Context dependency, the concept that outcomes of biological interactions shift in sign or magnitude depending upon genetic, abiotic, and biotic context, is core to understanding how interspecific interactions function and how they influence community dynamics and stability. In 2004, we published a paper examining the potential for context dependency in bark beetle-fungus symbioses (Klepzig and Six, 2004). Our review was stimulated by Bronstein’s (1994) paper wherein she posed several key questions for establishing when context dependency occurs. At the time of our paper, massive gaps in knowledge existed in how bark beetle-fungus symbioses functioned and we were only able to touch upon some of these questions. However, in the 17 years that have passed, an increase in interest in these systems, in part driven by massive climate-driven bark beetle outbreaks, has resulted in an explosion of new information. Concurrent with the expansion of literature on bark beetle-fungus symbioses, a growing recognition of the importance of mutualism in shaping biotic communities led to a greater understanding of context dependency in mutualisms in general. Consequently, we felt the time was right to revisit the degree to which context dependency operates in these ecologically and economically important insect-fungus interactions.

There is an increasing recognition that bark beetle (Curculionidae; Scolytinae)-fungus (primarily Ophiostomatales) symbioses vary in specificity and outcome and we now know that these differences play a major role in the extent to which they influence and shape the ecosystems within which they exist (Six, 2020a). Some, like the aggressive tree-killing Dendroctonus ponderosae-fungi mutualism (mountain pine beetle) are major drivers of ecosystem structure and function, while non-aggressive secondary beetles and fungi that colonize dying and recently killed trees play less influential roles (Six, 2020a). However, while the beetles are recognized to exhibit a broad range of behaviors, the fungal partners have often been reduced to simple binary categorizations as pathogen/non-pathogen or mutualist/antagonist, and their influence on the host beetle has been studied overwhelmingly in the context of the initial colonization phase of the tree and in tree defense induction. However, the longest period of interaction between a beetle host and its associated fungi occurs over the several months to year-long or longer period following attack. During this time, the fungi express a variety of characteristics that ultimately drive the outcome of the symbiosis. Variation in these characteristics and in the different ecological strategies of the fungi influences the type and strength of the symbiosis, the diversity of partners, and, as we shall discuss, the potential for context dependency.

Our treatment of context dependency in bark beetle-fungus symbioses is split into three sections. The first provides a brief review of what constitutes context dependency, the factors expected to influence it, and the evidence for (or against) it in mutualisms. In the second, we present descriptions of the various types of bark beetle-fungus symbioses and discuss evidence for or against context dependency in their outcomes. In the third, we identify remaining gaps and suggest avenues for further inquiry.

Throughout, we use symbiosis as a general term for organisms living in close association with one another regardless of whether outcomes are positive, neutral, or negative. We use the terms mutualism, commensalism, and antagonism to describe symbiotic outcomes in recognition that some symbioses may shift between these categories and that a strict designation of association type for some can be misleading (Bronstein, 1994). We also limit our treatment to the bark beetle-fungus symbioses that colonize conifers because their associations with fungi are distinct from those of angiosperm-colonizing species and because they are better studied.

Context dependency occurs when interaction outcomes shift in magnitude (effect size) and/or sign (+, 0, −) as a function of abiotic, biotic, or genetic context (Chamberlain et al., 2014). We follow here the broadly accepted definition provided by Hoeksma and Bruna (2015) that context dependency is a change in interaction net outcome; in other words, how one species influences the mean fitness correlates of another species. We also follow their reasoning that context dependency should be applied only to interactions between a particular pair of species and exclude changes in the performance of one species when it is paired with an alternate partner species. Thus, if one species pairs with more than one partner, each pairwise interaction should be considered independently. This appropriately restricts comparisons to those occurring within interactions rather than among interactions involving third-party symbionts or hosts although these can, and should be, considered under the umbrella of biotic context (see below).

Context dependency has often been assumed for mutualisms. In fact, mutualisms are routinely viewed as existing on an interaction continuum where outcomes slide between mutualism, commensalism, and antagonism (including parasitism and competition) as conditions (contexts) shift (Bronstein, 1994; Thrall et al., 2007). This view has been reinforced by a large body of research on mycorrhizae and legume-rhizobia and other mutualisms involving nutrient exchange (Hoeksma and Bruna, 2015). However, it is increasingly clear that many mutualisms do not operate on such continua (Lam and Chisholm, 2019). For example, by-product mutualisms, those where benefits accrue as part of the normal function of a partner without additional cost to the provider, and many highly specific obligate mutualisms either do not have alternate states or are highly constrained from shifting to commensalism or antagonism (Bronstein, 1994; Connor, 1995). One of the reasons sign shifts are seldom seen in mutualisms is that when there is a shift to antagonism this is quickly followed by the dissolution of the partnership or lineage extinction (Frederickson, 2017). Furthermore, even in mutualisms that lend themselves to context dependency, shifts in sign may not be as prevalent as was once believed (Karst et al., 2008; Chamberlain and Holland, 2009). A meta-analysis of studies on mycorrhizae found that fungal partners had routinely positive effects on plant growth regardless of shifts in soil nutrient availability (Karst et al., 2008). Another meta-analysis focusing on ant-plant protection mutualisms also found routinely positive effects for plants and only occasional shifts to neutrality (Chamberlain and Holland, 2009). In a broad comparison of mutualisms with plants, Morris et al. (2007) found that the mean effects of pollinator, bacterial, and arbuscular and ectomycorrhizal mutualisms seldom became negative or neutral and generally remained positive.

While shifts in magnitude rather than sign have been more commonly observed, certain types of mutualisms appear to lend themselves more to sign shifts than others. These include plant endophytes and secondary endosymbionts involved in protection mutualisms with insects (Hansen et al., 2007; Oliver et al., 2008). Some endophytic fungi in plants can shift from mutualistic to antagonistic depending on endophyte and/or plant genotype and environmental conditions (Faeth and Fagan, 2002). For example, shifts in soil nitrogen can influence whether endophytes are helpful in defending against herbivores or shift to parasitism (Faeth and Fagan, 2002). Facultative bacterial endosymbionts can increase host plant range, tolerance to heat stress, and provide protection against natural enemies for a number of insects (Oliver et al., 2008; Burke et al., 2010). However, the frequency of some protective secondary endosymbionts is high only when the host insect is under strong pressure but drops drastically when pressure is low suggesting that the symbionts become costly when protection is not needed (Hansen et al., 2007; Oliver et al., 2008).

To determine if context dependency occurs, it is necessary to quantify shifts in the costs and benefits of the interaction on both partners as abiotic, biotic, and genetic contexts vary. Unfortunately, explicit tests have seldom been applied outside of mycorrhizae, Rhizobium-legume, ant-plant, and pollinator systems (Hoeksma and Bruna, 2015). Additionally, such cost-benefit analyses are notoriously difficult in many mutualisms. However, particular characteristics of a mutualism provide clues as to whether and when context dependency may be expected and whether a sign change is involved or only changes in the effect size of benefits. Mutualisms are less likely to be context dependent when the environment is predictable, when symbiont richness is low, there is high symbiont reliability, and high quality rewards (Thrall et al., 2007). In particular, predictable and stable environments are thought to lead to increased closeness due to greater predictability of potential partners (Thrall et al., 2007). This, in turn, may lead to selection for increases in benefits over time and a greater potential for obligacy, especially if one or both partners provide access to highly-limiting resources or otherwise alter niche space in a way required for survival of one or both (Buser et al., 2014). On the other hand, an inability to consistently associate with a particular partner, either due to a diverse symbiont pool, high environmental variability, or both, results in a facultative symbiosis and reduces the potential for co-evolution, partner specificity, and dependence (Thrall et al., 2007).

Whether a mutualism is obligate or facultative is a major factor influencing its potential to be context dependent. Obligate mutualisms are constrained from shifting sign because a switch to commensalism or antagonism causes massive reductions in fitness and the extirpation of one or both partners. For obligate mutualisms, context dependency is more likely to be expressed via shifts in the effect size of benefits. For example, in a mutualism between a stinkbug and a bacterial symbiont, the symbiont is required for development to adulthood and the symbiosis never switches from positive to neutral or negative because this would result in the death of both partners. However, particular components of the biotic environment such as the species of host plant (potentially through differences in secondary chemistry or nutritional quality) can influence the fitness of the symbiont such that it cascades to affects the growth rate of the host bug (Couret et al., 2019). In contrast, some facultative symbionts may be beneficial in some contexts but neutral or antagonistic in others. These switches in outcome signs do not lead to extinctions because no partner is dependent on the other for survival.

Transmission mode also influences the potential for context dependency. Vertically-transmitted symbionts are the most likely to be specific and obligate and their transmission often involves specialized structures or behaviors that reinforce contact (Six, 2012; Fisher et al., 2016). On the other hand, horizontally-transmitted symbionts are often facultative and with substantial potential for hosts to make new acquisitions or replacements. There is also a tendency for horizontally-transferred symbionts to be antagonistic (pathogens, parasites, and competitors), while those that are vertically transmitted tend toward mutualism because symbiont fitness is dependent on host fitness (Yamamura, 1993; Herre et al., 1999). Seldom addressed in symbiosis theory are mutualisms that involve both modes of transmission or that commonly involve intra-specific and inter-specific host-switching. For example, Microtermes termites primarily transmit their fungal symbionts vertically yet frequent horizontal exchange occurs between colonies of the same species and even different species (van de Peppel and Aanen, 2020). Likewise, while some ambrosia beetles exhibit strong partner fidelity with, and vertical transmission of, particular ambrosia fungi, some may acquire fungi directly from their environment and possess different symbionts in different locations (Kostovcik et al., 2015; Rassati et al., 2019). In many cases where both vertical and horizontal transfers occur, there are taxonomic constraints such that symbionts shared among related hosts are also closely-related (Skelton et al., 2019). For instance, host specificity of symbiotic fungi associated with termites in the Macrotemitidae occurs at the genus level but not at the species level (van de Peppel and Aanen, 2020). In such cases, switching may be supported by common needs among hosts that can be provided by a number of related symbionts. Such a “semi-open system” operates as long as mechanisms exist for recognition of “insiders” and avoidance of “outsiders.”

Transmission mode also influences the evolution of costs and benefits to each partner and the potential for context dependency. In general, hosts are predicted to evolve reduced vertical transmission of symbionts that act as antagonists and increased vertical transmission of those that confer benefits (Yamamura, 1993). However, environmental context can play a major role in whether symbiont specificity and vertical transmission develops, even when an interaction results in a strongly mutualistic outcome. For example, mutualisms that operate across large spatial scales may be less likely to become specific and more likely to be context dependent. The geographic mosaic theory of coevolution predicts that fitness consequences of associations should vary across space and time and that selection will result in variable evolutionary outcomes (Thompson, 1994, 2005; Nuismer et al., 2000). This may especially be the case for mutualisms occupying heterogeneous landscapes. In spatially-structured populations that experience different biotic and abiotic contexts, a symbiont may be a mutualist in one environment and neutral or antagonistic in another, such as with the grass Agrostis hyemalis where the fungus Epichloe amarillans supports greater fecundity in arid locations, but not in areas with wetter conditions, or can decrease grass biomass in the presence of some microbial communities (Davitt et al., 2011; Brown and Akcay, 2018). In some cases, variable environmental conditions may support different symbionts (species or genotypes) in different host populations due to disparities in the needs and growth requirements of hosts, symbionts, or both (Brown and Akcay, 2018). Furthermore, high variability in environmental conditions can result in a variable symbiont pool decreasing the predictability of encountering and developing a specific association with any one symbiont. Such spatially-conditional mutualisms can result in host-symbiont conflict when hosts carrying locally-adaptive symbionts migrate from one population into another (Brown and Akcay, 2018).

Where conditions are stable and shift very little, context dependency in a mutualism is likely to be null or small (Hoeksma and Bruna, 2015). However, when instability reduces the ability of a partner to persist or causes shifts in costs and benefits for one or both partners, context dependency should be more prevalent. Abiotic, biotic, and genetic contexts may all vary to create shifts in symbiotic outcomes. Most symbioses operate under variable abiotic contexts, including, e.g., temperature, moisture, light, climate, and soil nutrient content. Thermal variability, for example, is a common factor influencing interaction strength and outcome. Most organisms are ectotherms and most symbioses involve one or more ectothermic partners. The high sensitivity of ectotherms to temperature can lead to strong and rapid responses in the physiological functions that underlie growth, survival, and the delivery of benefits to mutualist partners (Renoz et al., 2019). Likewise, water is required by all organisms and variability in its form and availability can be another major source of context dependent outcomes (Ruiz-Lozano, 2003). Nutrient acquisition is the driving force behind the formation of most mutualisms and can be influenced by a variety of environmental factors.

In some cases, a mutualism may be largely based on abiotic context such as those that occur between some secondary endosymbionts that confer protection from heat to their insect host (Heyworth et al., 2020). In many other cases, the abiotic context is crucial for aligning partners to function under a common set of conditions. For example, some corals “bleach” when temperatures exceed thresholds for the survival of their dinoflagellate algal symbionts (Cunning et al., 2015). This can be lethal; however, some corals can acquire new heat-tolerant symbionts to replace heat-intolerant ones, allowing the persistence of the symbiosis under new conditions (Cunning et al., 2015). Obligate, co-evolved mutualisms can be expected to possess partners well-aligned to operate under the typical conditions found in their environment, but persistence may be challenged during extreme events or under novel conditions (Six and Bentz, 2007). In contrast, facultative mutualisms are more fluid in their responses to change and may be more open to the acquisition of new partners as conditions shift (Batstone et al., 2018).

Biotic context encompasses all the organisms that the mutualism interacts with directly or indirectly including host plants for herbivores, other symbionts or hosts, predators, parasitoids, and competitors. These are termed third-party effects and they can have substantial consequences for symbiotic outcomes. For example, in ant-plant protection mutualisms with low ant species richness, there is no tendency to be context dependent while those with high ant species richness vary considerably in outcome (Chamberlain and Holland, 2009). With heritable facultative endosymbionts that confer protection to their host insects, coinfection with symbionts that provide different types of defense does not necessarily lead to protection against a broader array of enemies, but rather can result in greater variability in protection or even a reduction such as was seen in pea aphids coinfected by the bacterial symbionts, Hamitonella and Regiella (Weldon et al., 2020). In contrast, in mycorrhizal systems, the presence of more mycorrhizal fungal species and non-mycorrhizal microbes can increase the nutritional benefit accrued by host plants (Hoeksema et al., 2010). In some cases, predators, parasites, or parasitoids may exploit cues produced by one partner to exploit the other resulting in a decrease in fitness (Adams and Six, 2008; Boone et al., 2008). For example (O’Brien et al., 2018), in mutualisms that exploit plants, variability in the production of secondary chemical compounds within or among plant species can influence benefit delivery by microbial partners (Couret, et al., 2019). The range of third party effects that can occur is immense and they are often very important influences on mutualism outcomes.

Another consideration when assessing context dependency is whether cheaters and exploiters exist in the mutualism. Here, we follow the careful distinctions of Frederickson (2013) between the two. A cheater is a genotype or species that exhibits an adaptive uncooperative strategy, is evolutionarily derived from a cooperator, and reduces partner fitness relative to cooperators. Cheaters are distinct from inefficient partners (those that provide lower fitness benefits due to genotype or defect; Friesen, 2012) and parasites or exploiters of mutualisms which are similar in effect to cheaters but are not evolutionarily derived from cooperators (Frederickson, 2017). Cheaters, by definition, cause a net reduction in fitness. If cheating is influenced by context, this may allow for both a change in sign as well as a change in the magnitude of the interaction outcome. However, evidence for the existence of cheating in mutualisms is sparse (Frederickson, 2017). In contrast, exploiters are commonplace. Exploiters can be especially prone to context dependency and, as with other third parties that exert negative pressure on mutualisms, may be particularly powerful in influencing the dynamics of populations of mutualistic pairs.

Outcomes of symbioses can vary substantially across different genotypes of the symbiont and host. If different genotypes respond to abiotic or biotic contexts in disparate ways, this can result in differential provisioning of benefits. Such genetically based variability provides the potential and foundation for directional selection on mutualism traits (Hoeksma and Bruna, 2015; Stoy et al., 2020). Genotype x environment (GxE) interactions support directional selection in one partner, while genotype by genotype (GxG) interactions of host and symbiont result in co-evolution. Finally, genotype by genotype by environment (GxGxE) interactions support the development of geographic selection mosaics. In microbial symbionts, selection over time is expected to reduce genetic variability by removing all but the best partner genotype(s) (Hoeksma and Bruna, 2015). While this has been observed in many co-evolved mutualisms, it is not always the case (van de Peppel and Aanen, 2020). Genetic variation may be maintained, or its loss slowed, if the optimal mutualist partner remains substantially influenced by the abiotic or biotic environment (Bever, 1999). For example, while endosymbionts may lose large portions of their genomes because the host provides protection and genes to support many crucial functions (McCutcheon and Moran, 2011), ectosymbionts must often contend directly with the external environment and may need to retain full or near full genome function although overall allelic diversity may be reduced (Heath and Tiffin, 2007).

Genetic diversity in symbionts should influence their potential to respond to changing conditions in a context dependent manner. Low genetic diversity may result in greater reliability of benefits and a lower potential for cheating, but it also can constrain a host to particular environmental conditions and habitats because symbionts with low genetic diversity will have narrower ecological amplitudes. On the other hand, symbionts with greater genetic variation may allow for greater flexibility in a variable environment, but increase the potential for context dependent responses by reducing reliability in benefit delivery (Heath and Tiffin, 2007).

Finally, an important consideration is that partners are often subject to many contextual factors simultaneously and these can interact to influence outcomes. Clearly, many factors have the potential to influence outcomes in mutualisms.

The consequences of context dependency are not trivial as some mutualisms are powerful drivers of ecosystem structure and function and conditionality of interaction outcomes may have effects that extend far beyond direct effects on partners. While some mutualists are always mutualists and some parasites are always parasites, some symbionts, particularly facultative ones, may fall on a sliding scale. However, there is currently no evidence that bark beetle mutualisms change in sign, only in magnitude, and that the “closest” of these partnerships have evolved mechanisms to reduce context-dependency and stabilize benefit delivery. The bark beetle-fungus symbioses most likely to slide along the mutualism-agonism continuum are those involving facultative symbionts that occur in suites. However, we know little about these and they should be a priority in future work. Our understanding of bark beetle-fungus symbioses will also be improved by studying the “full” complement of fungi found in mutualisms with beetles. Many mutualisms are multipartite and involve two or more mutualist fungi. However, preconceived notions as to their function or value to the host have often led to a biased selection of one fungus over others, even in obligate mutualisms, leaving a gaping hole in our understanding of the mutualism as a whole. There has especially been the case with Type A beetles where studies are often biased toward the most virulent partner or the one best at provisioning nutrients. However, in obligate mutualisms, all partners are co-evolved with the host and are integral to the full function of the partnership.

One reason we lack a better understanding of many of these mutualisms is that they are notoriously difficult to manipulate in controlled experiments. However, the use of molecular community analyses and ecological stoichiometric approaches can help unravel the roles of fungi in providing nutrients and the use of isotopes can further aid in our understanding of where elements that end up in beetles originate. Knowledge of element source and demand can be used in modeling to predict how context dependency influences outcomes and to detect the point at which it destabilizes the system.

While we have come a long way in understanding these systems since our 2004 paper, we still have a way to go. We are also at a time when understanding the nature of both the context and the dependency in these mutualisms is needed more than ever so that we can understand how they will respond to anthropogenic change. Bark beetles and their mutualist fungi co-construct a niche that allows them to exist in a resource that is otherwise intractable or inaccessible. For the closest of these partnerships, this has resulted in some of the most influential agents of forest mortality in conifer forests worldwide (Six, 2020a). These fully ectothermic partnerships are influenced by climate at all levels indicating that shifting climatic context will have a major impact on their function and stability of their dependency, and how they influence forests into the future.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

DS developed the overall conceptual framework and wrote the initial draft. KK provided the feedback and additional development of conceptual arguments and co-wrote the final draft with DS. All authors contributed to the article and approved the submitted version.