Agricultural intensity and the trade-off between production and biodiversity

Get Complete Project Material File(s) Now! »

Two mutually dependent objectives

About 40% of the global land area is currently dedicated to agriculture (Ramankutty et al., 2008). It represents a substantial area of semi-natural habitat for wildlife species that cannot be ignored. Conversely, biodiversity benefits agriculture through ecosystem services1. This mutual dependency means agricultural lands must be involved in biodiversity conservation along with natural reserves (Rosenzweig, 2003).
In Europe, the conservation value of farmland has long been understood. Farming is his-torically old, which provided the time for a large pool of species to adapt and specialize to agricultural land uses (Benton et al., 2002). Extensively managed, permanent grasslands are among the habitats with high biodiversity levels (Baldock et al., 1993; Bignal & McCracken, 1996). Extensive agricultural use (e.g., moderate grazing, mowing) positively affects biodiversity (Watkinson & Ormerod, 2001) and it is necessary to maintain grassland habitats, otherwise lost by ecological succession. In certain countries of Eastern Europe, the abandonment of agricul-tural activities is as equally threatening to biodiversity as agricultural intensification (Verhulst et al., 2004). To achieve conservation in tropical regions, focus is on mitigating the detrimen-tal effects of agricultural expansion and obtaining minimally-impacted reserve areas. Recent studies, however, advocate for more research that addresses conservation in human-modified landscapes (Chazdon et al., 2009). Some ancient production systems can sustain many species associated with native forests (Ranganathan et al., 2008). In return, biodiversity could benefit agriculture.
High biodiversity is needed for the health and resilience of ecosystems. It also contributes to the capacity of ecosystems to sustain services (Loreau et al., 2001; Tilman et al., 2001). As an agroecosystem, agriculture is a provider of ecosystem services (e.g., provisioning of food, fiber, and fuel). As a human activity, agriculture is also a user of ecosystem services (Zhang et al., 2007). Several ecosystem services are essential to agricultural production: soil structure and fer-tility, nitrogen fixation (supporting services), pollination, and pest control (regulating services). Past models of agricultural intensification overlook the importance of ecosystem services and rely on technologies to fulfill some of their function (e.g., pesticides instead of natural preda-tion or parasitoidism). Intensive agriculture degrades biodiversity and has negative impacts on ecosystem services (Giller, 1997; Kremen et al., 2002; Bianchi et al., 2007). Substituting some of the impacted services by technology, like soil fertility and pollination, would have a tremendous economic cost (e.g., manual pollination, hydroponic crop production).
Agricultural intensification may be able to keep increasing production to satisfy food de-mand in the short-term. Its detrimental effects on biodiversity and ecosystems, however, would ultimately undermine ecosystem services and threaten production in the long-term (Foley et al., biodiversity objectives are thereby crucial.

What are the solutions for reconciliation?

Two general visions on how to solve the conflict between agricultural production and biodiversity exist. Sustainable intensification seeks to improve water and nutrient use efficiency, in order to keep increasing yield, while limiting environmental damages (Cassman, 1999; The Royal Society, 2009; Godfray et al., 2010). Agroecology relies on ecosystem services to partially achieve the functions that are currently fulfilled by chemical inputs (Altieri, 2002). Both options are likely to be adequate, either in combination or according to agroecosystem type and intensity context. Agroecology principles are effective in certain tropical agroforested systems (Perfecto et al., 1996). In Europe, however, agricultural systems are more intensive and artificial, needing bigger transformations to restore functional agroecosystems. European agriculture, however, has a powerful means for transformation: agricultural policy.
The Common Agricultural Policy (CAP) of the European Union (EU) historically only supported production. To meet the sustainability challenge, Agri-Environment Schemes (AESs) were introduced in the CAP in 1992. AESs propose subsidies to farmers, based on voluntary compliance, for adoption of management practices that reduce environmental pollution, and preserve biodiversity and landscapes. These practices correspond to extensification at local and landscape scales, such as reduced fertilization, reduced stocking rates in grasslands, use of hedgerows, and strip maintenance. In 1999, the CAP added a second “pillar” (in addition to production support) that was dedicated to rural development. AESs have not managed to elicit extensification through significant changes in agricultural practices, yet they permitted to limit further intensification (CNASEA et al., 2008). Although a significant budget increase occurred after 1992, the effectiveness of AESs at reversing the biodiversity decline in farmland is debatable (Kleijn et al., 2006; Le Roux et al., 2008; Princ´e et al., 2012). This lack of effectiveness may be explained by uptake rates of AESs that are too low and too spatially diffuse to elicit biodiversity benefits on a large scale (Kleijn & Sutherland, 2003; Whittingham, 2007). Most schemes are not well adapted to the agro-ecological context because they have large application gradients where their effects on species can vary (Whittingham et al., 2007). Spatial targeting has been suggested as a way to improve the effectiveness of AESs, which adapts and concentrates measures at points where they are expected to yield the highest environmental benefits (Piorr et al., 2009; Uthes et al., 2010).
Because it drives the trade-off between production and biodiversity, agricultural intensity is a variable that should be adjusted in space to reconcile two objectives. A quantitative description of the effects of agricultural intensity and its spatial allocation on biodiversity is needed for effective reconciliation; it will help to determine the intensity range that should be targeted by conservation policies and the intensity allocation strategy that should be adopted.

The land sparing/land sharing framework

A debate has been ongoing in the literature about two contrasting intensity allocation strategies aimed at reconciling production and biodiversity: land sparing and land sharing (also called wildlife-friendly farming) (Cassman, 1999; Trewavas, 2001; Fischer et al., 2008; Perfecto & Van-dermeer, 2008). Green et al. (2005) formalized this debate by developing a theoretical model that answers the following question: for a given level of agricultural production, what allocation
of area to different land uses maximizes biodiversity level? This model relies on the shape of the relationship between biodiversity and agricultural yield (Fig. I.2). A convex relationship hypothesis (full curve) means that biodiversity exponentially decreases with yield (i.e., loss of either unfarmed or very extensively managed habitats is the most detrimental to biodiversity). Under this hypothesis, land sparing would be the best strategy. Part of a region would be spared at nul or very low intensity to fulfill conservation objectives, while the remaining area would compensate for the loss of productive land with high yielding intensive farming. On the other hand, a concave relationship hypothesis (dashed curve) means that biodiversity declines slowly, as intensity starts to increase, but becomes severely impacted at high intensity levels. Under this hypothesis, land sharing would be the best strategy. The entire region would be farmed at moderate intensities because they could achieve satisfying performances for both production and biodiversity criteria.
Land sparing follows a logic of segregation for production and biodiversity objectives, while land sharing follows a logic of integration. The two strategies can also be linked to reconciliation visions described in Section 1.3. Land sparing requires sustainable intensification that mitigates negative externalities, such as pollution outside cultivated areas (e.g., nitrogen runoff, eutroph-ication) and global impacts (e.g., greenhouse gas emissions, climate change). Conversely, land sharing creates opportunity to reintegrate biodiversity and ecological processes into agroecosys-tems, as suggested by the principle of agroecology and multifunctionality.
Empirical evidence about the shape of the relationship between biodiversity and intensity is still limited. In Ghana and India, Phalan et al. (2011b) found most bird and tree species displayed convex negative relationships with yield. In Europe, Kleijn et al. (2009) found more convex negative relationships between plant species richness and nitrogen input intensity. Re-lying on the Green et al. (2005) model, several authors concluded, from these convex relation-ships, that land sparing would be the best land use allocation strategy (Gabriel et al., 2009; Phalan et al., 2011a; Godfray, 2011). The Green et al. (2005) model is a good starting point to explore sustainable intensity allocation strategies, yet it could benefit from some important improvements. In particular, the spatial arrangement of agricultural intensity is likely to have an effect on biodiversity, which has not been tested in previous studies that address the biodi-versity/intensity relationship.

The importance of intensity spatial arrangement

Allocation strategies, as considered by the Green et al. (2005) model, only include intensity levels and their relative proportions. The model does not account for spatial arrangement. Land sparing, however, corresponds to an aggregated arrangement because the two intensity extremes are segregated in space. Several studies show that mobile species are impacted by neighbor land uses, and, thus, by the spatial arrangement of land uses and their intensity. Two main mechanisms, taking place at different scales, can explain this effect.
The first mechanism concerns the role of different land uses of the agricultural landscape during the life cycle of a species. For some species, different land uses are needed to fulfill essen-tial, complementary requirements (e.g., nesting and foraging habitats, Blomqvist & Johansson 1995). Other land uses produce resources with different qualities, such as food resources with different levels of availability (i.e., Brotons et al. 2005). In both cases, complex spatial arrange-ment provides species access to different resources within their habitat range (Dunning et al., 1992). Conversely, some land uses can be dangerous to species, and their scattered distribution within a complex spatial arrangement is detrimental (Fahrig et al., 2011). They include land uses with either high predation risk or intensive land uses, where input use has a direct negative impact on non-targeted organisms (Freemark, 1995; Bradbury & Kirby, 2006). The impact of spatial arrangement of resources, and dangers, takes place on a relatively short time scale (one reproductive season) and, thus, at a relatively small spatial scale (the individual habitat range).
The second mechanism concerns the impact of different land uses during species metapopu-lation dynamics (Macarthur et al., 1962; Levins, 1969). Agricultural landscapes can be striking examples of the metapopulation conceptual framework, such as when cultivated land has frag-mented the original, natural habitat (Andr´en, 1994; Verboom et al., 1991). Some species can only persist in patches of semi-natural habitats: their metapopulation dynamics consist of local population dynamics within these patches and spatial dynamics of migrations and colonizations amongst them. For such species, the most important properties of land use spatial allocation are when large enough patches, to sustain local populations, and patches close enough, to sustain the metapopulation dynamics, are available (Hanski, 1994; Steffan-Dewenter, 2000). Even for species persisting in spared patches of semi-natural habitats, the intensity of the surrounding agricultural matrix impacts their dispersal abilities and their metapopulation dynamics (Sut-cliffe et al., 2003; Donald & Evans, 2006). Furthermore, this agricultural matrix can be used as lower quality habitat in some cases (Baillie et al., 2000; Perfecto & Vandermeer, 2002). When this occurs, a source-sink dynamic exists between semi-natural and agricultural habitats, and the quality of the agricultural matrix also becomes important (Foppen et al., 2000). The land sparing allocation strategy confines biodiversity objectives to natural habitats that serve as re-serves. For species having metapopulation dynamics, however, the intensity of the surrounding agricultural matrix also has great impact on their viability. The impact of spatial arrangement, through metapopulation dynamics, occurs over several generations and, therefore, involves larger spatial scales (Devictor & Jiguet, 2007).

Policy targeting to influence intensity allocation

A corollary question of the land sparing/sharing model is: how can policy measures effectively promote the production/biodiversity reconciliation and influence land use allocation? Knowing the shape of the biodiversity/intensity relationship and the effect of intensity spatial arrangement is necessary to determine the intensity level that should be targeted by conservation measures and, therefore, be most effective. Policy targeting could also be a way to adjust the spatial allocation of intensity.
Under the hypothesis of a convex relationship, conservation policies promoting less intensive practices will elicit higher biodiversity benefits within extensively managed areas (Fig. I.3). Measures targeted at extensive areas may be more effective because they reinforce the quality of areas that already have high biodiversity and resource potential (Kleijn & Sutherland, 2003; Whittingham, 2007). Conversely, measures could seek to improve the quality of agroecosystems in intensive areas where negative impacts on biodiversity are most severe (Primdahl et al., 2003). Whether extensive or intensive areas are the most suitable for policy targeting will remain unanswered as long as the biodiversity/intensity relationship is unknown. Moreover, the effects of the spatial arrangement of intensity need to be understood to achieve effective policy targeting. At the landscape scale, local conservation measures that promote extensive management practices often yield higher biodiversity benefits when the surrounding landscape is intensive (Tscharntke et al., 2005; Concepci´on et al., 2008). However, this effect can vary between cropland and grassland landscapes (Bat´ary et al., 2011a). At larger scales, conservation measures seem more effective when the surrounding small region (Gabriel et al., 2010) or country (Kohler et al., 2007; Bat´ary et al., 2010) is less intensive.
Most AESs use untargeted measures; therefore, the distribution of their uptake in space cannot be controlled. Uptake is based on voluntary compliance results in a rather random distribution, although some bias is observed. In particular, uptake tends to be higher where the adaptation cost for the measure is lower (Osterburg et al., 2001; Kleijn & Sutherland, 2003). Conversely, the spatial targeting of conservation measures could be a way to adjust the spatial allocation of intensity in order to promote either the segregation of extreme intensities (land sparing) or more heterogeneity with moderate intensities (land sharing). Knowing the spatial distribution of agricultural intensity is, therefore, important for targeting policies and modulating intensity allocations. For instance, Gabriel et al. (2009) mapped the proportion of organic farms in the UK and found that land sparing allocation at the country scale could be achieved by targeting the measures that endorse organic conversion in the areas where it is already aggregated.

READ  ACCOUNTING-BASED METHODS TO DETERMINE SHAREHOLDER VALUE

Adjusting intensity and its allocation: from concept to prac-tice

The biodiversity/intensity relationship, land sparing/sharing allocations, and the targeting of policy measures are connected. Together, they provide a conceptual framework for the recon-ciliation of production and biodiversity objectives in farmlands. How should one apply this conceptual framework to real cases?

Defining the land sparing and land sharing allocations

In real cases, scale, ecosystem type, and species influence the definitions of land sparing and land sharing intensity allocations.
Land sharing corresponds to a variability of land uses at a finer spatial scale (or grain) than in land sparing (Fischer et al., 2008). Indeed, several land sharing practices favor landscape heterogeneity: crop diversity, field margins, tree edges and clumps (Phalan et al., 2011a). How does one determine the level at which heterogeneity is considered land sparing? For instance, a field margin could be considered land sparing on a field scale, and either a grassland field or a tree clump could be considered land sparing on a farm scale. Phalan et al. (2011a) argue that both land sparing and land sharing should not be considered the same at different scales. The authors suggest that spared habitat should be sufficiently large to support viable populations. The land sparing definition, therefore, depends on species.
In tropical ecosystems, most of the endemic biodiversity occurs in the pristine forest (Brooks et al., 2002), an unexploited land cover which intensity can thereby be considered zero. Recent and brutal transition from forest to land uses exploited with moderate intensity, led to the loss of most species. In Europe, high biodiversity levels are found in several unexploited land areas (e.g., mountains, forests, wetlands, littoral zones). In contrast to tropical ecosystems, land use exploited with moderate intensity can also show very high biodiversity levels (e.g., permanent grasslands, Bignal & McCracken 1996). These two contrasted examples show that the type of ecosystem can influence what should be considered reserve habitat in a land sparing strategy. In Europe, most unexploited habitats already belong to reserves. Agricultural habitats represent a central conservation issue: they cover more than half of the non-urban area (CLC, 2006) and their farmland specialist species are at high risk from intensification (Section 1.1). Farmland species can not live in the unexploited part of the intensity gradient. The definition of a spared habitat is thus unclear: grasslands could be considered land sharing because they combine production and biodiversity objectives, or, alternatively, as land sparing because they also host unique biodiversity that should be preserved.
The land sparing and land sharing definitions are justified by the details (scale, species, ecosystem) of each case study. Adopting a scale relevant for both intensity allocation and policy targeting allows one to discuss the implications of results with regard to the above elements. The biodiversity/intensity relationship is the other component of the land sparing/sharing model that could vary according to each case study.

Computing the biodiversity/intensity relationship

What intensity measure?
Agricultural intensity can be defined as increased utilization or productivity of land (Netting, 1993), therefore, either output-oriented (production) or input-oriented (utilization) measures can be used to describe it (Turner & Doolittle, 1978; Dietrich et al., 2012). In the Green et al. (2005) model, yield is used to compute the biodiversity/intensity relationship. Yield provides direct insight into the trade-off between food production and biodiversity, yet management intensity is more likely to impact biodiversity.
Several studies have used yield as an indirect proxy of farming intensity because it reflects industrialization, specialization, and/or input use (Donald et al., 2001; Herzog et al., 2006a). Either the concave or convex negative density-yield functions of Green et al. (2005) represent either tolerant or less tolerant species to farming intensity. Although yield correlates with man-agement intensity, it also depends on pedo-climatic conditions. Most studies, therefore, address the effect of farming intensity on biodiversity in ways that directly focus on management prac-tices, in order to reveal their impact and understand the underlying mechanisms. Management practices can be intensive on two main input components: the chemical (or more generally mat-ter) inputs and the work inputs. Chemical input intensity concerns the higher use of fertilizers, pesticides, irrigation or seed at the field and farm scale (Giller, 1997). Management practices related to chemical inputs are interesting because they have a strong impact on biodiversity, but also on other environmental issues. In Europe, work inputs mainly relate to mechanization, which largely explains the important landscape simplifications associated with intensity (Bj¨ork-lund, 1999). With large sets of practices and landscape properties, a wide range of indicators have been used to address the effects of intensity on biodiversity. We summarize some of these indicators, along with their mechanisms of impact on bird species (Table S.s1 in Appendix).
Indicator categories focusing on particular practices or landscape properties (Table S.s1 in Appendix) are ecologically relevant because they involve underlying mechanisms of impact on biodiversity. They address, however, single mechanisms, and their narrow definition of intensity make them less relevant from an agricultural viewpoint. In particular, they are less related to yield, which is a disadvantage when studying the trade-off between food production and biodi-versity. Some studies have tried to build indicators that combine several intensity components. Herzog et al. (2006a) normalized nitrogen input, livestock density, and pesticide input into one indicator, and Pointereau et al. (2010) computed a score that accounted for management, crop diversity, and landscape components. The main difficulty for composite indicators is determin-ing how to combine the different components. Every method has disadvantages: The min/max normalization method produces relative values only; and a scored value requires arbitrary com-putational choices. The strength of such indicators, however, is that they simultaneously account for several ecologically relevant components and provide a more complete vision of intensity that is likely more closely related to yield.

Table of contents :

A General Presentation 
I Introduction 
1 Agricultural intensity and the trade-off between production and biodiversity
1.1 Two conflicting objectives
1.2 Two mutually dependent objectives
1.3 What are the solutions for reconciliation?
2 Adjusting intensity and its allocation to meet production and conservation objectives
2.1 The land sparing/land sharing framework
2.2 The importance of intensity spatial arrangement
2.3 Policy targeting to influence intensity allocation
3 Adjusting intensity and its allocation: from concept to practice
3.1 Defining the land sparing and land sharing allocations
3.2 Computing the biodiversity/intensity relationship
4 Research question
II General Approach 
1 Case study
2 Describing agriculture
2.1 The Input Cost/ha intensity indicator
2.2 Studying the aggregation of intensity among small agricultural regions
2.3 The grassland/arable land gradient and heterogeneity within small agricultural regions
3 The farmland bird community
3.1 Data from the French Breeding Bird Survey
3.2 The farmland bird community and its descriptors
4 Analyzing the effect of agriculture on the bird community
4.1 The correlative approach
4.2 Statistical methods
4.3 Exploring the effect of intensity allocation modifications
III Summary of the results 
1 Describing the spatial distribution of intensity at the French country scale, with SAR resolution
1.1 The distribution of intensity shows spatial structure
1.2 Moderate correlation between the intensity gradient and the land use gradient
2 Analyzing the response of the bird community to agricultural intensity
2.1 Homogeneity benefits specialists, heterogeneity benefits generalists
2.2 The response of birds to intensity: sharper in the extensive range, winner and loser species
3 The spatial aggregation of intensity influences the bird community/intensity relationship
4 Exploring optimal intensity allocations to overcome the production/biodiversity trade-off
4.1 Calibrations: intensity links biodiversity and production
4.2 Intensity allocations draw the trade-off between production and biodiversity, and reveal win-no-lose solutions
5 Targeting intensity changes
5.1 The spatial structure of intensity: several clusters with significant aggregation
5.2 Reaching optimal allocations: targeted intensity changes, opposite for extensification and intensification
IV Discussion 
1 Ph.D. contributions: generalization potentials and restrictions
1.1 Describing and mapping agricultural intensity
1.2 The farmland bird community focus
1.3 The scale studied
1.4 European perspectives
2 The land sparing/sharing framework applied in our case study: the importance of mixed strategies
2.1 The biodiversity/intensity relationship in the European context
2.2 The community level reveals how winners substitute losers
2.3 Accounting for spatial arrangement: land sparing can influence the biodiversity/ intensity relationship
2.4 The importance of mixed allocation strategies
3 Targeting conservation policies to improve their effectiveness
3.1 Current distribution of the conservation policies
3.2 What kind of targeting will lead to future improvements?
4 Perspectives
4.1 Down-scaling: the intensity changes at the farm level
4.2 Up-scaling: policy options to target intensity changes
4.3 From biodiversity to other criteria and ecosystem services
Supplementary material
s1 Agricultural intensity: several measures and effects
s2 The efficiency lever
B Articles
V A novel method for mapping agricultural intensity reveals its spatial aggregation: implications for conservation policies 
1 Introduction
2 Methods
2.1 Data
2.2 Input Cost/ha intensity indicator computation
2.3 Estimation method of the Input Cost/ha at SAR resolution
2.4 Testing the intensity aggregation
3 Results
3.1 Agricultural intensity and input categories distribution across production types
3.2 Intensity estimation with SAR resolution
3.3 Spatial aggregation of farming intensity
4 Discussion
4.1 Weakness and strength of the intensity indicator and its estimation
4.2 The spatial aggregation of intensity and its implications for conservation policies
5 Conclusions
References
VI Mixed benefits of compositional and configurational heterogeneity on farmland birds according to their habitat specialization 
1 Introduction
2 Methods
2.1 The French Breeding Bird Survey
2.2 Trait-based species groups
2.3 Land uses and heterogeneity
2.4 Statistical analysis
3 Results
3.1 Arable/grassland ratio vs heterogeneity
3.2 Effects of compositional and configurational heterogeneity on the bird community
4 Discussion
4.1 Implications for conservation
5 Conclusion
References
VII The spatial aggregation of agricultural intensity influences the variations of farm- land bird communities on a nationwide intensity gradient
1 Introduction
2 Methods
2.1 Agricultural intensity and its aggregation
2.2 Bird data
2.3 Statistical analysis
3 Results
3.1 Effects of agricultural intensity on the bird community
3.2 Interacting effect of intensity and aggregation
4 Discussion
4.1 Underlying mechanisms of the effect of intensity and its aggregation
4.2 The significant effect of intensity aggregation: implications for conservation
References
Supplementary material
s1 Effect of the other explanatory variables than intensity
s2 Linear vs non-linear models for the bird community/intensity relationship
s3 Correlation between intensity and land uses
s4 Detectability of the bird species
VIII Optimal targeting of agricultural intensity allocation to reconcile production, economy and biodiversity at the countrywide scale 
1 Introduction
2 Methods
2.1 Conceptual model
2.2 Data
2.3 Statistical calibrations
2.4 Simulations
3 Results
3.1 The trade-offs between criteria among all intensity allocations
3.2 The targeted intensity changes leading to optimal allocations
4 Discussion
4.1 Variables used
4.2 Complementarity with other approaches
4.3 Policy implications
References
Supplementary material
s1 Species list of the farmland bird community
s2 Calibrated relationships between the three criteria and intensity
s3 The trade-offs between the economy and alternative biodiversity criteria
General Bibliography

GET THE COMPLETE PROJECT

Related Posts