Being a facilitator can be costly: teasing apart reciprocal effects


Being a facilitator can come with a cost. That is what Christian Schöb and his team, in this issue of New Phytologist (pp. 95–105), demonstrated using globally distributed data from an impressively large set of field experiments and observations. In other words, if a species creates conditions that allow potential competitors to survive better, these competitors may not be so generous in return. This creates a sticky situation, in an evolutionary sense: why would any species facilitate a potential competitor? Since facilitation does persist in nature, the reciprocal cost found by Schöb et al. needs to be examined deeper. Indeed, the sign of the reciprocal effect is so important, some sources limit the definition of facilitation to include only commensalisms (+/0) or mutualisms (+/+) and explicitly exclude antagonisms such as parasitism (+/−) (Cain et al., 2011). This emphasizes the importance of understanding the feedback effects. But, how surprising is this negative feedback?

‘This creates a sticky situation, in an evolutionary sense: why would any species facilitate a potential competitor?’

Reciprocal effects – what are the expectations?

Bronstein (2009) indicated that if the reciprocal response is negative, then there should be selection against the benefactors’ negative effects, that is, the facilitator should show signs of reducing the costs due to the benefactors. This reduction could happen in many ways, with two of the most likely being a temporal ‘avoidance’, as occurs in primary successional situations, and a measurable anatomical or physiological response that ‘tolerates’ and so reduces the cost. Bronstein (2009) also pointed out that if a facilitative relationship is indeed parasitic, it would likely have the same selective pressures of parasitism. Amongst these, chief would be that there should be strong selection for the benefactors to not damage their facilitator too much since, like parasites, they still depend on their host. Thus, the expectation could be for at least three different responses in the facilitator (e.g. cushion plant) to negative reciprocal effects: avoid, tolerate and minimize absolute damage.

In their paper, Schöb et al. examined data on high elevation cushion plants derived from studies around the world and they looked at how they are affected by the community of plants that colonize them. Cushion plants are well known early colonizers of inhospitable sites (e.g. too cold, too dry, etc.). Their mere presence modifies the local abiotic conditions, allowing other plant species to germinate and grow within their short canopies, that is, facilitation. To date, however, few studies had looked at the growth, survival or fitness costs on the facilitator (i.e. the cushion) of having provided the conditions necessary for the benefactors to survive and grow. The authors’ objectives, however, were not to elaborate about the expectations of the reciprocal relationship. Examining the expectations here, it is apparent that all three can be seen in their results. In a temporal ‘avoidance’ response, one would expect that young cushions would show fewer costs from the benefactors. In their table 2, the strongest effect on seed density (the closest correlate of fitness) was a negative effect of cushion size, that is, larger (older) cushions had the lowest seed density. Furthermore, as for the ‘tolerate’ response, even with 100% cover of benefactors – that is complete cover of the host cushion plant – there was only a model estimated 30% decline in seed output (their table 1 and fig. 1d). Comparing this to other competitive situations where a shade intolerant species is 100% covered by other plants, these cushion plants showed good tolerance. Finally, for the ‘minimize damage’ response, the authors found a strong positive relationship between the cover of benefactors on the cushions and the number of fruits that a given flower produces (their fig. 1b). In other words, cushions that were invaded overproduced fruits per flower. While no mechanism was explicitly evaluated in their study, this was a dramatic ‘positive’ result by the cushions. This could be due to many things including phenotypic plasticity (e.g. putting more resources into fruit), increased attraction of generalist pollinators because of the benefactor's flowers (i.e. reciprocal facilitation), or simply compensatory fruit production (i.e. the plant would not actually have enough resources to mature all its flowers into fruits, so the loss of some flowers due to competition was partly compensated for in the remaining flowers). Thus, there may already be some selection for plasticity or else the magnitude of the competitive effect by the benefactors is not as strong as it could be. Thus, this increase in fruit output per flower minimized the cost of a reduced number of flowers per unit area.

What is the generally expected feedback when we see a facilitative relationship? Perhaps when we see a positive action, we expect that the response will be positive, as in mutualisms. While mutually beneficial situations among ‘facilitator’ and ‘benefactor’ are known (McIntire & Fajardo, 2011), evolution favours strategies and behaviours that improve heritable fitness, so there is no requirement to reciprocate. Ideally, we would be able to examine the relationships in the context of heritable fitness so we could identify the stable evolutionary outcome and understand the complexities of reciprocal interactions. Measuring fitness effects, of course, is a laudable goal for ecological studies. While this is likely difficult in many situations, recent work by Hart & Marshall (2013) provides a nice example that links population dynamics (i.e. the consequences of differential fitness), coexistence theory and the predictions of the Stress Gradient Hypothesis (SGH; Bertness & Callaway, 1994) by understanding the role of environmental context on lambda. In general, there is no a priori expectation for the sign of the reciprocal relationship because it depends on context and mechanisms. Thus, evaluating the mechanisms, whether they be competitive or facilitative, will help determine expectations.

How do we proceed?

To address the issue of reciprocal effects, there are at least three ideas that will enable us to better predict which sign they will have and understand their ecological and evolutionary function. First, the ‘reciprocal effects’ are just one of many interactions in a whole community, all of which are potentially important. The community is actually a large network of interactions that may begin with the pairwise neighbours, but that extend through a vast network of indirect effects (E. J. B. McIntire, unpublished; Levine, 1999), intransitive relationships (Laird & Schamp, 2006), and feedbacks, which can each be positive, neutral or negative with no default expectation.

Second, neither facilitation nor competition is actually a mechanism, that is, a species cannot ‘do’ competition. Rather, these are quantifiable and emergent outcomes of a wide range of mechanisms that could be genetic, physiological, anatomical, biochemical, numerical or otherwise (McIntire & Fajardo, 2014). Furthermore, we know that in general, facilitation and competition result from different mechanisms and these may be correlated or not and can be examined independently (see example in Fig. 1). For example, root competition for limited nutrients may exist among two plants while the same two plants are in a facilitative relationship for water because one plant shades the other in a semi-arid environment. Predicting the sign of the reciprocal effect, of course, will depend on the strength of the competition (if the facilitation is one way). There is no amount of theory that will indicate which of these will be stronger unless both mechanisms are known (i.e. nutrient and water stress). Importantly, Lin et al. (2012) demonstrated that violations of SGH predictions come when facilitation was asymmetric, that is, feedbacks were not positive. We will gain insight into several problems at once – such as understanding reciprocal relationships, better understanding of coexistence, diversity drivers – if we address the facilitation–competition dynamic separately and mechanistically, even in observation-only studies (McIntire & Fajardo, 2009). When many species across many ecosystems are involved, a trait-based approach is showing great promise as we learn more about the connections between ecological mechanism and species traits (Adler et al., 2013).

Figure 1.

Teasing apart bidirectional interactions between a facilitator (e.g. a nurse cushion plant) and a benefactor (growing with and benefiting from the facilitator). Shown is in a hypothesized two-species community across an environmental gradient, consistent with Schöb et al. (in this issue of New Phytologist, pp. 95–105). (a) Abundance of each species as could appear across the gradient using an observational survey. (b) Net interaction (e.g. facilitative, competitive or neutral) across gradient that could be uncovered by removal experiments of the system shown in (a). Vertical arrows indicate where transitions occur between interaction types of (e) and (c). The teasing apart of the underlying contributions in (b). (c) Abundance due to abiotic conditions only (black line) is lowered by a competitor (red line, competition effect-diagonal hatching) and raised by a facilitator (green line, facilitation effect-vertical hatching). These combine to create the benefactor abundances in (a). These can be teased apart using trait-specific experimental removals. For example, removing a stress facilitator and experimentally replacing only its stress amelioration role (e.g. with a physical wind break or shade cloth) would isolate the facilitation effect; leaving the facilitator and adding the resource that is limiting due to competition (e.g. fertilizer or water) would isolate the competition effect. (d) As with (c), but for facilitator abundance. Note the lack of facilitative effect here, but there could be a reciprocal facilitation effect. (e) Resulting transition between competition, parasitism and commensalism that follow from the above figures. Vertical dotted lines through all five subfigures indicate transitions between interaction types that correspond to (e); these can shift readily right and left depending on the species involved, and mechanisms of facilitation and competition. To better understand violations of the Stress Gradient Hypothesis predictions, the limiting resource may induce strong competition. The red competition lines will be lowered, thereby shifting parasitism and commensalism to the right, even if facilitation is identical.

Third, all interactions between species have both negative and positive components in each direction (e.g. +&−/+&−). Even the most obviously antagonistic relationships (+/−), such as herbivory or predation, have important instances where positive effects occur in both directions (+/+) via overcompensation (Paige & Whitham, 1987; Roos et al., 2007) or indirect effects. In the case of Schöb et al., the overall statistical relationship showed a negative effect of the benefactors on seed density (+/−), but numerous sites deviated from this noisy result. In these cases, the ‘benefactors’ growing within their cushion plant canopy maybe helped attract generalist pollinators to the cushion plant improving the seed output per flower, thereby reciprocating the facilitative relationship (+/+). So, the sign of the bidirectional effects depends on context (i.e. environment). Obviously, the positive or negative effects may be small in many situations and can be ignored. However, positive or negative effects may be transient and can readily switch sign. It is a poor assumption that reciprocal relationships are of a particular sign simply because of the sign of the ‘first’ relationship. Emerging work is suggesting that nurse communities, for example, may exist where many or perhaps all ostensibly competing species are also mutually increasing survival of their neighbouring competitors (J. Chacón & E. J. B. McIntire, unpublished). We are only beginning to understand how the reciprocal dynamics of positive and negative effects affect community composition and dynamics; keeping our expectations open about interaction sign is a key step as we explore further. Schöb et al. showed that a good null model for this relationship may be antagonism, which is likely a change for many who think of facilitation as primarily comprising mutualisms and commensalisms.