This dissertation focused on the hypothesis that plants choose the species composition of their associated mycorrhizal fungi as a part of their fitness-maximizing foraging strategy. The premises were simple: if the species of mycorrhizal fungi vary in effectiveness, and the plant is able to discriminate among them, then the plant should choose the composition that supports the highest plant growth. Moreover if the relative effectiveness of the fungi changes with environmental condition, then at each set of environmental conditions the plant should select a different composition of fungi. This hypothesis, first presented by Janos (1985), provides a mechanism for structuring communities of mutualists both within and between environments. Alternative mechanisms that may structure mycorrhizal fungal populations on individual plants include fungal competition and simple differences of rates of growth among the different fungal species. Fungal competition can cause differences in mycorrhizal fungal community structure along a light-intensity gradient only if fungi differ in their relative competitive abilities and the resulting competitive hierarchy changes with changes in light-intensity to the host plant. Likewise, differences in fungal growth rates can cause changes in mycorrhizal fungal community structure along a light-intensity gradient only if the growth-rate hierarchy among the fungi changes with changing light to the plant. Both of these mechanisms could potentially structure the mycorrhizal fungal community so that the most abundand fungi are the least beneficial to plant growth. This negative impact of mycorrhizal fungal community structure on plant growth would be a result of either a competitive ability - mutualistic effectiveness trade off (Hoeksema and Kummel, 2003), or a growth rate - mutualistic ability trade-off (Bever, 1999). The relationship between a plant and a mycorrhizal fungus can be characterized by an exchange of carbon sequestered by the plant for nutrients taken up by the fungus (Smith and Read 1997). Thus the relationship between a plant and its associated mycorrhizal fungi can be thought of as a biological market (Noe and Hammerstein 1994; Schwartz and Hoeksema 1998), where plants and fungi exchange resources, choose trading partners, advertise to one another, etc. This rich and complex context called for a thorough, formal, mathematical analysis of the plant choice hypothesis. The central question was: Can partner choice in the context of carbon for nutrient trading lead to coexistence among mycorrhizal fungal species, structure communities of mycorrhizal fungi within an environment, and cause changes in the community structure between environments? To address this question Stephen Salant and I wrote a model, presented in Chapter 2, where the growth of a plant depends on the carbon and nitrogen the plant has available after trading with its mycorrhizal fungal partners. The fungal partners are represented by their exchange functions, which determines how much carbon the plant has to relinquish as a payment for any given amount of nitrogen received from the fungus. The plant chooses the composition of its fungal trading partners and the amount of carbon allocated to each fungus so that the growth of the plant is maximized. The results of the analysis were dependent on the shape of the exchange function. For the most biologically realistic exchange functions, where the carbon payment on an additional unit of nitrogen increased with increasing amount of nitrogen received from the fungus, partner choice in the context of C for N trading allowed for species coexistence of mycorrhizal fungi, structured fungal communities through differential carbon allocation to each mutualistic fungal partner, and caused predictable changes in fungal community structure with increasing light availability to the host plant. The plant allowed fungal coexistence, structured fungal communities and changed fungal community structure by equating marginal costs of trading with the fungal partners. The basic prediction of ating marginal cost, and thus, by extension, the consequences of this to the fungal community, were upheld even when we allowed multiple plants connected by the same fungal mycelia to simultaneously choose their trading partners. Given the solid theoretical foundation of the plant choice hypothesis I turned to the field and examined whether patterns of the association between a plant and its mycorrhizal trading partners along a light intensity gradient were consistent with the plant choice hypothesis. My model system was the association between seedlings of balsam fir and Cenococcum and Lactarius , the two most abundant ectomycorrhizal fungi that associated with balsam fir seedlings in the Northern Hardwood forests of Sugar Island (Upper Peninsula of Michigan). Consistent with the plant choice hypothesis I found that species composition of mycorrhizal fungi were both correlated with and caused by the light availability to the seedling: relative abundance of Cenococcum decreased and relative abundance of Lactarius increased with increasing light availability. The result of a selectivity study was also consistent with the plant choice hypothesis: Cenococcum on shaded seedlings was over-represented, relative to its potential availability, where as Cenococcum on seedlings in higher light was variably both over-and under-represented. In contrast, in three of four studies, the relationship between plant growth and fungal abundance was inconsistent with the plant choice hypothesis: contrary to the prediction, seedlings dominated by Lactarius had higher than average rate of growth in low light, and lower than average rate of growth in high light. Seedlings dominated by Cenococcum did not differ from average in either low or high light. The results of these field studies presented a surprising contrast between the consistently close relationship between fungal community structure and light availability and the consistently non-adaptive relationship between fungal community structure and plant growth. This contrast can be interpreted in two ways. Either the plant is choosing fungi to increase some component of fitness other than growth, or mechanisms other than partner choice play a dominant role in structuring the fungal community. The two alternative mechanisms that could account for the field patterns were environment specific fungal competition, and environment-specific differences between growth rates of non-interacting fungal populations. These alternative explanations made it imperative to perform an experiment designed at distinguishing the mechanisms. The role that plant choice and the two alternative mechanisms play in determining the community structure of mycorrhizal fungi along a light gradient was addressed in a mesocosm (competition? experiment between Cenococcum and Lactarius. The data suggested that the community structure was determined by an interplay between fungal competition, fungal facilitation, and fungal growth-response to light availability of the host. In particular, in low light Cenococcum was facilitated by Lactarius, and Cenococcum inhibited Lactarius. However, in high light Cenococcum was inhibited by Lactarius, and Lactarius was inhibited by Cenococcum. Furthermore, a positive growth response of Lactarius to light availability caused increased abundance of Lactarius in high light. The results of fungal abundance in Cenococcum+Lactarius treatment were qualitatively identical to the field pattern: Cenococcum abundance was higher in low light, where as Lactarius abundance was higher in high light. Therefore, the results from this experiment are directly applicable to the field. The corroboration between this experiment and the field studies presented in Chapter 3 constitutes the first demonstration of a causal link between fungal interactions and fungal community structure in the field. The main evidence against the mechanism of plant choice in this experiment was the observation that Cenococcum switched from being a mutualist when in low abundance to being a parasite when in high abundance, and that l nts were not able to stop associating with Cenococcum at the abundance levels where Cenococcum was clearly parasitic. Also inconsistent with the plant choice hypothesis was the observation that the abundance at which Cenococcum started to behave like a parasite was lower in low light, where Cenococcum is in high abundance in the field. Lastly, inconsistent with the plant choice hypothesis was the non-significant negative association between fungal dominance in polyculture (and in the field) and fungal effects on plant growth in fungal monoculture: in low light, where Cenococcum dominated in the polyculture, Cenococcum supported lower plant growth than Lactarius in low light. However, in high light, where Lactarius was dominant in the polyculture and in the field, Lactarius supported lower plant growth than Cenococcucm. These results indicated that fungal competition was the most likely candidate mechanism behind the observed negative fungal interactions. However, the theory of fungal competition (Hoeksema and Kummel, 2003) predicts a trade-of between competitive ability and mutualistic effectiveness. In this study, there was no relationship between the ability to suppress other fungal species and the ability to support high plant growth. The effects of fungal composition on plant growth constitute only an indirect test of presence or absence of partner choice or competition. A split-root experiment that would either allow or not allow interaction of the fungi in the soil would provide a more direct test for the presence of fungal competition in the soil. Likewise a test for equalization of common marginal cost when dealing with several fungal species would provide the most directtest of the plant choice hypothesis. The complex nature of the results warrants further investigation, some of which is already under way. My ongoing and future research concentrates in four areas: 1) If the plant does not optimize fungal composition, does it at least optimize the use of the fungi once they colonize the plant? The most direct test of optimal use is the test of the ability of a plant to equalize marginal cost when trading with several species of fungi. I am planning this experiment, in collaboration with Stephen Salant and Jason Hoeksema, described in detail in chapter 2. Currently we are at the end of the preparatory phase. 2) What is the role of the plant in fungal competition? To address this question I have established a competition experiment between Cenococcum and Lactarius on agar at different glucose concentrations. This simulates the carbon availability along light gradient while taking the plant out of the system. The experiment will be harvested within two months. 3) How does plant choice interact with fungal competition? Jason Hoeksema and I have addressed this question theoretically in a patch-occupancy model (Hoeksema and Kummel 2003) and found that plant choice could lead to coexistence between strong competitors who are weak mutualists and weak competitors who are strong mutualists. I plan to address the question of the interaction between fungal competition and plant choice experimentally using a split-root design with permeable and impermeable barrier between the compartments. 4) The relationship between fungal community dynamics and plant performance. My further work on this topic will include finishing a study of nitrogen uptake with respect to fungal community structure in the field, and starting a new field study to determine the relationship between fungal community composition and plant survivorship.