Volume 110, Issue 1 p. 262-276
Free Access

Mycorrhizal symbiosis alleviates plant water deficit within and across generations via phenotypic plasticity

Javier Puy

Corresponding Author

Javier Puy

Department of Botany, Faculty of Sciences, University of South Bohemia, České Budějovice, Czech Republic

Zoology, School of Natural Sciences, Trinity College Dublin, Dublin 2, Ireland


Javier Puy

Email: [email protected]

Search for more papers by this author
Carlos P. Carmona

Carlos P. Carmona

Department of Botany, Institute of Ecology and Earth Sciences, University of Tartu, Tartu, Estonia

Search for more papers by this author
Inga Hiiesalu

Inga Hiiesalu

Department of Botany, Institute of Ecology and Earth Sciences, University of Tartu, Tartu, Estonia

Search for more papers by this author
Maarja Öpik

Maarja Öpik

Department of Botany, Institute of Ecology and Earth Sciences, University of Tartu, Tartu, Estonia

Search for more papers by this author
Francesco de Bello

Francesco de Bello

Department of Botany, Faculty of Sciences, University of South Bohemia, České Budějovice, Czech Republic

CIDE-CSIC, Valencia, Spain

Search for more papers by this author
Mari Moora

Mari Moora

Department of Botany, Institute of Ecology and Earth Sciences, University of Tartu, Tartu, Estonia

Search for more papers by this author
First published: 05 November 2021
Citations: 9

Handling Editor: Zeqing Ma


  1. Phenotypic plasticity is essential for organisms to adapt to local ecological conditions. It is expected that mutualistic interactions, such as arbuscular mycorrhizal (AM) symbiosis, mediate plant phenotypic plasticity, although it is not clear to what extent this plasticity may be heritable (i.e. transgenerational plasticity).
  2. We tested for plant plasticity within- and across-generations in response to AM symbiosis and varying water availability in a full factorial experiment over two generations, using a genetically uniform line of a perennial apomictic herb, Taraxacum brevicorniculatum. We examined changes in phenotype, performance and AM fungal colonization of the offspring throughout plant development.
  3. AM symbiosis and water availability triggered phenotypic changes during the life cycle of plants. Additionally, both factors triggered adaptive transgenerational effects especially detectable during the juvenile stage of the offspring. Water deficit and absence of AM fungi caused concordant plant phenotypic modifications towards a ‘stress-coping phenotype’, both within and across generations. Parental environment also affected AM fungal colonization of the offspring. Juvenile offspring of amply watered parents and adut offspring of mycorrhizal parents had increasing length of root colonized by AM fungi when they were under water stress.
  4. Synthesis. We show that, in addition to providing beneficial transgenerational effects in offspring traits proxies of fitness (such as increased biomass, survival or nutritional status), AM symbiosis can trigger transgenerational plasticity in anatomical and physiological traits related to resource-use acquisition, and further influence offspring AM fungal colonization. Our results show that AM symbiosis could improve plants' ability to cope with environmental stress, not only within, but also across generations.


Abiotic environment and prevailing biotic interactions select for the best-adapted individuals within and across species in terms of their functional traits (de Bello et al., 2012; Vellend, 2016). The probability of an individual to withstand this selection may depend on the nature and severity of the environmental condition, the heritable genetic variability, but also on its phenotypic plasticity (Bradshaw, 1965; Price et al., 2003). Phenotypic plasticity is the ability of a genotype to modify its performance and trait expression in response to the environment without altering the DNA sequence (Bradshaw, 1965; Price et al., 2003). These changes could operate within the life cycle of the organism subjected to those conditions (‘within-generation plasticity’). Additionally, they may be transmitted to the following generations via transgenerational effects (also called ‘across-generation or transgenerational plasticity’), meaning that the abiotic and biotic environments experienced by the parental generation influence the phenotype of the offspring (Herman & Sultan, 2011; Jablonka & Raz, 2009). Recent studies suggest that transgenerational effects could play a key role in the adaptive response of organisms to stressors, proven particularly essential during the juvenile stages (Dechaine et al., 2015; Lämke & Bäurle, 2017; Latzel et al., 2014; Puy, de Bello, et al., 2021). While the effect of abiotic conditions on transgenerational plasticity has been repeatedly demonstrated, little is known about the relative effect of transgenerational effects triggered by biotic conditions (Alonso et al., 2019; Puy, de Bello, et al., 2021), and even less about how they interact with abiotic factors.

Together with species' adaptations to environmental conditions in a site, biotic interactions are considered key drivers of plant community assembly (de Bello et al., 2012). Among these, mutualistic interactions such as mycorrhizal symbiosis are essential in determining, and potentially expanding, the realized niches of species (Gerz et al., 2018; van der Heijden et al., 2003; Peay, 2016). Arbuscular mycorrhizal (AM) symbiosis is a widespread mutualistic association between plant roots and fungi (from the subphylum Glomeromycotina; Spatafora et al., 2016). This association is considered mutually beneficial, since, in exchange for photosynthetic carbon, the AM fungi provide host plants with soil nutrients (mainly phosphates; Smith & Read, 2008), mitigate abiotic stress (e.g. making the host more tolerant to drought; Aroca et al., 2012; Augé et al., 2015; Doubková et al., 2013) and increase resistance to biotic stress, including pathogens (Pozo et al., 2015; Smith & Read, 2008). Besides the mutual effects of the AM interaction, the occurrence, abundance, activity and the final outcome of the interaction (from positive to negative) are known to be affected by multiple factors (Hoeksema et al., 2010; Johnson et al., 1997). These factors include intrinsic drivers, such as the genotype of both partners, or the plant developmental stage or sex of the host (Jones & Smith, 2004; Varga & Kytöviita, 2010), and external drivers such as the biotic environment (Šmilauer et al., 2020) or soil nutrient and water availability (Martínez-García et al., 2012; Pozo et al., 2015). For example, phosphorus or water deficiency in plants, at least when light is not limiting, generally stimulate AM colonization and influence the abundance of AM fungal structures (i.e. arbuscules, vesicles, etc.; Martínez-García et al., 2012; Pozo et al., 2015). However, it remains unclear whether the environmental conditions experienced in parental generations could affect the functioning of AM symbiosis of the offspring generation (De Long et al., 2019).

As illustrated before, the fitness benefits of plants in AM symbiosis are relatively well known (Lu & Koide, 1994; Smith & Read, 2008). However, it is unclear whether these benefits partly operate through phenotypic plasticity induced by the interaction (Vannier et al., 2015). AM symbiosis may lead to adaptive changes of plant morphological traits such as modifications in root architecture (Fusconi, 2014; Goh et al., 2013; Nuortila et al., 2004) or in functional traits associated with the plant resource-use strategy (also called plant economics spectrum; Reich, 2014). According to this, individuals shift towards a more resource-acquisitive strategy, characterized by faster above-ground growth and ‘cheaper’ and short-lived tissues (i.e. high specific leaf area, SLA; low leaf dry matter content, LDMC; high specific root length, SRL), when resources (such as nutrients, water or light) are abundant. By contrast, when resources are scarce, individuals shift towards a more resource-conservative strategy, displaying higher root biomass allocation and more structural and tougher tissues (i.e. low SLA, high LDMC, low SRL). Moreover, it also remains unclear whether AM symbiosis in a parental generation can trigger similar or different phenotypic changes in the offspring (i.e. transgenerational effects) and whether these changes are beneficial to the offspring (Koide, 2010; Varga et al., 2013). As such, it is crucial to assess the effect of biotic and abiotic drivers on within-generation and transgenerational plasticity combined, as biotic drivers can potentially modulate the effect of abiotic stress (González et al., 2017; Metz et al., 2015).

It should be noted that transgenerational effects can operate via two mutually non-exclusive mechanisms. One of the most important transmission mechanisms is via seed provisioning, that is, seed size, seed nutritional quality or hormonal balance (Dechaine et al., 2015; Germain et al., 2019; Herman & Sultan, 2011). Parental conditions and resource availability may determine the amount of resources provided by maternal plants in the embryos. Seed resources can be either passively determined by the available resources in the maternal environment (i.e. stressful parental conditions leading to reduced resources) or maintained (or even increased) if maternal plants actively compensate the stressful conditions (Violle et al., 2009). Additionally, and independently of seed properties, transgenerational effects can be mediated by environmentally induced heritable epigenetic modifications in the offspring (Herman & Sultan, 2011; Lämke & Bäurle, 2017). In this case, parental plants change the gene expression of their offspring plants in response to their environment, which allows the transmission of environmental information across generations (also referred as epigenetic (stress) memory; Chinnusamy & Zhu, 2009). In general, while transgenerational effects deriving from seed provisioning play a significant role during early stages of the development, they tend to fade away with time (Latzel et al., 2010).

Most of the existing studies showing beneficial transgenerational effects of having mycorrhizal parents (e.g. increasing offspring's biomass, survival, growth rate, nutrient concentration and seed production; Heppell et al., 1998; Koide, 2010; Varga, 2010; Varga et al., 2013) have not disentangled the effect of the two mechanisms mediating transgenerational effects separately (but see Varga et al. (2013) where seed mass was used as a covariate, or Varga and Soulsbury (2017, 2019) where DNA methylation was inspected as potential heritable epigenetic mechanism). Moreover, while the beneficial transgenerational effects have been mainly evident during the early stages of development of the offspring, existing research has not yet tested whether transgenerational effects of AM symbiosis persist further into the adult stage of the offspring, neither whether they modulate plant phenotypic and functional traits or influence AM fungal colonization.

We conducted a two-generation experiment to test for within- and across-generation plant plasticity in response to AM symbiosis using the perennial apomictic herb Taraxacum brevicorniculatum. Furthermore, in order to test whether both plasticity types interact with abiotic stress conditions, we included a watering regime treatment simulating hydric stress. By measuring (a) plant phenotypic traits, (b) performance (i.e. nutrition and biomass) and (c) AM fungal colonization of juvenile and adult offspring, we evaluated the persistence of the transgenerational effects throughout offspring development and tested whether within- and across-generation plant plasticity were adaptive, for example, improved the ability of the offspring to cope with water limitation.


2.1 Study material

Taraxacum brevicorniculatum Korol. (Asteraceae) is an apomictic polycarpic perennial plant (Kirschner et al., 2013). Like most species of the genus Taraxacum, it has a wide ecological niche, accepting all types of soils, pH and moisture levels (Luo & Cardina, 2012), and forms symbiosis with AM fungi (J. Puy, pers. obs. based on a preliminary study). In this study, we used genetically identical seeds collected from a population of plants grown under the same glasshouse conditions for several generations (collected and genetically identified by Kirschner et al., 2013). This strategy ensured little or no genetic and epigenetic variation in the plant material. Since T. brevicorniculatum is an obligate apomictic species, all seeds produced by a plant are effectively clones, thus enabling the study of plasticity within and across generations (Puy, de Bello, et al., 2021; Puy et al., 2018). In other words, all plants in the experiments were genetically identical, and after experiencing different conditions during the parental generation, the offspring only differed in non-genetic (i.e. non-genetic or epigenetic effects) inherited information.

2.2 Experimental setup

2.2.1 Parental generation

To induce the potential transgenerational effects related to mycorrhizal symbiosis and water availability, we first conducted an experiment in which the parental generation was grown under different conditions in a glasshouse for 3 months (April–July 2017). We grew 364 genetically identical individuals of T. brevicorniculatum in individual pots (7 × 7 × 18 cm), half inoculated with AM fungi (AM) and the other half without (NM). In addition, half of these individuals were grown with sufficient water and half under a water deficit scenario, see below for details.

The substrate consisted of 2:1 mixture of sterilized sand and natural soil to provide a good soil structure for the plant compensating for the high clay content of the soil. Natural soil was collected from 15 cm of the topsoil from five 2 × 2 m plots across a mesic meadow (a crop field abandoned in 2002) 30 km southeast of Tabor, 660 m a.s.l. (Vysočina region, Czech Republic, 49.331N, 15.003E) where Taraxacum sect. Ruderalia was present. The natural soil was first sieved using a 4-mm mesh sieve to remove any gravel, macrofauna or rhizomes from the soil, and then homogenized and pooled together. For the AM treatment, the natural soil containing indigenous microbial community was used; whereas for the NM treatment, the same soil was sterilized via γ irradiation (>25 kGy dose; McNamara et al., 2003). In an effort to isolate the influence of AM fungi and to compensate for the loss of other soil microbes due to sterilization, a microbial wash was also added (Koide & Li, 1989; Liang et al., 2015). We obtained the microbial wash by blending 6 kg of non-sterilized soil in 12 L of water and filtering the solution through 20-µm pore-size filter paper (Whatman® quantitative filter paper, Grade 41) broadly following Liang et al. (2015). Despite our intention to equalize the soil microbial communities other than AM fungi between AM and NM soil, the microbial wash may have not fully regenerated (quantitatively or qualitatively) the micro-organism populations in AM soil because of the founder effect of the newly added pro- and eukaryotic microbes (i.e. organisms that have passed through the 20-µm pore-size filter).

The AM and NM treatments were factorially combined with two levels of water availability. Half of the individuals were subjected to cycles of water deficit (simulating drought stress; W−), while the other half were watered regularly from the bottom ensuring the pot surface was always wet (control; W+). The water deficit treatment included periodic exclusion of watering until 50% of the individuals had wilted leaves, followed by 1-week recovery in control conditions. By the end of the experiment, the water deficit treatment comprised two water deficit pulses (the first started 12th May and the second 15th June) that lasted 3 weeks plus 1-week recovery each.

Prior to the establishment of the experiment, seeds were surface sterilized by immersion in 0.5% sodium hypochlorite solution (commercial bleach) for 20 min to avoid inoculation via seeds, and then germinated in Petri dishes. After 10 days of germination, the seedlings were transplanted individually into the pots specified above, with 91 replicates per treatment. After 3 months, we harvested all the plants except 15 randomly selected individuals from each of the four treatment combinations (AM W+, AM W−, NM W+, NM W−). These plants were kept for four more months and relocated to more benign conditions to promote seed production, since non-mycorrhizal plants in particular had not flowered yet at the time of the first harvest. Then, seeds of each plant were collected, and after measuring the average seed mass per plant, stored in cold conditions (2–4°C). The new conditions for promoting seed production did not comprise further water-deficit treatments to the plants (i.e. all were kept in the water control condition previously described) and included an addition of fertilizer (Kristalon; NPK 15–5–30 + 3Mg + 5S) at the concentration of 300 ppm once per month, and a 12-hr (20°C)/12 hr (10°C) light/darkness-and-temperature regime.

2.2.2 Offspring experiment

A similar glasshouse experiment to the one described above was repeated the following year (April–August 2018) with the seeds produced by the parental generation in the first experiment. The aim of the offspring experiment was to test for adaptive transgenerational effects of AM symbiosis and water availability on the offspring at their juvenile and adult stages. We tested this with a full factorial design where the offspring plants from each of the four parental treatments were exposed again to the four possible conditions (AM W+, AM W−, NM W+ and NM W−). Thus, the offspring experimental design resulted in 16 combinations: two parental mycorrhizal inoculations (Par. M: AM/NM) × two parental water availability levels (Par. W: W+/W−) × two offspring mycorrhizal fungal inoculations (Off. M: AM/NM) × two offspring water availability levels (Off. W: W+/W−). Since the seed mass of AM parents was on average lower than that of NM parents (see below Results of parental generation, Table S1) and seed provisioning could be a potential mechanism of transgenerational effects (Herman & Sultan, 2011), we controlled for it by classifying seeds from all parental treatments into five size categories. Then, we took the same number of seeds from each size-group in each parental treatment, resulting in a similar distribution of seed sizes between parental treatments. Thus, the offspring experimental design finally resulted in the 16 combinations × 5 seed size categories × 8 seedlings = 640 pots (Figure S1).

Plants were harvested at two different developmental stages. Half of the offspring plants were harvested 1.5 months after planting, at their juvenile stage; and the rest of the replicates were harvested 5 months after planting, at their adult stage before they flowered. Pots, substrate and watering regime were the same as in the parental experiment to ensure the most similar conditions. However, the first water deficit pulse in the offspring generation lasted 4 weeks instead of 3 weeks (the first one started on the 25th April and the second one, the 1st June) to ensure comparable effects on plants physiology (i.e. percentage of plants with wilted leaves). To facilitate the application of the treatments, four replicates of a parental treatment were placed in parallel, one in each offspring treatment (Figure S1).

2.3 Measured traits

We measured a set of important plant traits in both generations. For 41 plants per treatment of the parental generation (N = 164), we measured survival, total dry biomass (aerial plus root biomass), seed output (i.e. number of seeds) and seed mass at the time of harvest. For five randomly chosen plants per treatment (N = 20), we measured C, N and P concentration of the seeds, considered to be reliable indicators of seed reserves (Toorop et al., 2012). Total C and N concentrations were determined by dry combustion using an elemental analyser (CHNS Elemental Analyzer vario MICRO cube; Elementar Analysensysteme GmbH, Germany). Total P was determined by flow injection analysis.

For each plant in the offspring generation (N = 640), at the time of the respective harvest (i.e. juvenile and adult offspring harvest), we measured total dry biomass (aerial plus root biomass), and several above- and below-ground vegetative plant traits (Pérez-Harguindeguy et al., 2013). For each plant, two leaves were scanned for leaf area and weighed for fresh mass and dry mass after drying at 60°C (48 hr) to estimate specific leaf area (SLA; leaf area per dry mass, mm2/mg) and leaf dry matter content (LDMC; leaf dry mass per leaf fresh mass, mg/mg). Roots were carefully extracted, washed and a subsample of roots (6 cm2) was scanned at 600 dpi with an Epson Perfection 4990 scanner. Total root length, average root diameter (mm) and distribution of root length in different diameter classes were determined using the image analysis software WinRHIZO Pro, 2008 (Regent Instruments Inc.). After scanning, the root subsample and the rest of the root system were dried for 48 hr at 60°C and weighed. We used these measurements to estimate specific root length (SRL; root length per dry mass, m/g) and fine roots percentage (root length with a diameter <0.5 mm per total root length). Furthermore, we estimated root biomass allocation (i.e. root mass factor; RMF; root biomass per total biomass, g/g) after drying the remaining radicular part at 60°C (48 hr).

Additionally, we analysed the concentration of C, N and P in the leaves of two randomly chosen plants from the juvenile stage and eight plants from the adult stage per treatment (N = 32 and 128, respectively), following the methods described above. The root subsamples were stained with Chlorazol Black according to the protocol by Šmilauer et al. (2020). We quantified the AM fungal colonization of all individuals (i.e. also in NM plants to verify that no colonization was established) by measuring the percentage of root length colonized (%RLC) by AM fungal structures (arbuscules, vesicles and hyphae). The magnified gridline intersection method (McGonigle et al., 1990) was used with 400× magnification using a light microscope with a graticule inserted into the eyepiece. All the specific structures of AM fungi (arbuscules, vesicles and hyphae) that intersected the vertical line (i.e. root in horizontal position) were counted for at least 100 intersections per root sample. We further calculated the arbuscule:vesicle ratio (relative abundance of arbuscules per vesicles), suggested as an indicator of the fungal activity status and the relative cost or benefit of the fungus to the host plant (Braunberger et al., 1991; Titus & Lepš, 2000).

2.4 Statistical analysis

All analyses were carried out using R v3.2.3 (R Core Team, 2016) with α = 0.05 as significance level. In the parental generation, the effects of the mycorrhizal inoculation (two levels), the water availability (two levels) and their interaction were analysed by using linear effects models. Plant morphological traits were always log-transformed.

In the offspring generation, individuals were grouped into 16 different treatments (as a result of the combination of four factors with two levels each) depending on parental and offspring conditions. Two of the factors corresponded to parental conditions: mycorrhizal fungal inoculation treatment (Par. M: AM/NM) and water availability treatment (Par. W: W+/W−). The other two factors corresponded to offspring conditions: mycorrhizal fungal inoculation treatment (Off. M: AM/NM) and water availability treatment (Off. W: W+/W−). We analysed the effects of parental and offspring conditions on phenotypic morphological traits and performance (i.e. biomass and nutrient concentration) of the juvenile and adult offspring using linear effects models where the four experimental factors (two parental and two offspring conditions) and all their interactions were used as fixed effects. The effect of parental and offspring treatments on AM fungal colonization and arbuscule:vesicle ratio of the offspring were analysed using identical models but excluding the offspring mycorrhizal inoculation factor (Off. M) from the model due to the lack of AM fungal colonization in the NM offspring plants.

In all the models, we included seed mass and seed stoichiometry as covariates (i.e. as fixed effects) to control for differences between parental treatments in seed provisioning or quality provided by the maternal plants (Herman & Sultan, 2011; Toorop et al., 2012). The seed stoichiometry values were computed assigning the scores of the first axis of a principal components analysis (PCA) that combined the C, N and P chemical composition of the seeds and accounted for 60% of the variation. Seed stoichiometry was not correlated with seed mass (Pearson correlation = −0.18, p = 0.44, df = 18). Any effect of the parental conditions, either direct (Par. M or Par. W) or in interaction with the offspring conditions (Off. × Par.) that remained significant after removing the linear part of the maternal investment (seed mass and stoichiometry), was considered a transgenerational effect.

Additionally, to reduce the multi-trait space, we also analysed the effects of parental and offspring conditions on the integrated phenotype of the offspring using a multivariate approach (i.e. combining all morphological traits together). To do so, we first computed a multivariate dissimilarity matrix between each pair of the 640 individual plants using Gower dissimilarity based on log-transformed and scaled trait values (i.e. centred by the mean and divided by 1 SD unit of each trait) across all morphological traits together (total biomass, RMF, SLA, LDMC, SRL and % of fine roots). We then analysed the phenotypic differences using a PERMANOVA (McArdle & Anderson, 2001), where offspring development stage, the four experimental factors (two parental, and two offspring conditions) and their interactions, and the covariates (seed mass and seed stoichiometry) were used as predictors, estimating the significance value by 999 permutations.

Finally, we checked for correlations between plant traits of the offspring and their AM fungal colonization value to check which plant traits associate with AM fungal colonization and to examine whether these changes could partially explain the benefits of AM symbiosis to the host plant, meaning that they are adaptive. Pearson correlations are generally used, except for the correlations with AM fungal colonization traits considering all plants together where Spearman correlations are used.


3.1 Parental generation

In the parental generation, the water deficit treatment decreased T. brevicorniculatum total plant biomass and survival, but only on NM plants, with no effect on AM plants (Figure S2a,b; Table S1). AM fungal inoculation increased plant growth, survival and reproductive investment (i.e. number of seeds per unit plant biomass), with no effect on the total number of seeds produced per plant (Figure S2; Table S1). Additionally, seeds of AM plants were lighter than the ones of NM plants, and with higher N, P and C concentrations, although only the latter one was significant (Figure S2; Table S1). Nevertheless, the seed macronutrients stoichiometry, that is, C:N:P ratios, did not differ between AM and NM plants (Figure S2; Table S1).

3.2 Offspring phenotype and performance

Overall, offspring clearly differed in their multi-trait phenotypic variability at the two developmental stages (juvenile and adult). The offspring stage explained 40% of the trait dissimilarity between individual plants (PERMANOVA R2 = 0.40, p = 0.001), principally in size traits (Figure 1a). However, the offspring developmental stage also modulated how offspring treatments determined the position of the individuals in the plant economics spectrum. During the juvenile stage, both offspring treatments (offspring mycorrhizal fungal inoculation treatment, Off. M; and offspring water availability treatment, Off. W) significantly affected the multi-trait phenotypic variability (PERMANOVA R2 = 0.11, p < 0.01; Figure 1a and Table S2; Figure 2 and Table 1). Plant phenotypes were ordered following the offspring treatments: from smaller plants with higher biomass allocation into the roots (i.e. higher RMF), thicker leaves and roots (higher LDMC, lower SLA and SRL) and lower P concentration and C:N ratio of leaves under water deficit and absence of AM symbiosis to the opposite (i.e. bigger plants, higher RMF, SLA and SRL) under the more advantageous conditions: water control conditions and AM fungal inoculation (Figure 1a; Figure 2 and Table 1). During the adult stage, however, only the multi-trait phenotype of offspring under water deficit and absence of AM symbiosis (W-NM) differed from the rest. These plants had thinner and more absorptive roots (higher SRL and % Fine roots), and less root biomass allocation (RMF) and lower LDMC (Figures 1a and 3; Table S3).

Details are in the caption following the image
Multivariate description of offspring phenotypes (juveniles and adults) in a multidimensional scaling (MDS) map showing pairwise distances based on their morphological/functional traits in combination. Juvenile offspring plotted as circles and adult offspring as squares. (a) Influence of offspring treatments. Colour-coded points indicate offspring treatment, and shaded ellipsoids represent SD of each significant offspring treatment (PERMANOVA; Table S2): red—water-deficient offspring, blue—offspring under water control conditions; intense colour and solid line—mycorrhizal offspring, light colour and dashed line—non-mycorrhizal offspring. Black arrows represent the traits that contribute most to differences between individual phenotypes. (b) Influence of parental treatments. Colour-coded points indicate parental treatments: red—offspring of water-deficient parents, blue—offspring of parents that experienced water control conditions; intense colour—offspring of mycorrhizal parents, light colour—offspring of non-mycorrhizal parents. Ellipsoids represent SD of the significant parental treatment (only parental mycorrhizal treatment; PERMANOVA; Table S2): solid line—offspring of mycorrhizal parents, dashed line—offspring of non-mycorrhizal parents
Details are in the caption following the image
Effect of the offspring and parental treatments on plant phenotype characteristics of the juvenile offspring. (a) total plant biomass, (b) leaf P concentration, (c) leaf C:N ratio, (d) root mass factor, (e) leaf dry matter content, (f) specific root length and (g) fine roots percentage. The significant factors of each model with the directionality of each effect are shown in Table 1. Colour coding indicates the parental treatments: red—offspring of water-deficient parents, blue—offspring of parents that experienced water control conditions; intense colour—offspring of mycorrhizal parents, light colour—offspring of non-mycorrhizal parents. The bottom and top of the boxes are the 25th and 75th percentiles, respectively, the centred band is the median and the whiskers represent 1.5 times the length of the box further from the box limits or the maximum or minimum observation in the absence of outliers
TABLE 1. Summary of the results of linear model for main and interaction effects of offspring and parental treatments (the latter highlighted are in grey and in bold) on several characteristics of offspring plants during juvenile stage. Degrees of freedom followed by F and p values are given for all the effects analysed. Significant (p < 0.05) are shown in bold and colour highlighted indicating the direction of the effect: positive in blue or negative in red
Source of variation df Plant traits AMF colonization
Whole-plant traits Leaf traits Root traits
log (Total biomass) log (Root mass factor) (RMF) log (Leaf dry matter content) (LDMC) log (Specific leaf area) (SLA) Leaf P concentration Leaf N concentration Leaf C:N ratio log (Specific root length) (SRL) log (Average root diameter) % Fine roots Root length colonized by AMF (%RLC) Arbuscules:vesicles ratio
F p F p F p F p F p F p F p F p F p F p F p F p
Seed mass 1 8.11 <0.01 9.80 <0.01 0.003 0.96 4.89 0.03 0.17 0.70 0.01 0.91 0.22 0.65 0.21 0.64 5.60 0.02 1.06 0.30 0.38 0.54 1.82 0.18
Seed stoichiometry 1 0.38 0.53 1.08 0.30 0.96 0.33 0.05 0.83 1.02 0.35 0.10 0.76 0.04 0.85 0.04 0.84 0.14 0.71 0.15 0.70 0.01 0.92 0.02 0.90
Offspring mycorrhizal inoculation (Off. M) 1 38.33 <0.01 98.78 <0.01 163.77 <0.01 230.73 <0.01 232.11 <0.01 97.60 <0.01 33.43 <0.01 2.41 0.12 6.43 0.01 0.32 0.57
Offspring water availability (Off. W) 1 62.46 <0.01 341.79 <0.01 343.50 <0.01 195.66 <0.01 54.76 <0.01 4.32 0.08 0.82 0.40 245.86 <0.01 57.51 <0.01 0.91 0.34 5.38 0.02 10.82 <0.01
Parental mycorrhizal inoculation (Par. M) 1 6.10 0.01 7.39 <0.01 10.81 <0.01 3.90 0.049 3.36 0.11 0.06 0.82 0.07 0.81 0.004 0.95 1.15 0.28 4.24 0.04 1.54 0.22 1.10 0.30
Parental water availability (Par. W) 1 0.002 0.97 0.33 0.57 0.54 0.46 2.64 0.11 0.29 0.61 0.34 0.58 0.89 0.38 0.07 0.80 0.03 0.87 0.00 0.99 0.71 0.40 1.47 0.23
Off. M × Off. W 1 7.36 <0.01 8.50 <0.01 23.95 <0.01 19.27 <0.01 7.60 0.03 14.24 <0.01 3.95 0.09 0.25 0.62 4.66 0.03 0.73 0.39
Par. M × Par. W 1 3.43 0.07 1.73 0.19 0.16 0.69 0.58 0.45 0.66 0.45 0.06 0.82 0.87 0.39 0.27 0.60 1.85 0.17 4.63 0.03 0.37 0.55 1.47 0.23
Off. M × Par. M 1 0.77 0.38 0.02 0.90 0.001 0.98 0.14 0.71 0.53 0.50 0.40 0.55 0.53 0.49 0.05 0.82 0.04 0.84 0.29 0.59
Off. M × Par. W 1 0.01 0.91 0.33 0.57 0.59 0.44 1.52 0.22 0.01 0.92 1.03 0.35 0.99 0.36 0.50 0.48 0.37 0.54 0.47 0.49
Off. W × Par. M 1 0.07 0.80 1.84 0.18 1.75 0.19 0.72 0.40 6.08 0.048 0.47 0.52 0.93 0.37 0.43 0.21 0.04 0.84 0.46 0.50 1.05 0.31 0.03 0.87
Off. W × Par. W 1 0.01 0.93 0.18 0.67 0.0004 0.98 0.45 0.50 0.52 0.50 0.11 0.76 0.30 0.60 0.05 0.82 1.38 0.24 0.68 0.41 4.44 0.04 0.02 0.90
Off. M × Par. M × Par. W 1 3.79 0.05 0.12 0.73 0.38 0.54 0.71 0.40 0.09 0.77 3.70 0.10 1.97 0.21 4.19 0.04 5.74 0.02 5.57 0.02
Off. W × Par. M × Par. W 1 0.19 0.66 0.26 0.61 0.94 0.33 0.26 0.61 0.85 0.39 0.05 0.82 0.33 0.59 0.27 0.61 0.21 0.65 0.10 0.75 3.41 0.07 0.18 0.67
Off. M × Off. W × Par. M 1 0.52 0.47 2.27 0.13 0.44 0.51 0.003 0.95 6.08 0.048 0.53 0.49 0.85 0.39 0.14 0.71 3.42 0.07 1.96 0.16
Off. M × Off. W × Par. W 1 0.05 0.82 0.25 0.61 2.10 0.15 1.53 0.22 0.01 0.92 0.68 0.44 0.63 0.46 3.10 0.08 0.95 0.33 1.60 0.21
Off. (M × W) × Par. (M × W) 1 0.08 0.78 1.36 0.25 0.40 0.53 0.08 0.78 0.01 0.92 0.67 0.44 0.67 0.45 0.00 0.99 0.11 0.74 0.95 0.33
Details are in the caption following the image
Effect of the offspring and parental treatments on plant phenotype characteristics of the adult offspring. (a) total plant biomass, (b) leaf P concentration, (c) leaf C:N ratio, (d) root mass factor, (e) leaf dry matter content, (f) specific root length and (g) fine roots percentage. The significant factors of each model with the directionality of each effect are shown in Table S3. Colour coding indicates the parental treatments: red—offspring of water-deficient parents, blue—offspring of parents that experienced water control conditions; intense colour—offspring of mycorrhizal parents, light colour—offspring of non-mycorrhizal parents

We found transgenerational effects (i.e. parental conditions affecting offspring plants) after removing the effect of seed mass and seed stoichiometry (maternal seed provisioning effects). Following the multivariate analysis, we only found that parental mycorrhizal inoculation triggered transgenerational effects in the multi-trait phenotype (Figure 1b; Table S2). Similar transgenerational effects were found in all measured plant traits in the univariate analyses (Figure 2; Table 1). Juvenile offspring of mycorrhizal parents (Par. M) had in general lower biomass, lower allocation to the roots (lower RMF) but thinner roots (% Fine roots), and higher LDMC and leaf P concentration (Figure 1b; Figure 2a,b,d,e,g and Table 1). Except for biomass and LDMC, the direction of the response to the parental treatment was concordant with the response to the offspring treatment. For example, mycorrhizal offspring showed lower RMF and higher P concentration of leaves, and these effects were further pronounced when offspring also had mycorrhizal parents (Figure 2b,d and Table 1). Although only detected by the univariate analysis, the parental water conditions induced transgenerational effects on just three traits and, generally, in interaction with the offspring conditions (Par W × Off M × Par M; Table 1; Figure 2). We also detected significant transgenerational effects on plant traits triggered by the parental mycorrhizal inoculation in the adult stage, although they were less evident (Table S3). Adult offspring of mycorrhizal parents (Par. M) still had lower allocation to the roots (lower RMF; Figure 1b; Figure 3d and Table S3).

3.3 Offspring AM fungal colonization

The water availability treatments (Off. W) modified the AM fungal colonization of the juvenile offspring. Plants grown with water deficit cycles had less root length colonized by AM fungi (lower %RLC) but had higher arbuscule:vesicle ratio than control plants (Figure 4a,b and Table S2). Additionally, AM fungal colonization was affected by the parental treatments only when offspring plants had water deficit (Off. W × Par. W; Table 1). Offspring of parents that did not experience water deficit had higher %RLC than the ones of parents that had experienced water deficit (Figure 4a; Table 1). For the arbuscule:vesicle ratio, we did not detect significant transgenerational effects (Figure 4b; Table 1 and Figure S3).

Details are in the caption following the image
Effect of the offspring and parental treatments on AM fungal root colonization in juvenile stage (upper row) and adult stage (lower row): (a) and (c) percentage of root length colonized by AM fungi; (b) and (d) arbuscule:vesicle ratio. The significant factors of each model with the directionality of each effect are shown in Table 1 (juvenile) and Table S3 (adults). Colour coding indicates the parental treatments: red—offspring of water-deficient parents, blue—offspring of parents that experienced water control conditions; intense colour—offspring of mycorrhizal parents, light colour—offspring of non-mycorrhizal parents. No AM fungal colonization was detected in non-mycorrhizal offspring plant roots

During the adult stage, we found no significant difference in %RLC between the offspring water availability treatments (Off. W), although there was a lower arbuscule:vesicle ratio in plants with water deficit, reversed response compared with what happened during the juvenile stage (Figure 4c,d; Table S3 and Figure S3). Also, at this stage, we found that AM fungal colonization was affected by the parental treatments only when the offspring plants grew with water deficit cycles. The adult offspring of mycorrhizal parents (Off. W × Par. M) had higher %RLC than the ones of non-mycorrhizal parents (Figure 4c; Table S3).


In this study, we show the importance of AM symbiosis in triggering phenotypic plasticity in plants, both during their life cycle and in following generations. Such phenotypic changes can improve the ability of individuals to cope with environmental stress and likely increase the species’ realized niche. Here we show that mycorrhizal symbiosis could specifically trigger morphological changes related to resource use and resource acquisition strategies within generation and to succeeding generations. Furthermore, we provide evidence that AM fungal colonization of the offspring could be also affected by the conditions experienced by the parents. The transgenerational effects of mycorrhizal symbiosis and water availability were detected after controlling for differences in the amount and quality of resources provided in the seeds (Herman & Sultan, 2011), pointing to other potential drivers of environmentally induced heritable plasticity such as hormonal differences between offspring or epigenetic modifications.

4.1 Within-generational plasticity of offspring traits is developmental stage specific

The strong response of T. brevicorniculatum offspring plants to the conditions experienced during their life cycle (i.e. offspring mycorrhizal fungal inoculation and water availability treatments) shows the high level of plasticity of this plant species. However, we found that the response to these conditions differed in juvenile and adult phases, suggesting specific plant plasticity at different developmental stages (Coleman et al., 1994).

Measurements of different fitness-related characteristics (i.e. plant nutrition and growth) suggest that, generally, AM symbiosis improved plant performance and mitigated water deficit, since the benefit of being mycorrhizal increased with decreasing water supply (Aroca et al., 2012; Augé et al., 2015; Doubková et al., 2013). Regarding the nutritional effect, mycorrhizal fungal inoculation did dramatically offset drought stress and increased leaf P concentration at both offspring developmental stages, although it decreased leaf N concentration (Figures 2b,c and 3b,c; Table 1 and Table S3). These results reflect that, at least for some nutrients such as P, and despite being in water deficit conditions (which could also cause nutrient deficit, although we did not find this for N acquisition), mycorrhizal plants had better nutritional supply (Doubková et al., 2013; Lu & Koide, 1994). The unexpected effect of AM on N acquisition is striking, and could suggest that γ-irradiation might have increased some available forms of N (e.g. urn:x-wiley:00220477:media:jec13810:jec13810-math-0001 and urn:x-wiley:00220477:media:jec13810:jec13810-math-0002, summarized by McNamara et al., 2003) in sterile soils of NM treatment. It is also possible that the N was translocated mostly to other organs such as the storage taproot rather than to leaves, or that N immobilization in the mycorrhizal mycelium was not important enough (Reynolds et al., 2005). Nevertheless, plants were probably P- rather than N-limited because the additional acquisition of P caused by the mycorrhizal fungal inoculation translated into increased growth (i.e. plant biomass increase) despite a potential higher nutrient availability in sterile soil. This benefit allowed mycorrhizal plants to reach similar growth than non-stressed plants, despite experiencing water deficit conditions. In adult offspring, the benefit was more pronounced than in juveniles (Figure 3a vs. Figure 2a), probably due to a greater cost/benefit ratio during the juvenile stage (Johnson et al., 1997).

In both developmental stages, mycorrhizal fungal inoculation and water availability treatments induced significant changes in multiple plant phenotypic traits related to the resource-use strategy of the plant (also called plant economic spectrum; Reich, 2014), including both below-ground and above-ground traits (Fusconi, 2014; Goh et al., 2013; Nuortila et al., 2004). However, traits seem to respond differently depending on the developmental stage of the plant.

During the juvenile stage, water availability and mycorrhizal fungal inoculation triggered independent and additive trait plasticity in the same direction (Table 1; Table S2). Similar to findings of Shumway and Koide (1994), under reduced water availability and/or absence of AM fungi, plants shifted towards a more resource-conservative phenotype based on the plant economics spectrum framework (Reich, 2014). This phenotype is characterized by having greater below-ground biomass allocation (RMF), less photosynthetic but more water-use efficient leaves (greater LDMC, lower SLA), and thicker and more resistant roots (lower SRL; Figures 1 and 2 and Figure S4). These traits are expected to be beneficial when resources are scarce since they are associated with longer life span and enhanced water-use efficiency of the plant under water stress (Reich, 2014). Thus, the plastic response towards a conservative phenotype could improve T. brevicorniculatum ability to cope with water deficit. AM symbiosis seems to induce quantitative modifications in plant traits: the more root length is colonized by AM fungi (%RLC), the thinner the roots and higher the SLA of the host plants are (Figure S4a).

During the adult stage, the direction of plasticity changed compared to the response at the juvenile stage. In response to reduced water availability and/or absence of AM fungi, plants shifted towards a more resource-acquisitive phenotype. The plants decreased LDMC and C:N ratio and increased their SRL and percentage of fine roots, reflecting an adaptive phenotypic plasticity that improved resource uptake and compensated for the lack of AM symbiosis and the involvement of extraradical mycelium in plant resource uptake (Fusconi, 2014; Goh et al., 2013; Pozo et al., 2015). These different responses depending on plant developmental stage could explain why previous studies found variable effects of mycorrhizal symbiosis on plant phenotype (Johnson et al., 2012; Nuortila et al., 2004). Nevertheless, at both stages, we found strong positive correlation between root length colonized by AM fungi (%RLC) and SRL and strong negative one between %RLC and root diameter (Figure S4a,b).

4.2 Transgenerational effects on offspring phenotype and performance

Several studies had already shown that offspring from mycorrhizal parents can have greater performance and/or fitness, reflected in higher biomass, survival, growth rate and seed production (Heppell et al., 1998; Koide, 2010; Varga et al., 2013). This is partially confirmed in the current study. In regard to plant nutrition, we found that offspring of mycorrhizal parents had higher leaf P concentration than those of non-mycorrhizal parents (Figure 2b), suggesting that mycorrhizal symbiosis, besides directly providing soil nutrients to host plants, could also increase offspring nutrient uptake via transgenerational effects. In terms of biomass, we observed that having non-mycorrhizal parents was, however, beneficial (Figure 1a; Figure 2 and Table 1). This contrasting result in biomass (in relation with previous evidence) was probably due to the fact that in our study we experimentally controlled the effect of seed mass and stoichiometry, unlike previous studies (Dechaine et al., 2015; Germain et al., 2019; Herman & Sultan, 2011; but see Varga & Kytöviita, 2010). After removing the effect of differences in seed provisioning, the (‘remaining’) detected transgenerational effects might probably stem from epigenetic or hormonal differences between the offspring (Herman & Sultan, 2011; Rottstock et al., 2017; Varga & Soulsbury, 2017). We suggest that detailed molecular analyses of the plant material can complement this type of experiment in the future (e.g. checking for differences among offspring or in their colonized AM fungal community). Another possible way to confirm whether these effects were epigenetically controlled would be to modify the epigenetic signature of the plants via application of a demethylation agent (Puy, Carmona, Dvořáková, et al., 2021; Puy et al., 2018). However, it should be first tested whether the demethylation agent has any harmful effects on the formation of AM symbiosis.

Regarding the offspring phenotype, we found that the parental conditions greatly affected traits linked with the resource use and exploitation strategy of the plant (Reich, 2014). Moreover, this transgenerational plasticity (i.e. response to parental treatments) was concordant with the within-generational plasticity to the ongoing treatments. Plant traits shifted towards a ‘stress-coping phenotype’ under water deficit and absence of AM symbiosis during their life cycle (i.e. more conservative phenotype: increased the RMF and decrease SRL). Additionally, offspring individuals became even more conservative when their parents were under those stressful conditions (i.e. water deficit and absence of AM symbiosis; Figures 1b and 2). Thus, the transgenerational effects reinforced trait plasticity when individuals were grown in the same conditions as their parents. This can be seen as a ‘stress memory’, that is, a transmission of environmental information across generations, for example a recurring stress, that induces phenotypic modifications anticipating the offspring for it and improving offspring's ability to cope with predictable environment (Lämke & Bäurle, 2017). The higher biomass found in offspring of non-mycorrhizal parents could be a consequence of the heritable phenotypic plasticity, which is adaptive and enhances water-use efficiency of the offspring regardless of the environment they grow in. Also, it is important to emphasize that we found that transgenerational effects on phenotype were expressed early in the ontogeny (Figure 2) and faded away over offspring life development (Figure 3). This result reinforces the idea of transgenerational effects as an important factor promoting adaptation to repeated ecological conditions, especially during juvenile stages and establishment of communities (Dechaine et al., 2015; Latzel et al., 2014). Also, even though transgenerational effects are likely to fade away with time, the effects associated with differences in the seeds might fade away even faster (Latzel et al., 2010) as happened in our case (Table 1 and Table S3).

Even though both abiotic and biotic parental environments seemed to trigger transgenerational effects, we found that for our focal species, parental mycorrhizal fungal inoculation affected offspring plant traits more than the parental water availability treatments used in our experiment (Figure 1; Figure 2 and Table 1). The fact that the parental water treatment did not last until the seed formation and that its duration was shorter (i.e. just 2- to 3-week-long water deficit pulses) in comparison to the parental mycorrhizal treatment could potentially explain this result. However, it is also likely that water is a less crucial stressor for T. brevicorniculatum—it grows well under different water availability conditions (Luo & Cardina, 2012)—and therefore transgenerational effects of water availability could have not been evolutionary or ecological strongly constrained (Rendina González et al., 2018). In this sense, we propose to explore whether those adjustments are found across different species and genotypes, both experimentally and in natural populations; doing so would allow to expand the ecological relevance and realism of our study and provide a deeper understanding of the relevance of transgenerational plasticity in response to different biotic interactions.

4.3 Within- and transgenerational effects on AM fungal colonization

As expected, the environmental factors experienced by offspring during their life cycle (Off. W) affected the offspring AM fungal colonization. Contrary to our expectation (Martínez-García et al., 2012; Pozo et al., 2015), water deficiency did not stimulate root AM fungal colonization (since %RLC decreased). However, the proportion of arbuscules:vesicles increased in offspring with water deficit, suggesting that water deficiency can stimulate resource trading and AM mutualistic functioning since the arbuscules are the main structures where resource exchange takes place.

Moreover, we found that AM fungal colonization of the offspring could be influenced by the conditions experienced by the host plant parental generation. To our knowledge, this is the first study that shows transgenerational effects on AM fungal colonization (but see De Long et al., 2019 for differences in AM fungal structures). Specifically, we observed that only under water deficit conditions, the offspring of parents that experienced water scarcity had lower AM fungal colonization. Although this appears to contradict our initial hypothesis that resource scarcity (i.e. water deficit and NM parental treatments) would stimulate AM symbiosis of the offspring, we must note that offspring from water deficit and NM parents compensated by having increased root biomass. This means that this offspring had in total more roots colonized and greater number of arbuscules per individual than offspring from parents under ambient water conditions (see the calculations made by relativizing to the total root biomass of the plant—that is, root biomass × %RLC—Figure S5). Thus, we conclude that the result in general still supports the notion that transgenerational effects modify offspring towards the ‘stress-coping phenotype’ stimulating the establishment and activity of the AM symbiosis.

As also found for plant traits, the AM fungal colonization of the offspring was still influenced by the parental conditions during adult stage. This suggests that transgenerational effects could influence plant–AM fungi relationship and persist further than just the establishment and early stages of the symbiosis. At the adult stage, %RLC was affected by the parental mycorrhizal status so that offspring from mycorrhizal parents had higher %RLC under water deficit. These results suggest that mycorrhizal symbiosis could be promoted in the offspring when the parental generation had experienced mycorrhizal symbiosis.


We found that mycorrhizal symbiosis, alone and in combination with water availability, triggered phenotypic changes within and across generations on plant performance and AM fungal colonization. Water deficiency and absence of AM fungi triggered concordant plant phenotypic plasticity, towards a stress-coping phenotype, both within and across generations. This reflects an adaptive epigenetic mechanism that promotes rapid adaptation, and probably improves the ability of the species to cope with water deficit. In a context where the importance of individual and intraspecific variation of mycorrhizal plants and fungi in ecosystems is increasingly acknowledged (Johnson et al., 2012), our study provides evidence of a mechanism of phenotypic variation that has been neglected so far: transgenerational plasticity. These modifications confer competitive advantages to the next generation. Importantly, we show that transgenerational effects remained significant even after accounting for the differences in seed quality stocked by the mother, possibly pointing to hormonal differences or heritable epigenetic mechanisms as potential transmitters of these effects across generations.


We thank M. Applová, T. Jairus and N. G. Medina for technical assistance and N. Plowman for English revision. This research was financially supported by the Czech Science Foundation grant GACR (Grant No. GA17-11281S) and the European Union through the European Regional Development Fund (Dora Plus grant). J.P. was funded by the Irish Research Council Laureate Awards 2017/2018 (IRCLA/2017/60) to Yvonne Buckley. I.H., M.Ö., C.P.C. and M.M. received support by grants from the Estonian Research Council (PSG293, PUT1170, PRG1065) and by the European Regional Development Fund (Centre of Excellence EcolChange).


    J.P., C.P.C., I.H., M.Ö., F.d.B. and M.M. designed the research; J.P. performed the experiments; J.P. and C.P.C. analysed the data; J.P. wrote the main manuscript. All authors contributed substantially to revisions and gave final approval for publication.


    The peer review history for this article is available at https://publons.com/publon/10.1111/1365-2745.13810.


    The data that support the findings of this study are available on Figshare repository with the identifier https://doi.org/10.6084/m9.figshare.15124647 (Puy, Carmona, Hiiesalu, et al., 2021).