Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A methodological framework to facilitate analysis of ecosystem services provided by grassland-based livestock systems

A methodological framework to facilitate analysis of ecosystem services provided by... International Journal of Biodiversity Science, Ecosystem Services & Management, 2015 Vol. 11, No. 2, 128–144, http://dx.doi.org/10.1080/21513732.2015.1030695 A methodological framework to facilitate analysis of ecosystem services provided by grassland- based livestock systems a,b, a,b a,b M. Duru *, J.P. Theau and G. Martin a b INRA, UMR1248 AGIR, F-31320 Castanet-Tolosan, France; Université Toulouse, INPT, UMR AGIR, F-31029 Toulouse, France (Submitted 27 February 2014; accepted 26 February 2015; edited by Rob Alkemade) It remains challenging to describe ecosystem services (ES) provided by grassland diversity and their underlying drivers. Issues for characterising them are related to scale, knowledge and outcomes that are stakeholder-dependent. Forage production interests farmers, while C sequestration, species richness and the landscape mosaic interest society. To address these issues, we developed a methodological framework (MF) based on five grass functional types (GFTs) which enables usable indicators to be easily defined, such as the percentage of plants with a fast (Fast ) growth strategy in grasslands. The GFT MF consists of characterising plant functional diversity at field and farm levels, analysing its response to environmental and management drivers, and its effects on ES. The MF is applied to eight farms differing in their orientation and in their management intensity. Fast responds positively to the amount of fertiliser supplied and negatively to field elevation. At the GFT field level, Fast is positively correlated with herbage production and negatively with soil C content and species richness. GFT Within-farm grassland diversity of Fast allows examination of how animal feed requirements match available resources GFT and the landscape mosaic created. Our MF addresses grassland diversity with indicators derived from GFT, which allows summarising relations between environmental and management drivers and ES, then examines trade-offs between ES. Keywords: landscape; management; plant functional type; provision services; supporting services; trade-offs Introduction known, even though farmers promote it. For instance, Martin et al. (2009) show that functional plant diversity Policy-makers at European and regional levels are keen to in grasslands provides flexibility in the timing of grassland encourage more sustainable livestock systems, especially use. Similarly, Nozières et al. (2011) suggest that within- to reduce their environmental impacts and to enhance farm plant diversity decreases production costs by more ecosystem services (ES). This strategy encompasses max- closely matching grassland types to animal feed require- imising provision of ES by farms (Power 2010). In a ments, thereby improving stability of the livestock system. farming context, three main types of ES are distinguished Second, most grassland research produces results suitable (Swinton et al. 2007; Zhang et al. 2007): (i) services from for understanding effects of drivers on ES but struggles to agriculture, that is, provisioning services (e.g. forage or produce results that are easily used by stakeholders milk), (ii) non-market services, some authors distinguish- (Matthews et al. 2010). Thus, particular attention should ing regulating services (e.g. C storage) and cultural ser- focus on building relevant and appropriate methodological vices (e.g. attractive landscapes), and (iii) services to frameworks (MFs) that produce salient, legitimate and agriculture, that is, several supporting and regulating ser- credible information (Cash et al. 2003). Our objective is vices called ‘input services’ (Le Roux et al. 2008) because to evaluate strengths and weaknesses of a MF based on they allow a decrease in chemical inputs. Better under- characterising plant functional diversity to evaluate ES and standing of trade-offs and synergies among services rela- identify their main drivers, while addressing the above tive to their beneficiaries (e.g. farmers, society) is essential issues and requirements. for management and policy decisions (Fisher et al. 2009). On grassland-based livestock farms, ecosystem proper- Studying services provision, trade-offs and synergies in ties that provide ES depend largely on plant functional grassland-based livestock systems raises several issues. diversity (the presence or abundance of particular func- First, services are scale-dependent (Lavorel & Grigulis tional groups or traits) rather than on species diversity 2012) and are studied most at field and landscape levels (Hooper et al. 2005). Thereby there is growing consensus (Benton et al. 2003). The farm level is under-studied even that a plant trait-based methodology may well be able to though it is the level at which land-use and management address issues such as grassland ecophysiology and help decisions are made. Previous in-depth analysis of grass- manage some of the services that grasslands deliver to land-based livestock farms shows that biodiversity can humans (e.g. Diaz et al. 2007). The key hypothesis is also provide a management service, enabling farmers to that traits can simultaneously explain individual plant improve their management and working conditions responses to biotic and abiotic factors, with ecosystem (Lugnot & Martin 2013). However, this ES is not well *Corresponding author. Email: mduru@toulouse.inra.fr © 2015 Taylor & Francis International Journal of Biodiversity Science, Ecosystem Services & Management 129 properties underpinning ecosystem functions (Lavorel & examine management and environmental factors that Garnier 2002) and associated services (Lavorel et al. shape grassland functions and then analyse ES provided 2011). Plant functional diversity in structure and composi- by grasslands at the field, LMU and farm/landscape levels. tion is usually defined through two components: commu- Finally, we discuss which drivers have the greatest effect nity-level weighted mean of trait values or functional types on grassland functions and propose a synthetic representa- (i.e. groups of species sharing the same set of attributes) tion of relations between management and environmental (Colasanti et al. 2001) and functional divergence (FD) drivers, grassland functional structure and composition (Lavorel et al. 2011). FD describes trait diversity among and ES, which allows characterisation of trade-offs and species (or functional groups) that coexist within a com- synergies between ES for different beneficiaries. munity (De Bello et al. 2009). It is hypothesised to operate through functional complementarity (Petchey & Gaston 2006). For example, within-community diversity in plant Materials and methods height is expected to improve light capture (Vojtech et al. Development of the MF 2008), while diversity in leaf structural and chemical traits Linking driving forces, grassland functions and ES would reflect diversity in nutrient acquisition and retention We adapted a framework linking driving forces, grassland strategies (Gross et al. 2007). To characterise the relation functions, ES and beneficiaries of services developed by between biodiversity and ES, it is important to define Ape et al. (2012)(Figure 1). ES provided by grasslands indicators of ecosystem function that indicate the extent were distinguished according to the level at which they to which an ES can be provided (Ape et al. 2012), then were noticeable (field, farm and landscape) each of them ensure that stakeholders can easily use them. Therefore, being specific beneficiaries (Figure 1). For example, some several researchers (e.g. Van Der Biest et al. 2014) state management services are noticeable only at LMU or farm that there is a demand for simple MFs relying on indica- levels (emergent properties), while forage-production ser- tors to evaluate services. These indicators can be land-use vices are noticeable both at field and LMU levels proxies (Van Der Biest et al. 2014) or can be based on (Table 1). When farm level is considered, it is as a grassland functional structure and composition. We whole farm (Smukler et al. 2010), while the sub-levels explore the strength of the latter option. are key for understanding management services (Duru Methods requiring plant trait measurements are too et al. 2013). Furthermore, we distinguished the level at time-consuming and unfamiliar to stakeholders. which services are noticeable from the level at which data Therefore, we developed and used a simplified MF based should be recorded to assess them. For forage services the on grass functional types (GFTs) as indicators, in which levels can be similar, whereas for C storage the levels are stakeholders can easily characterise ES provided by grass- different (Table 1). In our study, we focus on field and land diversity in grassland-based livestock systems: for- farm levels. Upscaling, which was not considered in this age, management, input, environmental (species richness paper, requires integrating data (e.g. for C storage) or and soil C storage) and cultural aspects (aesthetic through spatially explicit modelling (e.g. for landscape attractive- characterisation of within–between-grassland diversity). ness). Farm orientation has only an indirect effect on To evaluate its suitability for addressing the above issues grassland functions through land use and management, and requirements, the methodology was applied to a range while environmental factors (e.g. climate, field and soil of management and environmental factors at different characteristics) have direct and indirect effects on grass- levels of organisation: field, land management unit land functions (Figure 1). (LMU) (i.e. parts of farms allocated to single groups of Trade-offs between services can occur for the same animals corresponding to single management units for beneficiary (e.g. forage and management services for production, feeding, health care, etc.) and a set of fields farmers) or between beneficiaries (e.g. forage services for at farm or landscape levels. The field level is needed to farmers, C storage for society). Arbitrating between ser- characterise the response of plant functional composition vices requires developing a small number of relevant to environmental factors and management practices. The indicators of grassland functions for assessing multiple LMU level is needed because averaging data at the farm services. For this, we distinguish grassland functions (i.e. level loses possible within-farm differences due to differ- the capacity of grassland ecosystems to provide services) ences in the management of animal groups (e.g. cows and from ES, which contribute to human well-being heifers). The level of a set of farm or landscape fields is (Raudsepp-Hearne et al. 2010). needed to assess services related to between-grassland field diversity. To cover a large range of management factors, the MF was applied to dairy and beef farms Selection of indicators of grassland functional structure differing in farming intensity, which influences grassland and composition vegetation (Andrieu et al. 2007). The paper is organised as follows. First, we describe Although credible, feedback from stakeholders shows that development of the MF, especially the choice of indicators the trait-based methodology to characterise plant func- of grassland functional structure and composition, and of tional diversity is not relevant in practice because it is management and environmental drivers. Second, we too time-consuming (Duru, Cruz & Theau 2010). To be 130 M. Duru et al. Figure 1. Relations between ES noticeable at different levels, their drivers and beneficiaries; dotted arrows indicate relations that were not studied. Table 1. Mapping of grassland properties to grassland services. Services Organisational level Indicators Types component at which services are for recording data of grassland functions of services provided noticeable for (potential services) by grassland (used targeted services) beneficiaries †† ‡ Forage Production and quality LMU, field (farmer) Field; LMU Fast (aggregated at Stocking rate GFT LMU level) Management Flexibility allowed by LMU, field (farmer) LMU (field ) Late (aggregated Spreading of harvest GFT temporality of at LMU level) dates herbage growth Adequacy between LMU (farmer) LMU/ farm GFT distribution Degree at which forage forage characteristics regarded to animal characteristics match and animal feed feed requirement and animal feed requirements inputs (fertiliser, requirements labour) (aggregated at LMU level) Input Fertility Field (farmer) Field Div N uptake Siomass production/ N GFT permitted by within- supplied field diversity (S) Environmental C sequestration Landscape Field Fast Soil C content GFT Species richness Field -> landscape Field Fast Sum Number of species GFT; GFT (society) Cultural Field mosaic Field -> landscape Set of fields, Fast × Div Visual within and GFT GFT (society) landscape mapped at different between fields levels diversity Notes: level not considered in this study. ‡reflect the forage production and its use intensity. Fast Late are the percentage of GFTs having a fast and a late growth strategy respectively. GFT GFT Sum is the percentage of grass species in biomass in a grassland. GFT relevant to decision-makers (Cash et al. 2003), that is, for characterising grassland vegetation (Duru, Theau, et al. the practical use of knowledge by farmers, we previously 2011). We classified grass species into functional types made a major change in the methodology for (Colasanti et al. 2001) rather than using continuous traits, International Journal of Biodiversity Science, Ecosystem Services & Management 131 even though the latter are expected to better represent The second management service studied depends on how changes in the intensity of processes (Lavorel & Garnier the farmer organises grassland diversity to match animal 2002). Based on leaf and phenological plant traits, five (A, feed requirements (Duru, Theau, et al. 2011). Typically, B, b, C and D) elementary GFTs were defined (Cruz et al. such a service can be assessed only at the LMU level. In 2010). These GFTs were ranked by leaf dry matter content fact, farmers combine fields allocated to different forage (LDMC), which increases from type A to D, and flowering crops into several assemblages, with each single assem- date, which is latest for types b and D (Duru et al. 2013). blage allocated to feed a particular herd batch. The number When combined, these elementary types reflect two major and nature of such assemblages are designed to meet growth strategies (Sun & Frelich 2011): fast (Fast :A objectives, such as to increase self-sufficiency of the sys- GFT and B) vs. slow (Slow : b, C and D) and early tem (i.e. the ratio of forage production to consumption), GFT (Early : A, B and C) vs. late (Late : b, D), which reduce operational costs or increase flexibility in organis- GFT GFT are meaningful for farmers (Duru et al. 2013). Based on ing work. Hence, we chose to investigate the farm as a set the observation that grass and dicotyledonous species that of such assemblages. To assess this, we ranked animal coexist in grassland communities display similar or con- feed requirements according to the targeted herbage qual- stant differences in plant-trait values (e.g. Ansquer et al. ity (assessed by Fast averaged at the LMU level): beef GFT (2009) for six plant traits and several ecosystem properties cow < milk cow for grazing, hay or silage; heifer < cow (Duru, Cruz & Theau 2010)), we focused on GFTs (for dairy farms). Assessing this management service expressed as the percentage of grass species in the herbage requires examining within–between-farm GFT distribution mass (Sum ). Consequently, we consider dicotyledo- of animal feed requirements and inputs (e.g. fertiliser, GFT nous species only as a whole to estimate their overall labour) as potential service. Therefore, we first compared impact on plant-community properties (Duru, Cruz & plant functional composition of LMUs within each farm, Theau 2010). FD (Div ) was estimated as the relative then examined whether there was a between-farm effect of GFT percentages of GFTs: the farm orientation or stocking density on each of the within-farm LMU effects. This can indicate whether a specialisation of plant types exists for certain land-use Div ¼ 1  p ; GFT i types, that is, grasslands fulfilling the same function in i1 the feeding system. For grasslands, input services are mainly related to the where p is the percentage of each of the coexistence of plant functional types or groups with differ- five GFTs; p ¼ 100: ent growth strategies within a community (e.g. Div ). It GFT was observed that this improves nutrient or light capture (e.g. Fornara & Tilman 2009). Focusing on fertility, we Selection of indicators of ES provided by grasslands examine if plant functional type diversity should result in Based on the four indicators of grassland functional compo- producing the same biomass with less fertiliser. We sition (Fast , Late ,Sum and Div ), forage ser- hypothesised that fertility services would be highest for GFT GFT GFT GFT grasslands with high within-field functional diversity. To vices were previously assessed at the plant-community level test this, we compared N uptake of grasslands with different (Duru, Ansquer, et al. 2010;Duruet al. 2013). Forage Div . production and herbage quality at the leafy stage are corre- GFT lated with the percentage of Fast , which have high growth For regulating services, we examined relations GFT rates and low lignin content. In this paper, we examine between Fast and soil C content and species richness. GFT whether these indicators (or combination of them) can be Fast is an indicator of weighted LDMC (Pakeman & GFT used to predict a larger set of services based on the Zhang Marriott 2010), which is correlated with soil C content. et al. (2007) classification, that is, ES from agriculture (here, Additionally, species richness is expected to decrease with forage production, regulating services and aesthetic value) decreasing nutrient availability (Ceulemans et al. 2013), and ES to agriculture (here, management and input services), for which Fast is also an indicator (Duru et al. 2013). GFT paying attention to the level(s) at which their assessment is Conversely, the likelihood of increased dicotyledonous most relevant. For each service, Table 1 lists the indicator species richness is strong when Sum is low, because GFT used for assessing potential services and the services used. grasslands usually have more dicotyledonous species than For forage production, in addition to measurements of stand- grass species. ing herbage mass at the leafy stage, we used the stocking rate In an agricultural context, permanent grasslands have calculated at the LMU level as a proxy. intrinsically high aesthetic value (Sanderson et al. 2013). Management services comprise two components. We Some vegetation features at field level such as flower assume that the percentage of Late is an indicator for diversity (Quetier et al. 2007) or vegetation-related feature GFT the timing flexibility for grassland use, because GFTs (landscape heterogeneity) are attractive for humans due to growing late in the season offer a large time window for their aesthetic value (Lindemann-Matthies et al. 2010). harvesting forage. Thus, we examine the relation between Although aesthetic preferences are highly subjective Late (potential service) and the range of harvest dates (Rodríguez-Ortega et al. 2014), we postulate that the het- GFT on farms throughout the growing season (used service). erogeneity of vegetation in phenology, height and colour at 132 M. Duru et al. different spatial levels (from within-field to a set of fields) Case study can provide raw data for assessing landscape attractiveness. Description We, thus, used two of the above indicators (Sum and GFT The study was performed in the Aubrac region in the Div ) to assess landscape diversity, especially its aes- GFT southern part of the French Massif Central (44.68°N, thetic component (Frank et al. 2013), at different levels. 2.85°E). The study area occupies approximately 40 km×20 km. Most of the area consists of unsown permanent grasslands used for dairy and beef livestock Selection of indicators of management and environmental systems. To ensure a large range of management and drivers environmental drivers (Figure 1), we chose farms with The strength of relations between drivers and services contrasting land-use types, with low or high stocking depends on the accuracy with which driver gradients are rates, and for which grassland fields range from 800 to characterised. They remain difficult to assess accurately, 1400 m a.s.l. To choose farms, we performed a pre-survey especially for managed grasslands, because most real with advisors and used data from the grey literature to situations are complex, differing in climate, soil and man- ensure a wide range of farm diversity. We preferred ana- agement (defoliation and fertilisation). These differences lysing a small number of greatly different farms in depth are reflected in the intensity of stress (mainly nutrients and rather than a larger number of representative farms in less temperature, even in the same biogeographic area) and depth. Thus, a total of eight farms, four beef (B) and four disturbance (defoliation intensity, frequency and time) dairy (D), were chosen. For each land-use type, two types that grasslands experience. Choice of management indica- of farm were distinguished according to their stocking tors was based on previous studies of permanent grass- rate. For example, farms 1 and 2 had higher stocking lands (Duru, Ansquer, et al. 2010; Duru et al. 2013). rates than farms 3 and 4 (Table 2). Numbers of cows and Plant nutrient availability and soil conditions (moisture heifers were converted into livestock units (LUs) on the content and pH) were assessed with the Ellenberg index basis of their live weights: a cow (650–700 kg live weight) (EIV, Ellenberg et al. 1992), which characterises species’ corresponded to 1 LU, while a heifer corresponded to 0.8 nutrient (N-EIV), soil moisture (M-EIV) and soil reactivity LU (2–3 years old) or 0.6 LU (1–2 years old). (R-EIV) preferences on a scale from 1 to 9. N-EIV was shown to be a better indicator of fertility than one based on plant N content, which is short-term and environment- Surveys, measurements and observations dependent (Duru et al. 2011a). Observations of species First, farms were surveyed for their main characteristics abundance allowed abundance-weighted EIVs to be calcu- (land area and production type) and management prac- lated for the entire grassland. The advantage of EIVs is tices. Grazing animals were categorised either as animals that they integrate plant species behaviour over many for production (milking cows in dairy systems, heifers, years (Schaffers & Sykora 2000). Furthermore, N-EIV is calves) or as animals for replacement or outside their strongly correlated with P and K availability in the topsoil production period (heifers and dry cows in dairy systems; (Ersten et al. 1998; Schaffers & Sykora 2000). beef suckler cows if calves are not sold at 9 months of age To characterise defoliation regimes, sward height was in beef systems, hereafter referred to as replacement ani- measured with a sward stick (Bossuet & Duru 1992) just mals). The feeding calendar for an average year was drawn before the farmer used the field for cutting or grazing. up, and spring and early summer land use was mapped Sixty measurements were collected for each grassland (Table 2). Then, field topology and topography (e.g. area, field. Cumulative daily temperature (Tsum) for each field elevation, distance to the cowshed, ease of access for cows was calculated from 1 February up to the field’s first use coming from the cowshed or a neighbouring field, (Duru, Ansquer, et al. 2010). Table 2. Characterisation of farm structure and management. Characteristics D1 D2 D3 D4 B1 B2 B3 B4 Structure Grassland area (ha) 54 58 77 69 105 70 115 160 Animal units (total) 60 56 70 52 106 78 87 117 Animal units (cow) 39 38 48 35 70 55 60 80 Stocking density (animal unit/ha) 1.1 1.1 1 0.8 1 1.1 0.8 0.7 Parcel area (ha) 2.5 1.8 3 2.5 3.2 2.5 5.7 6.9 Hay Cutting area/animal unit 0.4 0.45 0.44 0.43 0.37 0.4 0.41 0.36 Topping (%) 24 39 25 0 49 57 46 62 N fertilisation (kg/ha) 40 31 80 120 160 73 97 69 Summer grazing Stocking rate (animal unit/ha) 1.1 1.3 1.8 1.3 1.8 1.3 1.1 0.9 Spring grazing Turnout (DD) 397 410 410 410 450 450 320 450 N fertilisation (kg/ha) 14 29 46 63 15 24 0 0 Mixed diet in spring (days) 27 20 7 15 20 7 10 25 Notes: DD: degree-days from February1st; D: Dairy; B: Beef; 1, 2, 3, 4: farm number. International Journal of Biodiversity Science, Ecosystem Services & Management 133 Table 3. Measurement of driving forces affecting the provision of ES. Driving forces Indicator Level of analysis Measurement Farm orientation Dairy vs beef Farm Constituting of farm sampling and strategy Stocking rate Farm Constituting of farm sampling Animal performances objectives; farmer life style Not considered Environmental Topology Field Field distance from the cowshed factors Altitude, area Survey Soil moisture and reactivity Field Ellenberg index for moisture and reactivity Land management Fertilisation (amount) Field Survey Nutrient availability Field Ellenberg index for nitrogen Cutting management (date, topping) Field Survey Cutting management (height) Field Field measurement Turnout (date) LMU Survey Stocking rate (LU per ha) LMU Calculation from survey data Mixed diet in spring (days) Group of animals Calculation from survey data suitability for mechanisation) were recorded on the 169 determined by oxidation with potassium dichromate and fields of the 8 farms. Data were then mapped according to sulphuric acid (NF ISO 14235) (details of the normative amount of fertiliser applied (organic and inorganic), nature reference in AFNOR (1994)). The original method of De of the first use (e.g. grazing or cutting only, early grazing DM and De Boer (1959) was used to determine the that removes the apexes (topping)), date of first use and exhaustive botanical composition. type of grazing animals. To analyse factors affecting pro- ductivity, mean amounts of mineral N fertiliser and solid or liquid manure applied to each grassland field were Statistical analysis converted into kg, N applied per year. Data were recorded to describe farm orientation, farmer objectives, environ- ANOVA was performed at the field level to examine mental factors and management practices (Table 3). whether there was a significant effect of fertilisation rate A simplified adaptation of de De DM and De Boer and main use (grazing vs. cutting) on the three compo- (1959) method was used to characterise the plant func- nents of grassland functional composition (Fast , GFT Late and Sum ). Data expressed as percentages tional composition of the 169 grassland fields. One diag- GFT GFT were log-transformed to satisfy the conditions for onally orientated transect representative of vegetation ANOVA. Regression analyses were performed to express diversity was sampled between May and the beginning components of grassland functional composition accord- of June, depending on field elevation. Along the transect, ing to quantitative variables of management and the envir- 20 biomass samples were collected within 10 cm × 10 cm onment. By design, there is a parabolic relation between quadrats equidistantly distributed. Each biomass sample Div and the percentage of Fast : maximum func- was exhaustively sorted and a score from 0 to 6 was GFT GFT tional diversity is expected for moderate percentages of assigned to each species present: (0) species present but Fast , with minima at extreme percentages (Duru et al. not dominant, (1) species contributes at least one-sixth of GFT 2011b). Thus, response of Div to management and the biomass sampled (17%), and so on, up to (6) for a GFT environmental variables was calculated separately for species representing all the biomass sampled (100%) Fast ≥ 50% and Fast < 50% (Duru et al. 2013). (Theau et al. 2010). Although about half of the species GFT GFT Based on this relation, within–between-vegetation diver- present were recorded in all grassland fields, no significant sity for a set of fields was characterised by only two differences in plant-strategy distribution existed between variables: Fast and Sum . this simplified method and a point quadrat method (Daget GFT GFT & Poissonet 1971) done on a sub-sample of 60 grassland ANOVA was performed too to examine the effect of fields (Fallour et al. 2008). farm management strategy (i.e. stocking density) and field On the same subset of grassland fields (n =60), characteristics (i.e. distance of field from the cowshed, detailed data were recorded for herbage mass and N con- field elevation and area, soil pH and water availability) centration (to calculate N uptake), soil C content and on field management. Land use was categorised into three species richness. Biomass measurements were made just classes according to level of care requirements of animals: before the beginning of stem elongation of grass species cutting < heifer or beef cow grazing < dairy cow grazing. on four 0.25 m × 0.75 m quadrats in randomly distributed Some services (forage production and inputs) were each grassland community. Total N concentrations of dried directly expressed and predicted with regression analysis according to grassland composition. For vegetation hetero- and ground forage samples were determined by a CHN geneity, only Fast was mapped. To assess the degree to 2000 Analyser (LECO, St. Joseph, MI, USA). Three soil GFT which forage characteristics match animal feed require- samples per grassland were taken from the 0 to 5 cm layer ments, ANOVA was used to discern whether differences for measuring organic soil carbon content. This was 134 M. Duru et al. existed among LMUs within farms and whether there was always significantly higher for dairy farms than for beef an effect of farm strategy or stocking density on Fast . farms for both cut and grazed grasslands (Table 5). GFT Response of plant functional composition to management Results and environmental factors at the field level Drivers of grassland functions There were significant effects of fertilisation and main use Relations between grassland management and field (cutting vs grazing) on the percentages of Fast and GFT characteristics Late , and of field elevation on percentages of Sum GFT GFT Grassland management (fertiliser applied and herbage and Late (Table 6). Large differences were observed in GFT used) depended significantly on topology (distance from the response of Fast to N fertiliser rates for both grazed GFT the field to the cowshed, field area), topography (eleva- and cut grasslands (not shown), which indicates that envir- tion) and soil characteristics (pH and depth). However, onmental factors or biodiversity control N-use efficiency. these effects varied greatly according the component of A consistent, significant response of Fast to elevation GFT management considered (Table 4). In general, mown fields and R – and M-EIV was found for both land-use types are located at a lower elevation and on deeper soils. (Table 7). Based on regressions (Table 4) and the range of Grazed fields for dairy cows lie closest to the cowshed, variation observed for the three variables (N-EIV, plant while those for heifers or beef cows are at higher eleva- height and field elevation), nutrient availability had the tions and furthest from the cowshed. Grasslands used for greatest effect on the percentage of acquisitive species. animals requiring more care (grazing by dairy cows) were When N-EIV changed from 3.5 to 6.5 (the minimum and located at lower elevations and had deeper soils. maximum observed values being 3.1 and 6.9), the percen- Fertilisation rates decreased significantly with increasing tage of acquisitive grass species increased by 50%. In field elevation for three of the four types of grassland use contrast, when canopy height increased from 20 to (Table 4). For cut grasslands, fertilisation rates increased 80 cm or field elevation decreased by 400 m, the percen- with increasing field area. Otherwise, they showed no tage of acquisitive grass species increased by only around clear relation with the other factors studied. Indicators of 15%. Considering N-EIVs instead of N fertiliser and farming intensity (N fertilisation rate or N-EIV) were canopy height at harvest time to encompass both Table 4. Anova of five field characteristics for land use and fertiliser supply (independent variables). Variable df distance from the cowshed Altitude area R-EIV M-EIV F-value F-value F-value F-value F-value (P-value) (P-value) (P-value) (P-value) (P-value) Land use / 171 <0.0001 12,4 (0.001) 1.7 (0.17) 1.2 (0.26) 12 (0.02) Fertilisation supply cutting 80 0.62 (0.53) 11.7 < 0.0001 3.7 (0.029) 0.13 (0.78) 0.3 (0.66) (3 classes: 0; grazing by beef 17 0.12 (0.83) 3.5 (0.06) 1.2 (0.37) 3.4 (0.09) 1.1 (0.31) 0 < × < 75; cows †† >75 kg/ha/year) grazing by heifers 39 0.60 (0.58) 0.12 (0.85) 0.99 (0.21) 5.9 (0.03) 0.95 (0.40) grazing by dairy 35 2.1 (0.25) 35.1 (<0.0001) 1.27 (0.29) 2.1 (0.14) 0.1 (0.95) cows † †† Notes: P-values <0.05 are in bold; see column 2; anova for fertiliser supply was done separately for each land use; R- and M–EIV: Ellenberg indicator values for soil reactivity and moisture, respectively. Table 5. Mean values and ANOVA of N fertiliser and N-Ellenberg Indicator Value (N-EIV) for farm enterprise types (beef vs dairy) (independent variables) considering separately each of the three land-use types. Land use df Farm enterprise type Fertilisation rate (kg/ha) N-EIV cutting areas beef farm 71 5.7 81 dairy farm 100 5.9 F-value (P-value) 8.4 (0.0048) 4.2 (0.04) cow grazing areas 53 beef farm 15 5.1 dairy farm 42 5.7 F-value (P-value) 24.8 (<0.0001) 8.2 (0.006) heifer grazing areas 39 beef farm 17 5.4 dairy farm 44 5.1 F-value (P-value) 10.7 (0.002) 1.9 (0.18) Notes: P-values <0.05 are in bold; see column 2. International Journal of Biodiversity Science, Ecosystem Services & Management 135 Table 6. Mean values and ANOVA of three indicators of grassland composition (Fast Late Sum ) for fertilisation rates GFT GFT GFT (independent variables) considering separately main land-use types. Fertilisation (kg/ha) Land use (number of fields) Fast Late Sum GFT GFT GFT Cut grasslands 149 (26) 71 26 81 87 (19) 60 37 75 42 (30) 62 32 75 Grazed grasslands 63 (31) 60 32 75 33 (25) 45 26 70 9 (38) 28 28 60 ANOVA for land use and fertilisation Fertilisation 6.9 (0.0015) 3.37 (0.037) 3.6 (0.1) Land use 4.1 (0.007) 0.51 (0.76) 8.9 (0.0002) †† F-value (P-value) Field altitude: co-variable 46.1 (<0.001) 14.7 (0.0002) 1.1 (0.31) Notes: Fast and Late are the percentage of GFTs having a fast and a late growth strategy, respectively; Sum is the percentage of grass species in GFT GFT GFT †† biomass in a grassland; p values <0.05 are in bold; df = 171. Table 7. Regression analysis between Fast (percentage of GFTs having a fast growth strategy) and some management and GFT environmental variables. management variables environmental variables Management N fertiliser N-EIV Plant height altitude R-EIV M-EIV r se Grazed + (***) / / – (***) + (***) – (+) 0.55 (***) 13.0 Cut + (*) / / – (***) + (***) – (*) 0.44 (***) 12.4 Cut + grazed / + (***) + (***) – (***) + (**) – (*) 0.63 (***) 12.2 Notes: N, R- and M–EIV: Ellenberg indicator values for nutrient, soil reactivity and moisture, respectively; number of individuals: 169. +or – indicated the direction of effect. *P < 0.05; **P < 0.01; ***P < 0.0001. management types in the same model provided similar effect was not observed for cut grasslands, which had the consistent results. The percentage of Late was posi- highest fertiliser application rates (83.2 ± 47 kg N/ha). GFT tively correlated with N-EIV (<0.001) and Tsum Both regulating services studied had a significantly (P <0.05), and negatively with field elevation (P <0.05), negative correlation with Fast Species richness GFT. but the correlation was weaker (R = 0.22, P < 0.001). decreased significantly as Fast increased (Figure 2b), GFT To test the hypothesis that functional diversity and the correlation increased when considering Sum GFT depended on the levels of stress and disturbance, we (r = –0.75; P < 0.001), which had a significantly negative established two linear regressions according to whether correlation with species richness. Soil C content decreased the percentage of Fast was lesser or greater than 50%. significantly as the percentage of Fast increased GFT GFT We found significant effects (P < 0.001) of N fertilisation (Figure 2c). (positive) and field elevation (negative) on Fast < 50%, GFT and the opposite for both variables for Fast > 50%. The GFT same patterns were observed regardless of the percentage Services assessed at LMU and landscape levels of grass species. For cut grasslands, Div decreased as GFT For resource provision, a significantly positive correlation the percentage of Fast increased, and the reverse was GFT existed between stocking rate and the percentage of observed for grazed grasslands. Fast (P < 0.001; Figure 3a). For grazed grasslands GFT alone, mean stocking rate and Fast were highest for GFT dairy cows and lowest for heifers. However, there were differences in stocking rate for Fast > 50% for the same GFT Provision of ES by grasslands land use and farm strategy. Conversely, similar grassland Services assessed at the field level functional composition was found for all three types of Herbage mass just before the beginning of stem elongation animal groups (beef cows, dairy cows and heifers). A significantly was correlated with the percentage of Fast GFT minimum stocking rate of 0.5 animal units/ha was (Figure 2a). For the less fertilised grazed grasslands (N observed in the absence of Fast . For cut areas alone, GFT fertiliser =24.4 ± 26 kg N/ha), N uptake was significantly no relation was observed between stocking rate and the and positively correlated with Div (r = 0.40, n = 38, GFT percentage of Fast . The stocking rate depended mainly GFT P < 0.01). Including soil pH and moisture with EIVs on the proportion of the cut area which was topped in early increased the correlation (r = + 0.66, P < 0.001). This spring (r = +0.90; P = 0.077). However, stocking rate 136 M. Duru et al. (a) 7 2 = 0.53*** 0 20 40 60 80 100 (b) Fast (%) GFT 40 2 R = 0.36*** 020 406080 100 Fast (%) (c) GFT R = 0.44*** Figure 3. Relations between stocking rate and the percentage of GFT having an acquisitive growth strategy (Fast ), GFT 0 20406080 100 (a) (Y = 0.032X + 0.6); end of hay harvest date and the percentage Fast (%) GFT of GFTs having a late growth strategy (Late ), GFT (b) (Y = 21.6X + 388); data aggregated at the LMU level. Figure 2. Relations between the percentage of GFT having an acquisitive growth strategy (Fast ) and (a) standing herbage GFT mass (Y = −0.13X + 35.7), (b) species richness Within–between-field diversity (Y = 0.024X + 2.26) and (c) soil C content in the 05 cm layer (Y = −1.15X + 1.84). Vegetation heterogeneity can be assessed for sets of fields among farms or landscapes. Obviously, the more the within-field or between-field diversity, the more heteroge- positively correlated with N fertiliser rate and timing of the neous were the height and phenology (see Materials and first cut (r = +0.82; P < 0.05). Methods) of grasslands at both levels of organisation. For The flexibility component of management services example, the GFT with an acquisitive growth strategy was was represented as the mean percentage of Late at the GFT mapped for two farms (Figure A2). We observed that LMU level. It significantly correlated with the end date of almost 50% of the grassland area corresponds to fields the hay harvest (Figure 3b), because fields with high GFT with low GFT diversity (FastGFT: <30% or >70%), and diversity had longer durations of harvest operations (not that most of these grasslands are close together within the shown). Significant differences were observed in the per- landscape. centage of Fast among the three LMU types (cut, GFT At the farm level, cutting and cow grazing areas grazed by cows and by heifers) for six of the eight farms tended to have higher percentages of Fast in dairy GFT (Table 8). Except for farm D3, cut LMU had the highest farmsthaninbeef farms (Figure 4a). But this was not percentage of Fast (see Figure A1 for detailed descrip- GFT the case for heifers grazing areas. Within the range of tion of plant composition at the farm level). This indicates 4060% Fast ,the threeLMU typeswereobservedfor GFT that forage production and forage quality at the leafy stage both types of farm. In dairy farms, the heifer LMU (Fast ) were usually the highest for cut areas (except D3 GFT increased between-LMU differences due to its low per- and B1), followed by cow grazing areas (except for D1, centages of Fast , while for beef farms the cut LMU GFT B1, and B3). These data show consistent rankings of increased between-LMU differences due to it high per- animal feed requirements and type of vegetation allocated, centages of Fast .For theentiredataset,weconsid- GFT except for D1. There was a consistent effect of farm ered three components of plant diversity: Div ,Fast GFT GFT strategy on the percentage of Fast (P < 0.01) for all GFT and Sum (Figure 4b). Similar patterns between GFT LMUs (Table 8). Fast percentages were significantly GFT Div and Fast were observed regardless of the GFT GFT lower for beef farms. Among dairy farms and grazing percentage of grass species (Sum ), except for very GFT areas, there was a significant difference between Fast , GFT low Fast values. Comparison of analysis at field and GFT as was the case between farms for heifers. LMU levels (Figure 4a and b) shows that the LMU level –1 Soil C content (g/1000g) Species richness Standing herbage mass (t ha ) International Journal of Biodiversity Science, Ecosystem Services & Management 137 Table 8. Mean values and ANOVA of two indicators of grass- land composition (Fast and Late ) for the three land-use GFT GFT types (independent variables) considering each farm separately. Farm Df Land use Fast Late GFT GFT D1 17 cut area 71 25 cow grazing areas 38 23 heifer grazing areas 52 25 F-value (P-value) 2.3 (0.15) 0.9 (0.44) D2 21 cut area 67 37 cow grazing areas 50 37 heifer grazing areas 34 53 F-value (P-value) 7.18 (0.006) 3.2 (0.04) D3 19 cut area 84 10 cow grazing areas 87 13 heifer grazing areas 12 31 F-value (P-value) 28.6 (0.0007) 1.3 (0.24) D4 22 cut area 58 40 cow grazing areas 39 38 heifer grazing areas 10 43 F-value (P-value) 4.9 (0.02) 0.6 (0.73) B1 27 cut area 68 29 cow grazing areas 25 30 heifer grazing areas 53 36 F-value (P-value) 5.55 (0.01) 0.5 (0.78) B2 25 cut area 55 35 cow grazing areas 53 28 heifer grazing areas 39 37 F-value (P-value) 2.45 (0.10) 1.2 (0.45) B3 18 cut area 63 35 cow grazing areas 11 36 heifer grazing areas 19 44 Figure 4. Relations between within-field functional diversity F-value (P-value) 11.8 (0.003) 1.1 (0.51) (Div ) and the percentage of GFTs having an acquisitive B4 20 cut area 38 44 GFT growth strategy (Fast ); a: for data aggregated at the LMU cow grazing areas 1 34 GFT level (squares for cut grassland, circles and triangles for grassland heifer grazing areas 18 40 grazed by cows and heifers, respectively, open symbols for beef F-value (P-value) 8 (0.0002) 1 (0.55) farms, closed symbols for dairy farms) Notes: percentage in biomass; significant differences among land use within (Y = 0.0001X + 0.009X + 0.52); b: for all grassland fields, the a farm are indicated in bold. percentage of grass species in herbage biomass: >80% (closed Fast and Late are the percentage of GFTs having a fast and a late GFT GFT circles), 60–80% (crosses) or <60% (open circles). growth strategy, respectively; D: dairy farms; B: beef farms. in Duru, Cruz, Jouany, et al. (2010), or stocking rate, greatly structures grassland diversity. In other words, R = 0.78 in this paper), whereas management and envir- grassland diversity is greater between LMUs than within onmental drivers can explain only 44% or 59% of the them. Thus, differences in farm orientation and stocking variance, respectively, depending on whether the explana- density are required to maintain a grassland mosaic at the tory variables are observed or measured. Correlations landscape level (Figure A2 gives an example for two found elsewhere using plant traits instead of plant groups farms). are no higher (e.g. Lavorel et al. 2011; Duru et al. 2012). This justifies using plant functional types as a key tool for managing grassland functional composition and predicting Discussion ES. However, the method used to characterise grassland A plant functional-type-based MF for linking plant functional composition is highly simplified in com- management and ES parison to the measurement of plant traits. As is known, plant traits, plant functional groups or their proxies must Our results show that a MF based on a simplified char- be aggregated to assess certain ES, such as those provided acterisation of plant functional types allows a large set of by the landscape mosaic (Lavorel et al. 2011). However, ES at different organisational levels to be assessed, which aggregation does not have to reach the landscape level to is better than considering drivers alone. Regardless of the be able to understand effects of management and policies level considered, strong and weak correlations were found on within–between-grassland diversity. Thus, we found between plant functional composition and services (espe- that the LMU level performed better than the farm level cially forage services) and drivers, respectively. for understanding in depth the degree of heterogeneity in Regression analysis has shown that Fast alone can GFT management practices that impact ES. predict services well (e.g. herbage production, R = 0.69 138 M. Duru et al. The main advantage of the MF proposed is its ease of Based on the farm sample studied, we found that even a use by stakeholders. Its strength lies in not being limited to single farm can contain a wide range of within–between-field production services, as traditionally done in agronomy. functional plant diversity (Rudmann-Maure et al. 2008)and The concept of plant functional types is key because it is that contrasting land use within a farm can create a diversity appropriate for evaluating or predicting a wide range of ES of plant species as wide at observed at the landscape level. As and makes sense for farmers and other stakeholders. observed in a different context (Beyene et al. 2006), plant Farmers give positive feedback when their land is depicted functional type assemblages are the result of deliberate man- through plant functional types in a bar graph (Duru et al. agement choices resulting from farm enterprise type (Brodt 2011a) or a map (Figures A1 and A2, respectively). et al. 2006) and from assets and constraints such as available Moreover, several agricultural consultants have adopted facilities and field topography (Andrieu et al. 2007; Valbuena it, at least partly, to design grassland typologies at national et al. 2008;Martinetal. 2009). Differences in plant func- (Launay et al. 2011) or regional (Carrère et al. 2012) tional diversity at the LMU level are the result of land use and levels. They have a great interest in the MF because of farm enterprise type. Usually, cut grasslands have the highest its ability to reduce a large list of species into a small percentage of acquisitive types, first because they receive number of plant functional types in an effective commu- more fertiliser and have consistently higher N-EIV, and sec- nication tool. It addresses four key components of forage ond due to the direct effect of management practices on the services that fit well with farmers’ expectations (Duru, percentage of acquisitive types (Table 7). Dairy farms have a Cruz, Jouany, et al. 2010). higher percentage of acquisitive plant types for both cut and Below, we summarise and discuss our main findings grazed areas, which is consistent with the highest digestibil- about relations between grassland functional composition, ity of these plant types (Duru, Cruz & Theau 2010). management (next section) and ES, and examine trade-offs Between-farm comparisons can show whether the potential and synergies between ES (last section). exists to reduce the cost of feedstuffs. For example, the dairy farm D4 has similar milk production per cow (around 5000 kg per year) even though the percentage of Fast GFT Drivers of grassland functional composition and differed greatly: it was highest for D3 and lowest for D4 management (Table 8). Since obtaining a high percentage of Fast GFT At field level, the grasslands with the highest percentage requires high fertiliser input, this indicates that production of acquisitive species (Fast ), as well as the greatest costs could be reduced if enough land were available, espe- GFT abundance of grass species, responded significantly to cially if it is grazed, so as not to increase the workload. The management and certain environmental drivers (e.g. MF also detects discrepancies, for example, for dairy heifers field elevation). Fast increasedwithincreasingnutri- on farm D3 that used high-quality herbage. GFT ent availability (Wilson et al. 1999) and decreased with Field characteristics may explain between-farm differ- increasing temperature (Roche et al. 2004), which was ences in grassland functional diversity. In less-favoured negatively correlated with field elevation (Figure 5,top). areas, many farm-dependent constraints may occur These results are consistent with studies demonstrating (Andrieu et al. 2007; Martin et al. 2009). These include: that the features used to distinguish GFTs (specific leaf area, LDMC) are appropriate indicators of stress, in gen- (1) The proportion of grassland fields located near the eral (Harrison et al. 2010). Stress factors for nutrient cowshed, which affects the stocking rate, at least availability and mean temperature act in the same man- for dairy systems. In this way, D3, which has ner, favouring acquisitive species when stress is low. Our fewer pastures near the cowshed, has a higher results confirm those observed for a small (Duru, stocking rate than D4 (3.3 vs. 2.0 animal units Ansquer, et al. 2010) and a large (Martin et al. 2009) per ha). number of sites. Disturbances modify the effects of (2) The availability of summer pastures for animals stress, either reducing or amplifying them. For given with low feed requirements (dry cows, replace- climatic and soil conditions, mowing promotes acquisi- ment heifers), because these usually unfertilised tive species, while grazing promotes conservative spe- areas widen the range of GFTs encountered at cies. In other words, mowing reinforces the positive the farm level. For example, dairy farms D3 and effect of temperature and N on the abundance of acqui- D4 can use summer pastures that usually never sitive species (Figure 5, top). For functional diversity receive fertiliser, while this is not the case for D1 (here Div ), previous research on the intermediate and D2. GFT stress hypothesis (Vonlanthen et al. 2006) supports the (3) The use of modern harvesting equipment (round- idea that maximum diversity was observed only when baller), which reduces dependence on the weather, simultaneously considering stress and disturbance fac- shortens the harvest duration, and thus the range tors, as we also found. Additionally, we show that the of vegetation types observed (Benton et al. 2003). direction of effect for stress factors depends on the cur- There is little variability in functional composition rent dominant plant strategy; for example, N fertiliser between mown grasslands, except for farms B2 could decrease or increase Div and B4, which mowed summer pastures. GFT. International Journal of Biodiversity Science, Ecosystem Services & Management 139 Figure 5. A framework for linking environmental and management factors to grassland functional composition and a set of ES; the shape of each triangle indicates the direction of expected effect; rectangles indicate that effects depend on stakeholder viewpoints; FD: functional diversity. The results of this study clearly show that, as White et al. different strategies for resource acquisition leads to higher (2004) suggested, there is greater uniformity in plant types input efficiency for herbage production, as suggested by within LMUs than between them. Balancing constraints Fornara and Tilman (2009) and Dybzinski et al. (2008), on and goals, there is room for different degrees of farm-level the basis of species richness. For management services, we diversity of plant types. This means that some farmers found that between-farm differences are related to the choose to favour a certain range of plant types for eco- allocation of vegetation types or forage resources to dif- nomic (e.g. to reduce inputs) or labour reasons. ferent animal groups to save nutrients or feedstuff costs; Nevertheless, we did not find an effect of within-field for example, the capacity exists to reduce N fertiliser functional diversity on stocking rate as shown at the field without affecting animal production for some land-use level (Weigelt et al. 2009). As claimed by Sanderson et al. types, as seen above. Furthermore, our MF provides a (2004), the evidence for diversity effects is equivocal for simple method for comparing within–between-field plant pasturelands. However, stocking rate is probably a too functional diversity. It could help stakeholders determine coarse variable to reveal the effect of within-field func- the degree of grassland heterogeneity to promote tional diversity on potential complementarities between (Lamarque et al. 2011). plant types. The synthetic representation of relations between grassland functional composition and ES allows stake- holders to examine the main trade-offs they must consider Evaluation of trade-offs and synergies between ES (Figure 5, bottom). For farmers, among forage and man- The MF based on GFTs allows the evaluation of a large agement services, an opposite trend occurred between number of ES at different organisation levels and the measured herbage production and recorded production and analysis of main trade-offs between services. We found timing (r = −0.51, n = 24, P < 0.01), with Fast GFT Late as proxies, respectively (from data used for that a set of ES can be evaluated at the field level and GFT Figure 3a and b). However, at the field level, a nonlinear above (Figure 5, bottom for Fast ). Fast was a good GFT GFT relation between herbage production and yield flexibility proxy for forage production measured at the field level, was observed. At the LMU level, such trade-offs were not thus confirming results obtained in other regions (Duru, usually a problem for farmers because the priority services Cruz, Jouany, et al. 2010) or assessed through the stocking depended mainly on the animal group (cow vs. heifer) or rate at the LMU level (this paper). For low N rates, we the land management (grazing vs. cutting) considered. verified that the coexistence of plant functional types with 140 M. Duru et al. effects of land management on ecosystem services. Ecol Thus, the diversity of animal groups on a farm and of farm Indic. 21:110–122. enterprises in a region lead to a diversity of vegetation Benton TG, Vickery JA, Wilson JD. 2003. Farmland biodiversity, is types in a landscape which is enhanced by environmental habitat heterogeneity the key? Trends Ecol Evol. 18:182–188. factors such as field aspect and elevation. This explains Beyene A, Gibbon D, Haile M. 2006. Heterogeneity in land why this kind of diversity in agriculture creates a mosaic resources and diversity in farming practices in Tigray, Ethiopia. Agric Syst. 88:61–74. of vegetation types within and between farms that directly Bossuet L, Duru M. 1992. Estimating herbage mass by the contributes to landscape attractiveness, which is important sward-stick method. Fourrages. 131:283–300. for tourism (Junge et al. 2011) and indirectly and more Brodt S, Klonsky K, Tourte L. 2006. Farmer goals and manage- broadly to multiple ES (Smukler et al. 2010)(Figure 5). ment styles, implications for advancing biologically based agriculture. Agric Syst. 89:90–105. Carrère P, Orth D, Chabalier C, Seytre L, Piquet M, Landrieaux J, Rivière J. 2012. Une typologie multifonctionnelle des Conclusion prairies des systèmes laitiers AOP du Massif Central combi- In less-favoured areas such as mountains, farms have nant des approches agronomiques et écologiques. Fourrages. 209:9–22. high grassland diversity due to diversity in abiotic factors Cash DW, Clark W, Alcock F, Dickson NM, Eckley N, Guston (elevation), farm orientation (beef, dairy), enterprise DH, Mitchell RB. 2003. Knowledge systems for sustain- (cows, heifers) and management (grazing, cutting). To able development. Proc Nat Acad Sci USA. describe ES provided by grassland diversity and their 100:8086–8091. underlying drivers, we have developed a MF based on Ceulemans T, Merckx R, Hens M, Honnay O. 2013. Plant species loss from European semi-natural grasslands follow- GFTs. Due to its ease of use and credibility, this MF ing nutrient enrichment – is it nitrogen or is it phosphorus? should help agricultural experts and farm advisors under- Glob Ecol and Biogeo. 22:73–82. stand implications of different management choices on Colasanti RL, Hunt R, Askew AP. 2001. A self-assembling grassland diversity and on a large set of ES noticeable by model of resource dynamics and plant growth incorporating farmers, tourists and other members of society. The field plant functional types. Funct Ecol. 15:676–687. Cruz P, Theau JP, Lecloux E, Jouany C, Duru M. 2010. level may be sufficient for assessing their impact on plant Typologie fonctionnelle de graminées fourragères pérennes, diversity, while the land-use type and farm levels are still une classification multitraits. Functional typology of peren- needed to understand the drivers of management prac- nial forage grasses: a classification based on several charac- tices. Our MF can also help local policy-makers who teristics. Fourrages. 401:11–17. intend to support biodiversity with subsidies based on Daget P, Poissonet J. 1971. Une méthode d’analyse phytosocio- logique des prairies [A method for analyzing grassland phy- stocking-rate thresholds calculated for a set of fields or tosociology]. Ann Agronom. 22:5–41. theentirefarm. Ourresults clearly show that field and De Bello F, Thuiller W, Leps J, Choler P, Clément B, Macek P, farm levels are too small and too large, respectively. The Sebastia MT, Lavorel S. 2009. Partitioning of functional LMU level seems to be the right level for gathering data diversity reveals the scale and extent of trait convergence for management and/or vegetation, then engaging discus- and divergence. J Veg Sci. 20:475–486. De DM V, De Boer T. 1959. Methods used in botanical grassland sion between beneficiaries of ES to identify trade-offs research in the Netherlands and their application. Herbage and synergies. Abstr. 29:1. Diaz S, Lavorel S, McIntyre S, Falczuk V, Casanoves F, Milchunas DG, Skarpe C, Rush G, Sternberg M, Noy-Meir Disclosure statement I, et al. 2007. Plant trait responses to grazing – a global synthesis. Glob Change Biol. 13:313–341. No potential conflict of interest was reported by the authors. Duru M, Ansquer P, Jouany C, Theau JP, Cruz P. 2010. Suitability of grass leaf dry matter content for assessing the response of grasslands to land use and fertility. Ann Bot. Funding 106:823–831. This work was supported by the French National Research Duru M, Cruz P, Jouany C, Theau JP. 2010. Herb’type ,un Agency O2LA (Organismes et Organisations Localement nouvel outil pour évaluer les services de production fournis Adaptés) ANR-09-STRA-08. par les prairies permanentes. Herb’type : a new support for assessing agricultural services provided by permanent grass- lands. Prod Animales. 23:319–332. Duru M, Cruz P, Jouany C, Theau JP. 2011a. Combiner des References recherches en agroécologie et des dispositifs participatifs AFNOR. 1994. LA qualité des sols, vol.1. Recueil des normes. pour construire des outils d’évaluation des prairies perma- La Défense, Paris: AFNOR. nentes. Combining agroecological research and participatory Andrieu N, Josien E, Duru M. 2007. Relationships between approach to build tools for evaluating permanent grasslands. diversity of grassland vegetation., field characteristics and Cah Agricult. 20:223–234. land use management practices assessed at the farm level. Duru M, Cruz P, Jouany C, Theau JP. 2011b. Optimiser les Agric Ecosys Env. 120:359–369. services fourragers des prairies permanentes: des outils Ansquer P, Duru M, Theau JP, Cruz P. 2009. Convergence in pour caractériser et gérer la diversité floristique [Internet]. plant traits between species within grassland communities Fascade. 4p. Available from: http://www.sad.inra.fr/ simplifies their monitoring. Ecol Ind. 9:1020–1029. Duru M, Cruz P, Theau JP. 2010. A simplified method for Ape VO, Petz K, Alkemade R, Hein L, de Groot RS. 2012. characterising agronomic services provided by species-rich Framework for systematic indicator selection to assess grasslands. Crop Pasture Sci. 61:420–433. International Journal of Biodiversity Science, Ecosystem Services & Management 141 Duru M, Jouany C, Le Roux X, Navas ML, Cruz P. 2013. A Lavorel S, Grigulis K. 2012. How fundamental plant functional plant functional type approach tailored for stakeholders trait relationships scale-up to trade-offs and synergies in involved in field studies to predict ecosystem services pro- ecosystem services. J Ecol. 100:128–140. vided by grasslands. Renew Agricult and Food Syst. 1–16. Lavorel S, Grigulis K, Lamarque P, Colace MP, Garden D, Girel doi:10.1017/S1742170513000306. J, Pellet G, Douzet R. 2011. Using plant functional traits to Duru M, Theau JP, Cruz P. 2012. Functional diversity of species- understand the landscape distribution of multiple ecosystem rich managed grasslands in response to fertility : defoliation services. J Ecol. 99:135–147. and temperature. Basic Appl Ecol. 13:823–831. Le Roux X, Barbault R, Baudry J, Burel F, Doussan I, Garnier E, Duru M, Theau JP, Hossard L, Martin G, Cruz P. 2011. Diversité Herzog F, Lavorel S, Lifran R, Roger-Estrade J, et al. 2008. de la composition fonctionnelle de la végétation au sein Agriculture et biodiversité, valoriser les synergies. Paris: Inra d’une prairie et entre prairies. Characterisation of within editions. and between grassland functional diversity. Fourrages. Lindemann-Matthies P, Junge X, Matthies D. 2010. The influ- 205:61–73. ence of plant diversity on people’s perception and aesthetic Dybzinski R, Fargione JE, Zak DR, Fornara D, Tilman D. appreciation of grassland vegetation. Biol Conserv. 143:195– 2008. Soil fertility increases with plant species diversity 202. doi:10.1016/j.biocon.2009.10.003 in a long-term biodiversity experiment. Oecologia. Lugnot M, Martin G. 2013. Biodiversity provides ecosystem 158:85–93. services: scientific results versus stakeholders’ knowledge. Ellenberg H, Weber HE, Düll R, Wirth V, Werner W, Paulissen Reg Env Change. 13:1145–1155. D. 1992. Zeigerwerte von Pflanzen in Mitteleuropa. Scr Martin G, Cruz P, Theau JP, Jouany C, Fleury P, Granger S, Geobotanica. 18:1–258. Faivre R, Balent G, Lavorel S, Duru M. 2009. A multi-site Ersten ACD, Alkemade JRM, Wassen MJ. 1998. Calibrating study to classify semi-natural grassland types. Agricult Ellenberg indicator values for moisture, acidity, nutrient Ecosyst Env. 129:508–515. availability and salinity in the Netherlands. Plant Ecol. Matthews KB, Rivington M, Blackstock K, McCrum G, Buchan 135:113–124. K, Miller DG. 2010. Raising the bar? – The challenges of Fallour D, Theau JP, Corler K, Hossard L, Martin G, Jouany C, evaluating the outcomes of environmental modelling and Duru M, Cruz P. 2008. A simplified method to determine the software. Env Model & Softw. 26:247–257. abundance of grass functional groups in natural grasslands. Nozières MO, Moulin CH, Dedieu B. 2011. The herd, a source of Biodiversity and animal feed – Future challenges of flexibility for livestock farming systems faced with uncer- Grassland production – 22nd General Meeting of the tainties? Animal. 5:1142–1157. European Grassland Federation; Uppsal, Sweden. 93–95. Pakeman R, Marriott C. 2010. A functional assessment of the Fisher BR, Turner K, Morling P. 2009. Defining and classifying response of grassland vegetation to reduced grazing and ecosystem services for decision making. Ecol Econ. 68:643–653. abandonment. J Veg Sci. 21:683–694. Fornara DA, Tilman D. 2009. Ecological mechanisms associated Petchey OL, Gaston KJ. 2006. Functional diversity, back to with the positive diversity-productivity relationship in a N- basics and looking forward. Ecol Lett. 9:741–758. limited grassland. Ecology. 90:408–418. Power AG. 2010. Ecosystem services and agriculture, tradeoffs Frank S, Fürst C, Koschke L, Witta A, Makeschin F. 2013. and synergies. Phil Trans Royal Soc B, Biol Sci. Assessment of landscape aesthetics—validation of a land- 365:2959–2971. scape metrics-based assessment by visual estimation of the Quetier F, Lavorel S, Thuiller W, Davies AI. 2007. Plant-trait- scenic beauty. Ecol Indic. 32:222–231. doi:10.1016/j. based modeling assessment of ecosystem service sensitivity ecolind.2013.03.026 to land use change. Ecol Appl. 17:2377–2386. Gross N, Suding KN, Lavorel S, Roumet C. 2007. Raudsepp-Hearne C, Peterson GD, Bennett EM. 2010. Complementarity as a mechanism of coexistence between Ecosystem service bundles for analyzing tradeoffs in diverse functional groups of grasses. J Ecol. 95:1296–1305. landscapes. PNAS. 107:5242–5247. doi:10.1073/ Harrison S, Prentice I, Barboni D, Kohfeld K, Ni J, Sutra JP. pnas.0907284107. 2010. Ecophysiologicoal and bioclimatic foundations for a Roche P, Diaz-Burlinson N, Gachet S. 2004. Congruency analy- global plant functional classification. J Veg Sci. sis of species ranking based on leaf traits, which traits are the 21:300–317. more reliable? Plant Ecol. 174:37–48. Hooper DU, Ewel JJ, Hector A, Inchausti P, Lavorel S, Lawton Rodríguez-Ortega T, Oteros-Rozas E, Ripoll-Bosch R, Tichit M, JH, Lodge D, Loreau M, Naeem S, Schmid B, et al. 2005. Martín-López B, Bernués A. 2014. Applying the ecosystem Effects of biodiversity on ecosystem functioning, a consen- services framework to pasture-based livestock farming sys- sus of current knowledge and needs for future research. Ecol tems in Europe. Animal. 1–12. doi:10.1017/ Monogr. 75:3–36. S1751731114000421 Junge X, Lindemann-Matthies P, Hunziker M, Schüpbach B. Rudmann-Maure K, Weyand A, Fischer M, Stöcklin J. 2008. The 2011. Aesthetic preferences of non-farmers and farmers for role of land use and natural determinants for grassland vege- different land-use types and proportions of ecological com- tation composition in the Swiss Alps. Basic Appl Ecol. pensation areas in the Swiss lowlands. Biol Conserv. 9:494–503. 144:1430–1440. Sanderson MA, Skinner RH, Barker DJ, Edwards GR, Tracy BF, Lamarque P, Tappeiner U, Turner C, Steinbacher M, Bardgett Wedin DA. 2004. Plant species diversity and management of RD, Szukics U, Schermer M, Lavorel S. 2011. Stakeholder temperate forage and grazing land ecosystems. Crop Sci. perceptions of grassland ecosystem services in relation to 44:1132–1144. knowledge on soil fertility and biodiversity. Reg Env Sanderson MA, Archer D, Hendrickson J, Kronberg S, Liebig M, Change. 11:791–804. Nichols K, Schmer M, Tanaka D, Aguilar J. 2013. Launay F, Baumont R, Plantureux S, Farrie JP, Michaud A, Pottier Diversification and ecosystem services for conservation E. 2011. Prairies permanentes, des références pour valoriser agriculture: outcomes from pastures and integrated crop– leur diversité. éd. Paris: Institut de l’Elevage; 128 p. livestock systems. Renewable Agricult and Food Syst. Lavorel S, Garnier E. 2002. Predicting changes in community 28:129–144. composition and ecosystem functioning from plant traits, Schaffers AP, Sykora KV. 2000. Reliability of Ellenberg indi- revisiting the Holy Grail. Funct Ecol. 16:545–556. cator values for moisture, nitrogen and soil reaction, a 142 M. Duru et al. comparison with field measurements. J Veg Sci. Van Der Biest K, D’Hondt R, Jacobs S, Landuyt D, Staes J, 11:225–244. Goethals P, Meire P. 2014. EBI: an index for delivery of Smukler SM, Sánchez-Moreno S, Fonte SJ, Ferris H, ecosystem service bundles. Eco Indicat. 37:252–265. Klonsky K, O’Geen AT, Scow KM, Steenwerth KL, Vojtech E, Loreau M, Yachi S, Spehn EM, Hector A. 2008. Light Jackson LE. 2010. Biodiversity and multiple ecosystem partitioning in experimental grass communities. Oikos. functions in an organic farmscape. Agric Ecosys Env. 117:1351–1361. 139:80–97. Vonlanthen CM, Kammer PM, Eugster W, Bühler A, Veit H. Sun S, Frelich LE. 2011. Flowering phenology and height 2006. Alpine vascular plant species richness, the importance growth pattern are associated with maximum plant of daily maximum temperature and pH. Plant Ecol. height, relative growth rate and stem tissue mass 184:13–25. density in herbaceous grassland species. J Ecol. 99: Weigelt A, Weisser WW, Buchmann N. 2009. Biodiversity for 991–1000. multifunctional grasslands , equal productivity in high-diver- Swinton SM, Lupi F, Robertson GP, Hamilton SK. 2007. sity low-input and low-diversity high-input systems. Ecosystem services and agriculture: cultivating agricultural Biogeosciences. 6:1695–1706. ecosystems for diverse benefits. Ecol Econ. 64:245–252. White TA, Barker DJ, Moore KJ. 2004. Vegetation diversity, doi:10.1016/j.ecolecon.2007.09.020 growth, quality and decomposition in managed grasslands. Theau JP, Cruz P, Fallour D, Jouany C, Lecloux E, Duru M. Agric Ecosys Env. 101:73–84. 2010. Une méthode simplifiée de relevé botanique pour une Wilson PJ, Thompson K, Hodgson JG. 1999. Specific leaf area caractérisation agronomique des prairies permanentes. and leaf dry matter content as alternative predictors of plant Fourrages. 401:19–25. strategies. New Phytologist. 143:155–162. Valbuena D, Verburg PH, Bregt AK. 2008. A method to define a Zhang W, Ricketts TH, Kremen C, Carney K, Swinton SM. typology for agent-based analysis in regional land-use 2007. Ecosystem services and disservices to agriculture. research. Agricult Ecosys Env. 128:27–36. Ecol Econ. 64:253–260. International Journal of Biodiversity Science, Ecosystem Services & Management 143 Appendix Cut area 100% 80% 60% 40% 20% 0% D3 D1 B1 D2 B3 D4 B2 B4 Area grazed by cows 100% 80% 60% 40% B 20% 0% D3 D4 B2 D2 D1 B1 B3 B4 Area grazed by heifers 100% 80% 60% 40% 20% 0% B1 D1 B2 D2 B3 B4 D3 D4 Figure A1. Bar graphs for five GFTs ranked from fast growth strategy (A) to low growth strategy (D) for the dairy (Di) and beef (Bi) farms. Data were averaged for the whole grassland fields having the same use: cutting, grazing by cows and heifers. GFT (%) GFT (%) GFT (%) 144 M. Duru et al. Figure A2. Maps of vegetation diversity for two farms located in the same district. The four colours correspond to different percentages of GFT having an acquisitive growth strategy: w30% (white), 30–50 (light grey), 50–70(dark grey), >70% (black); the small map shows the location of the studied area within France. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png International Journal of Biodiversity Science, Ecosystem Services & Management Taylor & Francis

A methodological framework to facilitate analysis of ecosystem services provided by grassland-based livestock systems

Download PDF
17 pages

Loading next page...
 
/lp/taylor-francis/a-methodological-framework-to-facilitate-analysis-of-ecosystem-gGUWkDAGwA

References (80)

Publisher
Taylor & Francis
Copyright
© 2015 Taylor & Francis
ISSN
2151-3732
eISSN
2151-3740
DOI
10.1080/21513732.2015.1030695
Publisher site
See Article on Publisher Site

Abstract

International Journal of Biodiversity Science, Ecosystem Services & Management, 2015 Vol. 11, No. 2, 128–144, http://dx.doi.org/10.1080/21513732.2015.1030695 A methodological framework to facilitate analysis of ecosystem services provided by grassland- based livestock systems a,b, a,b a,b M. Duru *, J.P. Theau and G. Martin a b INRA, UMR1248 AGIR, F-31320 Castanet-Tolosan, France; Université Toulouse, INPT, UMR AGIR, F-31029 Toulouse, France (Submitted 27 February 2014; accepted 26 February 2015; edited by Rob Alkemade) It remains challenging to describe ecosystem services (ES) provided by grassland diversity and their underlying drivers. Issues for characterising them are related to scale, knowledge and outcomes that are stakeholder-dependent. Forage production interests farmers, while C sequestration, species richness and the landscape mosaic interest society. To address these issues, we developed a methodological framework (MF) based on five grass functional types (GFTs) which enables usable indicators to be easily defined, such as the percentage of plants with a fast (Fast ) growth strategy in grasslands. The GFT MF consists of characterising plant functional diversity at field and farm levels, analysing its response to environmental and management drivers, and its effects on ES. The MF is applied to eight farms differing in their orientation and in their management intensity. Fast responds positively to the amount of fertiliser supplied and negatively to field elevation. At the GFT field level, Fast is positively correlated with herbage production and negatively with soil C content and species richness. GFT Within-farm grassland diversity of Fast allows examination of how animal feed requirements match available resources GFT and the landscape mosaic created. Our MF addresses grassland diversity with indicators derived from GFT, which allows summarising relations between environmental and management drivers and ES, then examines trade-offs between ES. Keywords: landscape; management; plant functional type; provision services; supporting services; trade-offs Introduction known, even though farmers promote it. For instance, Martin et al. (2009) show that functional plant diversity Policy-makers at European and regional levels are keen to in grasslands provides flexibility in the timing of grassland encourage more sustainable livestock systems, especially use. Similarly, Nozières et al. (2011) suggest that within- to reduce their environmental impacts and to enhance farm plant diversity decreases production costs by more ecosystem services (ES). This strategy encompasses max- closely matching grassland types to animal feed require- imising provision of ES by farms (Power 2010). In a ments, thereby improving stability of the livestock system. farming context, three main types of ES are distinguished Second, most grassland research produces results suitable (Swinton et al. 2007; Zhang et al. 2007): (i) services from for understanding effects of drivers on ES but struggles to agriculture, that is, provisioning services (e.g. forage or produce results that are easily used by stakeholders milk), (ii) non-market services, some authors distinguish- (Matthews et al. 2010). Thus, particular attention should ing regulating services (e.g. C storage) and cultural ser- focus on building relevant and appropriate methodological vices (e.g. attractive landscapes), and (iii) services to frameworks (MFs) that produce salient, legitimate and agriculture, that is, several supporting and regulating ser- credible information (Cash et al. 2003). Our objective is vices called ‘input services’ (Le Roux et al. 2008) because to evaluate strengths and weaknesses of a MF based on they allow a decrease in chemical inputs. Better under- characterising plant functional diversity to evaluate ES and standing of trade-offs and synergies among services rela- identify their main drivers, while addressing the above tive to their beneficiaries (e.g. farmers, society) is essential issues and requirements. for management and policy decisions (Fisher et al. 2009). On grassland-based livestock farms, ecosystem proper- Studying services provision, trade-offs and synergies in ties that provide ES depend largely on plant functional grassland-based livestock systems raises several issues. diversity (the presence or abundance of particular func- First, services are scale-dependent (Lavorel & Grigulis tional groups or traits) rather than on species diversity 2012) and are studied most at field and landscape levels (Hooper et al. 2005). Thereby there is growing consensus (Benton et al. 2003). The farm level is under-studied even that a plant trait-based methodology may well be able to though it is the level at which land-use and management address issues such as grassland ecophysiology and help decisions are made. Previous in-depth analysis of grass- manage some of the services that grasslands deliver to land-based livestock farms shows that biodiversity can humans (e.g. Diaz et al. 2007). The key hypothesis is also provide a management service, enabling farmers to that traits can simultaneously explain individual plant improve their management and working conditions responses to biotic and abiotic factors, with ecosystem (Lugnot & Martin 2013). However, this ES is not well *Corresponding author. Email: mduru@toulouse.inra.fr © 2015 Taylor & Francis International Journal of Biodiversity Science, Ecosystem Services & Management 129 properties underpinning ecosystem functions (Lavorel & examine management and environmental factors that Garnier 2002) and associated services (Lavorel et al. shape grassland functions and then analyse ES provided 2011). Plant functional diversity in structure and composi- by grasslands at the field, LMU and farm/landscape levels. tion is usually defined through two components: commu- Finally, we discuss which drivers have the greatest effect nity-level weighted mean of trait values or functional types on grassland functions and propose a synthetic representa- (i.e. groups of species sharing the same set of attributes) tion of relations between management and environmental (Colasanti et al. 2001) and functional divergence (FD) drivers, grassland functional structure and composition (Lavorel et al. 2011). FD describes trait diversity among and ES, which allows characterisation of trade-offs and species (or functional groups) that coexist within a com- synergies between ES for different beneficiaries. munity (De Bello et al. 2009). It is hypothesised to operate through functional complementarity (Petchey & Gaston 2006). For example, within-community diversity in plant Materials and methods height is expected to improve light capture (Vojtech et al. Development of the MF 2008), while diversity in leaf structural and chemical traits Linking driving forces, grassland functions and ES would reflect diversity in nutrient acquisition and retention We adapted a framework linking driving forces, grassland strategies (Gross et al. 2007). To characterise the relation functions, ES and beneficiaries of services developed by between biodiversity and ES, it is important to define Ape et al. (2012)(Figure 1). ES provided by grasslands indicators of ecosystem function that indicate the extent were distinguished according to the level at which they to which an ES can be provided (Ape et al. 2012), then were noticeable (field, farm and landscape) each of them ensure that stakeholders can easily use them. Therefore, being specific beneficiaries (Figure 1). For example, some several researchers (e.g. Van Der Biest et al. 2014) state management services are noticeable only at LMU or farm that there is a demand for simple MFs relying on indica- levels (emergent properties), while forage-production ser- tors to evaluate services. These indicators can be land-use vices are noticeable both at field and LMU levels proxies (Van Der Biest et al. 2014) or can be based on (Table 1). When farm level is considered, it is as a grassland functional structure and composition. We whole farm (Smukler et al. 2010), while the sub-levels explore the strength of the latter option. are key for understanding management services (Duru Methods requiring plant trait measurements are too et al. 2013). Furthermore, we distinguished the level at time-consuming and unfamiliar to stakeholders. which services are noticeable from the level at which data Therefore, we developed and used a simplified MF based should be recorded to assess them. For forage services the on grass functional types (GFTs) as indicators, in which levels can be similar, whereas for C storage the levels are stakeholders can easily characterise ES provided by grass- different (Table 1). In our study, we focus on field and land diversity in grassland-based livestock systems: for- farm levels. Upscaling, which was not considered in this age, management, input, environmental (species richness paper, requires integrating data (e.g. for C storage) or and soil C storage) and cultural aspects (aesthetic through spatially explicit modelling (e.g. for landscape attractive- characterisation of within–between-grassland diversity). ness). Farm orientation has only an indirect effect on To evaluate its suitability for addressing the above issues grassland functions through land use and management, and requirements, the methodology was applied to a range while environmental factors (e.g. climate, field and soil of management and environmental factors at different characteristics) have direct and indirect effects on grass- levels of organisation: field, land management unit land functions (Figure 1). (LMU) (i.e. parts of farms allocated to single groups of Trade-offs between services can occur for the same animals corresponding to single management units for beneficiary (e.g. forage and management services for production, feeding, health care, etc.) and a set of fields farmers) or between beneficiaries (e.g. forage services for at farm or landscape levels. The field level is needed to farmers, C storage for society). Arbitrating between ser- characterise the response of plant functional composition vices requires developing a small number of relevant to environmental factors and management practices. The indicators of grassland functions for assessing multiple LMU level is needed because averaging data at the farm services. For this, we distinguish grassland functions (i.e. level loses possible within-farm differences due to differ- the capacity of grassland ecosystems to provide services) ences in the management of animal groups (e.g. cows and from ES, which contribute to human well-being heifers). The level of a set of farm or landscape fields is (Raudsepp-Hearne et al. 2010). needed to assess services related to between-grassland field diversity. To cover a large range of management factors, the MF was applied to dairy and beef farms Selection of indicators of grassland functional structure differing in farming intensity, which influences grassland and composition vegetation (Andrieu et al. 2007). The paper is organised as follows. First, we describe Although credible, feedback from stakeholders shows that development of the MF, especially the choice of indicators the trait-based methodology to characterise plant func- of grassland functional structure and composition, and of tional diversity is not relevant in practice because it is management and environmental drivers. Second, we too time-consuming (Duru, Cruz & Theau 2010). To be 130 M. Duru et al. Figure 1. Relations between ES noticeable at different levels, their drivers and beneficiaries; dotted arrows indicate relations that were not studied. Table 1. Mapping of grassland properties to grassland services. Services Organisational level Indicators Types component at which services are for recording data of grassland functions of services provided noticeable for (potential services) by grassland (used targeted services) beneficiaries †† ‡ Forage Production and quality LMU, field (farmer) Field; LMU Fast (aggregated at Stocking rate GFT LMU level) Management Flexibility allowed by LMU, field (farmer) LMU (field ) Late (aggregated Spreading of harvest GFT temporality of at LMU level) dates herbage growth Adequacy between LMU (farmer) LMU/ farm GFT distribution Degree at which forage forage characteristics regarded to animal characteristics match and animal feed feed requirement and animal feed requirements inputs (fertiliser, requirements labour) (aggregated at LMU level) Input Fertility Field (farmer) Field Div N uptake Siomass production/ N GFT permitted by within- supplied field diversity (S) Environmental C sequestration Landscape Field Fast Soil C content GFT Species richness Field -> landscape Field Fast Sum Number of species GFT; GFT (society) Cultural Field mosaic Field -> landscape Set of fields, Fast × Div Visual within and GFT GFT (society) landscape mapped at different between fields levels diversity Notes: level not considered in this study. ‡reflect the forage production and its use intensity. Fast Late are the percentage of GFTs having a fast and a late growth strategy respectively. GFT GFT Sum is the percentage of grass species in biomass in a grassland. GFT relevant to decision-makers (Cash et al. 2003), that is, for characterising grassland vegetation (Duru, Theau, et al. the practical use of knowledge by farmers, we previously 2011). We classified grass species into functional types made a major change in the methodology for (Colasanti et al. 2001) rather than using continuous traits, International Journal of Biodiversity Science, Ecosystem Services & Management 131 even though the latter are expected to better represent The second management service studied depends on how changes in the intensity of processes (Lavorel & Garnier the farmer organises grassland diversity to match animal 2002). Based on leaf and phenological plant traits, five (A, feed requirements (Duru, Theau, et al. 2011). Typically, B, b, C and D) elementary GFTs were defined (Cruz et al. such a service can be assessed only at the LMU level. In 2010). These GFTs were ranked by leaf dry matter content fact, farmers combine fields allocated to different forage (LDMC), which increases from type A to D, and flowering crops into several assemblages, with each single assem- date, which is latest for types b and D (Duru et al. 2013). blage allocated to feed a particular herd batch. The number When combined, these elementary types reflect two major and nature of such assemblages are designed to meet growth strategies (Sun & Frelich 2011): fast (Fast :A objectives, such as to increase self-sufficiency of the sys- GFT and B) vs. slow (Slow : b, C and D) and early tem (i.e. the ratio of forage production to consumption), GFT (Early : A, B and C) vs. late (Late : b, D), which reduce operational costs or increase flexibility in organis- GFT GFT are meaningful for farmers (Duru et al. 2013). Based on ing work. Hence, we chose to investigate the farm as a set the observation that grass and dicotyledonous species that of such assemblages. To assess this, we ranked animal coexist in grassland communities display similar or con- feed requirements according to the targeted herbage qual- stant differences in plant-trait values (e.g. Ansquer et al. ity (assessed by Fast averaged at the LMU level): beef GFT (2009) for six plant traits and several ecosystem properties cow < milk cow for grazing, hay or silage; heifer < cow (Duru, Cruz & Theau 2010)), we focused on GFTs (for dairy farms). Assessing this management service expressed as the percentage of grass species in the herbage requires examining within–between-farm GFT distribution mass (Sum ). Consequently, we consider dicotyledo- of animal feed requirements and inputs (e.g. fertiliser, GFT nous species only as a whole to estimate their overall labour) as potential service. Therefore, we first compared impact on plant-community properties (Duru, Cruz & plant functional composition of LMUs within each farm, Theau 2010). FD (Div ) was estimated as the relative then examined whether there was a between-farm effect of GFT percentages of GFTs: the farm orientation or stocking density on each of the within-farm LMU effects. This can indicate whether a specialisation of plant types exists for certain land-use Div ¼ 1  p ; GFT i types, that is, grasslands fulfilling the same function in i1 the feeding system. For grasslands, input services are mainly related to the where p is the percentage of each of the coexistence of plant functional types or groups with differ- five GFTs; p ¼ 100: ent growth strategies within a community (e.g. Div ). It GFT was observed that this improves nutrient or light capture (e.g. Fornara & Tilman 2009). Focusing on fertility, we Selection of indicators of ES provided by grasslands examine if plant functional type diversity should result in Based on the four indicators of grassland functional compo- producing the same biomass with less fertiliser. We sition (Fast , Late ,Sum and Div ), forage ser- hypothesised that fertility services would be highest for GFT GFT GFT GFT grasslands with high within-field functional diversity. To vices were previously assessed at the plant-community level test this, we compared N uptake of grasslands with different (Duru, Ansquer, et al. 2010;Duruet al. 2013). Forage Div . production and herbage quality at the leafy stage are corre- GFT lated with the percentage of Fast , which have high growth For regulating services, we examined relations GFT rates and low lignin content. In this paper, we examine between Fast and soil C content and species richness. GFT whether these indicators (or combination of them) can be Fast is an indicator of weighted LDMC (Pakeman & GFT used to predict a larger set of services based on the Zhang Marriott 2010), which is correlated with soil C content. et al. (2007) classification, that is, ES from agriculture (here, Additionally, species richness is expected to decrease with forage production, regulating services and aesthetic value) decreasing nutrient availability (Ceulemans et al. 2013), and ES to agriculture (here, management and input services), for which Fast is also an indicator (Duru et al. 2013). GFT paying attention to the level(s) at which their assessment is Conversely, the likelihood of increased dicotyledonous most relevant. For each service, Table 1 lists the indicator species richness is strong when Sum is low, because GFT used for assessing potential services and the services used. grasslands usually have more dicotyledonous species than For forage production, in addition to measurements of stand- grass species. ing herbage mass at the leafy stage, we used the stocking rate In an agricultural context, permanent grasslands have calculated at the LMU level as a proxy. intrinsically high aesthetic value (Sanderson et al. 2013). Management services comprise two components. We Some vegetation features at field level such as flower assume that the percentage of Late is an indicator for diversity (Quetier et al. 2007) or vegetation-related feature GFT the timing flexibility for grassland use, because GFTs (landscape heterogeneity) are attractive for humans due to growing late in the season offer a large time window for their aesthetic value (Lindemann-Matthies et al. 2010). harvesting forage. Thus, we examine the relation between Although aesthetic preferences are highly subjective Late (potential service) and the range of harvest dates (Rodríguez-Ortega et al. 2014), we postulate that the het- GFT on farms throughout the growing season (used service). erogeneity of vegetation in phenology, height and colour at 132 M. Duru et al. different spatial levels (from within-field to a set of fields) Case study can provide raw data for assessing landscape attractiveness. Description We, thus, used two of the above indicators (Sum and GFT The study was performed in the Aubrac region in the Div ) to assess landscape diversity, especially its aes- GFT southern part of the French Massif Central (44.68°N, thetic component (Frank et al. 2013), at different levels. 2.85°E). The study area occupies approximately 40 km×20 km. Most of the area consists of unsown permanent grasslands used for dairy and beef livestock Selection of indicators of management and environmental systems. To ensure a large range of management and drivers environmental drivers (Figure 1), we chose farms with The strength of relations between drivers and services contrasting land-use types, with low or high stocking depends on the accuracy with which driver gradients are rates, and for which grassland fields range from 800 to characterised. They remain difficult to assess accurately, 1400 m a.s.l. To choose farms, we performed a pre-survey especially for managed grasslands, because most real with advisors and used data from the grey literature to situations are complex, differing in climate, soil and man- ensure a wide range of farm diversity. We preferred ana- agement (defoliation and fertilisation). These differences lysing a small number of greatly different farms in depth are reflected in the intensity of stress (mainly nutrients and rather than a larger number of representative farms in less temperature, even in the same biogeographic area) and depth. Thus, a total of eight farms, four beef (B) and four disturbance (defoliation intensity, frequency and time) dairy (D), were chosen. For each land-use type, two types that grasslands experience. Choice of management indica- of farm were distinguished according to their stocking tors was based on previous studies of permanent grass- rate. For example, farms 1 and 2 had higher stocking lands (Duru, Ansquer, et al. 2010; Duru et al. 2013). rates than farms 3 and 4 (Table 2). Numbers of cows and Plant nutrient availability and soil conditions (moisture heifers were converted into livestock units (LUs) on the content and pH) were assessed with the Ellenberg index basis of their live weights: a cow (650–700 kg live weight) (EIV, Ellenberg et al. 1992), which characterises species’ corresponded to 1 LU, while a heifer corresponded to 0.8 nutrient (N-EIV), soil moisture (M-EIV) and soil reactivity LU (2–3 years old) or 0.6 LU (1–2 years old). (R-EIV) preferences on a scale from 1 to 9. N-EIV was shown to be a better indicator of fertility than one based on plant N content, which is short-term and environment- Surveys, measurements and observations dependent (Duru et al. 2011a). Observations of species First, farms were surveyed for their main characteristics abundance allowed abundance-weighted EIVs to be calcu- (land area and production type) and management prac- lated for the entire grassland. The advantage of EIVs is tices. Grazing animals were categorised either as animals that they integrate plant species behaviour over many for production (milking cows in dairy systems, heifers, years (Schaffers & Sykora 2000). Furthermore, N-EIV is calves) or as animals for replacement or outside their strongly correlated with P and K availability in the topsoil production period (heifers and dry cows in dairy systems; (Ersten et al. 1998; Schaffers & Sykora 2000). beef suckler cows if calves are not sold at 9 months of age To characterise defoliation regimes, sward height was in beef systems, hereafter referred to as replacement ani- measured with a sward stick (Bossuet & Duru 1992) just mals). The feeding calendar for an average year was drawn before the farmer used the field for cutting or grazing. up, and spring and early summer land use was mapped Sixty measurements were collected for each grassland (Table 2). Then, field topology and topography (e.g. area, field. Cumulative daily temperature (Tsum) for each field elevation, distance to the cowshed, ease of access for cows was calculated from 1 February up to the field’s first use coming from the cowshed or a neighbouring field, (Duru, Ansquer, et al. 2010). Table 2. Characterisation of farm structure and management. Characteristics D1 D2 D3 D4 B1 B2 B3 B4 Structure Grassland area (ha) 54 58 77 69 105 70 115 160 Animal units (total) 60 56 70 52 106 78 87 117 Animal units (cow) 39 38 48 35 70 55 60 80 Stocking density (animal unit/ha) 1.1 1.1 1 0.8 1 1.1 0.8 0.7 Parcel area (ha) 2.5 1.8 3 2.5 3.2 2.5 5.7 6.9 Hay Cutting area/animal unit 0.4 0.45 0.44 0.43 0.37 0.4 0.41 0.36 Topping (%) 24 39 25 0 49 57 46 62 N fertilisation (kg/ha) 40 31 80 120 160 73 97 69 Summer grazing Stocking rate (animal unit/ha) 1.1 1.3 1.8 1.3 1.8 1.3 1.1 0.9 Spring grazing Turnout (DD) 397 410 410 410 450 450 320 450 N fertilisation (kg/ha) 14 29 46 63 15 24 0 0 Mixed diet in spring (days) 27 20 7 15 20 7 10 25 Notes: DD: degree-days from February1st; D: Dairy; B: Beef; 1, 2, 3, 4: farm number. International Journal of Biodiversity Science, Ecosystem Services & Management 133 Table 3. Measurement of driving forces affecting the provision of ES. Driving forces Indicator Level of analysis Measurement Farm orientation Dairy vs beef Farm Constituting of farm sampling and strategy Stocking rate Farm Constituting of farm sampling Animal performances objectives; farmer life style Not considered Environmental Topology Field Field distance from the cowshed factors Altitude, area Survey Soil moisture and reactivity Field Ellenberg index for moisture and reactivity Land management Fertilisation (amount) Field Survey Nutrient availability Field Ellenberg index for nitrogen Cutting management (date, topping) Field Survey Cutting management (height) Field Field measurement Turnout (date) LMU Survey Stocking rate (LU per ha) LMU Calculation from survey data Mixed diet in spring (days) Group of animals Calculation from survey data suitability for mechanisation) were recorded on the 169 determined by oxidation with potassium dichromate and fields of the 8 farms. Data were then mapped according to sulphuric acid (NF ISO 14235) (details of the normative amount of fertiliser applied (organic and inorganic), nature reference in AFNOR (1994)). The original method of De of the first use (e.g. grazing or cutting only, early grazing DM and De Boer (1959) was used to determine the that removes the apexes (topping)), date of first use and exhaustive botanical composition. type of grazing animals. To analyse factors affecting pro- ductivity, mean amounts of mineral N fertiliser and solid or liquid manure applied to each grassland field were Statistical analysis converted into kg, N applied per year. Data were recorded to describe farm orientation, farmer objectives, environ- ANOVA was performed at the field level to examine mental factors and management practices (Table 3). whether there was a significant effect of fertilisation rate A simplified adaptation of de De DM and De Boer and main use (grazing vs. cutting) on the three compo- (1959) method was used to characterise the plant func- nents of grassland functional composition (Fast , GFT Late and Sum ). Data expressed as percentages tional composition of the 169 grassland fields. One diag- GFT GFT were log-transformed to satisfy the conditions for onally orientated transect representative of vegetation ANOVA. Regression analyses were performed to express diversity was sampled between May and the beginning components of grassland functional composition accord- of June, depending on field elevation. Along the transect, ing to quantitative variables of management and the envir- 20 biomass samples were collected within 10 cm × 10 cm onment. By design, there is a parabolic relation between quadrats equidistantly distributed. Each biomass sample Div and the percentage of Fast : maximum func- was exhaustively sorted and a score from 0 to 6 was GFT GFT tional diversity is expected for moderate percentages of assigned to each species present: (0) species present but Fast , with minima at extreme percentages (Duru et al. not dominant, (1) species contributes at least one-sixth of GFT 2011b). Thus, response of Div to management and the biomass sampled (17%), and so on, up to (6) for a GFT environmental variables was calculated separately for species representing all the biomass sampled (100%) Fast ≥ 50% and Fast < 50% (Duru et al. 2013). (Theau et al. 2010). Although about half of the species GFT GFT Based on this relation, within–between-vegetation diver- present were recorded in all grassland fields, no significant sity for a set of fields was characterised by only two differences in plant-strategy distribution existed between variables: Fast and Sum . this simplified method and a point quadrat method (Daget GFT GFT & Poissonet 1971) done on a sub-sample of 60 grassland ANOVA was performed too to examine the effect of fields (Fallour et al. 2008). farm management strategy (i.e. stocking density) and field On the same subset of grassland fields (n =60), characteristics (i.e. distance of field from the cowshed, detailed data were recorded for herbage mass and N con- field elevation and area, soil pH and water availability) centration (to calculate N uptake), soil C content and on field management. Land use was categorised into three species richness. Biomass measurements were made just classes according to level of care requirements of animals: before the beginning of stem elongation of grass species cutting < heifer or beef cow grazing < dairy cow grazing. on four 0.25 m × 0.75 m quadrats in randomly distributed Some services (forage production and inputs) were each grassland community. Total N concentrations of dried directly expressed and predicted with regression analysis according to grassland composition. For vegetation hetero- and ground forage samples were determined by a CHN geneity, only Fast was mapped. To assess the degree to 2000 Analyser (LECO, St. Joseph, MI, USA). Three soil GFT which forage characteristics match animal feed require- samples per grassland were taken from the 0 to 5 cm layer ments, ANOVA was used to discern whether differences for measuring organic soil carbon content. This was 134 M. Duru et al. existed among LMUs within farms and whether there was always significantly higher for dairy farms than for beef an effect of farm strategy or stocking density on Fast . farms for both cut and grazed grasslands (Table 5). GFT Response of plant functional composition to management Results and environmental factors at the field level Drivers of grassland functions There were significant effects of fertilisation and main use Relations between grassland management and field (cutting vs grazing) on the percentages of Fast and GFT characteristics Late , and of field elevation on percentages of Sum GFT GFT Grassland management (fertiliser applied and herbage and Late (Table 6). Large differences were observed in GFT used) depended significantly on topology (distance from the response of Fast to N fertiliser rates for both grazed GFT the field to the cowshed, field area), topography (eleva- and cut grasslands (not shown), which indicates that envir- tion) and soil characteristics (pH and depth). However, onmental factors or biodiversity control N-use efficiency. these effects varied greatly according the component of A consistent, significant response of Fast to elevation GFT management considered (Table 4). In general, mown fields and R – and M-EIV was found for both land-use types are located at a lower elevation and on deeper soils. (Table 7). Based on regressions (Table 4) and the range of Grazed fields for dairy cows lie closest to the cowshed, variation observed for the three variables (N-EIV, plant while those for heifers or beef cows are at higher eleva- height and field elevation), nutrient availability had the tions and furthest from the cowshed. Grasslands used for greatest effect on the percentage of acquisitive species. animals requiring more care (grazing by dairy cows) were When N-EIV changed from 3.5 to 6.5 (the minimum and located at lower elevations and had deeper soils. maximum observed values being 3.1 and 6.9), the percen- Fertilisation rates decreased significantly with increasing tage of acquisitive grass species increased by 50%. In field elevation for three of the four types of grassland use contrast, when canopy height increased from 20 to (Table 4). For cut grasslands, fertilisation rates increased 80 cm or field elevation decreased by 400 m, the percen- with increasing field area. Otherwise, they showed no tage of acquisitive grass species increased by only around clear relation with the other factors studied. Indicators of 15%. Considering N-EIVs instead of N fertiliser and farming intensity (N fertilisation rate or N-EIV) were canopy height at harvest time to encompass both Table 4. Anova of five field characteristics for land use and fertiliser supply (independent variables). Variable df distance from the cowshed Altitude area R-EIV M-EIV F-value F-value F-value F-value F-value (P-value) (P-value) (P-value) (P-value) (P-value) Land use / 171 <0.0001 12,4 (0.001) 1.7 (0.17) 1.2 (0.26) 12 (0.02) Fertilisation supply cutting 80 0.62 (0.53) 11.7 < 0.0001 3.7 (0.029) 0.13 (0.78) 0.3 (0.66) (3 classes: 0; grazing by beef 17 0.12 (0.83) 3.5 (0.06) 1.2 (0.37) 3.4 (0.09) 1.1 (0.31) 0 < × < 75; cows †† >75 kg/ha/year) grazing by heifers 39 0.60 (0.58) 0.12 (0.85) 0.99 (0.21) 5.9 (0.03) 0.95 (0.40) grazing by dairy 35 2.1 (0.25) 35.1 (<0.0001) 1.27 (0.29) 2.1 (0.14) 0.1 (0.95) cows † †† Notes: P-values <0.05 are in bold; see column 2; anova for fertiliser supply was done separately for each land use; R- and M–EIV: Ellenberg indicator values for soil reactivity and moisture, respectively. Table 5. Mean values and ANOVA of N fertiliser and N-Ellenberg Indicator Value (N-EIV) for farm enterprise types (beef vs dairy) (independent variables) considering separately each of the three land-use types. Land use df Farm enterprise type Fertilisation rate (kg/ha) N-EIV cutting areas beef farm 71 5.7 81 dairy farm 100 5.9 F-value (P-value) 8.4 (0.0048) 4.2 (0.04) cow grazing areas 53 beef farm 15 5.1 dairy farm 42 5.7 F-value (P-value) 24.8 (<0.0001) 8.2 (0.006) heifer grazing areas 39 beef farm 17 5.4 dairy farm 44 5.1 F-value (P-value) 10.7 (0.002) 1.9 (0.18) Notes: P-values <0.05 are in bold; see column 2. International Journal of Biodiversity Science, Ecosystem Services & Management 135 Table 6. Mean values and ANOVA of three indicators of grassland composition (Fast Late Sum ) for fertilisation rates GFT GFT GFT (independent variables) considering separately main land-use types. Fertilisation (kg/ha) Land use (number of fields) Fast Late Sum GFT GFT GFT Cut grasslands 149 (26) 71 26 81 87 (19) 60 37 75 42 (30) 62 32 75 Grazed grasslands 63 (31) 60 32 75 33 (25) 45 26 70 9 (38) 28 28 60 ANOVA for land use and fertilisation Fertilisation 6.9 (0.0015) 3.37 (0.037) 3.6 (0.1) Land use 4.1 (0.007) 0.51 (0.76) 8.9 (0.0002) †† F-value (P-value) Field altitude: co-variable 46.1 (<0.001) 14.7 (0.0002) 1.1 (0.31) Notes: Fast and Late are the percentage of GFTs having a fast and a late growth strategy, respectively; Sum is the percentage of grass species in GFT GFT GFT †† biomass in a grassland; p values <0.05 are in bold; df = 171. Table 7. Regression analysis between Fast (percentage of GFTs having a fast growth strategy) and some management and GFT environmental variables. management variables environmental variables Management N fertiliser N-EIV Plant height altitude R-EIV M-EIV r se Grazed + (***) / / – (***) + (***) – (+) 0.55 (***) 13.0 Cut + (*) / / – (***) + (***) – (*) 0.44 (***) 12.4 Cut + grazed / + (***) + (***) – (***) + (**) – (*) 0.63 (***) 12.2 Notes: N, R- and M–EIV: Ellenberg indicator values for nutrient, soil reactivity and moisture, respectively; number of individuals: 169. +or – indicated the direction of effect. *P < 0.05; **P < 0.01; ***P < 0.0001. management types in the same model provided similar effect was not observed for cut grasslands, which had the consistent results. The percentage of Late was posi- highest fertiliser application rates (83.2 ± 47 kg N/ha). GFT tively correlated with N-EIV (<0.001) and Tsum Both regulating services studied had a significantly (P <0.05), and negatively with field elevation (P <0.05), negative correlation with Fast Species richness GFT. but the correlation was weaker (R = 0.22, P < 0.001). decreased significantly as Fast increased (Figure 2b), GFT To test the hypothesis that functional diversity and the correlation increased when considering Sum GFT depended on the levels of stress and disturbance, we (r = –0.75; P < 0.001), which had a significantly negative established two linear regressions according to whether correlation with species richness. Soil C content decreased the percentage of Fast was lesser or greater than 50%. significantly as the percentage of Fast increased GFT GFT We found significant effects (P < 0.001) of N fertilisation (Figure 2c). (positive) and field elevation (negative) on Fast < 50%, GFT and the opposite for both variables for Fast > 50%. The GFT same patterns were observed regardless of the percentage Services assessed at LMU and landscape levels of grass species. For cut grasslands, Div decreased as GFT For resource provision, a significantly positive correlation the percentage of Fast increased, and the reverse was GFT existed between stocking rate and the percentage of observed for grazed grasslands. Fast (P < 0.001; Figure 3a). For grazed grasslands GFT alone, mean stocking rate and Fast were highest for GFT dairy cows and lowest for heifers. However, there were differences in stocking rate for Fast > 50% for the same GFT Provision of ES by grasslands land use and farm strategy. Conversely, similar grassland Services assessed at the field level functional composition was found for all three types of Herbage mass just before the beginning of stem elongation animal groups (beef cows, dairy cows and heifers). A significantly was correlated with the percentage of Fast GFT minimum stocking rate of 0.5 animal units/ha was (Figure 2a). For the less fertilised grazed grasslands (N observed in the absence of Fast . For cut areas alone, GFT fertiliser =24.4 ± 26 kg N/ha), N uptake was significantly no relation was observed between stocking rate and the and positively correlated with Div (r = 0.40, n = 38, GFT percentage of Fast . The stocking rate depended mainly GFT P < 0.01). Including soil pH and moisture with EIVs on the proportion of the cut area which was topped in early increased the correlation (r = + 0.66, P < 0.001). This spring (r = +0.90; P = 0.077). However, stocking rate 136 M. Duru et al. (a) 7 2 = 0.53*** 0 20 40 60 80 100 (b) Fast (%) GFT 40 2 R = 0.36*** 020 406080 100 Fast (%) (c) GFT R = 0.44*** Figure 3. Relations between stocking rate and the percentage of GFT having an acquisitive growth strategy (Fast ), GFT 0 20406080 100 (a) (Y = 0.032X + 0.6); end of hay harvest date and the percentage Fast (%) GFT of GFTs having a late growth strategy (Late ), GFT (b) (Y = 21.6X + 388); data aggregated at the LMU level. Figure 2. Relations between the percentage of GFT having an acquisitive growth strategy (Fast ) and (a) standing herbage GFT mass (Y = −0.13X + 35.7), (b) species richness Within–between-field diversity (Y = 0.024X + 2.26) and (c) soil C content in the 05 cm layer (Y = −1.15X + 1.84). Vegetation heterogeneity can be assessed for sets of fields among farms or landscapes. Obviously, the more the within-field or between-field diversity, the more heteroge- positively correlated with N fertiliser rate and timing of the neous were the height and phenology (see Materials and first cut (r = +0.82; P < 0.05). Methods) of grasslands at both levels of organisation. For The flexibility component of management services example, the GFT with an acquisitive growth strategy was was represented as the mean percentage of Late at the GFT mapped for two farms (Figure A2). We observed that LMU level. It significantly correlated with the end date of almost 50% of the grassland area corresponds to fields the hay harvest (Figure 3b), because fields with high GFT with low GFT diversity (FastGFT: <30% or >70%), and diversity had longer durations of harvest operations (not that most of these grasslands are close together within the shown). Significant differences were observed in the per- landscape. centage of Fast among the three LMU types (cut, GFT At the farm level, cutting and cow grazing areas grazed by cows and by heifers) for six of the eight farms tended to have higher percentages of Fast in dairy GFT (Table 8). Except for farm D3, cut LMU had the highest farmsthaninbeef farms (Figure 4a). But this was not percentage of Fast (see Figure A1 for detailed descrip- GFT the case for heifers grazing areas. Within the range of tion of plant composition at the farm level). This indicates 4060% Fast ,the threeLMU typeswereobservedfor GFT that forage production and forage quality at the leafy stage both types of farm. In dairy farms, the heifer LMU (Fast ) were usually the highest for cut areas (except D3 GFT increased between-LMU differences due to its low per- and B1), followed by cow grazing areas (except for D1, centages of Fast , while for beef farms the cut LMU GFT B1, and B3). These data show consistent rankings of increased between-LMU differences due to it high per- animal feed requirements and type of vegetation allocated, centages of Fast .For theentiredataset,weconsid- GFT except for D1. There was a consistent effect of farm ered three components of plant diversity: Div ,Fast GFT GFT strategy on the percentage of Fast (P < 0.01) for all GFT and Sum (Figure 4b). Similar patterns between GFT LMUs (Table 8). Fast percentages were significantly GFT Div and Fast were observed regardless of the GFT GFT lower for beef farms. Among dairy farms and grazing percentage of grass species (Sum ), except for very GFT areas, there was a significant difference between Fast , GFT low Fast values. Comparison of analysis at field and GFT as was the case between farms for heifers. LMU levels (Figure 4a and b) shows that the LMU level –1 Soil C content (g/1000g) Species richness Standing herbage mass (t ha ) International Journal of Biodiversity Science, Ecosystem Services & Management 137 Table 8. Mean values and ANOVA of two indicators of grass- land composition (Fast and Late ) for the three land-use GFT GFT types (independent variables) considering each farm separately. Farm Df Land use Fast Late GFT GFT D1 17 cut area 71 25 cow grazing areas 38 23 heifer grazing areas 52 25 F-value (P-value) 2.3 (0.15) 0.9 (0.44) D2 21 cut area 67 37 cow grazing areas 50 37 heifer grazing areas 34 53 F-value (P-value) 7.18 (0.006) 3.2 (0.04) D3 19 cut area 84 10 cow grazing areas 87 13 heifer grazing areas 12 31 F-value (P-value) 28.6 (0.0007) 1.3 (0.24) D4 22 cut area 58 40 cow grazing areas 39 38 heifer grazing areas 10 43 F-value (P-value) 4.9 (0.02) 0.6 (0.73) B1 27 cut area 68 29 cow grazing areas 25 30 heifer grazing areas 53 36 F-value (P-value) 5.55 (0.01) 0.5 (0.78) B2 25 cut area 55 35 cow grazing areas 53 28 heifer grazing areas 39 37 F-value (P-value) 2.45 (0.10) 1.2 (0.45) B3 18 cut area 63 35 cow grazing areas 11 36 heifer grazing areas 19 44 Figure 4. Relations between within-field functional diversity F-value (P-value) 11.8 (0.003) 1.1 (0.51) (Div ) and the percentage of GFTs having an acquisitive B4 20 cut area 38 44 GFT growth strategy (Fast ); a: for data aggregated at the LMU cow grazing areas 1 34 GFT level (squares for cut grassland, circles and triangles for grassland heifer grazing areas 18 40 grazed by cows and heifers, respectively, open symbols for beef F-value (P-value) 8 (0.0002) 1 (0.55) farms, closed symbols for dairy farms) Notes: percentage in biomass; significant differences among land use within (Y = 0.0001X + 0.009X + 0.52); b: for all grassland fields, the a farm are indicated in bold. percentage of grass species in herbage biomass: >80% (closed Fast and Late are the percentage of GFTs having a fast and a late GFT GFT circles), 60–80% (crosses) or <60% (open circles). growth strategy, respectively; D: dairy farms; B: beef farms. in Duru, Cruz, Jouany, et al. (2010), or stocking rate, greatly structures grassland diversity. In other words, R = 0.78 in this paper), whereas management and envir- grassland diversity is greater between LMUs than within onmental drivers can explain only 44% or 59% of the them. Thus, differences in farm orientation and stocking variance, respectively, depending on whether the explana- density are required to maintain a grassland mosaic at the tory variables are observed or measured. Correlations landscape level (Figure A2 gives an example for two found elsewhere using plant traits instead of plant groups farms). are no higher (e.g. Lavorel et al. 2011; Duru et al. 2012). This justifies using plant functional types as a key tool for managing grassland functional composition and predicting Discussion ES. However, the method used to characterise grassland A plant functional-type-based MF for linking plant functional composition is highly simplified in com- management and ES parison to the measurement of plant traits. As is known, plant traits, plant functional groups or their proxies must Our results show that a MF based on a simplified char- be aggregated to assess certain ES, such as those provided acterisation of plant functional types allows a large set of by the landscape mosaic (Lavorel et al. 2011). However, ES at different organisational levels to be assessed, which aggregation does not have to reach the landscape level to is better than considering drivers alone. Regardless of the be able to understand effects of management and policies level considered, strong and weak correlations were found on within–between-grassland diversity. Thus, we found between plant functional composition and services (espe- that the LMU level performed better than the farm level cially forage services) and drivers, respectively. for understanding in depth the degree of heterogeneity in Regression analysis has shown that Fast alone can GFT management practices that impact ES. predict services well (e.g. herbage production, R = 0.69 138 M. Duru et al. The main advantage of the MF proposed is its ease of Based on the farm sample studied, we found that even a use by stakeholders. Its strength lies in not being limited to single farm can contain a wide range of within–between-field production services, as traditionally done in agronomy. functional plant diversity (Rudmann-Maure et al. 2008)and The concept of plant functional types is key because it is that contrasting land use within a farm can create a diversity appropriate for evaluating or predicting a wide range of ES of plant species as wide at observed at the landscape level. As and makes sense for farmers and other stakeholders. observed in a different context (Beyene et al. 2006), plant Farmers give positive feedback when their land is depicted functional type assemblages are the result of deliberate man- through plant functional types in a bar graph (Duru et al. agement choices resulting from farm enterprise type (Brodt 2011a) or a map (Figures A1 and A2, respectively). et al. 2006) and from assets and constraints such as available Moreover, several agricultural consultants have adopted facilities and field topography (Andrieu et al. 2007; Valbuena it, at least partly, to design grassland typologies at national et al. 2008;Martinetal. 2009). Differences in plant func- (Launay et al. 2011) or regional (Carrère et al. 2012) tional diversity at the LMU level are the result of land use and levels. They have a great interest in the MF because of farm enterprise type. Usually, cut grasslands have the highest its ability to reduce a large list of species into a small percentage of acquisitive types, first because they receive number of plant functional types in an effective commu- more fertiliser and have consistently higher N-EIV, and sec- nication tool. It addresses four key components of forage ond due to the direct effect of management practices on the services that fit well with farmers’ expectations (Duru, percentage of acquisitive types (Table 7). Dairy farms have a Cruz, Jouany, et al. 2010). higher percentage of acquisitive plant types for both cut and Below, we summarise and discuss our main findings grazed areas, which is consistent with the highest digestibil- about relations between grassland functional composition, ity of these plant types (Duru, Cruz & Theau 2010). management (next section) and ES, and examine trade-offs Between-farm comparisons can show whether the potential and synergies between ES (last section). exists to reduce the cost of feedstuffs. For example, the dairy farm D4 has similar milk production per cow (around 5000 kg per year) even though the percentage of Fast GFT Drivers of grassland functional composition and differed greatly: it was highest for D3 and lowest for D4 management (Table 8). Since obtaining a high percentage of Fast GFT At field level, the grasslands with the highest percentage requires high fertiliser input, this indicates that production of acquisitive species (Fast ), as well as the greatest costs could be reduced if enough land were available, espe- GFT abundance of grass species, responded significantly to cially if it is grazed, so as not to increase the workload. The management and certain environmental drivers (e.g. MF also detects discrepancies, for example, for dairy heifers field elevation). Fast increasedwithincreasingnutri- on farm D3 that used high-quality herbage. GFT ent availability (Wilson et al. 1999) and decreased with Field characteristics may explain between-farm differ- increasing temperature (Roche et al. 2004), which was ences in grassland functional diversity. In less-favoured negatively correlated with field elevation (Figure 5,top). areas, many farm-dependent constraints may occur These results are consistent with studies demonstrating (Andrieu et al. 2007; Martin et al. 2009). These include: that the features used to distinguish GFTs (specific leaf area, LDMC) are appropriate indicators of stress, in gen- (1) The proportion of grassland fields located near the eral (Harrison et al. 2010). Stress factors for nutrient cowshed, which affects the stocking rate, at least availability and mean temperature act in the same man- for dairy systems. In this way, D3, which has ner, favouring acquisitive species when stress is low. Our fewer pastures near the cowshed, has a higher results confirm those observed for a small (Duru, stocking rate than D4 (3.3 vs. 2.0 animal units Ansquer, et al. 2010) and a large (Martin et al. 2009) per ha). number of sites. Disturbances modify the effects of (2) The availability of summer pastures for animals stress, either reducing or amplifying them. For given with low feed requirements (dry cows, replace- climatic and soil conditions, mowing promotes acquisi- ment heifers), because these usually unfertilised tive species, while grazing promotes conservative spe- areas widen the range of GFTs encountered at cies. In other words, mowing reinforces the positive the farm level. For example, dairy farms D3 and effect of temperature and N on the abundance of acqui- D4 can use summer pastures that usually never sitive species (Figure 5, top). For functional diversity receive fertiliser, while this is not the case for D1 (here Div ), previous research on the intermediate and D2. GFT stress hypothesis (Vonlanthen et al. 2006) supports the (3) The use of modern harvesting equipment (round- idea that maximum diversity was observed only when baller), which reduces dependence on the weather, simultaneously considering stress and disturbance fac- shortens the harvest duration, and thus the range tors, as we also found. Additionally, we show that the of vegetation types observed (Benton et al. 2003). direction of effect for stress factors depends on the cur- There is little variability in functional composition rent dominant plant strategy; for example, N fertiliser between mown grasslands, except for farms B2 could decrease or increase Div and B4, which mowed summer pastures. GFT. International Journal of Biodiversity Science, Ecosystem Services & Management 139 Figure 5. A framework for linking environmental and management factors to grassland functional composition and a set of ES; the shape of each triangle indicates the direction of expected effect; rectangles indicate that effects depend on stakeholder viewpoints; FD: functional diversity. The results of this study clearly show that, as White et al. different strategies for resource acquisition leads to higher (2004) suggested, there is greater uniformity in plant types input efficiency for herbage production, as suggested by within LMUs than between them. Balancing constraints Fornara and Tilman (2009) and Dybzinski et al. (2008), on and goals, there is room for different degrees of farm-level the basis of species richness. For management services, we diversity of plant types. This means that some farmers found that between-farm differences are related to the choose to favour a certain range of plant types for eco- allocation of vegetation types or forage resources to dif- nomic (e.g. to reduce inputs) or labour reasons. ferent animal groups to save nutrients or feedstuff costs; Nevertheless, we did not find an effect of within-field for example, the capacity exists to reduce N fertiliser functional diversity on stocking rate as shown at the field without affecting animal production for some land-use level (Weigelt et al. 2009). As claimed by Sanderson et al. types, as seen above. Furthermore, our MF provides a (2004), the evidence for diversity effects is equivocal for simple method for comparing within–between-field plant pasturelands. However, stocking rate is probably a too functional diversity. It could help stakeholders determine coarse variable to reveal the effect of within-field func- the degree of grassland heterogeneity to promote tional diversity on potential complementarities between (Lamarque et al. 2011). plant types. The synthetic representation of relations between grassland functional composition and ES allows stake- holders to examine the main trade-offs they must consider Evaluation of trade-offs and synergies between ES (Figure 5, bottom). For farmers, among forage and man- The MF based on GFTs allows the evaluation of a large agement services, an opposite trend occurred between number of ES at different organisation levels and the measured herbage production and recorded production and analysis of main trade-offs between services. We found timing (r = −0.51, n = 24, P < 0.01), with Fast GFT Late as proxies, respectively (from data used for that a set of ES can be evaluated at the field level and GFT Figure 3a and b). However, at the field level, a nonlinear above (Figure 5, bottom for Fast ). Fast was a good GFT GFT relation between herbage production and yield flexibility proxy for forage production measured at the field level, was observed. At the LMU level, such trade-offs were not thus confirming results obtained in other regions (Duru, usually a problem for farmers because the priority services Cruz, Jouany, et al. 2010) or assessed through the stocking depended mainly on the animal group (cow vs. heifer) or rate at the LMU level (this paper). For low N rates, we the land management (grazing vs. cutting) considered. verified that the coexistence of plant functional types with 140 M. Duru et al. effects of land management on ecosystem services. Ecol Thus, the diversity of animal groups on a farm and of farm Indic. 21:110–122. enterprises in a region lead to a diversity of vegetation Benton TG, Vickery JA, Wilson JD. 2003. Farmland biodiversity, is types in a landscape which is enhanced by environmental habitat heterogeneity the key? Trends Ecol Evol. 18:182–188. factors such as field aspect and elevation. This explains Beyene A, Gibbon D, Haile M. 2006. Heterogeneity in land why this kind of diversity in agriculture creates a mosaic resources and diversity in farming practices in Tigray, Ethiopia. Agric Syst. 88:61–74. of vegetation types within and between farms that directly Bossuet L, Duru M. 1992. Estimating herbage mass by the contributes to landscape attractiveness, which is important sward-stick method. Fourrages. 131:283–300. for tourism (Junge et al. 2011) and indirectly and more Brodt S, Klonsky K, Tourte L. 2006. Farmer goals and manage- broadly to multiple ES (Smukler et al. 2010)(Figure 5). ment styles, implications for advancing biologically based agriculture. Agric Syst. 89:90–105. Carrère P, Orth D, Chabalier C, Seytre L, Piquet M, Landrieaux J, Rivière J. 2012. Une typologie multifonctionnelle des Conclusion prairies des systèmes laitiers AOP du Massif Central combi- In less-favoured areas such as mountains, farms have nant des approches agronomiques et écologiques. Fourrages. 209:9–22. high grassland diversity due to diversity in abiotic factors Cash DW, Clark W, Alcock F, Dickson NM, Eckley N, Guston (elevation), farm orientation (beef, dairy), enterprise DH, Mitchell RB. 2003. Knowledge systems for sustain- (cows, heifers) and management (grazing, cutting). To able development. Proc Nat Acad Sci USA. describe ES provided by grassland diversity and their 100:8086–8091. underlying drivers, we have developed a MF based on Ceulemans T, Merckx R, Hens M, Honnay O. 2013. Plant species loss from European semi-natural grasslands follow- GFTs. Due to its ease of use and credibility, this MF ing nutrient enrichment – is it nitrogen or is it phosphorus? should help agricultural experts and farm advisors under- Glob Ecol and Biogeo. 22:73–82. stand implications of different management choices on Colasanti RL, Hunt R, Askew AP. 2001. A self-assembling grassland diversity and on a large set of ES noticeable by model of resource dynamics and plant growth incorporating farmers, tourists and other members of society. The field plant functional types. Funct Ecol. 15:676–687. Cruz P, Theau JP, Lecloux E, Jouany C, Duru M. 2010. level may be sufficient for assessing their impact on plant Typologie fonctionnelle de graminées fourragères pérennes, diversity, while the land-use type and farm levels are still une classification multitraits. Functional typology of peren- needed to understand the drivers of management prac- nial forage grasses: a classification based on several charac- tices. Our MF can also help local policy-makers who teristics. Fourrages. 401:11–17. intend to support biodiversity with subsidies based on Daget P, Poissonet J. 1971. Une méthode d’analyse phytosocio- logique des prairies [A method for analyzing grassland phy- stocking-rate thresholds calculated for a set of fields or tosociology]. Ann Agronom. 22:5–41. theentirefarm. Ourresults clearly show that field and De Bello F, Thuiller W, Leps J, Choler P, Clément B, Macek P, farm levels are too small and too large, respectively. The Sebastia MT, Lavorel S. 2009. Partitioning of functional LMU level seems to be the right level for gathering data diversity reveals the scale and extent of trait convergence for management and/or vegetation, then engaging discus- and divergence. J Veg Sci. 20:475–486. De DM V, De Boer T. 1959. Methods used in botanical grassland sion between beneficiaries of ES to identify trade-offs research in the Netherlands and their application. Herbage and synergies. Abstr. 29:1. Diaz S, Lavorel S, McIntyre S, Falczuk V, Casanoves F, Milchunas DG, Skarpe C, Rush G, Sternberg M, Noy-Meir Disclosure statement I, et al. 2007. Plant trait responses to grazing – a global synthesis. Glob Change Biol. 13:313–341. No potential conflict of interest was reported by the authors. Duru M, Ansquer P, Jouany C, Theau JP, Cruz P. 2010. Suitability of grass leaf dry matter content for assessing the response of grasslands to land use and fertility. Ann Bot. Funding 106:823–831. This work was supported by the French National Research Duru M, Cruz P, Jouany C, Theau JP. 2010. Herb’type ,un Agency O2LA (Organismes et Organisations Localement nouvel outil pour évaluer les services de production fournis Adaptés) ANR-09-STRA-08. par les prairies permanentes. Herb’type : a new support for assessing agricultural services provided by permanent grass- lands. Prod Animales. 23:319–332. Duru M, Cruz P, Jouany C, Theau JP. 2011a. Combiner des References recherches en agroécologie et des dispositifs participatifs AFNOR. 1994. LA qualité des sols, vol.1. Recueil des normes. pour construire des outils d’évaluation des prairies perma- La Défense, Paris: AFNOR. nentes. Combining agroecological research and participatory Andrieu N, Josien E, Duru M. 2007. Relationships between approach to build tools for evaluating permanent grasslands. diversity of grassland vegetation., field characteristics and Cah Agricult. 20:223–234. land use management practices assessed at the farm level. Duru M, Cruz P, Jouany C, Theau JP. 2011b. Optimiser les Agric Ecosys Env. 120:359–369. services fourragers des prairies permanentes: des outils Ansquer P, Duru M, Theau JP, Cruz P. 2009. Convergence in pour caractériser et gérer la diversité floristique [Internet]. plant traits between species within grassland communities Fascade. 4p. Available from: http://www.sad.inra.fr/ simplifies their monitoring. Ecol Ind. 9:1020–1029. Duru M, Cruz P, Theau JP. 2010. A simplified method for Ape VO, Petz K, Alkemade R, Hein L, de Groot RS. 2012. characterising agronomic services provided by species-rich Framework for systematic indicator selection to assess grasslands. Crop Pasture Sci. 61:420–433. International Journal of Biodiversity Science, Ecosystem Services & Management 141 Duru M, Jouany C, Le Roux X, Navas ML, Cruz P. 2013. A Lavorel S, Grigulis K. 2012. How fundamental plant functional plant functional type approach tailored for stakeholders trait relationships scale-up to trade-offs and synergies in involved in field studies to predict ecosystem services pro- ecosystem services. J Ecol. 100:128–140. vided by grasslands. Renew Agricult and Food Syst. 1–16. Lavorel S, Grigulis K, Lamarque P, Colace MP, Garden D, Girel doi:10.1017/S1742170513000306. J, Pellet G, Douzet R. 2011. Using plant functional traits to Duru M, Theau JP, Cruz P. 2012. Functional diversity of species- understand the landscape distribution of multiple ecosystem rich managed grasslands in response to fertility : defoliation services. J Ecol. 99:135–147. and temperature. Basic Appl Ecol. 13:823–831. Le Roux X, Barbault R, Baudry J, Burel F, Doussan I, Garnier E, Duru M, Theau JP, Hossard L, Martin G, Cruz P. 2011. Diversité Herzog F, Lavorel S, Lifran R, Roger-Estrade J, et al. 2008. de la composition fonctionnelle de la végétation au sein Agriculture et biodiversité, valoriser les synergies. Paris: Inra d’une prairie et entre prairies. Characterisation of within editions. and between grassland functional diversity. Fourrages. Lindemann-Matthies P, Junge X, Matthies D. 2010. The influ- 205:61–73. ence of plant diversity on people’s perception and aesthetic Dybzinski R, Fargione JE, Zak DR, Fornara D, Tilman D. appreciation of grassland vegetation. Biol Conserv. 143:195– 2008. Soil fertility increases with plant species diversity 202. doi:10.1016/j.biocon.2009.10.003 in a long-term biodiversity experiment. Oecologia. Lugnot M, Martin G. 2013. Biodiversity provides ecosystem 158:85–93. services: scientific results versus stakeholders’ knowledge. Ellenberg H, Weber HE, Düll R, Wirth V, Werner W, Paulissen Reg Env Change. 13:1145–1155. D. 1992. Zeigerwerte von Pflanzen in Mitteleuropa. Scr Martin G, Cruz P, Theau JP, Jouany C, Fleury P, Granger S, Geobotanica. 18:1–258. Faivre R, Balent G, Lavorel S, Duru M. 2009. A multi-site Ersten ACD, Alkemade JRM, Wassen MJ. 1998. Calibrating study to classify semi-natural grassland types. Agricult Ellenberg indicator values for moisture, acidity, nutrient Ecosyst Env. 129:508–515. availability and salinity in the Netherlands. Plant Ecol. Matthews KB, Rivington M, Blackstock K, McCrum G, Buchan 135:113–124. K, Miller DG. 2010. Raising the bar? – The challenges of Fallour D, Theau JP, Corler K, Hossard L, Martin G, Jouany C, evaluating the outcomes of environmental modelling and Duru M, Cruz P. 2008. A simplified method to determine the software. Env Model & Softw. 26:247–257. abundance of grass functional groups in natural grasslands. Nozières MO, Moulin CH, Dedieu B. 2011. The herd, a source of Biodiversity and animal feed – Future challenges of flexibility for livestock farming systems faced with uncer- Grassland production – 22nd General Meeting of the tainties? Animal. 5:1142–1157. European Grassland Federation; Uppsal, Sweden. 93–95. Pakeman R, Marriott C. 2010. A functional assessment of the Fisher BR, Turner K, Morling P. 2009. Defining and classifying response of grassland vegetation to reduced grazing and ecosystem services for decision making. Ecol Econ. 68:643–653. abandonment. J Veg Sci. 21:683–694. Fornara DA, Tilman D. 2009. Ecological mechanisms associated Petchey OL, Gaston KJ. 2006. Functional diversity, back to with the positive diversity-productivity relationship in a N- basics and looking forward. Ecol Lett. 9:741–758. limited grassland. Ecology. 90:408–418. Power AG. 2010. Ecosystem services and agriculture, tradeoffs Frank S, Fürst C, Koschke L, Witta A, Makeschin F. 2013. and synergies. Phil Trans Royal Soc B, Biol Sci. Assessment of landscape aesthetics—validation of a land- 365:2959–2971. scape metrics-based assessment by visual estimation of the Quetier F, Lavorel S, Thuiller W, Davies AI. 2007. Plant-trait- scenic beauty. Ecol Indic. 32:222–231. doi:10.1016/j. based modeling assessment of ecosystem service sensitivity ecolind.2013.03.026 to land use change. Ecol Appl. 17:2377–2386. Gross N, Suding KN, Lavorel S, Roumet C. 2007. Raudsepp-Hearne C, Peterson GD, Bennett EM. 2010. Complementarity as a mechanism of coexistence between Ecosystem service bundles for analyzing tradeoffs in diverse functional groups of grasses. J Ecol. 95:1296–1305. landscapes. PNAS. 107:5242–5247. doi:10.1073/ Harrison S, Prentice I, Barboni D, Kohfeld K, Ni J, Sutra JP. pnas.0907284107. 2010. Ecophysiologicoal and bioclimatic foundations for a Roche P, Diaz-Burlinson N, Gachet S. 2004. Congruency analy- global plant functional classification. J Veg Sci. sis of species ranking based on leaf traits, which traits are the 21:300–317. more reliable? Plant Ecol. 174:37–48. Hooper DU, Ewel JJ, Hector A, Inchausti P, Lavorel S, Lawton Rodríguez-Ortega T, Oteros-Rozas E, Ripoll-Bosch R, Tichit M, JH, Lodge D, Loreau M, Naeem S, Schmid B, et al. 2005. Martín-López B, Bernués A. 2014. Applying the ecosystem Effects of biodiversity on ecosystem functioning, a consen- services framework to pasture-based livestock farming sys- sus of current knowledge and needs for future research. Ecol tems in Europe. Animal. 1–12. doi:10.1017/ Monogr. 75:3–36. S1751731114000421 Junge X, Lindemann-Matthies P, Hunziker M, Schüpbach B. Rudmann-Maure K, Weyand A, Fischer M, Stöcklin J. 2008. The 2011. Aesthetic preferences of non-farmers and farmers for role of land use and natural determinants for grassland vege- different land-use types and proportions of ecological com- tation composition in the Swiss Alps. Basic Appl Ecol. pensation areas in the Swiss lowlands. Biol Conserv. 9:494–503. 144:1430–1440. Sanderson MA, Skinner RH, Barker DJ, Edwards GR, Tracy BF, Lamarque P, Tappeiner U, Turner C, Steinbacher M, Bardgett Wedin DA. 2004. Plant species diversity and management of RD, Szukics U, Schermer M, Lavorel S. 2011. Stakeholder temperate forage and grazing land ecosystems. Crop Sci. perceptions of grassland ecosystem services in relation to 44:1132–1144. knowledge on soil fertility and biodiversity. Reg Env Sanderson MA, Archer D, Hendrickson J, Kronberg S, Liebig M, Change. 11:791–804. Nichols K, Schmer M, Tanaka D, Aguilar J. 2013. Launay F, Baumont R, Plantureux S, Farrie JP, Michaud A, Pottier Diversification and ecosystem services for conservation E. 2011. Prairies permanentes, des références pour valoriser agriculture: outcomes from pastures and integrated crop– leur diversité. éd. Paris: Institut de l’Elevage; 128 p. livestock systems. Renewable Agricult and Food Syst. Lavorel S, Garnier E. 2002. Predicting changes in community 28:129–144. composition and ecosystem functioning from plant traits, Schaffers AP, Sykora KV. 2000. Reliability of Ellenberg indi- revisiting the Holy Grail. Funct Ecol. 16:545–556. cator values for moisture, nitrogen and soil reaction, a 142 M. Duru et al. comparison with field measurements. J Veg Sci. Van Der Biest K, D’Hondt R, Jacobs S, Landuyt D, Staes J, 11:225–244. Goethals P, Meire P. 2014. EBI: an index for delivery of Smukler SM, Sánchez-Moreno S, Fonte SJ, Ferris H, ecosystem service bundles. Eco Indicat. 37:252–265. Klonsky K, O’Geen AT, Scow KM, Steenwerth KL, Vojtech E, Loreau M, Yachi S, Spehn EM, Hector A. 2008. Light Jackson LE. 2010. Biodiversity and multiple ecosystem partitioning in experimental grass communities. Oikos. functions in an organic farmscape. Agric Ecosys Env. 117:1351–1361. 139:80–97. Vonlanthen CM, Kammer PM, Eugster W, Bühler A, Veit H. Sun S, Frelich LE. 2011. Flowering phenology and height 2006. Alpine vascular plant species richness, the importance growth pattern are associated with maximum plant of daily maximum temperature and pH. Plant Ecol. height, relative growth rate and stem tissue mass 184:13–25. density in herbaceous grassland species. J Ecol. 99: Weigelt A, Weisser WW, Buchmann N. 2009. Biodiversity for 991–1000. multifunctional grasslands , equal productivity in high-diver- Swinton SM, Lupi F, Robertson GP, Hamilton SK. 2007. sity low-input and low-diversity high-input systems. Ecosystem services and agriculture: cultivating agricultural Biogeosciences. 6:1695–1706. ecosystems for diverse benefits. Ecol Econ. 64:245–252. White TA, Barker DJ, Moore KJ. 2004. Vegetation diversity, doi:10.1016/j.ecolecon.2007.09.020 growth, quality and decomposition in managed grasslands. Theau JP, Cruz P, Fallour D, Jouany C, Lecloux E, Duru M. Agric Ecosys Env. 101:73–84. 2010. Une méthode simplifiée de relevé botanique pour une Wilson PJ, Thompson K, Hodgson JG. 1999. Specific leaf area caractérisation agronomique des prairies permanentes. and leaf dry matter content as alternative predictors of plant Fourrages. 401:19–25. strategies. New Phytologist. 143:155–162. Valbuena D, Verburg PH, Bregt AK. 2008. A method to define a Zhang W, Ricketts TH, Kremen C, Carney K, Swinton SM. typology for agent-based analysis in regional land-use 2007. Ecosystem services and disservices to agriculture. research. Agricult Ecosys Env. 128:27–36. Ecol Econ. 64:253–260. International Journal of Biodiversity Science, Ecosystem Services & Management 143 Appendix Cut area 100% 80% 60% 40% 20% 0% D3 D1 B1 D2 B3 D4 B2 B4 Area grazed by cows 100% 80% 60% 40% B 20% 0% D3 D4 B2 D2 D1 B1 B3 B4 Area grazed by heifers 100% 80% 60% 40% 20% 0% B1 D1 B2 D2 B3 B4 D3 D4 Figure A1. Bar graphs for five GFTs ranked from fast growth strategy (A) to low growth strategy (D) for the dairy (Di) and beef (Bi) farms. Data were averaged for the whole grassland fields having the same use: cutting, grazing by cows and heifers. GFT (%) GFT (%) GFT (%) 144 M. Duru et al. Figure A2. Maps of vegetation diversity for two farms located in the same district. The four colours correspond to different percentages of GFT having an acquisitive growth strategy: w30% (white), 30–50 (light grey), 50–70(dark grey), >70% (black); the small map shows the location of the studied area within France.

Journal

International Journal of Biodiversity Science, Ecosystem Services & Management – Taylor & Francis

Published: Apr 3, 2015

Keywords: landscape; management; plant functional type; provision services; supporting services; trade-offs

There are no references for this article.