Search for a command to run...
Understanding forces that structure plant communities is a key issue in plant community ecology. Traditionally, ecologists have focused on the absolute and the relative roles of competition and herbivory (Sih et al., 1985) and on the relative role of dispersal limitation vs biotic interactions (Huston, 1999), as well as on neutral models (Hubbell, 2001) and on their niche-based counterparts (Cadotte et al., 2008). Increasingly, mycorrhizal ecologists have argued that mycorrhizal fungi should also be considered as fundamentally important determinants of plant community composition. This conviction is motivated by a growing body of research which demonstrates that mycorrhizal fungi can have large effects on plant growth and may modify the strength of other interactions (van der Heijden, 2002), which is taken as proof that mycorrhizas should have a strong effect on plant community structure. We agree that the potential influence of mycorrhizal fungi on structuring plant communities has been conclusively shown. However, the realization of such potential under variable environmental conditions and complex biotic interactions, and the relative influence of the mycorrhizal symbiosis compared with other biotic and abiotic drivers, have only been addressed in a few studies (see Table 1) and remain uncertain for most systems. Interestingly, we find no studies where the relative influence of ectomycorrhizal or ericoid mycorrhizal fungi on plant community structure has been quantified. This may reflect the difficulty of measuring plant community outcomes (e.g. plant community diversity or evenness) in these symbioses, which tend to be formed by long-lived plants. Nonetheless, the widespread failure of ectomycorrhizal plants to establish in the absence of ectomycorrhizas suggests that the effect size of ectomycorrhizal fungi on plant community structure is likely to be large. Quantification, in this case, may only be possible through the creative utilization of individual-based models of forest community dynamics (e.g. SORTIE; Pacala et al., 1996). Similar debates about the relative importance of predation and competition in structuring plant communities have been addressed through syntheses and meta-analyses of published experimental studies (Sih et al., 1985; Gurevitch et al., 1992). These approaches have shown that for terrestrial plants, herbivores have more frequent and stronger effects on community structure than interspecific competition. Currently there is an insufficient number of published studies on the relative contribution of the mycorrhizal symbiosis (compared with other factors) to attempt a meta-analytical approach. Three general approaches can be used to increase our understanding of the role of mycorrhizal symbioses on plant community structure: Approach 1: The proportion of the variance in plant community composition that is attributed to the presence and composition of mycorrhizal fungi is a useful measure that would allow us to evaluate the mycorrhizal contribution to plant community composition. This requires the manipulation of mycorrhizal fungal communities in the field, either with fungal-addition treatments in landscapes where the fungi have been severely disturbed (Mendes et al., 2010), or with fungal-removal treatments using fungicides (Helgason et al., 2007), rotated cores with nylon mesh windows (Johnson et al., 2001) or the repeated growth of nonmycorrhizal plants (Stinson et al., 2006). It is noted that, similarly to other field manipulations, mycorrhizal field experiments also have their limitations. However, the concern of not having the perfectly controlled experiment should not deter us from using this approach. Similar constraints exist for manipulations of other biotic groups (e.g. competitors and herbivores), yet we have a much better understanding of the influence of these other groups only because of the long history of experimentally manipulating them in the field. Manipulations of mycorrhizal fungi should also be conducted across environmental gradients because many abiotic factors (e.g. temperature, moisture and fertility) are known to interact strongly with mycorrhizal diversity and functioning. In addition, other creative methods may be possible, which do not necessarily add fungi to a mycorrhizal-free soil substrate or attempt to completely eliminate fungi. Rather it may be useful to manipulate their composition and diversity in the field. For example, Opik et al. (2010) showed that there are probably two general categories of arbuscular mycorrhizal (AM) fungi: globally distributed generalist culturable taxa; and narrowly distributed specialized unculturable taxa. Therefore, it would be possible to increase the abundance and diversity of AM fungi and to manipulate the presence of generalist/specialist AM fungi in the ecosystem by culturing generalists and directly extracting specialists from the field. For example, in glasshouse experimental systems, inocula from arable and diverse natural ecosystems can be used, which may yield different fungal diversities and different proportions of generalist/specialist AM fungi. Similarly, in the field it would be possible to add either generalist culturable fungi (and thus decrease the share of specialists in diverse natural communities) or inocula from diverse natural ecosystems. Approach 2: To further determine the contribution of mycorrhizal symbioses relative to other interacting biotic and abiotic factors, experimental complexity will have to increase through the use of multifactorial designs. This will require manipulating the presence and composition of mycorrhizal symbionts along with other factors. We recommend that studies should first focus on comparing the effects of mycorrhizas with other biotic factors that have already been shown to have a major influence on plant communities (e.g. herbivory, competition and pathogens). As a follow-up, we then suggest extending such studies across environmental gradients (e.g. by conducting similar studies under different abiotic conditions, as in Approach 1). Such studies are easier to conduct in the glasshouse but need to be compared with patterns in the field. Several studies have taken this approach. For example, Grime et al. (1987) compared the effects of soil heterogeneity, grazing and mycorrhizal symbioses on plant diversity and found that the presence of mycorrhizal fungi explained a greater proportion of the variance in plant diversity than the other factors combined. Vogelsang et al. (2006) went further by manipulating the composition and diversity of mycorrhizal fungi, as well as soil phosphorus availability, and found that fungal composition and diversity explained a much greater proportion of the observed variance in plant diversity. Other studies have also taken a similar approach of studying the effect of mycorrhizal fungi in the context of other factors (see Table 1), and the proportion of variance explained by the effect of mycorrhizas ranges from 0 to > 50% (Fig. 1). One limitation of these glasshouse manipulation studies is that any observed effects may not scale to the field (Read, 2002), so we recommend that simultaneous glasshouse–field comparisons are also made. For example, Pringle & Bever (2008) set up parallel studies of plant response to mycorrhizal fungi in growth chambers and in the field and found that mycorrhizal effects were correlated between the two environments. It should also be emphasized that because mycorrhizal fungi are difficult to control in the field, they may not be consistently eliminated, either spatially or temporally. Thus, variance in field experiments may be underestimated. Parallel glasshouse–field analyses are also encouraged because some of the confounding factors can be set up as treatments in the glasshouse to yield more accurate estimates of variance. Percentage variance in plant community composition that is explained by competition among plants vs herbivory (Sih et al., 1985). We propose that a similar analysis using mycorrhizal fungi would be very informative (data from Table 1 indicate that mycorrhizal symbioses could explain 0–57% of the variance in plant community structure). Approach 3: Finally, inferences as to the relative importance of forces of plant community structure are derived from monitoring these potential forces, as well as plant community composition over time. This approach takes advantage of variation in plant and fungal community composition in the field, and uses measures of potential forces at time 1 as predictors of plant community composition at time 2 using the analytical framework of structural equation modelling (Shipley, 2002). This method is based on using natural or experimentally driven variation, or patterns through time, while measuring different variables that relate mycorrhizal abundance to plant community changes, via different mechanistic (direct and indirect) pathways. This approach could be better informed by analyses that explicitly compare the strength of different pathways (e.g. using pulse-label experiments and subsequent isotope analyses; or using the output of simulation community models). We recommend that the scientific community carefully choose good model field sites to focus on for long-term study, especially sites that may have fewer problems with confounding variables. In most cases, species richness and evenness have been the main metrics of plant community responses measured following mycorrhizal manipulations (Table 1). Further consideration should be given to other robust metrics that can be used to compare community outcomes across systems. These might include total productivity, ecosystem resilience and plant functional group composition (Lavorel et al., 1997). In conclusion, we recommend a redirection of mycorrhizal studies in plant community ecology, departing from the already answered question of whether mycorrhizal symbioses can affect plant community structure, and move on to the questions of how much they affect community structure under different conditions and in different systems. This requires an approach that not only centres on more complex experimental and survey designs, but also evaluates the proportion of variation explained by the other major factors. As in many other areas of ecology, it would help to move beyond the Fisherian approach of significance testing (noting the P-value; the effect is/is not significant) to ascertaining the variability explained by the components of more complex combinations of factors. In ANOVA terms this amounts to evaluating the Sum of Squares (e.g. Morin et al., 1988), but other, more complex, techniques are available (e.g. partition of variances of dissimilarities in Permutational Multivariate Analysis of Variance (Anderson, 2001), or Structural Equation Modelling (Shipley, 2002). In addition, there should be concerted efforts to replicate studies across host plant communities, environmental gradients and in a wide variety of ecosystems and biogeographical regions. Several ecology studies show that the importance or the intensity of biotic interactions vary between ecosystems, often in a systematic way (Sih et al., 1985; Bertness & Ewanchuk, 2002). For mycorrhizal studies, soil fertility may be an important determinant of the strength of mycorrhizal effects (Johnson, 2009), but studies across other environmental gradients may provide important insights on the relative contribution of mycorrhizal symbioses to plant community composition. In doing so, we would also recommend for studies to report a consistent measure of variables that are likely to affect the functioning of the symbiosis (such as soil nutrient levels and species identity). Relative importance might also be context dependent, being captured as a statistical interaction term among different factors. For example, the effect of mycorrhizal fungal community composition on plant diversity was dependent upon phosphorus addition (Vogelsang et al., 2006). Although quantification of relative influence is a significant challenge, we argue that only by doing so will mycorrhizal effects be integrated into a comprehensive understanding of plant community ecology. There are also important applied implications: only proper quantification of impact can justify economic investment into managing mycorrhizal symbioses in restoration, agriculture or forestry. While quantification raises the spectre that the mycorrhizal symbiosis may not be a significant driver of plant community dynamics in some or many ecosystems, even this would be useful in allowing mycorrhizal ecologists to refocus efforts. This collaboration was supported by the Australian Research Council-New Zealand Research Network for Vegetation Function (Working Group 61 – Plant–Soil Microbes Feedbacks). J.N.K. was supported by an NSERC Discovery grant, J.D.B. was supported by NSF grants DEB-0616891, DEB-0919434 and NIH-5 R01 GM092660, I.A.D. was supported by Ecosystem Resilience OBI of the FRST of NZ, M.C.R. was supported by DFG Priority Program 1374 Biodiversity Exploratories, A.M.K. was supported by the SNSF (grant PAOOA-119519), and M.M. and M.Z. were supported by ETF grants 7371, 7366 and Center of Excellence FIBIR.