Fungal Fight Club: phylogeny and growth rate predict competitive outcomes among ectomycorrhizal fungi

Abstract Ectomycorrhizal fungi are among the most prevalent fungal partners of plants and can constitute up to one-third of forest microbial biomass. As mutualistic partners that supply nutrients, water, and pathogen defense, these fungi impact host plant health and biogeochemical cycling. Ectomycorrhizal fungi are also extremely diverse, and the community of fungal partners on a single plant host can consist of dozens of individuals. However, the factors that govern competition and coexistence within these communities are still poorly understood. In this study, we used in vitro competitive assays between five ectomycorrhizal fungal strains to examine how competition and pH affect fungal growth. We also tested the ability of evolutionary history to predict the outcomes of fungal competition. We found that the effects of pH and competition on fungal performance varied extensively, with changes in growth media pH sometimes reversing competitive outcomes. Furthermore, when comparing the use of phylogenetic distance and growth rate in predicting competitive outcomes, we found that both methods worked equally well. Our study further highlights the complexity of ectomycorrhizal fungal competition and the importance of considering phylogenetic distance, ecologically relevant traits, and environmental conditions in predicting the outcomes of these interactions.


Introduction
Ectomycorrhizal fungi (EMF) are obligate plant mutualists that associate with 60% of all tree stems on Earth (Steidinger et al. 2019 ) and participate in nutrient trading that is especially important for woody temperate plants (Smith and Read 2008 ).These fungi are important e v en in the earl y sta ges of a host plant's life, as seedling survival and biomass are directly correlated with a host's ability to acquire fungal partners (Onguene and Kuyper 2002 ).Benefits of the mycorrhizal symbiosis for host plants include the physical extension of a host's resource pool, utilization of enzymes to access r ecalcitr ant nutrients, and e v en pr otection a gainst r oot pathogens (Leake et al. 2004 ).Because of these functions , EMF pla y a k e y role in forest carbon and nitrogen cycling and above-and belowground di versity (Leak e et al. 2004 ).Ad ditionall y, EMF ar e both taxonomically and functionally diverse.For example, EMF vary in enzymatic activity (Courty et al. 2005 ), mycelial growth (Cairney 1999 ), and host pr efer ence (Tedersoo et al. 2008 ).T hus , fungal community composition can affect both large-scale ecological cycles as well as the health of an individual host plant.Understanding how these communities assemble as a function of biotic and abiotic interactions is therefore paramount to the field of forest ecology.
Competition is a k e y factor structuring communities of EMF (Koide et al. 2005, Kennedy 2010 ).Ectomycorrhizal fungal comm unities ar e complex mosaics, e v en on the scale of meters (Taylor and Bruns 1999, Zhou and Hogetsu 2002, Anderson et al. 2014 ), and a single ectomycorrhizal plant host can harbor tens to hun-dreds of EMF in its rhizospher e (Bahr am et al. 2011, Thoen et al. 2019 ), e v en m ultiple genets of the same fungal species (Hortal et al. 2012 ).This diversity within small spatial scales would suggest that EMF may tend to occupy similar substrates or use ov erla pping soil resources.Consistent with this, many EM fungal comm unities a ppear to be structur ed by competition (Wu et al. 1999, Koide et al. 2005, Pickles et al. 2012 ).Ev en their r esponse to environmental factors like temperature and their relative investment in symbiotic and non-symbiotic tissues can be affected by the presence of a competitor (Hortal et al. 2016 ).Furthermore, different species of EMF differ in both competitive ability (Kennedy et al. 2007, Maynard et al. 2017 ) and mutualistic function (i.e.decomposition ability and enzymatic activity) (Lindahl andTunlid 2015 , Moeller andPeay 2016 ), and it has been shown that host plant performance can be impaired by associating with several competing fungi (Kennedy et al. 2007 ).Being able to predict the outcomes of competition between species will have implications both for fungal ecology and for the health of host plants (Kennedy et al. 2007, Hortal et al. 2017 ).
In addition to competition, abiotic factors can also structure these comm unities.Pr operties suc h as soil pH (Yamanaka 2003, Gryndler et al. 2017, Davison et al. 2021 ), temper atur e (Davison et al. 2021 ), and nutrient availability (Huggins et al. 2014, Sterkenburg et al. 2015 ) have all been observed to change the composition of mycorrhizal fungal communities.In particular, pH can affect fungal growth in vitro (Hung andTrappe 1983 , Yamanaka 2003 ) and in situ (Ge et al. 2017, Glassman et al. 2017 ).Possible mechanisms by which pH affects fungal performance include changes in enzymatic efficacy (Leake and Read 1990 ) and spore production and germination (Siqueira et al. 1984, Coughlan et al. 2000 ).pH also has been shown to vary in orders of magnitude across small spatial and temporal scales (Blossfeld et al. 2013 ), affecting micr obial comm unities (Lauber et al. 2009 ).T hus , an EM fungus ma y ha v e to ada pt to differ ent pH envir onments, both acr oss its own mycelium and as it disperses to new en vironments .T he extent to which an individual EM fungus succeeds in adjusting to these variations in pH may have po w erful effects on competitive outcomes and community structure in soils.
While it is important to understand how competition and abiotic factors individually affect ectomycorrhizal fungal community structur e, competitiv e inter actions and comm unity assembl y ha ppen in the context of the abiotic environment, and so, can be affected by environmental factors.For example, Mujic et al. ( 2016 ) demonstrated in a greenhouse study that abiotic soil conditions can change the outcomes of competitive interactions between fungi, with heterogeneous soil facilitating coexistence.Lik ewise, the competiti ve dominance of two EMF species ( Piloderma ) r e v ersed with the addition of wood ash to the growth substrate (Mahmood 2003 ).These studies made complex alterations to substr ate c hemistry, but ther e is also evidence that pH specifically is an important factor affecting competition between fungi.For example, it has been shown in yeast that competitive ability, thr ough secr eted toxins, has a narrow optimal pH window, suggesting that yeast community composition is controlled by environmental pH (Chen and Chou 2017 ).As shown, the interaction between pH and competition, specifically within EMF communities, needs further study.Since substrate colonization and competition for that substrate are integral parts of EM proliferation, understanding how pH affects its success will help predict the outcomes of niche development.
Recentl y, tr ait-based ecological a ppr oac hes hav e been pr ov en to be useful in defining the environments in which a fungus can survive (Koide et al. 2014 ), highlighting optimal conditions within those environments (Van Nuland and Peay 2020 ), and predicting the outcomes of ecological inter actions, suc h as competition (Maher ali and Klir onomos 2012 ).Leake and Read ( 1990 ) sho w ed that differences in the acid tolerance of extracellular proteinases betw een tw o ericoid fungi reflected their r espectiv e soil envir onments.Another fungal trait that gr eatl y affects competitiv e outcomes is growth rate, as effective substrate colonization is a k e y mechanism for priority effects in these communities (Fukami et al. 2010 ).Additionall y, EMF comm unities often a ppear to be structured by mycelial exploration type (Koide et al. 2014, Moeller et al. 2014, Bui et al. 2020 ).Luc kil y, k e y predicti ve traits are often phylogeneticall y conserv ed (Martin y et al. 2015 ) and thus v ary pr edictabl y with phylogen y.As a consequence, phylogenetic similarity can increase competition by intensifying niche overlap, as has been shown in bacterial (Tan et al. 2012 ), fungal (Taylor et al. 2014(Taylor et al. ), y east (P eay et al. 2012 ) ), and arbuscular mycorrhizal (Maher ali and Klir onomos 2012 ) comm unities .T her efor e, phylogenetic relatedness has been suggested as a strong predictor of competitive outcomes .T his link between phylogeny and competitive outcomes has not been thor oughl y explor ed in an EMF context, though it is likely that EMF communities are structured similarly.
Ther efor e, we set out to contrast the efficacy of phylogenetic relatedness versus a trait-based approach to predict EMF responses to competition, pH, and their interaction.We hypothesized that phylogenetic distance would be the best predictor of competition and that all fungi would perform better (have higher gr owth r ates) at a lo w er pH.In our study, we gr e w fiv e EMF in single and pairwise culture assays on media at two different pH levels.We found that both phylogenetic relatedness and growth rate predicted the outcomes of competition equally well, which suggests a benefit to considering both a ppr oac hes in future work.

Methods
In order to address our hypotheses, we performed an experiment comparing fungal growth on culture plates.We grew fungi in single culture with an intraspecific competitor (self versus self, or SvS, control), and with an interspecific competitor (competition plates) at two different media pH le v els ( Supplementary Fig. S1 ).We quantified measures of growth and competitive ability in order to de v elop a network of competitive outcomes among these EMF.

Fungal cultures
To compare competitive outcomes across a broad phylogenetic range of EMF, we used five fungal cultures originally isolated from North America and Europe between 1976 and 2019 ( Supplement ary Table S1 ): four basidiomycetes ( Amanita muscaria , Hebeloma cylindrosporum , Laccaria bicolor , and Paxillus involutus ) and one ascomycete ( Cenococcum geophilum ).All of these fungi are broadly distributed host generalists (LoBuglio 1999 , Marmeisse et al. 2004, Geml et al. 2006, Hedh et al. 2008, Plett et al. 2015 ), and they (or their congeners) likely co-occur on the root systems of trees in natur e (Bahr am et al. 2011 ).Prior to our experiment, we maintained cultures on agar plates in a dark cupboard at room temper atur e (23 • C) and tr ansferr ed them e v ery two months to sustain growth.

Experimental design
To explore the effects of competition and pH on the growth of EMF, we inoculated these five fungi in a pairwise factorial experiment on two media with differing pH for a total of 20 combinations (10 pairs x 2 pH le v els = 20 competition treatments).We placed culture plugs (7/32 diameter, about 5.6 mm) on plates equidistant from the central axis ( Supplementary Fig. S2 ).Single control plates consisted of a single plug placed in the center of the plate (5 fungi x 2 pH le v els = 10 single control treatments).Intraspecific control plates consisted of two plugs of the same fungi placed in the same way as pair plates for a total of 10 combinations (5 fungi x 2 pH le v els = 10 SvS contr ol tr eatments).All tr eatments wer e grown with 10 replicates ( n = 400).
We used Modified Melin-Norkrans (MMN) medium amended with casein hydr ol ysate and a micronutrient solution ( Suppleme ntary Table S2 ).We autoclaved the phosphate-containing salts and agar/carbon substrates as separate solutions, and mixed afterw ar ds, to pr e v ent the formation of peroxides (Kawasaki and Kama gata 2017 ).Befor e autoclaving, we adjusted half the media to pH 5.6 and the other half to pH 7.0 using 1 M HCl and 1 M NaOH.These pH treatments were chosen to reflect soil pH near trees within the University of California's Sedgwick Reserve, at whic h ectomycorrhizal comm unities hav e been pr e viousl y shown to respond strongly to soil pH (Runte et al. 2021 ;Soil Survey Staff).
We stored assay plates in darkness at room temperature and took growth rate measures by regularly tracing the colony circumfer ence (Fig. 1 A).Thr oughout the experiment, we r emov ed plates exhibiting irr egular gr owth, including contamination, a dislodged plug, or dormancy ( Supplementary Fig. S3 ).We took pictures of all plates at the end of the experiment and used the area tool in the software ImageJ (Schneider et al. 2012 ) to estimate each area measurement.

Phylogenetic analysis and distance calculations
In order to determine the phylogenetic distances between fungi, we ran a multigene phylogenetic analysis with one outgroup taxon ( B .dendrobatidis ).We extracted genomic DNA from all fungal cultures, amplified the ITS region with primers ITS1F and ITS4 (Gardes and Bruns 1993 ), and sequenced amplicons with Sanger sequencing at MCLABS (South San Francisco, CA, USA).We then r etrie v ed sequences for the ITS region of our outgroup species and three additional genes ( TOP1 , TEF1 , and RPB2 ) from GenBank (Benson et al. 2013 ), known to be useful DNA barcoding markers (Lücking et al. 2020 ).We used MAFFT (Katoh et al. 2002 ) to align the sequences, estimated the phylogeny with the program PhyML (Guindon et al. 2010 ) using the TN93 substitution model selected with the Smart Model Selection tool (Lefort et al. 2017 ), and extracted the patristic distance matrix from the tree for later model calculations .T he tr ee closel y r esembled published phylogenies of EMF (Kohler et al. 2015 ).We then used the PhyML tool PRESTO (Guindon and Gascuel 2003 ) to visualize the phylogeny (Fig. 1 B).

Quantifying competition
To calculate growth rates across control and paired treatments, we fit a linear model to the natural log of colony area as a function of experimental day [ lm(log(area measurement) ∼day) ].To estimate maximum growth rates, we restricted our model fit to the time window of exponential growth (the first 3-4 days of each experiment).We used the measur ed maxim um colon y ar ea for colony size analyses.All further analyses were conducted with both the growth rate and colony size metrics, but, for coherence, only the results regarding the growth rate metric were included in the main text (see Colony Size Supplement).Since the SvS control plates consisted of two identical plugs, we r andoml y c hose one of the plugs for further analysis.
To quantify the effects of interspecific competition, we calculated the effect of competition (EoC) metric as the ratio of growth rate on competition plates versus SvS control plates [( growth in competition / growth on SvS control plate) ], modeled after the plantsoil feedback metric (Heinze et al. 2016 ).We then modeled EoC as a function of fungal ID, competitor ID, and pH treatment [ lm ( EoC ∼ fungal ID * competitor ID * pH treatment )] to test for significant interactions between these parameters ( Supplementary Table S8 ).
Since both phylogenetic distance and EoC values are pairwise metrics, we computed a pairwise distance metric for our growth measures .T his allo w ed for more logical models assessing the predictive po w er of gro wth metrics on EoC and comparing the difference in efficacy of growth metric versus phylogenetic-based models.Specifically, we calculated the difference between the mean SvS contr ol gr owth r ate in both pH conditions for each pair of fungi (i.e.gr owth r ate distance A. muscaria & C. geophilum = | mean( A. muscaria SvS growth rates in pH 7) − mean( C. geophilum SvS growth rates in pH 7) | ).
To classify the type of interaction at the front where competing fungi met, we also calculated an index of anta gonism (IoA) scor e based on methods in Wicklow and Hirschfield ( 1979 ) ( Supplement ary Fig. S4 ).This metric assigns a point value to the intensity of antagonism one fungus exhibits on its opponent fr om observ ation of physical distance and ov er gr owth at the contact zone.For example, a higher point value is aw ar ded to a fungus that grows on top of its opponent after meeting in the middle, while the fungus that is overtaken receives a low score ( Supplementary Fig. S4B ).We used this as a proxy for the effectiveness of a fungus's competitive ability.
To visualize the observed competitive hierarchical interactions, we created network plots as complete directed graphs with the R pac ka ge visNetwor k (Almende et al. 2019 ).We calculated the thickness of the control network edges as the growth rate of the fungi ov er the gr owth r ate of the hypothetical opponent.We calculated the thickness of the EoC network edges as the percentage of times that a fungus had a larger EoC value than its opponent across all r eplicate plates.Finall y, we calculated the thic kness of the IoA network plots as the av er a ge IoA score of the plugs in each pairing.

Sta tistical anal ysis and modeling
We conducted analysis of variance tests (ANOVA) and Tuk e y Honest Significant Differences (Tuk e yHSD) tests to find the differences in gr owth measur es between (1) pH tr eatments for single contr ols, (2) pH treatments for SvS controls, (3) pH treatments for competition plates, and (4) control and competition plates under the same pH ( Supplementary Table S3 -7 ).
To investigate whether phylogenetic distance or growth rate better explained a strain's response to competition, we built linear mixed effects models using the lme4 pac ka ge (Bates et al. 2015 ).Each model predicted the EoC on growth rate using pH as a fixed effect along with either phylogenetic distance or gr owth r ate distance, and with fungal identity and plate identity as crossed random effects.For the growth rate distance model, because each pair of fungi had a unique growth rate distance at each pH value, the pH:growth rate distance interaction term was excluded; for phylogenetic distance, the distance value did not vary with pH, so the interaction term was retained: [ lmer(log(EoC) ∼ growth rate distance Due to the limited number of species in our phylogeny, we additionally ran the linear mixed effects models without C. geophilum and then again without C. geophilum and P. involutus to investigate their robustness.We compared these models using a chi-square test computed by the stargazer function from the Stargazer pac ka ge (Hlav ac 2018 ).We then used the pac ka ge lmerTest (Kuznetsov a et al. 2017 ) to generate P -v alues for eac h of the models and the r.squared.GLMM function, from the package MuMIn (Barto ń 2020 ), to generate pseudo-R 2 values for the model equations.To summarize and visualize these data, we created summary tables for our models with the package Stargazer , and we used the package ggeffects to plot our models (Lüdecke 2018 ).Additionally, to make sure our linear mixed effects models did not have correlated main effects, we modeled the correlation betw een gro wth rate distance and phylogenetic distance [ lmer(phylogenetic distance ∼ growth rate distance + (1 | pH) ].We performed all analyses in R (version 4.1.1)(R Core Team 2021) using the RStudio interface (version 1.4.1717)(RStudio Team 2021).

pH significantly affects EMF growth
Regular (e.g.uncontaminated, positive) growth ( Supplementary F ig. S3 ) was observed on 368 out of the total 400 plates (92%).Fungal growth rates were significantly affected by pH in both single (ANOVA: F 1,89 = 19.35,P = 3.01e −05 ) and SvS (ANOVA: F 1,94 = 11.92,P = 8.34e −04 ) controls.In the single controls, the gr owth r ates for A. muscaria , C. geophilum , and H. cylindrosporum were all significantly lo w er under more acidic conditions (Fig. 2 ), while the other two fungi ( L. bicolor and P. involutus ) wer e m uc h less affected by pH.The SvS controls showed similar results, except that A. muscaria additionall y gr e w poorl y in the neutr al pH condition (Fig. 2 ).Although most tested fungi gr e w poorl y on the acidic media r elativ e to the neutral media, H. cylindrosporum was the most dr amaticall y affected.Its av er a ge gr owth r ate on the acid media was 46.4% of its rate when grown in single culture and 42.8% when grown in intraspecific competition.By contrast, the growth of P. involutus was unaffected by media pH in both contr ol tr eatments (Fig. 2 ).The growth of the five fungi on the SvS plates was used to model single species growth r ates, fr om whic h we calculated gr owth r ate distances for later modeling.

The effects of competition vary with fungal identity and pH
When grown in interspecific competition, fungal growth rates varied in ways that depended on the identity of the fungus, its competitor, and the pH of the plate (linear model: F 39,322 = 34.16,P = 2.20e −16 , adjusted R 2 = 0.782, Supplementary Table S8 ).Most fungi gr e w mor e slowl y when competing with other str ains, with L. bicolor being the most consistently inhibited by competition (Fig. 3 D).This pattern was far from universal, ho w ever: the gro wth rate of A. muscaria was mostly unaffected by the presence of competitors (Fig. 3 A), while C. geophilum gener all y gr e w faster in the presence of other species r elativ e to its controls (Fig. 3 B).Intriguingl y, H. cylindrosporum r esponded differ entl y to competition depending on the pH of the media: at pH 7, it was negativ el y affected b y competition regar dless of competitor identity, while at pH 5.6, it gr e w slightl y faster when competing with A. muscaria than when growing on its own (Fig. 3 C).

Competiti v e networks reveal complex interactions and lack of hierarchical structure
The network plots highlight the complexity of the competitive interactions between the five fungi.Based on the results of the control plates, L. bicolor and P. involutus seem to be the highest performers at pH 5.6, and P. involutus and H. cylindrosporum at pH 7 (Fig. 4 A and B).These r esults ar e inconsistent with fungal performance under competition.At pH 5.6, L. bicolor was most negativ el y affected by competition, while C. geophilum was dominant in terms of positive performance under competitiv e str ess (Fig. 4 C).At the neutral pH, H. c ylindrosporum becomes the w eakest performer, and A. muscaria joins C. geophilum at the upper end of the performance spectrum (Fig. 4 C).
In certain pairings, the outcome of competition switched based on what pH environment the competition took place in.Under the more acidic condition, both P. involutus and H. cylindrosporum outcompeted L. bicolor .This outcome r e v ersed for both pairings in the neutral condition (Fig. 4 C), with L. bicolor instead outperforming the other two fungi.Furthermore, H. cylindrosporum outgr e w A. muscaria at pH 5.6, while at pH 7, A. muscaria was the dominant fungus (Fig. 4 C).These results are inconsistent with the predictions from the control plates, which suggest that L. bicolor , under pH 5.6, should have a faster growth rate than H. cylindrosporum, and H. cylindrosporum , under pH 7, should have a faster growth rate than A. muscaria (Fig. 4 A and B).
The IoA metric sho w ed little variance between the pH treatments (Fig. 4 D).Hebeloma cylindrosporum remained dominant, while C. geophilum remained the weakest competitor.The performance of the other three fungi was all similar between pH conditions.

Both growth rate and phylogenetic models have similar predicti v e efficacy
We found that both phylogenetic distance and gr owth r ate distance predicted the effects of competition with similar efficacy (ANOVA: AIC gr owth r ate dist = −139.516,AIC phylogenetic dist = −131.227,Chi-sq = 0, P = 1, Table 1 ).In both models, pH had a significant effect on the correlation between predictive and response variables (gr owth r ate distance model: estimated v alue = −0.103,P = 2.05e −06 ; phylogenetic distance model: estimated value = −0.111,P = 0.033, Fig. 5 ).Additionally, w e found that gro wth rate distance and phylogenetic distance were not significantl y corr elated (linear model: P = 0.791) ( Supplementary Table S9 ), allowing us to conclude that these variables can be treated independently.When Asterisks next to lines r epr esent significant differences in growth rate in that comparison.[All significant differences were determined by Tuk e y HSD tests; " * * * " P < 0.001, " * * " P < 0.01, " * " P < 0.05].
investigating models with reduced taxa, we similarly found that ther e wer e no differ ences in pr edictiv e po w er betw een the tw o models ( Supplementary Table S10 -11 ).

Colony size growth metric produces similar results
Fungal colony sizes were significantly affected by pH in both single (ANOVA: F 1,89 = 38.45,P = 1.78e −08 ) and SvS (ANOVA: F 1,94 = 30.90,P = 2.53e −07 ) controls.When grown in competition, fungal colony sizes varied in ways that depended on the identity of the fungus, its competitor, and the pH of the plate (linear model: F 39,322 = 35.04,P = 2.20e −16 , adjusted R 2 = 0.786, Colony Size Supplementary Table S7 ).The network plots made using the colony size metric still show a lack of hier arc hical structure, though H. cylindrosporum replaces C. geophilum as the dominant fungus when considering EoC wins (see Colony Size Supplement).Results differ most when considering the linear mixed effects models.For colony size, we found that phylogenetic distance predicted the effects of competition better than colony size distance (ANOVA: AIC colony size dist = 300.651,AIC phylogenetic dist = 290.172,Chi-sq = 12.479, P = 4.12e −04 ).When investigating models with reduced taxa, we found that with four taxa (without C. geophilum ), ther e wer e no differ ences in pr edictiv e po w er betw een the tw o models, and with three taxa (without C. geophilum and P. involutus ), phylogenetic distance again predicted the effects of competition better than colony size distance (see Colony Size Supplement).

Discussion
Outcomes of EM fungal competition can alter forest ecology at both individual (e.g.tree health) and landscape-wide (e.g.biogeochemical cycling) scales.Howe v er, fe w studies of EM competition (compared to the high diversity of fungal taxa co-occurring on tree root systems) exist from which to derive predictions of competitive outcomes .T herefore , we set out to compare two pr edictiv e methods of understanding how EM fungi perform in competition.Our results suggest that incorporating either phylogenetic relatedness or a growth measure approach may provide useful insight in predicting the effects of competition on fungal performance.Ov er all, while pH did str ongl y influence competitive performance, interactions between the five isolates were complex, highlighting the need for a better understanding of fungal competitive mechanisms and their reaction to environmental changes.In this way, our findings parallel other studies of fungal competition that show context-dependent outcomes (Kennedy 2010, Mujic et al. 2016 ) that can vary with fungal growth habits and environmental tolerances (Marín et al. 1998, Chen and Chou 2017, Maynard et al. 2019 ).
While, in our experiment, phylogenetic distance predicted competitiv e inter actions equall y well to the gr owth r ate metric, there is much debate in the literature surrounding this topic.Some suggest that trait-based approaches should be most effectiv e in pr edicting these kinds of outcomes (Mahon et al. 2021 ) because tr aits r ele v ant to competition may v ary acr oss linea ges in ways that are poorly captured by the underlying phylogeny (Cadotte et al. 2017 ).Ho w e v er, others ar gue for combining a ppr oac hes (Mayfield andLevine 2010 , Cadotte et al. 2017 ) because it is difficult to experimentally address the entire scope of ecologicall y r ele v ant tr aits and their individual r oles in sha ping competitive outcomes (Cadotte 2013 ).To address this conflict, Mayfield and Levine ( 2010 ) posit that phylogenetic relatedness would positiv el y corr elate to the effects of competition only if phylogenetic structure captured differences in niche preferences such that closely related species more harshly exclude one another.Alternativ el y, if phylogenetic relatedness was correlated to dif-Figure 3. Fungal responses to growth medium pH and interspecific competition.The EoC (effect of competition) metric, calculated as the log ratio of gr owth r ate on competition plates vs. contr ol plates [(gr owth in competition/gr owth on contr ol plate)], differs depending on the pH in whic h the competition occurs [pH 5.6 (r ed), pH 7 (blue)].Fungi ar e order ed fr om most competitiv e (top) to least competitiv e (bottom).Opponent fungi are denoted at the bottom: A. muscaria (left), C. geophilum (middle left), H. cylindrosporum (middle), L. bicolor (middle right), and P. involutus (right).Asterisks above the box plots r epr esent significant differences in growth rate between the control and competition plates.Asterisks above the bars represent significant differences in growth rate between pH treatments.[All significant differences were determined by Tuk e y HSD tests; " * * * " P < 0.001, " * * " P < 0.01, " * " P < 0.05].
fer ences in competitiv e tr aits, one would expect closel y r elated species to have similar competitive adv anta ges and ther efor e outcompete distantl y r elated species (Mayfield and Le vine 2010 ).Our results support the second hypothesis, as phylogenetic distance negativ el y corr elated with the EoC metric, meaning that mor e distantl y r elated species experienced str onger negativ e competitiv e effects.Finally, though our work suggests that both trait and phylogenetic models have similar competence, futur e r esearc h may find that increasing the number of r ele v ant tr aits studied will pr o-vide a substantially more po w erful predictor for competitive effects.
The differences in efficacy between our two growth measures, gr owth r ate and colon y size, in pr edicting competitiv e outcomes further illustrate the complex nature of interspecific interactions.We found that the effects of competition on growth rate were predicted equally w ell b y initial growth rate and phylogenetic distance, but phylogeny was the better predictor of final colony size (Colony Size Supplementary Table S1 ).In an ecological context,  these growth measures can represent a proxy for two types of competition: initial growth rate, representing exploitation competition, and final colony size, representing interference competition (Bod d y and Hisco x 2016 , Smith et al. 2018 ).While an organism's ability to colonize its environment, determined in part b y gro wth rate , ma y be less dependent on phylogenetic distance, we found that the longer-term outcome of the physical interaction between competing fungi (colony size) was more dependent on phylogen y, perha ps because phylogen y pr edicts unmeasur ed traits like chemical defenses or the close proximity of the fungi on Petri dishes intensifying interference competition.This result corr obor ates pr e vious work that posited the importance of phylo-genetic relatedness increases at smaller ecological scales, specifically with individual biotic interactions (Carboni et al. 2013 ).Both e volutionary r elatedness and tr ait differ ences may influence competition, but our results suggest their effects depend on the specific competitive outcome measured.
While most competitive outcomes between species pairs were negative, C. geophilum experienced significant facilitation in half of its interspecific pairings.As C. geophilum is most distantl y r elated to the other four fungi, this finding supports the idea that phylogenetically distant species are more likely to experience facilitative biotic interactions (Valiente-Banuet and Verdú 2007 ).This finding contradicts the overall trend in our data of intensifying competi-tion with declining relatedness, suggesting that the sign of species interactions may reverse at sufficient phylogenetic distance .F ew studies have shown evidence for facilitation in EM fungi (Mamoun and Oliver 1993, Koide et al. 2005, Pickles et al. 2010, Gorzelak et al. 2012 ), perhaps because mycorrhizal interactions are predominantly studied in the context of competition (Valiente-Banuet and Verdú 2007 ).Further, the degree of taxonomic variation in studies of species interactions can generally limit the type of interaction that may be detected.Specifically, if the strength of competition scales with phylogenetic relatedness, then studies may need to include species that range in relatedness (including across higher taxonomic ranks) to be able to detect the full spectrum of negative to positive interactions (Philippot et al. 2010 ).Though our study used five EM fungi with varying degrees of taxonomic relatedness for this reason, we saw few differences between the power of our models to predict competitive interactions when considering taxa of differing r elatedness.Futur e studies constructed in a phylogeneticall y div erse and nested way can help us to identify the phylogenetic scales at whic h competitiv e vs. facilitativ e inter actions dominate.Importantl y, our r esults r eflect inter actions between only one strain of each species tested.Because intraspecific variation can exceed interspecific variation in some cases (Johnson et al. 2012, Hazard et al. 2017 ), a definitive test of phylogeny-function r elationships would r equir e r eplication at the str ain le v el.Future work should prioritize c har acterizing fungal tr ait v ariability within species while also examining differences between species.
Although the interactions we observed among these fungal str ains pr ovide intriguing insight into belowground competition, extr a polating to field conditions r equir es an understanding of how competition will change with environmental context.Our data show the complexity of this inter play.We observ ed significant changes in growth rate in both control treatments (Fig. 2 ) and EoC values (Fig. 3 ) based on pH, with a higher contr ol gr owth r ate in the neutral condition.The maintenance of intracellular pH (Bignell 2012 ) and modification of environmental pH (Krishna Sundari and Adholeya 2003 ) are important functions in fungi and involve an arsenal of enzymes , transporters , and signaling molecules .Fungal gr owth r ate can be negativ el y affected b y lo w pH, as shifts to acidic conditions, e v en as small as 0.3 pH units, r epr ess phospholipid biosynthesis genes and, ther efor e, membr ane biosynthesis (Kane 2016 ).We also observed certain competitive outcomes flip betw een the tw o pH tr eatments, whic h displays the wide range of fungal responses to differing pH.Additionally, the heterogeneity of soil pH likel y constr ains wher e particular fungi can grow and so determines the identities of potential competitors.Our data further show that the identity of these competitors is a k e y determinant of the competitive success of fung i (Fig. 3 ).Fung i have a plethor a of mec hanisms by whic h they can dir ectl y and indir ectl y compete for territory (Bod d y and Hiscox 2016 ).These include str ategies of a ggr ession (i.e.secr eted or ganic compounds and anta gonistic gr owth) and defense (i.e. the formation of physical barriers and the upregulation of nutrient acquisition and stress regulatory mechanisms) (Bod d y and Hisco x 2016 ).While our study shows evidence of some of these mechanisms ( Supplementary Fig .S1 ), the specific competitiv e r epertoir es of these fiv e EMF ar e r elativ el y unclear, and further study of EMF competition may r e v eal m utualism-specific tools.Additionall y, in the Petri dish context of this experiment, a ppar ent anta gonistic ov er gr owth could hav e instead r epr esented mor e benign coexistence, as long as both fungi still had access to non-limiting media resources.For example, all tested basidiomycetes r outinel y ov er gr e w C. geophilum without clearly impairing its growth ( Supplementary Fig. S1 ).The mycelial morphology of ectomycorrhizal fungi is known to vary with substr ate c hemistry (Dic kie et al. 1998 ), a pr ocess whic h ma y ha ve impacted the growth of coinoculated fungi in this experiment.It is difficult to extr a polate fr om these conditions (i.e.no mycorrhizal association, no beneficial or antagonistic soil microbes) to a field soil, but these fungal interactions are certainly complex and envir onmentall y v ariable.Finally, our work suggests that different measures of competition can pr oduce differ ent competitiv e hier arc hies (Fig. 4 ).Both our EoC and IoA metrics displayed lar gel y differ ent network plots ( C. geophilum dominant under EoC and H. cylindrosporum dominant under IoA), corr obor ating pr e vious work that posits fungi differentially prioritize competitive and colonizing abilities (Smith et al. 2018 ).The fact that different fungi are competitively superior, depending on the competition metric used, could help explain the observed diversity of coexisting EMF communities on small spatial scales (Anderson et al. 2014 ).The complexity of competition and coexistence r equir es further study, but priority effects, competition-colonization tradeoffs, and niche partitioning are all likely to contribute to ECM community assembly across diverse envir onments (K ennedy 2010 ).
To conclude, we tested the usefulness of modeling interspecific interactions between EMF using differences in growth measures and phylogenetic distance by culturing five fungi in single, intraspecific, and pairwise interspecific treatments.Additionally, we replicated our experiment at two pH le v els to explor e the effect of pH on ECM fungal competitive interactions.Our results show that both phylogeny and growth measures can be useful for predicting the results of EMF competition and that pH str ongl y influences these inter actions.Futur e r esearc h should focus on identifying, whic h fungal tr aits most str ongl y affect biotic interactions and how often variation in these traits is ca ptur ed adequately by phylogenetic analyses .Furthermore , due to our study design's div er gence fr om the natur al EMF ecological context, fu-tur e r esearc h within the scope of ECM competitiv e inter actions should examine how in-vitro observations hold up under the context of the mycorrhizal symbiosis, with a focus on the effects of envir onmental v ariables and the presence of facilitativ e inter actions.Since its discovery, the mycorrhizal symbiosis has proven to be an integral component of for est comm unity structur e .T he complexity of the biotic and abiotic interactions in both the development and maintenance of this symbiosis r equir es m uc h further study, but doing so will allow us to a ppl y a more informed conceptual fr ame work to the structuring of terr estrial ecological communities.

Figure 1 .
Figure 1.Ov ervie w of the experimental design.(A) Examples of the three layouts of experimental plates.Concentric lines denote the colony circumference used to measure growth at different time points.(B) Phylogeny of the five EMF with one outgroup ( Batrachochytrium dendrobatidis ).Numbers on nodes denote posterior probabilities.(C) Examples of variation in growth characteristics of each fungal species [A = A. muscaria , C = C. geophilum , H = H.cylindrosporum , L = L. bicolor , P = P. involutus ].

Figure 2 .
Figure 2. Fungal responses to growth medium pH and intraspecific competition.Growth rate of single and SvS controls for A. muscaria , C. geophilum , H. cylindrosporum , L. bicolor , and P. involutus at two pH le v els.Lines connecting points r epr esent useful comparisons.Err or bars r epr esent standard err or.Asterisks next to lines r epr esent significant differences in growth rate in that comparison.[All significant differences were determined by Tuk e y HSD tests; " * * * " P < 0.001, " * * " P < 0.01, " * " P < 0.05].

Figure 4 .
Figure 4. Fungal competitive networks vary by metric and shift with growth medium pH.Network plots depict av er a ge inter actions between fungi based on four measures: single control growth rate (A), SvS control growth rate (B), EoC wins (C), and the IoA score (D).The width of arrows for each measure was calculated, respectively, as the ratio of the mean growth rate over the mean growth rate of the theoretical opponent (for both controls), the number of competition plates on which a fungus had a higher EoC value than its opponent, and the average IoA score for each fungus in interspecific competition.The size of the nodes is proportional to the sum of all outgoing arrow widths [pink = L. bicolor , green = C. geophilum , beige = P. involutus , purple = A. muscaria , orange = H.cylindrosporum ; pH 5.6 (top) and pH 7 (bottom)].

Figure 5 .
Figure 5. Phylogenetic distance and growth rate distance model predictions on EoC.(A) Graphical results of the linear mixed effects model when using gr owth r ate distance to pr edict the effects of competition on gr owth r ate.(B) Gr a phical r esults of the linear mixed effects model when using phylogenetic distance to predict the effects of competition on growth rate.[pH 5.6 (red), pH 7 (blue); ribbons represent standard error].

Table 1 .
ANOVA table comparing the two linear mixed effects models.P < 0.05; * * P < 0.01; * * * P < 0.001.The first column depicts summary statistics for the model that uses growth rate distance to predict the EoC on growth rate .T he second column depicts summary statistics for the model that uses phylogenetic distance to predict the EoC on gr owth r ate.Numbers in par entheticals denote standard error.