Facilitation as a ubiquitous driver of biodiversity


  • Eliot J. B. McIntire,

    1. Canadian Forest Service, Natural Resources Canada, Victoria, BC, Canada
    2. Canada Research Chair – Conservation Biology and Modeling, Département des sciences du bois et de la forêt, Université Laval, Québec, QC, Canada
    Search for more papers by this author
  • Alex Fajardo

    Corresponding author
    1. Centro de Investigación en Ecosistemas de la Patagonia (CIEP) Conicyt–Regional R10C1003, Universidad Austral de Chile, Coyhaique, Chile
    Search for more papers by this author



Models describing the biotic drivers that create and maintain biological diversity within trophic levels have focused primarily on negative interactions (i.e. competition), leaving marginal room for positive interactions (i.e. facilitation). We show facilitation to be a ubiquitous driver of biodiversity by first noting that all species use resources and thus change the local biotic or abiotic conditions, altering the available multidimensional niches. This can cause a shift in local species composition, which can cause an increase in beta, and sometimes alpha, diversity. We show that these increases are ubiquitous across ecosystems. These positive effects on diversity occur via a broad host of disparate direct and indirect mechanisms. We identify and unify several of these facilitative mechanisms and discuss why it has been easy to underappreciate the importance of facilitation. We show that net positive effects have a long history of being considered ecologically or evolutionarily unstable, and we present recent evidence of its potential stability. Facilitation goes well beyond the common case of stress amelioration and it probably gains importance as community complexity increases. While biodiversity is, in part, created by species exploiting many niches, many niches are available to exploit only because species create them.

‘Simultaneous competition and beneficence [facilitation] can have major impacts on plant community structure’. Hunter & Aarssen (1988)

‘…community assembly mechanisms determining the nature and magnitude of facilitative effects on the composition of species and diversity of a community are not well understood’. Schöb et al. (2012)

I. Introduction

Interactions between organisms have long been considered a major driver in the structuring and organization of natural communities (Hairston et al., 1960; Tilman, 1982; Mittelbach, 2012). Specifically, interactions within trophic levels have been widely deemed to be a key element of these drivers through their effect on coexistence and biodiversity (Chesson, 2000; Silvertown, 2004). Kropotkin (1902), Clements (1936), and Allee et al. (1949) showed that communities comprised a combination of positive and negative interactions between species. As the paradigm within which the field of ecology primarily operates is the struggle for existence, negative interactions (e.g. competition) were repeatedly found to be most important for species diversity, save for neutrality (Hubbell, 2001), in spite of calls and evidence to the contrary (e.g. Hay et al., 2004). The role of mutualisms in biodiversity – those interactions that are traditionally identified as occurring between very divergent evolutionary lines, such as mitochondria in eukaryotic cells, pollinators and plants, corals, mycorrhizas, and lichens, or very different trophic levels, such as cleaner fish–client relationships – is well known and widely appreciated (see mutualism chapters in recent editions of ecology texts, e.g. Cain et al., 2011; Mittelbach, 2012). Consequently, the focus of this paper will be on the effects on biodiversity of positive interactions among same-trophic guilds, commonly referred to as facilitation, that have been dwarfed by the attention given to negative interactions (May, 1981; Gross, 2008).

The work of Hunter & Aarssen (1988) can be considered the first attempt to gather evidence on diverse types of positive plant interactions, at the time called beneficence relationships. However, it was not until the seminal work of Bertness & Callaway (1994) that facilitation was explicitly introduced as a biotic process that can have important consequences in ecological communities, particularly in stressful environments (i.e. the stress gradient hypothesis (SGH)). The SGH predicts that positive species interactions are more important in biologically and physically stressful habitats than in benign habitats (Bertness & Callaway, 1994). Thereupon it commenced a renaissance of experimental and observational studies focusing on the occurrence of facilitation and the testing of the SGH in a variety of apparently stressful systems, including water-stressed systems and alpine and salt marsh systems, which repeatedly supported the predictions of the SGH (Callaway, 2007; He et al., 2013). Facilitation has been defined as ‘an interaction in which the presence of one species alters the environment in a way that enhances growth, survival and reproduction of a second species' (Bronstein, 2009). Although this definition appears to synthesize well the original concept of facilitation and stressful conditions (cf. Bertness & Callaway, 1994), it clearly does not limit facilitation to stressful conditions (see Holmgren & Scheffer, 2010). As such, several rapidly increasing bodies of ecological study examining positive interactions and their diverse modes of action have begun to suggest that facilitation may play more than just an incidental role in structuring plant communities through the alleviation of stressful conditions. Several reviews on facilitation have been published (Callaway, 1995; Brooker et al., 2008; He et al., 2013) and new hypotheses have been proposed (Valiente-Banuet & Verdú, 2007; Kikvidze & Callaway, 2009; McIntire & Fajardo, 2011; Soliveres et al., 2011) that together have expanded our understanding of the implications that facilitation has for community ecology (Stachowicz, 2001; Bruno et al., 2003; Callaway, 2007; Brooker et al., 2008). Yet, in our opinion, these reviews and new hypotheses still underestimate the role of facilitation in community ecology because they have primarily focused on the effects of individuals in creating obvious structures and moderating abiotic stresses, which represent only a fraction of the mechanisms and ecosystems where facilitation actually occurs (cf. Callaway, 2007). That is, in spite of over 20 yr of intensive research, and some attempts to include positive plant interactions in ecological theory (e.g. Bruno et al., 2003; Michalet et al., 2006), within-trophic facilitation still appears to persist as a special case contributor to the maintenance of species diversity, rather than a ubiquitous driver of diversity at all scales. We believe that this persistence is, first, a consequence of the clear importance of antagonisms in simple experimental conditions and, secondly, a consequence of the disparate mechanisms by which positive interactions occur, dissociating them from one another. As a result, the lack of unity linking the diverse mechanisms creating positive within-trophic effects on biodiversity may be at fault.

1. Facilitation, a ubiquitous process driving biodiversity

Until now, species have been perceived as resource users and thus competitors. While this is true, they are also, more generally, condition modifiers (Fig. 1). They change the distribution of abiotic resources and change the biotic dynamics. Because all species' niches differ across a different high-dimensional resource space and can be constrained by different high-dimensional factors, the new, local conditions will change the relative success of species, and favour a different array of species. For example, if a species uses light, it creates lower light conditions for all immediate neighbours growing beneath and changes the competitive hierarchies of those species in the understory (e.g. shade-tolerants outcompete shade-intolerants); when a species uses nitrogen it locally reduces the nitrogen content and may change the carbon : nitrogen ratio, favouring different species; if a species ‘uses’ a generalist pollinator, it may attract more pollinators, encouraging pollen-limited species, and so on for all resources and conditions. The usual assumption is that the use of resources has a negative effect on immediate neighbours. For community dynamics, however, it is not the absolute sign of the effect, but the relative strength of that effect that matters. For example, stress- or shade-tolerant species may have reduced growth under stressful conditions or under a forest canopy, respectively, but their reduction is less than for stress- or shade-intolerant species, so new communities form (e.g. understory communities) from the local species pool. Thus, biodiversity can be increased by any species presence through the modification of the available niches as compared to the abiotic environment alone, at times increasing the alpha diversity and more generally increasing the beta diversity (Fig. 1).

Figure 1.

Ability of facilitative mechanisms to modify a site's conditions created by the environment and by competition, allowing species diversity to change. We depict the multidimensional environmental conditions at a site without biotic drivers (dashed circles) and modifications to them produced by biotic drivers, either competition alone (central green circle) or where facilitation and competition act together (outer green shapes). Small coloured circles represent the multidimensional niche requirements of a hypothetical sample of species, where colour is unique by species. The facilitator species is not depicted as it may or may not be at the same scale as the site (e.g. a tree modifies the site conditions below it and does not fully experience those modified conditions). At a given site, when competition alone is considered, available multidimensional conditions are a reduced subset of the environmental conditions (green circle is smaller than dashed circle). When facilitative mechanisms are added, local conditions will be shifted in a variety of ways depending on mechanisms (large arrows leading outwards from the large grey circle). For a species to occur, its niche requirements must be available within the site conditions. Depending on the magnitude of the facilitative changes, new species may appear, previously present species may be lost, or some species will be maintained as their niche requirements are present, lost or maintained (appearance, disappearance or maintenance of small coloured circles). To show that facilitative drivers have the ability to push the site conditions outside of the multidimensional space created by the abiotic environment alone, the green shapes may extend well outside of the dashed circles (e.g. novel habitat), depending on the strength of the particular facilitative mechanisms acting. The direction of the expansion of conditions is arbitrary. For simplicity, we omit biotic drivers other than competition and facilitation.

We define the facilitation effects on biodiversity as any increase in a diversity measure that results directly or indirectly from the modifications of biotic and abiotic conditions caused by any and all species' presence. We spend the remainder of the text showing many mechanisms by which these modifications occur and the vast diversity of species on the planet that results from them. By doing this, we unify a broad array of disparate mechanisms of net positive within-trophic interactions into the single ecological concept known as facilitation. Our proposed set of mechanisms is more inclusive than that used in previous reviews and, as a result, we present evidence that facilitation is actually a major driver of biodiversity across ecosystems. These mechanisms include widely known drivers such as the mitigation of abiotic stressors (e.g. SGH) and novel habitat creation, but also less widely appreciated ones, such as intransitive interactions, indirect interactions, being small compared to one's neighbours, heterogeneity creation and resource sharing. Indeed, many species modify conditions creating net positive effects even in productive or nonstressful locations.

We structure our paper in six sections. First, we highlight empirical, analytical and simulation studies that show that a wide range of direct and indirect mechanisms change the biotic or abiotic conditions and explicitly contribute to increase or maintain diversity and are thus facilitative, and we highlight the partial duality between competition and facilitation as a result of indirect effects. Secondly, we discuss briefly the importance of facilitation over evolutionary time and its evolutionary stability even under conditions that are considered unlikely from a strict understanding of competition. Thirdly, we discuss some reasons why this full appreciation of positive effects has been overlooked until now. Fourthly, we examine new efforts at understanding facilitation using plant functional traits. Fifthly, we discuss the predictability and testability of the different and disparate mechanisms of facilitation that drive biodiversity, with particular reference to the trait approach. Finally, we show why this understanding of facilitation can help in conservation, restoration and management. In this paper, we omit discussion of mutualisms because they are widely known and appreciated. Similarly, we restrict our discussion to sessile organisms; however, all organisms change the local conditions, and we show several examples in tables and figures of animal facilitation. Rather, we focus on the ways in which positive interactions affect topics traditionally dominated by competitive relationships and thus we constrain our discussion to same-trophic level net positive effects, consistent with how facilitation is generally defined (McIntire & Fajardo, 2011).

II. Facilitative mechanisms increasing diversity

All facilitative mechanisms act by benefactor species changing the biotic or abiotic conditions for facilitated species. We highlight several known mechanisms and note that our particular classification of the mechanisms may not be mutually exclusive; that is, amelioration of stressful conditions could be conceived by some as novel habitat creation. However, we present them to align with our historical understanding of facilitation. Furthermore, we update and subsume Hunter & Aarssen's (1988) nine types of, at the time, beneficiary mechanisms with new literature and new understanding of the mechanisms. Like competition (e.g. direct interference competition and indirect exploitative and apparent competition), it is clear that facilitation encompasses both direct and indirect mechanisms. Facilitation may be complex and its indirect effects may be difficult to tease apart experimentally (Callaway, 2007, p. 117). Yet, difficulty detecting such effects does not deny their existence; in many cases they can be well understood (Czárán et al., 2002; Brooker et al., 2008).

1. The direct mechanisms of facilitation

Positive interactions play a critical, but underappreciated, role in ecological communities by reducing physical or biotic stresses in existing habitats and by creating new habitats on which many species depend Stachowicz (2001)

We identify five broad mechanisms of direct facilitation as drivers of diversity, of which two have dominated the literature (Fig. 1). These all cause changes in the biotic or abiotic conditions, altering the available multidimensional niches, allowing for increased diversity (Hacker & Gaines, 1997; Michalet et al., 2006). The first two are stress amelioration and novel habitat creation (Stachowicz, 2001; Callaway, 2007). Three others, less commonly associated with facilitation, are creation of habitat complexity, access to resources, and service sharing. For space considerations, we elaborate somewhat on only habitat complexity and resource sharing because they have recent additions in the literature that provide new insights (Table 1; see citations therein for the others). We provide descriptions of types of service sharing (Table 1) and point to some literature on these. As these direct mechanisms are the best understood in the literature and, furthermore, it is likely that ecologists will continue to identify further mechanisms by which the biotic and abiotic changes to a community increase or maintain diversity, we limit the discussion here.

Table 1. The direct mechanisms of facilitation. Net positive effects of neighbouring species occur through changing the abiotic or biotic conditions, resulting in an alteration of site conditions for the facilitated species (Fig. 1)
MechanismExample mechanismsLiterature
Abiotic stress ameliorationSoil warming in cold or shading in hot conditions (e.g. stress gradient hypothesis; Fig. 2a)Bertness & Callaway (1994), Bruno et al. (2003), Brooker et al. (2008), Soliveres et al. (2011)
Novel ecosystems

Ecosystem engineers or ecosystem constructors (Fig. 2b)

Structural support (Fig. 2c)

Primary and secondary succession

Shade (e.g. forest understory)

Niche construction

Connell & Slatyer (1977), Chapin et al. (1994), Jones et al. (1994), Odling-Smee et al. (2003), Badano et al. (2006), Brooker et al. (2008), Gedan & Bertness (2010), Kylafis & Loreau (2011), Lowman & Schowalter (2012)
Habitat complexity (heterogeneity)

Spatial and temporal complexity created by the physical presence of species

Organism size and shape (e.g. larger species and species architecture)

Accretion of soil organic matter

Tilman (1982), Pugnaire et al. (2004), Aarssen et al. (2006), Bartels & Chen (2010), Schöb et al. (2012)
Service sharing

Pollination efficiency via interspecific masting synchrony (Fig. 3d)

Animal dispersal efficiency via interspecific masting synchrony

Predator satiation (e.g. mixed species herds or flocks, and interspecific fruit masting)

Social learning and transfer of information (e.g. about predation risks)

Host defence overriding

Janzen (1974), Sinclair (1985), Kelly & Sork (2002), Shibata et al. (2002), Danchin et al. (2004), Seppänen et al. (2007), Smith et al. (2011)
Access to resources

Keystone modifiers

Interspecific social learning and transfer of information about resources (e.g. habitat selection; Fig2d)

Hydraulic lift

Decomposition and nutrient cycling from neighbouring plants

Neighbouring species regeneration

Resource sharing via bet-hedging

Low light, per se, in understories

Mills et al. (1993), Levine (1999), Nilsson & Wardle (2005), Zou et al. (2005), Li et al. (2007), McIntire & Fajardo (2011), Tarroux & DesRochers (2011), Uitdehaag (2011)

First, in stress amelioration, the common situation is that individuals of species A ameliorate stressful conditions for individuals of species B, thus creating the niche requirements of the latter (Fig. 1; the green shape extends beyond environmental conditions). Consequently, the presence of species A (a nurse or foundational plant) has positive net effects for species B. Formalized in the SGH, this process creates the conditions required by the facilitated individuals (Bruno et al., 2003; Brooker et al., 2008; Gross, 2008); in stressful environments facilitators allow more stress-intolerant species to persist (Fig. 2a,b; e.g. Butterfield et al., 2013). Several studies suggest that the impact of the SGH is probably underestimated because we are only just realizing the implications of what constitutes a ‘relative’ stress gradient; stress is species-specific, it does not affect a community as a whole (Körner, 2003), resulting in stress-induced positive interactions being detected across wider than expected gradients (Holmgren & Scheffer, 2010; Soliveres et al., 2011). In the context of facilitation driving diversity, the benefactor species used some array of resources, but concurrently modified the conditions such that the limiting resources, for example moisture, were no longer limiting for its neighbours.

Figure 2.

Some direct mechanisms of facilitation. (a) Plants can ameliorate stressful conditions for other plants, framed under the stress gradient hypothesis. The photograph shows Acacia caven (Fabaceae), which forms savannas in semi-arid regions of central Chile; under its canopy other species, such as Leucocoryne ixioides (Liliaceae), Stellaria arvalis (Caryophyllaceae) and Olsynium junceum (Iridaceae), benefit from lower evapotranspiration. (b) Species create novel environments for other species, a process that sometimes involves nurse plants (also ‘ecosystem engineers’ or ‘ecosystem constructors’). The photograph shows the alpine cushion Azorella madreporica in the central Andes of Chile, which hosts many other species including (inset) Hordeum comosum, Poa spp. (Poaceae) and Loasa sigmoidea (Loasaceae), species that survive poorly outside of cushions. (c) Structural support allows numerous species of vines and epiphytes to exist. The photograph shows Hydrangea serratifolia (Hydrangeaceae), a common vine in the Valdivian rainforest of Chile that grows on huge trees, such as Aextoxicon punctatum (Aexotoxicaceae) and Eucryphia cordifolia (Cunoniaceae). (d) Interspecific social learning and transfer of information about predators is known to occur and increase survival. The photograph shows Eastern North America Carolina chickadee (Poecile carolinensis, in inset), which responds positively to and clearly distinguishes the different predator threats encoded in titmouse (Baeolophus bicolor) calls they hear. Sources: (a) C. Torres, (b) A. Fajardo, (c) A. Saldaña and (d) Wikimedia commons.

Secondly, under novel habitat creation, a number of species and species types alter the biotic or abiotic conditions in ways that are large enough that we chose to identify the new conditions as ‘novel habitat’ (Fig. 1; the green shape can extend well outside of environment). They have been called various names, including ecosystem constructors, ecosystem engineers and foundation species (Jones et al., 1994; Odling-Smee et al., 2003; Butterfield et al., 2013). Within this category, there are well-known examples, such as corals, mangrove trees, salt marsh foundation species and cushion plants (Stachowicz, 2001; Bruno et al., 2003; Badano et al., 2006), as well as other less appreciated examples, such as trees and giant kelps (e.g. Brooker et al., 2008). What was not fully appreciated in these studies is that the effect of habitat creation occurs at all scales, going well beyond the normal notions of ‘novel habitat creation’. For example, plant species create spatial heterogeneity as a result of their spatial structures and resource use creating more available niches for other species (Schöb et al., 2012). While these examples have enlarged our awareness of the ubiquity of stress amelioration and novel habitat creation, these are just two of several mechanisms of facilitation.

Thirdly, according to the resource heterogeneity hypothesis, increasing habitat complexity for a given area increases the number of species that can coexist, resulting in higher species diversity (Tilman, 1982; Mittelbach et al., 2001; Fig. 1; a shift in the green shape, which can be quite complex). Often omitted, the source of the heterogeneity is commonly other species, which are therefore providing facilitative effects. Schöb et al. (2012) demonstrated that a key facilitative mechanism in their system was the creation of heterogeneity, per se, distinct from niche space expansions and shifts. Similarly, nurse plants can increase the variability of available niches (Soliveres et al., 2011). The mere presence of relatively large species, for example, can allow a higher diversity of physically small species as a result of heterogeneity creation by large plants (Aarssen et al., 2006). Furthermore, the creation of resource or sheltering heterogeneity may lead to an increase in niche segregation and competition intransitivity (see the section on ‘The indirect mechanisms of facilitation’), which will ultimately result in an increase of species richness (Tielbörger & Kadmon, 2000; Soliveres et al., 2011).

Fourthly, individuals can also increase access to resources for other individuals via several mechanisms (Table 1; Fig. 1; this may cause the green shape to extend outside of environment). The mechanism of resource sharing, which appears to contradict expectations based on competition, can occur directly (McIntire & Fajardo, 2011) or indirectly via mycorrhizas, for example (Simard et al., 1997). McIntire & Fajardo (2011) showed that otherwise competing trees (i.e. close neighbours) appeared to share resources through merging of woody tissues, with the result that merged stems had higher survival and growth. Another similar example involves the transfer of phosphorus from one plant to another via mycorrhizal hyphae (Newman & Ritz, 1986; Simard et al., 1997) or raising the water table allowing access to deep water (Zou et al., 2005). While our study did not explain how this could be evolutionarily stable, Uitdehaag (2011) demonstrated mathematically that resource sharing among competitors would occur as part of a bet-hedging strategy, particularly when conditions experienced by the competitors are negatively correlated. The importance, therefore, of resource sharing would depend on the scale of heterogeneity of resources. Overall, however, our understanding of this is weak and there is much room for theoretical advances here.

2. The indirect mechanisms of facilitation

If ‘webs’, ‘hierarchies’, or ‘loops’ of strong competitive effects within communities can elicit strong positive indirect effects, then plant species may be quite interdependent even though all pairwise interactions are competitive Callaway (2007)

Net positive effects can occur via compounding of multiple negative effects, that is, indirect effects or ‘the enemy of my enemy is my friend’ (Table 2; Figs 1, 3). The emergence of these net positive effects through multiple negative proximate relationships has been, however, largely ignored in competition and facilitation research (but see Tielbörger & Kadmon, 2000; Saccone et al., 2010; Soliveres et al., 2011). As a special case of indirect effects where the pairwise competition is asymmetric, intransitive competition – colloquially known as the ‘rock-paper-scissors’ model – and intransitive networks have also been recognized as having a major role in species coexistence (Levine, 1999; Czárán et al., 2002; Laird & Schamp, 2006; Allesina & Levine, 2011). Competitive intransitiveness is essentially a horizontal analogue of trophic cascades (Hairston et al., 1960) that can be more complex because the networks can be infinitely long through feedbacks on the self. It can allow individuals within a species to help each other (Thorpe et al., 2009), can promote coexistence (Laird & Schamp, 2009) and can prevent extinctions (Verdú & Valiente-Banuet, 2008), and the facilitation resulting from the intransitivities may be the most likely outcome in complex systems (Allesina & Levine, 2011; E. J. B. McIntire, unpublished). An important intransitive competitive driver is the greater relative strength of intraspecific competition compared with interspecific competition, well known as one of the crucial premises for species coexistence stabilizing mechanisms that maintain species diversity (Chesson, 2000; Levine & HilleRisLambers, 2009). New work is showing that species diversity may be increased via intransitive competition because of multiple limiting factors (Allesina & Levine, 2011). This mechanism effectively enhances and stabilizes species richness. While intransitive competition via the relative strength of intraspecific over interspecific competition can readily create positive effects and lead to the inclusion of more species in the community, the link to facilitation has not been considered (Stachowicz, 2001; but see Gross, 2008) or has been explicitly denied (p. 178 in Mittelbach, 2012).

Table 2. The intransitive processes of facilitation can create net positive effects that occur via compounding of multiple negative effects
Competitive intransitiveness: the rock-paper-scissors modelIt is known that in both animals and plants competition networks with pairwise interactions between species are not transitive. This is like the ‘rock-paper-scissor’ game applied to competition, with no universal scale of competitiveness. Of particular interest, competitively weaker species can persist because of this phenomenon (Fig. 3a,b), effectively reversing predictions of competitive exclusion niche modelsHairston et al. (1960), Levine (1999), Frean & Abraham (2001), Czárán et al. (2002), Verdú & Valiente-Banuet (2008), Laird & Schamp (2009), Thorpe et al. (2009), Allesina & Levine (2011), Levi & Wilmers (2012)
Intraspecific competitionCompetitively dominant species are known to limit their own abundances more than those of competitively inferior species, thus stabilizing coexistence; an individual of a given species competes with its spatially immediate conspecifics (high niche overlap), potentially reducing the growth of itself, encouraging the existence of immediate other-species neighbours (low niche overlap). This process is distinct from niche separation because it is the intraspecific competition that drives the coexistence, allowing species of relatively close niches to coexist (Fig. 3c)Chesson (2000), Levine & HilleRisLambers (2009)
Indirect effectsWhen competition occurs and relationships are not fixed (i.e. species A may outcompete species B under some conditions, but not others), then competition will propagate, creating indirect positive effects that are complex to evaluate. However, these indirect effects may be strong where species richness is highDethier & Duggins (1984), Vandermeer et al. (1985), Thompson et al. (1991), Levine (1999), E.J.B. McIntire, unpublished
Figure 3.

Examples of species and ecosystems where positive effects are likely common. (a) Tropical forests have several facilitative processes contributing to high biodiversity. First, trees are ‘ecosystem constructors’ creating structures for species types including vines and epiphytes. Trees also create shade allowing the proliferation of shade-tolerant species, a massive collection of species that exists as a result of the net positive balance between the direct negative effects of canopy shading and the indirect positive effects of canopy shading on non-shade-tolerants. Furthermore, these forests are probably comprised of large intransitive competition networks which result in net positive effects on species and biodiversity. The picture shows a tropical forest in Martinique, in the Caribbean Sea. (b) Shade-tolerant Myrceugenia planipes (Myrtaceae) growing under the dense canopy of the Valdivian rainforest of southern Chile. (c) Intraspecific competition occurring in a meadow in the French Alps (Vanoise massif, Pralognan la Vanoise) probably enhances species richness, allowing many species, including Eryngium alpinum (Apiaceae), Campanula barbata (Campanulaceae) and Festuca paniculata (Poaceae), to coexist. (d) Masting species sometimes have synchronized flowering across species (community masting), conveying mutual benefits via predator satiation; the picture shows a mixed temperate deciduous Japanese forest of Fagus crenata (small seedling in the right inset) and Castanea crenata (fruits in the inset). Sources: (a, d) Wikimedia commons, (b) A. Fajardo, and (c) I. Till-Bottraud.

Indirect facilitation has been described for decades (Dethier & Duggins, 1984; Vandermeer et al., 1985; Levine, 1999), but its importance has been unclear because it has been difficult to identify its effects empirically. Also, there has been no null expectation in the sign of a multitude of direct competitive effects. In a recent study (E. J. B. McIntire, unpublished), it was shown that indirect facilitation via the propagation of negative competitive effects is indeed the null expectation of many net negative interactions. Distinct from intransitive networks, it was shown that this positive net effect also occurred even where competitive relationships were not fixed (i.e. could change according to context) and even where species were indistinguishable (E. J. B. McIntire, unpublished), a situation that may exist according to the neutral theory of biodiversity (Hubbell, 2001). In these cases, it was sufficient that variation exist among individuals for diversity to be increased, consistent with Clark (2010), as a consequence of stochasticity. In this case, however, the mechanism occurred via the propagation of the negative effects. Where species niche requirements differed, richness was increased as a result of the well-known ratio of intraspecific to interspecific competition (Chesson, 2000). Even here, however, individual variation further enhanced the species richness in those conditions, particularly with a diverse species pool (E. J. B. McIntire, unpublished).

Indirect and intransitive facilitation undermines the identity of facilitation as a strict opposite to competition. Indeed, if competition networks result in net positive effects as a result of competitive mechanisms, such as resource exploitation, then they are both competitive and facilitative. It is unclear, at the moment, whether indirect competition or intransitivity can change the site conditions outside of those of the environment (Fig. 1; the green shape does not extend beyond environment) or whether that would only occur via the compounding of indirect and direct mechanisms (e.g. access to resources combined with indirect competition). Recent work on the relative prominence of indirect competition that increases diversity suggests that positive effects are the most likely outcome of intransitive competitive networks (Allesina & Levine, 2011). The experimental and empirical support for the importance of indirect and intransitive effects is growing rapidly, but it is nevertheless proving a challenge. As with other efforts in science, difficulty in measuring does not undermine the importance of the process; our challenge is to design new tests of these hypotheses that are capable of detecting their importance.

III. Facilitation as an evolutionary driver in proximate interactions

Positive interactions are important in evolutionary time, as demonstrated by the existence of widespread species groups and ecosystems in which facilitation is important (Table 3). Lortie et al. (2004) synthesized the findings of numerous studies showing that interdependence in communities, some adaptively based, is ubiquitous, introducing the integrated community concept. Furthermore, invasional meltdown, where pre-existing positive relationships are recreated in a new ecosystem, sometimes increasing or stabilizing diversity, attests to the importance of facilitation over large scales (Adams et al., 2003; Simberloff, 2006). A recent study showed that facilitative relationships among ‘competitors’ can be selected for on very short time and spatial scales in arid environments as climate starts to shift (Kéfi et al., 2008). In this case, the selection occurred primarily in species that only had short distance dispersal capabilities. In another study, stem merging was shown to occur readily between spatial neighbours of the same species (McIntire & Fajardo, 2011), a phenomenon that caused the individuals to avoid the ubiquitous transition from facilitative to competitive processes over ontogenetic time. In the context of natural selection, the trait allowing merging – and its possible link to origins of plant grafting – was probably selected for because of the dramatic increase in individual-level survival enjoyed within the merged groups (McIntire & Fajardo, 2011).

Table 3. Species groups that are commonly or near-obligate facilitators or benefactors (many examples below fall under more than one description, allowing for a more comprehensive description of the prevalence of facilitation)
Species groups or ecosystem typeFacilitator or facilitatedDescription
Shade-tolerant speciesFacilitatedShade tolerance, defined as the ability to grow and survive in less than full-sunlight conditions, is a trait that occurs widely and is a key driver for a massive diversity of plants (Halpern & Spies, 1995; Fig. 3b). These species occur under the canopy of tall woody plants or algae, for example. They exist largely because the negative direct effects of canopy shading are weaker than the positive effects resulting from the same shading on competitor species (Pages & Michalet, 2003). In most cases, shade-tolerant plant species are excluded competitively when there is no shade provided by overstory species (Sher et al., 2002)
Keystone modifiersFacilitatorKeystone modifiers, such as large herbivores or napped sapsuckers, provide resources for others of the same trophic levels (Mills et al., 1993)
Parasitic speciesFacilitatorWith a role similar to that of some herbivores, parasitic plants may preferentially affect a competitive dominant, such as sandalwood (Santalum paniculatum), mistletoe (Viscum album), or Cuscuta salina (Callaway & Pennings, 1998), and thus lower its performance and increase species richness and coexistence (Watson, 2009)
Epiphytes and vinesFacilitatedThese groups are near-obligate recipients of facilitative effects (although some vines grow on rocks or other structures; Fig. 2c). These include vast arrays of diversity including many mosses, lichens, ferns and other vascular plants (Sillett & Antoine, 2004). Lianas have recently been shown to support, and thus facilitate, themselves (Leicht-Young et al., 2011)
Forests, coral reefs, mangroves, salt marshes, and deserts as example communitiesBothFacilitation at the ecosystem level is ubiquitous. A large fraction of global biodiversity lives in ecosystems that are entirely facilitative (Callaway, 2007)
Ecosystem engineersFacilitatorSpecies such as corals, woody plants (e.g. trees), cushion plants (Fig. 2b) or algal forests (e.g. kelps) create structures, and others, such as salt marsh founders, change conditions and create novel ecosystems (e.g. mangroves, coral reefs, and salt marshes) that otherwise would not exist. These are responsible for important changes to the abiotic environment, creating vast species diversity
Nurse plantsFacilitatorIn many ecosystems, including alpine and arctic ecosystems, nurse plants create the sole substrate available for other plants to germinate and grow (Fig. 2b)
Communicating speciesBothPublic information sharing among animals and even plants is known to be widespread, with large but unknown effects on diversity (Danchin et al., 2004)
‘Smaller species'FacilitatedBecause of the structural heterogeneity produced by larger species in a community, there is a greater diversity of small species than would otherwise exist (Aarssen et al., 2006; Bartels & Chen, 2010)

The obvious question about this latter type of positive interaction is: wouldn't a cheater that takes resources from merged stems, but does not give back, win out? An earlier simulation study gave a possible answer to this question: within a world of cheaters, close competitors, and variable environments, both cheater and facilitators can be selected for and be readily maintained depending on the abiotic environment (Travis et al., 2005). In other words, there are probably cheaters throughout the system, but the cheater does not take over under some, maybe many, conditions. If, in a hypothetical world, individuals within a species could only be takers or givers, then takers would win. In the real world, the two strategies can co-exist within a population. Other factors that may reduce the impact of cheating in facilitative interactions are related to partner fidelity and partner choice mechanisms, like in mutualism (Axelrod & Hamilton, 1981; Simms et al., 2006).

Interestingly, another study showed that recent species lineages conserve the regeneration niche (via nurse tree syndrome) of older and distant lineages, showing that facilitation-induced biodiversity gains over evolutionary time were maintained (Valiente-Banuet & Verdú, 2007). In such cases, facilitation is crucial in determining community composition by enhancing long-term biodiversity. In a recent study, it has also been shown that alpine, foundational cushion species have functioned as micro-refugia by facilitating less stress-tolerant lineages in severe environments world-wide (Butterfield et al., 2013).

IV. Why has facilitation been just recently added to ecological theory?

From Gause's day [1934] onward, the discussion of coexistence and community assembly rules has been framed almost exclusively in terms of competition, despite numerous but largely unsuccessful attempts to broaden the discussion to include additional mechanisms. Hubbell (2005)

1. The historical context

Because much of the 20th Century's development of ecological theory was firmly anchored in Darwin's ‘struggle for existence’ concept, the idea of antagonism between individuals through the preponderance of negative density-dependence processes prevailed in the elaboration of models explaining species coexistence and biodiversity. Lotka–Volterra's model of interspecific competition, Gause's competitive exclusion principle, and Hutchinson's and MacArthur's works on the multidimensional niche and limited similarity are good examples of this, which were exclusively based on negative interactions between species (Bruno et al., 2003; Mittelbach, 2012). According to Hubbell's quote, ecologists have been preoccupied with a single explanation for species diversity (i.e. competition) and this has precluded us from conceiving other, also plausible, explanations of the ways in which patterns of species diversity are shaped.

Concurrently, using a ‘facilitation’ version of the Lotka–Volterra competition model, Gause & Witt (1935) found that changing the sign of the interspecific effect models’ component from negative to positive tended ‘to lead to silly solutions in which both [facilitators and facilitated] populations undergo unbounded exponential growth, in an orgy of mutual benefaction’ (reviewed in Mittelbach, 2012). Among other things, we believe that this silly result came from keeping the intraspecific density-dependent effect negative. This niche-similarity tenet of intraspecific negative density-dependent process may be questioned as a universal law in natural communities. A recent attempt to add explicit positive direct effects to this simple model showed quite a different result. Even though the individual positive effects included were weaker than the individual negative effects, coexistence occurred as a result of the positive effects (Gross, 2008). Supporting this point, in recent empirical studies, same-cohort conspecifics (i.e. strong niche overlap) did not compete but facilitated each other's survival (a cushion plant species in Cerfonteyn et al., 2011; seedlings of a tree species in Fajardo & McIntire, 2011; liana species in Leicht-Young et al., 2011; a biannual forb species in S. R. Biswas & H. H. Wagner, unpublished), challenging the notion that individuals have to compete when they encounter other similar-niche individuals. These examples constitute evidence for context-specific positive density dependence, with the same pattern seen in the Allee effect (Allee et al., 1949); that is, when the population density is low, the addition of new individuals correlates positively with individual fitness. In the case of Nothofagus pumilio seedlings (Fajardo & McIntire, 2011), facilitation at the intraspecific level not only occurred at the establishment stage but continued through later stand stages, including the self-thinning (i.e. maximum competition) stage (McIntire & Fajardo, 2011). Introducing context specificity (i.e. low population size, abiotic stress, novel structures, etc.) to intraspecific facilitation shines a new light on this 80-yr-old model.

2. Is facilitation predisposed to be undetected?

In competition-focused research fields, such as ecology and economics, facilitation may be undetected because it appears weaker than competitive mechanisms (Gross, 2008). For thousands of years of human communities, sellers of goods who are competing with one another have congregated in markets to sell to the public. Clearly, the individual sellers are competing with one another for the customer. But they are also facilitating each other because they draw in purchasers that would otherwise be unavailable. In this case, these two mechanisms act at different scales. If a researcher looks at direct interactions, there is clearly competition: the sellers are not giving each other anything directly. However, the entire stage on which the competition can occur is a result of the facilitative interactions. It is also likely that there are indirect or intransitive effects occurring; having more competitors limits you and your competitors, resulting in possible positive effects.

Under the niche paradigm, diversity is achieved by each species occupying a unique combination of a multidimensional niche. If an ‘invader’ species arrives in the community, it may be able to thrive because there is a niche available (the empty niche hypothesis for invasion success; Stachowicz & Tilman, 2005). However, recent studies have demonstrated that the same observation can be caused by an undetected direct facilitative mechanism: organisms themselves increase the number of niches by increasing habitat complexity (Schöb et al., 2012). In other words, are the species exploiting more niches, or are there more niches to exploit? To differentiate these two mechanisms probably requires at least a trait-based analysis, which has only recently been related to plant interactions (e.g. Schöb et al., 2012; Spasojevic & Suding, 2012; see the section on ‘Facilitation and the plant functional programme’). Unless explicitly evaluated and tested, the observations are identical whether the mechanism is abiotic niche filling or heterogeneity creation via facilitation. In studies addressing intransitive competition, there are both competitive and facilitative effects, and thus there will likewise be a detection bias.

The abiotic stress amelioration may be the easiest facilitative mechanism to confirm because, at extreme drought or temperatures, species perish quickly in an experiment. If, in contrast, the facilitative mechanism creates more heterogeneity allowing, say, more species covering a higher diversity of specific leaf area (SLA) to coexist, the proximate metric to measure is SLA dispersion, even if it is a coexistence mechanism and does lead eventually to greater survival of a greater diversity of species. The more readily measurable ‘metric’ of rapid experimental mortality creates a positive detection bias for tests of the SGH. Given that 20 yr ago little was known about facilitation in stressful environments, and that today we know that facilitation creates widespread productivity and diversity increases virtually every time we look in these environments, it is possible that facilitation via a variety of mechanisms is widespread in all systems.

V. Facilitation and the plant functional trait programme

In recent years, myriad studies have focused on a plant functional trait programme, with the main goal being to understand how traits mediate community assembly and coexistence and then to use this understanding to predict the effects of global change on biodiversity (Suding et al., 2005). Functional trait research has shown that trait variation creates fitness advantages through environmental filtering processes while simultaneously preventing competitive exclusion via niche partitioning. Evidence for the latter comes from the great variation in trait values found within most communities and from studies showing that trait values of co-occurring species are overdispersed relative to expectations from null models (e.g. Westoby et al., 2002; Kraft et al., 2008). What has been less explored is that such within-community high trait dispersion may be in part a result of facilitation (see Gross et al., 2009; Schöb et al., 2012; Spasojevic & Suding, 2012). There is growing evidence that facilitation also increases trait dispersion at the community level. Distinct from the overdispersion created by niche partitioning, under facilitation-driven overdispersion, the coexistence will be of functionally contrasting species (e.g. Callaway, 2007; Gross et al., 2009; Butterfield & Briggs, 2011). Facilitator species (Table 3) mostly respond to abiotic environment filtering, so they possess the functional traits constrained by the environmental conditions occurring at larger scales than the community. In contrast, facilitated species mostly respond to biotic environment filtering represented by the modified conditions the facilitator species imposes, this being at a scale defined by the facilitator's scope (i.e. a biotic environmental filter within a large-scale abiotic environmental filter). Thus, facilitated species will possess functional traits according to these micro-scale conditions but not necessarily to those found in the region: facilitator species' positive effects not only increase the number of species (see Butterfield et al., 2013) but also increase trait dispersion in the local community (Gross et al., 2013; Fig. 1). Thus, distinguishing the cause of trait divergence will require an understanding of the type of traits that occur; that is, the mechanisms responsible for this trait divergence will reflect the facilitation mechanisms outlined here.

Although existing work on trait-based approaches represents important advances in community ecology, it lacks predictive power to understand the impact of any perturbation in community structuring. Thus, a shift from a phenomenological approach towards a more mechanistic approach is currently needed and encouraged (Adler et al., 2013). Butterfield & Briggs (2011) experimentally found that, in an arid system, shrub species facilitated by a nurse's canopy showed functional traits related to a resource-conservative strategy but other colonizer species (nurse-independent) showed traits related to a more resource-acquisitive strategy. Similarly, Schöb et al. (2012) found, in an alpine system, a significant trait differentiation between species facilitated by cushion species and others growing in bare ground: the former showing a more resource-conservative strategy and the latter showing a more resource-acquisitive strategy (higher leaf dry matter content (LDMC) and SLA). Although these studies demonstrate the functional causes and consequences of facilitation in community structuring, they also suggest that relevant traits may be idiosyncratic, that is, environment-dependent (Butterfield & Callaway, 2013). These examples of where studies have pursued the particular traits responsible for the patterns observed show great promise in distinguishing facilitation-based trait divergence from niche-based trait divergence. But it is with an understanding of all the mechanisms we have presented in this study by which local environmental conditions and constraints are modified that a functional understanding of facilitation will occur.

VI. Predictability and testability

Testing this notion of facilitation as a driver of biodiversity will have greatest success if two things are achieved: (1) understanding the resource conditions, both at the site level and at the species level, and (2) allowing for indirect effects, either biotic (e.g. competition propagation) or abiotic (e.g. when changes in conditions cause some species to disappear but others to appear). Using a trait-based approach will help with both. For example, detecting a lower growth rate in a potential facilitated species as a result of the shading from a ‘benefactor’ is not sufficient to reject the hypothesis of facilitation as a driver of biodiversity. In this case, if that lower growth rate in the shade is associated with the appearance of two or more new species that are shade-tolerant, using a trait-based approach, facilitation has had its positive effect in increasing diversity as predicted.

More specifically, each facilitative mechanism will require a different type of test. Some are well understood and are widely tested (e.g. primary succession and shade tolerance) and others are not. Some have been tested more recently and are becoming mature and accepted hypotheses (e.g. SGH). Others, however, have only recently been explored (e.g. nontransitive competition networks, hydraulic lift, resources sharing, and habitat complexity). The more mature hypotheses have stronger predictions and have been widely supported. Testing some of the less mature hypotheses may require different types of experiments from the traditional pairwise competitive trial. Mathematical models of intransitive competition are maturing and there are numerous alternative models that provide concrete predictions (e.g. Frean & Abraham, 2001; Laird & Schamp, 2009; E. J. B. McIntire, unpublished) that can be tested in controlled experiments. New work using traits to show the mechanisms of facilitative interactions shows great promise (Schöb et al., 2012). Other multi-species models indicate that the fraction of a community richness that comprises mutualistic versus exploitative relationships is greatest with low dispersal ability (Filotas et al., 2010), consistent with empirical work in arid systems (Kéfi et al., 2008). However, resource sharing still has several empirical observations (McIntire & Fajardo, 2011; Tarroux & DesRochers, 2011) but only two mathematical models that we are aware of that describe how it can persist on evolutionary time-scales (Travis et al., 2005; Uitdehaag, 2011).

Although species coexistence models based on niche theory have provided critical insights into how trade-offs (e.g. niche differences) can promote higher levels of species diversity, these insights have been mostly based on simple systems or experiments (e.g. two annual species). In reality, these competition-based models have provided limited evidence for niche differences in more complex systems (Clark, 2010), and have rarely provided general principles about many-species communities (Simberloff, 2004). Indeed, evidence suggests that competition-based models fail to explain species diversity in tropical forests where high numbers of species with very similar niches and life forms coexist. Likewise, in communities subject to invasion, invaders without apparent niche differences are able to successfully establish (MacDougall et al., 2009). While there is some support for the neutral model in highly diverse forests (Hubbell, 2001), there is also evidence rejecting the neutral model even in tropical forests (McGill et al., 2006; Ricklefs & Renner, 2012) and that there are coherence problems with the theory (Clark, 2012). Further empirical tests that isolate individual mechanisms and that allow for detection of facilitative mechanisms are badly needed.

VII. Conservation, restoration and management

For a long time, intercropping systems, including agroforestry (Young, 1997), have been used to improve yield rates over monocultures. Intercropping can increase yield rates through niche complementarity mechanisms but also via facilitative mechanisms such as improved nutrient-use efficiency (Zhang & Li, 2003; Li et al., 2007). For conservation or restoration applications, the challenge for practitioners is identifying which positive interactions are present, and how to take advantage of them to maintain or increase biodiversity. In semi-arid and marine systems, facilitation is already being widely used in restoration (e.g. Gómez-Aparicio et al., 2004; Halpern et al., 2007), including recently appreciated intraspecific facilitation (Madsen et al., 2012). These abiotic stress amelioration applications appear to be readily implementable with immediate results. Climate or niche envelope approaches to understanding species responses to future climate lack biotic interactions in general, and positive interactions as a rule. Successful predictions will probably require an understanding of positive interactions (Lortie et al., 2004). Promoting ecological resilience will look different if it is intransitive competition networks that are promoting diversity. In a root grafting study (Tarroux et al., 2010), researchers may have resolved a major cause of ‘thinning shock’: trees whose neighbours are thinned mechanically show stagnant or declining growth (Vincent et al., 2009) and can show high mortality rates (Harrington & Reukema, 1983), in spite of the increased space, light and access to nutrients. Newly cut trees that were root-grafted converted the observed resource sharing into resource sinks (Tarroux et al., 2010). But, as ecologists and land managers world-wide address ecosystem integrity, understanding the potential positive effects in the system can change the strategies used.

VIII. Conclusions and next steps

The unification under this single facilitation concept of several disparate mechanisms, which have previously been explored separately within the ecological literature, will create three benefits in our effort to understand the structuring and dynamics of communities. First, the high relative importance of positive effects on diversity in relation to negative effects needs to be better appreciated and explicitly studied without appealing to simplified communities that may bias the detection of negative effects. Each time we depart from simple pairwise interaction frames of study, evidence is mounting that we will detect positive effects as the net outcome in plant interactions. Thus, we can better explain and understand this type of result and perform more appropriate studies if the null hypothesis of competitive interactions is not strictly negative. Secondly, it is wrong to assume that negative effects are the only stable result of plant interactions; there are numerous mechanisms that are evolutionarily stable that promote positive interactions within trophic levels. By identifying processes and traits that drive positive interactions it will become clear that positive effects are ubiquitously stable. Thirdly, we demonstrated that facilitation is only sometimes the opposite of competition, because facilitation involves net positive effects. The result of this enemy-of-my-enemy-is-my-friend phenomenon is net positive effects. As there is no absolute correct scale in ecology, both approaches are valid and important. At times they act at the same scale (e.g. much of the SGH and intransitive competition), and at times they act at different scales (e.g. overstory trees facilitating shade-tolerant herbs which compete among themselves; Schöb et al., 2013). This result allows a deeper understanding of all the mechanisms of positive effects in communities.

The importance of facilitation for diversity, including intransitive and indirect mechanisms, is still unknown, although recent evidence suggests that it may be particularly important in high-diversity ecosystems (Allesina & Levine, 2011; E. J. B. McIntire, unpublished). Few studies have attempted to quantify the number of species that have been added to a community because of facilitative interactions. Where it has been attempted, Hacker & Gaines (1997) estimated that at least 35% of the observed species diversity in a salt marsh was present as a result of direct facilitative interactions, Valiente-Banuet & Verdú (2007) estimated that 90–100% of species of interest were facilitated over evolutionary time, Stone & Roberts (1991) found that 20–40% of effects from direct competition were beneficial in a simulation experiment, and E. J. B. McIntire (unpublished) found that indirect facilitation accounts for > 50% of species richness under a wide range of conditions. The number of species that would be classified as shade tolerant, an example which includes understory plants, and may be the majority of the floristic diversity (Halpern & Spies, 1995), would give an idea of the biodiversity consequences of overstory facilitation. As for the facilitative effects of competitive intransitivity, there is an enormous potential biodiversity effect of this facilitation that may be responsible for a large fraction of species in an ecosystem (Callaway, 2007). The challenge now is to quantify the relative importance of facilitative mechanisms for maintaining global biodiversity by designing experiments and studies that can, in principle, detect net positive interactions so that we can resolve questions such as: are the species exploiting more niches, or are they creating more niches to exploit?


Financial support came from the Centre d'études de la forêt (www.cef-cfr.ca) for a visiting scientist grant to A.F. and the Canada Research Chair program (EJBM). A.F. also acknowledges financial support from a FONDECYT Project No. 1120171. We also appreciate the help of A. Valiente-Banuet and P. Adler who kindly reviewed a preliminary version of this paper. Finally, we thank A. Saldaña, I. Till-Bottraud and C. Torres for providing photographs of their study sites.