Macroevolutionary decline in mycorrhizal colonization and chemical defense responsiveness to mycorrhization

Summary Arbuscular mycorrhizal fungi (AMF) have evolved associations with roots of 60% plant species, but the net benefit for plants vary broadly from mutualism to parasitism. Yet, we lack a general understanding of the evolutionary and ecological forces driving such variation. To this end, we conducted a comparative phylogenetic experiment with 24 species of Plantago, encompassing worldwide distribution, to address the effect of evolutionary history and environment on plant growth and chemical defenses in response to AMF colonization. We demonstrate that different species within one plant genus vary greatly in their ability to associate with AMF, and that AMF arbuscule colonization intensity decreases monotonically with increasing phylogenetic branch length, but not with concomitant changes in pedological and climatic conditions across species. Moreover, we demonstrate that species with the highest colonization levels are also those that change their defensive chemistry the least. We propose that the costs imposed by high AMF colonization in terms of reduced changes in secondary chemistry might drive the observed macroevolutionary decline in mycorrhization.


INTRODUCTION
Mutualisms among species shape ecological communities and ecosystem dynamics worldwide. [1][2][3] Among the most ancient and widespread mutualistic symbioses are the associations between plant roots and mycorrhizal fungi. Mycorrhizae acquire photosynthates directly from plants, and in exchange they expand the plants' foraging capacities, 4 therefore stimulating plant growth and resistance against abiotic and biotic stressors. 5,6 In addition to facilitate nutrient acquisition, arbuscular mycorrhizal fungi (AMF), particularly, have been shown to alleviate plant stress caused by abiotic environmental conditions, 7,8 and also biotic stress such as herbivory or pathogen attack. 6 However, although widespread and putatively beneficial, plant-AMF interactions can vary from being highly mutualistic to parasitic. In this latter scenario, the mycorrhizae can inhibit plant growth by imposing strong energetic costs to the plants. 9,10 To date, ecologists still struggle to understand the sources of variation in mycorrhization levels and their effects on plant phenotypes. Sources of variation include biotic factors such as the plants and fungal genetic make-up, or abiotic factors such as climatic conditions. 10,11 If the association with AMF is a phylogenetically constrained trait, closely related species should sustain similar mycorrhization intensities, and respond similarly to mycorrhization. Although at higher taxonomic levels, such as across plant families, a phylogenetic constraint has been found for plant responsiveness to AMF colonization, the intensity of colonization and influence on ecosystem functioning, 2,12 it remains unclear whether this selection exists at a finer taxonomic level, such as across species belonging to the same genus. On the other hand, distantly related plants inhabiting similar environments might favorably associate with components of the local fungal community, independent of their evolutionary history. 13 Therefore, AMF colonization intensity and plant responsiveness to AMF colonization can be driven by either shared evolutionary history, 12 or ecological-niche convergence because of shared abiotic and the biotic factors. 14 Although patterns of plant responses to AMF are relatively well documented for plant growth or the allocation of biomass to different organs across species with different life histories, 10,15 our understanding of how plants respond to AMF by changing their chemical phenotype, within biogeographical or phylogenetic variation, is largely lacking. 16,17 Plants produce a plethora of molecules that are not directly linked to primary metabolism, referred to as secondary or specialized metabolites, that serve to mediate interactions with the biotic and abiotic environments. 18 For instance, such specialized metabolites combine with plant physical defenses to shape the plant defensive phenotype against a wide range of herbivores and pathogens. 19 AMF can modify plant specialized metabolite production in both roots and aboveground plant parts, [20][21][22] in turn modifying plants' interactions with herbivores, pathogens, and higher trophic levels. 23,24 However, how plant evolutionary history and ecological factors modulate the effect of AMF on plant chemical defense traits remains uncertain.  Table S1). The tips of the phylogeny are colorcoded based on the root-to-tip distance of each species (lighter colors indicate shorter branch length), see also Figure S5. The size of the dot on the tips of the phylogeny is proportional to the number of arbuscules (arbuscular mycorrhizal fungi (AMF) colonization intensity) of each of the 24 Plantago species. (B) represents the distribution around the globe of the 24 Plantago species used in the study. (C) and (D) represent the correlation between AMF colonization intensity phylogenetic branch length, and pedo-climatic niche distance, respectively. Dots are coded based on the root-to-tip distance of each species, with lighter colors indicating shorter branch length. Blue line and grey-shaded area around the line represent 95% confidence level interval for predictions from a linear model. (E) and (F) represent, for total leaf phosphorus (leaf P mg mg À1 ) and carbon to nitrogen ratio (CN % biomass), respectively, the average effect of AMF colonization intensity represented as the Cohen's D effect size between mycorrhized and nonmycorrhized plants. Red boxes (High) show the average of five species that are highly mycorrhized (see red arrows in Figure 1A), whereas green boxes (Low) show the average of five low mycorrhized species (see green arrow in Figure 1A). Boxplots represent the minimum (black dots), first quartile (black line), median (bold black line), third quartile (black line), and maximum (black dots) of data distribution. Letters above boxes indicate significant difference between groups (PGLS; p < 0.05). ll OPEN ACCESS 2 iScience 26, 106632, May 19, 2023 iScience Article In this study, we question how phylogenetic history and pedo-climatic convergence affect AMF colonization intensity and responsiveness in relation to plant growth and chemical defense traits. To this end, we studied patterns of AMF colonization and plant responses across 24 species in the genus Plantago L. (Figure 1A), by growing them from seed in a common environment with and without inoculation of four common and widespread AMF species; Rhizophagus irregularis, Funneliformis mosseae, Claroideoglomus claroideum, and Diversispora celata. The genus Plantago is exceptionally well-studied phylogenetically and chemotaxonomically. 25,26 A large fraction of the estimated 250 species in the genus have been shown to produce a diverse array of monoterpenoid derived iridoid glycosides (IGs), which are recognized as chemotaxonomic markers for Plantago. 25 IGs are generally regarded to function as allelochemicals and are involved in plant defenses with antimicrobial and/or antiherbivore properties. 27,28 Moreover, the genus Plantago is distributed worldwide and species are found in habitats ranging from desertic-mediterranean to temperate-continental climate niches. 29 Because this range of habitats are associated with variable environments, herbivore and pathogen communities, ecological convergence may have shaped AMF-Plantago interactions, as well as growth patterns and specialized metabolites' production, making the genus an excellent case for this investigation. By adopting a phylogenetic-multifunctional approach, we asked: (1) How are mycorrhizal colonization levels across species related to phylogenetic history or edaphic and climatic conditions? (2) How does AMF colonization affect plant growth and IG variation? (3) What is the relationship between colonization levels and AMF-mediated plant growth or chemical changes? First, we predicted that if AMF stimulate growth and defense traits, resource-poor and stressful environments would select for species with high investment in AMF colonization intensities. Second, because of evolutionary tendencies for family-wide losses in plant-AMF association, rather than recent gains, we predicted that over time, plant clades in the genus Plantago would tend to reduce their association with AMF. Finally, because AMF are known to stimulate plant growth and chemical defences, 30 we predicted that species with low mycorrhization levels are also less plastic in responses to AMF colonization such that they exhibit reduced responsiveness to mycorrhization.
2) Phosphorus (P) and carbon to nitrogen (CN) amounts in leaves across species -We found that high mycorrhizal species, on average when mycorrhized, had a 30% increase in P content in their leaves, whereas low mycorrhizal species has a 10% decrease in P content ( Figure 1E;  3) AMF effect on growth traits and iridoid glycosides (IGs) production -We found that AMF changed the overall growth of plants by reducing total plant biomass by 10.7% (see negative effect in Table 1, Figure 2A), and by reducing plant biomass allocation to roots compared to shoots by 7.8%, compared to non-mycorrhizal control plants ( Figure 2B and Table 1). However, we found no effect of AMF treatment on the leaf dry matter content (LDMC) and specific leaf area (SLA) ( Figures 2C and 2D, Table 1). A total of 27 IGs were found across all species and treatments (Table S2), and the MCMCglmm models indicated that AMF had no overall effect on the total abundance, and richness of IGs across all species (Table 1).

OPEN ACCESS
iScience 26, 106632, May 19, 2023 3 iScience Article 4) AMF effect on multivariate plant growth and chemical space -Climatic and soil differences in the Plantago species' niches were not found to correlate with neither the growth traits nor with the IGs matrix (Table S3). In contrast, we found that closely related species, regardless of the mycorrhizal status or ecological niche preferences, have IGs profiles that are more similar than distantly related species (Table S3), and we found no phylogenetic signal for the growth-related traits. Together, these patterns suggest a strong phylogenetic signal on the chemical profile regardless of inoculation/colonization with AMF. Through PERMANOVAs we found a species and a species by AMF interaction effect on the plant growth traits matrix ( Figure S4A, Table S4), and species effect on the IGs matrix ( Figure S4B, Table S4).

5)
Plant growth and chemical defense responsiveness to AMF colonization -No significant correlation was found between plant growth-related phenotypic changes induced by AMF and AMF colonization level ( Figure 3A, PGLS: r = 0.001, t = À0.004, p = 0.99, and phylogenetic signal for responsiveness of growth trait: l = 0.25, p l = 0.33). However, we found that when species are highly mycorrhized, the species produced a more similar IG chemical profile structure than for species with low levels of mycorrhization ( Figure 3B, PGLS: r = 0.52, t = À3.5, p = 0.007, and phylogenetic signal for responsiveness of chemical trait l = 0.92, p l = 0.001). Finally, we found a negative relationship between growth and IG responsiveness to AMF colonization ( Figure 3C, PGLS: r = 0.46, t = À2.5, p = 0.022), showing that those species which are more different in their IGs profiles when mycorrhized are those that are more similar in their growth-related phenotypes iScience Article

DISCUSSION
We observed that across Plantago species, arbuscule colonization intensity decreases with increasing phylogenetic root-to-tip distance, but AMF colonization intensity was independent of the pedo-climatic niche of the species. Moreover, contrary to our prediction, species displaying the highest AMF colonization intensity changed their chemical profiles the least when colonized with AMF. Together, these effects resulted in a genus-wide trade-off between plant growth-related responses and the plant IGs production-responses to mycorrhizal colonization.

Mycorrhizal colonization intensity declines over evolutionary time
Predicting across-species declines in mycorrhization might seem counterintuitive, if for instance we assume that all plant-mycorrhizae associations are beneficial for plants, and should therefore be favored by natural selection. 31  iScience Article literature indeed indicate that AMF can be categorized along a parasitism to mutualism spectrum. 32,33 Therefore, when the costs for plants of maintaining AMF in their roots become too high, natural selection during speciation events should favor individuals that form reduced symbiotic partnerships with AMF. Of interest, across the wider phylogeny of angiosperms, it was shown that multiple non-mycorrhizal families (e.g., Brassicaceae, Crassulaceae) generally derive from a mycorrhizal ancestor, 34 whereas the reverse, the acquisition of mycorrhization from a non-mycorrhizal ancestor, has never been detected and is considered unlikely to happen. 35 Moreover, the complete loss of mycorrhizae should first go through a weakening in colonization intensity. 35 As we show, it appears that genetic relatedness, and the evolutionary trends therein, control the level of mycorrhization more than other factors, as we have shown here for pedo-climatic conditions. Particularly, we observed that species coming from similar climatic conditions, which may be more likely to grow in similar resource conditions, have disparate colonization intensities, thus suggesting a lack of ecological convergence within the Plantago genus for mycorrhization intensity. On the other hand, a study aiming at identifying predictors of global patterns of AMF colonization intensity emphasized the role of several climatic components and soil nutrient as drivers of root colonization intensity. 36 By extracting coarse values (about 1 3 1 Km pixels) for soil and climatic variables, we indeed might have blurred finer levels of variation in microclimate and pedological forms, which would have masked within-species variation, ultimately accentuating the observed lack of  Figure S4B) based on iridoid glycosides production for each Plantago species non-mycorrhized or with AMF. (C) shows the correlation between AMF-mediated growth changes and chemical changes across species. Dots are coded based on the root-to-tip distance of each species, with lighter colors indicating shorter branch length. Blue line and greyshaded area around the line represent 95% confidence level interval for predictions from a linear model. Significance of the correlations is estimated using phylogenetic generalized least squares (PGLS) models. iScience Article among-species signal. Future studies should thus further address whether within species, levels of mycorrhization remain relatively constant (i.e., within species variation is smaller than across species variation), or whether the mycorrhization levels of different ecotypes is the result of a local adaptation to local climatic, and edaphic conditions, as well as the local fungal community.
Alternatively, different forces rather than climate adaptation, such as plants life history traits in relation to ecological succession, have also been proposed for studying variation in AMF colonization intensity and plant responsiveness to AMF. 15 In our dataset, we have included six annual (or biennial) and 18 perennial species. Because there is indication that younger plants tend to sustain higher mycorrhizal colonization, 37 we could expect perennials to be less mycorrhized, because they should reach maturity later, and they have more time to shed AMF structure from their roots. On the other hand, because perennials need to optimize their fitness over long periods of time, we could expect that the root-system-expanding properties of an AM network to be more helpful to perennial plants than to annuals. We indeed, found an indication that perennials have higher arbuscule colonization than annuals and that they are higher or not different for total or vesicle colonization (see Figure S7and Table S5). Other ecological factors, such as specific soil traits that were not available to us for this continent-wide analysis (e.g., CNP ratio, cation-exchange capacity), 38 or plant species invasion potential, 39 might also drive species-level variation in AMF colonization intensity. Declines in AMF intensity are thus common, and are related to phylogenetic history, however, the weak phylogenetic signal itself measured for AMF colonization intensity (l = 0.42) suggests that other, not here measured, ecological factors, life-history traits, and plant-belowground traits are also driving across-species variation in how plants favor and maintain robust interactions with AMF communities.
Root traits themselves might also predict variation in mycorrhization intensity. 40 For instance, plants that possess more fine root branching and shorter absorptive first-order roots harbor less AMF, whereas plants with thicker roots rely more on AMF to compensate for low absorptive surface. Accordingly, plants-AMF symbioses can be placed along a ''collaborative gradient''. 41 Plantago spp. might also follow such dynamics, because we have shown that higher level of arbuscules lead to higher phosphorus uptake, whereas lover levels of arbuscules characterize species with lower benefits in term of phosphorus uptake. Whether these effects, mediated by plant functional traits, also follow phylogenetic lines remains to be fully elucidated. A large-scale study showed that the best predictor of variation for the majority of root functional traits is the phylogenetic relatedness of the plants. 42 This phylogenetic structuring of root functional traits may be present at a finer taxonomic across Plantago spp. and may partially explain the observed AM-arbuscules colonization de-escalation. However, to find the mechanistic reasons of why AMF colonization weakens in more derived species remains challenging. A future step in this direction might be to measure the selective pressure on the genes driving the symbiosis, such as STR2 and STR1. 43 Plant growth and chemical defense responsiveness to AMF iScience Article or pathogens. 49 Because we only measured constitutive IGs on healthy plants, measuring the priming or higher inducibility effect was not feasible for this study, but undoubtedly feasible in future comparative phylogenetic studies using artificial elicitors (e.g., jasmonic acid or salicylic acid), or controlled herbivory manipulations. 50 Nonetheless, we observed strong phylogenetic signal for IGs diversity across species, independently of whether plants were mycorrhized or not. This observation is in line with past findings, which demonstrated that IGs can be used as taxonomic markers for the different subgenera of the genus Plantago. 25 In contrast, average species' pedo-climatic niche did not correlate with IGs similarity, thus suggesting that phylogeny is a stronger driver of IGs diversity than the ecological niche of each species.
Furthermore with respect to IGs, we observed that in comparing AMF-colonized versus uncolonized plants, those taxa that had higher levels of AMF colonization showed less change in IG diversity. In other words, the degree of AMF colonization across species, which is to some extent phylogenetically controlled, also correlated with the degree to which a plant changes chemical defense profile when mycorrhized. More specifically, we would intuitively predict that higher AMF colonization, potentially providing more resources, would result in greater diversity of IGs, but our findings are in opposition to this prediction. This suggests two avenues for further research. First, the ability of plants to allow high degrees of AMF colonization is a pleiotropic trait that trades off with the ability to produce more diverse and more abundant IGs. This could be explained for example by a carbon allocation trade-off, because IGs are carbon-based molecules, and as we have discussed above, AMF can be carbon-sinks for many plant species. 51,52 Whether this is true for other molecule classes, such as the nitrogen-containing alkaloids, needs to be confirmed in other systems. Second, and along other trade-off lines, we have here observed that AMF-induced changes in growth negatively correlate with changes in IGs. This might indicate that when mycorrhized, plant species are constrained to allocate resources to either functional traits related to growth, or to chemical defenses, but not to both simultaneously. Acrossspecies growth-defense trade-offs have been widely postulated, 53 which might also vary depending on the ecological context, such as the availability of resources, 54 and thus, indirectly, the ability of maintaining beneficial symbioses with AMF. 55 Therefore, whether within-or across species AMF-mediated growth-defense trade-offs are equally maintained needs further enquiry across a broader range of systems.
We show that AMF colonization intensity in Plantago species is tightly linked to the evolutionary history of the genus, and that more recently evolved species are on average less colonized than more ancestral species. The intensity of AMF colonization in turn dictates the degree of resources each species can allocate to growth or to chemical defenses. The history of land plants is intimately linked to the 400 M years old symbiosis with mycorrhizal fungi. It therefore appears that more recently evolved lineages tend to decrease their dependence to AMF, as shown here, as well as novel genera of families, as shown elsewhere, are relaxing the intensity of such associations. The causes of such declines in AMF-plant mutualism might be sought through resource-allocation trade-offs, physiological or evolutionary constraints, or through artificial selection experiments such as those naturally done during crop domestication.

Limitations of the study
We here show that more recently evolved species are less mycorrhized. One limitation in relation to these findings, which we recommend including in future experiments, is to explore these effects by looking at arbuscular mycorrhizal symbiosis essential genes (e.g., STR2 and STR1) across multiple species to detect if these genes are under relaxed selective pressure in the low AMF colonized Plantago species. Secondly, we detected no effect of the macro-pedo-climatic niche of each species on AMF colonization intensity. However, finer-scale analyses might indicate a direct effect of soil properties on intraspecific variation for those Plantago species that have a wide distribution (e.g., P. major). Third, the correlation between AMF colonization and phylogenetic branch length was significant for arbuscular colonization but only a tendency was found for the correlation between phylogenetic branch length and total AMF root colonization. Further studies need to investigate the robustness of this relationship, especially because the occurrence of arbuscules is a more dynamic measure of AMF root colonization and depends on root age. Finally, future work should relate the observed findings, here obtained from a greenhouse experiment, with natural AMF colonization rates in field settings.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

ACKNOWLEDGMENTS
We thank all the botanical gardens that supplied Plantago seed material. We thank Lara De Gianni and Molly Baur for help with data collection. We are grateful to Laurent Oppliger and the staff of the Neuchâ tel Botanical Garden for helping with plant maintenance during the experiment. This manuscript is supported by a Swiss National Science foundation grant (31003A_179481) to SR.

DECLARATION OF INTERESTS
The authors declare no competing interests.   All raw and analyzed data deposited This paper https://datadryad.org/stash/share/UssC9sdnx-

OPEN ACCESS
The genus Plantago is an optimal lineage of plants for investigating patterns of plant growth and defense responsiveness to AMF inoculation for multiple reasons. First, the genus Plantago has a worldwide distribution (see ''method details'' below) with several taxa occurring in temperate and Mediterranean biomes, as well as in the cold-desert regions. 57 In the tropics, species are known to occur at high elevations, and several species occur as single island endemics 56,57 ( Figure 1B, Table S1). Second, all Plantago species tested to date are highly mycotrophic. 58 Third, many species have been shown to produce IGs. 25,26,59 For this study, we obtained seeds of 24 Plantago species representing each of the four major Plantago clades, 60 including representatives from all continents (Table S1). As shown in Table S1, most of the seeds for the species tested originated from one unique source, therefore our within-species variation for all traits measured, including mycorrhization levels, was rather low. While this represents a major caveat for these type of macro-ecological studies, one major assumption that we make here is that within-species variation should be generally lower than across species variation, and therefore, which genotypes we chose should impact little the major findings across species, at the macroevolutionary scale [see e.g., [61][62][63] ]. Moreover, although our sampling covers 12% of all species in the genus, 29 it includes a broad spectrum of the phylogenetic (covering 10 out of 18 major Plantago sections, Figure S1), geographic and ecological diversity of the genus (Figures S2 and S3). Sequencing and phylogenetic reconstruction is described in the ''method details'' section.

Fungal organisms: Arbuscular mycorrhizal fungi inoculum
Two-weekold plants were inoculated with a mixture of four, broadly distributed and co-occurring AMF species: R. irregularis, F. mosseae, C. claroideum (order: Glomerales) and D. celata (order: Diversisporales). All species have been shown to have a global distribution, 64 iScience Article containing extra-radical spores and AMF-colonized root fragments. Un-inoculated, control plants were treated with the same substrate mixture, but free of AMF. 66 We acknowledge that the different Plantago species used in the study might have co-evolved with different species or strains of AMF. 67 However, despite the variable outcome of the plant-AMF symbiosis depending on the identity of the host plant or the symbiont, and the host selectivity for the AMF partners that occurs under natural conditions, 68 it has been previously shown that under greenhouse conditions, almost any AMF is able to colonize any mycorrhizal plant species to some extent. 69 Hence, both for practical reasons, and because we are working at the level of the same host plant genus with a recent evolutionary history (last estimation for most ancient Plantago divergence is approximately 16.7 Ma 56 ), we opted to standardize the AMF community using a common inoculum for every plant species.

Common garden experiment
We ran a greenhouse common garden experiment in semi-controlled condition to measure the effect of AMF colonization on plant growth and IGs production. All seeds were germinated in Petri dishes laminated with moist filter papers at room temperature and in dark conditions. After germination, seedlings were transplanted into 13 cm width 3 10 cm height plastic pots in a mixture of low nutrient substrate autoclaved twice at 121 C for 20 min, the two cycles separated by 48h rest, composed of two-thirds of quartz sand and one-third of homemade compost soil (pH = 7.64, bioavailable p = 4 mg/kg, organic matter = 35%, CN = 11.5, CEC -Ca -Mg -K -Na -Al (cmol (+) /kg) = 29.5-0.016 -0.002-0.0005 -0-0). Before transplantation, the soil of each pot was homogenized with either the AMF inoculum (5% of the volume of the pot = 65 mL of inoculum) or the same substrate without AMF inoculum (the control). Half of the plants (n = 5-7 plants per species) received the AMF-containing substrate, while the other half (n = 5-7 plants per species) served as controls (see Table S1 for the number of replicates per species). Plants were grown for two months (July-August 2016) in an automatic vent-opening greenhouse at the Botanical Garden of Neuchâ tel, Switzerland, under natural temperature and light conditions, in a fully-randomized scheme. Plants were watered every three days.

Climatic and soil niche reconstruction
Occurrence records with geographical coordinate data for all species in our study were extracted from the Global Biodiversity Information Foundation (GBIF; http://data.gbif.org). Erroneous records were removed from the dataset. Ten of the 19 Chelsa climatic measures 70 Figure S1) were extracted for each species using the raster package 71 in R (version 3.6.1). 72 To avoid covariation of predictors and reduce the dimensionality of the climatic niche of the species we further condensed all the climatic variables using a Principal Component Analysis (PCA). The first two axis of the PCA explained together 83% of the variation. According to Horn's Parallel Analysis performed in the paran package, 73 the first axis was strongly correlated with temperature and precipitation (50% of variation explained), and therefore was selected for downstream analyses ( Figure S2). Using the same geographical occurrence data used for climatic data extraction, soil physiochemical information was extracted from the Global Gridded Soil Information Database (SoilGrids -ISRIC world Soil Information. 74 The median value of the 19 available variables at the depth interval ranging from 5 to 15 cm depth, the predominant bioactive zone for roots, were extracted. To avoid covariation of predictors and reduce the dimensionality of the soil niche of the species, we further condensed all the soil variables using a PCA. O and soil pH KC) were retained. The first two axes of the PCA explained 80% of the variation, and axes strongly correlated with organic carbon and pH (first axis 61% of variation explained, moving from organic and acid soils to non-organic basic soils along axis 1) and soil texture and CEC (second axis 19% of variation, from high nutrient clay soils to low nutrient sandy soils along axis 2) ( Figure S3). Both axes were retained and used as a proxy of the soil niche of each species, based on Horn's Parallel Analysis. To measure potential phylogenetic collinearity across pedo-climatic variable, we performed partial mantel tests (multi.mantel function in the phytools package 75

Sequencing and phylogenetic reconstruction
For phylogenetic reconstruction, the ITS and trnlF introns were chosen for this study based on these two regions showing enough resolution across the major sub-genera in the genus Plantago. 56,60 DNA sequences for seventy-four species, including the species Litorella uniflora L. were included in our full phylogenetic analyses (Data S1; Figure S1). Seventeen sequences were newly generated for this study (and uploaded to GenBank), and 85 and 45 sequences, respectively, previously generated in Rønsted et al. 60 Figure S1) was pruned to the 24 species used for analyses using function fix.poly in the RRphylo pacakage. 79,80

AMF colonization intensity
Before oven-drying, soil particles were carefully cleared from roots and 1g of fresh young/secondary roots from each individual root system were randomly cut and stored at À20C until the staining procedure was carried out to visualize AMF. Root staining consisted of 1) root clearing by KOH 10% during 10 minat 90 C bath, 2) rinsing the KOH solution and acidifying with vinegar (5% acidity) at room temperature for 5 min, and 3) staining with a solution of 5% -blue ink/vinegar. The stained roots were submerged in glycerol in Eppendorf tubes and stored at 4 C until slide preparation was undertaken. To estimate overall AMF colonization, ten root fragments of 1.5 cm length were placed vertically on a microscope slide. A solution of polyvinyl lacto-glycerol (PVLG), prepared by mixing 100 mL lactic acid, 100 mL ddH2O, 10 mL glycerol, and 16g polyvinyl alcohol powder at 80 Cº for 4 h, was added on the root fragments for microscopy visualization and preservation. AMF colonization intensity was estimated on n = 5 or 6 individual root system per Plantago species per AMF treatment (only fine roots were scored also non-mycorrhizal plant were scored for mycorrhizal colonization) following the intersection method. 80

Iridoid glycosides (IGs) identification and quantification
For IGs analyses, one young yet fully-expanded leaf per plant was oven-dried at 40 C for 48 h and ground to powder using an MM400 Retch TissueLyser (Qiagen, Hilden, Germany). Next, 10 mg of leaf powder per plant was extracted with 1.5 mL methanol, and the supernatant was diluted five times by adding 800 mL of MilliQ water to 200 mL of pure extract. IGs were separated using ultra-high performance liquid chromatography-time of flight mass spectrometry UHPLC-QToF-MS using an Acquity BEH C18 column from Waters (50 3 2.1 mm, 1.7 mm particle size) following the same protocol as in Bakhtiari et al. 81 Absolute amounts of IGs were determined by external calibration using six standard solutions of catalpol (for catalpol quantification) and aucubin (for all other IGs quantification) at 0.2, 0.5, 2, 5, 10 and 20 mg mL-1. Concentrations were normalized to plant weight and expressed as mg mg-1 dry weight. IGs were identified based on their retention time and chemical formula by comparing them to previous chemical descriptions of Plantago

Plant functional trait measurement, root colonization
At the end of the two-months growing period, the following plant functional traits related to growth were measured according to Pé rez-Harguindeguy et al. 36 ; 1) root biomass (g dry weight (DW), 2) shoot biomass (g DW), 3) total plant biomass (g DW), 4) root-shoot ratio, 5) specific leaf area (SLA, mm 2 mg À1 ), 6) leaf dry matter content (LDMC, mg g À1 ), 7) chlorophyll content (SPAD). Specifically, SLA was calculated by dividing the area of the youngest fully-expanded leaf, estimated using ImageJ software (https://imagej.nih.gov/ij/) by its dry biomass. LDMC was calculated by dividing the dry biomass of the same leaf by its water-saturated fresh biomass. Chlorophyll content was measured on three youngest fully-expanded leaves per plant using an SPAD-502Plus chlorophyll meter (Konica Minolta, Investment Ltd., Tokyo, Japan). Finally, the whole plant was oven-dried at 40 C for 48h for measuring dry aboveground biomass, dry root biomass, and quantifying IGs in the leaves (see below). Next, to assess whether AMF colonization intensity could be related to variation in elemental composition (C, N, P) of Plantago leaves, we selected the five most (red arrows in Figure 1A) and the five least (green arrows in Figure 1B) colonized Plantago species across the phylogeny under investigation.

IGs diversity and abundance indexes
We quantified the diversity (i.e., the number of individual componds) and abundance of all iridoid glycosides (IGs) found in all plants growing with and without AMF (see Table S2).

Statistical analyses
All statistical analyses were conducted using R (version 4.0.2). 72 1) Variation in AMF colonization levels across species: phylogenetic and environmental drivers -First, we assessed the effect of Plantago species on colonization intensity using, after testing for model assumptions, one-way ANOVA. Next, we assessed the correlation between both phylogenetic history and environmental niche on AMF colonization levels. To test the effect of phylogenetic history, we used the root-to-tip distance (i.e., patristic distance calculated as the number of substitutions per site extracted with the distRoot function in the adephylo package. 82 We regressed root-to-tip distance against AMF colonization levels using a linear model (lm). The phylogenetic signal was also calculated for the three AMF colonization variables using the function phyloSignal in the package phylosignal. 83 For the environmental effect we performed a coinertia analysis between the retained climatic and pedological variables to check for a shared pedo-climatic structure using the coin function in the 'vegan' package. 84 When significant, the coinertia analysis indicates a significant co-structuration (i.e., correlation) between the pedological and the climatic environments. Accordingly, we observed a significant positive correlation between the soil and climate matrices (r = 0.68, p = 0.001 based on 999 permutations), indicating, particularly along the first axis of the coinertia analysis, that Plantago species growing in colder and drier environments are also found to grow in deep soils with high pH, high bulk density, while species growing in warmer and more humid environments are also found to growing fertile soils. We regressed the coinertia axis 1 against AMF colonization levels using phylogenetic generalized least squares (PGLS) models, with the l values estimated by maximum likelihood (ML) (pGLS function in the 'caper' package 85 ).
2) P and carbon-to-nitrogen ratio (CN) -To measure the effect of AMF colonization on P or CN in Plantago leaves, we first calculated effect sizes between non-mychorrized and mycorrhized plants, for each species individually, using Cohen's d metric, 86 as estimated with the 'effsize' package. 87 In our case, a positive effect size indicates that the plants with AMF increase their P or N content compared to control, non-mycorrhized, plants. We next calculated group effect (high versus low mycorrhizal species) using phylogenetic generalized least squares (PGLS) models, with the l values estimated by maximum likelihood (ML) (pGLS function in 'caper' package 85 ) on the pruned phylogenetic tree.
3) AMF effect on univariate plant growth and chemical traits -We tested the overall effect of AMF inoculation on growth traits (total biomass, root-shoot ratio, SLA, and LDMC), and chemical defense ll OPEN ACCESS iScience Article