Competition for mutualistic resources
Since mutualisms are defined as interactions that confer reciprocal benefits to two species, they are widely depicted as shown in Figure 1A. The partners are linked by positively labeled arrows, the top arrow in the figure indicating that species M1 confers benefits to M2, and the bottom arrow indicating that M2 confers benefits to M1. This abstraction of mutualism hides as much as it reveals, however. What are these benefits and is the exchange as simple and reciprocal as the figure might imply?
Figure 1. Competition for mutualistic resources. (A) The traditional net effects diagram of mutualism. The arrows show the net reciprocal benefits (+/+) of the interaction between two mutualistic species (M1 and M2). (B) A resource-based diagram of mutualism. Mutualists M1 and M2 produce resources R1 and R2, respectively, and consume the resource produced by the partner mutualist. The arrows show resource production (black) and resource consumption (red). (C) Mutualism when low-quality mutualists have a competitive advantage. M1 has now been decomposed into two individuals or two species from a mutualist guild (M1a and M1b). M1a is a better mutualist since it offers a large amount of resources to M2 (thick arrow from M1a to R1). However, M1b is a better competitor for the resources produced by M2 (thick arrow from R2 to M1b). Here, the competitive advantage comes from interference competition (dashed inhibition arrow between M1b and M1a's R2 consumption arrow). (D) Mutualism when high-quality mutualists have a competitive advantage. Again, M1a is a better mutualist by producing large amounts of R1. However, it is now a better competitor for R2 (thick arrow from R2 to M1a). In this case, the competitive advantage results from partner control by M2 (dashed inhibition arrow between M2 and M1b's R2 consumption arrow).
Download figure to PowerPoint
Figure 1A illustrates the effects of a single bout of commodity exchange, but not its underlying mechanism. While not shown here, competition can be included quantitatively in net effects diagrams through path coefficients.16 Stanton's use of path analysis to quantify the interactions between and within guilds of mutualists was an important step in extending the study of mutualism beyond pairs of species. However, path analysis is still a top-down, phenomenological approach that does not identify the causes of the measured effects. In contrast, in Figure 1B, we try to capture more of the mechanism of commodity exchange underlying mutualism through a bottom-up depiction. Now, instead of the arrows connecting two species, the arrows point to a commodity that is then delivered to the partner. These commodities may be, in the language used above, either rewards or services; we group these into a single category, called resources (R). While we recognize their differences, we take this approach because in the context of exchange, they share two critical features: rewards and services both can be costly to offer to partners, and both can be competed for. In Figure 1B, there are still two species (M1 and M2), but now there are also two resources, R1 and R2; the arrows now are resource flows rather than (as in Fig. 1A) net effects. Black arrows indicate resources produced by a mutualist, and red arrows those that are consumed (either actually or metaphorically) by a mutualist. Thus, in Figure 1B, M1 produces R1, a resource that is consumed by M2; M2 in turn produces R2, a resource that is consumed by M1. As an example, M1 might be a plant that produces nectar (R1), which is consumed by M2, a bee, that produces the resource of pollen transport (R2) that is used profitably by M1.
A major advantage to depicting mutualism as shown in Figure 1B rather than Figure 1A is that it makes clear how competition can lie at the heart of these interactions. In Figure 1A, it is not evident what could be competed for; in Figure 1B, it is clear that it is the resources, R1 and R2. Indeed, the best documented way in which there is interplay between competition and mutualism is that there is competition for the commodities mutualists produce. Consider pollination, the most thoroughly studied mutualism. Floral visitors are well documented to compete intraspecifically for nectar, for example, by adjusting their foraging routes in relation to the activities of conspecifics on flowers.17,18 Competition may also pit individuals of different species against one another, a phenomenon particularly well-investigated between honeybees and native bees that they may be displacing.19 On the other side of the interaction, it is common for plants to fiercely compete both intraspecifically and interspecifically for pollinators, as Mitchell et al. have thoroughly reviewed;20 competition for pollination has clearly shaped the evolution of flower sizes and numbers, floral reward chemistry and volume, and both visual and olfactory cues.20–23
Adding competition for resources explicitly into Figure 1B yields many possible outcomes, two of which are shown in Figure 1C and D. First, we decompose one mutualist (M1) into multiple individuals or species; we will explore how these entities (e.g., different pollinator individuals or species, or different nectar-producing plant individuals or species) compete for mutualistic resources. We will call these M1a and M1b. These two entities consume a shared resource, R2, and both produce a second resource, R1. Via these resources, both interact with a partner species, M2. However, M1a and M1b are not identical. The different thicknesses of the lines from M1a and M1b to R1 indicate that one entity produces more of R1 than does the other. Similarly, the different thicknesses of the lines from R2 to M1a and M1b indicate that one entity consumes more of R2 than does the other.
Figure 1C illustrates the situation in which a superior competitor is an inferior mutualist and competitive interactions play out in a way likely to be harmful to the shared partner. In Figure 1C, M1b provides less of R1 to the shared partner M2; thus, we define it as an inferior mutualist. However, M1b uses more of R2 than does the superior mutualist, M1a. M1a's lower consumption of R2 is due in some way to the presence of M1b: M1b might be actively interfering with M1a's consumption, or it may simply be consuming R2 faster or more efficiently, leaving less behind. We illustrate this effect with a dashed inhibition arrow running from the superior competitor M1b to the arrow connecting the shared resource R2 to M1a. To put this in words, the shared partner is stuck with a relatively low-quality mutualist able to reduce the success of better-quality mutualists. (This is quite realistic biologically. For example, Bennett and Bever demonstrated that the most beneficial mycorrhizal fungus species for Plantago lanceolata is the worst competitor for root space, whereas the worst fungal mutualist is the best competitor for P. lanceolata roots.9) This is a situation that might lead to the low-quality mutualist dropping resource provision altogether and becoming an exploiter of the system, leading one to question how mutualisms embodying this structure are able to persist evolutionarily. We discuss this in more detail below.
In Figure 1D, we illustrate the situation in which it is the superior mutualist that holds the competitive advantage. In contrast to the situation shown in Figure 1C, this can clearly benefit the shared partner. Here, M1a, the mutualist that consumes more of the shared resource R2, also provides more of resource R1 to the partner species M2. We have illustrated this outcome as being the result of actions of M2: it has exerted some kind of “partner control” that reduces the ability of the inferior mutualist M1b to compete for the resources it provides. This situation has been argued to permit mutualisms to persist in the presence of cheaters, as we discuss below. As in the case shown in Figure 1C, it is not difficult to identify biological examples of these relationships. Adam documented such a case in the interaction between cleaner fish and their “clients.”10 Clients (butterflyfish) are able to selectively associate with cleaners (wrasses) that provide them with the highest quality service (parasite removal). Thus, clients confer a competitive advantage to the best cleaners. Indeed, cleaners provide better service when competitors are present as this is the only way that they will be chosen by hosts.
Figure 1C and D are only two possible types of competitive interactions within guilds of interacting mutualists. Many other networks can be envisioned, and indeed are well documented in the literature. For example, we have not considered here that competitive advantage and mutualistic quality can both be functions of population size, and thus can vary over ecological time scales (e.g., Refs. 24 and 25). There is clearly much more to explore. Our overall point is simply that making the resource exchange underlying mutualism explicit (Fig. 1B), and clarifying which of these resources are competed for, which partners hold the competitive advantage, and which are the best mutualists (Fig. 1C and D), reveal a fascinating range of possible ecological and evolutionary ramifications that are completely obscured in the simple, standard net effects-based way of viewing mutualisms (Fig. 1A).
Competition between mutualists and exploiters
Almost all mutualisms are afflicted with individuals and species that gain the benefits that mutualism offers, while investing little or nothing in return.26–28 A perennial question about mutualism is how it can persist ecologically and evolutionarily in the face of these organisms (hereafter, exploiters) that would seem to be at an advantage. To answer this question, it is essential to think beyond the comparative effects of exploiters and mutualists on their shared partners. These are relatively well studied. One also needs to consider the nature and outcome of competition between exploiters and the species that share that partner. This issue has barely been addressed in the growing literature on cheating within mutualism (but see, for instance, Refs. 13 and 29–31).
First, it is necessary to illustrate the interactions in question, as we did in Figure 1, for mutualisms in the absence of exploitation. Parallel to Figure 1A, Figure 2A gives the standard, net effects-based illustration of a mutualism that is associated with an exploiter, E. The arrows are labeled to indicate that E benefits from M1 but is detrimental to it, and that E and M2 are detrimental to each other (since they share a partner). A well-known example is the well-studied network of interactions among plants, pollinators, and nectar-robbers, floral visitors that feed on nectar but do not pick up or deposit pollen.32 Nectar-robbers (E) and pollinators (M2) both interact with plants (M1), but only M2 confers a benefit to M1. As in Figure 1A, no mechanisms are shown.
Figure 2. Competition between mutualists and exploiters. (A) A net effects diagram. The core mutualism between M1 and M2 is the same as in Figure 1A. An exploiter (E) has been added. E gains a benefit (+) from M1, but inflicts a net cost (–) on M1, since E does not reciprocate any benefit. Both E and M2 gain benefits from M1, thus they are competitors (–/–) for these benefits. (B) A resource-based diagram. The core mutualism is the same as in Figure 1B. The added species, E, does not produce any resources. However, E consumes R1 thereby competing with M2 for this resource. (C) Exploited mutualism when the exploiter has a competitive advantage. E is a better competitor for R1. Here, the competitive advantage comes from interference competition. (D) Exploited mutualism when partner control gives a competitive advantage to the mutualistic partner. M2 is able to gain more of R1 due to partner control by M1. (E) Exploited mutualism when mutualists are superior competitors. Again, M2 is able to gain more of R1; however, it is now because M2 interferes directly with E. For the resource-based diagrams, (B–E), arrows represent resource production (black arrows), resource consumption (red arrows), increased resource production or consumption (thick arrows), and interference with resource consumption (dashed inhibition arrows).
Download figure to PowerPoint
The other panels in Figure 2 take a resource-exchange rather than a net effects-based approach to these interactions, as did Figure 1B–D. The resources of exchange are added into Figure 2B, parallel to Figure 1B. Note that the M1-R1-M2-R2 network in Figure 2B is identical to that shown in Figure 1B. However, an exploiter species has now been added. Like M2, E consumes the resource R1, but unlike M2, it does not provide the resource R2 to the partner M1. Interestingly, moving to a resource-exchange perspective has served to simplify the net effects depiction (Fig. 2A) by making it clear what it is that species share, exchange, and compete for. In the case of plant—pollinator–nectar-robber interactions, for example, Figure 2B clarifies that pollinators (M2) and nectar-robbers (E) both utilize a resource, nectar (R1), but that only pollinators deliver a resource (R2), pollen transport, to the shared partner.
Parallel to Figure 1C and D, Figure 2C–E shows distinct ways in which competition for a resource shared between a mutualist and exploiter can occur and be mediated. Each suggests distinct ecological and evolutionary outcomes, and each captures a phenomenon represented in the empirical literature.
In Figure 2C, the exploiter is competitively superior at obtaining the resource from the shared mutualist. (We refer readers to the discussion of Figure 1C for an explanation of how to read these effects based on the colors, arrow thicknesses, and arrow patterns.) This gives rise to a situation in which mutualists can potentially be competitively excluded by exploiters, raising the obvious question of if and how mutualism can persist under these conditions. Examples can be found in nature. For example, Dohzono et al. studied a case in which a nectar-robbing bumble bee, Bombus terrestris, competes with native pollinating bumble bees for a shared nectar-producing plant, Corydalis ambigua, in Japan.33 Once nectar-robbers are present, pollinators abandon C. ambigua for other nectar resources, to the detriment of the plant. It is interesting that in this case, the exploiter is an introduced species. Will this plant, or at least its mutualism with native pollinators, be able to persist over the long term? This is more than an abstract evolutionary question. It highlights that an understanding of competitive hierarchies among native and introduced species may shed light on the conditions under which mutualisms will be able to persist and evolve in the face of anthropogenic change.34
Figure 2C may thus seem to be an ecologically and evolutionarily fragile situation for mutualism. However, close study of several mutualist–exploiter interactions has revealed that the shared partner has some ability to control the exploiter in a way that shifts the competitive advantage toward the mutualistic partner. We illustrate this general situation in Figure 2D. As a biological example, Kiers and colleagues have elegantly demonstrated that plants are able to discriminate among and differentially deliver resources to symbiotic Rhizobium bacteria that produce relatively more fixed nitrogen for them;35,36 similar control mechanisms may exist within other plant rhizosphere mutualisms.37 A wide variety of control mechanisms have been suggested under various names (e.g., sanctions, punishment, and partner choice). Given the potential importance of these mechanisms in allowing mutualism to persist in the face of exploitation, a large body of theory has been developed to examine when each mechanism is likely to evolve and how it would function (e.g., Refs. 27 and 38–40). We consider this issue in more depth below.
It is clear, however, that partner control mechanisms are not always necessary to explain how competing mutualists and exploiters are able to coexist. One obvious case is when it is the mutualist rather than the exploiter that holds an innate competitive advantage. This situation is illustrated in Figure 2E. Good empirical evidence comes from mutualisms between certain tropical plants and the highly specialized ants that inhabit them. Some of these ants are mutualistic, aggressively defending their plants from herbivore attack, whereas others occupy the plant and provide no defense. Exploiter ants have in several cases been reported to be competitively inferior to mutualistic ants; when mutualists invade a plant that exploiters occupy, it is commonly the exploiters that are displaced.41,42 Their superior ability to locate unoccupied plants results in a competition–colonization trade-off that allows them to persist even in the face of their evident disadvantage when challenged.41,42 Furthermore, certain mutualistic ants possess dietary specializations that allow them to more efficiently use the food that plants provide them.29,31 Of course, it is possible that the chemical makeup of this resource has evolved as a partner-control mechanism that shifts the competitive balance toward mutualists, illustrating that it can be difficult to empirically distinguish interaction networks as similar as those shown in Figure 2D and E.
Thus, to understand when we would expect to see the evolution of sanctions and punishments in mutualisms afflicted with exploitation, it is critical to examine whether exploiters hold a competitive advantage over the mutualists with which they share a resource, or vice versa. This has only rarely been investigated. The usual assumption has been that control mechanisms are always essential for mutualism to persist in the face of exploitation.
Competition between mutualists
In the previous two sections, we have considered the two best understood ways in which competition and mutualism interact: when individuals or species compete for resources provided by a shared mutualist and when mutualists and exploiters compete for resources from a shared partner. A much less studied phenomenon is when mutualistic partners are also competitors for resources. When mutualists occupy different trophic levels, as is generally the case,2 resource sharing is not expected. Thus, for instance, plants and pollinators do not share and compete for resources, nor do ants and the plants they defend. However, a number of less well-known mutualisms involve species that occupy the same trophic level and thus are likely to compete for access to shared resources.43 Müllerian mimicry in butterflies provides a good illustration. In these cases, mutual benefit is derived by a shared resemblance that “trains” predators to recognize them and, because they are distasteful or toxic, to avoid consuming them (e.g., Ref. 44). These individuals are likely to compete for food and other resources, however. A similar situation can be found in interspecific group foraging: individuals benefit either by shared predator vigilance or by increased access to food.45,46 However, they also may compete for the food they locate.47
We know of no standard way of illustrating these relationships via net effects arrows as in Figures 1A and 2A. We attempt to do this in Figure 3A. Here, the arrows are labeled +/– because the partners are simultaneously competing (hence, are in a minus/minus interaction) and benefiting each other (hence, are in a plus/plus interaction). This is not very satisfying. The difficulty of finding a way to illustrate these interactions is symptomatic of a general problem with using net effects to capture them: mutualism and competition occur simultaneously, and whether they add up to net effects that are positive for one, both, or neither partner is likely dependent on many system-specific and context-dependent factors.48 Path coefficients could be used to quantify the net effects;16 however, the individual contributions of mutualism and competition would be lost, making it difficult to translate the relationship into a mechanistic model.
Figure 3. Competition between mutualists. (A) A net effects diagram. M1 and M2 are simultaneously engaged in mutualism (+/+) and competition (–/–). The true net effects could be either positive or negative, depending on whether mutualism or competition dominates, respectively. (B) A resource-based diagram. The core mutualism is the same as in Figure 1B. There is now also a third resource (R3) that is consumed by both M1 and M2. (C) Asymmetric competition between mutualists. M1 is a better competitor for the shared resource and interferes with consumption of R3 by M2. While not explicitly shown, competition for R3 could change production of the mutualistic resources, R1 and R2. For the resource-based diagrams, (B, C), arrows represent resource production (black arrows), resource consumption (red arrows), increased resource consumption (thick arrow), and interference with resource consumption (dashed inhibition arrow).
Download figure to PowerPoint
These interactions are much more effectively captured once resources are illustrated explicitly, as seen in Figure 3B. Now it is clear that M1 and M2 interact mutualistically via resources R1 and R2, exactly as in Figures 1B and 2B. What is different here is that there is a third resource, R3, for which M1 and M2 compete. To frame the group foraging example described above in these terms, investment of time and energy into predator vigilance might be the resource of exchange. Indeed, the resources R1 and R2 are in this case the same thing. R3 in this example might be a shared food resource.
The central question for understanding the persistence of mutualism in this scenario is when the magnitude of competition will outweigh the magnitude of mutualism or vice versa, and, if competition is stronger, what the fate of mutualism is likely to be. Clearly, we cannot project the evolutionary consequences for mutualism in an interaction network like this without explicitly considering and measuring competition.
Figure 3C illustrates one way in which competition could be manifest. In this case, M1 is the superior competitor for the shared resource, suppressing M2's use of it. When framed this way, the question about the persistence of mutualism becomes refocused as a question about the persistence of M2. Will the detriment M2 experiences from its reduced access to R3 outweigh the benefit it receives from M1, via the mutualistic component of their interaction? And, how will these combined effects feed back on M1? Competition could logically lead loose mutualisms of this type to dissolve as predicted in a model of Ranta et al.48
Relevant empirical data on these questions are few. Hino found that five of six bird species studied changed their foraging behavior when in mixed-species flocks compared to when foraging alone.49 Interestingly, feeding rates were higher in mixed flocks. Although this may indicate an absence of competition for food, and in fact an increase in food availability, the authors point out that it could also be an effect of kleptoparasitism or social learning, either of which could be the result of intense competition among species.
As in all previous cases we have described, our figures capture some but not all of the complexity of how competition and mutualism can interact. In Figure 3B, competition is for a resource extrinsic to mutualism (R3). However, competition may also occur between M1 and M2 for mutualistic resources (R1 and R2). For example, two studies in marine habitats have found that two fish species collaborate to locate food, but then appear to compete to consume it.47,50 To capture these and other complex competition–mutualism interactions (e.g., Ref. 51), some important modifications to our figures would be required. The ones we show, however, provide a starting point for how to conceptualize these phenomena.
Competitive advantage as a benefit of mutualism
The best studied benefits of mutualism are transportation (e.g., of pollen by pollinators), protection (e.g., of aphids by ants), and nutrition (e.g., of plants by Rhizobium bacteria). However, other mutualistic benefits are well documented. Among these are beneficial alterations of a partner's competitive environment, and this is the final intersection of mutualism and competition that we will consider. Here are two empirical examples. Hartnett et al. explored how competitive interactions among prairie plants might be mediated by mutualistic mycorrhizal fungi.52 They demonstrated that competitive dominance of one grass species, Andropogon gerardii, depends on it having access to mycorrhizae. Thus, in this case, a mutualist confers traits that give its partner a competitive advantage. As another example, Stachowicz and Hay studied interactions between herbivorous crabs and the coralline algae upon which they live.53 They showed that crabs feed upon fouling seaweeds that, if unchecked, would overgrow the coralline algae; the algae provide a place for crabs to live. In this case, then, a mutualist actively interferes with a competitor to the partner's advantage.
The net effects figure shown in Figure 4A summarizes interactions of this general type. Note that in this case, there might or might not be a mutualism between M1 and M2 in the absence of the competitor C. With reference to the two examples above, plants and mycorrhizae are likely to be mutualists even in the absence of competitors as there are other benefits of this interaction. However, the crabs and algae studied by Stachowicz and Hay would likely not be.53 Such context dependency—that is, a mutualistic outcome that occurs only in a limited set of environments—is almost impossible to capture in a net effects-based figure such as Figure 4A. A resource-based figure permits us to do this. Furthermore, it allows us to recognize important differences between phenomena exemplified in these two empirical cases and to consider how competition may function in each of them. For this reason, we treat them separately below.
Figure 4. Competitive advantage as a benefit of mutualism. (A) A net effects diagram. The core mutualism is the same as in Figure 1A. A third species (C) has been added that competes (–/–) with M1. M2 interferes (–) with C, thus giving an indirect benefit to M1, in addition to any direct benefits of the core mutualism. (B) A resource-based diagram. The core mutualism is the same as in Figure 1B. Additionally, M1 consumes a resource (R3) that is also consumed by C. (C) Competitive advantage is a secondary benefit of mutualism. By consuming R2, M1 becomes a superior competitor for R3 and is able to interfere with consumption of R3 by C. (D) A resource-based diagram of context-dependent mutualism. M1 produces R1, which is consumed by M2. However, M2 does not produce any resource that can be consumed directly by M1 (R2 has been removed from the core mutualism). Instead, the benefit provided by M2 is context-dependent and requires the presence of C. (E) Competitive advantage is the only benefit of mutualism. M2 is an antagonist of C and interferes with its consumption of R3. Consequently, M1 is able to increase consumption of R3. For the resource-based diagrams, (B–E), arrows represent resource production (black arrows), resource consumption (red arrows), increased resource production or consumption (thick arrows), and interference with resource consumption (dashed inhibition arrows).
Download figure to PowerPoint
Figures 4B and C illustrate the case in which one, but not the only, benefit of mutualism is the suppression of competitors. In Figure 4B, we again see the same core mutualism as in Figures 1B, 2B, and 3B. Added to it is a competitor (C) that shares a different resource (R3) with mutualist M1. (If C shared R1 or R2 with a mutualist, we would consider it to be an exploiter of the M1–M2 mutualism, and the scheme shown in Fig. 2B would be more appropriate.) As in Figures 1B, 2B, and 3B, mutualist M1 gains a direct benefit from consuming R2. In addition, as illustrated in Figure 4C, consuming R2 gives M1 a competitive advantage over C for the shared resource R3. The example provided by Hartnett et al.,52 described above, fits this scenario. Note again that it is not M2 (mycorrhizae) that suppress the competitor; M1 (the plant) does this, but only when their mutualists M2 are present. Competitive suppression is thus an indirect benefit provided by mycorrhizae accompanying the direct benefits of nutrient provision. Evidently, it can be extremely important in its own right, however. For example, Wilson and Hartnett show that community-scale plant diversity may be increased if mycorrhizae augment a subordinate species’ performance in competition with a dominant one.54
Figure 4D illustrates the case when an interaction is mutualistic only in the presence of a competitor, that is, when alteration of the competitive environment is the only benefit of a mutualism. In this case, there is no core mutualism resembling those in Figures 1B, 2B, and 3B. Here, one species (M2) consumes a resource (R1) provided by a partner (M1), but there is no reciprocal benefit (i.e., there is no resource R2). As in Figure 4B, the resource-providing partner M1 competes for another resource (R3) with a competitor C. In Figure 4D, the relationship between M1 and M2 is not mutualistic: these interactions are often referred to as commensal (i.e., +0 rather than ++) or facilitative, or sometimes more generically as “positive interactions” (e.g., Refs. 55 and 56). Alternatively, if R1 is costly to produce and no benefit is returned for its provision, then this interaction could be antagonstic (+–). It can become mutualistic, however, via the mechanism illustrated in Figure 4E. Here, M2 alters the environment in a way that shifts the balance of competition between M1 and C in favor of M1. This scenario matches that described by Stachowicz and Hay,53 in which crabs benefit from coralline algae (they are provided with a substrate on which to live and feed), but algae only benefit from the crabs when fouling seaweeds are present and crabs remove them.
Context-dependent outcomes like this one are now widely recognized as one of the most ubiquitous ecological features of mutualism, regardless of their natural history.3 The evolutionary implications of context dependency, in contrast, have barely been considered (but see Ref. 57). Figure 4 clearly suggests that understanding when mutualisms and mutualistic outcomes arise may depend upon documenting the competitive environment in which they occur.