Plant architecture, its diversity and manipulation in agronomic conditions, in relation with pest and pathogen attacks
- First Online:
- Cite this article as:
- Costes, E., Lauri, P.E., Simon, S. et al. Eur J Plant Pathol (2013) 135: 455. doi:10.1007/s10658-012-0158-3
- 800 Views
Plant architectural traits have been reported to impact pest and disease, i.e., attackers, incidence on several crops and to potentially provide alternative, although partial, solutions to limit chemical applications. In this paper, we introduce the major concepts of plant architecture analysis that can be used for investigating plant interactions with attacker development. We briefly review how primary growth, branching and reiteration allow the plant to develop its 3D structure which properties may allow it (or not) to escape or survive to attacks. Different scales are considered: (i) the organs, in which nature, shape and position may influence pest and pathogen attack and development; (ii) the individual plant form, especially the spatial distribution of leaves in space which determines the within-plant micro-climate and the shoot distribution, topological connections which influence the within-plant propagation of attackers; and (iii) the plant population, in which density and spatial arrangement affect the micro-climate gradients within the canopy and may lead to different risks of propagation from plant to plant. At the individual scale, we show how growth, branching and flowering traits combine to confer to every plant species an intrinsic architectural model. However, these traits vary quantitatively between genotypes within the species. In addition, we analyze how they can be modulated throughout plant ontogeny and by environmental conditions, here considered lato sensu, i.e. including climatic conditions and manipulations by humans. Examples from different plant species with various architectural types, in particular for wheat and apple, are provided to draw a comprehensive view of possible plant protection strategies which could benefit from plant architectural traits, their genetic variability as well as their plasticity to environmental conditions and agronomic manipulations. Associations between species and/or genotypes having different susceptibility and form could also open new solutions to improve the tolerance to pest and disease at whole population scale.
Plant architectural traits have been reported to impact pest and disease, hereafter referred to as attackers, incidence on several crops and to potentially provide alternative, although partial, solutions to limit chemical applications (Baccar et al. 2011). The impact of architectural traits spread through several spatial scales: (i) the organs, in which nature, surface and shape influence plant-attacker interactions at tissue level, (ii) the plant form and topology which influence within-plant propagation of attackers and part of the microclimate, and (iii) the plant population, in which density and spatial organization determine plant to plant attacker propagation, as well as the microclimate. The relative importance of traits and scales depends on the attacker and the plant species, especially its size, life span, architectural and morphological types (annuals, bi-annuals, perennials and/or herbs, bushes, liana, trees) and the changes in dimensions and types of organs that occur during its ontogeny.
Plant architectural analysis deals with the description and understanding of plant organization and structural changes over time (Hallé et al. 1978). Initially dedicated to tropical forest trees, architectural analysis has been successfully applied to a number of temperate species (e.g., Lauri and Térouanne 1995; Stecconi et al. 2010), including annuals (Moulia et al. 1999). This discipline analyses the repeated emergence of new organs and tissues from apical, axillary and secondary meristems, driven by both internal processes and external (environmental) conditions. Even though the repetitive and decentralized nature of plant organogenesis confers a large plasticity to plants, their architecture is highly organized throughout ontogeny and at different scales: the phytomers (including the leaf, its subtending internode and the attached axillary meristem) which appear in succession within axes, the relative positioning of laterals along main axes at successive branching orders, as well as the reiteration process (White 1979; Nozeran 1984; Watson et al. 1995).
Here, we briefly review the main processes involved in plant architecture construction (see Barthélémy and Caraglio 2007 for further details), focusing on their involvement in attacker development and dissemination. Indeed, plant architecture and attacker development are closely related, due to both dynamic processes and reciprocal effects: the plant constitutes the attacker habitat and/or provides it with resources, therefore constraining its life conditions and development; conversely, the attacker may affect plant growth and architecture through sap consumption, leaf area reduction, removal of apices, etc. When considering the impact of structural aspects on attacker development, architectural traits have been investigated through concepts such as plant sectoriality and heterogeneity (Marquis 1996; Orians and Jones 2001), habitat complexity (Lawton 1983; Langellotto and Denno 2004) and connectivity, i.e. direct contacts between individual plants (Skirvin and Fenlon 2003). We end this review by a brief outline on quantification and modelling of plant architecture in relation to plant-attacker complex interactions.
Elementary processes responsible for plant architecture construction
Primary growth results from cell division and elongation in apical meristems, at leafy shoot and root tips. At the shoot apical meristem (SAM), leaf primordia are initiated with an almost stable time interval, called the plastochron. Initiation takes place at particular positions around the central part of SAM, resulting in particular and almost stable azimuthal insertion of leaves around the shoot. The divergence angle between consecutive leaves is called phyllotaxy and may display different configurations depending on the species and axis type (e.g. spiral or decussate phyllotaxy). In dense canopies, such as annual crops or within-tree volumes, leaf orientation may significantly differ from initial leaf insertion, due to active reorientation during growth, e.g. for light interception.
While leaf primordia develop, the stem segments between successive leaves elongate giving birth to internodes and new meristematic tissues are constituted at the axil of leaf petioles. The time between the appearances of two successive leaves, is called the phyllochron and in most species is longer than the plastochron. Phyllochron and internode elongation, which defines the distance between successive leaves, are the two key components of shoot growth rate. At a longer time scale such as a growing season, the growth can be continuous or rhythmic, i.e. characterized by periods of primary growth cessation between periods of active growth. Primary growth characteristics (phyllotaxy, phyllochron, rhythmicity, elongation rate, determinate vs undeterminate) and their combination define the geometrical and topological distances between organs within an axis. They also contribute to the ability of a shoot to escape attackers which develop and disseminate from an initial infection/infestation site in the plant. Notably, the relative rates of leaf emission, leaf and internode extension on the one hand, and infestation/infection rate on the other hand make plant escape possible. This escape can be viewed as a race between plant growth and attacker dispersion for foliar attackers, or source-sink flow combined to a dilution for viruses (Roberts et al. 1997; Bodin-Ferri et al. 2002).
Because young and tender leaves are usually the most susceptible to attackers, the time of development and its final morphology influence the leaf susceptibility period to pathogens and appetence for insects, as exemplified by the ontogenic resistance to apple scab caused by Venturia inaequalis (MacHardy 1996). There is a large spectrum of leaf morphology, with different shapes, symmetry, composed of leaflets or not, venation, etc. (see Bell 1991). However, one of the main leaf characteristics for interaction with pathogens is certainly its surface structure. The presence and thickness of a cuticle may protect the leaves from pathogen penetration. By contrast, the presence of hairs, trichomes, domatia and any complex structure of the leaf area is likely to modify pest-natural enemy interactions and also increases alternative food resources such as pollen grains or fungal spores caught into hairs (Roda et al. 2003). Moreover, the physiological properties of leaves, for instance the sugar, water and nitrogen (N) contents, constitute a resource that attackers require for their own development (see Ney et al. 2012). In many plant species, leaf morphology and physiological properties change during plant ontogeny. Changes in leaf morphology referred to as heteroblasty (for a recent review see Zotz et al. 2011) increase the complexity of the relationships between plant and its attackers and affect the within-plant variation in leaf susceptibility to attackers over time.
Secondary growth results from division of cambial cells giving birth to secondary phloem outwards the stem and to secondary xylem and wood inwards (Lachaud et al. 1999). Secondary growth has a major role in the diameter growth of axes in Gymnosperms and Dicotyledonous plants. As a consequence, its dynamics is of major importance for determining shoot form and its evolution over time. Indeed, the progressive loading of shoots due to wood formation, weight of lateral organs such as leaves, fruits and branches, leads to bending or drooping of axes and contributes to the plant 3D geometry (Moulia and Fournier 2009). This leads to modifications of geometrical distances between organs during a season and of within-tree micro-climate, therefore impacting indirectly organ accessibility for plant-attackers.
Branching corresponds to the development of axillary shoots—issuing from axillary meristem activity—and is responsible for the high level of structural complexity of polyaxial plants compared to monoaxial plants. This process allows polyaxial plants to enlarge their space occupation, displaying a large range of topological and geometrical organization. A large variability of branching strategies has been observed in plants which involves different time schedules and/or locations for axillary meristem development.
Many investigations on branching process have focused on apical dominance and bud dormancy. Apical dominance is defined as the control exerted by the shoot apex on the outgrowth of axillary buds. It involves correlative mechanisms mediated by auxins, cytokinins and strigolactones (Cline 2000; Gomez-Roldan et al. 2008), and the nutritional status of the axillary buds (Champagnat 1989; Leyser 2009). Its morphological consequence is the inhibition of axillary buds during the growing season they are formed. In annual plants, apical dominance is always partial and combined to environmental effects. In perennials as well, some axillary buds can develop during the growing season in which they are formed, producing so-called sylleptic shoots (Champagnat 1954). In addition, in perennial plants and in climates characterized by a cold period, all buds enter into dormancy during autumn and winter. Except in the case of sylleptic development, axillary buds develop after having passed through these dormancy stages (paradormancy, endodormancy and ecodormancy; Lang et al. 1987), giving birth to so-called proleptic (or delayed) shoots. Their final development stage strongly depends on the bud position along the parent shoot: among the configurations that have been observed, the most common are acrotony and basitony.
Reiteration exists in a majority of plants, except in some gymnosperms for instance in Pinus genus. In annual monocotyledon plants, for instance in wheat, rice or sorghum, it constitutes the main branching process through tillering that leads to a more complex plant constituted by several equivalent axes, among which it may become difficult to distinguish between the main axis and tiller (Doust 2007). Tillering is a main component of cereal production and a key advantage to compensate damages from attackers. Indeed, whatever the process leading to new buds development, sequential branching or reiteration, the ability of plants to generate new axes and de-multiply their structure gives them a large adaptability to environmental conditions and to traumas (e.g. Gruntman and Novoplansky 2011). This ability is crucial to escape from herbivores (including pests but also mammals such as goats) and pathogen attacks or to compensate after damages by promoting the development of healthy parts. Branching and reiteration also contribute to the ability of plants to recover or even reinforce their defences after attacks. As both branching and reiteration affect the “walking” pathways of foraging pests in search of (new) resources within a branch or a tree, the branching complexity resulting from high branching orders decreases the colonization rate (Casas and Djemai 2002), and therefore decreases the within-branch or tree infestation by pests as recently shown for aphids (Dysaphis plantaginea) in apple trees (Simon et al. 2012). In forest trees, the structural complexity also alters the foraging patterns of herbivores and affects predation rates (Riihimäki et al. 2006). Indeed, all organisms permanently or temporarily present in the plant (e.g. plant attackers but also their antagonists) are affected by plant architecture that constrains resource location and pathways (Casas and Djemai 2002; Langellotto and Denno 2004).
Flowering transition is a key step in plant architectural development. The genetic and physiological mechanisms responsible for flowering transition in SAM have been widely studied in the plant model Arabidopsis thaliana. They involve endogenous factors such as hormonal balance and environmental factors, especially temperature and light (both photoperiod and light quality) (e.g. Mouradov et al. 2002; Boss et al. 2004). The number of nodes developed is also a possible architectural component of SAM flowering transition, through information accumulated about the distances of an apex from the roots (Sachs 1999). Although there are fundamental differences between annual and perennial plants in their lifespan, the underlying genetic control of flower induction and floral organ formation appears to be roughly conserved among them (Tan and Swain 2006).
Flowering occurrence strongly impacts on and correlates with vegetative development depending on the induction period and the within-plant location of the floral meristems. In plants with terminal flowering, the irreversible differentiation of the SAM into flowers or inflorescences provokes a determinate growth (notice that determinate growth can also result from SAM death or differentiation into a specialized organ (tendril, spine, etc.; see Barthélémy and Caraglio 2007). This is observed in A. thaliana and many crop species, in which the phyllochron and the period of flowering transition determine the number of phytomers developed on flowering axes (Slafer and Rawson 1995; Nanda et al. 1995). In polycarpic species, such as the apple tree, terminal flowering can be followed by a sympodial branching (Crabbé 1987). In plants with axillary flowering (that may have or not an indeterminate growth), flowering is part of the branching pattern: depending on whether there are several axillary, or a single, bud(s), and how many among them can flower at each node, different combinations and shoot structures have been observed (Guédon et al. 2001). This is especially true in the Rosaceae family in which species belonging to Prunus or Malus genus provide a range of flowering and branching patterns (Costes et al. 2006).
Flowers and fruits constitute choice organs for plant attackers. During flowering, many angiosperms have a remarkable reproductive strategy for attracting pollinators. However, pollinators are not sterile and airborne pathogens can also be introduced into nectar. Considering nectar is a rich medium in which microorganisms could grow, the low level of reported infections of the gynoecium, has been attributed to particular properties of this complex biological fluid containing attractive compounds such as flavonoids and proteins (Liu and Thornburg 2012), but also having the potential function of inhibiting microbial growth (Carter and Thornburg 2004). As previously discussed for leaf attractiveness and role in resource acquisition, many fruits and seeds are attractive and favour plant dispersion by animals, but they also are targets for attackers and thus include a number of proteins involved in plant defence to abiotic and biotic stresses (van Loon et al. 2006).
The shoot morphology, the amount of growth and branching, as well as the nature of developed organs (e.g. vegetative or reproductive) strongly depend on the plant development stage and ontogeny (Nozeran 1984). Even though there is still a debate on the progressive or abrupt nature of changes occurring during ontogeny (Poethig 1990), discrete temporal phases are frequently distinguished during development of both annual and perennial plants (Guédon et al. 2007). Whether annual or perennial, the ability of a plant to react to damages by promoting new organ development either by growth, branching or reiteration is expected to strongly depend on its developmental stage (e.g. Prioul et al. 2004).
In crop plants, ontogeny concerns many aspects of organs developed during a growth cycle. The morphological and cuticular characteristics of leaves (Sylvester et al. 2001; Dornbush et al. 2011), the internode length, the probability of branching and the number of phytomers of branches or tillers show marked gradients along the shoot, with an abrupt change between organs that grow in the juvenile phase compared to those growing in the mature phase. In maize, the sexual character of the apical inflorescence (male/female) depends on axis position (Moulia et al. 1999). These changes have been shown to correlate with the amount of growth regulators such as gibberellins that are present during organ development (Evans and Poethig 1995). Gradual changes are also related to changes in environmental and nutritional conditions, combining seasonal changes to those linked to canopy development. As a consequence, the plant plasticity (see below) is tightly related to plant ontogeny.
In perennials, ontogeny concerns complex patterns that develop over years: changes in the morphological characteristics of plant entities have been reported in a number of forest and fruit tree species, for instance the number of phytomers and laterals per growth unit (GU) or annual shoot (Costes et al. 2003; Renton et al. 2006), the relative proportion of leaf versus stem biomass (Lauri and Térouanne 1991; Sabatier and Barthélémy 1999; Suzuki 2002). Morphogenetic gradients also involve the occurrence and recurrence of flowering. The first flowering occurrence defines the end of the juvenile phase and the beginning of the mature phase (Hackett 1985). In apple trees, juvenile vegetative phase, adult vegetative and reproductive phases have been shown to follow one another through complex patterns of dependencies between vegetative and flowering growth units (Costes and Guédon 2012). These patterns result, at least partly, from the periodicity of fruiting occurrence within the tree structure, referred to as biennial bearing in many fruit trees (Monselise and Goldschmidt 1982). However, the pattern of variation in flowering occurrences is often more complex than a simple succession of “off” and “on” years, as recently described in the apple tree (Guitton et al. 2012). In an ecological or forest context, the variation in fruits and seeds production has been shown to follow a multi-annual pattern and to correlate with the variation in attacker populations. This has led to the mast depression assumption (Selas 1997; Kelly and Sork 2002), which considers that two selective factors favour the evolution of masting: increased pollination efficiency in wind-pollinated species, and satiation of seed feeders.
Diversity of plant architecture and architectural models
The observation of the aerial meristem activity in many species led Hallé and co-workers (Hallé et al. 1978) to propose a classification of plant architectures into architectural models which account for the diversity of all higher plants through a limited number of developmental patterns, defined at the whole plant and axis scales (for a review see Barthélémy and Caraglio 2007). The general classification of architectural models also provides a conceptual framework for analyzing plant architecture and has led to accurate observations of the development dynamics, at the whole tree scale. These investigations rely on the postulate that each species exhibits its own and genetically defined architecture. In other words, the five morphological criteria (growth, branching, morphological differentiation and orientation of axes, flowering location; Hallé et al. 1978) are intrinsic properties of each species and correspond to a genetic potential which may be modulated—but not totally modified—by environmental conditions.
The evidence of genetic variation of architectural traits within a given species has been highlighted by studies that have compared different cultivars (Lauri et al. 1995) and by the description of mutants for morphological traits (e.g., compact habit, Lapins 1974). Moreover, the genetic determinism of plant architectural traits has been demonstrated by studies on segregating populations that have been performed on plant models (Tisne et al. 2008), for both annuals (e.g. Prioul et al. 2004) and perennials (De Wit et al. 2002; Segura et al. 2008; Hu and Scorza 2009). The fact that architectural traits are inherited by descendants opens new avenues for selection programs with the objective to select plants having more appropriate geometrical and/or structural properties in order to tolerate damages (see Ando et al. 2007).
Root architecture and its impact on plant defense against attackers
Roots are crucial for water and nutrients supply to plants and for their anchorage into the soil. Root activity impacts both the aerial development of the plant and the physiochemical and biological status of the surrounding soil. In particular, the structure of the root system plays a major role in sink and storage functions, deposition and excretion of biochemical compounds and association with symbiotic organisms (Pagès 2002). Because of the absence of morphological markers along root axes that made it difficult to access to the root systems, their architectural analysis has been initiated at a later period than aerial parts (Atger and Edelin 1994). Among the numerous soil constraints that can strongly affect root development, water content has been the most often considered trait (Bruckler et al. 2004). In a recent review, Hodge et al. (2009) have illustrated the main genetic and environmental factors leading to root system development and functions, as well as their large plasticity depending on soil constraints. Recent studies and reviews have highlighted that roots play an active role in plant defence against attackers, by sensing the soil environment, through root-shoot signalling and by constituting a dynamic reserve for regrowth (see Erb et al. 2009 for a recent review). This role is tightly linked to the below and above ground interactions after attacks that are mediated by plants and occur between the members of the complex communities that are associated with a given plant species (van Dam and Heil 2011). These interactions can be of antagonistic, synergistic or neutral nature for one, several or all members. Root-derived plant toxins are key aspects of these above and below ground plant interactions, and a number of them have been described after plant attacks, such as nicotine and different alkaloids (Erb et al. 2009). Another aspect of above-below ground interactions corresponds to the root capability to constitute reserve organs for assimilates after plant attack, which has been proved in numerous species such as maize, ryegrass or poplar. Moreover, several examples have demonstrated that roots are not only passive receptors of storage products but can change their sink strength after leaf attack (Erb et al. 2009).
Even though the architectural traits are intrinsic characteristic of each particular species, whether annual or perennial, environmental conditions, including biotic, abiotic conditions and manipulations performed by humans, can modulate individual development (Sultan 2000). This generates a large variability in plant architecture, resulting from the plasticity of underlying processes: growth can stop earlier or later, branching can be reduced in number and its nature can change. Organ dimensions and form can be modulated accordingly.
Effect of environmental conditions
Environmental conditions, especially temperature, light, rainfall and their ongoing changes with global warming have the ability to modulate many developmental processes such as flowering (and the number of organs developed prior to this transition), branching, tropisms or organ dimensions, and phenology (Nicotra et al. 2010). Temperature is responsible for vernalisation on annual plant seeds, and for growth stop and bud dormancy in perennials. After seedling or bud burst, temperatures influence shoot primary growth rate. These effects have been widely described and have led to the use of thermal time as time unit for representing dynamics of plant or shoot development under different temperature conditions (Bonhomme 2000). The use of thermal time however assumes that the response to temperature is linear, which is acceptable only for a restricted range of temperature conditions. So the use of calendar time, together with non-linear temperature-response functions is a more general approach (e.g. Parent et al. 2010), which is particularly appropriate for modelling temperature gradients and/or encompassing different forms of life, such as plants and their attackers, with specific temperature response functions.
The role of light on morphogenesis is multifold and interactions between light and plant architecture are particularly complex because of mutual impacts. Light impacts on plant architecture take place through a number of pathways, including carbon and N availability as light drives carbon (via photosynthesis) and N acquisition (via the transpiration driven xylem flux of nitrate and N compounds). Conversely, within the canopy, the 3D position of plant organs modifies the light environment of their neighbours (Sinoquet et al. 1991). Moreover, light quantity and quality directly impact organ formation, in particular leaves where structure strongly differs between shaded and lighted conditions (Hanba et al. 2002; Bruschi et al. 2003). Thereafter, the main function of leaves, especially photosynthesis and transpiration are highly dependent on the intercepted light. These functions in turn impact the plant capability to maintain its growth and support the carbon and N costs of its structure maintenance. In addition, plants are able to sense the light qualitative change, especially the Red: Far Red (R: FR) ratio, and modulate their architecture accordingly to compete for light (Ballaré 1999; Casal et al. 2004). It has been shown that R:FR signal promoting competition between plants down-regulates plant defences against attackers (Izaguirre et al. 2006).
Effect of manipulations by humans
By contrast to environmental conditions that are only poorly controlled nor resulting from elaborated strategies, plant management by humans is performed with precise aims, in particular to optimize yield. Diverse techniques used to manipulate plants, in changing the spatial and the temporal arrangements of plant’s organs may also affect attackers’ dynamics. Irrigation and fertilization have obvious and well-known effects on growth with an increase of vegetative growth, for both stems and leaves, roughly proportional to N dose up to a given threshold. Fertilization and irrigation permit increasing crop productivity. Nitrogen and water are essential components required for cell division and expansion, and for photosynthesis. A common effect of both is to promote the vegetative growth of plants—increasing leaf and stem extension and shoot branching and thus creating the condition of a denser canopy, with a higher production, but also representing a higher amount of substrate for plant attackers. Moreover N and water are also elements directly needed by plant attackers. Fertilization may result in a higher N content of the leaves, especially when done in the last stages of plant development, i.e. when plasticity is reduced. Leaf richness in N is for instance closely related to the amount of spores produced by rust lesions (Robert et al. 2004). In general, fertilized crops are considered as more favourable to disease development. This is usually attributed to both higher leaf N content and higher leaf density that increase contacts between leaves and favour the persistence of water on the organ surface after rains. Indeed the presence of liquid water is required for the germination of a number of fungi and some of them, such as Septoria spp., also use it to propagate from leaf to leaf (Lovell et al. 1997). So besides the indirect effect through canopy development it is likely that irrigation water brought by aspersion may also directly favour disease development.
For annuals, sowing dates are chosen to adapt plant biological cycles to climatic conditions, for instance to obtain flowering and harvest before summer high temperatures, or for young seedlings to escape spring frosts. Changing sowing date also changes the conditions of the race between plants and their pathogens. For instance, in winter-wheat, early sowing is performed to strengthen stand establishment before winter. However it may also favour some diseases such as Septoria spp., because of the conjunction of wet autumnal conditions which favour fungus development while fungicides are poorly efficient at low temperatures with high vegetative crop development which favours the initial development of epidemics. Sowing date, especially in day-length sensitive species, also changes the phyllochron and the number of leaves per shoot. In winter-wheat, an early sowing promotes a larger number of vegetative phytomers but the number of extended internodes is unchanged and the rate of leaf emergence is lower (Dornbush et al. 2011). This favours the propagation of Septoria spp. from leaf to leaf during stem extension in spring. Finally early sowings are known to increase the risk of Septoria epidemics in winter wheat (reviewed by Baccar et al. 2011).
In perennials, beyond species and cultivar tree habit, tree training may strongly modify architectural traits. The fruit-tree is undoubtedly the perennial which is the most manipulated: dwarfing rootstocks are used to control the aerial part development and promote early entrance into production (Lockard and Schneider 1981), whereas pruning and training are widely practiced to optimize tree shape and regularity of production (Forshey et al. 1992). Indeed, tree training constitutes a powerful, but still under study, lever to modulate attackers’ development within the tree. However, research work mixing plant physiology and pathology or entomology and addressing interactions between plant structure and attackers is poorly developed (Price et al. 1980; see Simon et al. 2007 and Kührt et al. 2006 at a fine and coarse scale, respectively).
The direct removal of infected or infested vegetative and reproductive organs through heading or thinning cuts of parts of shoots.
The length of the growth period which affects the length of the period when susceptible organs are present in the tree. This mechanism is especially true in the case of ontogenic resistance of organs to infection/infestation.
More recently a direct effect of tree topology has been evidenced in the apple: aphid infestation, which significantly decreases when the degree of branching increases (Simon et al. 2012). In this latter case the hypothesis is that this architectural trait directly influences resource localization and access because of more complex within-plant pathways.
From individuals to populations and plot design
In addition to manipulations at the individual plant scale, a number of agro-technical choices are usually performed at the population and field scales. In particular, planting density is known to affect attackers’ development in both annual and perennial crops. As well documented for annuals and perennials, neighbouring plants cause dramatic phenotypic responses especially at the architectural level (Callaway et al. 2003). The dissemination rate of the attacker is related to the distance, accessibility and/or connectivity between resource organs, (i.e. direct contacts between individual plants) beside other physiological parameters also affected by plant density. The same parameters also affect natural enemies, as exemplified by Randlkofer et al. (2010): parasitoid behaviour differed in experimental vegetation (grass) structures differing in plant density, height and connectivity.
The cultivation in association (or mixed cropping) of different varieties or species is an ancient method permitting the reduction of pesticides and fertilizers, which was formerly used when inputs were not available or scarce. It is a well-established practice in subsistence agriculture, especially in tropical regions. Briefly, source of gain comes from complementarities and differences between species or varieties in their uptake and use of resources, their genetic resistances to attackers, the hosting of attackers’ enemies. An additional stability may come from compensatory growth of some components in the cultivation mixture, when conditions are unfavourable to some of them. Even though the difficulty to automatize the tasks required to grow and harvest plant associations has limited its use in industrialized cropping systems, such associations are regaining interest with the concern for plant production sustainability. According to Mundt (2002), practical difficulties linked to plant mixtures have been overestimated, whereas a better resistance to attackers is one proven advantage of mixed systems (Finckh et al. 2000; Mundt 2002; Vallavieille-Pope 2004). However, many other aspects are likely to be involved and interact for determining yield that may be either increased or depressed in comparison with standard agronomic systems. Analyzing results obtained with cereal mixtures, Kiær et al. (2009) have confirmed the overall trend towards an increase in yield, and have shown that association of varieties that differ in (i) yield production (ii) disease resistance and (iii) weed suppression were the major factors determining the gain. The review by Malézieux et al. (2009) also outlines the benefits of multi-crop systems to enhance overall productivity and ecosystem services such as pest suppression. For instance, intercropping tomatoes and lentils decreases tomato infestation by various pests (Saha et al. 2001). A main concern is thus to decide which traits of the cultivars or species to be grown must be associated to promote synergies and avoid unbalanced competition that would result in the suppression of one component by another one. The architecture of species, which highly regulates how they will share resources in space and time, is amongst the most important characteristics.
In perennials, the choice on plot design and associations are crucial because of their long term impact on production. In coffee trees, an epidemiologic survey on coffee berry disease (CBD), caused by Colletotrichum kahawae, highlighted that maintenance pruning, removal of mummified berries, and mixed cropping with shade plants are cultural practices which create environmental conditions that limit CBD development (Bedimo et al. 2007). In particular, coffee trees located under the shade of fruit trees were significantly less attacked than those located in full sunlight. Similarly, fruit tree orchards are heterogeneous habitats comprising tree rows, a grass layer in the alleys or as understorey vegetation, and sometimes the presence of companion plants. Such spatial heterogeneity affects how pests and pathogens localize their host-plant or reach a susceptible organ, respectively. It also increases alternative resources for most natural enemies, but constrains patterns of prey capture and parasitism, with positive or negative effects depending on the foraging ability of the natural enemy and the spatial scale (Gingras and Boivin 2002; Langellotto and Denno 2004). Besides, to increase the genetic diversity of current orchards generally planted with one clone, the mixture of various cultivars differing in pest and disease susceptibility but also differing in their temporal architectural management could help limit attackers’ development (Didelot et al. 2007). Last, because plant-mediated effects affect both attackers and natural enemies, such perspective deserves to be considered in a more general framework of orchard re-design, maximizing both bottom-up and top-down processes to control attackers.
Contribution of quantification and modelling approaches
From the 1980s, the development of computational tools and the progressive application of architectural concepts to agronomic plants have stressed the need for quantifying and simulating plant architectural development. Simulations have emerged after Borchert and Honda’s work (1985) and the L-Systems language emergence and development in the context of plant growth simulation (Lindenmayer 1968; Prusinkiewicz and Lindenmayer 1990). A large number of simulation systems have been developed so far, that have been classified in Godin et al. (2005) as descriptive models, either driven by meristem activity or top-down geometrical approach, and reactive models. This latter category also includes the most recent functional-structural plant models (FSPM, see Vos et al. 2009 for a recent definition). However, whatever the model category, accounting for observed plant organization, at different scales and over plant ontogeny, remains a real difficulty which has made necessary the development of quantitative and statistical analyses.
The proposal of a mathematical background, benefiting from the concepts of architectural analysis previously developed, has allowed the capture of plant organization into data bases. This consists in representing the multi-scale organization of plants into multi-scale tree graphs (MTG, Godin and Caraglio 1998) and encoding them by a generic but flexible method, designed by end-users for each plant and application. MTGs also allow attributing quantitative variables to plant entities. A particular attention has been dedicated to the integration in MTGs of 3D coordinates and angles that are collected on different plant entities, especially leaves, during plant 3D digitalization (Sinoquet et al. 1991; Godin et al. 1999). Tree graphs, sequences or variables can then be extracted from MTGs, to be further analysed with dedicated tools, such as those available in the VPlants module of OpenAlea platform (Pradal et al. 2008). Sequence analyses with Markovian models have revealed patterns, regularities and dependencies between adjacent entities across a range of applications such as the quantification of branching patterns in apple cultivars (Renton et al. 2006), with different rootstocks (Seleznyova et al. 2003), and for decomposing variance of annual shoot length into ontogenetic, climatic and individual components in forest trees (Guédon et al. 2007).
A rather similar methodological background has been applied to root systems where quantitative analysis has also explored different methods (Smit et al. 2000; Danjon and Reubens 2008). Among them, digitizing techniques and X-Ray tomography have been successfully used for collecting 3D root structures including geometry and topology, the second one having the advantage of being non-destructive. Specific variables for root systems have also been proposed to characterize root geometry, for instance the length of the top unbranched zone (Lecompte et al. 2001). The combination of methods for measuring and exploring the root structure with assumptions on root development rules has allowed modelling of root growth and architecture (Pagès et al. 2000).
Today, 3D plant modelling has realized quite a high level of achievement in the simulation of the dynamics of plant architecture using statistical approaches or empirical models fitted to field measurements. Such models can be used to investigate a range of processes in plant-environment interactions, several of them being keys for plant-pathogen interactions: phylloclimate (Chelle 2005), rain interception and splash (Bassette and Bussière 2008), pesticide deposition (Dorr et al. 2008). The building of mechanistic models has also resulted in a better understanding of the environmental plasticity of annual plants, e.g. for branching (Evers et al. 2011), N distribution (Berthelot et al. 2012), patterning of organ size (e.g. Louarn et al. 2010); and partly mechanistic 3D models have been used to test hypothesis on the underlying mechanisms. However, we are only in the very early steps of FSPM building that would integrate along the annual crop cycle the range of environmental regulations required to mechanistically simulate plants behaving as real ones (Fournier and Andrieu 1999; Kang et al. 2012; Evers et al. 2007). In perennial crops, stochastic models estimated from architectural databases have been integrated in FSPM for several fruit species (Costes et al. 2008; Lopez et al. 2008; Cieslak et al. 2011). These approaches, by mixing mechanistic and stochastic models, have the capability to simulate the spatial heterogeneity and variation over years that define the “maze” the attackers will face (as discussed in Casas and Djemai 2002). In these cases again, there are still tremendous efforts to deploy for collecting and analyzing large data sets and provide a comprehensive and quantified overview of the effects of different climatic or agronomic practices on tree development and plasticity. Also, at plot scale, models mixing spatial and temporal patterns of plant resistance and attackers deployment in different scenarios of spatial distribution and genetic resistance pyramiding versus separation have also recently emerged but without considering the 3D plant organization (Sapoukhina et al. 2009).
The use of 3D models of plant development interfaced with attackers is quite recent and is a scientific field under active development. Models are likely to permit the integration of different sources of complexity, to combine them and simulate plant development in different climatic or attack scenarios and compared the model outputs to observations (e.g. Skirvin 2004; Calonnec et al. 2008). However presently, the few 3D plant models that have been linked to pathogen dispersion or insect movement (see e.g. Hanan et al. 2002; Calonnec et al. 2008; Baccar et al. 2011) are structural models that simulate essentially the time course of 3D structure, with little functional aspects. Simulation with such models, in which architectural traits and/or climate scenarios vary have proven useful for investigating how the complex time-and-space-dynamics of crop structure modulate an epidemic, and identifying critical periods in the growth cycle or critical traits that favour pathogen propagation or promote disease escape. In the ECHAP project (Robert C. Pers. Comm.) 3D plants models are used with pathogen dispersion and pesticide deposition models to investigate how to optimize simultaneously plants architectural traits and pesticide application strategies. Indeed, trait per trait analysis is very difficult to carry out through experiments since even quasi-isogenic plants differ by multiple traits due to the feedbacks in plant functioning. On the other end, the power of model simulations remains hampered by the difficulty to predict the architectural plasticity in response to environmental conditions, or the variation in nutritive value of plant tissues and their effect on pathogen cycle. Thus the efforts mentioned above towards incorporating more functional aspects in plant models should result in a much wider range of applications of these approaches eventually permitting to anticipate the consequences of technical choices or environmental changes. In parallel, it seems promising to introduce in these approaches more detailed models of the pathogen cycle and its interaction with tissue properties (cuticle, C and N content) and pesticides.
In face of the complexity of genotype x environment x agro-technical practices interactions, the determinants of plant architectural plasticity appear poorly understood and still in a very qualitative way. Moreover, several plant organization levels and different time scales are involved in plant attacker interactions. Thus, developing new knowledge is highly desirable and would contribute to the emergence of innovative strategies to limit plant attacker infestation. Due to the complexity of processes, the number of scales to be considered, the interlacing of spatial and temporal levels, and the existence of long term and/or antagonist effects, integrative points of view based on modelling approaches are required. A tremendous domain of research is still to develop for representing plant and attacker interactions. In particular, deterministic models are missing that would enable us to predict the behavior of heterogeneous canopies and would be a guide for choosing or breeding the traits to be associated. A large avenue of research has thus just be opened and must be enlarged in the coming years, mixing multidisciplinary approaches for the conception of new modeling approaches and to cope with the complexity of 3D plant architecture and processes related to plant-attacker relationships.