Within-species phylogenetic relatedness of a common mycorrhizal fungus affects evenness in plant communities through effects on dominant species

Arbuscular mycorrhizal fungi (AMF) have been shown to influence plant community structure and diversity. Studies based on single plant–single AMF isolate experiments show that within AMF species variation leads to large differential growth responses of different plant species. Because of these differential effects, genetic differences among isolates of an AMF species could potentially have strong effects on the structure of plant communities. We tested the hypothesis that within species variation in the AMF Rhizophagus irregularis significantly affects plant community structure and plant co-existence. We took advantage of a recent genetic characterization of several isolates using double-digest restriction-site associated DNA sequencing (ddRADseq). This allowed us to test not only for the impact of within AMF species variation on plant community structure but also for the role of the R. irregularis phylogeny on plant community metrics. Nine isolates of R. irregularis, belonging to three different genetic groups (Gp1, Gp3 and Gp4), were used as either single inoculum or as mixed diversity inoculum. Plants in a mesocosm representing common species that naturally co-exist in European grasslands were inoculated with the different AMF treatments. We found that within-species differences in R. irregularis did not strongly influence the performance of individual plants or the structure of the overall plant community. However, the evenness of the plant community was affected by the phylogeny of the fungal isolates, where more closely-related AMF isolates were more likely to affect plant community evenness in a similar way compared to more genetically distant isolates. This study underlines the effect of within AMF species variability on plant community structure. While differential effects of the AMF isolates were not strong, a single AMF species had enough functional variability to change the equilibrium of a plant community in a way that is associated with the evolutionary history of the fungus.

138 Gp1 -5B, ESQLS69, LPA54; Gp3 -A4, DAOM229457, DAOM240409; Gp4 -A1, 139 DAOM197198-CZ, DAOM240159 (Fig. 1a). The delineation of this variation was based on 140 previous ddRADseq data generated from DNA of 59 isolates of R. irregularis isolated from 141 several geographical locations and comprising 6888 sites in the genome where single 142 nucleotide polymorphisms (SNPs) occurred. Each of the 9 isolates was first grown in an 143 identical in vitro environment for 5 months [24]. In order to create a natural soil inoculum for 144 each isolate [25], we inoculated P. lanceolata with 500 in vitro-produced spores in ten 0.5 145 litre pots in a sterile autoclaved Oil-Dri® granular clay (Oil Dri corporation of America) with 146 quartz sand in a 3:2 ratio. Ten P. lanceolata were mock inoculated with water in order to 147 produce a substrate representing a mock-inoculated treatment. The P. lanceolata plants were 148 grown for five months and then the soil of the ten pots per isolate were mixed to create the 9 149 inoculum stocks and substrate for the mock-inoculated treatment.   (Fig. 1b). The position was randomly chosen except that plants of the same 164 functional group were never planted next to each other but always at the most distant location.
165 Thus, Arrhenatherum elatius was always the most distant plant from Festuca pratensis, 166 Trifolium pratense was always the most distant plant from Lotus corniculatus and Prunella 167 vulgaris was always the most distant plant from Knautia arvensis (Fig. 1b).
168 Fourteen mycorrhizal treatments and one non-mycorrhizal (NM) treatment were applied to 169 the plant communities and were replicated ten times (Fig. 1a). We inoculated each pot of the 9 170 single inoculation treatments (treatments 1-9) with 100g of inoculum. Treatments with a co-171 inoculation of three isolates (Gp1, Gp3 and Gp4; treatments 10-12) received 33.3g of 172 inoculum of each isolate. In the case of Gp3+Gp4 (treatment 13) we used 16.6g of inoculum 200 The AMF colonization evenness was calculated in the same way in order to assess equality or 201 inequality of AMF colonization among the six plant species within a mesocosm. This could AMF relatedness and plant community structure 11 202 be considered as a proxy for AMF preferences within a mesocosm. Significant correlations 203 between mesocosm variables, as well as the relationships between single plant data and 204 mesocosm averages, were assessed using Pearson's product moment and polynomial 205 regressions.
206 The effect of plant species and mycorrhizal treatments on AMF colonization, plant 207 responsiveness and number of flowers produced were analyzed using linear mixed-models 208 (lmer) with the pot as a random factor. In order to test for significant differences an analysis 209 of variance was used on the models. Significant pairwise comparisons were assessed using the 210 difflsmeans function. Significant differences among AMF treatments towards plant 211 community metrics were assessed using a one-way ANOVA and a Tukey HSD post-hoc test.  (Table 1a). Isolates A4 (Gp3) and ESQLS69 (Gp1) were significantly the lowest 245 colonizers and the three isolates of Gp4: A1, DAOM240159 and DAOM197198-CZ were the 246 highest colonizers (Fig. 2a). R. irregularis, independent of isolate identity, colonized a 247 significantly greater proportion of the roots of K. arvensis, P. vulgaris and T. pratense than 248 the dominant plant, F. pratensis (Fig. 2b).   above-ground dry mass evenness, root dry mass, above-ground dry mass, total productivity, flower 257 production at 78 and 105 days, and AMF colonization evenness. 283 and a significant increase in the relative contribution of all the other plants (Table S1; Fig.   284 3d). . All nodes showed a support 302 value of 100, clearly separating the nine isolates into three main groups (Fig. 1a). This tree 303 was then used for phylogenetic signal analyses.

Source of
304 Testing for phylogenetic signals with plant species data and plant community metrics 305 There was a significant within-species AMF phylogenetic signal in AMF colonization of F.
306 pratensis in two out of the five tests (Fig. 4a), with a globally highest colonization by Gp4 307 isolates. Similarly, there was a significant within-species AMF phylogenetic signal in the 308 mycorrhizal responsiveness of F. pratensis (Fig. 4b) in three tests out of the five. There was 309 also a significant within-species AMF phylogenetic signal in the mycorrhizal responsiveness 310 of K. arvensis for one test out of five (Fig 4b). 319 At the community level, the overall mean AMF colonization, RDM, ADM, TDM and flower 320 production did not reveal a significant within-species AMF phylogenetic signal (Fig. 4c). In 321 contrast, there was a significant within-species AMF phylogenetic signal in mean mycorrhizal 322 responsiveness of the community in four out of five tests (Fig. 4c). There was no detectable 323 within-species AMF phylogenetic signal in AMF colonization evenness (Col-evenness) (Fig.   324 4d). There was a significant within-species AMF phylogenetic signal in terms of community 325 evenness (ADM-evenness) in three tests out of five (Fig 4d).