Intraspecific variation in mycorrhizal response is much larger than ecological literature suggests

Mycorrhizal response is the most common metric for characterizing how much benefit a plant derives from mycorrhizal symbiosis. Traditionally, ecologists have used these metrics to generalize benefit from mycorrhizal symbiosis in plant species, ignoring the potential for plant intraspecific trait variation to alter the outcome of the mutualism. In order for mean trait values to be useful as a functional trait to describe a species, as has been attempted for mycorrhizal response traits, interspecific variation must be much larger than intraspecific variation. While the variation among species has been extensively studied with respect to mycorrhizal response traits, variation within species has rarely been examined. We conducted a systematic review and analyzed how much variation for mycorrhizal growth and nutrient response typically exists within a plant species. We assessed 28 publications that included 60 individual studies testing mycorrhizal response in at least five genotypes of a plant species, and we found that intraspecific trait variation for mycorrhizal response was generally very large and highly variable depending on study design. The difference between the highest and lowest growth response in a study ranged from 10% to 350% across studies, and 36 of the studies included species for which both positive and negative growth responses to mycorrhizae were observed across different genotypes. The intraspecific variation for mycorrhizal growth response in some of these studies was larger than the variation documented among species across the plant kingdom. Phosphorus concentration and content

testing mycorrhizal response in at least five genotypes of a plant species, and we found that intraspecific trait variation for mycorrhizal response was generally very large and highly variable depending on study design.The difference between the highest and lowest growth response in a study ranged from 10% to 350% across studies, and 36 of the studies included species for which both positive and negative growth responses to mycorrhizae were observed across different genotypes.The intraspecific variation for mycorrhizal growth response in some of these studies was larger than the variation documented among species across the plant kingdom.Phosphorus concentration and content was measured in 17 studies and variation in phosphorus response was similar to variation in growth responses.We also found that plant genotype was just as important for predicting mycorrhizal response as the effects of fungal inoculant identity.Our analysis highlights not only the potential importance of intraspecific trait variation for mycorrhizal response, but also the lack of research that has been done on the scale of this variation in plant species.Including intraspecific variation into research on the interactions between

INTRODUCTION
Traditionally, community ecology overlooks how the structure and function of ecosystems are impacted by intraspecific trait variation (Violle et al., 2012).Researchers often use species mean trait values as parameters in theoretical approaches, with no measure of variation around this mean.The usage of these values assumes that intraspecific variation is negligible compared with interspecific variability, although this is rarely the case (Albert et al., 2010).In fact, when variation in plant functional traits was measured across five plant species and 56 different sites, intraspecific trait variation exceeded that of between-species trait variation in 9 out of 14 traits (Tautenhahn et al., 2019).Current understanding of the role of intraspecific trait variation in ecosystems suggests that, while variation between species is greater than within species, intraspecific trait variation can be more important than interspecific variation when assessing species interactions and community processes (Des Roches et al., 2018;Siefert et al., 2015).The impact of intraspecific variation is as important as species variation in scenarios in which indirect interactions between species alter community composition, such as plant-microbe interactions that occur belowground (Des Roches et al., 2018;Fridley & Grime, 2010).For example, plant genetic variation in pine trees affects root microbiome function, altering nutrient cycling and potentially feeding back to plant growth in the surrounding community (Pérez-Izquierdo et al., 2019).Despite evidence that variation within plant species can dramatically alter interactions with other species and the surrounding environment, there is currently limited knowledge of how plant intraspecific variation affects one of the most fundamental components of the root-associated microbiome, mycorrhizal fungi.
Mycorrhizal fungi form one of the oldest and perhaps most important symbioses with plants (Brundrett & Tedersoo, 2018;Delaux et al., 2015).They are a major driver of plant diversity, composition, and productivity in many ecosystems by providing plants access to scarce and unavailable nutrient resources in exchange for carbon sugars, which in turn increases the growth and fitness of the plant (Bauer et al., 2012;van der Heijden et al., 1998).
Outside the field of ecology, some progress has been made in understanding how plant genetic variation affects mycorrhizal symbiosis.Using genome-wide association mapping, researchers have identified gene variants within plant species that affect mycorrhizal colonization rate (Davidson et al., 2019;Stahlhut et al., 2021) and plant response to mycorrhizal symbiosis (Lehnert et al., 2018).Other researchers have found distinct gene expression patterns in both plant genotypes and fungal isolates during mycorrhizal symbiosis, suggesting the presence of genes responsible for specificity in response (Savary et al., 2020;Watts-Williams, Emmett, et al., 2019).These findings show that traits related to a plant's response to mycorrhizal symbiosis are variable within species, polygenic, and heritable.These attributes of the mycorrhizal response indicate that intraspecific trait variation for mycorrhizal response is likely to be important for both plant and fungal community resilience and ecosystem processes (Barab as & D'Andrea, 2016;Johnson et al., 2012).
A plant's response to mycorrhizal symbiosis can range from parasitic to obligately mutualistic (Johnson et al., 1997).The mycorrhizal response is known to be dependent on phosphorus limitation, nitrogen limitation, and mycorrhizal fungi identity, which confounds attempts to generalize mycorrhizal response within a species (Bever et al., 2009;Hoeksema et al., 2010;Savary et al., 2020).The effect of plant genotype has been less studied, but it can also affect the mycorrhizal response, primarily through differences in nutrient allocation strategies, partner specificity, and local adaptation (Delavaux & Bever, 2022;Koch et al., 2017;Riley et al., 2019;Wagg et al., 2015).Despite knowing that intraspecific trait variation for mycorrhizal response in plants is also variable, most studies do not include plant classification levels beyond species or assess the relative importance of different sources of variation in mycorrhizal response.
In this paper, we assess the plant intraspecific trait variation for the mycorrhizal response.Our first goal was to document the degree of intraspecific trait variation for a mycorrhizal response within plant species.We drew data from publications that tested mycorrhizal biomass and phosphorus response of five or more genotypes, ecotypes, or populations of a plant species in a single mycorrhizal species inoculum environment, so that we could document the range and distribution of intraspecific trait variation for mycorrhizal response.We compared these data to the range of species means across the plant kingdom.Then, we also used a subset of these studies to determine the relative importance of plant genotype compared with fungal inoculant identity, which is currently known to affect mycorrhizal response.Finally, we assessed the correlation between biomass response and phosphorus response within plant species in order to determine how biomass and phosphorus response were correlated across different studies.With these analyses, we illustrate the potential impact of plant intraspecific variation on the measured quality of mycorrhizal interactions.We use these analyses to generate a greater understanding of the potential role intraspecific trait variation plays in the functioning of mycorrhizal symbiosis.

METHODS
In order to evaluate the degree of intraspecific trait variation for the mycorrhizal response, we conducted a repeatable search of peer-reviewed publications in the Web of Science.We utilized broad search terms in order to capture any publication that may have been relevant to the scope of the analysis (Appendix S1).The final search was conducted on 11 July 2022 and resulted in 1471 total publications.
Publications were included based on three inclusion criteria.First, publications must measure growth or nutrient response to mycorrhizal colonization.These response metrics require both mycorrhizal and control plants to be grown in sterile background soil with otherwise similar plant growth conditions.Second, each publication must measure mycorrhizal response in at least five different genotypes of a plant species, an arbitrary minimum selected to provide a reasonable estimate of within-species variation.Finally, inoculated treatments must only contain a single fungal isolate tested across all plant genotypes.This is to ensure that the measurement of growth can be directly attributed to the interactions between the plant genotype and the fungal isolate being tested, as plant and fungal isolate can favor certain partners and allocate differentially in mixed fungal-isolate environments (Bever et al., 2009;Hoeksema et al., 2010;Savary et al., 2020).Publications that tested multiple single isolate inocula were considered as separate parallel experimental contrasts with a common control.Publications that included the same plant genotypes and fungal isolates in different nutrient environments (e.g., high and low soil phosphorus concentrations) were also considered as separate experimental contrasts, if control plants for each nutrient environment were included.
Two researchers independently evaluated each search result by screening the title, abstract, and full text.A conservative screening of the titles excluded publications that were not relevant to the study in topic or scope.The abstracts of the remaining publications were screened for relevance.The remaining publications were read to determine whether they met all of the inclusion criteria.At this stage, publications were excluded for one of the following reasons: plants were tested with a mixed inoculum or unsterilized soil, they did not meet the minimum threshold of five plant genotypes, they did not have a properly sterilized control for growth response comparisons, they did not include a measure of plant benefit that could be used for this study, or they were meta-analyses that did not include unique data (Appendix S2).In total, 24 publications from the literature search met all inclusion criteria and were included in the analysis.We screened the citations of all publications that made it past the abstract review for relevance and the full text of any relevant citations was also examined.This screen added four additional older publications that were not indexed by the Web of Science database, for a total of 28 publications used in the final analysis.

Data extraction
Once the screening process was complete, data were extracted from each of the publications.Three focal variables were identified for data extraction.The first variable was mycorrhizal growth response (MGR), calculated as the difference between dried aboveground biomass of inoculated and control plants in each experimental contrast, relativized by the control (Hetrick et al., 1992).The second and third variables, mycorrhizal phosphorus response (MPR) of phosphorus concentrations and phosphorus content of aboveground plant tissue (which are referred to as MPR-concentration and MPR-content throughout this paper), were calculated using the same formula as MGR.Phosphorus concentration reflects the phosphorus status of shoot tissues produced, while the content reflects a measure of total phosphorus taken up and incorporated into the shoot tissues.For all three of these variables, results are bounded by −1 and infinity, such that when a genotype has an MGR equal to 1, the inoculated plant has an aboveground dried biomass that is twice as large as that of the control plant, and when a genotype has a value of −0.5, the inoculated plant has an aboveground biomass that is half that of the control.A value of zero indicates that the control and inoculated plants of a genotype performed the same.To calculate these metrics, the biomass and phosphorus data for inoculated and control plants were extracted from the publications and calculated to ensure data consistency, even when data for MGR or MPR were presented as well.Generally, publications included biomass and phosphorus data in either the text or in the supplemental material of the paper.If data were not available within tables in the publication or in the supplemental material, data were extracted from figures within the publications using WebPlotDigitizer v. 4.4 (Rohatgi, 2020).
The proportion of total genetic variation represented by the selection of genotypes in each publication was impossible to ascertain accurately, but more representative studies (those that assess larger numbers of genotypes) are inherently more likely to contain higher degrees of variation for the traits of interest.To account for this, we chose to distinguish between publications that had low, medium, and high representations of genetic variation.High diversity publications were defined as having more than 15 genotypes that were explicitly chosen to be highly representative of species' diversity, such as pre-established diversity panels created for assessing trait variation.Medium diversity publications were defined as having either more than 15 genotypes that were not chosen to be representative of the entire species or included fewer than 15 genotypes that were explicitly chosen to represent the entire species.Low diversity publications were defined as those that included fewer than 15 genotypes that were not chosen to be representative of the species diversity.The designation to these categories was based explicitly on the information given about genotype selection in the text of the publication.While it would be ideal to weight publications by the proportion of genetic variation included in each study, because this information was not available and these bins are rough approximations of diversity, we have chosen instead to report the average intraspecific trait variation for MGR for each diversity classification.
We have chosen to report the total range of variation for each study instead of reporting calculated metrics like coefficients of variation for mycorrhizal responsiveness for three important reasons.First, while the studies that included high diversity genotype selections may encapsulate most of the within-species variation for mycorrhizal responsiveness in a given environment, coefficients of variation for growth response are likely to vary greatly across different nutrient environments (e.g., Wang et al., 2020).Second, because most of the studies included in this analysis are cultivated species, a single estimate of total intraspecific variation based on these species may not reflect the actual diversity within larger subsets of species.Third, the coefficient of variation highly depends on the mycorrhizal dependency on genotypes, because the potential benefit is bounded by infinity, whereas the potential negative effects of mycorrhizal symbiosis are bounded by −1.We have instead included the average range of variation for the high, medium, and low diversity selections.The error inherent in this estimate is large, as we were unable to control for many variables that are known to affect the variability in response to mycorrhizal symbiosis such as nutritional conditions and mycorrhizal identity (Bever et al., 2009;Hoeksema et al., 2010;Savary et al., 2020), but we used this estimate as a guide for potential variation in mycorrhizal response in controlled experimental conditions.
In order for a trait to be usefully scored at a species level for studies that rely on quantifying interspecific variation, like in community ecology and macroevolution, interspecific variation should be greater than intraspecific variation.To assess whether the variation across species for MGR was larger than the variation within species, we compared the range of intraspecific trait variation for mycorrhizal response with the range of response in two extensive datasets, MycoDB (Chaudhary et al., 2016;version 4) and Reinhart et al. (2017).MycoDB is a large database of MGR generated through many separate inoculation experiments and includes 348 plant species.While the species means for MGR were not generated in the same environmental conditions, this database contains the most thorough reporting of mycorrhizal response across the plant kingdom.Reinhart et al. (2017) include 69 native and invasive Northern Plains plant species with limited genetic diversity grown with a multispecies inoculant and is one of the largest datasets for the mycorrhizal response for species grown in the same experimental conditions.This dataset contains species found in a single ecoregion, which removes any confounding effects of comparing across ecosystem types and provides a full scale of species responses in natural conditions.The range of variation in MGR measured as species means in these two datasets was compared with the range of intraspecific variation in MGR across studies that met our inclusion criteria.
We sought to determine whether the effect of plant genotype was as important as another well established driver of mycorrhizal response, mycorrhizal fungal identity.To do this, we used data from studies that used two or more fungal treatments to calculate the median absolute deviation (MAD) of plant genotype and fungal inoculant using the "mad" function in the stats package in R version 4.1.1(R Core Team, 2021).For the purpose of this study, the MAD estimates for treatments represent the total variation in mycorrhizal response that can be attributed to a particular treatment within a study and, as such, should only be compared within a study.We used a paired sample t-test to determine whether the MAD for plant genotype from each study was significantly different from the MAD for either phosphorus or fungal treatment.A nonsignificant result would indicate that the dispersion around the mean is not significantly different for plant genotype and fungal or phosphorus treatment, suggesting similar scales of effects on mycorrhizal response.
To assess the relationship between biomass response and phosphorus uptake response, Pearson correlation coefficients (r) and 95% confidence intervals were calculated between MGR, MPR using plant total P content, and MPR using total P concentration in plant tissue for each study that included nutrient data.The correlations were used as effect sizes in meta-analyses to calculate mean correlation coefficients between these three traits.Correlation coefficients were converted using Fisher's z-transformation and used in a random effects model, with effect sizes weighted by the inverse variance method.The metaanalysis of correlations was implemented with the "metacor" function in the meta package using restricted maximum likelihood to estimate between-study variance (Schwarzer, 2007).The between-study variance was estimated using a random effects model with a Hartung-Knapp variance adjustment (r) (Hartung & Knapp, 2003;Viechtbauer, 2005).Results are presented using backtransformed Pearson correlation coefficients (r).

RESULTS
Data from 28 peer-reviewed publications fit the inclusion criteria and were used in this analysis.For MGR, in total, 60 experimental contrasts were included in the final analysis (Table 1).For MPR, 17 experimental contrasts included both phosphorus content and concentration.All publications used for MPR analyses were also used in the MGR analysis, such that the former contrasts are a subset of the latter.These publications were published from 1994 to 2022, and 14 out of 28 were published after 2015.The selected publications included as few as five to as many as 334 genotypes.The genetic variation represented by the plant genotypes used in each publication was typically not assessed, although some publications did use established diversity panels that included representative genotype selections from their focal species.Generally, publications that included high amounts of variation tended to be more recent and the genotypes were generally selected to represent genetic diversity that would allow for adequate screening for mycorrhizal traits in a species (Figure 1).Publications that included medium or low amounts of variation often selected important cultivars, selections from major groups within the species, or genotypes that had been used in previous publications.
In total, nine plant families and four fungal families are included in this analysis.More than half of the experimental contrasts used plant species from the Poaceae family, including three publications that used rice (Oryza sativa) and three that used corn (Zea mays).Four publications used species from the Fabaceae, two of which were Medicago truncatula (Schultz et al., 2010;Watts-William, Emmett, et al., 2019).The other seven families represented in this analysis (Amaranthaceae, Amaryllidaceae, Asteraceae, Linaceae, Pinaceae, Rosaceae, and Rubiaceae) are only represented by one publication each.Except for the publications that examined response in Picea abies, M. truncatula and Triticum turgidum subsp.dicoccoides (Mari et al., 2003;Schultz et al., 2010;Watts-William, Emmett, et al., 2019;Yücel et al., 2009), all plant species used in this analysis are commonly cultivated.Most experimental contrasts used arbuscular mycorrhiza (AM) fungal species belonging to the family Glomeraceae, with the most commonly used species being Rhizophagus irregularis (Błaszk., Wubet, Renker & Buscot) C. Walker & A. Schüsler.Three contrasts used Gigaspora margarita, a member of the Gigasporaceae.Despite using broad search terms, the only ectomycorrhizal plant and fungal species in selected publications was P. abies with its symbiont Laccaria bicolor (Mari et al., 2003).

Degree of variation in MGR
Across the 60 experimental contrasts that measured biomass of control and inoculated plants, the range of variation in MGR ranged from 0.1 to 20.7, representing a 10% to a 2270% range in plant growth response to AM fungi among different genotypes of a plant species (Figure 2A).The median range of MGR within a species was 0.90.The three publications with the largest amount of intraspecific trait variation for MGR (Eltigani et al., 2022;Tawaraya et al., 2001;Vestberg, 1992) included genotypes that were highly responsive to mycorrhizal fungi and genotypes that were not, despite the fact that these contrasts included genotype selections that were not highly representative of the genetic diversity within the plant species.Of the 60 experimental contrasts, 36 had plant genotypes with a positive MGR as well as genotypes with a negative MGR in the same environment.Among the contrasts that include highly diverse genotype selections (Davidson et al., 2019;Ellouze et al., 2016;Suzuki et al., 2015;Watts-William, Cavagnaro, et al., 2019;Watts-William, Emmett, et al., 2019;Yücel et al., 2009), the variation across genotypes for MGR is 1.91 by experimental contrast and 1.45 by publication (Figure 2B).In all experimental contrasts, including high diversity genotype selections, there were genotypes that responded positively and genotypes that responded negatively to inoculation treatments.
Unsurprisingly, variation in MGR was higher in interspecific plant selections.In the global database of Note: Sample size represents the no.plant genotypes included in each comparison.The diversity classifications are based on information given about genotype selection in the text of each publication.High diversity publications were defined as having more than 15 genotypes and that were explicitly chosen to be highly representative of a species' diversity, such as a pre-established diversity panels created for assessing trait variation.Medium diversity publications were defined as having either more than 15 genotypes that were not chosen to be representative of the entire species or included fewer than 15 genotypes that were explicitly chosen to represent the entire species.Low diversity publications were defined as those that included fewer than 15 genotypes that were not chosen to be representative of the species diversity.
MGR, MycoDB, the range of MGR was 6.74 for 348 represented plant taxa (Chaudhary et al., 2016).The range of MGR across 95 native and invasive Northern Plains plant taxa was 5.60 when tested in similar environmental conditions using multispecies inoculum (Reinhart et al., 2017).Although these ranges were larger than most studies assessing intraspecific variation in MGR, three studies had larger intraspecific variation in MGR than the database containing 348 species from across the globe grown in variable conditions, and four studies had larger intraspecific variation than interspecific variation among 69 species found in a single ecosystem grown in the same environmental conditions (Eltigani et al., 2022;Tawaraya et al., 2001;Vestberg, 1992).
Eight publications tested plant responses to multiple fungal isolates.When MAD were paired with plant genotype mean absolute differences, they were not significantly different for fungal inoculation treatment (t = 0.002, df = 7, p-value = 0.99) (Figure 3).This indicates that the effect of plant genotype on MGR can be as large as the effect of fungal inoculant identity.There was not enough power to test whether there were interactive effects between plant genotype and fungal inoculant identity, but that does not mean that these G × G interactions are not important.The high diversity studies Watts-William, Cavagnaro, et al. ( 2019

Degree of variation in MPR
Fifteen experimental contrasts from 12 publications assessed intraspecific variation for phosphorus accumulation in plant tissues for both inoculated and uninoculated plants (Figure 4).Generally, the variation in MPR was as large or larger than the variation in MGR.We considered two measures of MPR: (1) the total amount of phosphorus contained in inoculated relative to uninoculated plants (MPR-content) and ( 2) the concentration of phosphorus in plant tissue (MPR-concentration).The median variation in MPR-content was 1.04 and the median variation for MPR-concentration was 0.81.Allium fistulosum exhibited the largest variation in MPR, with ranges of 42.7 and 7.6 for MPR-content and MPR-concentration, respectively (Tawaraya et al., 2001).
Because an increase in plant nutrient acquisition through the mycorrhizal symbiosis may result in either (1) the production of additional biomass (as seen with MGR), (2) an increase in the nutrient status of existing tissues, or (3) both of these phenomena, it is important to capture the effects of mycorrhizal inoculation on both total nutrient content in the plants (MPR-content) and concentration of nutrients in the tissue (MPR-concentration).The weighted correlation coefficient for MGR and MPR-content was positive and did not overlap zero (r = 0.79, lower 95% confidence interval = 0.53, upper 95% confidence interval = 0.91) (Figure 5).This result suggests that the main consequence of additional plant phosphorus uptake The variation in mycorrhizal growth response (MGR) in genotypes of a given plant species, separated by experimental contrast.The experimental contrasts are ordered by the range and colored by low (pink), medium (green), and high (blue) diversity levels.MycoDB (Chaudhary et al., 2016) and Reinhart et al. (2017) show the interspecific diversity in MGR across the plant kingdom and in prairie species respectively.Three studies had larger variation in MGR than the database containing 348 species from across the globe grown in variable conditions and four studies had larger variation than 69 species found in a single ecosystem grown in the same environmental conditions (Eltigani et al., 2022;Tawaraya et al., 2001;Vestberg, 1992).The boxes represent the first and third quartiles and the whiskers extend to the highest and lowest values.Points represent outliers (more than 1.5× interquartile range from the first or third quartile) in each study.One outlier from Tawaraya et al. ( 2001) (MGR = 20.7) was not included in this figure.Numbers to the left of the species name indicate the study that the data were drawn from Table 1.(B) The range of variation in MGR in experimental contrasts that included low, medium, and high amounts of variation.Tawaraya et al. (2001), range of MGR = 20.7, is an outlier and is not shown in (B).
was additional plant growth, resulting in higher aboveground biomass.In contrast, the weighted correlation coefficient for MGR and MPR-concentration was negative and also did not overlap zero (r = −0.33,lower 95% confidence interval = −0.51,upper 95% confidence interval = −0.14)(Figure 5), suggesting that the total phosphorus concentration in aboveground tissues decreases as the aboveground biomass increases.These two results combined show that mycorrhizal symbiosis generally increases biomass productivity by producing more plant tissues with lower nutrient concentrations.The weighted correlation coefficient for MPR-content and MPR-concentration was variable and overlaps zero (r = 0.41, lower 95% confidence interval = −0.12,upper 95% confidence interval = 0.76) (Figure 5), indicating differences in allocation to growth versus luxury consumption (i.e., taking up more nutrients than needed for growth) across experimental contrasts (Riley et al., 2019).For example, in corn, while the MPR metrics are positively correlated with each other when the genotypes are inoculated with G. margarita, they are negatively correlated when the same genotypes are inoculated with R. irregularis (Khalil et al., 1994).

DISCUSSION
Although researchers have traditionally studied the effects of mycorrhizal symbiosis at the interspecific scale, this analysis displays the extensive amount of trait variation for the mycorrhizal response that can also be found within plant species.While it is no surprise that individuals within a species are variable, the scale of the intraspecific trait variation for mycorrhizal response is larger than previously recognized.Our findings show that, even when fungal genotype and environmental conditions are controlled for, the variation across genotypes can be F I G U R E 3 The median absolute differences (MAD) between mean mycorrhizal growth responses (MGR) for fungal inoculant identity for all studies that measured MGR in multiple fungal inoculants).Each point is paired with the respective MAD for plant genotype effect, measured in each study.Plant genotype effect was not significantly different from fungal inoculant (t = 0.002, df = 7, p-value = 0.99).Study number (as referenced in Table 1) is reported beside each line.
F I G U R E 4 The variation in mycorrhizal phosphorus response (MPR) using (A) concentration of P in plant leaf tissue and (B) using P content in aboveground plant tissue in genotypes of a given plant species.The experimental contrasts are ordered by the range and colored by low (pink), medium (green), and high (blue) diversity levels.The boxes represent the first and third quartiles and the whiskers extend to the highest and lowest values.Points represent outliers (more than 1.5× interquartile range from the first or third quartile) in each study.The numbers to the left of the species name indicate the study that the data were drawn from (Table 1).almost as great as the variation in species means.The large amount of intraspecific variation for this trait limits the application of MGR as a continuous functional trait in community ecology, as traits used at this scale are expected to be generalizable at the species level (McGill et al., 2006).These results complement the work of previous researchers who showed the effects that plant ecologists typically study (e.g., plant-soil feedbacks and plant community dynamics) are as strongly affected by intraspecific variation as by interspecific variation (Des Roches et al., 2018;Pérez-Izquierdo et al., 2019).The presence of large amounts of intraspecific variation for mycorrhizal response mirrors the large degree of variation found in plant functional traits (Tautenhahn et al., 2019).While some species may exhibit little variation in the trait, the mycorrhizal response is often highly variable within species and, as such, the inclusion of estimates of variation inherent within species means should be used in future models that include mycorrhizal effects.
Variation in MPR was, in many cases, as large as variation in MGR, but how plants allocated increased phosphorus depends on the plant and fungal genotype.Resource limitation is a major driver of local adaptation to mycorrhizal fungal communities in plants (Johnson et al., 2010), and this adaptation may result in genotypic differences in a plant that affect MGR through a difference in nutrient allocation strategy (Riley et al., 2019).Future efforts should continue to quantify mycorrhizal responsiveness of other plant functional traits, like resource allocation, in order to solidify our understanding of the role of belowground mutualisms in the plant economic spectrum.This approach would probably provide a better predictive framework for a mycorrhizal response, especially given that the phylogenetic signal for responsiveness traits is low and the traits are often divergent among closely related species (Reinhart et al., 2017).It is also important to note that symbiont identity can also affect nutrient allocation strategies.In one scenario (e.g., G. margarita in corn), improved nutrient supply increases shoot phosphorus content, which drives additional shoot growth (although at a proportionally slower pace), and thus increases phosphorus concentration in aboveground tissues.In an alternate scenario (e.g., R. irregularis in corn), stimulation of shoot growth by other mechanisms drives shoot growth at a faster pace than improved nutrient supply to the shoot, such that while the overall phosphorus content increases, the phosphorus concentration of the shoot tissues declines (Khalil et al., 1994).These patterns indicate that stimulation of plant growth by mycorrhizal symbiosis may occur either by improved phosphorus supply or via other mechanisms (e.g., hormonal signaling [Liu et al., 2018;Pons et al., 2020]; nitrogen and water uptake and disease resistance [Delavaux et al., 2017], nutrient dilution effect with increasing plant productivity [Welti et al., 2020]) and that the relative degree of stimulated increase in growth versus increase in nutrient supply among mycorrhizal fungal species (or genotype) may determine plant MGR and MPR dynamics.Given the diversity of nutrient and biomass responses to mycorrhizal fungi, quantifying the impact of different mycorrhizal species on plant physiology could F I G U R E 5 Effect sizes for the correlations between mycorrhizal growth response (MGR) and mycorrhizal phosphorus response (MPR) using P content and P concentration.Each square represents effect size (correlation coefficient, r) and whiskers represent 95% confidence intervals.The size of each square represents the weight of the study in analysis.The overall effect for each correlation (pooled weighted correlation coefficient) is represented by the black diamond.The references for each experimental contrast can be found in Table 1.
help researchers understand how ecological processes mediated by plant carbon-nutrient balance, like herbivory or litter decomposition, are affected by mycorrhizal symbiosis as a whole as well as with different mycorrhizal functional types.
Through this study, we highlight the lack of research on intraspecific variation for mycorrhiza response, particularly in phylogenetically and functionally diverse plant and mycorrhizal fungi species.Cultivated plant species, members of Poaceae, and model arbuscular mycorrhizal fungi were overrepresented in research on intraspecific variation in mycorrhizal symbiosis.Of the nine studies that used high diversity genotypes, only one included a noncultivated Poaceae species (Watts-Williams, Emmett, et al., 2019) and only two included mycorrhizal species that did not belong to Glomeraceae (Mari et al., 2003;Watts-Williams, Cavagnaro, et al., 2019).As a result, any estimates of average variation in mycorrhizal response would probably be heavily biased toward these overrepresented plant and fungal families, which may not be representative of the within-species variation in mycorrhizal response across the plant kingdom.We currently lack any information about the diversity of responses in most plant families, including many of the largest and most ecologically important.Also, because it was impossible to estimate the total genetic diversity represented by the selection of genotypes used in each study, these results may be drastically underestimating the diversity found within a given plant species.Some studies only used a small or regional selection of genotypes, which led to a lower degree of variation for MGR.Most of the studies also used crop species, many of which have gone through domestication bottlenecks that have a decreased genetic diversity of the cultivated species relative to wild progenitors (Flint-Garcia, 2013).It is possible that the degree of variation is even larger within wild species, particularly those with extensive ranges or those that occupy diverse environments.Additionally, as artificial selection typically favors faster and larger growth in annual crops, the correlation between MGR and MPR may be more variable than the meta-analysis of correlation coefficients suggests (Figure 5).The fungal species represented in this study were highly concentrated into a handful of genera and families, and there is limited information on the effects of plant genetic variation on nonmodel arbuscular mycorrhizal species and other types of mycorrhizal symbioses.Given the outsized role that some of these groups have in ecosystem services (e.g., ectomycorrhizal and ericoid fungi in cold climates [Soudzilovskaia et al., 2019]) and species fitness (e.g., orchid mycorrhizal fungi in the Orchidaceae [Li et al., 2021]), there is an urgent need to investigate how genetic variation affects mycorrhizal response across the vast diversity of plant and mycorrhizal fungi species.
The presence of such substantial variation in MGR and MPR is encouraging for researchers that are trying to maximize the benefits of mycorrhizal symbionts to a plant, but it also creates a formidable challenge to understanding these interactions.Identifying genes associated with mycorrhizal response in different plant species, fungal species, and environments can elucidate the genetic underpinnings of mycorrhizal response and plant-fungal specificity and we can incorporate these findings into current breeding programs.Even less understood is the importance of genetic variation for the mycorrhizal response for species in natural environments.To address this gap of knowledge, we should understand the source of this trait variation and how this variation is structured across populations and environments (Moran et al., 2016;Siefert et al., 2015).Although incorporating intraspecific variation for the mycorrhizal response into models is still difficult with our limited knowledge, by further expanding our understanding of variation for mycorrhizal response to include different taxonomic scales, environments, and ecosystems, we can better predict the relative importance of within-species variation on mycorrhizal response (Westerband et al., 2021).
In summary, this paper uncovers a startling amount of intraspecific variation for traits commonly used to measure the plant benefit from mycorrhizal symbiosis.We challenge a core assumption underpinning much of the research done in community ecology on plantmycorrhizal interactions that intraspecific variation in plant response to mycorrhizal symbiosis can be appropriately ignored.It is becoming increasingly apparent that a plant's response to mycorrhizal symbiosis relies heavily on the genotype of both partners.To move forward with improving ecological theory, as well as enable applied efforts to increase plant-positive responses to mycorrhizal fungi, we need to understand both the genetic underpinnings of this genotypic specificity and the major drivers of mycorrhizal response.Ultimately, individuals within a plant species vary greatly in their response to mycorrhizal fungi and further incorporation of this variation into research on the interactions between plants and their symbionts can increase our understanding of plant coexistence and ecological stability.

DATA AVAILABILITY STATEMENT
This paper utilizes data from other sources (see Table 1 for details).The compiled data used for analyses (Stahlhut et al., 2023) ) and Erdi ̇nç et al. (2017) reported MGR for each genotype with multiple different fungal species and visually appear to have G × G interactions (Appendix S3).

F
I G U R E 1 Overview of the studies used in this analysis.(A) Cumulative number of publications included in this analysis by publication year and diversity classification.(B) Distribution of plant families by publication.(C) Distribution of fungal families by experimental contrast.Note that some publications used more than one fungal family, but all publications used only one plant species.
Final publications used in this study.