Eco-Evolutionary Dynamics of Agricultural Networks

CHAPTER SIX

Eco-Evolutionary Dynamics of Agricultural Networks: Implications for Sustainable Management Nicolas Loeuille
,†,1, Sébastien Barot{, Ewen Georgelin
,†, Grigorios Kylafis
,†, Claire Lavigne} *Laboratoire EcoEvo, UMR 7625, UPMC, Paris, France † Laboratoire Ecologie des Populations et des Communaute´s, USC INRA 2031, Paris, France { Laboratoire BIOEMCO, UMR 7618, IRD, Paris, France } Laboratoire Plantes et Syste`mes de culture Horticoles, UR1115, INRA, Avignon Cedex, France 1 Corresponding author: e-mail address: [email protected]

Contents 1. Introduction 2. Within Field, Applying Evolutionary Perspectives to the Selection of Agricultural Species 2.1 General effects of domestication and selection 2.2 Beyond the one trait approach: Accounting for trade-offs 2.3 Adapting to local practices and conditions: The importance of diversity 2.4 Beyond individuals: The influence of selection processes on the community and ecosystem context 3. Disturbances Due to Agriculture: Implications for Eco-Evolutionary Dynamics Within Surrounding Ecosystems 3.1 Nutrient enrichment and its ecological and evolutionary consequences in agricultural landscapes 3.2 Chemical warfare in agricultural landscapes: The ecological and evolutionary consequences of pesticide use 3.3 Effects of altering species composition and relative abundance of species in agricultural landscapes 4. Accounting for Spatial Heterogeneities: Dispersal, Fragmentation, and Evolution in Agricultural Landscapes 4.1 Characteristics of agricultural landscapes, past, present and future 4.2 Consequences of agricultural landscape structure in terms of gene flow 4.3 Consequences of spatial modifications from a demographic point of view 4.4 Consequences of spatial structure for pairwise co-evolution 4.5 Beyond pairwise interactions: Consequences of eco-evolutionary dynamics for community structure and composition 4.6 Land sparing versus land sharing, from an evolutionary point of view Advances in Ecological Research, Volume 49 ISSN 0065-2504 http://dx.doi.org/10.1016/B978-0-12-420002-9.00006-8

#

2013 Elsevier Ltd All rights reserved.

340 344 344 348 351 353 360 361 365 372 376 376 380 385 388 390 393

339

340

Nicolas Loeuille et al.

5. Perspectives and Challenges Acknowledgements Appendix A. Evolution of the investment into nutrient uptake, effects on emergent functioning Appendix B. Mixing group selection and individual selection in co-evolutionary models Appendix C. Effects of enrichment on the control of biomass within a tri-trophic food chain when the herbivore evolves Appendix D. Eco-evolutionary dynamics in a plant–herbivore–predator confronted to insecticides Appendix E. Evolution of specialization rate of the pest and its ecological consequences Glossary References

395 398 398 401 402 407 413 416 417

Abstract Community and ecosystem ecology are paying increasing attention to evolutionary dynamics, offering a means of attaining a more comprehensive understanding of ecological networks and more efficient and sustainable agroecosystems. Here, we review how such approaches can be applied, and we provide theoretical models to illustrate how eco-evolutionary dynamics can profoundly change our understanding of agricultural issues. We show that community evolution models can be used in several contexts: (1) to improve the selection of agricultural organisms within the context of their ecological networks; (2) to predict and manage the consequences of agricultural disturbances on the ecology and evolution of ecological networks; and (3) to design agricultural landscapes that benefit from network eco-evolutionary dynamics, but without negative impacts. Manipulation of landscape structure simultaneously affects both community ecological dynamics (e.g., by modifying dispersal and its demographic effects) and co-evolution (e.g., by changing gene flows). Finally, we suggest that future theoretical developments in this field should consider appropriate co-evolutionary models and ecosystem services.

1. INTRODUCTION Models and experiments in ecology have increasingly incorporated evolutionary dynamics in the context of community structure and ecosystem functioning (Agrawal et al., 2012; Bra¨nnstro¨m et al., 2012; Loeuille and Loreau, 2009). Such models have been coined ‘community evolution models’ (Bra¨nnstro¨m et al., 2012; Loeuille and Loreau, 2009). They link evolutionary processes driven by individual fitness to emergent properties of ecological networks, linking small-scale processes to large-scale patterns.

Eco-Evolutionary Dynamics in Agroecosystems

341

Evolutionary dynamics affect important structural attributes of ecological communities, such as connectance or the number of trophic levels of a food web, via the evolution of body size (Loeuille and Loreau, 2005) or adaptive foraging (Beckerman et al., 2006). Similarly, the structure of mutualistic networks (e.g., plant–pollinator or plant–seed disperser) largely depends on the co-evolution of partner species (Bascompte et al., 2006; Nuismer et al., 2012). Evolution also modifies ecosystem functioning, in ways that are particularly relevant to agricultural systems: biomass, productivity, nutrient cycling or ecosystem sustainability are all potentially affected. It can also produce counter-intuitive effects (Abrams and Matsuda, 2005) and the interplay of ecological and evolutionary dynamics (hereafter, eco-evolutionary dynamics) changes the distribution of nutrients within ecosystems (Boudsocq et al., 2011), the way nutrients are recycled (De Mazancourt et al., 2001; Loeuille et al., 2002) and their spatial distribution (Loeuille and Leibold, 2008a). Evolution also constrains the resilience of ecological assemblages (Kondoh, 2003; Loeuille, 2010a, b), thereby affecting the sustainability of the system. Agricultural development is based on a selective process (Gepts, 2004), so it is particularly important to incorporate rigorous evolutionary thinking into it. Human populations have long chosen species and favoured traits that increase the productivity of cultivated organisms or make their cultivation easier. Because selective processes are at the heart of agricultural developments, it should be possible to implement current progresses linking trait evolution with emergent properties of ecosystems, to improve selection and the management of agricultural landscapes. An immediate difficulty, however, results from differences in the ways selective processes are considered in ecological versus agricultural contexts. Because evolutionary concepts are often phrased differently in ecology and agriculture, precise definitions of key concepts are necessary (see Glossary). Selection in agriculture is driven by the choice, conscious or otherwise, of cultivated species and of their characteristics (Meyer et al., 2012). Such a choice is often based on total biomass or yield (Perales et al., 2004; Vigouroux et al., 2011) and the process de facto relies on a group selection criterion (Denison et al., 2003, but see Duputie´ et al., 2009). In contrast, evolutionary ecology usually considers phenotypic changes based on individual or gene fitness (Dieckmann and Law, 1996; Fisher, 1930; Hamilton, 1964). Thus, fitness in community evolution models can be complex, due to the need to account for multispecies assemblages and interaction networks (Ito and Ikegami, 2006; Loeuille and Loreau, 2005), as well as various disturbance scenarios (Loeuille and Loreau, 2004; Norberg et al., 2012).

342

Nicolas Loeuille et al.

Here, we will discuss what agriculture can learn from these community evolution models, as well as associated experimental and empirical works in evolutionary ecology. Particularly, are there lessons to be learned to reconcile economic goals (e.g., an appropriate level of food production) with the desire to conserve ecological networks? A few reviews have already shown how evolutionary arguments can be applied to agricultural systems (Denison, 2012; Denison et al., 2003; Thrall et al., 2011), by providing case studies and examples of the many ways in which we can use evolutionary ecology to design a better agriculture. Here, although some arguments will be based on single-species evolution studies, we will focus on tackling the links between evolutionary and community contexts, drawing extensively on community evolution models and ecological theories. We divide our central theme into two parts. (i) we ask how do community evolution models help us to guide the choice of cultivated species, traits or genes for a sustainable agriculture? This requires the assessment of how these choices affect the co-evolution of species within associated ecological networks and their consequences in terms of yield and sustainability. (ii) we ask, can we use community evolution models to predict the consequences of disturbances due to agriculture on surrounding ecosystems? Such disturbances include fertilization, pesticide use, changes in species community composition and landscape modifications. Empirical observations suggest that each of these disturbances has far reaching implications for the ecology and evolution of the natural communities that abut agricultural systems (Klein et al., 2007; Tscharntke et al., 2005). We also ask whether current ecological and evolutionary theory can provide guidelines to limit the consequences of these disturbances and preserve ecological networks. It is necessary to point out that agricultural systems vary considerably in how they create heterogeneities and disturbances (Male´zieux, 2011; Massol and Petit, 2013, Chapter 5 of this volume). From a low-intensity gatherer mode of farming to intensive industrialized agriculture, a continuum exists along which disturbances increase. Male´zieux (2011) distinguishes ‘traditional agriculture’ from ‘intensive agriculture’ and warns that much of the surface currently occupied by agriculture is not, despite common misconceptions, intensively managed. He also proposes to manage agriculture through ‘ecological intensification’, relying on services provided by ecological networks (Dore´ et al., 2011; Male´zieux, 2011). Here, we will try to state for each example whether it is applicable to intensive or traditional agriculture, within agricultural fields or in surrounding ecosystems. Justifications for questions (i) and (ii) can take several forms. One important motivation is the conservation of species diversity. Agricultural

Eco-Evolutionary Dynamics in Agroecosystems

343

development has triggered marked declines in species diversity (Robinson and Sutherland, 2002), although some species, such as those associated with downland, heathlands and other semi-natural or farmed landscapes, have benefited (Eriksson, 2012). The maintenance of (at least some) diversity is important, not only for the intrinsic value of species, but also because it generally enhances ecosystem functioning (Loreau et al., 2001), and another important justification emerges from this point. When ecosystems within the field and in the surrounding landscape are functioning effectively, human populations receive important economical benefits, commonly called ‘ecosystem services’ (Costanza et al., 1997; Raffaelli and White, 2013; Bohan et al., 2013, Chapter 1 of this volume). Several of these ecosystem services directly benefit agriculture, such as pollination (Klein et al., 2007; Tscharntke et al., 2005), biological control of pests by pathogens and predators (Crowder et al., 2010; Macfadyen et al., 2011; Thrall et al., 2011) and the efficient recycling of nutrient and maintenance of soil fertility (Young and Crawford, 2004). Community assembly and co-evolution have built the ecological networks that provide these services, so community evolution models should be able to explain the impact of agriculture on them. We organize the text in three different parts, focusing on evolutionary ecology within fields, around them, and then across the whole agricultural landscape. In the first part, within agricultural fields, we ask how evolutionary ecology in general and community evolution models in particular can guide the choice of cultivated species and of their traits, to account simultaneously for agricultural yield and ecological sustainability. In the second part, concerning adjacent ecosystems, we propose that community evolution models can be used to discuss the consequences of agricultural disturbances on the ecology and evolution of natural networks. The third part, integrating agricultural fields and their adjacent ecosystems, investigates how spatial evolutionary models may guide the landscape management of agricultural systems. In each part, we build the arguments by increasing levels of organizational complexity, moving from a species perspective (often the cultivated species), to pairwise interactions, then to community structure and finally to ecosystem functioning. A table summarizing main arguments and key references introduces each part. At several junctures, we illustrate our statements by developing new models tightly focused on a key question. The main result of each model is then presented in a box, while detailed hypotheses and analyses are provided in appendices. In the fourth and final part, we finish by indicating possible future developments.

344

Nicolas Loeuille et al.

2. WITHIN FIELD, APPLYING EVOLUTIONARY PERSPECTIVES TO THE SELECTION OF AGRICULTURAL SPECIES 2.1. General effects of domestication and selection Domestication is the outcome of artificial selection that leads to increased adaptation of plants and animals to cultivation by humans (Gepts, 2004). The most detailed studies concern crop domestication (Gepts, 2004; Harlan et al., 1973) and we will focus on this literature here as a complete review is beyond the scope of the present chapter. Instead, we will focus on the main phenotypic traits that have been selected via domestication then show that most selected strategies would perform poorly in natural selection settings. Inevitably, such a selection procedure requires extensive use of continued inputs to maintain the cultivated species and traits, posing a sustainability problem. About 2500 plant species have been domesticated (Meyer et al., 2012) in an ongoing process that is rooted in the transition from hunter-gatherers to settled agriculture during the Neolithic. The first steps of domestication were probably unconscious: wild plants were simply translocated across newly man-made environments, which altered selective pressures. However, further selection of varieties has probably been conscious for a very long time. Artificial selection inspired much of Darwin’s theory of evolution by natural selection (Darwin, 1859) and he later dedicated a whole book to this subject (Darwin, 1868) after being struck by the strength and pace of artificial selection and the obvious link it makes between present selective pressures and subsequent evolution. Many different traits are often selected in several domestication events, defining a domestication syndrome (Gepts, 2004; Harlan, 1992; Meyer et al., 2012; Purugganan and Fuller, 2009). In cereals, two types of traits are essential: an ability to germinate deep in the soil in open, disturbed habitats (e.g., large seeds, loss of dormancy) and traits that facilitate harvesting (e.g., non-shattering seeds, maturation synchrony, reduced branching, and resistance to tilling). Domestication also involved the loss of secondary metabolites, allowing seeds or fruits to taste better or to be more easily digested (Meyer et al., 2012). Changes in the reproductive strategy are frequently encountered, from outcrossing to selfing or from sexual to vegetative reproduction. Such changes made further selection easier, as they increase reproductive isolation and facilitate the rapid fixation of selected traits.

345

Eco-Evolutionary Dynamics in Agroecosystems

Take home messages and key references for section 2 Agriculture and crop Ecology and evolutionary ecology Section breeding

References

Meyer et al., 2012 Purugganan and Fuller, 2009

Section Domestication and crop 2.1 breeding is a diversified evolutionary process impacting many plant traits, favouring easy cultivation and higher yields

Domestication and artificial selection have selected traits that could not evolve through natural selection, due to differences in selective regimes

Section Crop breeding often involves group selection 2.2 based on collective properties of crops such as yield. Such a selection has favoured fast-growing crops, incurring costs in terms of competitive abilities

Existing allocation trade-offs Denison between growth rates and et al., 2003 many different traits (e.g., competitive ability, defences) constrain the evolution of these traits

Section Crop breeding has resulted 2.3 in some homogenization of crop plants associated with the development of agricultural practices allowing homogeneous environmental conditions through high inputs. An unsettled controversy exists regarding the degree of desired local adaptation of crops to different agricultural practices. Modern crop breeding has led to a drastic reduction in within field genetic diversity

Natural selection selects diversified strategies as soon as environment is not totally homogeneous. Ecology increasingly views genetic diversity as an important factor for ecosystem functioning

Thrall and Burdon, 2003 Murphy et al., 2007 Van Bueren et al., 2008 Crutsinger et al., 2006

Section Results from community 2.4 evolution models suggest that crop breeding could be constrained by trade-offs between yield and sustainability

Allocation and ecological trade-off are likely to result in trade-offs at the community or ecosystem scale, including between different ecosystem services

Fussman et al., 2007 Strauss et al., 2002 Box 6.1 Box 6.2

346

Nicolas Loeuille et al.

The selective process changes widely among agricultural systems: in traditional cropping systems, farmers manage local landraces based on networks of seed exchanges (Pautasso et al., 2013), whereas since the end of the nineteenth century, selection has been increasingly undertaken by specialized farmers, public organisms and private companies. This specialization of roles has allowed the development of sophisticated methods using the latest developments of biological sciences: genetics, molecular markers and biotechnologies, and the development of a global multinational agroindustry (Dennis et al., 2008; Pennisi, 2008). Such modern changes in selection methods are intertwined with the development of modern agriculture based on mechanization, and the use of fertilizers and pesticides. Crop breeding has targeted varieties with higher yields and three major staple crops—wheat, rice, and corn—provide about 45% of human-ingested calories worldwide. The grain yield of these cereals doubled during the twentieth century, modern crop breeding methods playing a key role in this increase (Richards, 2000). A trait related to yield, the harvest index (ratio of the grain biomass to the whole aerial biomass), has also doubled, reaching values as high as 0.5 in most cereals (Hay, 1995; Richards, 2000). Increases in the harvest index have been obtained by selecting increased allocation to reproductive parts and decreased allocation to leaves and stems. The development of dwarfed varieties during the Green Revolution has led to shorter and stiffer stems that are less sensitive to falling over, or lodging (caused by bending or breakage of the stem or root problems), even at high fertilization rates. Prior to the 1970s, these trends were mostly driven by ‘defect elimination’ and ‘selection for yield’ (Donald, 1968). The first corresponds to removing major flaws, such as weak straws for cereals or fragile skins for tomatoes. The second is based on the need to increase production, regardless of the plant traits involved (Donald, 1968). Harvest indices have seemingly plateaud (Hay, 1995), and crop breeding now targets traits that appear most promising to increase yield: resistance to pests and pathogens and to unfavourable conditions (Witcombe et al., 2008), such as drought (Pennisi, 2008) or saline soils (Richards, 1992). This is crucial for exploiting new, less favourable, land and to cope with global changes (e.g., climate change, soil degradation). Other selected traits focus on the efficient acquisition and recycling of resources. Such research aimed at optimizing root systems (De Dorlodot et al., 2007) and at increasing their capacity to take up phosphorus (Gahoonia and Nielsen, 2004). Artificial selection of nutrient use also goes beyond individual plant traits to consider the complex relationships with soil microbial communities. Nutrient management is

Eco-Evolutionary Dynamics in Agroecosystems

347

a key point for future agricultural sustainability because: (1) mineral fertilizers are not produced sustainably (phosphorus is dug in mines, nitrogen is industrially fixed using fossil fuels as an energy source); (2) fertilizers leak from agroecosystems, altering functioning and contributing to greenhouse gas emissions (e.g., N2O) (Vitousek et al., 1997). Artificial selection could also aim to increase photosynthetic rates (Long et al., 2006; Richards, 2000), but when expressed per unit land area these have mostly increased through inputs (irrigation, inorganic fertilizer). Whether higher photosynthetic rates through crop breeding are possible is still hotly debated (Denison et al., 2003), and it would require manipulation of many genes simultaneously, while most modern crop breeding usually targets just a few genes. Domestication and artificial selection alter the composition of communities at all spatial scales: out of 2500 domesticated species, 20 major crops cover about 12% of the land surface (Leff, 2004), while the worldwide number of seed plant species is estimated at about 260,000. About 22% of continental surfaces are used as pastures and rangelands planted to some extent with domesticated grasses (Gle´min and Bataillon, 2009). In addition to reducing species diversity, modern agriculture and crop breeding have developed a restricted number of pure or hybrid lines so that genetic diversity is drastically reduced, both locally and globally (Haudry et al., 2007). Reduced genetic diversity can happen early in the domestication due to bottlenecks, with early farmers using a limited number of plant individuals as progenitors (Reif et al., 2005). A second step in the genetic erosion occurred when pure or hybrid lines were preferred to local landraces. Genetic diversity in bread wheat cultivated in a region of France, for instance, has halved since 1878 (Bonneuil et al., 2012). On the contrary, farmers in traditional systems often increase genetic diversity through the creation of local landraces adapted to different local uses, tastes, agricultural practices, and environments (Elias et al., 2001; Pautasso et al., 2013). Can evolutionary ecology shed new light on the whole process of domestication and crop breeding and can it improve crop breeding and the sustainability of agriculture? Evolutionary arguments are not always present in the recent crop breeding literature (but see Harlan et al., 1973). For example, whole review papers on crop breeding do not use the term ‘evolution’ in its Darwinian sense (Good et al., 2004; Zhang, 2007). Only some relatively recent initiatives have appealed to more evolutionary-oriented thinking in agriculture and crop breeding (Denison, 2012; Denison et al., 2003; Gepts, 2004; Thrall et al., 2011).

348

Nicolas Loeuille et al.

Artificial selection has led to species and varieties that depend on humans to reproduce, as they have lost the suitable reproduction and dispersal strategies, and to grow, with the need for nutrient addition and physical protection. It tends to produce organisms whose traits would usually be disfavoured by natural selection. The loss of traits allowing independent reproduction is only possible because artificial selection is based on group level criteria, such as yield (Denison et al., 2003). Such traits incur low individual fitness and would be expected to disappear through natural selection. For instance, non-shattering stems would be disfavoured in nature, because seed would not fall to the ground, but are systematically selected in harvesting systems where humans want to collect the seed before the seed is lost (Harlan et al., 1973). In some cases, improving crop varieties has reversed the effects of past natural selection. Thinking of natural selection therefore allows an a priori assessment of future crop breeding avenues: Denison et al. (2003) hypothesize that natural selection is likely to have already optimized the major physiological pathways of plants (e.g., photosynthesis), so significant further improvement through crop breeding is unlikely. Because the units of selection are different in artificial and natural selection, most evolutionary models developed in plant ecology have no straightforward links with crop breeding. Some theoretical models study the evolution of traits relevant to crops, including growth rate, nutrient uptake efficiency (Boudsocq et al., 2011), defences against herbivores (Loeuille and Leibold, 2008a), seed size (Geritz et al., 1999), plant height (Falster and Westoby, 2003), and shoot–root ratio (Vincent and Vincent, 1996). However, many crop traits are entirely anthropocentric and essentially irrelevant in natural systems, such as those linked to the taste of the food or allowing an easier harvest (e.g., non-shattering grains).

2.2. Beyond the one trait approach: Accounting for trade-offs While we have listed above a list of traits targeted by artificial selection, they should not be considered in isolation. Trade-offs may cause negative correlations between two or more traits, so that considering just one does not give an adequate description of selection dynamics. This concept is at the very heart of most evolutionary models and the influence of trade-offs has been systematically investigated in this context (De Mazancourt and Dieckmann, 2004; Loeuille and Loreau, 2004). Trade-offs also exist between traits relevant to artificial selection and their analysis is critical in terms of agricultural

Eco-Evolutionary Dynamics in Agroecosystems

349

management, although they are often overlooked in such a context (Denison, 2012; but see Thrall, 2013). Trade-offs can emerge due to allocation constraints of resources, time, or space. For instance, resources allocated to growth cannot be used for reproduction or to produce defensive compounds against herbivores or pathogens (Herms and Mattson, 1992). In the same vein, opening stomata allows photosynthesis but increases losses of water by evapotranspiration. We list below the trade-offs that are commonly considered in evolutionary ecology and that are relevant to agroecosystems. In ecology, selection for yield could be related to the r–K theory (Pianka, 1970), which can emerge from existing trade-offs between individual traits (Rueffler et al., 2006a) High yield, fast growth rates (r strategies) are thus traded-off against traits that improve competitive ability (K strategies), such as large size, long generation time, so that crop breeding is likely to produce poor competitors (Denison et al., 2003; Donald, 1968; Weiner, 2004). Crop individuals have higher yield because they spare the resource otherwise required to interfere efficiently with other individuals. If only conspecifics are present, having poor-competitor r strategists in turn decreases competition, and yield is further increased (Donald, 1968). It also requires a release from competition with wild plants. This type of selection is therefore dependent on tillage and herbicides. The loss of competitiveness favours short stems (dwarfed varieties), upright and erect leaves and lower individual leaf surface area, whereas natural selection usually selects against these traits. Recognition of the growth/competition trade-off raises the question of the optimal strategy for achieving more sustainable agriculture. In a succession context, fast-growing r strategies are progressively replaced by K strategies. In agriculture, artificial selection maintains r strategies in the field. From an evolutionary and ecosystem ecology point of view, this maintenance at early stages of succession cannot be a sustainable strategy for two reasons. First, this situation is far from equilibrium, so that the system is susceptible to abrupt changes, low resilience and invasions (Odum, 1953). Second, this maintenance of r strategies requires large inputs. Lower-yielding phenotypes would allow lower rates of fertilization, as suggested by a theoretical model (Zhang et al., 1999) and experiments (Gersani et al., 2001). They could also tolerate cultivation on marginal soils or under stressed conditions. Growth rate is also often negatively correlated with defences against herbivores or pathogens, as resources allocated to growth cannot be used for both purposes (Herms and Mattson, 1992). Also, the allocation of mineral

350

Nicolas Loeuille et al.

nutrient to roots or leaves should attract pathogens and herbivores (Mattson, 1980; White, 2005). Most often, in nature, intermediate strategies along the growth/defence trade-off are selected (De Mazancourt et al., 2001; Loeuille and Leibold, 2008a; Loeuille and Loreau, 2004; Loeuille et al., 2002). Crops selected on yields in the absence of pests are again at one extreme of the continuum. They clearly depend on inputs (pesticides, fungicides) to compensate, again raising sustainability issues. Trade-offs also limit the improvements that are possible through crop breeding, suggesting great difficulties in enhancing photosynthetic rate (Denison et al., 2003). Without accounting for trade-offs, some scientists have proposed the breeding of a Green Super Rice (Zhang, 2007), with increased: (1) resistance to herbivores, (2) resistance to drought, (3) nutrient use efficiency, (4) grain quality, and (5) yield. Most of these traits are likely to be linked via antagonistic constraints, for instance due to resource allocation. As a more general illustration of the lack of consideration given to these constraints, we have searched ISI Web of Science using the terms ‘crop breeding’ and ‘trade-off’ (in keywords, title, and abstract) and only found 27 articles. Ideally, for any selection procedure, existing trade-offs among targeted traits should be assessed before deciding which combination is more desirable, and indeed feasible, using a hierarchical set of criteria (yield, crop quality, sustainability of the production, etc., e.g., Quilot-Turion et al., 2012). In extreme cases, targeting one trait can lead to the loss of others (Ellers et al., 2012; Ostrowski et al., 2007), and reduced genetic diversity in crop plants can increase the likelihood of such losses. Examples of traits that are threatened by modern crop breeding include: (1) the ability to interact with mycorrhizae (Zhu et al., 2001), (2) symbiotic nitrogen fixation (Kiers et al., 2007), (3) inhibition of nitrification (Subbarao et al., 2006), and (4) the ability to benefit from earthworm activities (Noguera et al., 2011). These four examples suggest that trade-off choices have consequences for interspecific interactions, and, by extension, ecological networks. Artificial selection of higher growth rates may therefore have a general effect of decreasing benefits from interactionrelated traits. In natural ecosystems, plants rely heavily on such traits, which evolved in complex four-dimensional settings involving soil organic matter and interactions with micro- and macro-organisms (Puga-Freitas et al., 2012; Shahzad et al., 2012). Selection of such traits can play a critical role for a more sustainable management of agriculture, as once interaction traits are present, ecological processes can partly replace artificial inputs, via ecological intensification (Dore´ et al., 2011; Male´zieux, 2011).

Eco-Evolutionary Dynamics in Agroecosystems

351

Many trade-offs have already been documented in crop plants: between seed size and seed number (Sadras, 2007), between nutrient use efficiency and nutrient stress tolerance (Maia et al., 2011), between water and phosphorus acquisition (Ho et al., 2005) or between performance at low and high salinity levels (Richards, 1992). The growth/ defence trade-off seems to be widespread in wild plants (Lind et al., 2013) and has been investigated in a number of crop plants (e.g., tomato: Le Bot et al., 2009; sunflower: Mayrose et al., 2011), although no general overview exists for the latter. There is an emerging consensus, however, that fertilization affects herbivores (Butler et al., 2012) and pathogens (Huber and Watson, 1974), as well as plant growth, in agricultural systems. This trade-off is thus critical as it may provide an important key to conceive alternatives to pesticide use. The potential trade-off between nutrient uptake, at high and low concentrations of fertilizer, is controversial both for wild plants and for crops (Craine, 2009), and results to date have produced little or no supporting evidence (Hasegawa, 2003; Reich et al., 2003), suggesting that the same crop varieties may be cultivated regardless of the fertilization practice. This assessment of costs, and of trade-offs in general, is central to evolutionary ecology and for agronomy. In ecology, the results of community evolution models depend mostly on the shape of trade-offs, but such information is often unavailable (De Mazancourt and Dieckmann, 2004; Loeuille and Loreau, 2004). Artificial selection in agriculture could provide very important data to address this gap, creating a synergy between the two disciplines.

2.3. Adapting to local practices and conditions: The importance of diversity Diverse conditions, in space or time, typically favour a diversification of strategies through natural selection (Thrall and Burdon, 2003; Thompson, 1999), unless strong homogenizing are operating via high gene flow impeding local specialization (Harrison and Hastings, 1996; Day, 2001; Loeuille and Leibold, 2008b; Urban et al., 2008), or if generalists are equally fit as specialists, regardless of environmental conditions. This last condition implicitly assumes an absence of trade-off. These conditions are very stringent, so that diversification due to environmental variation is expected to happen in most natural conditions. This contrasts markedly with the situation in intensive agricultural systems, with a few species and a low genetic diversity. From an evolutionary point of view, this homogenization of cultivated plants can be interpreted as

352

Nicolas Loeuille et al.

either we have indeed managed to select highly generalist no-cost varieties, or the heterogeneity of conditions in which crops grow is insufficient to support many specialized varieties. The second explanation seems a better fit to our understanding of agricultural systems, for several reasons. First, a temporal correlation exists between diversity decline of cultivated plants and homogenization of their growing conditions, which is obtained by tillage, herbicides and pesticides that remove biotic interactions, while irrigation and mineral fertilization smooth out variations in resource conditions. A more sustainable agriculture would require management that changes modern agricultural practices and crop breeding approaches simultaneously (Tilman, 1999), but because both belong to the same technological regime, this creates inertia that impedes the development of alternative strategies (Vanloqueren and Baret, 2009). A direct corollary is that more sustainable crop varieties should fit better to their local environment and practices. Interactions between genotypes and environmental conditions determine crop performance (Van Bueren et al., 2008), and differences among varieties have been found for systems where there is variation in intercropping practices (O’leary and Smith, 1999), mineral fertilization (Atlin and Frey, 1990; but see Hasegawa, 2003), or soil salinity (Kelman and Qualset, 1991). Yields of 35 genotypes of soft white winter wheat were compared between organic and conventional practices, and a significant interaction between the genotype and cropping system was evident in four out of five locations (Murphy et al., 2007). This suggests that varieties to be used in organic farming should be selected for under the conditions of organic farming (Przystalski et al., 2008), whereas at present it currently largely relies on ‘conventional’ varieties. Breeding particular varieties for this type of agriculture could increase yields and improve organic agriculture sustainability (Wolfe et al., 2008). Since a better adaptation to local environments and practices should enhance sustainability and yields, such a strategy should inevitably be favoured, although the costs entailed by such local adaptation also need to be accounted for, such as those related to economies of scale, labour, agronomic management, and crop handling. The loss of genetic diversity of cultivated organisms is detrimental to agriculture sustainability because it limits the scope of potential future evolution (Dieckmann and Law, 1996; Lande, 1979). This, in addition to the fact that farmers in intensive systems no longer produce their own seeds, will compromise future local adaptation in crops. In traditional cropping systems, farmers adjust the landraces to suit their own farm. It may be argued that

Eco-Evolutionary Dynamics in Agroecosystems

353

modern tools of artificial selection (e.g., genetic engineering) can offset the lack of on-farm local evolution, but local adaptation involves many, often unknown, niche dimensions, and understanding its significance requires a predictive framework we still lack. In traditional agriculture, on the other hand, on-farm selection and seed exchange networks allow the necessary evolution of crops in the face of global change (Bellon et al., 2011) and the evolution of resistance to pathogens (Paillard et al., 2000). Mechanisms identified from ecological biodiversity-ecosystem functioning studies also suggest how diversity loss can be detrimental to agricultural sustainability. In natural ecosystems, higher plant species diversity produces higher biomass, primary production, as well as more resilient systems (Loreau et al., 2001). Different species react to environmental variations, often in uncorrelated ways, which decreases the overall variance of the community (Yachi and Loreau, 1999). This ‘portfolio’ mechanism, postulated for species diversity, also applies to genetic diversity (Vellend and Geber, 2005). Genetic diversity within species also influences ecosystem functioning (Crutsinger et al., 2006; Kotowska et al., 2010), and in agroecosystems, for instance, a mixture of genotypes with different resistance levels may impede a pathogen outburst by reducing the overall risk of contamination (Zhu et al., 2000; Bohan et al., 2013, Chapter 1 of this volume).

2.4. Beyond individuals: The influence of selection processes on the community and ecosystem context Choices of domesticated species and artificial selection of their traits will ultimately affect the overall structure of ecological networks. Indeed, the selection imposed on cultivated organisms, often the dominant species of their ecosystem, affects interspecific interaction in multiple ways. We previously described trade-offs based on allocation costs, but while these are undoubtedly important drivers of the evolution of some traits (Herms and Mattson, 1992), the evolution of other traits incurs a different type of cost. They may be constrained by the balance of various ecological interactions (thereafter, ecological costs), rather than by energy allocation within individuals. Strauss et al. (2002) propose that anti-herbivore defences can have different allocation and/or ecological costs. Importantly, ecological costs directly link trait selection to community aspects. For instance, Mu¨llerScha¨rer et al. (2004) suggest that although some phenolic compounds are effective against generalist herbivores, they may attract specialists. In principle, a plant species can then be invasive simply because their specialists are

354

Nicolas Loeuille et al.

absent, reducing the costs of these defences to zero. Evolution of defences that incur ecological costs affect herbivore insect communities in complex ways (Courtois, 2010), and they can also affect higher trophic levels or mutualistic species. In tobacco plants, nicotine compounds, usually considered to be primarily selected as herbivore defences (Krieger et al., 1971), are negatively linked to the pollinator reliance of the plant (Adler et al., 2012). Defences also affect nectar quality, so that herbivore repulsion is traded-off against pollinator attraction (Adler and Bronstein, 2004). Some volatile compounds serve as a defence but they can also act as cues for herbivore behaviours, leading to attraction to damaged plants for example, and signals to insects at higher trophic levels and pollinators (Poelman et al., 2008; Van Zandt and Agrawal, 2004; Xiao et al., 2012). Artificial selection can directly modify defensive traits (Carrie`re et al., 2010) and domestication often involves modification of palatability or digestibility of the cultivated organism (Meyer et al., 2012). This has indirectly selected against secondary metabolic compounds that are used for defence. How a modification of these defences affects the surrounding ecological network will, in turn, depend on the ecological costs associated with these traits. The importance of ecological trade-offs goes beyond insect communities, as other taxa and soil maintenance and recycling processes are also involved. Tannin production, again a classical defensive compound, slows down litter recycling by affecting the soil microbial loop (Grime et al., 1996; Whitham et al., 2003). Community evolution models also suggest that defences affect recycling processes via the herbivore loop (De Mazancourt et al., 2001). Defences could decrease root association with mycorrhizae, impairing the plant’s ability to take up mineral nutrients (de Roman et al., 2011). Plant defensive traits may therefore be linked to the long-term maintenance and recycling efficiency of soils, a feature that underpins the sustainability of agricultural yields (Male´zieux, 2011). More recently, an experimental study has even shown that defences may be traded-off against competitive ability (Agrawal et al., 2012), thereby coupling two traditionally separate issues in agriculture: competition from weeds and control of pests. Community evolution models illustrate ways in which evolution under ecological costs can affect network properties. For instance, Loeuille and Leibold (2008a) introduced a model in which two defensive strategies are incorporated, one incurring an allocation cost and the other having an ecological cost. Because of ecological costs, under high nutrient conditions, evolution of defences within the food web decreases species diversity and the network also becomes more compartmentalized, with well-separated

Eco-Evolutionary Dynamics in Agroecosystems

355

food chains. Evolution constrained by ecological costs is also usually less positive for community resilience than when it is constrained by allocation costs (Loeuille, 2010a). While such models suggest that ecological costs can have considerable implications for community structure and stability, empirical data to determine their consequences in an agricultural context are scarce. Recent observations on rice, however, indicate that such implications can be far-reaching (Xiao et al., 2012), as artificial selection of anti-herbivore defensive strategies may indeed modify the structure and interactions of the complete insect community, including parasitoids and predators. The direct implementation of defensive traits in the absence of an assessment of ecological costs for the maintenance of associated ecological networks should therefore be treated with caution. Evolution also affects ecosystem functioning and services (Fussman et al., 2007), and, as an example, selection for the high yield and fast-growing strategies favoured in agriculture can lead to a ‘tragedy of the commons’, in which evolution of acquisition of shared resources leads to the overexploitation and eventual extinction of these resources (Hardin, 1968) with negative consequences for the sustainability of the system. Emergent properties of the system such as biomass or mineral nutrient availability are affected by the evolution of such traits in complex ways (Boudsocq et al., 2011; Loeuille et al., 2002). For example, in the model by Boudsocq et al. (2011) (see also Box 6.1), plant evolution of nutrient uptake may lead to three contrasting evolutionary outcomes depending on costs: (1) an evolutionary equilibrium is reached, (2) evolution negatively affects the population viability, and (3) accumulation of mineral nutrient, eventually followed by a switch to another limiting nutrient or factor. From an agricultural perspective, only option (1) is sustainable. Outcome (2) shows how evolution of plant strategy can affect biodiversity maintenance, while outcome (3) highlight how it can have important consequences for ecosystem functioning (here, changes in the limiting nutrient). Let us focus on the first case, evolutionary equilibrium, whereby natural selection leads to intermediate rates of nutrient uptake at which the availability of mineral nutrient is minimal and the selected strategy does not maximize primary productivity or plant biomass. Modifying the model by Boudsocq et al. (2011) to account more closely for artificial selection (Box 6.1) and using primary productivity (yield, a common target for crop breeding) as a criterion for selection then leads to higher mineral nutrient availability accompanied by a faster rate of loss from the system, which creates a sustainability cost.

356

Nicolas Loeuille et al.

BOX 6.1 Natural selection, artificial selection, and the functioning of ecosystems The model (Boudsocq et al., 2011) is based on a simple representation of nutrient cycling between three compartments (primary producers, dead organic matter, and limiting mineral nutrient) in an open ecosystem (see the detail of model in Appendix A). Nutrient availability is strictly determined by the quantity of mineral nutrient. Nutrient losses occur from the mineral nutrient pool, due to diffusion. To predict the evolution of investment into nutrient uptake, we use the adaptive dynamics framework (Dieckmann and Law, 1996; Geritz et al., 1999). We consider a trade-off between nutrient uptake and plant nutrient turnover due to root and leaf mortality. The existence of this trade-off is supported by many mechanisms and observations. For example, a higher investment into the root system to increase nutrient uptake may (1) decrease the allocation of resources to defences against herbivores (Herms and Mattson, 1992), (2) increase the mineral nutrient concentration in the plant biomass which stimulates litter decomposition (Endara and Coley, 2011), (3) require the production of many thin roots with a high turnover (Eissenstat et al., 2000). Three types of evolutionary dynamics are possible depending on the tradeoff shape: (1) evolution of plant traits lead to an accumulation of nutrient so that another nutrient type eventually becomes limiting, (2) evolution leads to a runaway towards higher uptake of nutrient, which asymptotically leads to plant extinction, (3) an evolutionary equilibrium is reached with an intermediate investment in nutrient uptake. Analytical computations show that the availability of mineral nutrient always decreases along evolutionary dynamics. Neither the biomass of primary producers nor primary productivity are maximized at the evolutionary equilibrium (Boudsocq et al., 2011). Focusing on cases that lead to an evolutionary equilibrium, we illustrate these results in Fig. 6.1, that shows how two compartments (mineral nutrient pool, plant biomass) and primary productivity change depending on the value of the evolving trait s, the investment into nutrient uptake. Natural selection yields an s value (s
, vertical solid line) that minimizes mineral nutrient availability (N
, horizontal solid line), thus minimizing mineral nutrient losses happening through diffusion out of the system. Can this model help predicting the result of the artificial selection in crop plants? A possible scenario is that crop breeding is managed to maximize primary productivity (f0) and thus drives crop plants towards the s0 value (doted vertical lines). Compared with the previous ‘natural selection’ scenario, such a selection increases mineral nutrient availability (N0 , horizontal dotted line) and associated nutrient losses (Fig. 6.1A). Therefore, our results suggest that artificial selection per se, even in the absence of fertilization, may modify nutrient diffusion in ways that are negative for agricultural sustainability. The three other panels (B, C, D) of Fig. 6.1 display the same type of results after an increase in fertilization (panel B), an increase in plant biomass export to mimic harvesting (C), or an increase in both (D). These changes increase the effect of artificial selection, further enhancing nutrient losses. This result is not trivial.

357

Eco-Evolutionary Dynamics in Agroecosystems

BOX 6.1 Natural selection, artificial selection, and the functioning of ecosystems—cont'd A

B ¢

0

¢

0

F

0

0

0

F0

¢

C

¢

D 0

0

¢

¢

F

¢

0

F

¢

0

Figure 6.1 Each panel describes variations in the nutrient stock of primary producer (V 0), the stock of mineral nutrient (N0) and primary productivity (F0) at their ecological equilibriums as a function of the investment into nutrient uptake (s). In the displayed cases an evolutionary equilibrium is reached (vertical solid line, s
) and natural selection has been show to minimize N (solid horizontal cases, N
). Crop breeding can be supposed to maximize primary productivity (vertical dotted line (s0 )). This value of the investment into nutrient uptake leads to an availability of mineral nutrient N0 (doted horizontal line). In all cases crop breeding leads to a higher availability of mineral nutrient (N0 > N
). Panel A is based on the same parameters as in the original article (fertilization, 6 kg nutrient ha1 year1; biomass exportation, 0.1 year1)(see original Fig. 6.2 in Boudsocq et al., 2011). Panel B uses a higher fertilization rate (30 kg nutrient ha1 year1). Panel C uses a higher exportation rate of plant biomass (0.5 year1). Panel D uses both a higher fertilization and a higher exportation rate. Fertilization and crop harvesting can intuitively be thought as factors increasing nutrient losses at the ecological scale, but crop breeding could have mitigated such effects. We find the contrary: crop breeding further increases nutrient losses. Our results illustrate that the intensification of agriculture (increase in harvest index and mineral fertilization) interacts with crop breeding to determine the sustainability of agriculture (here characterized by nutrient losses). This result is useful since the mitigation of nutrient losses is generally viewed independently either from the point of view of agricultural practices (e.g., Gardner and Drinkwater, 2009) or of crop breeding (e.g., Good et al., 2004).

358

Nicolas Loeuille et al.

Such results raise an intriguing possibility: evolutionary dynamics can lead to negative relationships between ecosystem services (here primary production versus nutrient conservation). While such negative relationships have already been noted (Bennet et al., 2009), their fundamental causes as well as their consequences for ecosystem management, ecological engineering and agriculture are mostly unknown. We suggest that they may naturally emerge from artificial selection. In agricultural networks, species co-evolve with the cultivated plant. We now consider a system in which cultivated species would be selected through artificial selection and a wild herbivore species would follow natural selection based on individual fitness (Box 6.2). The initial community

BOX 6.2 Mixing group and individual selection to study the co-evolution of agroecosystem networks Most often, community evolution models that tackle co-evolution use individual fitness to determine evolutionary dynamics. In the case of agriculture, to study the co-evolution of a cultivated plant species and of the species from its surrounding network, one may want to base the evolutionary dynamics of the plant species on an artificial group selection criterion, and the evolutionary dynamics of other species based on individual fitness. We know of only one mixed model of this kind (Fletcher and Doebeli, 2013). We illustrate some possible implications of this idea using the model described fully in Loeuille et al. (2002), which studies the co-evolution between defence investment by the plant, and herbivore attack rate. The plant–herbivore interaction is modelled using a Lotka–Volterra function, whose rate depends on the two traits. Plant defences incur a cost in terms of growth while herbivore investment in consumption incurs a mortality cost. In its initial version, based on natural selection, the plant–herbivore co-evolution model leads to a set of strategies (plant defences, herbivore attack rate) that allow the coexistence of plants and herbivores, although some oscillations may exist around the equilibrium state (Fig. 6.2A). If plant selection now depends on total biomass, for instance because the farmer selects phenotypes producing higher yields, higher defences are always favoured, and co-evolutionary dynamics are profoundly affected. Eventually, such a co-evolution causes the extinction of the herbivore (evolutionary murder, Fig. 6.2B, see also Appendix B). This incurs an economic benefit if the herbivore is a crop pest. On the other hand, the argument is fairly general, so that the co-evolutionary dynamics associated with artificial selection can also produce unwanted extinctions. Of course, this model is largely simplified compared to real agricultural systems, in which herbivores have not disappeared (though they are usually less abundant). In any case, it illustrates a mechanism through which co-evolution community structure may be profoundly affected by artificial selection.

Eco-Evolutionary Dynamics in Agroecosystems

359

BOX 6.2 Mixing group and individual selection to study the co-evolution of agroecosystem networks—cont'd

Figure 6.2 Plant herbivore co-evolution depending on the selection regime. In panel A, plant defences co-evolve with herbivore attack rate depending on individual fitness. The evolutionary outcome (black filled circle) is at the intersection of the herbivore isocline (light grey) and of the plant isocline (black). Herbivore evolutionary dynamics always converge towards the isocline (plain line). For the plant isocline, evolutionary dynamics can be convergent (plain line) or divergent (dotted line). When the isoclines intersect in the plain part, evolutionary dynamics will lead to the evolutionary singularity. In the dotted part, dynamics may cycle around the singularity. In all instances, the full system plant–herbivore survives (i.e., co-evolution stays within existence boundaries, dashed curve). For more details see Loeuille et al. (2002). Panel B: Same model, but plant selection is based on biomass production instead of individual fitness. Under such conditions, more defended morphs are always favoured (Appendix B). This creates an ever increasing trait value for the plant that leads to the extinction of the herbivore (dashed curve), that is, an evolutionary murder (Dercole et al., 2006).

co-evolution model, where both species evolved out of natural selection, predicts the maintenance of the complete community. Such maintenance is unlikely when plants are subjected to group selection, where higher defences are favoured, because they allow a higher plant biomass by reducing herbivory. Eventually, such high defences cause the extinction of the herbivore. Therefore, artificial selection regime can have important implications not only for evolutionary dynamics, but also for side-effects on the community structure and conservation issues of non-targeted species. Such results of mixing artificial and natural selection pose interesting new questions. As evolution affects the system’s functioning, a possible target for crop breeding would be not only the plant species or traits, but also some part of its connected ecological network. A clear example is the experiment described in Swenson et al. (2000), who grew Arabidopsis in pots whose soil was chosen from an ecosystem selection perspective. At each generation,

360

Nicolas Loeuille et al.

pots were inoculated with soil samples selected from the biomass of an Arabidopsis individual of the previous generation. Group selection then involved all organisms present in the inoculate soil based on plant yield. Individual plant biomass increased rapidly through this selection process. From a theoretical point of view, such outcomes raise intriguing questions, such as what sets of species of the network should we incorporate in this ecosystem selection processes? How many species or functional groups should be involved? What constrains the efficiency of the selection process? All of these aspects remain largely unknown.

3. DISTURBANCES DUE TO AGRICULTURE: IMPLICATIONS FOR ECO-EVOLUTIONARY DYNAMICS WITHIN SURROUNDING ECOSYSTEMS Lessons from eco-evolutionary models can applied when considering the effects of agricultural disturbances (Gould, 1991; Thrall et al., 2011; Verbruggen and Kiers, 2010), which typically correspond to strong selective pressures on wild organisms, so fast evolution should be expected when suitable variability exists for associated heritable traits. It is critical to account for these eco-evolutionary dynamics, as they affect the ultimate survival of these organisms in the landscape (Ferrie`re et al., 2004; Loeuille and Leibold, 2008b) and produce indirect effects that propagate through ecological networks. Alternatively, adaptation can also take place via plasticity. Phenotypic plasticity may limit the evolution of genetic responses to disturbances, by modulating individual fitness (Ghalambor et al., 2007; Hendry et al., 2011). Also, the degree of plasticity is itself under selection and ultimately depends on the disturbance regime (frequency, amplitude and predictability). We first list a few of the disturbances associated with agriculture. This is not meant to be an exhaustive or exclusive list. Rather, we focus on disturbances whose implications may be assessed using community evolution models. As modern approaches of crop selection often rely on minimizing environmental variation through a wide use of nutrient additions and a strong control of enemies, these efforts to simplify the system correspond to strong selective pressures on the ecological networks in which the cultivated organism is embedded. There are three main disturbances, which we will now focus on in turn: nutrient addition, pesticide use, and habitat alteration.

361

Eco-Evolutionary Dynamics in Agroecosystems

Take home messages and key references for section 3 Agriculture and crop Ecology and evolutionary ecology Section breeding

References

Section Agriculture exerts strong 3.1 selective pressures on wild animals and plants through direct inputs and habitat modification

Fast evolution is expected for variable traits, given the strong disturbances Co-evolution creates indirect effects that propagate throughout the network

Thrall et al., 2011 Loeuille and Leibold, 2008a

Section Fertilization creates global 3.2 nutrient enrichment Evolutionary impacts of enrichment on communities modify important ecosystem services

Nutrient enrichment modifies top-down controls within food chains by altering the selection regime of defences

Oksanen et al., 1981 Herms and Mattson, 1992 Loeuille and Loreau, 2004 Box 6.3

Section Pesticides exert strong selective pressures that 3.3 trigger the evolution of resistances. This incurs large economic costs Side-effects exist on nontargeted species

Extra-mortality evolutionary models can be used to discuss such disturbances Evolution of resistance depends on the community context

Mallet, 1989 Abrams and Matsuda, 2005 Box 6.4

Section The homogenization of 3.4 biotic and abiotic conditions by agriculture should mostly select for specialized interactions Evolution of specialization affects pest management and ecosystem services

Evolution of specialization depends on species abundance distributions in communities Evolution of specialization affects the propagation of disturbances in the system

Poisot et al., 2011 Dore´ et al., 2011 Symondson et al., 2002 Box 6.5

3.1. Nutrient enrichment and its ecological and evolutionary consequences in agricultural landscapes Most natural ecosystems being limited by either N, P or K (Boudsocq et al., 2012; Vitousek and Howarth, 1991), and agroecosystems are no exception. Fertilization mainly consists of these nutrients. They are transported in surrounding ecosystems through abiotic factors (diffusion, transport by water) or through biotic factors (e.g., metaecosystem effects: Loreau et al., 2003). Nutrient additions profoundly change the diversity and composition of

362

Nicolas Loeuille et al.

natural communities, in terrestrial and aquatic systems (Tilman, 1999; Xia and Wan, 2008) and are one of the major players in current global changes (Langley and Megonigal, 2010; Reich, 2009). From an evolutionary point of view, nutrient additions have direct selective effects, by alleviating costs of life history or physiological traits that are energetically expensive. They also generate indirect selective effects on traits related to interspecific interactions, as they modify the composition and structure of ecological networks. We will analyse first the direct effect (allocation costs), and then turn our attention to assessing indirect effects (community composition and diffuse co-evolution). To illustrate how the evolution of traits with allocation costs alters the dynamics of communities, the best starting point is to study the qualitative effects it has on food chains. Comparing the impact of nutrient enrichment on food chain dynamics without and with evolution will enable us to assess its role in community structure and composition. If a food chain is supplied with increased resources, two possibilities exist. If biomass at each trophic level is bottom-up controlled—meaning that competition for resources is the major driver of ecological dynamics—then biomasses at all trophic levels will increase, because enrichment relaxes competition constraints. This type of pattern holds well in some ecosystems (Chase, 2003; White, 2005). In other instances, when the biomass at a given level is determined by predation exerted on it (top-down control), the biomass at the top level of the food chain will increase during enrichment, as well as all odd levels from the top-down, while the biomasses at other trophic levels remain constant or decrease (Hairston et al., 1960; Oksanen et al., 1981). Again, empirical patterns can be found to support such top-down control predictions (Persson et al., 1992; Ripple and Betscha, 2012). Figure 6.3 in Box 6.3 illustrates such an ecological impact of nutrient enrichment. Of course, the picture is inevitably more complicated when omnivory and generalist feeding, which are common features of most ecological networks, blur trophic levels into continua, but the same general principles apply. Biological control of crop pests by natural enemies is one of the many ecosystem services on which the development of a sustainable agriculture relies (Costanza et al., 1997; Crowder et al., 2010; Dore´ et al., 2011; Male´zieux, 2011). Top-down control models therefore provide conditions under which biological controls should operate. For instance, considering herbivore as pests, nutrient enrichment would increase pest biomass when the food chain has two levels (plant–herbivore) or an even number of levels

Biomass

BOX 6.3 Herbivore pest evolution under nutrient enrichment: Implications for biological control (e.g., predator) (Fig. 6.3). Otherwise, Theplant–herbivore–predator–super dynamics of food chains are fairly well understood. Particularly, when top-we expect control tonutrient act, andenrichment herbivoresincreases to be maintained at of constant downbiological controls dominate, the biomass the levels their enemies enrichment process. top by compartment and during of everythe alternate level below. The biomass of other levels remains unchanged or decreases (Hairston et al.,the 1960;conditions Oksanen et under al., Unfortunately, evolutionary dynamics erode 1981, see also Fig. 6.3, top). The strength of top-down controls strictly determine which this top-down control is expected to happen. Many phenotypic traits constraining top-down controls have allocation costs, so that their selection regime may change when energetic constraints are relaxed. Allocation costs are for instance well described for some plant defences (Herms and Mattson, 1992; Strauss et al., 2002). If nutrient enrichment selects for more defended plants, energy is less easily transmitted up the food chain, so that top-down controls are decreased (Loeuille and Loreau, 2004). This is also true for other traits, such as vigilance, where a possible cost is the foregone time and energy from activities not related to resource acquisition (Illius and Fitzgibbon, 1994). Again, if nutrient enrichment makes the resource more abundant or more nutritious, higher vigilance can be favoured (Armstrong, 1979; Leibold, 1996), which decreases energy transmission through the food chain. Similar arguments could be made for the evolution of time partitioning between productive and unproductive habitat (Schmitz et al., 2004; Grabowski and Kimbro, 2005) or for the evolution of body size, a trait 120 some degree of predator avoidance (Emmerson and whose increase allows P 100 incurring metabolic costs (Kleiber, 1961). Raffaelli, 2004), while 80 Using a simple, Lotka–Volterra-based food chain model, Loeuille and H Loreau (2004) show60how the evolution of plants, evol the evolution of herbivores, or the co-evolution of the two, affect top-down controls. Their 40 results show, for instance, that evolution of increased plant defences under 20 H nutrient enrichment weakens top-down controls. A more relevant question 0 200 insect 100 150 250 pest evolution 300 from an agricultural perspective is whether is affected input by nutrient enrichment and, if so, Nutrient what are the possible consequences for Figure 6.3control. Top, foodInchains: corresponds compartments biomass biological Box black 6.3, we extend tothe Loeuille &whose Loreau model increases, white to compartments not affected by the enrichment. For the ‘herbivore to evolution answer this question, by showing that in a plant–herbivore–predator food scenario’, enrichment selects larger values of the herbivore defence trait, chain, evolution of herbivore defences decreases top-down controls exerted depicted by an arrow on the right side of the herbivore level. Note that in the food onchain pests. implication agricultural systems Herbivores is that the on One the left,immediate predators exert a biologicalfor control on the herbivores. do not increase during enrichment. However, when possible, this is no advantage gained on one side (increased yieldevolution throughis nutrient addition) longer true. Enrichment allows the evolution of the herbivore to decrease the has a direct cost to an a priori unrelated ecosystem service (biological control). strength of biological controls. Herbivore biomass then benefits from the enrichIn addition these increases. direct effects of nutrient enrichment the selection ment and theirto biomass Bottom, a simulation showing theon herbivore bioof mass traits,(inindirect evolutionary effects(in areblack) also expected. Nutrient enrichment grey) and predator biomass for the ‘ecology alone’ scenario (dashed for theof‘herbivore evolution’ scenario (plain lines).and Note that, affects thelines) totaland number species of the community (Chase Leibold, starting from an initial nutrient input level of 100, the net effect of the evolution 2002; Kassen et al., 2000) as well as the number of trophic levels of herbivores on the weakening of the biological control can be fully perceived (Oksanen et al., 1981) so that interactions are globally changed within the as the distance between the two grey lines. For the analytical demonstration and parameter setting of Fig. 6.3, see Appendix C. Continued

364

Nicolas Loeuille et al.

BOX 6.3 Herbivore pest evolution under nutrient enrichment: Implications for biological control—cont'd the possibility of biological controls, an important ecosystem service (Costanza et al., 1997). A crucial question is whether evolutionary dynamics strengthen or undermine top-down controls when nutrient enrichment takes place. Using an adaptive dynamics approach applied to a Lotka–Volterra food chain model, Loeuille and Loreau (2004) studied how the biomass variation pattern described above changed, depending on different plant–herbivore co-evolution scenarios. One of their conclusions was that plant evolution significantly weakens the top-down control, because nutrient enrichment selects for higher defence levels that decrease herbivore impacts. This result is not especially relevant within the agricultural context where plant trait is often more constrained by human action than by direct selection from herbivores, but a more interesting question arises about how top-down control is changed considering the evolution of herbivore pests facing their natural predators. The action of such predators, that is, the biological control, may change due to herbivore evolution, as modified by nutrient additions. We study this issue by extending the model of Loeuille and Loreau (2004), adding a predator level and studying the evolution of herbivore defences. In Appendix C, we show that the weakening of top-down controls observed in the original model also applies in such longer food chains. That is, herbivores evolve higher anti-predator defences during the enrichment, which decreases the biological control (see Fig. 6.3). The result is fairly general. It does not depend on the trade-off functions that are used, just requires that the defensive trait of the herbivore has an allocation cost. Identity of the defensive trait can be of many types, from chemical anti-predator defences (development of toxicity) to changes in behaviour (increased vigilance, use of refugia, etc.). As in Loeuille and Loreau (2004), co-evolution scenarios—whether they are plant–herbivore, herbivore–predator co-evolution, or co-evolution of the three species—are likely to yield much more complicated results. Note however that studying herbivore evolution is a logical first step given the agricultural context. Herbivore pests are often insects or small invertebrates, whose evolution has been shown to be very rapid in response to agricultural change (Gould, 1991; Carrière et al., 2010): evolution of species at other levels of the food chain may very well be slower.

network. Because of such changes, the diffuse co-evolution of the entire ecological network is likely to be affected. It is, of course, more difficult to predict how such changes in the network will impact evolutionary dynamics, but some recent models have linked diffuse co-evolutionary dynamics to community structure or ecosystem functioning. Loeuille and Loreau (2005) explain that, in a food web based on predator–prey body size co-evolution that considers interference competition, nutrient enrichment

Eco-Evolutionary Dynamics in Agroecosystems

365

yields more trophic levels and diversity. Bra¨nnstro¨m et al. (2012), in a related model, show that it also affects the strength of disruptive selection on body size in ecological networks, thereby favouring ecological speciation. Results of increased diversity, numbers of trophic levels and of changes in connectance are also found in a co-evolutionary model that relies on many phenotypic traits evolving independently (Drossel et al., 2001). Note that the optimistic message that may be inferred from these studies—that nutrient enrichment increases community diversity—should be taken with caution, because species niches in these models rely on a single energy axis, so that enrichment simply inflates the available niche space for diversification. In nature, at high nutrient loads, negative side-effects such as toxicity or anoxia ultimately come into play to halt such increases in diversity (Harnik et al., 2012; Justic et al., 1995).

3.2. Chemical warfare in agricultural landscapes: The ecological and evolutionary consequences of pesticide use A second disturbance that creates important selective pressures is the use of insecticides, herbicides, or other chemical pesticides. While usually targeted at specific pests, pesticide toxicities are usually tested on just a few other species and they may have side-effects on a far larger number across different trophic levels (Crowder et al., 2010; McMahon et al., 2012; Robinson and Sutherland, 2002), some of which, such as pollinators, may provide important ecosystem services (Klein et al., 2007). Direct effects can be observed when sensitivity to the pesticide evolves due to selective pressures exerted by chemicals. Alternatively, because this disturbance propagates through the ecological network and affects its structure and diversity, it modifies the selective pressures indirectly throughout the community, altering the course of diffuse co-evolution. The use of pesticides is an important disturbance associated with agriculture on local to global scales (Matson, 1997) and by generating extra mortality on wild organisms, pesticides have detrimental consequences for ecological dynamics and overall ecosystem functioning (Matson, 1997). Impacts have been noted on many groups such as birds (Carson, 1962), pollinators (Potts et al., 2010,) and natural enemies of pests (Bianchi et al., 2006; Geiger et al., 2010). Decreases in such groups erode their associated ecosystem services (Bianchi et al., 2006; Potts et al., 2010), yet few studies have considered how associated evolutionary dynamics affect ecological networks.

366

Nicolas Loeuille et al.

The potential for an evolutionary response to pesticides varies among wild organisms associated with agricultural landscapes. Examples of rapid evolutionary responses abound in the agricultural literature (Palumbi, 2001; Thrall et al., 2011), where resistance to pesticides is a classical textbook example of how continuous selection pressures on a trait with generally simple genetic determinism results in rapid allele change. The international survey of resistance in weeds (http://www.weedscience.com) listed 217 species (129 dicots and 88 monocots) that had evolved resistance to 21 of the 25 known herbicide sites of action and to 148 different herbicides. A similar pattern is observed in insect pests, with an extremely rapid increase in the number of species showing resistance to at least one insecticide since the 1950s (Mallet, 1989). Approximately 8000 cases of resistance to 300 insecticide compounds are now reported in more than 500 insect species (Arthropod Pesticide Resistance Database; http://www.pesticideresistance.com; Whalon et al., 2008). Similarly, 300 cases of field resistance to 30 fungicides have been reported in 250 species of phytopathogenic fungi (Fungicide Resistance Action Committee database; http://www.frac.info; Bourguet et al., 2013). That evolutionary responses are less well-known in non-pest organisms could simply reflect the fact that they are largely ignored in such instances (Pelosi et al., 2013), because of the lesser economic concern they represent. It is, however, not expected that all species will evolve in response to pesticides: even when a variable and heritable response trait exists, some processes can limit its effective evolution. Plastic responses can play a role in the emergence of resistance to pesticides, by alleviating fitness costs (Hendry et al., 2011). For organisms with a plastic behaviour that can simply avoid crops with insecticides selective pressures for increased resistance will be weak. Such exploitation of alternative habitats can, however, incur indirect costs, for instance when such habitats have fewer resources. When plasticity actually impedes the evolution of pesticide resistance, it still raises important evolutionary questions: namely, when do we expect the evolution of such plasticity and what are the associated costs? A general idea is that plasticity is favoured when selective constraints are variable in space and time (Agrawal, 2001; Lind and Johansson, 2007). Because pesticide use may be highly variable in time, being most prevalent during infestation periods, and because semi-natural habitats are not directly treated, creating a spatial heterogeneity, plastic responses should commonly evolve. This view is supported by the existence of many plastic behaviours allowing some resistance to pesticides (Gould, 1991). Changes in the oviposition and host fidelity behaviour of corn rootworm, for instance, allowed it to counter crop rotations (Gray et al., 2009).

Eco-Evolutionary Dynamics in Agroecosystems

367

Demographic constraints, such as population density or generation time, also affect the evolutionary response to pesticides. This idea can be illustrated using evolutionary rescue models (Bell and Gonzalez, 2011; Gomulkiewicz and Holt, 1995). A population facing a disturbance can go extinct when the extra mortality decreases its per capita growth rate sufficiently. On the other hand, selection of less vulnerable phenotypes will occur, increasing average fitness. Evolutionary rescue is determined by the race between these two processes (Gomulkiewicz and Holt, 1995): extinction or rescue depends on the initial density of the population and on the generation time of the species. Small populations or long-lived organisms, having small variability in the response trait, are most at risk. Their densities following pesticide use may fall under a critical value, where demographic stochasticity increases the risk of extinction (Gomulkiewicz and Holt, 1995). However, species whose effective population size is large (therefore possibly harbouring high genetic variability), or generation times are short (so that new mutations can appear more readily), may escape extinction when adaptation happens fast enough to increase average fitness values before the population viability threshold is reached. Most pests probably fall in this category, as they have short generation times and high population densities. Pollinator populations, on the other hand, have lower densities and longer life cycles. Evolutionary rescue therefore provides a potential explanation for the often-observed selection of resistance to insecticides in pests (Bourguet et al., 2013; Mallet, 1989), while many pollinator populations collapse (Beismeijer et al. 2006). Another example is the evolution of pesticide resistance in fungi, which is determined by the pathogen’s evolutionary potential, as set by its life cycle and reproduction system (Barrett et al., 2008; McDonald and Linde, 2002). Most existing works related to pesticides tackle the conditions for the evolution of higher resistance in pest species (Bourguet et al., 2013; Gould, 1991; Mallet, 1989; Thrall et al., 2011), reflecting the huge economic costs incurred (Palumbi, 2001). Evolutionary models can then suggest managements that delay the emergence of these resistances, such as high-dose/refuge or pyramid strategies that use the ensemble activity of multiple compounds (Bourguet et al., 2013). A complete review of this aspect is beyond the scope of the present chapter and has been covered elsewhere (Bourguet et al., 2013; Carrie`re et al., 2010). These approaches, however, mostly focus on one species, the targeted pest, ignoring the broader ecological network (Tabashnik et al., 2004; Vacher et al., 2003). Many observations suggest that the community context does affect the evolution of resistance. For example, resistance to insecticides in mosquitoes negatively affects the resistance to parasites, so that evolutionary dynamics ultimately

368

Nicolas Loeuille et al.

depends on the overall parasite load within the population (Agnew et al., 2004). Resistance in the peach-potato aphid increases its sensitivity to predators due to a decreased response to alarm pheromones (Foster et al., 2003). Through such ecological costs, the predation context affects evolution of resistance. Evolution of resistances also affects pest densities, allowing them to attain large numbers, and these effects can propagate through ecological networks, creating top-down or bottom-up effects. Because pests interact both directly and indirectly with many other species, density-dependent effects will eventually impinge on the eco-evolutionary dynamics of other wild species of the community. We have little information on such eco-evolutionary dynamics within the pesticide context, but another corpus of models, developed in fisheries science, have studied the evolutionary dynamics of communities in which a subset of species faces extra mortality, via harvesting. Even though these models are not explicitly designed for agricultural systems, they are often stated in general equation systems that can be applied outside the fisheries context, although such an analysis is still restricted to the mortality effects of pesticides. Other sub-lethal effects (e.g., modification of fecundity or of consumption rates) require more explicit and detailed models. Abrams and Matsuda (2005) studied a two-species model that we have translated to a herbivore pest and its host plant, where the former suffers extra mortality (e.g., due to pesticides) and plant nutrient uptake rate evolves. Nutrient uptake rate is supposedly positively correlated with plant vulnerability, because of an allocation trade-off between defences and growth. The model shows that herbivore density can increase when its own mortality increases. This counter-intuitive result occurs because herbivore density initially decreases, causing selection of higher vulnerabilities of the plant. Because the plant is more vulnerable, the density of herbivore subsequently increases. This latter positive evolutionary effect eventually exceeds the direct negative effects of extra mortality (Fig. 2 in Abrams and Matsuda, 2005). This result has interesting agricultural implications because it suggests an evolutionary mechanism through which pesticides eventually increase pest abundances. Abrams and Matsuda suggest that plastic responses could produce similar qualitative trends. The model would also be suitable to describe the evolution of wild plants consumed by the pest. When increasing pesticide application, pest density increases until a threshold is reached, but beyond this value the pest abruptly goes extinct. In a four-compartment model that can be translated as two plants sharing a limiting nutrient and a pest herbivore, Abrams (2009) again studied how the response of the pest to extra mortality is affected by adaptation in the plant. Plant

Eco-Evolutionary Dynamics in Agroecosystems

369

evolution can again create an increase in pest density when pesticide is used. Compared with the simple food chain model of Abrams and Matsuda (2005), the evolution of the second plant species extends the range of pesticide use the pest can suffer before going extinct. A possible agricultural implication is that evolution of a lesser vulnerability in wild plants can improve the existence conditions of a pest confronted with pesticides. While an exact test of such patterns has not been done, existing data comparing the evolution of weeds within fields and out of fields can provide an interesting point of comparison (Ellstrand et al., 2010). Evolutionary effects of extra mortality can also affect the stability of natural communities, which may increase or decrease, depending on how pesticide impacts relate to population densities (Witting, 2002). Another important result of fisheries research is that increased mortality can select for earlier maturation and smaller body size at maturation (Allendorf et al., 2008; Barot et al., 2004; Enberg et al., 2009; Law, 2000). Because of high mortality, those who mature faster are more likely to reproduce and are therefore favoured. Similarly, because the pesticide use exerts extra mortality, the same selective mechanism may apply, with fast-reproducing strategies being favoured, although we could not find any experiment or observation that has tested this idea. An obvious difficulty lies in the fact that confounding factors exist: having a fast life cycle in the first place makes it more likely for a species to be a pest, and it may also facilitate the evolution of resistance. This prediction of selection for earlier reproduction may be dependent on the community context. Ga˚rdmark et al. (2003) used a model in which a prey population consisting of three-age classes suffers extra mortality from harvesting and is consumed by a predator species. They studied the evolution of the age at reproduction (age 2 or 3; age 1 being juveniles), depending on the stage being predated. Substituting the prey species with a pest species and harvesting with pesticide extra mortality, in the absence of predation evolution of early maturation is favoured. However, reproduction is delayed to age 3 when extra mortality acts on age 1 and predation on age 2. Thus, the two mortality-related selective pressures can interact antagonistically (Ga˚rdmark et al., 2003). Importantly, it is unlikely that only pests are affected by pesticide use, and impacts on other members of the network are likely to modify associated ecosystem services as a result. To illustrate this, we modelled a plant– herbivore pest–predator food chain where both the pest and the predator are vulnerable to pesticides. For the sake of simplicity, we considered that only the herbivore evolves in response to pesticides and that evolution of resistance incurs a cost in growth rate (allocation cost). The model and principal results are detailed in Box 6.4.

BOX 6.4 Evolution of resistance to insecticides in a tri-trophic food chain We model a three-species community consisting of a plant attacked by a herbivore pest, consumed by a predator and in which insecticide use affects both the herbivore and the predator. Detailed equations and analysis of the model are given in Appendix D. We assume that the herbivore evolves in response to insecticide use. Its susceptibility incurs an allocation cost: when susceptibility decreases, herbivore reproduction decreases (Carrière et al., 1994). A possible mechanism is that conversion efficiency is reduced, due to an allocation in detoxification metabolism (Després et al., 2007). We study three different community scenarios: the predator is absent from the community, the predator has a low susceptibility to insecticides, and the predator has a high susceptibility to insecticides. We use the adaptive dynamics framework to study the impact of herbivore evolution on the three-species food chain (see Appendix D). Figure 6.4 shows how evolved resistance changes when pesticide use (parameter l) increases. Figure 6.5 displays the variations in densities for the three community scenarios and depending on whether herbivore evolution happens or not. As intuitively expected, resistance is selected for at higher inputs of insecticides (Fig. 6.4). Herbivore adaptation delays the extinction of both

Figure 6.4 Influence of insecticide intensity on the value of herbivore susceptibility trait at the evolutionary equilibrium, for the three different community scenarios: ‘No predator’, ‘Low susceptibility of predator’ and ‘High susceptibility of predator’. Parameters values: r ¼ 2; I ¼ 0.2; w ¼ 0.2; a ¼ 0.5; dh ¼ 0.5; g ¼ 0.5; dp ¼ 0.1; c(s) ¼ exp(0.1 s); j(s) ¼ exp(0.3 s); Low h ¼ 0.1; High h ¼ 1. For parameter definitions and details of the equations, see Appendix D.

BOX 6.4 Evolution of resistance to insecticides in a tri-trophic food chain—cont'd

Figure 6.5 Densities of the three species at the ecological equilibrium (left) or ecoevolutionary equilibrium (right) for different intensities of insecticide application and for three community scenarios. Continuous lines show plant density V 0, long-dashed lines herbivore density H0 and dotted lines predator density P0. In these graphs, all species coexist in areas labelled ‘1’, ‘2’ indicates plant–herbivore coexistence and ‘3’ parameter sets for which only the plant exists. Parameters values are the same as in Fig. 6.4. When there is no evolution the susceptibility trait for the herbivore is fixed at 0 (alternative values do not alter the qualitative results). the herbivore and the predator with increasing load of insecticides (Fig. 6.5). However, when the predator is not strongly affected by pesticides,less evolved resistance appears in the herbivore. This last result can be explained by density-dependent mechanisms: when the predator has a low susceptibility, the plant is at a higher density because of the predator exerts top-down control on the herbivore. It is then beneficial for the herbivore to be more susceptible (but with a high conversion efficiency) as the efficient consumption of abundant plants offsets mortality costs. When the predator is highly susceptible to pesticide, its density is decreased or it goes extinct, so that the density-dependent advantage disappears. Community context and evolutionary dynamics thereby interact to determine evolution of pests subjected to increased mortality.

372

Nicolas Loeuille et al.

Overall, resistance levels increased with insecticide use, but predator presence decreased the evolved level of resistance and pest density. Therefore, conservation of predators not only maintains an important ecosystem service through biological controls, but it may also limit the evolution of resistance in the pest. Note, however, that this double service is lost when the predator is highly sensitive to pesticides. Such effects are produced by the interaction between density-dependent ecological processes (here, predation) and trait dynamics due to species evolution. They affect both the magnitude of species responses and the qualitative direction of such responses, as reflected in densities and trait values. In more complex networks, evolutionary effects of pesticide can propagate through many trait-dependent and density-dependent pathways (Wootton, 1994; Yodzis, 2000). Predicting the associated diffuse evolution is difficult, especially because most models are based on trophic interactions, while pesticide use may also affect non-trophic compartments (e.g., pollinators). The interplay between different interaction types is a growing research area in ecology and evolution (Altermatt and Pearse, 2011; Fontaine et al., 2011). Antagonistic and mutualistic interactions influence each other, both through ecological, density-dependent effects and through selective pressures (Adler et al., 2012; Herrera et al., 2002). This is particularly relevant for agricultural landscapes, whose sustainability relies on ecosystem services that come from different interaction types. Biological control is linked to trophic interactions acting on pests, while pollination services rely on mutualistic interactions. Pesticides may affect all different interaction types, coupling the different ecosystems services through eco-evolutionary dynamics.

3.3. Effects of altering species composition and relative abundance of species in agricultural landscapes A third disturbance corresponds to changes in habitat, species abundance, distribution, and composition due to agricultural activities, which can have multiple effects. By developing open landscapes at the expense of natural habitats, agriculture modifies landscape composition; at more local scales, community diversity and composition are also affected. Agricultural communities usually contain just a few dominant species (the cultivated ones), represented by a few artificially maintained genotypes (Purugganan and Fuller, 2009), so the base of the food web changes differs markedly from the natural state that typically contains many plant species.

Eco-Evolutionary Dynamics in Agroecosystems

373

The impact of crop management practices on communities is wellknown: for instance, changes in the relative abundance of crop and weed plants can generate bottom-up effects changing arthropod community structure (Lenardis et al., 2011). On the other hand, insects can directly reduce weed biomass in favour of the crop biomass, and vice versa (Maron and Crone, 2006; Norris and Kogan, 2005). Such changes in species biomass at the different trophic levels alter selective pressures, which, in turn, affect the degree of specialization in species interactions. The topic of specialization has recently attracted considerable attention from community and ecosystem ecologists (Forister et al., 2012; Poisot et al., 2011) and at least three factors influence the relative fitness of generalists and specialists have been identified. First, evolution of specialization depends on trade-off shapes, with weaker trade-offs, that is, low cost of switching from one interaction to another, favouring generalists (Levins, 1962; Rueffler et al., 2006b; 2007). Second, specialization can arise from ecological opportunities created in complex networks (Forister et al., 2012), such as plant-mediated enemy-free space in tri-trophic systems favouring specialization in herbivore insects (Singer and Stireman, 2005). Third, temporal variation in fluctuating and highly disturbed environments favours generalists while environmental constancy favours specialists (Levins, 1968, Futuyma and Moreno, 1988). Existing models predict specialization under high resource predictability (little or no seasonality), abundance and diversity (Roughgarden, 1972; Van Valen, 1965). Behavioural selectivity and plasticity, however, modulate the effects of spatial and temporal heterogeneity and can promote the evolution of specialization (Poisot et al., 2011). In a mathematical model, Ravigne et al. (2009) suggested that the joint evolution of local adaptation and habitat choice leads to specialization regardless of the shape of trade-offs, provided that the cost of habitat selection is not too high. The general trend of agricultural intensification has been simplification in terms of reduced diversity of species (few crop species, mostly one crop plant per field), genes (few varieties, genetic homogeneity within each cultivar and mostly one cultivar by field), and management practices (few active ingredients or tillage systems). Consequently, the transition from traditional to intensive agriculture should favour specialized interactions, although this may be partially mitigated by crop rotation. While these ideas require experimental investigation, two lines of observational evidence provide empirical support. At the species level, outbreaks of pests feeding on the dominant crop happen more readily in these landscapes. The apparently strong

374

Nicolas Loeuille et al.

modularity (i.e., weakly interlinked subsets of species, strongly connected internally) observed in species networks within agricultural systems suggests a high degree of specialization (Macfadyen et al., 2011). Traditionally, modular structure has been thought a stabilizing factor, slowing down the spread of disturbance between modules (May, 1972; Krause et al., 2003, but see Allesina and Tang, 2012), but this has been challenged recently, as it appears that although modularity can enhance stability in trophic networks whereas, a highly connected and nested architecture, resulted from generalized interactions, can promote community stability in mutualistic networks (Bascompte et al., 2006, The´bault and Fontaine, 2010). Specialization has important implications for biological control in agricultural systems. Traditionally, it has been assumed (Howarth, 1991) that an effective biological control should be highly host-specific, based on the resource concentration hypothesis that consumers are more likely to prefer the most abundant resource (Root, 1973). Experimental and theoretical evidence, however, suggests that in the case of pest predators, generalists can generate better pest control services (Flaherty, 1969), especially in pest-rich communities where generalist predators maintain low prey populations via resource partitioning (Symondson et al., 2002). Less effort has been devoted to the study of biological weed control in crop fields. On the one hand, the introduction of insects has been proposed as a mechanism of weed control (McEvoy 2002), but it has also been hypothesized that the top-down effects of a generalist insect on a crop can be diluted via the presence of weed species (Root, 1973). Therefore, a generalist insect–weed– crop system may be easy to manipulate to favour a crop or other beneficial plant species. What remains unclear, however, is how eco-evolutionary dynamics can affect specialization of the herbivore and hence, biological weed control. A simple model of the interaction between a herbivore pest and two plant species, one crop and one weed, competing for a common resource (Box 6.5) can be proposed here, in line with the diamond shape commonly applied in general ecological scenarios (Armstrong, 1979; Leibold, 1996). This structure is also ideal for studying the interaction between weeds and insect pests. Indeed, experimental observations suggest that insect pest outbreaks may be negatively affected by the presence of weeds, because they may host predators or parasitoids (Altieri, 1981). All else being equal, results of the diamond model suggest that a crop species, despite being a poor resource competitor, can coexist with a competitive weed, due to its herbivore-resistance mechanism favoured by breeding efforts. Biological weed control in such a system can be readily achieved via fertilization (see Box 6.5). The generated

BOX 6.5 Degree of specialization of pests and its eco-evolutionary effects: Implications for weed and pest management A major challenge in ecology is to understand how species can coexist despite being limited by the same biotic or abiotic factors (Hutchinson, 1961; Tilman, 1982; Chase and Leibold, 2003). In theory, species coexistence requires that the abundance of each species is limited by a different (biotic or abiotic) factor. A generalist herbivore may represent this additional limiting factor and insure the coexistence of two plant species competing for a common limiting resource (e.g., a mineral nutrient). This interaction configuration matches fairly well the case of a generalist insect feeding on two plant species, a crop and a weed sharing a common limiting resource. Damage to the crop (weed) leads to decreased use of resources by the crop (weed). Resources not used by the crop (weed) theoretically become available to weeds (the crop), resulting in increased weed (crop) growth (Norris and Kogan, 2005). The question is how the eco-evolutionary dynamics of such a system can help to design effective weed and pest management. At low resource inputs (small I) the model predicts that competitively superior weeds dominate because densities of herbivore would be insufficient to reverse their competitive advantage. At high resource input (high I), however, the crop dominates because it can support a high herbivore density that in its turn suppresses weed to low biomass. In other words, the pest functions as an agent of biological control on the weed at high fertilization levels. Therefore, an agronomist can readily assure a high crop yield by merely increasing nutrient input and/or by inflicting a high mortality rate on the weed (e.g., through herbicide use). The efficiency of this biological control mechanism, however, can be compromised by the evolution of herbivore specialization. A strong selective pressure exists for a relative increase in herbivore’s specialization rate on the more abundant/profitable crop species (see Appendix E). A consequence of this evolution of specialization on the abundant crop is that the pest-insect no longer exerts biological control at high fertilization (Fig. 6.6B and C). Moreover, the evolution of specialization enhances the top-down control on crop plants, which results in the positive response of the pest herbivore biomass to fertilization (Fig. 6.6A).

Pest biomass

A

0.0

0.2

0.4

0.8

1.0

C

Crop biomass

Weed biomass

B

0.6

0.0

0.2

0.4

0.6

0.8

Nutrient input (I )

1.0

0.0

0.2

0.4

0.6

0.8

1.0

Nutrient input (I)

Figure 6.6 Illustration of changes in the biomasses of the pest, crop and weed species along a gradient of nutrient with (continuous lines) and without evolution (dotted lines). For the analytical demonstration and parameter setting, see Appendix E.

376

Nicolas Loeuille et al.

increase in crop biomass, however, will likely affect the degree of specialization of the herbivore. The herbivore species responds adaptively to the relative decrease of the weed population by increasing its specialization rate on the most abundant crop population. Thus, the evolution of specialization, via density-dependent mechanisms or plasticity, may ultimately compromise the efficiency of biological control practices.

4. ACCOUNTING FOR SPATIAL HETEROGENEITIES: DISPERSAL, FRAGMENTATION, AND EVOLUTION IN AGRICULTURAL LANDSCAPES 4.1. Characteristics of agricultural landscapes, past, present and future Large changes in agricultural landscapes have happened in the last 60 years: on a global scale, cultivated areas (1.5 109 ha in 2003) increased by 13% between 1961 and 2003 and irrigated land has doubled (Paillard et al., 2010). In some regions of Asia and South-America, the ‘Green Revolution’ favoured the transition from subsistence agriculture to industrial agriculture, and similar changes took place in North America and parts of Europe. Through increased inputs and breeding of modern varieties, the productivity of the major crops improved, reducing malnutrition in some Take home messages and key references for section 4 Agriculture and crop Ecology and evolutionary ecology Section breeding

References

Section Ecological intensification of 4.1 agriculture would lead to more fragmented and diverse agricultural areas. Impact on fragmentation of natural areas depends on the land-sparing/land-sharing choice

Habitat heterogeneity and Tscharntke steepness of environmental et al., 2005 gradients affect ecoevolutionary dynamics

Section The spatial distribution of 4.2 fields affects gene flows from cultivated species to wild populations and among wild populations. This subsequently affects pest and wild species evolution

The spatial organization of the patches of a metacommunity influences gene flows and the subsequent evolutionary dynamics

Haygood et al., 2003 Plantegenest et al., 2007

377

Eco-Evolutionary Dynamics in Agroecosystems

Take home messages and key references for section 4—cont'd Agriculture and crop Ecology and evolutionary Section breeding ecology

References

Section Conservation biological 4.3 control of pests aims at promoting movement of individuals from seminatural areas to fields. Spillover of individuals from crops to natural habitats also happens. This impacts the ability of species to adapt to crop/natural habitats

Dispersal between patches impacts community composition and related eco-evolutionary dynamics. Amount of spillover and habitat characteristics also affect evolution of dispersal and species adaptation to new habitat

Rand et al., 2006 Blitzer et al., 2012

Section Agriculture would benefit 4.4 from co-evolution scenarios that would foster the evolution of mutualistic interactions between crops and other organisms

Landscapes lead to a mosaic of coldspots and hotspots of co-evolution, linked by the dispersal of either one or the two species

Thompson, 1999 Hochberg et al., 2000

Section Agricultural activities create Eco-evolutionary dynamics gradients of productivity lead to higher diversity and 4.5 more complex networks due to fertilization when gradients of productivity are smooth and in intermediate productivity locations Section Land sharing is likely to be the safest option for 4.6 agriculture in terms of ecoevolutionary dynamics

Loeuille and Leibold, 2008a Doebeli and Dieckmann, 2003

Based on metapopulation/ metacommunity models, land sharing may better limit the evolution of pest resistance, maintain mutualistic interactions, maintain diversity

regions and allowing other regions to reach food self-sufficiency (Evenson and Gollin, 2003), albeit sometimes at the cost of increased social inequality (Jarosz, 2012). The industrialization of agriculture has had large impacts on the structure (composition and spatial configuration, Turner, 1989) of the landscape, whose composition (i.e., the nature of land covers/uses) has become less diverse as local

378

Nicolas Loeuille et al.

crops or landraces have been replaced by a few highly productive cereals (wheat, corn, and rice that now represent 40% of cultivated lands; Tilman, 1999). The local intensification of agriculture and increased inputs also reduced plant diversity in cultivated areas through improved weed control and reduced necessity for set-aside fallows. The proportion of natural or semi-natural areas in landscapes were reduced and these areas became more fragmented, partly because of an increase in cultivated area, but also because of land reallocation procedures to increase field sizes and facilitate mechanization. This has destroyed the edge habitats that surrounded small fields and acted as corridors between different parts of the natural habitat (Davies and Pullin, 2007), which, combined with intensive agricultural practices in cultivated fields, sharpened the boundaries between different habitat types. The industrialization of agriculture has also altered the temporal dynamics of land use, replacing stable perennial cultivated areas, such as orchards or pastures, with arable land (Male´zieux, 2011; Tscharntke et al., 2005) and a fast habitat turnover through crop rotations. Asynchrony of sowing and harvest dates among crops further creates a ‘hidden heterogeneity’ of the crop mosaic, such that a given landscape exhibits very different structures both within and across years (Vasseur et al., 2013), including more frequent periods with no vegetation cover (Male´zieux, 2011). The low sustainability of intensive agricultural development has been widely documented, in particular based on its negative environmental impacts (Flohre et al., 2011; Geiger et al., 2010). New paradigms have been proposed, such as ‘sustainable agricultural intensification’ (Lee et al., 2006) or ‘ecological intensification’ (Dore´ et al., 2011; Griffon, 2010; Male´zieux, 2011). Based on some principles of agroecology (Altieri, 1989, 1999), these approaches propose that part of a response to global food security challenges, in the context of increasing population size and climatic variability, relies on an increase of biodiversity at the agricultural landscape scale and within fields, and they should also increase the provision of additional ecosystem services, beyond that of food production (Macfadyen et al., 2012; Smith et al., 2012). In contrast to the limited range of management techniques used in intensive systems, at the landscape level, this ecological intensification requires more diverse cropping systems (De Schutter, 2010; Paillard et al., 2010). Within fields, higher levels of aboveground diversity could be provided, either by growing spatial associations of species, possibly in an agroforestry framework (Ratnadass et al., 2012; Smith et al., 2012), or by increasing the diversity of crops over time by diversifying rotations and/or including intercrops (Altieri, 1999). Increasing crop intra-specific genetic diversity, by increasing the numbers of genotypes per species grown either within or across fields, is another suggested solution

Eco-Evolutionary Dynamics in Agroecosystems

379

(Goldringer et al., 2006; Tooker and Frank, 2012). Because the development of ecological intensification involves large changes in agricultural management, the associated (initial) economic costs must be considered. Recent studies suggest that current trends in yield may not be sufficient to meet future demands (Ray et al., 2013), so ideally ecological intensification should not decrease yields relative to the present situation (Tilman et al., 2011). To achieve this will require considerable development in scientific knowledge, training of farmers and more sophisticated landscape planning (Dore´ et al., 2011, Me´die`ne et al. 2011). The predicted changes in agricultural systems are likely to increase the diversity and the fragmentation of agricultural habitats, but whether the converse will be true for natural and semi-natural habitats will depend on the options taken during landscape planning. Future agricultural management is often seen as a choice between land sparing versus land sharing. With an increasing population to feed, the land-sparing solution proposes to increase the local productivity of fields by any means necessary, including high inputs, so that remaining ecosystems can be preserved. Following this approach, the result would be very intensive, nature unfriendly, agricultural fields on one side, and spared patches of wildlife on the other side, the two being as clearly separated as possible (Fig. 6.7, left). In the land-sharing solution, the management of agricultural ecosystems aims at being wildlife friendly (less intensive production, less inputs, etc.), with the land being ‘shared’ by human and nature. Meeting increased food demands then needs larger agricultural surfaces. If the land-sparing solution is chosen (Fig. 6.7, left), the amplitude of environmental gradient increases, as all

Figure 6.7 Impact of land-sparing and land-sharing management options on the spatial gradient of environmental conditions. On each panel, exploited areas are in dark, natural systems in light. In the case of land sparing, the difference between the two habitats is large and they are well separated in space. In the land-sharing scenario, agricultural exploitation is made closer to natural conditions and agricultural fields are embedded in a matrix of natural habitats.

380

Nicolas Loeuille et al.

patches, both natural and agricultural, are either extremely suitable or extremely unsuitable for any species and the environment becomes more spatially aggregated. In the land-sharing situation (Fig. 6.7, right) these scenarios are reversed. Agricultural fields are managed as lightly as possible so that variations between the field and the surrounding ecosystems are minimized: the ecotone is less extreme. Also, semi-natural habitat is kept within agricultural landscapes, so that the environment is less spatially aggregated.

4.2. Consequences of agricultural landscape structure in terms of gene flow By fuelling local genetic variability or, conversely, by introducing locally maladapted genes, gene flows are major constraints on eco-evolutionary dynamics. They interact with local selection processes to determine the amount of evolutionary change. Also, they depend on the level of spatial heterogeneity and species dispersal abilities. Here, we will give a brief review of the ways the spatial organization can constrain gene flows, by considering how gene flows have affected the domestication and artificial selection process, how genes flow from cultivated species to surrounding ecosystems, and how agricultural fragmentation affects gene flows among natural populations. Finally, we will highlight the implications of gene flows for the eco-evolutionary dynamics of agricultural pests. Gene flows in cultivated–wild/weed complexes have long been recognized as an important process for crop domestication and as a source of crop improvement in traditional agriculture (de Wet and Harlan, 1975; Elias et al., 2001). Most crops indeed have wild relatives with which they can hybridize readily. Ellstrand (2003) lists 49 cultivated plant species for which there is genetic evidence of spontaneous hybridization with wild relatives among which wheat, corn and rice and 12 of the 13 most important crops in the world hybridize with their wild relatives (Ellstrand et al., 1999). Increases in cultivated areas and anthropogenic modifications of wild species habitats have promoted the likelihood of new contact zones between crops and related species (Crispo et al., 2011). When the crop and its relatives co-occur, the structure of the agricultural landscape determines the amount of crop-to-wild gene flow by affecting their relative spatial distributions. It is only recently, however, with the use of GM crops, that the impact of the spatial distribution of crops and their relatives on such gene flow has been thoroughly investigated and that attempts have been made to quantify it. Quantification has proved difficult because landscape-scale experiments are logistically challenging and gene flow events are rare. Historical gene flow estimates based on population genetic structure cannot be used

Eco-Evolutionary Dynamics in Agroecosystems

381

effectively because of the dynamical spatial distribution of crops over the landscape (Holderegger et al., 2010; Sork et al., 1999). Most effort has thus been devoted to measures of contemporaneous gene flow, as measured, for example, in Brassica napus (Hall et al., 2000), Brassica rapa (Warwick et al., 2003; Schafer et al., 2011), Beta vulgaris (Arnaud et al., 2010), and Agrostis stolonifera (Snow, 2012). Experiments have also been conducted to estimate the rate of crop-to-wild gene flow (e.g., for oilseed rape Jorgensen et al., 2009), some of which have addressed the spatial distribution of plants (Klinger et al., 1992; Massinga et al., 2003). Quantitatively, results depend on the parental species and genotypes, especially their selfing rate, as well as on the experimental design (e.g., plant densities and management) (e.g., Jorgensen et al., 2009). The amount of gene flow also depends on the relative abundance of the crop versus its wild relative (e.g., Giddings, 2000, Lavigne et al., 2002). For species that are pollinated by wind or both insects and wind, more insight concerning the impact of landscape structure may come from models describing gene flows between fields planted with the same crop (review in Beckie and Hall, 2008; Kuparinen, 2006). Qualitatively, most studies typically report leptokurtic gene flow curves, with most cross-pollination events occurring over short distances and few events at large distances. Modelling approaches indicate that gene flow will be maximized in landscapes where the wild relative populations are small and nearby (Klein et al., 2006a,b). They also indicate that, in species with sufficient long distance pollen dispersal (i.e., fat-tailed dispersal kernels, Kot et al., 1996), gene flow with distant wild relative populations will largely depend on the overall abundance of the crop over the landscape (Lavigne et al., 2008), while for short distance dispersing species, it will mainly depend on the proximity of the nearest field (Klein et al., 2006a,b). Larger relative wild populations will also be less impacted because of the protection effect of their local pollen cloud, that is, it is easier to find mates of the same species (Lavigne et al., 2008; Rhymer and Simberloff, 1996). Other landscape characteristics, such as its physical heterogeneity (barren land, high hedgerows), may either promote or limit gene flow from crops (Dupont et al., 2006; Manasse, 1992; Morris et al., 1994; Reboud, 2003). Predicting in detail how the structure of the agricultural landscapes affects crop-to-wild gene flow in insect-pollinated species is particularly challenging because pollinator movements may depend on the specific spatial distribution of habitats (Hadley and Betts, 2012). The fate of introgressed genes will depend on a balance between frequency of gene introgression and selection for the ‘cultivated’ gene in the hybrids and their progeny in the wild population (e.g., Nagylaki, 1975).

382

Nicolas Loeuille et al.

If the introgressed gene is selected for, it can result in genetic assimilation and possible losses of diversity in wild populations. Assimilation is also expected for detrimental genes when migration rates are high (e.g., Felsenstein, 1976; Haygood et al., 2003; Huxel, 1999; Nagylaki, 1975; Slatkin, 1985). Introgressed genes can negatively affect the fitness of the hybrids or subsequent generations, placing wild populations at risk of extinction via demographic swamping (Haygood et al., 2003; Kwit et al., 2011; Levin et al., 1996). Finally, if hybrids or their progeny exhibit higher fitness than their parents, they may become weedy or invasive. Given the huge, and increasing, areas that have been planted with cultivated crops and, given their ability to hybridize with wild relatives, invasions in natural areas are surprisingly infrequently reported. Situations where hybridization led to invasive or weedy species becoming agricultural pests are far more frequently reported, possibly because more easily observed and of larger economic interest. These include, for instance, Johnson grass, weedy rice and weedy beet (Ellstrand et al., 2010). Genetic assimilation is less easily observed because introgressed genes may have little or no impact on the wild species phenotype (Rhymer and Simberloff, 1996), yet reports of introgression of cultivated genes in the wild species (or subspecies) are numerous (e.g., Arias and Rieseberg, 1994; de la Cruz et al., 2005; Linder et al., 1998; Rieseberg et al., 1999; Warwick et al., 2003; Whitton et al., 1997). The relatively infrequent reports of genetic assimilation in wild relatives of crops or in invasive species that descend from crops may be caused by the generally poor adaptation of crops outside fields (de Wet and Harlan, 1975) and the low fitness of hybrids with cultivated traits. Indeed, low hybrid fitness, rather than promoting genetic assimilation of wild populations, increases the risk of local extinction because of increased demographic stochasticity and maladaptation (Ellstrand, 1992; Simberloff, 1988). The exact cause of extinction may then be difficult to determine a posteriori. The modified spatial structure of the agricultural landscape also affects existing gene flows between semi-natural habitats embedded within it, as they are prone to shrinkage and fragmentation. In theory, this should isolate wild populations from each other, increase their risk of extinction by a simultaneous increase of demographic stochasticity and of genetic drift limiting their evolutionary potential and favouring inbreeding depression (e.g., Simberloff, 1988; Stockwell et al., 2003; Young et al., 1996). These processes are, in particular, expected in species that are specialists of semi-natural environments, and which have low reproductive rates and limited dispersal abilities (Ryall and Fahrig, 2006). For these species, habitat fragmentation

Eco-Evolutionary Dynamics in Agroecosystems

383

results in a loss of connectivity among populations (Kindlmann and Burel, 2008). The overall trend for reduced diversity in populations from fragmented landscapes has been confirmed using a meta-analysis, which further highlighted that the level of diversity also decreased with age since fragmentation (Aguilar et al., 2008). However, habitat fragmentation does not always lead to genetic isolation of populations, and landscape genetics has helped to clarify this issue (Manel et al., 2003). In some landscapes, semi-natural elements, such as hedges, can connect remnant forest patches, although the effectiveness of such corridors largely depends on the species considered (Davies and Pullin, 2007), and some species are also capable of movement across the open habitats of agricultural landscapes (Kanuch et al., 2012). In certain cases, gene flows by pollen may even be enhanced in fragmented landscapes, as is often the case for trees (Bacles and Jump, 2011), and in some species with long distance dispersal, isolated individuals may receive more diverse pollen than those growing in large groups (Ismail et al., 2012; Klein et al., 2006a,b). A different pattern emerges for insect-pollinated species (Hadley and Betts, 2012), for which the combined effects of habitat loss and fragmentation tend to have negative impacts, probably because of pollen limitation. The most susceptible species are self-incompatible species that need outcrossing from genetically distinct individuals (Aguilar et al., 2006). Finally, some generalist species can live in both the semi-natural areas of agroecosystems and in the fields, provided that pesticide intensity is not too high (Rand et al., 2006; Samu and Szinetar, 2002) and for them habitat opening through agricultural conversion is not necessarily negative and may even make for more connected landscapes (e.g., Eriksson, 2012; Ne`ve et al., 2008). While gene flows among natural populations affect their densities, and in some cases threaten their conservation, gene flows are also modified for pest species, with possibly large economic implications. Landscape structure influences not only the dynamics of pests but also their evolutionary dynamics (Burdon and Thrall, 2008; Plantegenest et al., 2007). Evolutionary consequences of the spatial distribution of crops over the landscape have been largely investigated in the case of the evolution of pest resistance to pesticides (insecticides, fungicides, herbicides) and of variety resistances by pathogens. The rapid increase of resistance in general is due to the widespread use of pesticides that globally increases the selection pressure for resistance and to the mode of action of compounds that focus on specific targets. Similarly, varietal resistances by pathogens have broken down, due to the deployment over large areas of a same monogenic resistance to a pathogen resulting in a

384

Nicolas Loeuille et al.

rapid increase in virulence alleles: currently estimated at 4–6 years on average in western countries (McDonald and Linde, 2002). The spatial context also provides solutions for the management of the evolution of resistance. Deployment strategies of plants or pesticides have been developed to cope with the rarefaction of new pesticide molecules and the high cost for producing varieties resistant to pathogens or, more recently, insecticidal GM crops. When the pesticide resistance (or virulence gene) is already present in the population, or when it appears quickly, as is the case for many pesticides, spatial strategies aim to reduce the overall selection pressure and slow the progression of the resistance (respectively, virulence) alleles or, even, to stabilize their frequency at a low level where resistance (respectively, virulence) has an associated fitness cost (Bourguet et al., 2013; Vacher et al., 2003). These strategies are based on establishing mosaics (also named mixtures) of resistance genes/pesticide treatments over the landscape at different spatial scales (within or between fields), so that pest populations experience heterogeneous selection pressures. The optimal spatial scale of patches within the mosaic depends on the dispersal ability of the pest/pathogen (Bourguet et al., 2013). When the resistance is not already present in the populations, as in GM crops, high-dose refuge strategies have been proposed and tested (Bourguet et al., 2013; Gould, 1991; Lu et al., 2012). High doses are meant to severely reduce pest densities to slow the emergence of resistance, be it monogenic or polygenic, and to make resistance functionally recessive. Refuges are most commonly areas that are not treated with pesticides, not planted with GM crops, or not planted with varieties resistant to a particular pathogen. They favour sensitive (respectively avirulent) individuals of the pest species when there is a fitness cost to resistance (respectively, virulence) (Bourguet et al., 2013; Carrie`re et al., 2010). Dispersal of individuals between refugia and non-refugia zones is essential (Carrie`re et al., 2010; Gould, 1998), so that their use actually alters gene flows within the landscape. Pests may also have hosts in natural populations, and wild or weedy plant species that act as a ‘reservoir’ of inocula when the crop plant is absent in the landscape are important for pathogen dynamics and evolution (Burdon and Thrall, 2008). In fact, the optimal strategy to promote the durability of varieties depends on the importance of reservoirs as sources of inocula relative to field-to-field dispersal. When reservoirs are important pest sources, high cropping ratios of resistant varieties should be promoted because the pathogen population may be maintained at a low level (Fabre et al., 2012). Wild or weedy plants may also include the host where pest

Eco-Evolutionary Dynamics in Agroecosystems

385

sexual reproduction occurs, which can accelerate pest evolution rates (Plantegenest et al., 2007). In practice, controlling the spatial structure of varieties or treatments over landscapes can be difficult when different farmers own discrete parcels of land. The scale of ecological processes then differs from the scale of management (scale mismatch hypothesis, Pelosi et al., 2010). Some large-scale management programmes have nevertheless been implemented in co-ordinated refuge strategies for GM crops (Carrie`re et al., 2012) or in landscape designs aimed at enabling efficient pest control (Bianchi et al., 2006).

4.3. Consequences of spatial modifications from a demographic point of view Consequences of individual movements across agricultural landscapes go beyond gene flows, and ‘spillover’ also modifies densities and species interactions in both the cultivated and wild habitats. Most studies have focused on movements from the wild to the cultivated habitats, with the aim to protect the ‘integrity’ of crops (Rand et al., 2006): field edges, for example, were long considered as a reservoir of weeds or diseases by farmers (Norris and Kogan, 2000). Semi-natural habitats are also seen as a source of pest enemies that can act as biological control agents. The promotion of predator and parasitoid spillovers from semi-natural areas into the crops is one aim of conservation biological control programmes (Landis et al., 2000). Landscape management options designed to promote biological control are based on increasing the proximity of cultivated to semi-natural areas, so that pest enemies active in fields may find additional food resources, refugia, or overwintering sites in adjacent semi-natural areas (Bianchi et al., 2010; Brosi et al., 2008). This increase of contact zones may, however, have evolutionary consequences for populations inhabiting semi-natural areas, by changing their interactions within their species assemblage (Knight et al., 2005). A recent review also highlighted the frequency of the opposite direction of spillover, from crops to semi-natural areas for a wide range of functional groups (Blitzer et al., 2012), which will depend on species traits, such as dispersal ability and attack rate on the pest prey (Chalak et al., 2010), and on similarity between cultivated and semi-natural areas (Opatovsky et al., 2010; Prevedello and Vieira, 2010). It may have dramatic consequences for ecological networks when crops are the dominant land cover and when they harbour larger populations of organisms than semi-natural habitats. Consequences are direct when herbivore

386

Nicolas Loeuille et al.

spillover reduces the size of semi-natural plant populations (Louda et al., 2003; McKone et al., 2001) or when generalist predators or parasitoids considered beneficial in the fields attack non-pest herbivores (Gladbach et al., 2011; Rand and Louda, 2006). Pathogens that build up large populations in fields may also affect wild related species, and wild plants may be affected by pollinator spillover from the crops. Indeed, cultivated areas may provide rich resources for pollinators, as in the case of mass flowering crops, such as bean or oilseed rape, and large pollinator populations may build up (Hanley et al., 2011; Westphal et al., 2003), which could benefit wild plant populations. However, such changes of the pollinator community may also disfavour certain competing wild pollinator species and their associated plants in semi-natural habitats next to crops (Dieko¨tter et al., 2010). The interplay of such changes in the composition of ecological communities and of gene flows crucially affects the eco-evolutionary dynamics within agricultural landscapes, and is likely to affect species coexistence (Urban et al., 2008) and the overall structure of interaction webs (Loeuille and Leibold, 2008a; Rossberg et al., 2008). Eco-evolutionary dynamics are also affected by changes in fragmentation due to agricultural activities. A directly impacted trait is dispersal: several models show that the spatial heterogeneity of environmental conditions (Mathias et al., 2001; Parvinen, 2002) or the fragmentation of the landscape in patches of asymmetric size (Hanski and Mononen, 2011, Massol et al., 2011) favour the evolution of high variability in intra-specific dispersal strategies. These theoretical results are consistent with observations in natural populations (Hanski et al., 2004). Agricultural activities, by modifying the level of fragmentation, change the selective pressures that maintain such variation. Beyond a certain level of fragmentation, positive evolutionary effects are likely to be compensated by increased local extinctions, so fragmentation effects are likely to be negative overall. A recent model simulating the impact of climate change shows that evolution of dispersal in a very fragmented landscape did not allow the species to escape extinction (Kubisch et al., 2013), while it did at intermediate fragmentation levels. These examples suggest that fragmentation caused by agricultural activities not only has evolutionary—and sometimes positive—implications for dispersal, but also should be seen in the light of other disturbances. The spatial heterogeneity created by agricultural activities raises the question of the conditions under which a species that inhabits the natural

Eco-Evolutionary Dynamics in Agroecosystems

387

habitat will adapt to the agricultural field and be maintained there. Clearly, this is of critical importance for developing sustainable agriculture and accounting for species conservation. A suitable modelling framework for addressing this question is the one developed for source–sink landscapes (Pulliam, 1988), in which some areas can maintain populations even in the absence of dispersal (sources), while others do not (sinks). The blackhole sink model by Holt et al. (2003) is particularly relevant here: it shows, in a fixed landscape, that if a population has a source (e.g., the natural habitat) and a sink (e.g., surrounding agricultural fields), then, in the absence of habitat selection, the population eventually adapts to survive in the sink. The worse the conditions are in the sink, the longer it takes for adaptation to occur, although higher dispersal from the source accelerates this rate, so adaptation can be quite abrupt and rapid. These results are particularly relevant for considering the erosion of diversity in agricultural landscapes (Robinson and Sutherland, 2002). From the black-hole sink model, one would expect that species from surrounding habitats will eventually adapt to agricultural locations and that diversity would be restored in the long run. Indeed, adaptations to new agricultural landscapes have been observed in many species (Eriksson, 2012), such as those adapted to open habitats maintained by agricultural activities. However, in high input, intensive agricultural fields, it is likely that adaptation to the black-hole sink is delayed by the extreme conditions in the sinks (i.e., the fields) so that diversity recovery is slower and may be outpaced by extinctions (Gomulkiewicz and Holt, 1995). In the light of the black-hole sink model, restoring diversity is much more feasible in traditional agriculture, where the conditions within fields and in surrounding landscapes are not so different, so that adaptation to agricultural conditions can happen faster. Restoration of diversity is important not only for the conservation of species, as diverse networks of species interactions most often have a better functioning and stability (Loreau et al., 2001) and may in turn provide more efficient and stable ecosystem services to agricultural fields (Crowder et al., 2010). The black-hole sink model can also help to explain why cultivated species so rarely invade natural ecosystems. Through artificial selection, cultivated species have often evolved a suite of traits that make them non-viable in natural conditions (Meyer et al., 2012; Purugganan and Fuller, 2009), so the surrounding ecosystems are strong black-hole sinks for them. Adaptation and invasion of these habitats would therefore be very slow and, since the evolution of these cultivated species is often constrained by human

388

Nicolas Loeuille et al.

populations in ways that are unrelated to the sink conditions, there is no reason to expect that the within-sink adaptation will ever take place.

4.4. Consequences of spatial structure for pairwise co-evolution Metapopulation models such as the black-hole sink model consider the evolution of a single species in space, and by ignoring species interactions they focus on spatial heterogeneities and fragmentation. However, most ecosystem services on which agriculture relies, such as pollination or biological control, depend on species interactions, so we need to know how the spatial effects of agricultural activities affect pairwise co-evolution within communities. A suitable framework for addressing this is the ‘geographic mosaic of co-evolution’ (Thompson, 1999, 2005), which is based on the idea that the conditions of co-evolution between two species are highly variable in space. In some places, called ‘hotspots’, the two species interact strongly and their evolution is mostly explained by their pairwise interaction. In other places (‘coldspots’), the co-evolution is less tight due to local conditions. The landscape consists of a mosaic of coldspots and hotspots, linked by the dispersal of either one or the two species (Gomulkiewicz et al., 2000). Several interaction types have been modelled in this framework, such as host–parasite interactions (Nuismer and Kirkpatrick, 2003; Nuismer et al., 2000), predator–prey interactions (Hochberg and van Baalen, 1998), or mutualistic interactions (Gomulkiewicz et al., 2003). A key strength of the geographic mosaic of co-evolution is that it goes beyond this simple classification of interaction types, so mutualistic and antagonistic interactions need not be treated separately, and can be considered as part of a continuum. The empirical situation that motivated the geographic mosaic of co-evolution explains how this new approach arose (Thompson, 1999): in the interaction between a moth (Greya politella) and a plant of the genius Lithophragma, the female moth transports pollen from flower to flower when they oviposit within the corolla, thereby providing a basis for a mutualistic interaction. The larvae subsequently eat a few flowers, decreasing plant fitness. Therefore, the balance of the interaction, positive or negative, varies depending on other local conditions. If many other pollinators are available, the interaction may be seen as negative. If pollinators are rare, the interaction can be seen as mutualistic, as the moth then provides a much-required service. The nature, strength and degree of reciprocity of the pairwise co-evolution will change accordingly, making the landscape a geographic mosaic of co-evolution (Thompson, 1999). Theoretical works on the mosaic of co-evolution have investigated this continuum of antagonistic and mutualistic

Eco-Evolutionary Dynamics in Agroecosystems

389

interactions, as well as their dynamics in space and time (Gomulkiewicz et al., 2000, 2003; Nuismer et al., 1999, 2000, 2003). This evolution along a continuum between antagonistic and mutualistic interactions has potentially huge implications for agriculture. Sustainable agriculture would, ideally, rely on ecosystem services produced by mutualistic interactions, so any co-evolution scenario that turns mutualistic interactions into antagonistic ones would be detrimental. Considering this, the geographic mosaic of co-evolution gives two important insights regarding the long-term management of interactions. First, the long-term maintenance of mutualism depends on local energetic constraints. Hochberg et al. (2000) investigated how co-evolution makes the nature of a symbiosis change along gradients of productivity. They showed that mutualistic interactions are expected in poor environments, while richer places should harbour strong parasitism. If one considers the interaction between plants and mycorrhizae or between plants and soil microbes in general, the association should evolve to be mutualistic in poor environments, but antagonistic when too much nutrient would be added, due to evolution or adaptation of either plants or mycorrhizae. In a geographic mosaic of co-evolution, nutrient addition could therefore indirectly disrupt an important ecosystem service. Because the model by Hochberg et al. (2000) is fairly general, the warning may seem speculative. However, two models more tightly focused on this particular issue (De Mazancourt and Schwartz, 2010; Thrall et al., 2007) and reviews of empirical data (Verbruggen and Kiers, 2010) support this general idea. The second important insight is that just a few strongly selecting hotspots can drive co-evolution at the landscape scale (Gomulkiewicz et al., 2000), and there is ample evidence of strong selection in agricultural systems (Gould, 1991). If an agricultural field is a hotspot for the co-evolution of an interaction, then co-evolution happening there would have important ramifications for co-evolution within surrounding ecosystems. Returning to the example of the maintenance of plant–microbe mutualism, this raises a puzzling question: if fertilization within fields were to turn some mutualistic interactions into negative ones, might it also disrupt interactions in surrounding ecosystems? An interesting example in which evolution under agricultural practices may lead to large changes in wild populations concerns pollination. In intensive landscapes, fragmentation of wild habitats and use of insecticides have altered pollinator diversity and densities, disrupting the ecological link between pollinators and wild plants. Under such conditions, evolutionary models predict the selection of more self-fertilization in wild plants (Massol and Cheptou, 2011), with large implications for the genetic structure, further evolution and ecological dynamics of plant species.

390

Nicolas Loeuille et al.

4.5. Beyond pairwise interactions: Consequences of eco-evolutionary dynamics for community structure and composition Agricultural development creates strong spatial heterogeneities in habitats and species pools: semi-natural areas coexist with agricultural fields and community structure changes markedly over space. Total diversity, the number of trophic levels, and the overall productivity or biomass of the community are all structural and functional aspects that vary between the field, its margins, and the surrounding habitats. Models accounting for spatial heterogeneities and variations in community structure can help us understand their implications for the evolutionary and ecological dynamics of species assemblages (e.g., community evolution models along gradients of productivity may explain how spatial heterogeneities in nutrient enrichment affect community stability and complexity). One relevant result of such models is that the emerging diversity eventually maintained throughout the landscape depends on the amplitude of the gradient of productivity. In a sympatric speciation model, Doebeli and Dieckmann (2003) showed that maximum diversity is obtained at intermediate gradient amplitude. If the initial gradient is shallow, increasing its steepness creates niche opportunities that are positive for the emergence of diversity. Increasing the steepness of the gradient further leads to adjacent locations having increasingly contrasted settings. Any dispersal then produces maladaptation, and such negative gene flows eventually impede the diversification process and select for generalists. This illustrates that the management of productivity gradients, a central challenge in contemporary agriculture, is likely to affect the maintenance of species diversity at landscape scales, as well as the scope for future speciation. Such results do not seem to necessarily depend on the precise settings of a particular model (seee.g.,Day,2001;Kawata,2002;Massol,2012): in avery different model based on predator–prey co-evolution, Hochberg and van Baalen (1998) also showed how the productivity gradient affects the maintenance of phenotypic diversity.Thetotaltraitdiversitymaintaineddependedontheslopeoftheproductivity gradient and peaked in areas of intermediate productivity. In an unrelated model, Loeuille and Leibold (2008a) studied the evolution of plant defences in response to specialist and generalist herbivores, along a gradient of productivity. As with the other models, regional diversity was higher when the steepness of the productivity gradient was intermediate. Local diversity was again highest and trophic structure most complex in areas of intermediate productivity. In Box 6.6, we discuss more thoroughly how this model can be used to explore

Eco-Evolutionary Dynamics in Agroecosystems

391

BOX 6.6 How enrichment and fragmentation interact, illustration using a plant defence-based evolutionary model Contemporary ecosystems are often exposed to multiple stressors, but how these interact is usually difficult to assess. In the case of nutrient enrichment and fragmentation, an important point is that local selective pressures are modified, but dispersal is also altered. Dispersal and local selective pressures interact in complex ways (Urban et al., 2008). Part of this interaction stems from gene flows that constrain the genetic variation on which natural selection can apply. Other mechanisms are also involved, such as the modification of community structure due to the incoming dispersal of new species or when dispersal creates densitydependent effects. A key condition to tackle this interaction is to consider simultaneously community structure, evolution and space. One example model that fits these requirements is the one by Loeuille and Leibold (2008a), which studies the evolution of two types of plant defences. One defence is efficient against all herbivores, but has an allocation cost that decreases plant growth rate. The second defence type is efficient against generalist herbivores, but may attract specialist herbivores, therefore having an ecological cost. Space is divided in 12 patches along a gradient of productivity, connected by passive dispersal of herbivores and plants. Evolutionary dynamics can yield diversification of the defensive strategies, but such a diversification critically depends on the level of dispersal. If dispersal is very high, the plant population has only highly defended morphs (no diversification), else three functional groups coexist within the metacommunity (low defences, intermediate defences, high defences). In the latter case, highly defended morphs are in richer patches, lowly defended morphs in poor patches and intermediate defences are observed mainly in intermediate patches. The degree of spatial overlap of these three functional groups depends again on the level of dispersal in the metacommunity. As shown in Fig. 6.8, these evolutionary dynamics constrain the architecture of the local food webs. Using these results (Fig. 6.8), it is possible to illustrate how fragmentation— implemented as a decrease of dispersal—and nutrient enrichment interact to affect the local structure of food webs. Arrow 1 shows that the effects of fragmentation can be impressive while you stand in a location of poor productivity. At first, fragmentation has a positive effect on diversity, because it increases spatial structure along the productivity gradient. Subsequently, diversity is lost as only locally adapted morphs are maintained in the patch. Arrows 1 and 2 show how this effect of fragmentation depends on the productivity of the patch: in a rich patch (arrow 2), fragmentation does not affect the local food web at all as such patches are dominated by highly defended morphs regardless of the dispersal level. Continued

392

Nicolas Loeuille et al.

BOX 6.6 How enrichment and fragmentation interact, illustration using a plant defence-based evolutionary model—cont'd

Figure 6.8 Food web architecture as constrained by the evolution of plant defences in a metacommunity. Each web is supported by a limiting inorganic nutrient (triangle) on which plants (circles) feed, supporting herbivores (squares). The white circle is a plant species whose evolution is not considered. For the other (evolving) plant, the shade of grey gives the quantity of defence the plant has at evolutionary equilibrium, from lowly defended (pale grey) to highly defended (black). At the start of the simulations, the white herbivore is a generalist feeding on both plants while the black herbivore specializes on the evolving plant species. The initial food web architecture therefore resembles the middle web of the left column. Arrows (1) and (2) show two contrasted fragmentation scenarios, arrows (3) and (4) show two enrichment scenarios, as detailed in the text. Adapted from Loeuille and Leibold (2008a). Similarly, effects of enrichment will depend on the level of fragmentation in the system. If the dispersal is very small (arrow 3), increasing the richness of the patch first increases diversity, but then decreases the complexity of the food web. Note that such a bell shape relationship between complexity and productivity is commonly observed in ecology, in models as in experiments or empirical works (Kassen et al., 2000; Mittelbach et al., 2001; Chase and Leibold, 2002). However, it is not observed when enrichment happens in a very well-connected metacommunity (arrow 4). Then, neither local diversity nor local trophic structure is modified by nutrient enrichment.

Eco-Evolutionary Dynamics in Agroecosystems

393

the management of existing interactions between nutrient enrichment and fragmentation, two common side-effects of agricultural activities.

4.6. Land sparing versus land sharing, from an evolutionary point of view The pros and cons of land sparing versus land sharing have been discussed in detail from many different viewpoints: economic, ecological, agricultural, political, etc. (Fischer et al., 2011; Green et al., 2005; Hodgson et al., 2010; Tscharntke et al., 2005). Here, we ask whether one of these two solutions is overall likely to perform better than the other, considering eco-evolutionary dynamics. This is particularly important because these management strategies involve large-scale modifications of landscapes over several decades, a timescale over which evolution needs to be considered, given the strength of selection agriculture imposes on natural systems. Before we explore the benefits and costs of the two options, let us reconsider the spatial consequences of the two choices, which differ in the amplitude of the environmental gradient (large for the sparing strategy, lower for the sharing strategy), and the level of spatial aggregation (high for sparing, lower for sharing) (Fig. 6.7). Considering these characteristics, land sharing appears to be a more sensible option in the light of eco-evolutionary models, for three main reasons. The first concerns the management of pest evolution. In the land-sparing solution (Fig. 6.7, left panel), high pesticide inputs equate to high selective pressures on the enemies of the cultivated species, so they are likely to evolve resistance in just a few generations (Carrie`re et al., 2010; Gassmann et al., 2009; Gould, 1991). The use of no-pesticide refuges is already widely applied to fight the spread of resistance (Carrie`re et al., 2010), based on many evolutionary models (Bourguet et al., 2013; Tyutyunov et al., 2008; Vacher et al., 2003). Under land-sharing management, patches of natural habitats readily provide many refugia in which resistance will be counter-selected. In the land-sparing choice, agricultural fields are aggregated, so high selective pressures and limited availability of and connectivity with refugia make the appearance and maintenance of resistance much more likely. The lessons of evolutionary models for the maintenance of diversity and the complexity of communities also suggest that land sharing is a safer choice. As evolution along a gradient of productivity of intermediate steepness is more favourable to the emergence and maintenance of species diversity (Doebeli and Dieckmann, 2003), locations of intermediate productivities should contain maximum phenotypic diversity (Hochberg and van

394

Nicolas Loeuille et al.

Baalen, 1998) and the most complex food webs (Loeuille and Leibold, 2008a, Box 6.6). The gradient associated with land sparing is likely to be too extreme to maintain diversity, and intermediate locations are absent (Fig. 6.7). Land sharing, on the other hand, provides a shallower gradient with many locations of intermediate productivity, creating a spatial structure that may harbour high diversity. A third reason why land sharing may be a better solution is linked to the results of the geographic mosaic of co-evolution. Due to gene flows, hotspots can drive the co-evolution at larger scales, even when rare, if selection is sufficiently strong (Gomulkiewicz et al., 2000). Under land-sparing management, it is likely that this condition is met because very intensive agriculture concentrates on part of the landscape, so selective pressures there will be high for any species present. Hence, if these exploited locations are hotspots for some species co-evolution—not an unlikely scenario because few species will coexist in such high-input, fragmented landscapes, so their reciprocal interactions will be relatively more important—these will influence co-evolutionary dynamics throughout the landscape. Essentially, this means that what is considered ‘land sparing’ may not really be sparing species co-evolution in space. We must point out, however, that the black-hole sink model (Holt et al., 2003) gives a more nuanced answer concerning the relative benefits of land sharing and land sparing. For instance, with the spatial heterogeneities depicted on Fig. 6.7 for a species that is best suited to the natural area in the land-sparing situation, the cultivated area is likely to be very unsuitable, creating a very strong black-hole sink. Adaptation to the sink will be very slow to arise, a drawback from a conservation point of view (although from a land-sparing perspective, these areas were not designed to support biodiversity in the first place). In land-sharing scenarios, cultivated areas are more wildlife-friendly and closer to natural conditions, so they may still be sinks, but less strong. Adaptation should therefore be faster, allowing some species to persist in the exploited areas. Now consider a cultivated species, for which the reverse reasoning can be made. In land-sparing scenarios, this species will never adapt to the natural environment, because it will be a very black-hole sink, and because the species evolves through artificial selection. But in the land-sharing scenarios, evolution of the cultivated species will have to account for conditions closer to the ones existing in semi-natural areas. Environmental conditions vary less spatially, so some of the cultivated species may adapt to the natural locations or gene introgression may become more frequent. Therefore, from an evolutionary point of view, it is quite

Eco-Evolutionary Dynamics in Agroecosystems

395

possible that land sharing eventually produces new ‘invasive species’ (or at least large introgressions) in the embedded semi-natural areas.

5. PERSPECTIVES AND CHALLENGES In this chapter, we have taken recent insights gained from general ecoevolutionary models and applied them to agricultural perspectives. On the one hand, attempting such an exercise represents a stretch because community evolution models are general, while agricultural questions are precise and embedded in real field situations. However, as exemplified by the land-sharing versus land-sparing debate, there is a need for agriculture to set a general strategy for sustainability to complement smaller-scale political decisions. General questions and debates do also exist regarding agriculture sustainability, how it is constrained, and community evolution models can be useful when these general questions arise. It is therefore desirable to try to adapt certain features of these mechanistic models to the agricultural context, to obtain more appropriate answers to these questions. The spatial heterogeneity of environmental conditions in agricultural landscapes is often quite different from that assumed in community evolution models. Some locations are agricultural fields and some others are natural habitats and transitions between the two are quite abrupt. Variations in location type and environmental characteristics should be modelled accordingly, rather than on a continuous environmental gradient, a choice that is often favoured in community evolution models (Dieckmann and Doebeli, 1999; Doebeli and Dieckmann, 2003; Norberg et al., 2012). Also, simply varying environmental conditions may not be sufficient. In agricultural fields, the cultivated species often account for most of the local biomass, so to model efficiently such situations, it is important to have the dominant species and species composition varying—again abruptly—across the landscape. The metacommunity framework is a suitable starting point for developing such evolutionary approaches (Urban et al., 2008), but adapting it fully to suit agricultural settings more effectively will involve some further efforts (Massol and Petit, 2013, Chapter 5 of this volume). Another possible issue concerns the modelling of evolutionary dynamics. Individual-based selection drives evolution in wild species, while cultivated species are most often selected based on yield, a group property. In Box 6.2, we study co-evolution, combining individual selection for the wild species and group selection for the cultivated species in a simple two-species model. The results differed greatly from the initial individual selection model, and

396

Nicolas Loeuille et al.

the robustness of the simple two-species community decreased markedly. To understand the co-evolution between cultivated species, based on human choice, and wild species, these differences in selection regimes need to be built into appropriate models in the future. If we can simultaneously adapt landscape structure and selection regime to agricultural constraints, the geographic mosaic of co-evolution could provide important new perspectives on pest management (Bousset and Che`vre, 2013). Another necessary improvement is to gain better assessments of evolutionary speed driven by artificial selection for cultivated species relative to other species in the landscape. Superficially, it would seem that artificial selection, with breeding programmes and genetic tools, would produce faster evolution for cultivated species, with the evolutionary dynamics of wild species lagging behind, but this is not necessarily true. In fact, the development of new traits, new seeds or new species for agriculture is faced with many constraints, ranging from the research time needed to develop these, to the health regulations to sell them. Once on the market, it is in the interest of companies that sell them to maintain them for a sufficient amount of time and in a sufficiently large number of places, so that the initial investment is compensated. Clearly, such marketing constraints favour the replacement of many genotypes by just a few recently engineered (Bonneuil et al., 2012). This decrease in genetic diversity in turns diminishes the potential for fast evolution in the future. Sociological issues may also be involved: the reluctance of some populations regarding some engineering tools (e.g., GMO) has clearly slowed down their use in some parts of the world. Similarly, legislation (e.g., on the exchange of seeds) also constrains the speed of evolution for cultivated species. For all these reasons, evolutionary speed may not be so fast as is commonly perceived for most cultivated species. It is, however, abrupt, as it may introduce new genes or traits on very large scales in very short times when new varieties are chosen. That intensive agriculture still uses so much input emphasizes that artificial selection has failed to provide cultivated species adapted to local conditions and their variations. Evolution of wild species on the other hand is very variable in speed and depends on the genetic variability, population size, mating system and generation time of the organism involved (Frankham, 1996; Leimu et al., 2006; Loeuille, 2010b; Soule´, 1976). It is important to bear in mind here that agricultural pests, whether they are soil pathogens (Burdon and Thrall, 2009), insect herbivores (Carrie`re et al., 2010) or weeds (Gould, 1991), all have large population sizes and short generation time, allowing them to adapt rapidly to changes imposed by

Eco-Evolutionary Dynamics in Agroecosystems

397

agriculture. Many other species (pollinators, predators, etc.) do not have such traits and therefore cannot evolve at a similar pace, which might help to partially explain their declines in agricultural landscapes. The assessment of evolutionary speed is clearly critical for the associated modelling and if species are expected to evolve on similar time scales, a co-evolutionary model is most appropriate. On the contrary, if a subset of species evolves sufficiently slowly, it is possible to ignore their evolution for short or intermediate term questions, decreasing the complexity of the model and increasing its robustness. The development of adequate models can help to address important challenges. One of them is the consideration of agricultural change in the context of other global scale disturbances. A large literature exists on the effects of climate change, and the understanding of multispecies system responses to global change has been the subject of considerable research (Hagen et al., 2012; Ledger et al., 2013; O’Gorman et al., 2012; Raffaelli and White, 2013). Understanding of the evolutionary aspects and dispersal constraints, of climate change, are also just starting to emerge (Le Galliard et al., 2012; Norberg et al., 2012; Kubisch et al., 2013; Moya-Laran˜o et al., 2012). Because agriculture drives changes in habitats, in species composition and in fragmentation, it interacts strongly with climate change disturbances and joint studies of these two aspects are clearly urgently needed. This interaction between agricultural modifications and CO2-related disturbances also involves fertilization management, and several experiments have shown how CO2 and nutrient loads interact to create unexpected responses in wild communities (Langley and Megonigal, 2010; Reich, 2009). Finally, a lot of work remains to be done to understand how ecoevolutionary dynamics affect our perception of ecosystem services. As shown in Boxes 6.3 and 6.5, it is quite possible that the selective pressures we exert on a community (e.g., nutrient enrichment or weed removal) affect an a priori unrelated ecosystem service (biological control). Ecosystem services constrain the future sustainability of agriculture, but also the extent to which we can rely on ecological intensification (Dore´ et al., 2011; Male´zieux, 2011) or the benefits we would gain from land sharing. Trade-offs existing between different phenotypic traits may naturally couple different ecosystem services together: for instance, plant chemistry simultaneously affects the repulsion of some herbivores, the attraction of others (Mu¨ller-Scha¨rer et al., 2004; Van Zandt and Agrawal, 2004), the attraction of enemies (Poelman et al., 2008), and of pollinators (Adler et al. 2006; Adler et al., 2012). These chemical traits clearly couple pest management with

398

Nicolas Loeuille et al.

pollination and biological control ecosystem services, and some defences, such as tannins, simultaneously decrease plant growth rate and change the litter degradability (Grime et al., 1996; Whitham et al., 2003). Evolution then affects two ecosystem services: food provision, and long-term maintenance of soil organic matter. In the light of evolutionary trade-offs, it is important to develop models that investigate this coupling of various ecosystem services. Adequate community evolution models could be a great step forward, and agricultural issues could therefore provide important inspiration for the development of future theoretical models in evolutionary ecology. The implications of eco-evolutionary dynamics are increasingly being considered in theoretical and applied ecology: The American Naturalist recently devoting a complete issue on this subject, for instance. It opens many new doors to important contemporary ideas whose social and economic implications could be very valuable in the near future. For instance, the development of adaptive dynamics models, and more generally the consideration of eco-evolutionary feedbacks opened many new doors in our understanding of anti-biotic resistances (Day, 2001), of the evolution of parasite virulence (Alizon and Lion, 2011) or of vaccination strategies (Gandon and Day, 2007) in the field of biomedical research into human health. Similarly, the development of eco-evolutionary approaches in an agricultural context can unveil many new of the currently unknown consequences of artificial selection, the management of crop pests, and the sustainability of ecosystem services.

ACKNOWLEDGEMENTS In addition to INRA, IRD, UPMC and CNRS that routinely support our research, the authors acknowledge the financial support of the Region Ile de France (DimAstrea grant allocated to N. L. and G. K.) and of the INRA metaprogram SMaCH (C. L.). We thank Franc¸ois Massol and an anonymous referee for their interesting comments on an earlier version of this text.

APPENDIX A. EVOLUTION OF THE INVESTMENT INTO NUTRIENT UPTAKE, EFFECTS ON EMERGENT FUNCTIONING The simple three-compartment model describes the recycling of a limiting nutrient between primary producers (V), dead organic matter (D), and the stock of inorganic nutrient (N). Details can be found in the original article (Boudsocq et al., 2011). The nutrient is recycled within the

Eco-Evolutionary Dynamics in Agroecosystems

399

ecosystem via internal recycling rates: the primary producer mortality rate (dv), the mineralization rate of dead organic matter (md), and the uptake of the mineral nutrient by V (g). Nutrient inputs are fixed and occur in organic (Rd) and mineral (I) forms. Nutrients are lost from the ecosystem with fixed rates, respectively lv, ld, and ln for the V, D and N compartments. lv accounts for losses of nutrients through fires in terrestrial ecosystems or harvest in agricultural systems. Dead organic matter and its content in nutrient can be lost through fires, erosion, and leaching (dissolved organic matter) (ld). Mineral nutrients are lost through leaching or denitrification (ln). Finally, if the limiting nutrient is nitrogen and if there are nitrogen-fixing primary producers (leguminous plants, some cyanobacteria), we assume that fixation leads to an inflow of nutrient proportional to the primary producer biomass (fv). In addition, we define primary productivity as: ’ ¼ gNV þ fv V

ðA1Þ

The system of differential equations reads: dN ¼ md D þ I  ðgV þ ln ÞN dt dV ¼ gNV  ðdv þ lv  fv ÞV ðA2Þ dt dD ¼ dv V þ Rd  ðmd þ ld ÞD dt Using Eq. (A2), the formulas for the non-trivial ecological equilibrium can be derived: N0 ¼

V0 ¼

D0 ¼

dv þ lv  fv g md Rd þ I  ln N 0 md þ ld ðdv þ lv  fv Þð1  bÞ dv Rd þ ðI  ln N 0 Þ dv þ lv  fv

ðA3Þ

ðmd þ ld Þð1  bÞ

dv md that can be considered as the recycling efficiency dv þ lv  fv md þ ld of the ecosystem. with b ¼

400

Nicolas Loeuille et al.

We assume that the evolutionary dynamics is constrained by a trade-off between the uptake of nutrient and nutrient losses by plants: g ¼ g0 ebs dv ¼ dv0 ecs

ðA4Þ

s is the evolving trait that affects positively the investment into nutrient uptake. b and c determine, respectively, the benefit and the cost of additional investment into mineral nutrient uptake. To predict evolutionary dynamics, we use the Adaptive Dynamics framework (Dieckmann and Law, 1996; Geritz et al., 1998). In adaptive dynamics, the relative fitness of a rare mutant sm in a population s at its ecological equilibrium emerge from the demographic dynamics:
1 dVm

WVm ðsm ,sÞ ¼ ¼ gðsm ÞN 0 ðsÞ  ðdv ðsm Þ þ lv  fv Þ ðA5Þ Vm dt
Vm !0 We thus get: WVm ðsm ,sÞ ¼ g0 ebsm

dv0 ecs þ lv  fv  ðdv0 ecsm þ lv  fv Þ g0 ebs

ðA6Þ

To reach an evolutionary equilibrium the selection gradient must be null:   @WVm ðsm ,sÞ ¼ dv0 ecs ðb  c Þ þ bðlv  fv Þ ¼ 0 ðA7Þ @sm sm !s If lv > fv and c > b, there is thus a unique evolutionary equilibrium or singular strategy:   1 bðlv  fv Þ ðA8Þ s
¼ ln c ðc  bÞdv0 We focus here on this particular case (realized R
) but two other cases are possible: tragic R
(b > c, runaway evolution that leads to the asymptotic extinction of the species) or explosive R
(c > b and fv > lv, the evolutionary dynamics lead to a situation where the limiting nutrient accumulates up to a level where it is no longer limiting). Once the evolutionary singularity is determined (Eq. A8), it is possible to determine whether this singularity is convergent (in the sense that selective pressures favour strategies that are closer to the evolutionary equilibrium)

Eco-Evolutionary Dynamics in Agroecosystems

401

and whether it is invasible (in the sense that alternative strategies can invade the evolutionary singularity). Convergence and invasibility in adaptive dynamics emerge from the second derivatives of the relative fitness (Geritz et al., 1998). Given that: ! @ 2 WVm ¼ bc ðlv  fv Þ @s2 sm !s!s
ðA9Þ ! @ 2 WVm ¼ bc ðlv  fv Þ @s2m
sm !s!s

it is shown that the singular strategy is convergent and non-invasible: it is a continuously stable strategy (CSS). This means that the evolutionary dynamics eventually lead to this strategy and stop there. dN 0 @N 0 ds ¼ Finally, it is possible to show that the sign of is always negdt @s dt ative (see appendix 4 in Boudsocq et al., 2011), so that the singular strategy s
minimizes the availability of mineral nutrient and nutrient losses. Expressing  0  0 @’ and @V
@s @s s!s
, it can also be shown that the CSS neither maximizes s!s

primary productivity nor plant biomass (see Appendix 5 in Boudsocq et al., 2011). Finally, for the numerical simulations (Fig. 6.1A of the present chapter), we use fv ¼ 0.01 year1, lv ¼ 0.1 year1, ln ¼ 0.05 year1, ld ¼ 0.0077 year1, dV0 ¼ 0.275 year1,g0 ¼ 0.0137 ha kg1 year1, md ¼ 0.0766 year1, I ¼ 6 kg ha1 year1, Rd ¼ 6.7 kg ha1 year1, b ¼ 1, c ¼ 1.735.

APPENDIX B. MIXING GROUP SELECTION AND INDIVIDUAL SELECTION IN CO-EVOLUTIONARY MODELS The model described in Loeuille et al. (2002) considers a nutrient explicit version, including recycling processes, of plant–herbivore co-evolution. As shown in the original article, plant population at equilibrium is determined by the herbivore characteristics: V0 ¼ (dh/w), where dh is the intrinsic mortality rate of herbivores and w is the herbivore consumption rate. The model further assumes that plant

402

Nicolas Loeuille et al.

defences sv decrease this consumption rate w, while herbivore trait sh increases the investment of the herbivore in consumption, at the expense of mortality costs. Incorporating such traits, one gets: V0 ¼

dh ðsh Þ : wðsh ,sv Þ

Because defences decrease the consumption rate w, in terms of phenotype variation, selecting phenotypes sv that produce higher total yield V0 means selecting higher values of sv. We point out that this result is true for many other models describing plant–herbivore interaction, provided that they use prey-dependent functional responses for plant consumption and that herbivore mortality is density-independent.

APPENDIX C. EFFECTS OF ENRICHMENT ON THE CONTROL OF BIOMASS WITHIN A TRI-TROPHIC FOOD CHAIN WHEN THE HERBIVORE EVOLVES We extend the model of Loeuille and Loreau (2004) by adding a predator level. This makes the model much more relevant to some agricultural questions, as biological control exerted by enemies of herbivores is primordial for a sustainable agriculture. The ecological dynamics of the system follow the set of ordinary differential equations: 8 dN   > ¼ I  ln N þ ð1  vv Þdv V þ ð1  vh Þdh H þ 1  vp dp P  gNV > > > dt > > > > > dV > > > < dt ¼ V ðgN  dv  wH Þ dH > > ¼ H ðwV  dh  aP Þ > > > dt > > > >   dP > > > : dt ¼ P aH  dp ðC1Þ where variables N, V, H and P correspond to the pool of limiting inorganic nutrient, plant biomass, herbivore biomass and predator biomass, respectively. All compartments are expressed in mass units of limiting nutrient. I corresponds to the mass of nutrient input per unit of time in the system, ln is the diffusion rate of the nutrient out of the system, vx is the proportion

Eco-Evolutionary Dynamics in Agroecosystems

403

of the nutrient of species x that is recycled within the system, dx is the turnover rate of the biomass of species x (which combines the basic death rate of individuals, but also nutrient excreted by individuals, as well as dead tissues) and g, w, a represent the individual consumption rates of plants, herbivores and predators, respectively. By setting the right side of Eq. (C1) to zero it is possible to determine the coexistence equilibrium. This equilibrium reads: N0 ¼

wH 0 þ dv g

  aðIg  dv ln Þ þ dp dh g vp  vh  wdp ln V ¼ gwdp vp þ gadv vv dp H0 ¼ a 0

P0 ¼

ðC2Þ

wV 0  dh a

Here exponent 0 stands for ‘ecological equilibrium’. It is possible to show that, as soon as nutrient input is sufficient to maintain the coexistence equilibrium, that is to say, as soon as P0 > 0, this coexistence equilibrium is stable. The critical value of nutrient input needed to maintain the full food chain is:   wdp ln þ gdh dp vh ln dh vh I> þ dv þ ðC3Þ ga g w Once this value is reached, further enrichment affects the equilibrium biomass of the food chain as: @N 0 ¼0 @I @V 0 a ¼ @I adv vv þ wdp vp @H 0 ¼0 @I

ðC4Þ

@P 0 w ¼ @I adv vv þ wdp vp Equation (C4) is simply obtained by differentiating Eq. (C2) with respect to nutrient input rates.

404

Nicolas Loeuille et al.

Equation (C4) indicates that nutrient enrichment benefits plant and predator biomass, while insect pests (here herbivores) are controlled by their enemies. Therefore, the ecological dynamics of the system produces a desirable outcome from an agricultural point of view, with the cultivated organism benefitting from the enrichment and enemies are kept in check through biological control. Such variations of the nutrient stocks are completely consistent with those uncovered in previous works on food chains (Loeuille and Loreau, 2004; Oksanen et al., 1981). Now, we consider that herbivores evolve. We assume a trait x that corresponds to the investment of herbivores in defences against their predators. We voluntarily keep this trait as general as possible. It may simply be some toxicity or morphologies that would help the herbivore to escape predation, but also behaviours that would allow herbivores to escape their predators (vigilance, use of alternative habitats, etc.). The only assumption we make is that evolution towards higher values of this trait decreases the growth or reproduction of the herbivore. In mathematical terms, we assume that trait x can be any real value, and it affects simultaneously w and a such that both are now decreasing functions of the trait. That is: x2R w ðxÞ < 0 a0 ðxÞ < 0 0

ðC5Þ

Again, to maintain the desired level of generality, we do not explicit functions a and w so that the following results do not depend on a priori fixed trade-off shapes. We study the evolution of x using adaptive dynamics (see Appendix A for the details of this technique). Individual fitness readily emerge from population dynamics (Eq. C1):
1 dHm

W ðxm ,xÞ ¼ ¼ wðxm ÞV 0 ðxÞ  dh  aðxm ÞP 0 ðxÞ ðC6Þ Hm dt
Hm !0 Evolutionary singularities are therefore determined by:
@W ðxm ,xÞ

¼ 0 , w0 ðx
ÞV
 a0 ðx
ÞP
¼ 0 @xm
xm !x

ðC7Þ

In Eq. (C7), x
corresponds to the value of the phenotype at the evolutionary singularity and V
and P
correspond to the plant and predator biomasses at this singular trait. Computing the exact position of x
requires

Eco-Evolutionary Dynamics in Agroecosystems

405

defining more precisely the trade-off, that is to say, to define precisely the functions a(x) and w(x). This is unnecessary for our argument, so we will just assume that at least one value of x exists that satisfies Eq. (C7). Would it not be the case, evolutionary dynamics would produce ever increasing or ever decreasing values of x. Eventually, such runaway dynamics would induce the extinction of the predator (evolutionary murder) or of the herbivore (evolutionary suicide). Assuming that x
exists, it is now necessary to check whether evolution will lead to an end at this point or not. The strategy cannot be invaded provided:
@ 2 W ðxm ,xÞ

< 0 , w00 ðx
ÞV
 a00 ðx
ÞP
< 0 ðC8Þ @xm 2
xm !x!x
The convergence condition can be determined from the following condition:

@ 2 W ðxm ,xÞ

@ 2 W ðxm ,xÞ

þ <0 @xm 2
xm !x!x
@xm @x
xm !x!x


!
@V

P @P
@x x <0 ðC9Þ , w00 ðx
ÞV
 a00 ðx
ÞP
þ a0 ðx
Þ 


V @x x
If the evolutionary singularity satisfies Eqs. (C8) and (C9) simultaneously, then the evolutionary singularity is called a CSS. Selection then drives x towards the singular strategy x
and eventually stops at this point. Note that conditions (C8) and (C9) are not necessarily fulfilled. The trade-off shape and density-dependent effects of the trait x play a critical role in that respect. Then, it is possible that evolution runs away from the strategy. Alternatively, it is possible that evolution leads to the singularity, but that it is invasible, so that a diversification occurs through branching processes. These possibilities are interesting, but we can only understand how enrichment affects the evolutionary equilibrium when this equilibrium exists and is reached, so that we focus only on CSS cases. More complicated dynamics are beyond the scope of the present analysis. So we assume that conditions (C8) and (C9) are met. Phenotype x then eventually stabilizes at x
, defined by Eq. (C7), and we disturb the system by increasing nutrient inputs. Again, differentiating equilibrium conditions (C2) together with the evolutionary equilibrium condition (C7), one gets the following variations:

406

Nicolas Loeuille et al.

@N
a0 H
dh @x
¼ @I gaV
@I @V
aV
ðw00 V
 a00 P
Þ @x
¼ @I @I a0 dh @H
a0 H
@x
¼ @I a @I
@P w @V
¼ @I a @I @x
a2 a0 gdh V
  ¼  02 2 @I a dh dp ðln þ gvh V
Þ  a2 gV
ðw00 V
 a00 P
Þ wdp vp þ adv vv ðC10Þ Recalling from condition (C5) that a0 < 0, and from condition (C8) that w V
a00 P
< 0, and recalling that other parameters and variables of the model are all positive, it is easy to see from the first four lines of Eq. (C10) that N
, V
, H
and P
all vary in the same direction as x
when nutrient input rate I is increased. Again from inequalities (C5) and (C8), it is easy to see that the right-hand side of the last line of Eq. (C10) is positive, so that x
increases when I increases. It follows from all these considerations that when nutrient enrichment takes place, all biomasses increase, regardless of the trophic level, and that the herbivore trait of investment in defence increases too. Therefore, while the herbivore pest is controlled by its predator from an ecological dynamics point of view (see Eq. C4), it is no longer controlled when we allow evolution of defences for this herbivore. Its population does increase throughout enrichment (Eq. C10), due to this evolutionary effect. The combination of nutrient enrichment and of the evolutionary process decreases the biological control exerted by herbivores. Next to the qualitative analysis given here, an example of such an effect is displayed on Fig. 6.3 in Box 6.3. To realize this figure, we had to fix a trade-off shape. We assumed that: 00

aðxÞ ¼ a0 eka x wðxÞ ¼ w0 ekw x

ðC11Þ

Trade-off Eqs. (C11) have the advantage that, by varying ka and kw one can get a linear, convex or concave relationship between a and w, so that the complete variety of evolutionary outcomes are possible, including CSSs. Parameter values we use for the figure are the following: ln ¼ 0.06,

Eco-Evolutionary Dynamics in Agroecosystems

407

vv ¼ 0.13, vp ¼ 0.3, vh ¼ 0.3, dv ¼ 0.95, dp ¼ 0.75, dh ¼ 0.4, g ¼ 2.7, w0 ¼ 1.4, a0 ¼ 4.5, kw¼0.12, ka¼0.13.

APPENDIX D. ECO-EVOLUTIONARY DYNAMICS IN A PLANT–HERBIVORE–PREDATOR CONFRONTED TO INSECTICIDES Using ordinary differential equations we study the influence of insecticides on a three-species community as the following: 8 1 dV > > ¼ r  aV  wH ðD1Þ > > V dt > > > > < 1 dH ðD2Þ ¼ cwV  aP  dh  lj H dt > > > > > 1 dP > > ðD3Þ > : P dt ¼ gaH  dp  lh The model describes a community composed of one plant V, one herbivore H, and one predator P. The plant dynamics (D3) is constrained by the intrinsic growth rate (r) and intra-specific competition a. Hence, in the absence of the herbivore, the plant has a logistic growth. We used Holling type I functional responses to model the consumption interactions, w being the per capita attack rate of herbivores, c the energy conversion efficiency of the herbivore, a the per capita attack rate of predators, and g the conversion efficiency of this predator. Herbivore and predator dynamics also decrease through interaction-independent death rates (respectively, dh and dp) and through an additional mortality term, due to the action of insecticides. The parameter l represents the intensity of insecticide application and j and h correspond to the sensitivities of herbivores and predators, respectively. The two populations are therefore confronted to the same perturbation, to which they respond differently because of their specific sensitivities. By setting the time derivatives in Eqs. (D1), (D2) and (D3) to zero, the equilibrium points of the three-species model can be fully determined. The plant–herbivore–predator model has four equilibrium points. Existence condition requires that species densities at equilibrium are positive. We also assess the stability of these points, with the Routh–Hurwitz criterion. The trivial equilibrium S1 ¼ (V0 ¼ 0; H0 ¼ 0; P0 ¼ 0) always exists, but it is unstable when r > 0, a situation that is most often true given that the plant is cultivated.

408

Nicolas Loeuille et al.

The plant equilibrium point S2 ¼ (V0 ¼ r/a; H0 ¼ 0; P0 ¼ 0) always exists and it is stable when the perturbation by insecticides is strong enough to prevent the herbivore to invade the system. This requires: aðlj þ dh Þ > wcr

ðD4Þ

The third equilibrium point allows the coexistence of plant and herbivores. At this equilibrium S3, plants and herbivores have the following density equations: lj þ dh wc aðlj þ dh Þ þ wcr H0 ¼ w2 c V0 ¼

ðD5Þ ðD6Þ

Out of Eq. (D5) one clearly sees that the plant biomass is always positive. Therefore, existence conditions are based on the positivity of herbivore biomass. This requires that: aðlj þ dh Þ < wcr

ðD7Þ

The Routh–Hurwitz criterion shows that this plant–herbivore equilibrium point is stable when mortality incurred by predator population through pesticides is high:     agaðdh þ ljÞ > w w dp þ hl þ agr c ðD8Þ Finally, it is possible that all three species coexist at one equilibrium, where density equations are   w dp þ hl þ agr 0 ðD9Þ V ¼ aga dp þ hl H0 ¼ ðD10Þ ag     w w dp þ hl þ agr c  agaðdh þ ljÞ 0 P ¼ ðD11Þ a2 ga All the three species have positive densities when predator biomass (D11) is positive. This is ensured by the following condition:     ðD12Þ agaðdh þ ljÞ < w w dp þ hl þ agr c The Routh–Hurwitz criterion for a system of three ordinary differential equations shows that the system is stable when condition (D12) is satisfied.

Eco-Evolutionary Dynamics in Agroecosystems

409

We now assume to study how these ecological equilibriums change when more insecticides are used. By computing the partial derivatives of equilibrium equations with respect to l, we can solve analytically how the densities of the different species will vary with an increase in perturbation intensity l. When the community does not contain the predator, the partial derivatives of Eqs. (D5) and (D6) with respect to l are @V 0 j ¼ @l wc 0 @H aj ¼ 2 @l wc

ðD13Þ ðD14Þ

Thus, when the intensity of perturbation increases, plant biomass increases while herbivores are negatively affected. We similarly study changes in the full coexistence equilibrium (Eqs. D9– D11): @V 0 wh ¼ @l aga 0 @H h ¼ @l ag 0 2 @P w hc þ agaj ¼ @l a2 ga

ðD15Þ ðD16Þ ðD17Þ

At the coexistence equilibrium, the plants and predator biomasses decrease with an increase in insecticide use, while herbivores increase. Comparisons of Eqs. (D13 and D14) and of Eqs. (D15–D17) highlight the fact that the efficiency of pesticide use strongly varies with the number of trophic levels. When the system is composed of two levels, then using pesticide yields the intended result, namely the decrease of the pest population and the increase in yield. When the system is composed of three trophic levels and when the top trophic level is sensitive to insecticide, using them erodes the biological control of pests and has adverse effects from an agricultural point of view. Now that we have obtained the effects of insecticide disturbances on the ecological equilibrium and highlighted that qualitative variations are dependent on the number of trophic levels, we study how evolution of agricultural pests (here, herbivores) change such qualitative outcomes.

410

Nicolas Loeuille et al.

As pointed out in the main text, the emergence of resistance in pests is often associated with fitness costs (Bourguet et al. 2004; Carrie`re et al., 1994; Gassmann et al., 2009). Particularly, studies have shown that resistance most often decrease the allocation in growth or reproduction (Bourguet et al. 2004; Carrie`re et al., 1994). To consider this trade-off, we add in our plant–herbivore–predator community a sensitivity trait for the herbivore that affects both its mortality due to insecticides and its reproduction rate. The model becomes 8 1 dV > > ¼ r  aV  wH ðD18Þ > > V dt > > > > < 1 dH ðD19Þ ¼ c ðsÞwV  aP  dh  l ðsÞj H dt > > > > > 1 dP > > ðD20Þ > : P dt ¼ gaH  dp  lh The two functions j(s) and c(s) are positive, increasing function of s. The lower the value of s, the more resistant the pest is, but the higher the reproduction cost. We choose exponential functions for c and j, because they satisfy the different assumptions and also for mathematical convenience: c ðsÞ ¼ c0 expðvsÞ jðsÞ ¼ j0 expðzsÞ

ðD21Þ ðD22Þ

The ratio v/z controls the shape of the trade-off. By varying their relative values, we can have a large panel of trade-off shapes, from convexity to concavity. When the trait of interest and the associated trade-off are set, we use adaptive dynamics to determine the resulting eco-evolutionary dynamics (see Appendix A for an introduction of this technique). We first study the case where only plants and herbivores coexist. The fitness of a rare mutant s0 in a resident population of trait s can then be determined:
1 dHm

0 ¼ c ðs0 Þ wV 0 ðsÞ  dh  ljðs0 Þ ðD23Þ W1 ðs , sÞ ¼ dHm dt
Hm !0 with V0 ¼ (lj(s) þ m)/ac(s), the density of the plant at the equilibrium fixed by the resident trait. The associated fitness gradient that determines the direction of change of the trait is

411

Eco-Evolutionary Dynamics in Agroecosystems


@W1 ðs0 , sÞ

¼ vdh þ ljðsÞðv  zÞ @s0
s0 !s

ðD24Þ

From Eq. (D24), two qualitative possibilities emerge, depending on the trade-off shapes. If v  z the fitness gradient is always positive. The sensitivity of herbivore s always increases during the evolution. When v < z, the fitness gradient can be positive or negative depending on the resident trait value. Singular strategies may then be obtained, at which the gradient of fitness is null. Such a singular strategy is given by the following equation:   1 vdh
s1 ¼ log ðD25Þ j0 lðz  vÞ z As in Appendices A and C, we study the invasibility and convergence characteristics of this singular strategy. Recalling that we have v < z, we obtain:
@ 2 W1 ðs0 , sÞ

dh vzðv  zÞ ¼ <0 ðD26Þ
2 0 zv @s s0 !s!s
1

Because this second derivative is negative, the singular strategy cannot be invaded by surrounding mutants. It can be easily shown that:
@ 2 W1 ðs0 ,sÞ

¼0 ðD27Þ @s0 @s
s0 !s!s
1

The sum of Eqs. (D26) and (D27) is always negative, which ensures that the singular strategy is convergent, that is, evolution drives the system towards this point. The evolutionary singularity is therefore a CSS. We now characterize the evolutionary dynamics when all three species coexist. The invasion fitness of a mutant herbivore is: W2 ðs0 , sÞ ¼ wV 0 c ðs0 Þ  aP 0 ðsÞ  dh  ljðs0 Þ The fitness gradient becomes
@W2 ðs0 ,sÞ

¼ wV 0 c 0 ðsÞ  lj0 ðsÞ @s0
s0 !s The singular strategy, at which the gradient vanishes is   1 j0 lz
log s2 ¼ vz c0 wvV 0

ðD28Þ

ðD29Þ

ðD30Þ

412

Nicolas Loeuille et al.

Note that, in Eq. (D28), plant biomass is not dependent on the resident trait. Therefore, it is easy to see that @ 2W(s0 , s)/@s0 @s is equal to zero. Then, as in the plant–herbivore case, characteristics of the evolutionary singularity only depend on the second derivative with respect to s0 :

 
 @ 2 W2 ðs0 , sÞ

2 0 2  z ðD31Þ ¼ v wV c s lj s2 2 2
@s0 s0 !s!s
2 Inserting Eq. (D30) in Eq. (D31) shows that it is a decreasing function of z and that it vanishes for z ¼ v. Because at the evolutionary singularity, the system is also at equilibrium from an ecological point of view, wV 0c(s
) ¼aP0(s
) þdh þ lj(s
). Equation (D31) becomes
        @ 2 W2 ðs0 , sÞ

ðD32Þ ¼ v2 aP 0 s
2 þ dh þ v2  z2 lj s
2 2
0 @s s0 !s!s
2 Equation (D32) is positive for v > z. Therefore, the second partial derivative of invasive fitness with respect to 0 s is positive for v > z, vanishes for v ¼ z and is negative for v < z. We conclude from this that the singular strategy in this model is a repellor when the trade-off is convex (v > z) and a CSS when the trade-off is concave (v < z). The repellor situation leads to a runaway dynamics on which we cannot easily study the influence of increased pesticides. For the CSS cases, on the other hand, evolution eventually settles the trait at s2
and we can study the effects of increased pesticide use by linearizing around this equilibrium situation. To study the influence of increased pesticide use, we differentiate equations of the singular strategies s1
and s2
(Eqs. D25 and D30) regarding pesticide use parameter l. When the system only contains plants and herbivores, effects of pesticide use on the position of the singular strategy are determined by the following equation: @s1
1 ¼ @l lz

ðD33Þ

In such instances, the singular strategy always decreases when pesticide use increases. From an agricultural point of view, it means that we expect evolution of further resistance in all instances when more insecticide is sprayed.

413

Eco-Evolutionary Dynamics in Agroecosystems

When the three species coexist, variations in the singular strategy with pesticide use are determined from the following equations: wdp  agr @s
2   ¼   @l l w dp þ hl  agr ðv  zÞ

ðD34Þ

@s2
wl   ¼  @h w dp þ hl þ agr ðv  zÞ

ðD35Þ

Note that again, given condition of coexistence (D12), when the singular strategy is a CSS, it decreases with l, meaning that increased pesticide use allows for evolution of higher levels of resistance. The sensitivity of the predator also affects on the singular strategy. The singular strategy decreases with predator sensitivity h. In agricultural terms, it clearly indicates that sensitive predators allows for higher levels of resistance in the pest crop.

APPENDIX E. EVOLUTION OF SPECIALIZATION RATE OF THE PEST AND ITS ECOLOGICAL CONSEQUENCES We tailor the model of Leibold (1996) to fit the agricultural scenario of an insect-pest feeding on two plant types, one weed and one crop, which in their turn compete over a common soil resource. The ecological dynamics of this system follow the set of ordinary differential equations:    

dH ¼ H hc xc þ xοc Vc þ hw xw þ xοw Vw  dh dt  

dVc ¼ Vc vc gc R  wc þ woc H  dc dt  

dVw ¼ Vw vw gw R  ww þ wow H  dw dt dN ¼ I  lN  gc NVc  gw NVw dt

ðE1Þ

Where woi and wi (i ¼ c, w) denote the plant and herbivore-dependent parts of the plant–herbivore interaction; when the plant is maximally defended, woi ¼ 0. Note that we suppose these density-dependent effects are additive. The herbivore-dependent part wi can also be interpreted as degree of specialization on the target plant i. Other variables and parameters have been defined in Appendix C. The only additional parameters appearing in this

414

Nicolas Loeuille et al.

model are the conversion efficiencies in the functional responses of the pest and the two plant species (hc, hw, vc, vw). Here, we make some fairly realistic and general assumptions about the crop and the weed species. First, crop most often requires fertilization to grow and survive, so that we assume the crop is poor competitor for the resource comdw dc pared to the weed (i.e., < ). We also assume that the crop species is vw gw vc gc relatively less vulnerable to herbivores owing to its enhanced herbivore resistance favoured during breeding efforts (woc < < wow so that ww þ wow > wc þ woc). In other words, while the crop is mostly limited by the resource, the weed is mostly limited by the pest. Theoretically, the stable coexistence of the two competing plant species is assured when the following two conditions are met: (i) the ratio of resource effects and herbivore effects on the per capita growth of the crop species is higher than the same ratio of effects on the vc gc vw gw per capita growth of the weed (i.e., > , where Xi ¼ wi þ wοi ), Xc Xw (ii) the ratio of weed impacts on herbivore and resource growth is higher than hw Xw H hc Xc H > ) (Leibold, 1996). the same ratio of crop impacts (i.e., gw R gc R The coexistence equilibrium of the set of Eqs. (E1) reads as follows:     0 1 hw Xw I  lN 0 hc Xc I  lN 0 dc dw dh  B Xc  Xw C 0 dh  gw N0 g N0 0 CV ¼    c  V N0 ¼ B ¼ w @vc gc vw gw A c hc Xc hw Xw hw Xw hc Xc  gc gw   Xc Xw gc gw gw gc 0 1 dc dw  B v g v g v g v c w w gw C wB c c C H0 ¼ c ðE2Þ @ v g v c w gw A c Xc Xw  Xc Xw Note that the equilibrium biomass of the crop (weed) increases @Vw0 @Vc0 < 0, > 0) (decreases) linearly with the resource input rate I (i.e., @I @I whereas the equilibrium biomasses of the resource and the pest remain @H 0 @R0 unchanged ( , ¼ 0). @I @I Now, we consider that the pest evolves. In specific, we assume that the degree of specialization on the crop (wc) coevolves with the degree of specialization on the weed (ww) in a traded-off manner. This trade-off can be

Eco-Evolutionary Dynamics in Agroecosystems

415

explained in terms of energetic limitations: the energy spent in searching and handling of one plant type comes to a cost for searching and handling of the other one. Again we assume that a trait x affects both traits and ww such that w0 c ðxÞ > 0, then w0 w ðxÞ < 0. Next, we follow the methodological steps of adaptive dynamics, as introduced in appendix A. First, we define the relative fitness of a rare variant xm in a population x as follows:     ðE3Þ WH ðxm ,xÞ ¼ hc wc ðxm Þ þ wοc Vc0 þ hw ww ðxm Þ þ wοw Vw0  dh The evolutionary singularity of the above fitness function is obtained by setting the first derivative of the above fitness function to zero:
@Wh ðxm ,xÞ

¼ 0 , hc w0c ðx
ÞVc
þ hw w0w ðx
ÞVw
¼ 0 ðE4Þ @xm
xm ¼x¼x
where x
denotes the singular strategy and Vc
, Vw
correspond to the equilibrium biomass of the crop and the weed at the singularity, respectively. Then, the invasibility of the singular strategy x
can be determined by the following condition:
@ 2 Wh ðxm ,xÞ

< 0 , hc w00c ðx
ÞVc
þ hw w00w ðx
ÞVw
< 0 ðE5Þ
@x2m
xm ¼x¼x The convergence stability of the singular strategy can be determined from the following condition:

@ 2 Wh ðxm ,xÞ

@ 2 Wh ðxm ,xÞ

þ <0
@x2m @xm @x
xm ¼x¼x
xm ¼x¼x
ðE6Þ
 

@Vc

Vc
@Vw


00

00

<0  , hc wc ðx ÞVc þ hw ww ðx ÞVw þ hc @x
x
Vw
@x
x
It follows that the singular strategy represents a CSS when both specialization rates saturate with increasing x (i.e., w00c ðxÞ,w00w ðxÞ are negative). We focus on the CSS case and again look at the adaptive changes generated in the equilibrium biomasses of the model compartments when we perturb the system by changing input rate I (e.g., due to fertilization). Noting, as in the other appendices, the ecological equilibrium by 0 and the evolutionary equilibrium by
, the adaptive responses of species biomass to the above perturbation were calculated to have the following signs (explicit formulas are not provided because of their complexity):

416

Nicolas Loeuille et al.

@H
@V 0 @V
@V 0 @V
@R
> 0, c > c > 0, w < w < 0, >0 @I @I @I @I @I @I

ðE7Þ

and @H
@Vc0 @Vc
@Vw0 @Vw
@R
< 0, > > 0, < < 0, <0 @dw @dw @dw @dw @dw @dw Several things are worth noticing in the above line of inequalities: first, the perturbation parameter I generates a positive effect on trait x, which translates to a positive effect on the trait wc and a negative one on the trait ww. Second, the ecological effect of a change in I on plant biomasses is moderated by the evolution of specialization (See Fig. 6.6). Therefore, consistent with results detailed in other appendices, these last results suggest that using purely ecological models may give over-optimistic results compared to the complete eco-evolutionary feedbacks. Third, the evolution of specialization enforces the top-down control of the herbivore, which results in a positive response of the herbivore’s biomass to fertilization. To realize Fig. 6.6, we had to fix a trade-off shape. We assumed that: 1=z wc ðxÞ ¼ wmax c x 1=z ww ðxÞ ¼ wmax , where x 2 ½0,1 w ð1  xÞ

ðE8Þ

The parameter z measures the strength of the trade-off; when z > 1 the trade-off is weak (w00c ðxÞ,w00w ðxÞ are negative) and when z < 1 the trade-off is strong (w00c ðxÞ,w00w ðxÞ are positive). Parameter values we use for the Fig. 6.6 are the following: hc ¼ 0.5, hw ¼ 0.7, wc max ¼ 1, wc 0 ¼ 0.01, ww max ¼ 1, ww 0 ¼ 0.1, dh ¼ 0.1, dc ¼ 0.4, dw ¼ 0.2, vc ¼ 0.5, vw ¼ 1, gc ¼ 0.6, gw ¼ 0.3, l ¼ 0.1, z ¼ 1.5.

GLOSSARY Ecological network a complex set of species linked by interactions of different types: direct/indirect, trophic/non-trophic, mutualistic/antagonistic. The complexity characterizing an ecological network constrains species ecological dynamics and adaptive responses to disturbances. Fitness (individual) number of reproducing offspring left by an individual during its lifetime. Adaptability ability of a population to cope with environmental variations through genetic change (natural selection, gene flows), phenotypic plasticity or behavioural change. Natural selection a gradual, non-random process by which phenotypic/genotypic traits change in frequency as a function of fitness differences among their bearers. Important features of selection include: Dimensionality number of selective pressures operating simultaneously,

Eco-Evolutionary Dynamics in Agroecosystems

417

Intensity difference between maximum and minimum fitness within the population. It depends on environmental changes/fluctuation, changes in the ecological context, gene flow, etc. Group selection (in the case of agriculture, Artificial Group Selection) selection based on the fitness of the group rather than individual fitness; this may result in fixation of traits disadvantageous to the individual itself, provided there is some heritability of the group property under selection. Artificial selection human-driven selection of organisms favouring those with desirable characteristics, for example, for cultivation and use. These characteristics may be negatively linked to organism fitness. Adaptive change/response a phenotypic change (with a genetic or non-genetic basis) that improves the fitness of individuals relative to average fitness within the population; it can be limited by allocation or ecological trade-offs. Allocation trade-off beneficial changes in one trait cause detrimental changes in another due to energetic or time constraints within an individual. Ecological trade-off beneficial changes in a species response to one interaction incur costs from another interaction (e.g., plant defences against herbivores may have detrimental effects on pollination); the higher the complexity of an ecological network the more important these trade-offs are. Domestication outcome of the artificial selection process linked with cultivation by humans of plants and animals. Diffuse co-evolution allelic/phenotypic changes occurring within the frame of an ecological network, that is, species evolve in response to a number of other species of the network, each of which is also evolving in response to another set of species (Janzen, 1980). Contemporary (or rapid) evolution allelic/phenotypic change occurring in natural ecosystems on a short time frame Eco-evolutionary feedback situation in which the ecological context constrains natural selection, while resulting evolution affects the ecological context through changes in species density, species interactions or the environment.

REFERENCES Abrams, P.A., 2009. Adaptive changes in prey vulnerability shape the response of predator populations to mortality. J. Theor. Biol. 261, 294–304. Abrams, P.A., Matsuda, H., 2005. The effect of adaptive change in the prey on the dynamics of an exploited predator population. Can. J. Fish. Aquat. Sci. 62, 758–767. Adler, L., Bronstein, J., 2004. Attracting antagonists: does floral nectar increase leaf herbivory? Ecology 85, 1519–1526. Adler, L.S., Wink, M., Distl, M., Lentz, A.J., 2006. Leaf herbivory and nutrients increase nectar alkaloids. Ecol. Lett. 9, 960–967. Adler, L.S., Seifert, M.G., Wink, M., Morse, G.E., Turlings, T., 2012. Reliance on pollinators predicts defensive chemistry across tobacco species. Ecol. Lett. 15, 1140–1148. Agnew, P., Berticat, C., Bedhomme, S., Sidobre, C., Michalakis, Y., 2004. Parasitism increases and decreases the costs of insecticide resistance in mosquitoes. Evolution 58, 779–786. Agrawal, A., 2001. Phenotypic plasticity in the interactions and evolution of species. Science 249, 321–326.

418

Nicolas Loeuille et al.

Agrawal, A.A., Hastings, A.P., Johnson, M.T.J., Maron, J.L., Salminen, J.P., 2012. Insect herbivores drive real-time ecological and evolutionary change in plant populations. Science 338, 113–116. Aguilar, R., Ashworth, L., Galetto, L., Aizen, M.A., 2006. Plant reproductive susceptibility to habitat fragmentation: review and synthesis through a meta-analysis. Ecol. Lett. 9, 968–980. Aguilar, R., Quesada, M., Ashworth, L., Herrerias-Diego, Y., Lobo, J., 2008. Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches. Mol. Ecol. 17, 5177–5188. Alizon, S., Lion, S., 2011. Within-host parasite cooperation and the evolution of virulence. Proc. R. Soc. B 278, 3738–3747. Allendorf, F.W., England, P.R., Luikart, G., Ritchie, P.A., Ryman, N., 2008. Genetic effects of harvest on wild animal populations. Trends Ecol. Evol. 23, 327–337. Allesina, S., Tang, S., 2012. Stability criteria for complex ecosystems. Nature 483, 205–208. Altermatt, F., Pearse, I.S., 2011. Similarity and specialization of the larval versus adult diet of European butterflies and moths. Am. Nat. 178, 372–382. Altieri, M.A., 1981. Weeds may augment biological control of insects. Calif. Agric. 35, 22–23. Altieri, M.A., 1989. Agroecology—a new research and development paradigm for world agriculture. Agric. Ecosyst. Environ. 27, 37–46. Altieri, M.A., 1999. The ecological role of biodiversity in agroecosystems. Agric. Ecosyst. Environ. 74, 19–31. Arias, D.M., Rieseberg, L.H., 1994. Gene flow between cultivated and wild sunflowers. Theor. Appl. Genet. 89, 655–660. Armstrong, R.A., 1979. Prey species replacement along a gradient of nutrient enrichment: a graphical approach. Ecology 60, 76–84. Arnaud, J.F., Fenart, S., Cordellier, M., Cuguen, J., 2010. Populations of weedy crop-wild hybrid beets show contrasting variation in mating system and population genetic structure. Evol. Appl. 3, 305–318. Atlin, G.N., Frey, K.J., 1990. Selecting oat lines for yield in low-productivity environments. Crop. Sci. 30, 556–561. Bacles, C.F.E., Jump, A.S., 2011. Taking a tree’s perspective on forest fragmentation genetics. Trends Plant Sci. 16, 13–18. Barot, S., Heino, M., O’brien, L., Dieckmann, U., 2004. Reaction norms for age and size at maturation: study of the long term trend (1970–1998) for Georges Bank and Gulf of Maine cod stocks. Ecol. Appl. 14, 1257–1271. Barrett, L.G., Thrall, P.H., Burdon, J.J., Linde, C.C., 2008. Life history determines genetic structure and evolutionary potential of host-parasite interactions. Trends Ecol. Evol. 23, 678–685. Bascompte, J., Jordano, P., Olesen, J.M., 2006. Asymmetric coevolutionary networks facilitate biodiversity maintenance. Science 312, 431–433. Beckerman, A.P., Petchey, O.L., Warren, P.H., 2006. Foraging biology predicts food web complexity. Proc. Natl. Acad. Sci. U. S. A. 103, 13745–13749. Beckie, H.J., Hall, L.M., 2008. Simple to complex: modelling crop pollen-mediated gene flow. Plant Sci. 175, 615–628. Bell, G., Gonzalez, A., 2011. Adaptation and evolutionary rescue in metapopulations experiencing environmental deterioration. Science 332, 1327–1330. Bellon, M.R., Hodson, D., Hellin, J., 2011. Assessing the vulnerability of traditional maize seed systems in Mexico to climate change. Proc. Natl. Acad. Sci. U.S.A. 108, 13432–13437. Bennet, E.M., Peterson, G.D., Gordon, L.J., 2009. Understanding relationships among multiple ecosystem services. Ecol. Lett. 12, 1–11.

Eco-Evolutionary Dynamics in Agroecosystems

419

Bianchi, F.J.J.A., Booij, C.H.J., Tscharntke, T., 2006. Sustainable pest regulation in agricultural landscapes: a review on landscape composition, biodiversity and natural pest control. Proc. R. Soc. Lond. B 273, 1715–1727. Bianchi, F., Schellhorn, N.A., Buckley, Y.M., Possingham, H.P., 2010. Spatial variability in ecosystem services: simple rules for predator-mediated pest suppression. Ecol. Appl. 20, 2322–2333. Biesmeijer, J.C., Roberts, S.P.M., Reemer, M., Ohlemu¨ller, R., Edwards, M., Peeters, T., Schaffers, A.P., et al., 2006. Parallel declines in pollinators and insect-pollinated plants in britain and the Netherlands. Science 313 (5785), 351–354. http://dx.doi.org/ 10.1126/science.1127863. Blitzer, E.J., Dormann, C.F., Holzschuh, A., Klein, A.-M., Rand, T.A., Tscharntke, T., 2012. Spillover of functionally important organisms between managed and natural habitats. Agric. Ecosyst. Environ. 146, 34–43. Bonneuil, C., Goffaux, R., Bonnin, I., Montalent, P., Hamon, C., Balfourier, F., Goldringer, I., 2012. A new integrative indicator to assess crop genetic diversity. Ecol. Indic. 23, 280–289. Boudsocq, S., Barot, S., Loeuille, N., 2011. Evolution of nutrient acquisition: when adaptation fills the gap between contrasting ecological theories. Proc. R. Soc. Lond. B 278, 449–457. Boudsocq, S., Niboyet, A., Lata, J.C., Raynaud, X., Loeuille, N., Mathieu, J., Blouin, M., Abbadie, L., Barot, S., 2012. Plant preference for ammonium versus nitrate: a neglected determinant of ecosystem functioning? Am. Nat. 180, 60–69. Bourguet, D., Guillemaud, T., Chevillon, C., Raymond, M., 2004. Fitness costs of insecticide resistance in natural breeding sites of the Mosquito culex pipiens. Evolution 58 (1), 128–135. Bourguet, D., Delmotte, F., Franck, P., Guillemaud, T., Reboud, X., Vacher, C., Walker, A.S., Consortium, R.E.X., 2013. The skill and style to model the evolution of resistance to pesticides and drugs. Evol. Appl. 3, 375–390. Bousset, L., Che`vre, A.M., 2013. Stable epidemic control in crops based on evolutionary principles: adjusting the metapopulation concept to agro-ecosystems. Agr. Ecosyst. Environ. 165, 118–129. Brosi, B.J., Armsworth, P.R., Daily, G.C., 2008. Optimal design of agricultural landscapes for pollination services. Conserv. Lett. 1, 27–36. Burdon, J.J., Thrall, P.H., 2008. Pathogen evolution across the agro-ecological interface: implications for disease management. Evol. Appl. 1, 57–65. Burdon, J.J., Thrall, P.H., 2009. Coevolution of plants and their pathogens in natural habitats. Science (New York, N.Y.) 324 (5928), 755–756. http://dx.doi.org/10.1126/ science.1171663. Bra¨nnstro¨m, A˚., Johansson, J., Loeuille, N., Kristensen, N., Troost, T., Hille, R., Dieckmann, U., 2012. Modeling the ecology and evolution of communities: a review of past achievements, current efforts, and future promises. Evol. Ecol. Res. 14, 601–625. Butler, J., Garratt, M.P.D., Leather, S.R., 2012. Fertilisers and insect herbivores: a metaanalysis. Ann. Appl. Biol. 161, 223–233. Carrie`re, Y., Deland, J.P., Roff, D.A., Vincent, C., 1994. Life-history costs associated with the evolution of insecticide resistance. Proc. R. Soc. Lond. B 258, 35–40. Carrie`re, Y., Crowder, D.W., Tabashnik, B.E., 2010. Evolutionary ecology of insect adaptation to Bt crops. Evol. Appl. 3, 561–573. Carrie`re, Y., Ellers-Kirk, C., Hartfield, K., Larocque, G., Degain, B., Dutilleul, P., Dennehy, T.J., Marsh, S.E., Crowder, D.W., Li, X.C., Ellsworth, P.C., Naranjo, S.E., Palumbo, J.C., Fournier, A., Antilla, L., Tabashnik, B.E., 2012. Largescale, spatially-explicit test of the refuge strategy for delaying insecticide resistance. Proc. Natl. Acad. Sci. U.S.A. 109, 775–780.

420

Nicolas Loeuille et al.

Carson, R., 1962. Silent Spring. Houghton Mifflin, Boston, USA. Chalak, M., Hemerik, L., van der Werf, W., Ruijs, A., van Ierland, E.C., 2010. On the risk of extinction of a wild plant species through spillover of a biological control agent: analysis of an ecosystem compartment model. Ecol. Model. 221, 1934–1943. Chase, J.M., 2003. Strong and weak trophic cascades along a productivity gradient. Oikos 101, 187–195. Chase, J.M., Leibold, M.A., 2002. Spatial scale dictates the productivity-biodiversity relationship. Nature 416, 427–430. Chase, J.M., Leibold, M.A., 2003. Ecological Niches: Linking Classical and Contemporary Approaches. University of Chicago Press, Chicago, USA. Costanza, R., d’Arge, R., De Groot, R., Farber, S., Grasso, M., Hannon, B., Limburg, K., Naeem, S., O’Neill, R.V., Paruelo, J., Raskin, R.G., Sutton, P., van den Bel, M., 1997. The value of the world’s ecosystem services and natural capital. Nature 387, 253–260. Courtois, E., 2010. Le parfum de la de´fense des plantes en foreˆt amazonienne. PhD Thesis. University Paul Sabatier, Toulouse, France. Craine, J.M., 2009. Resource Strategies of Wild Plants. Princeton University Press, Princeton, USA. Crispo, E., Moore, J.S., Lee-Yaw, J.A., Gray, S.M., Haller, B.C., 2011. Broken barriers: human-induced changes to gene flow and introgression in animals. Bioessays 33, 508–518. Crowder, D.W., Northfield, T.D., Strand, M.R., Snyder, W.E., 2010. Organic agriculture promotes evenness and natural pest control. Nature 466, 109–113. Crutsinger, G.M., Collins, M.D., Fordyce, J.A., Gompert, Z., Nice, C.C., Sanders, N.J., 2006. Plant genotypic diversity predicts community structure and governs an ecosystem process. Science 313, 966–968. Darwin, C., 1859. The Origin of Species. John Murray, London, UK. Darwin, C., 1868. Variation of Animals and Plants Under Domestication. John Murray, London, UK. Davies, Z.G., Pullin, A.S., 2007. Are hedgerows effective corridors between fragments of woodland habitat? An evidence-based approach. Landsc. Ecol. 22, 333–351. Day, T., 2001. Population structure inhibits evolutionary diversification under competition for resources. Genetica 112–113, 71–86. De Dorlodot, S., Forster, B., Pages, L., Price, A., Tuberosa, R., Draye, X., 2007. Root system architecture: opportunities and constraints for genetic improvement of crops. Trends Plant Sci. 12, 474–481. de la Cruz, E.P., Gepts, P., GarciaMarin, P.C., Villareal, D.Z., 2005. Spatial distribution of genetic diversity in wild populations of Phaseolus vulgaris L. from Guanajuato and Michoacan, Mexico. Genet. Resour. Crop Evol. 52, 589–599. De Mazancourt, C., Dieckmann, U., 2004. Trade-off geometries and frequency-dependent selection. Am. Nat. 164, 765–778. De Mazancourt, C., Schwartz, M.W., 2010. A resource ratio theory of cooperation. Ecol. Lett. 13, 349–359. De Mazancourt, C., Loreau, M., Dieckmann, U., 2001. Can the evolution of plant defence lead to plant-herbivore mutualism? Am. Nat. 158, 109–123. de Roman, M., Fernandez, I., Wyatt, T., Sahrawy, M., Heil, M., Pozo, M.J., 2011. Elicitation of foliar resistance mechanisms transiently impairs root association with arbuscular mycorrhizal fungi. J. Ecol. 99, 36–45. De Schutter, O., 2010. UN Report submitted by the Special Rapporteur on the right to food A/HRC/16/49. www.srfood.org/images/stories/pdf/officialreports/20110308_a-hrc16-49_agroecology_en.pdf. de Wet, J., Harlan, J., 1975. Weeds and domesticates: evolution in the man-made habitat. Econ. Bot. 29, 99–107.

Eco-Evolutionary Dynamics in Agroecosystems

421

Dennis, E.S., Ellis, J., Green, A., Llewellyn, D., Morell, M., Tabe, L., Peacock, W.J., 2008. Genetic contributions to agricultural sustainability. Philos. Trans. R. Soc. B 363, 591–609. Denison, R.F., 2012. Darwinian Agriculture: How Understanding Evolution Can Improve Agriculture. Princeton University Press, Princeton, USA. Denison, R.F., Kiers, E.T., West, S.A., 2003. Darwinian agriculture: when can humans find solutions beyond the reach of natural selection? Q. Rev. Biol. 78, 145–168. Dercole, F., Ferrie`re, R., Gragnani, A., Rinaldi, S., 2006. Coevolution of slow fast populations: evolutionary sliding, evolutionary pseudo equilibria and complex Red Queen dynamics. Proc. R. Soc. Lond. B 273, 983–990. Despre´s, L., David, J.-P., Gallet, C., 2007. The evolutionary ecology of insect resistance to plant chemicals. Trends Ecol. Evol. 22, 298–307. Dieckmann, U., Law, R., 1996. The dynamical theory of coevolution: a derivation from stochastic ecological processes. J. Math. Biol. 34, 579–612. Dieckmann, U., Doebeli, M., 1999. On the origin of species by sympatric speciation. Nature 400, 354–357. Dieko¨tter, T., Kadoya, T., Peter, F., Wolters, V., Jauker, F., 2010. Oilseed rape crops distort plant-pollinator interactions. J. Appl. Ecol. 47, 209–214. Doebeli, M., Dieckmann, U., 2003. Speciation along environmental gradients. Nature 421, 259–264. Donald, C.M., 1968. The breeding of crop ideotypes. Euphytica 17, 385–403. Dore´, T., Makowski, D., Male´zieux, E., Munier-Jolain, N., Tchamitchian, M., Tittonell, P., 2011. Facing up to the paradigm of ecological intensification in agronomy: revisiting methods, concepts and knowledge. Eur. J. Agron. 34, 197–210. Drossel, B., Higgs, P.G., McKane, A.J., 2001. The influence of predator-prey population dynamics on the long-term evolution of food web structure. J. Theor. Biol. 208, 91–107. Dupont, S., Brunet, Y., Jarosz, N., 2006. Eulerian modelling of pollen dispersal over heterogeneous vegetation canopies. Agric. For. Meteorol. 141, 82–104. Duputie´, A., Massol, F., David, P., Haxaire, C., McKey, D., 2009. Traditional Amerindian cultivators combine directional and ideotypic selection for sustainable management of cassava genetic diversity. J. Evol. Biol. 22, 1317–1325. Eissenstat, D.M., Wells, C.E., Yanai, R.D., Whitbeck, J.L., 2000. Building roots in a changing environment: implications for root longevity. New Phytol. 147, 33–42. Elias, M., Mckey, D., Panaud, O., Anstett, M.C., Robert, T., 2001. Traditional management of cassava morphological and genetic diversity by the Makushi Amerindians (Guyana, South America): perspectives for on-farm conservation of crop genetic resources. Euphytica 120, 143–157. Ellers, J., Kiers, E.T., Currie, C.R., McDonald, B.R., Visser, B., Jaenike, J., 2012. Ecological interactions drive evolutionary loss of traits. Ecol. Lett. 15, 1071–1082. Ellstrand, N.C., 1992. Gene flow by pollen—implications for plant conservation genetics. Oikos 63, 77–86. Ellstrand, N.C., 2003. Current knowledge of gene flow in plants: implications for transgene flow. Philos. Trans. R. Soc. B 358, 1163–1170. Ellstrand, N.C., Heredia, S.M., Leak-Garcia, J.A., Heraty, J.M., Burger, J.C., Yao, L., Nohzadeh-Malakshah, S., Ridley, C.E., 2010. Crops gone wild: evolution of weeds and invasives from domesticated ancestors. Evol. Appl. 3, 494–504. Ellstrand, N.C., Prentice, H.C., Hancock, J.F., 1999. Gene flow and introgression from domesticated plants into their wild relatives. Annu. Rev. Ecol. Syst. 30, 539–563. Emmerson, M.C., Raffaelli, D., 2004. Predator–prey body size, interaction strength and the stability of a real food web. J. Anim. Ecol. 73, 399–409.

422

Nicolas Loeuille et al.

Enberg, K., Jørgensen, C., Dunlop, E.S., Heino, M., Dieckmann, U., 2009. Implications of fisheries-induced evolution for stock rebuilding and recovery. Evol. Appl. 2 (3), 394–414. http://dx.doi.org/10.1111/j.1752-4571.2009.00077.x. Endara, M.J., Coley, P.D., 2011. The resource availability hypothesis revisited: a metaanalysis. Funct. Ecol. 25, 389–398. Eriksson, O., 2012. Species pools in cultural landscapes—niche construction, ecological opportunity and niche shifts. Ecography 36, 403–413. Evenson, R.E., Gollin, D., 2003. Assessing the impact of the green revolution, 1960 to 2000. Science 300, 758–762. Fabre, F., Rousseau, E., Mailleret, L., Moury, B., 2012. Durable strategies to deploy plant resistance in agricultural landscapes. New Phytol. 193, 1064–1075. Falster, D.S., Westoby, M., 2003. Plant height and evolutionary games. Trends Ecol. Evol. 18, 337–343. Felsenstein, J., 1976. The theoretical population genetics of variable selection and migration. Annu. Rev. Genet. 10, 253–280. Ferrie`re, R., Dieckmann, U., Couvet, D., 2004. Evolutionary Conservation Biology. Cambridge University Press, Cambridge, UK. Fischer, J., Bata´ry, P., Bawa, K.S., Brussaard, L., Chappell, M.J., Clough, Y., Daily, G.C., et al., 2011. Conservation: limits of land sparing. Science 334, 593. Fisher, R.A., 1930. The Genetical Theory of Natural Selection. Clarendon Press, Oxford, UK. Flaherty, D.L., 1969. Ecosystem trophic complexity and densities of the Willamette mite, Eotetranychus willamettei Ewing (Acarina: Tetranychidae). Ecology 50, 911–916. Fletcher, S.B., Doebeli, M., 2013. Towards a general theory of group selection. Evolution 67, 1561–1572. Flohre, A., Fischer, C., Aavik, T., Bengtsson, J., Berendse, F., Bommarco, R., Ceryngier, P., Clement, L.W., Dennis, C., Eggers, S., Emmerson, M., Geiger, F., Guerrero, I., Hawro, V., Inchausti, P., Liira, J., Morales, M.B., Onate, J.J., Part, T., Weisser, W.W., Winqvist, C., Thies, C., Tscharntke, T., 2011. Agricultural intensification and biodiversity partitioning in European landscapes comparing plants, carabids, and birds. Ecol. Appl. 21, 1772–1781. Fontaine, C., Guimara˜es, P.R., Ke´fi, S., Loeuille, N., Memmott, J., van der Putten, W.H., van Veen, F.J.F., The´bault, E´., 2011. The ecological and evolutionary implications of merging different types of networks. Ecol. Lett. 14, 1170–1181. Forister, M.L., Dyer, L.A., Singer, M.S., Stireman, J.O., Lill, J.T., 2012. Revisiting the evolution of ecological specialization, with emphasis on insect-plant interactions. Ecology 93, 981–991. Foster, S.P., Young, S., Williamson, M.S., Duce, I., Denholm, I., Devine, G.J., 2003. Analogous pleiotropic effects of insecticide resistance genotypes in peach-potato aphids and houseflies. Heredity 91, 98–106. Frankham, R., 1996. Relationship of genetic variation to population size in wildlife. Conserv. Biol. 10, 1500–1508. Fussman, G.F., Loreau, M., Abrams, P.A., 2007. Eco-evolutionary dynamics of communities and ecosystems. Funct. Ecol. 21, 465–477. Futuyma, D.J., Moreno, G., 1988. The evolution of ecological specialization. Annu. Rev. Ecol. Syst. 19, 207–233. Gahoonia, T.S., Nielsen, N.E., 2004. Root traits as tools for creating phosphorus efficient crop varieties. Plant Soil 260, 47–57. Gandon, S., Day, T., 2007. The evolutionary epidemiology of vaccination. J. R. Soc. Interface 4, 803–817. Ga˚rdmark, A., Dieckmann, U., Lundberg, P., 2003. Life-history evolution in harvested populations: the role of natural predation. Evol. Ecol. Res. 5, 239–257.

Eco-Evolutionary Dynamics in Agroecosystems

423

Gardner, J.B., Drinkwater, L.E., 2009. The fate of nitrogen in grain cropping systems: a meta-analysis of 15N field experiments. Ecol. Appl. 19, 2167–2184. Gassmann, A.J., Carrie`re, Y., Tabashnik, B.E., 2009. Fitness costs of insect resistance to Bacillus thuringiensis. Annu. Rev. Entomol. 54, 147–163. Geiger, F., Bengtsson, J., Berendse, F., Weisser, W.W., Emmerson, M., Morales, M.B., Ceryngier, P., Liira, J., Tscharntke, T., Winqvist, C., Eggers, S., Bommarco, R., Part, T., Bretagnolle, V., Plantegenest, M., Clement, L.W., Dennis, C., Palmer, C., Onate, J.J., Guerrero, I., Hawro, V., Aavik, T., Thies, C., Flohre, A., Hanke, S., Fischer, C., Goedhart, P.W., Inchausti, P., 2010. Persistent negative effects of pesticides on biodiversity and biological control potential on European farmland. Basic Appl. Ecol. 11, 97–105. Gepts, P., 2004. Crop domestication as a long-term selection experiment. In: Janick, J. (Ed.), Long-Term Selection: Crops, Animals, and Bacteria. Plant Breed. Rev. 24, John Wiley & Sons, NJ, USA, pp. 1–40, Part 2. Geritz, S.A.H., Kisdi, E., Mesze´na, G., Metz, J.A.J., 1998. Evolutionary singular strategies and the adaptive growth and branching of the evolutionary tree. Evol. Ecol. 12, 35–57. Geritz, S.A.H., Van Der Meijden, E., Metz, J.A.J., 1999. Evolutionary dynamics of seed size and seedling competitive ability. Theor. Popul. Biol. 55, 324–343. Gersani, M., Brown, J.S., O’brien, E.E., Maina, G.M., Abramsky, Z., 2001. Tragedy of the commons as a result of root competition. J. Ecol. 89, 660–669. Giddings, G., 2000. Modelling the spread of pollen from Lolium perenne. The implications for the release of wind-pollinated transgenics. Theor. Appl. Genet. 100, 971–974. Ghalambor, C.K., McKay, J.K., Carroll, S.P., Reznick, D.N., 2007. Adaptive versus nonadaptive phenotypic plasticity and the potential for contemporary adaptation in new environments. Funct. Ecol. 21, 349–407. Gladbach, D.J., Holzschuh, A., Scherber, C., Thies, C., Dormann, C.F., Tscharntke, T., 2011. Crop-noncrop spillover: arable fields affect trophic interactions on wild plants in surrounding habitats. Oecologia 166, 433–441. Gle´min, S., Bataillon, T., 2009. A comparative view of the evolution of grasses under domestication. New Phytol. 183, 273–290. Goldringer, I., Prouin, C., Rousset, M., Galic, N., Bonnin, I., 2006. Rapid differentiation of experimental populations of wheat for heading time in response to local climatic conditions. Ann. Bot. 98, 805–817. Gomulkiewicz, R., Holt, R.D., 1995. When does evolution by natural selection prevent extinction? Evolution 49, 201–207. Gomulkiewicz, R., Thompson, J.N., Holt, R.D., Nuismer, S.L., Hochberg, M.E., 2000. Hot spots, cold spots, and the geographic mosaic theory of coevolution. Am. Nat. 156, 156–174. Gomulkiewicz, R., Nuismer, S.L., Thompson, J.N., 2003. Coevolution in variable mutualisms. Am. Nat. 162, S80–S93. Good, A.G., Shrawat, A.K., Muench, D.G., 2004. Can less yield more? Is reducing nutrient input into the environment compatible with maintaining crop production? Trends Plant Sci. 9, 597–605. Gould, F., 1991. The evolutionary potential of crop pests. Am. Sci. 79, 496–507. Gould, F., 1998. Sustainability of transgenic insecticidal cultivars: integrating pest genetics and ecology. Annu. Rev. Entomol. 43, 701–726. Grabowski, J.H., Kimbro, D.L., 2005. Predator-avoidance behavior extends trophic cascades to refuge habitats. Ecology 86, 1312–1319. Gray, M.E., Sappington, T.W., Miller, N.J., Moeser, J., Bohn, M.O., 2009. Adaptation and invasiveness of Western corn rootworm: intensifying research on a worsening pest. Annu. Rev. Entomol. 54, 303–321. Green, R.F., Cornell, S.J., Scharlemann, J.P.W., Balmford, A., 2005. Farming and the fate of wild nature. Science 307, 550–555.

424

Nicolas Loeuille et al.

Griffon, M., 2010. Pour des agricultures e´cologiquement intensives. Editions de l’Aube, La Tour-d’Aigue, France. Grime, J.P., Cornelissen, J.H.C., Thompson, K., Hodgson, J.G., 1996. Evidence of a causal connection between anti-herbivore defence and the decomposition rate of leaves. Oikos 77, 489–494. Hadley, A.S., Betts, M.G., 2012. The effects of landscape fragmentation on pollination dynamics: absence of evidence not evidence of absence. Biol. Rev. 87, 526–544. Hagen, M., Kissling, W.D., Rasmussen, C., De Aguiar, M.A.M., Brown, L.E., Carstensen, D.W., Alves-Dos-Santos, I., Dupont, Y.L., Edwards, F.K., Genini, J., Guimara˜es Jr., P.R., Jenkins, G.B., Jordano, P., Kaiser-Bunbury, C.N., ´ ., Morellato, L.P.C., Ledger, M.E., Maia, K.P., Marquitti, F.M.D., Mclaughlin, O O’Gorman, E.J., Trøjelsgaard, K., Tylianakis, J.M., Vidal, M.M., Woodward, G., Olesen, J.M., 2012. Biodiversity, species interactions and ecological networks in a fragmented world. Global change in multispecies systems part 1. Adv. Ecol. Res. 46, 89–210. Hairston, N.G., Smith, F.E., Slobodkin, L.B., 1960. Community structure, population control and competition. Am. Nat. 94, 421–425. Hall, L., Topinka, K., Huffman, J., Davis, L., Good, A., 2000. Pollen flow between herbicide-resistant Brassica napus is the cause of multiple-resistant B-napus volunteers. Weed Sci. 48, 688–694. Hamilton, W.D., 1964. The genetical evolution of social behaviour. J. Theor. Biol. 7, 1–52. Hanley, M.E., Franco, M., Dean, C.E., Franklin, E.L., Harris, H.R., Haynes, A.G., Rapson, S.R., Rowse, G., Thomas, K.C., Waterhouse, B.R., Knight, M.E., 2011. Increased bumblebee abundance along the margins of a mass flowering crop: evidence for pollinator spill-over. Oikos 120, 1618–1624. Hanski, I., Era¨lahti, C., Kankare, M., Ovaskainen, O., Sire´n, H., 2004. Variation in migration propensity among individuals maintained by landscape structure. Ecol. Lett. 7, 958–966. Hanski, I., Mononen, T., 2011. Eco-evolutionary dynamics of dispersal in spatially heterogeneous environments. Ecol. Lett. 14, 1025–1034. Hardin, G., 1968. The tragedy of the commons. Science 162, 1243–1248. Harlan, J.R., 1992. Crops and Man. American Society of Agronomy, Madison, USA. Harlan, J.R., De Wet, J.M.J., Price, E.G., 1973. Comparative evolution of cereals. Evolution 27, 311–325. Harnik, P.G., Lotze, H.K., Anderson, S.C., Finkel, Z.V., Seth Finnegan, S., Lindberg, D.R., Liow, L.H., Lockwood, R., McClain, C.R., McGuire, J.L., O’Dea, A., Pandolfi, J.M., Simpson, C., Tittensor, D.P., 2012. Extinctions in ancient and modern seas. Trends Ecol. Evol. 27, 604–617. Harrison, S., Hastings, A., 1996. Genetic and evolutionary consequences of metapopulation structure. Trends Ecol. Evol. 11, 180–183. Hasegawa, H., 2003. High-yielding rice cultivars perform best even at reduced nitrogen fertilizer rate. Crop. Sci. 43, 921–926. Haudry, A., Cenci, A., Ravel, C., Bataillon, T., Brunel, D., Poncet, C., Hochu, I., Poirier, S., Santoni, S., Gle´min, S., David, J., 2007. Grinding up wheat: a massive loss of nucleotide diversity since domestication. Mol. Biol. Evol. 24, 1506–1517. Hay, R.K.M., 1995. Harvest index: a review of its use in plant breeding and crop physiology. Ann. Appl. Biol. 126, 197–216. Haygood, R., Ives, A.R., Andow, D.A., 2003. Consequences of recurrent gene flow from crops to wild relatives. Proc. R. Soc. Lond. B 270, 1879–1886. Hendry, A.P., Kinnison, M.T., Heino, M., Day, T., Smith, T.B., Fitt, G., Bergstrom, C.T., Oakeshott, J., Jørgensen, P.S., Zalucki, M.P., Gilchrist, G., Southerton, S., Sih, A.,

Eco-Evolutionary Dynamics in Agroecosystems

425

Strauss, S., Denison, R.F., Carroll, S.P., 2011. Evolutionary principles and their practical application. Evol. Appl. 4, 159–183. Herms, D.A., Mattson, W.J., 1992. The dilemma of plants: to grow or to defend. Q. Rev. Biol. 67, 283–335. Herrera, C.M., Medrano, M., Rey, P.J., Sanchez-Lafuente, A.M., Garcia, M.B., Guitian, J., Manzaneda, A.J., 2002. Interaction of pollinators and herbivores on plant fitness suggests a pathway for correlated evolution of mutualism- and antagonism-related traits. Proc. Natl. Acad. Sci. U. S. A. 99, 16823–16828. Ho, M.D., Rosas, J.C., Brown, K.M., Lynch, J.P., 2005. Root architectural tradeoffs for water and phosphorus acquisition. Funct. Plant Biol. 32, 737. Hochberg, M., van Baalen, M., 1998. Antagonistic coevolution over productivity gradients. Am. Nat. 152, 620–634. Hochberg, M.E., Gomulkiewicz, R., Holt, R.D., Thompson, J.N., 2000. Weak sinks could cradle mutualistic symbioses—strong sources should harbor pathogenic symbioses. J. Evol. Biol. 13, 213–222. Hodgson, J.A., Kunin, W.E., Thomas, C.D., Benton, T., Gabriel, D., 2010. Comparing organic farming and land sparing: optimizing yield and butterfly populations at a landscape scale. Ecol. Lett. 13, 1358–1367. Holderegger, R., Buehler, D., Gugerli, F., Manel, S., 2010. Landscape genetics of plants. Trends Plant Sci. 15, 675–683. Holt, R.D., Gomulkiewicz, R., Barfield, M., 2003. The phenomenology of niche evolution via quantitative traits in a ‘black-hole’ sink. Proc. R. Soc. Lond. B 270, 215–224. Howarth, F.G., 1991. Environmental impacts of classical biological-control. Annu. Rev. Entomol. 36, 485–509. Huber, D.M., Watson, R.D., 1974. Nitrogen form and plant disease. Annu. Rev. Phytopathol. 12, 139–165. Hutchinson, G.E., 1961. The paradox of the plankton. Am. Nat. 95, 137–145. Huxel, G.R., 1999. Rapid displacement of native species by invasive species: effects of hybridization. Biol. Conserv. 89, 143–152. Illius, A.W., Fitzgibbon, C., 1994. Costs of vigilance in foraging ungulates. Anim. Behav. 47, 481–487. Ismail, S.A., Ghazoul, J., Ravikanth, G., Shaanker, R.U., Kushalappa, C.G., Kettle, C.J., 2012. Does long-distance pollen dispersal preclude inbreeding in tropical trees? Fragmentation genetics of Dysoxylum malabaricum in an agro-forest landscape. Mol. Ecol. 21, 5484–5496. Ito, H.C., Ikegami, T., 2006. Food web formation with recursive evolutionary branching. J. Theor. Biol. 238, 1–10. Janzen, D.H., 1980. When it is coevolution. Evolution 34, 611–612. Jarosz, L., 2012. Growing inequality: agricultural revolutions and the political ecology of rural development. Int. J. Agric. Sustain. 10, 192–199. Jorgensen, R.B., Hauser, T., D’Hertefeldt, T., Andersen, N.S., Hooftman, D., 2009. The variability of processes involved in transgene dispersal-case studies from Brassica and related genera. Environ. Sci. Pollut. Res. 16, 389–395. Justic, D., Rabalais, N.N., Turner, R.E., 1995. Stoichiometric nutrient balance and origin of coastal eutrophication. Mar. Pollut. Bull. 30, 41–46. Kanuch, P., Jarcuska, B., Schlosserova, D., Sliacka, A., Paule, L., Kristin, A., 2012. Landscape configuration determines gene flow and phenotype in a flightless forest-edge grounddwelling bush-cricket, Pholidoptera griseoaptera. Evol. Ecol. 26, 1331–1343. Kassen, R., Buckling, A., Bell, G., Rainey, P.B., 2000. Diversity peaks at intermediate productivity in a laboratory microcosm. Nature 406, 508–512. Kawata, M., 2002. Invasion of vacant niches and subsequent sympatric speciation. Proc. R. Soc. Lond. B Biol. Sci. 269, 55–63.

426

Nicolas Loeuille et al.

Kelman, W.M., Qualset, C.O., 1991. Breeding for salinity-stressed environments: recombinant inbred wheat lines under saline irrigation. Crop. Sci. 31, 1436–1442. Kiers, E.T., Hutton, M.G., Denison, R.F., 2007. Human selection and the relaxation of legume defences against ineffective rhizobia. Proc. R. Soc. Lond. B 274, 3119–3126. Kindlmann, P., Burel, F., 2008. Connectivity measures: a review. Landsc. Ecol. 23, 879–890. Kleiber, M., 1961. The Fire of Life: An Introduction to Animal Energetics. Wiley, New York, USA. Klein, E., Lavigne, C., Gouyon, P.H., 2006a. Mixing of propagules from discrete sources at long distance: comparing a dispersal tail to an exponential. BMC Ecol. 6, 3. Klein, E.K., Lavigne, C., Picault, H., Renard, M., Gouyon, P.H., 2006b. Pollen dispersal of oilseed rape: estimation of the dispersal function and effects of field dimension. J. Appl. Ecol. 43, 141–151. Klein, A.-M., Vaissie`re, B.E., Cane, J.H., Steffan-Dewenter, I., Cunningham, S.A., Kremen, C., Tscharntke, T., 2007. Importance of pollinators in changing landscapes for world crops. Proc. R. Soc. Lond. B 274, 303–313. Klinger, T., Arriola, P.E., Ellstrand, N.C., 1992. Crop-weed hybridization in radish (Raphanus sativus)-effects of distance and population size. Am. J. Bot. 79, 1431–1435. Knight, T.M., McCoy, M.W., Chase, J.M., McCoy, K.A., Holt, R.D., 2005. Trophic cascades across ecosystems. Nature 437 (7060), 880–883. http://dx.doi.org/10.1038/ nature03962. Kondoh, M., 2003. Foraging adaptation and the relationship between food-web complexity and stability. Science 299, 1388–1391. Kot, M., Lewis, M., Van Den Driessche, P., 1996. Dispersal data and the spread of invading organisms. Ecol. Lett. 77, 2027–2042. Kotowska, A.M., Cahill, J.F., Keddie, B.A., 2010. Plant genetic diversity yields increased plant productivity and herbivore performance. J. Ecol. 98, 237–245. Krause, A.E., Frank, K.A., Mason, D.M., Ulanowicz, R.E., Taylor, W.W., 2003. Compartments revealed in food-web structure. Nature 426, 282–285. Krieger, R.I., Feeny, P.P., Wilkinson, C.F., 1971. Detoxification enzymes in the gut of catterpillars: an evolutionary answer to plant defences. Science 172, 579–580. Kuparinen, A., 2006. Mechanistic models for wind dispersal. Trends Plant Sci. 11, 296–301. Kubisch, A., Degen, T., Hovestadt, T., Poethke, H.J., 2013. Predicting range shifts under global change: the balance between local adaptation and dispersal. Ecography 36 (8), 873–882. http://dx.doi.org/10.1111/j.1600-0587.2012.00062.x. Kwit, C., Moon, H.S., Warwick, S.I., Stewart, C.N., 2011. Transgene introgression in crop relatives: molecular evidence and mitigation strategies. Trends Biotechnol. 29, 284–293. Lande, R., 1979. Quantitative genetic analysis of multivariate evolution, applied to brain: body size allometry. Evolution 33, 402–416. Landis, D.A., Wratten, S.D., Gurr, G.M., 2000. Habitat management to conserve natural enemies of arthropod pests in agriculture. Annu. Rev. Entomol. 45, 175–201. Langley, J.A., Megonigal, J.P., 2010. Ecosystem response to elevated CO2 levels limited by nitrogen-induced plant species shift. Nature 466, 96–99. Lavigne, C., Klein, E.K., Couvet, D., 2002. Using seed purity data to estimate an average pollen mediated gene flow from crops to wild relatives. Theor. Appl. Genet. 104, 139–145. Lavigne, C., Klein, E.K., Mari, J.F., Le Ber, F., Adamczyk, K., Monod, H., Angevin, F., 2008. How do genetically modified (GM) crops contribute to background levels of GM pollen in an agricultural landscape? J. Appl. Ecol. 45, 1104–1113. Law, R., 2000. Fishing, selection, and phenotypic evolution. ICES J. Mar. Sci. 57, 659–668.

Eco-Evolutionary Dynamics in Agroecosystems

427

Le Bot, J., Be´nard, C., Robin, C., Bourgaud, F., Adamowicz, S., 2009. The ‘trade-off ’ between synthesis of primary and secondary compounds in young tomato leaves is altered by nitrate nutrition: experimental evidence and model consistency. J. Exp. Bot. 60 (15), 4301–4314. Le Galliard, J.-F., Re´my, A., Ims, R.A., Lambin, X., 2012. Patterns and processes of dispersal behavior in arvicoline rodents. Mol. Ecol. 21, 505–523. Ledger, M.E., Brown, L.E., Edwards, F.K., Hudson, L.N., Milner, A.M., Woodward, G., 2013. Extreme climatic events alter aquatic food webs: a synthesis of evidence from a mesocosm drought experiment. In: Woodward, G., O’Gorman, E.J. (Eds.), Advances in Ecological Research, vol. 48, Chapter 6. Academic Press, pp. 343–395. Lee, D.R., Barrett, C.B., McPeak, J.G., 2006. Policy, technology, and management strategies for achieving sustainable agricultural intensification. Agr. Econ. 34, 123–127. Leff, B., 2004. Geographic distribution of major crops across the world. Glob. Biogeochem. Cy. 18, GB1009. Leibold, M.A., 1996. A graphical model of keystone predators in food webs: trophic regulation of abundance, incidence, and diversity patterns in communities. Am. Nat. 147, 784–812. Leimu, R., Mutikainen, P., Koricheiva, J., Fisher, M., 2006. How general are positive relationships between plant population size, fitness and genetic variation? J. Ecol. 94, 942–952. Lenardis, A.E., Morvillo, C.M., Gil, A., de la Fuente, E.B., 2011. Arthropod communities related to different mixtures of oil (Glycine max L. Merr.) and essential oil (Artemisia annua L.) crops. Ind. Crop Prod. 34, 1340–1347. Levins, R., 1962. Theory of fitness in a heterogeneous environment. 1. Fitness set and adaptive function. Am. Nat. 96, 361–373. Levins, R., 1968. Evolution in Changing Environments. Princeton University Press, Princeton, NJ. Levin, D.A., FranciscoOrtega, J., Jansen, R.K., 1996. Hybridization and the extinction of rare plant species. Conserv. Biol. 10, 10–16. Lind, M.I., Johansson, F., 2007. The degree of adaptive phenotypic plasticity is correlated with the spatial environmental heterogeneity experienced by island populations of Rana temporaria. J. Evol. Biol. 20, 1288–1297. Lind, E.M., Borer, E., Seabloom, E., Adler, P., Bakker, J.D., et al., 2013. Life-history constraints in grassland plant species: a growth-defence trade-off is the norm. Ecol. Lett. 16, 513–521. Linder, C.R., Taha, I., Seiler, G.J., Snow, A.A., Rieseberg, L.H., 1998. Long-term introgression of crop genes into wild sunflower populations. Theor. Appl. Genet. 96, 339–347. Loeuille, N., 2010a. Influence of evolution on the stability of ecological communities. Ecol. Lett. 13, 1536–1545. Loeuille, N., 2010b. Consequences of adaptive foraging in diverse communities. Funct. Ecol. 24, 18–27. Loeuille, N., Leibold, M.A., 2008a. Ecological consequences of evolution in plant defences in a metacommunity. Theor. Popul. Biol. 74, 34–45. Loeuille, N., Leibold, M.A., 2008b. Evolution in metacommunities: on the relative importance of species sorting and monopolization in structuring communities. Am. Nat. 171, 788–799. Loeuille, N., Loreau, M., 2004. Nutrient enrichment and food chains: can evolution buffer top-down control? Theor. Popul. Biol. 65, 285–298. Loeuille, N., Loreau, M., 2005. Evolutionary emergence of size-structured food webs. Proc. Natl. Acad. Sci. U. S. A. 102, 5761–5766.

428

Nicolas Loeuille et al.

Loeuille, N., Loreau, M., 2009. Emergence of complex food web structure in community evolution models. In: Verhoef, H.A., Morin, P.J. (Eds.), Community Ecology. Oxford University Press, Oxford, UK, pp. 163–179. Loeuille, N., Loreau, M., Ferrie`re, R., 2002. Consequences of plant-herbivore coevolution on the dynamics and functioning of ecosystems. J. Theor. Biol. 217, 369–381. Long, S.P., Zhu, X.-G., Naidu, S.L., Ort, D.R., 2006. Can improvement in photosynthesis increase crop yields? Plant Cell Environ. 29, 315–330. Loreau, M., Mouquet, N., Holt, R.D., 2003. Metaecosystems: a theoretical framework for a spatial ecosystem ecology. Ecol. Lett. 6, 673–679. Loreau, M., Naeem, S., Inchausti, P., Bengtsson, J., Grime, J.P., Hector, A., Hooper, D.U., Huston, M.A., Raffaelli, D., Schmid, B., Tilman, D., Wardle, D.A., 2001. Biodiversity and ecosystem functioning: current knowledge and future challenges. Science 294, 804–808. Louda, S.M., Pemberton, R.W., Johnson, M.T., Follett, P.A., 2003. Nontarget effects—the Achilles’ Heel of biological control? Retrospective analyses to reduce risk associated with biocontrol introductions. Annu. Rev. Entomol. 48, 365–396. Lu, Y., Wu, K., Jiang, Y., Guo, Y., Desneux, N., 2012. Widespread adoption of Bt cotton and insecticide decrease promotes biocontrol services. Nature 487, 362–365. Macfadyen, S., Cunningham, S.A., Costamagna, A.C., Schellhorn, N.A., 2012. Managing ecosystem services and biodiversity conservation in agricultural landscapes: are the solutions the same? J. Appl. Ecol. 49, 690–694. Macfadyen, S., Gibson, R.H., Symondson, W.O.C., Memmott, J., 2011. Landscape structure influences modularity patterns in farm food webs: consequences for pest control. Ecol. Appl. 21, 516–524. Maia, C., Dovale, J.C., Fritsche-Neto, R., Cavatte, P.C., Miranda, G.V., 2011. The difference between breeding for nutrient use efficiency and for nutrient stress tolerance. Crop Breed. Appl. Biotechnol. 11, 270–275. Male´zieux, E., 2011. Designing cropping systems from nature. Agron. Sustain. Dev. 32, 15–29. Mallet, J., 1989. The evolution of insecticide resistance: have the insects won. Trends Ecol. Evol. 4, 336–340. Manasse, R.S., 1992. Ecological risks of transgenic plants: effects of spatial dispersion on gene flow. Ecol. Appl. 2, 431–438. Manel, S., Schwartz, M.K., Luikart, G., Taberlet, P., 2003. Landscape genetics: combining landscape ecology and population genetics. Trends Ecol. Evol. 18 (189), 197. Maron, J.L., Crone, E., 2006. Herbivory: effects on plant abundance, distribution and population growth. Proc. R. Soc. Lond. B 273, 2575–2584. Massinga, R.A., Al-Khatib, K., St Amand, P., Miller, J.F., 2003. Gene flow from imidazolinone-resistant domesticated sunflower to wild relatives. Weed Sci. 51, 854–862. Massol, F., 2012. A framework to compare theoretical predictions on trait evolution in temporally varying environments under different life cycles. Ecol. Complex. in press. Available online 27 June 2012. Massol, F., Cheptou, P.O., 2011. Evolutionary syndromes linking dispersal and mating system: the effect of autocorrelation in pollination conditions. Evolution 65, 591–598. Massol, F., Duputie´, A., David, P., Jarne, P., 2011. Asymmetric patch size distribution leads to disruptive selection on dispersal. Evolution 65, 490–500. Massol, F. and Petit, S., 2013. Chapter five - Interaction networks in agricultural landscape mosaics. Adv. Ecol. Res., 49, 291–338. Matson, P.A., 1997. Agricultural intensification and ecosystem properties. Science 277, 504–509. Mathias, A., Kisdi, E., Olivieri, I., 2001. Divergent evolution of dispersal in a heterogeneous landscape. Evolution 55, 246–259. Mattson, W.J., 1980. Herbivory in relation to plant nitrogen content. Annu. Rev. Ecol. Syst. 11, 119–161.

Eco-Evolutionary Dynamics in Agroecosystems

429

May, R.M., 1972. Will a large complex system be stable. Nature 238, 413–414. Mayrose, M., Kane, N.C., Mayrose, I., Dlugosch, K.M., Rieseberg, L.H., 2011. Increased growth in sunflower correlates with reduced defences and altered gene expression in response to biotic and abiotic stress. Mol. Ecol. 20 (22), 4683–4694. McDonald, B.A., Linde, C.C., 2002. Pathogen population genetics, evolutionary potential and durable resistance. Annu. Rev. Phytopathol. 40, 349–379. McEvoy, P.B., 2002. Insect-plant interactions on a planet of weeds. Entomol. Exp. Appl. 104, 165–179. McKone, M., McLauchlan, K.K., Lebrun, E.G., McCall, A.C., 2001. An edge effect caused by adult corn-rootworm beetles on sunflowers in tallgrass prairie remnants. Conserv. Biol. 15, 1315–1324. McMahon, T.A., Halstead, N.T., Johnson, S., Raffel, T.R., Romansic, J.M., Crumrine, P.W., Rohr, J.R., 2012. Fungicide-induced declines of freshwater biodiversity modify ecosystem functions and services. Ecol. Lett. 15, 714–722. Me´die`ne, S., Valantin-Morison, M., Sarthou, J.P., de Tourdonnet, S., Gosme, M., Bertrand, M., Roger-Estrade, J., Aubertot, J.N., Rusch, A., Motisi, N., Pelosi, C., Dore, T., 2011. Agroecosystem management and biotic interactions: a review. Agron. Sustain. Dev. 31, 491–514. Meyer, R.S., DuVal, A.E., Jensen, H.R., 2012. Patterns and processes in crop domestication: an historical review and quantitative analysis of 203 global food crops. New Phytol. 196, 29–48. Mittelbach, G.G., Steiner, C.F., Scheiner, S.M., Gross, K.L., Reynolds, H.L., Waide, R.B., Willig, M.R., Dodson, S.I., Gough, L., 2001. What is the observed relationship between species richness and productivity? Ecology 82, 2381–2396. Morris, W.F., Kareiva, P.M., Raymer, P.L., 1994. Do barren zones and pollen traps reduce gene escape from transgenic crops. Ecol. Appl. 4, 157–165. Moya-Laran˜o, J., Verdeny-Vilalta, O., Rowntree, J., Melguizo-Ruiz, N., Montserrat, M., Laiolo, P., 2012. Climate change and eco-evolutionary dynamics in food webs. Adv. Ecol. Res. 47, 1–80. Mu¨ller-Scha¨rer, H., Schaffner, U., Steinger, T., 2004. Evolution in invasive plants: implications for biological control. Trends Ecol. Evol. 19, 417–422. Murphy, K.M., Campbell, K.G., Lyon, S.R., Jones, S.S., 2007. Evidence of varietal adaptation to organic farming systems. Field Crops Res. 102, 172–177. Nagylaki, T., 1975. Conditions for the existence of clines. Genetics 80, 595–615. Ne`ve, G., Barascud, B., Descimon, H., Baguette, M., 2008. Gene flow rise with habitat fragmentation in the bog fritillary butterfly (Lepidoptera: Nymphalidae). BMC Evol. Biol. 8, 84. Noguera, D., Laossi, K.R., Lavelle, P., Cruz De Carvalho, M.H., Asakawa, N., Botero, C., Barot, S., 2011. Amplifying the benefits of agroecology by using the right cultivars. Ecol. Appl. 21, 2349–2356. Norberg, J., Urban, M.C., Vellend, M., Klausmeier, C.A., Loeuille, N., 2012. Eco-evolutionary responses of biodiversity to climate change. Nat. Clim. Change 2, 1–5. Norris, R.F., Kogan, M., 2000. Interactions between weeds, arthropod pests, and their natural enemies in managed ecosystems. Weed Sci. 48, 94–158. Norris, R.F., Kogan, M., 2005. Ecology of interactions between weeds and arthropods. Annu. Rev. Entomol. 50, 479–503. Nuismer, S.L., Kirkpatrick, M., 2003. Gene flow and the coevolution of parasite range. Evolution 57, 746–754. Nuismer, S.L., Thompson, J.N., Gomulkiewicz, R., 1999. Gene flow and geographically structured coevolution. Proc. R. Soc. Lond. B 266, 605–609. Nuismer, S.L., Thompson, J.N., Gomulkiewicz, R., 2000. Coevolutionary clines across selection mosaics. Evolution 54, 1102–1115.

430

Nicolas Loeuille et al.

Nuismer, S.L., Thompson, J.N., Gomulkiewicz, R., 2003. Coevolution between hosts and parasites with partially overlapping geographic ranges. J. Evol. Biol. 16, 1337–1345. Nuismer, S.L., Jordano, P., Bascompte, J., 2012. Coevolution and the architecture of mutualistic networks. Evolution 67, 338–354. Odum, E.P., 1953. Fundamentals of Ecology. Saunders, Philadelphia, USA. O’Gorman, E.J., Pichler, D.E., Adams, G., Benstead, J.P., Cohen, H., Craig, N., Cross, W.F., Demars, B.O.L., Friberg, N., Gı´slason, G.M., Gudmundsdo´ttir, R., Hawczak, A., Hood, J.M., Hudson, L.N., Johansson, L., Johansson, M.P., ´ lafsson, J.S., Junker, J.R., Laurila, A., Manson, J.R., Mavromati, E., Nelson, D., O Perkins, D.M., Petchey, O.L., Plebani, M., Reuman, D.C., Rall, B.C., Stewart, R., Thompson, M.S.A., Woodward, G., 2012. Impacts of warming on the structure and functioning of aquatic communities. Adv. Ecol. Res. 47, 81–176. Oksanen, L., Fretwell, S.D., Arruda, J., Niemela, P., 1981. Exploitation ecosystems in gradients of primary productivity. Am. Nat. 118, 240–261. O’leary, N., Smith, M.E., 1999. Breeding corn for adaptation to two diverse intercropping companions. Am. J. Alternat. Agric. 14, 158–164. Opatovsky, I., Pluess, T., Schmidt-Entling, M.H., Gavish-Regev, E., Lubin, Y., 2010. Are spider assemblages in fragmented, semi-desert habitat affected by increasing cover of agricultural crops? Agric. Ecosyst. Environ. 135, 233–237. Ostrowski, E.A., Ofria, C., Lenski, R.E., 2007. Ecological specialization and adaptive decay in digital organisms. Am. Nat. 169, E1–E20. Paillard, S., Goldringer, I., Enjalbert, J., Doussinault, G., de Vallavieille-Pope, C., Brabant, P., 2000. Evolution of resistance against powdery mildew in winter wheat populations conducted under dynamic management. I: is specific seedling resistance selected? Theor. Appl. Genet. 101, 449–456. Paillard, S., Treyer, S., Dorin, B., 2010. Prospective Agrimonde: Sce´narios et de´fis pour nourrir le monde en 2050. Ed. Quae, Versailles, France. Palumbi, S.R., 2001. Evolution—humans as the world’s greatest evolutionary force. Science 293, 1786–1790. Parvinen, K., 2002. Evolutionary branching of dispersal strategies in structured metapopulations. J. Math. Biol. 45, 106–124. Pautasso, M., Aistara, G., Barnaud, A., Caillon, S., Clouvel, P., Coomes, O.T., Deleˆtre, M., Demeulenaere, E., De Santis, P., Do¨ring, T., Eloy, L., Emperaire, L., Garine, E., Goldringer, I., Jarvis, D., Joly, H.I., Leclerc, C., Louafi, S., Martin, P., Massol, F., McGuire, S., McKey, D., Padoch, C., Soler, C., Thomas, M., Tramontini, S., 2013. Seed exchange networks for agrobiodiversity conservation. A review. Agro. Sust. Develop. 33, 151–175. Pelosi, C., Goulard, M., Balent, G., 2010. The spatial scale mismatch between ecological processes and agricultural management: do difficulties come from underlying theoretical frameworks? Agric. Ecosyst. Environ. 139, 455–462. Pelosi, C., Barot, S., Capowiez, Y., Hedde, M., Vandenbulcke, F., 2013. Pesticides and earthworms. A review. Agron. Sustain. Dev. 1–30. Pennisi, E., 2008. Plant genetics: the blue revolution, drop by drop, gene by gene. Science 320, 171–173. Perales, H.R., Benz, B.F., Brush, S.B., 2004. Corn diversity and ethnolinguistic diversity in Chiapas, Mexico. Proc. Natl. Acad. Sci. U. S. A. 102, 949–954. Persson, L., Diehl, S., Johansson, L., Andersson, G., Hamrin, S.F., 1992. Trophic interactions in temperate lake ecosystems: a test of food chain theory. Am. Nat. 140, 59–84. Pianka, E.R., 1970. On r- and K-selection. Am. Nat. 104, 592–597. Plantegenest, M., Le May, C., Fabre, F., 2007. Landscape epidemiology of plant diseases. J. R. Soc. Interface 4, 963–972.

Eco-Evolutionary Dynamics in Agroecosystems

431

Poelman, E.H., Van Loon, J.J.A., Dicke, M., 2008. Consequences of variation in plant defence for biodiversity at higher trophic levels. Trends Plant Sci. 13, 534–541. Poisot, T., Bever, J.D., Nemri, A., Thrall, P.H., Hochberg, M.E., 2011. A conceptual framework for the evolution of ecological specialisation. Ecol. Lett. 14, 841–851. Potts, S.G., Biesmeijer, J.C., Kremen, C., Neumann, P., Schweiger, O., Kunin, W.E., 2010. Global pollinator declines: trends, impacts and drivers. Trends Ecol. Evol. 25, 345–353. Prevedello, J.A., Vieira, M.V., 2010. Does the type of matrix matter? A quantitative review of the evidence. Biodivers. Conserv. 19, 1205–1223. Przystalski, M., Osman, A., Thiemt, E.M., Rolland, B., Ericson, L., Ostergard, H., Levy, L., Wolfe, M., Bu¨chse, A., Piepho, H.P., Krajewski, P., 2008. Comparing the performance of cereal varieties in organic and non-organic cropping systems in different European countries. Euphytica 163, 417–433. Puga-Freitas, R., Barot, S., Taconnat, L., Renou, J.P., Blouin, M., 2012. Signal molecules mediate the impact of the earthworm Aporrectodea caliginosa on growth, development and defence of the plant Arabidopsis thaliana. PLoS One 7, e49504. Pulliam, R.H., 1988. Sources, sinks and population regulation. Am. Nat. 132, 652–661. Purugganan, M.D., Fuller, D.Q., 2009. The nature of selection during plant domestication. Nature 457, 843–848. Quilot-Turion, B., Ould-Sidi, M.M., Kadrani, A., Hilgert, N., Genard, M., Lescourret, F., 2012. Optimization of parameters of the ‘Virtual Fruit’ model to design peach genotype for sustainable production systems. Eur. J. Agron. 42, 34–48. Raffaelli, D., White, P.C.L., 2013. Ecosystems and their services in a changing world: an ecological perspective. In: Woodward, G., O’Gorman, E.J. (Eds.), Advances in Ecological Research, vol. 48, Chapter 1. Academic Press, pp. 1–70. Rand, T.A., Louda, S.A., 2006. Spillover of agriculturally subsidized predators as a potential threat to native insect herbivores in fragmented landscapes. Conserv. Biol. 20, 1720–1729. Rand, T.A., Tylianakis, J.M., Tscharntke, T., 2006. Spillover edge effects: the dispersal of agriculturally subsidized insect natural enemies into adjacent natural habitats. Ecol. Lett. 9, 603–614. Ratnadass, A., Fernandes, P., Avelino, J., Habib, R., 2012. Plant species diversity for sustainable management of crop pests and diseases in agroecosystems: a review. Agron. Sustain. Dev. 32, 273–303. Ravigne, V., Dieckmann, U., Olivieri, I., 2009. Live where you thrive: joint evolution of habitat choice and local adaptation facilitates specialization and promotes diversity. Am. Nat. 174, E141–E169. Ray, D.K., Mueller, N.D., West, P.C., Foley, J.A., 2013. Yield trends are insufficient to double global crop production by 2050. PLoS One 8 (6), e66428. Reboud, X., 2003. Effect of a gap on gene flow between otherwise adjacent transgenic Brassica napus crops. Theor. Appl. Genet. 106, 1048–1058. Reich, P.B., 2009. Elevated CO2 reduces losses of plant diversity caused by nitrogen deposition. Science 326, 1399–1402. Reich, P.B., Buschena, C., Tjoelker, M.G., Wrage, K., Knops, J., Tilman, D., Machado, J.L., 2003. Variation in growth rate and ecophysiology among 34 grassland and savanna species under contrasting N supply: a test of functional group differences. New Phytol. 157, 617–631. Reif, J.C., Zhang, P., Dreisigacker, S., Warburton, M.L., Van Ginkel, M., Hoisington, D., Bohn, M., Melchinger, A.E., 2005. Wheat genetic diversity trends during domestication and breeding. Theor. Appl. Genet. 110, 859–864. Rhymer, J., Simberloff, D., 1996. Extinction by hybridization and introgression. Annu. Rev. Ecol. Syst. 27, 83–109. Richards, R.A., 1992. Increasing salinity tolerance of grain crops: is it worthwhile? Plant Soil 146, 89–98.

432

Nicolas Loeuille et al.

Richards, R.A., 2000. Selectable traits to increase crop photosynthesis and yield of grain crops. J. Exp. Bot. 51, 447–458. Rieseberg, L.H., Kim, M.J., Seiler, G.J., 1999. Introgression between the cultivated sunflower and a sympatric wild relative, Helianthus petiolaris (Asteraceae). Int. J. Plant Sci. 160, 102–108. Ripple, W.J., Betscha, R.L., 2012. Large predators limit herbivore densities in northern forest ecosystems. Eur. J. Wildlife Res. 58, 733–742. Robinson, R.A., Sutherland, W.J., 2002. Post-war changes in arable farming and biodiversity in Great Britain. J. Appl. Ecol. 39, 157–176. Root, R.B., 1973. Organization of a plant-arthropod association in simple and diverse habitats: the fauna of collards (Brassica Oleracea). Ecol. Monogr. 43, 95–124. Rossberg, A.G., Ishii, R., Amemiya, T., Itoh, K., 2008. The top-down mechanism for body mass-abundance scaling. Ecology 89, 567–580. Roughgarden, J., 1972. Evolution of niche width. Am. Nat. 106, 683–718. Rueffler, C., Egas, M., Metz, J.A.J., 2006a. Evolutionary predictions should be based on individual-level traits. Am. Nat. 168, E148–E162. Rueffler, C., Van Dooren, T.J.M., Metz, J.A.J., 2006b. The evolution of resource specialization through frequency-dependent and frequency-independent mechanisms. Am. Nat. 167, 81–93. Rueffler, C., Van Dooren, T.J.M., Metz, J.A.J., 2007. The interplay between behavior and morphology in the evolutionary dynamics of resource specialization. Am. Nat. 169, E34–E52. Ryall, K.L., Fahrig, L., 2006. Response of predators to loss and fragmentation of prey habitat: a review of theory. Ecology 87, 1086–1093. Sadras, V.O., 2007. Evolutionary aspects of the trade-off between seed size and number in crops. Field Crops Res. 100, 125–138. Samu, F., Szinetar, C., 2002. On the nature of agrobiont spiders. J. Arachnol. 30, 389–402. Schafer, M., Ross, A., Londo, J., Burdick, C.A., Lee, E.H., Travers, S.E., Van de Water, P.K., Sagers, C.L., 2011. The establishment of genetically engineered canola populations in the US. PLoS One 6, e25736. Schmitz, O.J., Krivan, V., Ovadia, O., 2004. Trophic cascades: the primacy of trait-mediated indirect interactions. Ecol. Lett. 7, 153–163. Shahzad, T., Chenu, C., Repinc¸ay, C., Mougin, C., Ollier, J.L., Fontaine, S., 2012. Plant clipping decelerates the mineralization of recalcitrant soil organic matter under multiple grassland species. Soil Biol. Biochem. 51, 73–80. Simberloff, D., 1988. The contribution of population and community biology to conservation science. Annu. Rev. Ecol. Syst. 19, 473–511. Singer, M.S., Stireman, J.O., 2005. The tri-trophic niche concept and adaptive radiation of phytophagous insects. Ecol. Lett. 8, 1247–1255. Slatkin, M., 1985. Rare alleles as indicators of gene flow. Am. Nat. 39, 53–65. Smith, J., Pearce, B.D., Wolfe, M.S., 2012. A European perspective for developing modern multifunctional agroforestry systems for sustainable intensification. Renew. Agr. Food Syst. 27, 323–332. Snow, A.A., 2012. Illegal gene flow from transgenic creeping bentgrass: the saga continues. Mol. Ecol. 21, 4663–4664. Sork, V.L., Nason, J., Campbell, D.R., Fernandez, J.F., 1999. Landscape approaches to historical and contemporary gene flow in plants. Trends Ecol. Evol. 14, 219–224. Soule´, M.E., 1976. Ayala, F.J. (Ed.), Molecular Evolution. Sinauer Associates, Sunderland, MA, pp. 60–77. Stockwell, C.A., Hendry, A.P., Kinnison, M.T., 2003. Contemporary evolution meets conservation biology. Trends Ecol. Evol. 18, 94–101.

Eco-Evolutionary Dynamics in Agroecosystems

433

Strauss, S.Y., Rudgers, J.A., Lau, J.A., Irwin, R.E., 2002. Direct and ecological costs of resistance to herbivory. Trends Ecol. Evol. 17, 278–285. Subbarao, G.V., Ito, O., Sahrawat, K.L., Berry, W.L., Nakahara, K., Ishikawa, T., Watanabed, T., Suenagae, K., Rondonf, M., Raof, I.M., 2006. Scope and strategies for regulation of nitrification in agricultural systems-challenges and opportunities. Crit. Rev. Plant Sci. 25, 303–335. Swenson, W., Wilson, D.S., Elias, R., 2000. Artificial ecosystem selection. Proc. Natl. Acad. Sci. U. S. A. 97, 9110–9114. Symondson, W.O.C., Sunderland, K.D., Greenstone, M.H., 2002. Can generalist predators be effective biocontrol agents? Annu. Rev. Entomol. 47, 561–594. Tabashnik, B.E., Gould, F., Carrie`re, Y., 2004. Delaying evolution of insect resistance to transgenic crops by decreasing dominance and heritability. J. Evol. Biol. 17, 904–908. The´bault, E´., Fontaine, C., 2010. Stability of ecological communities and the architecture of mutualistic and trophic networks. Science 329, 853–856. Thompson, J.N., 1999. Specific hypotheses on the geographic mosaic of coevolution. Am. Nat. 153, S1–S14. Thompson, J.N., 2005. The geographic mosaic of coevolution. University of Chicago Press, Chicago, IL. Thrall, P.H., 2013. Darwinian agriculture: how understanding evolution can improve agriculture R. Ford Denison 2012. Evol. Appl. 6, 408–410, Princeton University Press. Thrall, P.H., Burdon, J.J., 2003. Evolution of virulence in a host-pathogen metapopulation. Science 299, 1735–1737. Thrall, P.H., Hochberg, M.E., Burdon, J.J., Bever, J.D., 2007. Coevolution of symbiotic mutualists and parasites in a community context. Trends Ecol. Evol. 22, 120–126. Thrall, P.H., Oakeshott, J.G., Fitt, G., Southerton, S., Burdon, J.J., Sheppard, A., Russell, R.J., Zalucki, M., Heino, M., Denison, R.F., 2011. Evolution in agriculture: the application of evolutionary approaches to the management of biotic interactions in agro-ecosystems. Evol. Appl. 4, 200–215. Tilman, D., 1982. Resource Competition and Community Structure. Princeton University Press, Princeton, NJ. Tilman, D., 1999. Global environmental impacts of agricultural expansion: the need for sustainable and efficient practices. Proc. Natl. Acad. Sci. U. S. A. 96, 5995–6000. Tilman, D., Balzer, C., Hill, J., Befort, B.L., 2011. Global food demand and the sustainable intensification of agriculture. Proc. Natl. Acad. Sci. 108 (50), 20260–20264. Tooker, J.F., Frank, S.D., 2012. Genotypically diverse cultivar mixtures for insect pest management and increased crop yields. J. Appl. Ecol. 49, 974–985. Tscharntke, T., Klein, A.M., Kruess, A., Steffan-Dewenter, I., Thies, C., 2005. Landscape perspectives on agricultural intensification and biodiversity-ecosystem service management. Ecol. Lett. 8, 857–874. Turner, M.G., 1989. Landscape ecology—the effect of pattern on process. Annu. Rev. Ecol. Syst. 20, 171–197. Tyutyunov, Y., Zhadanovskaya, E., Bourguet, D., Arditi, R., 2008. Landscape refuges delay resistance of the European Corn Borer to Bt-Maize: a demo-genetic dynamic model. Theor. Popul. Biol. 74 (1), 138–146. http://dx.doi.org/10.1016/j.tpb.2008.05.005. Urban, M.C., Leibold, M.A., Amarasekare, P., De Meester, L., Gomulkiewicz, R., Hochberg, M.E., Klausmeier, C.A., Loeuille, N., de Mazancourt, C., Norberg, J., Pantel, J.H., Strauss, S.Y., Vellend, M., Wade, M.J., 2008. The evolutionary ecology of metacommunities. Trends Ecol. Evol. 23, 311–317. Vacher, C., Bourguet, D., Rousset, F., Chevillon, C., Hochberg, M.E., 2003. Modelling the spatial configuration of refuges for a sustainable control of pests: a case study of Bt Cotton. J. Evol. Biol. 16, 378–387.

434

Nicolas Loeuille et al.

Van Bueren, L.E.T., sterga˚rd, H., Goldringer, I., Scholten, O., 2008. Plant breeding for organic and sustainable, low-input agriculture: dealing with genotype-environment interactions. Euphytica 163, 321–322. Van Zandt, P.A., Agrawal, A.A., 2004. Community-wide impacts of herbivore-induced plant responses in Milkweed (Asclepias syriaca). Ecology 85, 2616–2629. Van Valen, L., 1965. Morphological variation and width of ecological niche. Am. Nat. 99, 377–390. Vanloqueren, G., Baret, P.V., 2009. How agricultural research system shape a technological regime that develops genetic engineering but locks out agroecological innovations. Res. Policy 38, 971–983. Vasseur, C., Joannon, A., Aviron, S., Burel, F., Meynard, J.-M., Baudry, J., 2013. The cropping systems mosaic: how does the hidden heterogeneity of agricultural landscapes drive arthropod populations? Agric. Ecosyst. Environ. 166, 3–14. Vellend, M., Geber, M.A., 2005. Connections between species diversity and genetic diversity. Ecol. Lett. 8, 767–781. Verbruggen, E., Kiers, T.E., 2010. Evolutionary ecology of mycorrhizal functional diversity in agricultural systems. Evol. Appl. 3, 547–560. Vigouroux, Y., Barnaud, A., Scarcelli, N., Thuillet, A.C., 2011. Biodiversity, evolution and adaptation of cultivated crops. C. R. Biol. 334, 450–457. Vincent, T.L.S., Vincent, T.L., 1996. Using the ESS maximum principle to explore rootshoot allocation, competition and coexistence. J. Theor. Biol. 180, 111–120. Vitousek, P.M., Aber, J.D., Howarth, R.W., Likens, G.E., Matson, P.A., Schindler, D.W., Schlesinger, W.H., Tilman, D.G., 1997. Human alteration of the global nitrogen cycle: sources and consequences. Ecol. Appl. 7, 737–750. Vitousek, P., Howarth, R., 1991. Nitrogen limitation on land and in the sea: how can it occur? Biogeochemistry 13, 87–115. Warwick, S.I., Simard, M.J., Legere, A., Beckie, H.J., Braun, L., Zhu, B., Mason, P., SeguinSwartz, G., Stewart, C.N., 2003. Hybridization between transgenic Brassica napus L. and its wild relatives: Brassica rapa L., Raphanus raphanistrum L., Sinapis arvensis L., and Erucastrum gallicum (Willd.) OE Schulz. Theor. Appl. Genet. 107, 528–539. Weiner, J., 2004. Ecology—the science of agriculture in the 21st century. J. Agric. Sci. 141 (3–4), 371. http://dx.doi.org/10.1017/S0021859603003605. Westphal, C., Steffan-Dewenter, I., Tscharntke, T., 2003. Mass flowering crops enhance pollinator densities at a landscape scale. Ecol. Lett. 6, 961–965. Whalon, M.E., Mota-Sanchez, D., Holligworth, R.M., 2008. Global Pesticide Resistance in Arthropods. Oxford University Press, Oxford, UK. White, T.C.R., 2005. Why Does the World Stay Green? Nutrition and Survival of Plant Eaters. CSIRO Publishing, Collingwood, Australia. Whitham, T.G., Young, W.P., Martinsen, G.D., Gehring, C.A., Schweitzer, J.A., Shuster, S.M., Wimp, G.M., Fischer, D.G., Bailey, J.K., Lindroth, R.L., Woolbright, S., Kuske, C.R., 2003. Community and ecosystem genetics: a consequence of the extended phenotype. Ecology 84, 559–573. Whitton, J., Wolf, D.E., Arias, D.M., Snow, A.A., Rieseberg, L.H., 1997. The persistence of cultivar alleles in wild populations of sunflowers five generations after hybridization. Theor. Appl. Genet. 95, 33–40. Witting, L., 2002. Evolutionary dynamics of exploited populations selected by density dependent competitive interactions. Ecol. Model. 157, 51–68. Witcombe, J.R., Hollington, P.A., Howarth, C.J., Reader, S., Steele, K.A., 2008. Breeding for abiotic stresses for sustainable agriculture. Philos. Trans. R. Soc. B 363, 703–716. Wolfe, M.S., Baresel, J.P., Desclaux, D., Goldringer, I., Hoad, S., Kovacs, G., Lo¨schenberger, F., Miedaner, T., sterga˚rd, H., van Bueren, E.T.L., 2008. Developments in breeding cereals for organic agriculture. Euphytica 163, 323–346.

Eco-Evolutionary Dynamics in Agroecosystems

435

Wootton, J.T., 1994. The nature and consequences of indirect effects in ecological communities. Annu. Rev. Ecol. Syst. 25, 443–466. Xia, J., Wan, S., 2008. Global response patterns of terrestrial plant species to nitrogen addition. New Phytol. 179, 428–439. Xiao, Y., Wang, Q., Erb, M., Turlings, T.C.J., Ge, L., Hu, L., Li, J., Han, X., Zhang, T., Lu, J., Zhang, G., Lou, Y., 2012. Specific herbivore-induced volatiles defend plants and determine insect community composition in the field. Ecol. Lett. 15, 1130–1139. Yachi, S., Loreau, M., 1999. Biodiversity and ecosystem productivity in a fluctuating environment: the insurance hypothesis. Proc. Natl. Acad. Sci. U. S. A. 96, 1463–1468. Yodzis, P., 2000. Diffuse effects in food-webs. Ecology 81, 261–266. Young, A., Boyle, T., Brown, T., 1996. The population genetic consequences of habitat fragmentation for plants. Trends Ecol. Evol. 11, 413–418. Young, I.M., Crawford, J.W., 2004. Interactions and self-organization in the soil-microbe complex. Science 304, 1634–1637. Zhang, D.Y., Sun, G.J., Jiang, X.H., 1999. Donald’s ideotype and growth redundancy: a game theoretical analysis. Field Crops Res. 61, 179–187. Zhang, Q., 2007. Strategies for developing Green Super Rice. Proc. Natl. Acad. Sci. U. S. A. 104, 16402–16409. Zhu, Y.-G., Cavagnaro, T.R., Smith, S.E., Dickson, S., 2001. Backseat driving? Accessing phosphate beyond the rhizosphere-depletion zone. Trends Plant Sci. 6, 194–195. Zhu, Y., Chen, H., Fan, J., Wang, Y., Li, Y., Chen, J., Fan, J.X., Yang, S., Hu, L., Leung, H., Mew, T.W., Teng, P., Wang, Z., Mundt, C.C., 2000. Genetic diversity and disease control in rice. Nature 406, 718–722.