enhance the nutritive value of the diet as well as to maintain possible toxic concentrations of plants below critical levels (Provenza and Villalba 2010). Certain plants can also have health benefits for the animals. For example, legumes contain condensed tannins that may cause increased production of milk and wool, improve the lambing percentage and reduce bloating risk as well as intestinal parasites (Min et al. 2003). In addition, Martin et al. (2010) point out that adding tannin-rich legumes to animal diets may decrease rumen methanogenesis and thus the production of the greenhouse gas methane. As reducing methane production during rumination also means decreasing energy losses by the animals, this is interesting from a production point of view as well. So far, the importance of diverse grasslands in this respect is not completely understood.

Thus, despite unclear productivity effects, plant richness may have positive effects on product quality, animal health, nutrient and water retention as well as production stability. The latter may be especially important for sustainable production under changing climatic conditions, but has so far mainly been studied in experimental plots.

Selectively grazing animals preferrably feed on certain pasture areas (horizontal selection) or plant parts (vertical selection) (Arnold and Dudzinski 1978; Elsässer 2000). Given a free choice, they select a mixed diet rather than chosing one fodder species only (Villalba and Provenza 2009). The chosen biomass usually has higher concentrations of nitrogen, phosphorus and energy than avoided material (Wales et al. 1998). Despite the variability in quality and digestibility of herbage on offer in time and space, ruminants aim to select herbage with fairly constant digestibility (Fulkerson et al. 2007; Garcia et al. 2003). Therefore, the degree of selectivity changes with the quality of the herbage on offer. The animals have to resolve the trade-off between feeding on preferred food and the energy required to forage for that food (Rook et al. 2004; Utsumi et al. 2009). A higher selectivity has been found when preferred patches were aggregated (Dumont et al. 2002).

The intensity of vertical selectivity differs between animal species and is related to the actual mechanical way of fodder uptake. Cattle take up plant material with their prehensile tongue into the mouth where it is pressed against the dental plate of the upper jaw and torn off with a move of the head. They can graze tall herbage more easily than sheep because of their physical size (Hodgson 1990; Wilmshurst et al. 2000). Cattle might select separate leaves merely from tall plants, while sheep and goats with their narrower and more pointed muzzles graze more fastidiously and readily select individual leaves and other plant parts (Animut and Goetsch 2008; Arnold and Dudzinski 1978; Dumont 1997).

Besides determining the potential bite selection of an animal, the body size also influences the size of a feeding station, i.e. the area a standing grazer can reach with its head (Table 2). A cluster of feeding stations with the same intake rate is defined as a grazing patch. The size of this feeding patch depends on the size of the animal as well as the heterogeneity, biomass and quality of fodder available. Thus, the size and selectivity of the animal in interactions with the heterogeneity of the sward will lead to a mosaic of areas with different spatial and temporal dimensions of defoliation (Table 2).

Sight helps the grazing animal to position itself towards the other animals and the environment, but is less important in selecting the diet. In experiments, sheep with their eyes bandaged selected a diet similar to that of sheep allowed to see. However, the preference for certain grassland plants changed when touch, smell and taste were impaired (Arnold and Dudzinski 1978). Animals familiar with a sward or forage were quicker in finding their preferred feeding patches (Bailey and Sims 1998) and using the available forage (Flores et al. 1989a, b), suggesting an influence of learning in patch selection (Dumont 1997).

Besides a spatial and qualitative dimension of selective grazing, there is also a temporal dimension that influences the structure of the sward and helps to establish a mosaic of more or less frequently defoliated patches. Thus, the previous meal an animal had seems to have an influence on the preference for the next one (Dumont 1997; Mote et al. 2008). From experiments on extensive grazing it was concluded that there was a strong diurnal pattern of selectivity: Dumont et al. (2007) found a marked preference of cattle for short, highly digestible bites in the morning and an increased consumption of bite types requiring a greater rumination effort during the second half of the day. Bites of short mixed vegetation consisting of grasses and herbs were generally grazed preferentially, regardless of the offer and time of day (Dumont et al. 2007).

Plant species on a pasture usually exhibit two defence strategies: resistance to (avoidance) and tolerance of herbivory (Briske 1996). Resistance refers to the ability of a plant to reduce the amount of damage. This means reducing the probability and intensity of defoliation by morphological traits like thick hair, sharp leaf blades (silica) and chemical defences. This group is classified as facultative weeds and weed grasses if they are potentially edible (Opitz von Boberfeld 1994). Among these are Holcus lanatus, Deschampsia caespitosa and Ranunculus repens. Also unwanted poisonous and non-edible plants like Equisetum palustre, Cirsium palustre or Juncus effusus show this defence mechanism and may compete successfully for space and nutrients if no agronomic measures are taken (Moretto and Distel 1997, 1999). Tolerance is the ability of a plant to react to defoliation by rapid regrowth and recovery without a reduction in fitness. In this case, growing points for regeneration are located below the grazing level at the shoot basis or along stolons and storage roots may contribute to survival after intense defoliation (Herben and Huber-Sannwald 2002).

Disturbances by the grazer can shift the competition conditions among plants, as varying defoliation frequencies lead to different optima in adaptation to grazing. Generally, intensive grazing will induce the formation of a dense, well-tillered sward (Frame 1992; Matthew et al. 2000; Nelson 2000). As a result, the vegetation composition usually differs between tall and short sward areas (e.g. Correll et al. 2003) and indicator species for the extremes in grazing, i.e. selective undergrazing and selective overgrazing, can be determined (Opitz von Boberfeld 1994).

The treading of grazing animals can have two effects: it may cause compaction of the topsoil and it can create open gaps without vegetation cover. According to Jacob (1987), the tread of a cattle of 600 kg causes a pressure of 4–5 kg cm⁻² on the topsoil. The resulting compaction may lead to retarded water infiltration and gas diffusion, increasing the risk of surface runoff and elevated emissions of gases like the greenhouse gas nitrous oxide (Menneer et al. 2005; Mulholland and Fullen 1991; Oenema et al. 1997; van Groenigen et al. 2005). However, compaction can also have positive effects: it is expected that treading might compensate for the prohibition of rolling in spring on nature protected grassland (Benke and Isselstein 2001).

Damages of the vegetation leading to patches of bare soil may offer space for propagation of seeds from the seed bank and invasion by other species. This can be desirable, but can also promote the growth of unwanted species. Kohler et al. (2006) found that gaps were colonized by species with small seeds, unspecialized seed dispersal, a persistent seed bank and high vegetation spread. The role of other grazing effects (feeding, dung deposition and trampling) on the recolonisation was only secondary, modifying the competition between recolonisers.

Plant species react differently to treading. Jacob (1987) found that Poa annua had increasing yield proportions at heavily frequented pasture gate areas while proportions of H. lanatus decreased. In line with this, Graf Bothmer (1953) ascribed a community at a zone close to pasture gates of permanent pastures showing highest frequency and dominance of P. annua, Polygonum aviculare, Plantago major and Lolium perenne to larger influences of treading in these areas.

The grazing animal transforms vegetation biomass into animal biomass and performance; however, with considerable losses and a rather low efficiency. In cattle, about 75–95% of the ingested N is returned via excreta (Whitehead 1995). In this transformation, nutrients are redistributed from relatively large areas where the animals feed to small excreta patches. These excreta patches have high input of nutrients, but also experience a grazing pattern different to the rest of the pasture area.

Dung patches might cover 5–10% of the grazed area each year in dairy farming, but the affected area can be much greater and, depending on weather conditions, be one to six times the covered area (Bao et al. 1998; Bastiman and van Dijk 1975; Haynes and Williams 1993). Herbage growing in the vicinity of dung patches is unattractive to stock, also due to the dung smell, and is avoided unless the grazing pressure is very high (Frame 1992; Gillet et al. 2010). This behaviour is explained by hygienical/sanitary advantages of avoidance (Hutchings et al. 1998). As a result, micro-areas with a tall sward develop, especially under extensive grazing.

Urine patches can cover up to 24% (at 700 cow-days ha⁻¹) of the pasture and the area affected may be up to double that size (Haynes and Williams 1993; Whitehead 2000). The vegetation at urine patches may be grazed preferentially (Day and Detling 1990; Steinauer and Collins 2001), probably due to high concentrations of minerals in the herbage.

The nutrient return with excreta is large. It is unevenly distributed within the pasture and often accumulates at feeding, rest and water places (König 2002; Owens et al. 2003). This results in further differentiation in sward structure and soil conditions. In the process of grazing and excretion, a decoupling of major plant nutrients takes place. Usually, more K is excreted in urine than in dung (Whitehead 2000); while P is mainly excreted in dung. A certain amount of N is excreted with dung, the rest with urine (e.g. Schellberg et al. 2007). Thus, the more N cattle take up, the higher the ratio of N in urine versus N in dung (Whitehead 1995).

On urine patches, legumes are especially negatively affected. White clover competes only poorly for mineral N with grasses and is more susceptible to scorch. N₂ fixation can be markedly depressed in the urine patch (Ball et al. 1979; Ledgard et al. 2001). Therefore, urine patches become grass dominated (Ledgard et al. 1982), but the degree of clover reduction and N₂ fixation is dependent on the time of urine application as well as the clover content of the sward (Ball et al. 1979; Ledgard et al. 1982). Thus, Norman and Green (1958) did not find an effect of a single urine application on the botanical composition of a pasture.

Dung patches may lead to an increase in the total yield of grasses around the pats (MacDiarmid and Watkin 1971; Norman and Green 1958). This effect was shown to be stronger when the excretion was combined with defoliation. Underneath the cow pat, the vegetation died (MacDiarmid and Watkin 1971). Dung patches were found to decrease species turnover and thus have a stabilizing effect on plant composition in their direct surroundings in mountain pastures (Gillet et al. 2010).

The development of a specific sward structure is induced by the behaviour of the grazing animal as discussed above and by agricultural management (pasture maintenance) on a background of site characteristics. Important with respect to grazing management is the grazing intensity, grazing system and the type and breed of grazing animal. The effects of grazing are further modified and partly determined by the level of nutrient input (fertilization; additional feeding), and the intensity of intermittent management like cutting or topping, rolling and harrowing, usually intended to decrease grazing effects. However, these secondary management effects will not be considered in more depth here.

High grazing intensity has often been blamed for negative effects on diversity (Dumont et al. 2009; Henle et al. 2008; Plantureux et al. 2005; Vallentine 2001). With increasing intensity, animals become less selective in the choice of their diet in order to obtain sufficient intake (Dumont et al. 2007). Thus, defoliation will be more homogeneous than on less intensively grazed paddocks, creating less diverse niches. Furthermore, the frequency of defoliation will be high, allowing only pasture plants adapted to this to survive. With very high grazing pressure, animals may harm vegetation points by removing too much biomass, especially from preferred plant species. This happens more easily by animals being able to remove biomass close to the soil, such as horses, sheep or goats rather than cattle (Animut and Goetsch 2008; Benavides et al. 2009; Menard et al. 2002). With high grazing intensity, effects due to treading and gap creation will also be more serious. In contrast to selective grazing, gap creation and compaction will not be maximal at low grazing pressures, but increase with increasing intensity. However, colonisation of new gaps will be retarded with high grazing intensity due to frequent disturbances of newly emerging propagules. Excreta patches will affect larger pasture areas (White et al. 2001) and more nutrients can be lost by run-off, leaching or gaseous losses. However, increased grazing pressure decreases the size of dung pats as the animals tend to feed closer to and sooner after an excretion event.

The grazing system may have large effects on diversity, even if the annual stocking density is the same for different systems. Most important in this respect are rotational grazing and permanently stocked pasture. Permanently stocked pasture requires less work from the farmer, as the animals are put on the pasture in spring and removed at the end of the grazing season. In rotational grazing, animals have less space per unit of time, but are transferred to a new paddock at regular time intervals. Thus, at a given time, the stocking density is higher with rotational grazing, but the vegetation is then allowed time to recover until the animals rotate back to the same paddock. Therefore, the pressure on preferred species is less intense than in permanently stocked pastures (Pavlu et al. 2003). It has been found that grazing at intermediate intensity may allow more plants to get to the flowering stage (Correll et al. 2003; Sahin Demirbag et al. 2009) and may thus have positive effects on the vegetation, but also on the abundance of insects (Dumont et al. 2009; Kruess and Tscharntke 2002).

As permanently stocked pastures can only be grazed with relatively few animals to allow them to find enough fodder even in times of little vegetation growth, different areas develop with very different frequency of use. The seasonal vegetation development of a continuously grazed pasture (set stocking) in temperate areas can be divided into three parts, namely the spring/early summer period, the summer, and the late summer/autumn period based on the development of herbage mass (Jacob 1987). Figure 1 gives an overview of the interactions of grazing cattle and sward structure during a grazing period. The spring/early summer period is characterized by a surplus of herbage mass of good quality allowing a high performance of livestock. As grazers initially use only relatively small areas on continuously grazed pastures with set stocking, other areas develop into a generative state where feed quality deteriorates. During the summer period, grazing cattle therefore have to invest time to select herbage and are also forced to use overripe parts of the pasture. As a result, performance of the individual animal decreases (Baumont et al. 2000). Towards the end of the grazing period, in late summer/autumn, the relation between herbage on offer (standing crop) and intake by the grazing cattle synchronizes again. At this time, the variability in quality and sward height is reduced, causing less need for the animal to select. This will allow, weather conditions permitting, a moderate increase in animal performance during that period. Overall, preferred patches are defoliated very frequently and experience the same pressure as on pastures with high grazing intensity. However, other pasture areas are hardly influenced by the animals during long parts of the grazing season. Here, competition between species will drive diversity development. Usually, farmers would choose to cut or mulch surplus vegetation at the end of a grazing season.

The type of grazing animal has important implications for phytodiversity, especially due to different feeding preferences. The mechanical prerequisites for selective grazing and their differences between animal species have already been discussed above. Requirements of the animals for energy and quality further determine their influence on the vegetation. Impacts due to treading and excretion vary between species. Treading is especially important where a lot of weight is carried on a small area or where animals are very mobile. Apart from small differences in nutrient retention between animal species, excretion mainly differs with respect to the amounts excreted at a given time and the distribution of excreta patches. Thus, depending on the size of the pasture, horses may show latrine behaviour, excreting always at the same points (Lamoot et al. 2004), while cattle may distribute excreta more evenly over the pasture area (White et al. 2001). This has implications for the nutrient return to the plants and mining of nutrients versus accumulation at other places.

Interestingly, the choice of the breed, apart from size and weight restrictions, seems generally to be of less importance in cattle (Fraser et al. 2007; Isselstein et al. 2007), but effects have been reported for sheep and goats (Osoro et al. 2007, 2002). Larger breeds might achieve better performance rates but have higher requirements for maintenance (protein, energy, minerals etc.).

Different effects of grazers on swards are sometimes utilized in co-grazing. Thus, grazing by goats has been found to have positive effects on following sheep grazing, as the proportion of clover in the pasture increased (del Pozo et al. 1998). Sheep may feed on dung pats of cattle and vice versa, decreasing the amount of nutrient and pasture space lost (Abaye et al. 1994; Forbes and Hodgson 1985; Fraser et al. 2007). Co-grazing may also lead to increased daily liveweight gains of both animal species involved (Nolan and Connolly 1989). A combination of species in co-grazing may lead to the development of a more uniform sward with respect to height. However, due to the distinct effects on plant species by selective grazing, treading and excretion, the underlying heterogeneity might be larger with co-grazing, allowing the creation of more diverse niches.

To sum up, grazing is regarded as a most efficient way of utilizing and maintaining less intensive and semi-natural grasslands. However, the interactions of soil and site characteristics, hydrology, plant communities, and grazing management are complex and the situation is often further complicated by restrictions in grazing time, nutrient return and market demands. A thorough understanding of the grazing process will help to properly address the problems arising in a specific environmental/agricultural/socio-cultural context and to combine benefits of extensive grazing concepts for improved or maintained biodiversity, landscape scenery, soil protection and farm income (Soder et al. 2007). In order to achieve these tasks, it is likely that management restrictions need to be adapted to local conditions, especially by adjusting grazing intensity to productivity, by allowing some form of nutrient return or by mulching, to avoid cases where the process of selective grazing might lead to abandonment of parts of the pasture. In a complex situation like extensive grazing what may be beneficial for one objective may have damaging consequences for another (Mills et al. 2007).