No globally consistent effect of ectomycorrhizal status on foliar traits


Author for correspondence:

Nina Koele

Tel: +64 3 321 9700



  • The concept that ectomycorrhizal plants have a particular foliar trait suite characterized by low foliar nutrients and high leaf mass per unit area (LMA) is widely accepted, but whether this trait suite can be generalized to all ectomycorrhizal clades is unclear.
  • We identified 19 evolutionary clades of ectomycorrhizal plants and used a global leaf traits dataset comprising 11 466 samples across c. 3000 species to test whether there were consistent shifts in leaf nutrients or LMA with the evolution of ectomycorrhiza.
  • There were no consistent effects of ectomycorrhizal status on foliar nutrients or LMA in the 17 ectomycorrhizal/non-ectomycorrhizal pairs for which we had sufficient data, with some ectomycorrhizal groups having higher and other groups lower nutrient status than non-ectomycorrhizal contrasts. Controlling for the woodiness of host species did not alter the results.
  • Our findings suggest that the concepts of ectomycorrhizal plant trait suites should be re-examined to ensure that they are broadly reflective of mycorrhizal status across all evolutionary clades, rather than reflecting the traits of a few commonly studied groups, such as the Pinaceae and Fagales.


Plants have evolved different forms of mycorrhizal associations, including arbuscular mycorrhizas, ectomycorrhizas and ericoid mycorrhizas, as important adaptations to assist in nutrient uptake. It is generally believed that ectomycorrhizal plants occur predominantly on low-nutrient soils (Lambers et al., 2008). Read (1991) hypothesized that mycorrhizal types are associated with specific ecosystem and soil attributes, ranging from arbuscular mycorrhizal plants with easily decomposable litter on slightly acidic soils to ectomycorrhizal and ericoid mycorrhizal plants with highly recalcitrant litter on strongly acidic soils. Through a reinforcing feedback between soil conditions and plant traits, Read (1991) suggested that a set of foliar traits should be associated with each mycorrhizal type. For ectomycorrhizal plants, for example, it is assumed that they have long-lived leaves, high nutrient resorption during foliar senescence and recalcitrant litter. In turn, these traits can cause a positive feedback of recalcitrant litter inhibiting the mineralization of nutrients (Read, 1991). Thus, it should be possible to predict foliar traits and soil processes from mycorrhizal status. Supporting this hypothesis, Cornelissen et al. (2001) found that ectomycorrhizal plants were slower growing than arbuscular mycorrhizal and non-mycorrhizal plants based on a dataset from the Sheffield region in the UK. The observations of Read (1991) and Cornelissen et al. (2001) suggest that ectomycorrhizal roots, low leaf nutrients and high LMA (leaf mass per unit area) should occur as a trait syndrome on the slow-return end of the leaf economics spectrum (Reich et al., 1992, 1997; Wright et al., 2004). A large investment of the plant in the ectomycorrhizal symbiosis is needed to obtain nutrients from poor mineral soils and recalcitrant litter. Therefore, it is expected that ectomycorrhizal plants will have lower foliar nutrients than their non-ectomycorrhizal relatives.

One potential difficulty in assessing links between mycorrhizal status and other plant traits is a strong bias in the mycorrhizal literature. Dickie & Moyersoen (2008) noted that more than 80% of the literature on ectomycorrhiza up to 2008 was based on only two plant groups (Pinaceae, 62%; Fagales, 22%). Thus, it is possible that concepts of ‘ectomycorrhizal’ traits may be more indicative of traits shared between the Pinaceae and Fagales (e.g. recalcitrant litter, woodiness) than an actual link with mycorrhizal status. A similar criticism can be raised of quantitative analyses not considering phylogeny, such as Cornelissen et al. (2001), where, for example, seven members of the Fagales were treated as independent replicates to test links between mycorrhizal status and foliar nutrients. With recent advances in the understanding of plant phylogeny (Stevens, 2001) and mycorrhizal status (Brundrett, 2009), together with the development of extensive global plant trait databases (Wright et al., 2004; Kattge et al., 2011), these limitations can now at least be partially resolved.

Here, we test whether the evolution of ectomycorrhiza is linked to consistent changes in plant functional traits when comparing ectomycorrhizal plant clades with their closest ancestral non-ectomycorrhizal clade. We use an extremely large global dataset (= 11 466 samples across 2968 species) of leaf nutrients and LMA (g m−2) combined with literature records of mycorrhizal status for plant genera. Specifically, we test whether plant traits vary with mycorrhizal status within phylogenetic pairs (paired contrasts) or across all plants including phylogenetic corrections (null model and independent contrast tests). For context, we also test whether plant traits vary with mycorrhizal status irrespective of phylogeny. We also test whether woody species bias the outcome of comparisons among mycorrhizal groups, as many, but not all, ectomycorrhizal plants are woody.

We focus exclusively on ectomycorrhizal plants as this is the only major root symbiosis to have evolved multiple times, making phylogenetic comparisons possible. Ectomycorrhizas have evolved in a wide range of groups, spanning most of the phylogeny of land plants (Fig. 1). For example, ectomycorrhizal plant groups include gymnosperms (Gnetum, Pinaceae), a number of at least partially herbaceous groups (Polygonum, Kobresia, some Cistaceae and Rosaceae), as well as numerous woody angiosperms (e.g. Fagales, Myrtaceae, Dipterocarpaceae). In contrast, arbuscular mycorrhizas, which evolved during the colonization of land by plants (Schüßler et al., 2001; Parniske, 2008; Ercolin & Reinhardt, 2011), and nitrogen-fixing symbioses and orchid and ericoid mycorrhizas are all restricted to a single phylogenetic origin in plants (Soltis et al., 1995; Brundrett, 2009).

Figure 1.

Schematic diagram of the phylogenetic spread of ectomycorrhizal plants across the plant kingdom, showing ectomycorrhizal clades (closed circles), closest non-ectomycorrhizal comparison (open circles) and with selected other clades shown for context (grey text). Groups representing nested or paraphyletic comparisons rather than true sister taxa are shown as triangles (see Table S1 and Methods S1 for details). #G gives the approximate number of genera in each group. Group names (right) are used throughout the text and other figures.

Materials and Methods

Mycorrhizal status

We searched the literature for reports of ectomycorrhiza, relying heavily on reviews to indicate possible ectomycorrhizal clades (Harley & Harley, 1987a,b; Bruns & Shefferson, 2004;Wang & Qiu, 2006; Brundrett, 2009). Based on previous experiences with errors in these reviews (Dickie et al., 2007), once an indication of ectomycorrhizal status was obtained, we examined the primary literature to evaluate the validity and taxonomic breadth of that claim. In some cases, we excluded putative ectomycorrhizal groups where the evidence was weak, the occurrence of ectomycorrhiza extremely rare or highly atypical plant/fungal associations had been labelled as ectomycorrhizal, particularly if both the plant and fungal partners were not commonly considered as ectomycorrhizal. All intentional exclusions are detailed in Supporting Information Methods S1.

The phylogenetic level at which mycorrhizal status was determined varied from species (e.g. Polygonum viviparum) to family (e.g. Pinaceae) or order (e.g. Fagales). As many species have not been examined for mycorrhizal status, we inferred mycorrhizal status from the base of the phylogeny where possible; that is, if all reports of mycorrhizal status within a genus, family or order were ectomycorrhizal, we assumed that the remaining species in that group were ectomycorrhizal. Evidence of ectomycorrhizal formation based solely on artificial conditions was not generally considered.

We considered the ability of a plant to form ectomycorrhiza as a discrete, independent trait regardless of whether the plant also formed arbuscular mycorrhiza, although we discuss dual mycorrhizal plants briefly in the interpretation of the results. This is a departure from a previous analysis in which ‘dual’ ectomycorrhizal/arbuscular mycorrhizal plants were considered to be distinct from solely ectomycorrhizal or solely arbuscular mycorrhizal plants (Cornelissen et al., 2001). Our rationale is two-fold. First, many of the supposed ‘strictly ectomycorrhizal’ taxa have been found to form arbuscular mycorrhiza (Horton et al., 1998; Dickie et al., 2001), calling into question the validity of treating these taxa as strictly ectomycorrhizal. Second, there is no evidence that the gain of ectomycorrhiza is genetically linked to the loss of ability to associate with arbuscular mycorrhizal fungi (although there may be ecological reasons why the gain of ectomycorrhiza tends to reduce or eliminate arbuscular mycorrhizal infection; Lodge & Wentworth, 1990). The full list of identified contrasts is given in Methods S1, with the results summarized in Fig. 1.

Plant traits

We generated a large global dataset of leaf percentage nitrogen (%N) and percentage phosphorus (%P) and LMA by updating the compilations of published studies presented in Reich & Oleksyn (2004), Wright et al. (2004) and Reich et al. (2010). Our dataset was supplemented with data from New Zealand (Richardson et al., 2004, 2008, 2012; Mason et al., 2010; Freschet et al., 2011) and China (He et al., 2008) and we specifically searched for trait data for genera that were under-represented in the global dataset (Waterman et al., 1980; Bowman, 1994; Chidumayo, 1994; Long et al., 1999; Cai et al., 2009; Iheanacho & Udebuani, 2010; Rentería & Jaramillo, 2010). Several of the published studies added were compilations from the same sources, and so we removed double entries before calculating a mean value of N, P, N : P and LMA for each species in the dataset. When synonyms existed for a genus, these were renamed to the current name based on the online version of Vascular Plant Families and Genera published by Kew Botanical Gardens (, and ectomycorrhizal status was checked for conflicts between the synonymous genera. If the mycorrhizal status of a genus was unknown, trait data of that genus were not included in the analyses. Species life form was inferred from the trait literature and identified as ‘woody’, including all plants forming woody tissue, and ‘non-woody’. We checked whether the effect of ectomycorrhizal status on foliar traits was biased by woody status.


For all phylogenetic analyses, we used the Angiosperm Phylogeny Website (Stevens, 2001) as our primary source and Vascular Plant Families and Genera as our source of genera within families ( Where we have referenced other taxonomic sources, these are cited directly. For each ectomycorrhizal group, we contrasted leaf traits with the closest phylogenetic group of non-ectomycorrhizal status (in most cases, arbuscular mycorrhizal; in four cases, non-mycorrhizal). In some cases, these represent monophyletic sister group contrasts, but, in other cases, we have used paraphyletic or nested pairs (Ackerly, 2000), particularly where an ectomycorrhizal group is contrasted with arbuscular mycorrhizal congeners (e.g. Polygonum, Kobresia). In three cases (Faboideae, Caesalpinioideae, Myrtaceae), we have extended this to a comparison of confamilial polyphyletic groups. For example, all known ectomycorrhizal Faboideae are compared with all known arbuscular mycorrhizal Faboideae. This is a pragmatic solution to a combination of unresolved phylogenies, insufficient replication in some groups and to avoid the over-representation of groups in which ectomycorrhiza have evolved multiple times within a single subfamily or family. A table of all genera and species per contrast for which we had plant trait data, and their mycorrhizal and woodiness status, can be found in Supporting Information Table S1.

For each phylogenetic pair, trait data were extracted from the global dataset based on the known ectomycorrhizal or non-ectomycorrhizal status of the genera. Most data were matched at the genus level, with the exceptions of the Pomaderreae (subgenus) and the genera Kobresia, Acacia and Polygonum, where matches were made at the species level.

Statistical analyses

We first tested for differences in each trait between ectomycorrhizal and non-ectomycorrhizal species irrespective of phylogeny, using all species of known mycorrhizal status in the global leaf trait dataset. For this test, we used ANOVA on log-transformed trait data.

Second, we conducted a phylogenetic paired comparison to test for consistent changes in plant traits with mycorrhizal status. Ackerly (2000) has suggested that paired comparisons are appropriate for testing relationships between continuous dependent traits (e.g. foliar nutrients) and discrete independent traits (e.g. mycorrhizal status). Within each phylogenetic pair, we used a two-tailed Wilcoxon two-sample rank sum test to compare foliar traits between ectomycorrhizal and non-ectomycorrhizal species. This test accommodates small and uneven group sizes and does not assume that the data follow a normal distribution. If there were fewer than three species in either the ectomycorrhizal or non-ectomycorrhizal clade, we calculated a 95% confidence interval for the larger group and determined whether the mean or single observation of the smaller group fell outside this confidence interval. Finally, we omitted all non-woody species from the analyses and re-ran the two-tailed Wilcoxon two-sample rank sum test or 95% confidence interval to test for a bias of woodiness in the comparisons of foliar traits within phylogenetic pairs.

Third, to test for an overall effect of mycorrhizal status constrained by phylogeny, we calculated the instances in which foliar %N or %P was greater in the ectomycorrhizal contrast than in the non-ectomycorrhizal contrast, and used a sign test to determine whether ectomycorrhizal and non-ectomycorrhizal species were significantly different in foliar %N or %P (Ackerly, 2000). Because a non-significant sign test result is not strong evidence to accept the null hypothesis (Kelly & Purvis, 1993), we performed a t-test to examine whether the difference in group means for foliar %N and %P between ectomycorrhizal and non-ectomycorrhizal species was significantly different from zero. Furthermore, because some groups have vastly different sample sizes of ectomycorrhizal and non-ectomycorrhizal observations, we re-ran the sign tests and the t-tests only with groups that had at least 10 observations for each ectomycorrhizal and non-ectomycorrhizal group.

Finally, as the paired group comparisons do not fully integrate phylogenetic branch lengths, we applied two additional phylogenetic corrected approaches: (1) using computer null model simulation (Garland et al., 1993), as implemented in the procedure phy.anova in package geiger in R (Harmon et al., 2009); and (2) using comparative analysis by independent contrasts, as implemented in the procedure brunch in package caper in R (Orme et al., 2012). The null model approach tests the observed trait distribution against 1000 simulated null distributions based on Brownian motion. The independent contrast method finds non-nested, independent contrasts within the phylogeny, ensuring that no data are used more than once. Both of these procedures permit a categorical independent variable in testing trait covariation within a phylogenetic context.


We identified 19 phylogenetic pairs of ectomycorrhizal and non-ectomycorrhizal groups (Fig. 1, Methods S1), 17 of which contained at least one species for which foliar trait data were available to test the effect of ectomycorrhizal status. The numbers of genera, species (observations) and percentage of woody species are shown in Table 1.

Table 1. Summary of differences in leaf traits between ectomycorrhizal and non-ectomycorrhizal species within 17 contrasts
Group nameMycorrhizal statusNo. generaNo. speciesWoody (%)Nitrogen P valuesPhosphorus P valuesN : P P valuesLMA P values
  1. Group names are described in Fig. 1. For each group, we present the numbers of genera and species for each ectomycorrhizal and non-ectomycorrhizal contrast, the percentage of woody species and the results of statistical tests. Where n ≥ 3 species, we used a Wilcoxon test and provide the P value. Where n < 3, we used 95% confidence intervals to derive a P value. EM, ectomycorrhizal; LMA, leaf mass per unit area; ND, no data; NS, not significant.

Kobresia EM1100.050.050.05ND
Acacia EM131000.05NSNSND
Uapaca EM13100NSNSNSND
Tilia EM15100NS0.022NSNS
Polygonum EM110NSNSNSNS
Coccoloba EM131000.00690.036NSNS

Plant traits without consideration of phylogeny

Across all plant species, when not considering phylogeny, foliar %P was significantly higher, and, N : P ratio and LMA were significantly lower in ectomycorrhizal than in non-ectomycorrhizal plants (Table 2): foliar %P = 0.11 ectomycorrhizal (n = 423) vs foliar %P = 0.09 non-ectomycorrhizal (n = 827); N : P = 13.84 ectomycorrhizal (n = 422) vs N : P = 16.18 non-ectomycorrhizal (n = 827); LMA = 125.45 g m−2 ectomycorrhizal (n = 168) vs LMA = 150.23 g m−2 non-ectomycorrhizal (n = 408). Foliar %N was not different between mycorrhizal types (foliar %N = 1.55 for ectomycorrhizal (n = 422) and 1.53 for non-ectomycorrhizal (n = 827) plants).

Table 2. Statistical P values for the significance of ectomycorrhizal status as a predictor of foliar traits using uncorrected ANOVA and four phylogeny-corrected approaches
TraitUncorrected ANOVAaPaired comparison, sign testPaired comparison, t-testNull model simulationIndependent contrasts
All groupsbGroups n > 10All groupsbGroups n > 10
  1. a

    Irrespective of phylogeny.

  2. b

    For sign test and t-test, we performed the test for all phylogenetic pairs and for a restricted dataset of groups where n > 10. LMA, leaf mass per unit area. **, significant at 0.01, ****, significant at 0.0001

Foliar N (%)0.340.8010.290.810.900.97
Foliar P (%)< 0.001****0.800.690.090.330.570.20
Foliar N : P< 0.001****10.380.170.780.460.28

Plant traits within phylogenetic pairs

Across the 17 phylogenetic pairs, there were no consistent differences in foliar traits when comparing ectomycorrhizal with non-ectomycorrhizal clades according to the Wilcoxon tests. Of the 17 phylogenetic pairs, foliar %N and %P were significantly higher in ectomycorrhizal than in non-ectomycorrhizal species in three pairs, significantly lower in five pairs and not significantly different in nine pairs (Fig. 2, P values in Table 1). The N : P ratio (Table 1) of ectomycorrhizal species was significantly higher than that of non-ectomycorrhizal species in two pairs, lower in two pairs and not significantly different in 13 pairs. The LMA of ectomycorrhizal species was higher than that of non-ectomycorrhizal species in four pairs, lower in one pair and similar in five pairs. For seven pairs, insufficient data were available to perform comparisons of LMA (Table 1). These results were not influenced by the sample size of the individual ectomycorrhizal and non-ectomycorrhizal comparisons, as differences in ectomycorrhizal and non-ectomycorrhizal foliar traits plotted against differences in sample size between the ectomycorrhizal and non-ectomycorrhizal clades all revolved around zero.

Figure 2.

Mean percentage foliar nitrogen (N) (a) and phosphorus P (b) in ectomycorrhizal clades (y axis) and their closest non-ectomycorrhizal comparison (x axis) for all phylogenetic groups with sufficient trait data. Mean values above the 1 : 1 line have higher foliar nutrients in the ectomycorrhizal clade than in the non-ectomycorrhizal clade. Closed circles represent groups in which the foliar trait differed significantly according to the Wilcoxon test or the 95% confidence interval between the ectomycorrhizal and non-ectomycorrhizal clades; open circles represent groups in which the foliar trait did not differ significantly within clades. Bars represent the standard error of the ectomycorrhizal (vertical bars) and non-ectomycorrhizal (horizontal bars) clades. Aca, Acacia; Cae, Caesalpinioideae; Coc, Coccoloba; Dip, Dipterocarpaceae; Dry, Dryadeae; Fab, Faboideae; Fag, Fagales; Gne, Gnetaceae; Kob, Kobresia; Myr, Myrtaceae; Pin, Pinaceae; Pis, Pisonieae; Pol, Polygonum; Pom, Pomaderreae; Sal, Salicaceae; Til, Tilia; Uap, Uapaca.

According to the sign tests, foliar %N, %P, N : P ratio and LMA were not significantly different between ectomycorrhizal and non-ectomycorrhizal clades: nine of 17 groups had higher foliar %N in ectomycorrhizal than in non-ectomycorrhizal clades (three of six groups when only groups with > 10 observations were considered), and eight of 17 groups had higher foliar %P in ectomycorrhizal than in non-ectomycorrhizal clades (four of six groups when only groups with > 10 observations were considered). Eight of 16 groups had a higher N : P ratio in ectomycorrhizal than in non-ectomycorrhizal clades (one of five groups when only groups with > 10 observations were considered), and seven of 11 groups had higher LMA in ectomycorrhizal than in non-ectomycorrhizal clades (three of four groups when only groups with > 10 observations were considered). Strengthening this observation, a t-test showed that the differences were not significantly different from zero between ectomycorrhizal and non-ectomycorrhizal foliar %N, %P, N : P ratio and LMA (Table 2).

The two phylogenetic tests (null model simulation and independent contrasts) supported the overall lack of significance of mycorrhizal status as a predictor of foliar %N, foliar %P, LMA and N : P ratio (Table 2).

Influence of woody species on foliar traits

Of all species of known mycorrhizal status for which trait data were available, 77% were woody and 23% were non-woody. The investigation of all species in the dataset, compared according to their woodiness status (woody or non-woody plants) irrespective of phylogeny and mycorrhizal status, revealed that woody plants had significantly lower foliar %N and %P and significantly higher N : P ratio and LMA compared with non-woody plants.

Within phylogenetic pairs, most contrasts are purely woody or non-woody, so that a bias of woodiness on foliar traits in the mycorrhizal contrasts is not an issue. In the phylogenetic contrasts Faboideae, Fagales, Uapaca, Tilia and Coccoloba, the ectomycorrhizal groups are 100% woody, but the non-ectomycorrhizal outgroups are mixed woody and non-woody. To test for a bias of woodiness, we reanalysed differences between ectomycorrhizal and non-ectomycorrhizal species omitting the non-woody species. Most phylogenetic contrasts did not show a bias from woodiness, that is, the same significant differences between ectomycorrhizal and non-ectomycorrhizal species as when both woody and non-woody species were considered. However, for Tilia, which showed no significant differences between ectomycorrhizal and non-ectomycorrhizal clades for all species, foliar %N was significantly higher in woody ectomycorrhizal species than in woody non-ectomycorrhizal species (%N = 2.31 for ectomycorrhiza vs %N = 1.83 for non-ectomycorrhiza, PDF = 38 = 0.098). For Coccoloba, the exclusion of non-woody species meant that ectomycorrhizal and non-ectomycorrhizal woody species did not show significant differences in foliar %N, whereas, with the inclusion of non-woody species, ectomycorrhizal species had significantly lower foliar %N than did non-ectomycorrhizal species. For foliar %P, the phylogenetic contrasts Fagales and Coccoloba no longer showed significant differences between ectomycorrhizal and non-ectomycorrhizal species when considering woody species only. The N : P ratio and LMA were not affected by the woodiness of the species.


We found no consistent effects of mycorrhizal status on foliar traits within phylogenetic pairs, directly countering the suggestion that ectomycorrhizal plants have a low-nutrient trait syndrome (Read, 1991; Cornelissen et al., 2001). Similar to Cornelissen et al. (2001), we found significant differences in foliar %P, N : P ratio and LMA between ectomycorrhizal and non-ectomycorrhizal species when not considering phylogeny. However, when phylogeny was considered, there was no evidence of consistent differences in foliar traits between ectomycorrhizal clades and their nearest non-ectomycorrhizal relatives. The lack of correlation between ectomycorrhizal status and foliar traits was also supported by null model simulation and phylogenetic independent contrast tests. Of particular interest is the observation that ectomycorrhizal Pinaceae have higher, not lower, foliar %N and %P when compared with non-ectomycorrhizal Pinales. On the basis of these results, we suggest that the hypothesis that foliar traits are linked to mycorrhizal status should be rejected.

The development of the hypothesis that mycorrhizal status is linked to foliar nutrients is partly based on the observation of organic matter accumulation under some ectomycorrhizal trees (Read, 1991). Orwin et al. (2011) have shown that ectomycorrhizal organic nutrient uptake alone can cause soil organic matter accumulation, regardless of litter quality. Simple parsimony would argue that, where mycorrhizal status can directly affect organic matter accumulation, this explanation should be preferred over one invoking linked suites of above- and below-ground-traits. Further, extensive and persistent organic matter accumulation occurs under woody arbuscular mycorrhizal plants as well (Wardle et al., 2008). Nevertheless, Michelsen et al. (1996, 1998) have shown that in ecosystems in which the mineralization of nutrients from organic matter is slow (e.g. tundra, subarctic), ectomycorrhizal fungi could be of major importance for N uptake of plants.

One of the issues in considering the effects of mycorrhizal status on plant traits is that many ectomycorrhizal plants are woody, whereas many non-ectomycorrhizal plants are herbaceous. For example, Cornelissen et al. (2001) compared woody species with non-woody species without accounting for phylogeny, and this may have partly obscured the actual effect of mycorrhizal status. In our phylogenetic contrasts, woodiness as a bias between ectomycorrhizal and non-ectomycorrhizal species was not a problem, because the contrasts were, in most cases, either both completely woody or non-woody. In a few groups of mixed woodiness, significant differences changed, either appearing or disappearing when only woody species were used for the contrasts. This did not, however, alter our overall findings. Similarly, our results were not driven by the inclusion of groups that have been routinely reported to form both ectomycorrhiza and arbuscular mycorrhiza. For example, the dual-status Salicaceae, Acacia and Myrtaceae groups showed no consistent effects of mycorrhizal status compared with groups that have been considered as more strictly ectomycorrhizal (e.g. Pinaceae, Fagaceae).

Despite our inability to find consistent effects of mycorrhizal status on foliar traits, the evolution of ectomycorrhizal root systems across multiple phylogenetic clades suggests that the ectomycorrhizal trait creates a significant advantage to plants. This is further supported by the widespread dominance of ectomycorrhizal plants across vast areas of temperate and boreal forests, as well as significant areas of tropical forests – a dominance achieved despite the fact that ectomycorrhizal plants only comprise 2% of the plant flora (Brundrett, 2009). One possibility is that the type of advantage conferred by mycorrhizal status differs across clades. Although we found no consistent effect of mycorrhizal status on foliar nutrients, for all three of the comparisons in which significant positive effects of ectomycorrhizal symbiosis on foliar %N occurred, the nearest non-ectomycorrhizal group had low foliar %N (< 2.0%, Fig. 2). This may suggest that, in groups with low-nutrient foliage, the evolution of ectomycorrhiza allows plants to become more competitive on low-nutrient sites, reflected in higher foliar nutrients compared with non-ectomycorrhizal plants. In contrast, four of five of the ectomycorrhizal groups with significantly lower foliar %N had non-ectomycorrhizal relatives with foliar %N > 2.0% (Fig. 2). One possible explanation is that, in groups with high-nutrient foliage, the evolution of ectomycorrhizal associations may be more related to niche expansion, permitting ectomycorrhizal species to expand into lower nutrient soils (Lambers et al., 2008). Another explanation is that, in groups with high-nutrient foliage, the ectomycorrhizal symbiosis provides benefits other than increased nutrients, such as better water uptake or protection against soil pathogens.

An understanding of the linkages between plant traits and mycorrhizal symbionts is an important goal, and here our results are largely negative: refuting any link between ectomycorrhizas and above-ground foliar traits. Nonetheless, we believe that progress can be made in understanding how ectomycorrhizas influence ecosystems, either through modelling of mycorrhizal processes (e.g. Orwin et al., 2011) or further global trait comparisons. In particular, our dataset was limited by being a sampling of available data, rather than a targeted sampling for mycorrhizal status comparison. More targeted sampling of specific phylogenetic groups and explicit linking of mycorrhizal status with soil nutrient status are essential for further advances in this area.


This research was funded by the New Zealand Ministry of Science and Innovation to Landcare Research via the Ecosystem Resilience Outcome-Based Investment (Contract C09X0502) and the National Vegetation Survey (NVS) Databank. P. Reich was supported by the US National Science Foundation LTER Program (DEB 0620652) and the Wilderness Research Foundation. We thank D. Peltzer, P. Bellingham and anonymous reviewers for a constructive review.