Introduction

Extreme drought events are becoming common and widespread across the biosphere as a result of anthropogenic climate change (IPCC 2023; Liu et al. 2018). The effects of such drought events on plant communities can be dramatic (Luo et al. 2019; Ploughe et al. 2018; Stampfli and Zeiter 2004). For instance, drought can act as a strong environmental filter to eliminate plant species that lack traits for drought tolerance from the plant community (Engelbrecht et al. 2007; Moeslund et al. 2013; Tilman and Haddi 1992). This allows plants with certain traits to persist during extreme drought and subsequently thrive due to reduced competition and surges in nutrient availability upon rewetting (Cleland et al. 2013; Leitner et al. 2017). With extreme drought events becoming more widespread and pronounced with climate change (IPCC 2023; Lange et al. 2020), understanding the mechanisms that underlie plant species recovery after extreme drought is critical to predict and manage ecosystem responses.

The persistence and recovery of plants during and after drought can further depend on their interaction with neighbouring plants (Cadotte and Tucker 2017; Kraft et al. 2014). Numerous recent studies have shown how intraspecific interactions alter plant responses to drought, resulting in either an amplification of negative drought responses (Foxx and Fort 2019; Guo et al. 2020) or facilitation through improved drought tolerance (Wang and Callaway 2021; Zhang et al. 2017). Neighbouring conspecific plants can strongly impede each other’s persistence and recovery during and after drought through competition for space, nutrients and light (Foxx and Fort 2019; Guo et al. 2020). On average, such intraspecific interactions can be several folds stronger than interspecific competition in co-occurring plants, as these plants have greater niche overlap, which limits plant performance (Adler et al. 2018). Moreover, at higher plant densities, intraspecific plant competition could lead to reductions in average shoot and root biomass due to limited space and nutrient availability (Postma et al. 2021). This reduction of plant growth, especially root growth, can exacerbate the effects of a disturbance event, such as drought, by impairing water uptake, thereby inducing density-dependent mortality of plants (Casper and Jackson 1997). Examining intraspecific plant interactions is essential for understanding the responses of plant populations to drought. Specifically, as these interactions determine resource availability, competition, facilitation, and ultimately influence the recovery potential of plants within ecosystems.

Intraspecific plant competition can lower soil nutrient availability with subsequent effects on plant-soil biota interactions, such as symbiotic interactions between plant and mycorrhizal fungi (Ayres et al. 2006; Koide 1991). Arbuscular mycorrhizal (AM) fungi have been extensively studied to understand how their positive symbiotic relationships can mitigate the negative effects of drought on the host plant (Augé 2001; Jongen et al. 2022; Worchel et al. 2013). Fungal extraradical mycelia cover a surface area 10- to 1000-times larger than that of root hairs, making mycorrhizal fungi highly efficient in taking up water and nutrients (Goltaph et al. 2008; Marjanović and Nehls 2008). By infecting and spreading within the root cortical cells of host plants, AM fungi form a symbiotic relationship with plants where nutrients such as phosphorus (P) and nitrogen (N) and water are exchanged for photosynthesized carbon and lipid (Wang et al. 2017). Plants have been found to acquire up to 80% of their essential N or P through this symbiosis (van der Heijden et al 2008) and several meta-analyses have consistently shown that AM fungi can ameliorate the drought stress on plant performance (Delavaux et al. 2017; Hawkins and Crawford 2018; Jayne and Quigley 2014; Kivlin et al. 2013). However, studies investigating the effects of AM fungi on intraspecific competition often show a diminished beneficial effect of AM fungal colonisation compared to communities with interspecific competition (Tedersoo et al. 2020; Guo et al. 2022). This shift in response to mycorrhizal colonisation under intraspecific competition is likely because mycorrhizal fungi intensify competition between plants which overlap in niche and nutrient requirements (Tedersoo et al. 2020; Guo et al. 2022). In contrast, under interspecific competition, mycorrhizal fungi can promote the performance of weaker competitors and dampen competitive interactions (Hart et al. 2003; Wagg et al. 2011). Yet, we know little about how intraspecific plant competition and plant-mycorrhizal symbiosis can interactively affect plants’ responses during and after drought events (Birhane et al. 2014; Hawkins and Crawford 2018; Zhang et al. 2011).

Drought can further amplify negative intraspecific competition within plant populations, which could weaken the benefits provided by mycorrhizal fungi to host plants (Hawkins and Crawford 2018). Alternatively, AM fungi can also relax drought-induced amplification of intraspecific plant competition. For instance, the same AM fungi that negatively affected biomass production of plants in ambient water conditions by intensifying intraspecific competition also reduced intraspecific competition in drier soil conditions, subsequently benefitting plants (Duan et al. 2021; Zhang et al. 2011). The benefits of AM fungi to plant populations during and after a drought may vary depending on mycorrhizal responses to changes in water availability and the intensity of intraspecific plant competition (Birhane et al. 2014; Duan et al. 2021; Hawkins and Crawford 2018; Meisner et al. 2013; Zhang et al. 2011).

Here, we conducted a growth chamber experiment to study the post-drought recovery of two congeneric Centaurea plants; a common European resident plant (Centaurea jacea) and its congener range-expanding plant (Centaurea stoebe), which is expanding its geographic range from southern Europe to northern Europe in recent years (Wilschut et al. 2019). As a result of ongoing climate change, many species are expanding their native range to track their favourable climatic conditions (Anderson 2015; Walther et al. 2002). However, the ability of plants to expand their range is often constrained by the novel biotic and abiotic conditions of the new habitat (Morriën et al. 2010; Spence and Tingley 2020). Range-expanding plants like C. stoebe arriving from more arid environments may profit over native plants in drought conditions (Yang et al. 2022), although this may further depend on how intraspecific plant competition limits the C. stoebe growth in the presence of AM fungi.

We therefore experimentally manipulated the presence of an AM fungi species (Rhizophagus irregularis) and created a density gradient (to create a gradient of intraspecific plant competition) within C. jacea and C. stoebe. Through this, we aim to investigate the interactive effects of AM fungal colonisation and intraspecific plant competition on their species-specific drought recovery. We hypothesize greater post-drought recovery with decreasing intraspecific plant competition. We also hypothesize greater drought recovery of plants in the presence of AM fungi, though this AM fungi mediated recovery will be dampened at high plant densities (high intraspecific competition).

Materials and methods

Study species

Plants

Centaurea jacea and Centaurea stoebe are herbaceous plants and belong to the family of Asteraceae. Centaurea jacea is a perennial flowering plant that is native and widespread throughout Europe. Centaurea stoebe is a biennial or short-lived perennial flowering plant that is also native to Europe but is expanding its northern European range due to climate warming (Broennimann et al. 2014; Lauber et al. 2018). Given that previous studies have shown both common and distinct responses of these two plant species to climate change manipulations despite being closely related (Koorem et al. 2021; Quist et al. 2020; Wilschut et al. 2019), we chose these plants to advance the current understanding of plant recovery after drought by exploring their intraspecific interactions and mutualistic interactions with mycorrhizal fungi.

Arbuscular mycorrhizal fungi

We inoculated our study soils with Rhizophagus irregularis, previously known as Glomus intraradices (Stockinger et al. 2009), a well-known model AM fungi species from the family Glomeraceae (Krüger et al. 2012; Tisserant et al. 2013). Rhizophagus irregularis can colonise the roots of numerous plant species, such as our study Centaurea species (Bunn et al. 2014; Thakur et al. 2019). As such, it is described as a generalist coloniser of plants with a widespread distribution (Basiru et al. 2021; Savary et al. 2018).

Experimental design

Seeds of both plant species (C. jacea and C. stoebe) were obtained from a seed company (UFA Samen, Switzerland) and were stored at 4 °C before germination. For surface sterilisation, the seeds were bleached for 15 min in a 30% bleach solution (commercial bleach with sodium hypochlorite) and rinsed with deionised water subsequently. The germination was initiated on a moist filter paper (using deionised water) in Petri dishes kept in the dark for one week at room temperature (average of ~ 20–22 °C). Subsequently, seedlings were transferred carefully into a multi-pot tray containing sterilised soils (CAPITO line, Landi, Switzerland). We sterilized soils in an autoclave (Systec VX-150, Systec GmbH & Co., Germany) twice at 121 °C for 20 min, and the two cycles separated by at least 48 h to target more resistant fungal species that opportunistically spread in the soil. The seedlings in the multi-pot trays grew for one week in the climate chambers at 20 °C/16°C at 16 h day (i.e., with light) and 8 h night (i.e., dark) conditions. One-week-old seedlings were then transplanted into 0.7 L pots (10 × 10 × 11 cm) containing either the sterilised substrate or the same substrate inoculated with AM fungi.

The soil used in our experiment (both for germination and the main experiment) were a mixture of 50% quartz sand (particle size = 0.3–0.7 mm), 40% universal potting soil (Terre Suisse AG, Switzerland) and 10% perlite (abiotic properties of the substrate: Ph = 6.7, organic matter = 3.4%, N = 0.004%, C = 0.034%, Pbioavailable = 96 mg/kg). The soil was hand-mixed after bigger particles – such as stones, clay and wood – were removed from the potting soil with a coarse-meshed sieve of 0.5 cm mesh. Soil mixtures were also sterilised in an autoclave twice at 121 °C for 20 min, and the two cycles were separated by at least 48 h exactly in the same way as the soil used for the germination of plants. The autoclaved soil was then distributed into the plant pots (height = 120 mm, diameter = 140 mm), with a total of 800 g of dry weight in each pot.

For the colonisation of plant roots with mycorrhizal fungi, we used MYC 800 (Andermatt Biocontrol, Switzerland), a powder that is commonly used as a solid fertiliser containing germinating spores of R. irregularis. As a supporting substrate, the MYC 800 powder consists of 80% kaolin and 20% diatomite. One gram of this product provides approximately 800 propagules (mainly spores). We inoculated AM fungi treatments with 2 g of this powder (i.e., ~ 1600 spores of AM fungi). The inoculum was mixed into the substrate before planting the seedlings in order to enable faster contact with the root surfaces of the plant. To control for AM fungi-associated microbes present in the inoculum, we collected a microbial wash by filtering the same amount of inoculum used for the AM fungi treatment with 6 L deionised water through a 25 μm mesh net. The size of the mesh was large enough to allow microbes to pass through and small enough to prevent contamination of mycorrhizal spores and hyphal fragments (Błaszkowski et al. 2008; Taktek et al. 2015). Each pot assigned to non-mycorrhizal (control) treatment received 50 mL (corresponding to the amount of inoculum added to mycorrhizal treatment pots) of the microbial wash when watered for the first time. Analysis of root mycorrhizal colonisation in plants that were grown in soils without AM fungi confirmed that our sterilised soils (added with AM fungi-associated microbes) were free of AM fungal spores.

Seedlings ranging between 1 and 2 cm in height were transplanted into pots with densities ranging from one to five, with and without AM fungal inoculation (Fig. 1). Pots were then randomly allocated to four tables in two climate chambers with identical light and temperature settings. In both climate chambers, the plants were exposed to the following growing conditions: 16 h of daytime at 20 °C with a light intensity of ~ 13,500 lx and 8 h of night-time at 16 °C. The room’s relative humidity (RH) was approximately 50% during the day and about 80% during the night. Temperature, light intensity and room’s air RH were constantly monitored to account for differences on the four tables in the two climate chambers (Supplementary Fig. 1). We let the plants grow for a total of nine weeks, within which half the pots were exposed to an extreme drought event (Fig. 1). When not subjected to drought, pots were continuously watered every four to five days with 100 ml of deionised water. For the drought treatment, the plants were watered with the same amount for the first three weeks. These three weeks were to enable plants to establish themselves in the soil, but also to facilitate the root colonisation by R. irregularis, which is usually well established after around the third or fourth week from the initial colonisation (Corkidi et al. 2004). After that, drought treatments received no water at all for the next three weeks, as shown in Fig. 1. We withheld water for three weeks in these treatments to simulate an extreme drought event, pushing many plants to their wilting point as soil water content reached 0% (volumetric water content) (Fig. 1). Following this drought period, plants were allowed to recover by rewetting the pots, which was carried out by a regular addition of deionised water (the same way for no drought treatments). Soil moisture was regularly checked with a Soil Moisture Meter TDR 150 (FieldScout, Spectrum Technologies Inc., USA) at two depths (3.8 and 7.6 cm) on 24 extra pots (one for each treatment combination) (Fig. 1) in order to monitor soil water availability across treatments without disturbing the main treatment units. Each treatment combination was replicated six times, resulting in a total of 240 pots (2 plant species x 5 densities x 2 AM fungi treatments x 2 drought treatments x 6 replicates) and 720 plant individuals.

Fig. 1
figure 1

Schematic representation of experimental design with plant density and extreme drought as our main treatments. Temporal soil water content of the pot under drought and control treatments are indicated in the lower panel (data shown from extra pots, details in methods). Plants in different drought and density treatments were inoculated or left un-inoculated with the mycorrhizal fungus Rhizophagus irregularis

Full size image

Harvest and response variables

The height of each plant was recorded as the distance from the soil surface to the highest point of the upstretched longest leaves every week during the experimental period. Measurement of chlorophyll content was taken before and after the extreme drought of the two youngest fully expanded healthy leaves per plant using a SPAD-502 Chlorophyll Meter (Konica Minolta, Tokyo 100–7015, Japan). After two weeks of post-drought recovery and on the ninth week of the experiment, final measurements of the plant height and chlorophyll content were taken again, and plants were harvested.

During the harvest, the aboveground tissue of each plant was removed just above the soil level, and a single young fully expanded leaf from each plant was cut at the base of the petiole to later measure plant leaf traits. Roots were meticulously washed in order to remove attached substrates, and root samples of about 1 g (fresh weight) were taken from each pot for mycorrhizal colonisation and root trait measurements. The fresh weight of the remaining root was weighed and then dried in an oven for 3 days at 40 °C along with the plant shoot to measure the dry biomass of each plant. Due to the intertwining of roots, it was not possible to measure the dry biomass of each individual plant, as such, the total root biomass per pot was divided by the number of plant individuals to express the average plant root biomass per individual. The fresh leaf and root samples were weighed and scanned using an Epson Perfection V850 Pro Scanner, and were analysed using ImageJ and RhizoVision Explorer v2.0.3 (Rasband 1997; Seethepalli et al. 2021, respectively), to collect data on specific leaf area (SLA; leaf area divided by its dry weight), root diameter and specific root length (SRL; root area divided by root weight). Specific root length was estimated as the ratio of root length to its dry mass. Leaf samples were also dried as described above to calculate the leaf dry matter content (LDMC) as the leaf fresh weight divided by their dry biomass (Cornelissen et al. 2003).

We also estimated carbon, nitrogen and their ratio (C: N ratio) of belowground and aboveground plant organs by dry combustion of ground root and leaf material using a CN elemental analyser (CNS-Analyzer: Elementar vario EL cube, Elementar Analysensysteme GmbH, Langenselbold, Germany) following the Micro-Dumas combustion method (Stewart et al. 1963). Sample preparation, prior to C and N analyses, consisted of grounding of one young fully expanded dry leaf (for density treatment with more than 1 individual, only a leaf from a random individual was chosen) and root samples (for density treatment with more than 1 individual the pool of root of each pot) material using tissue lyser machine (QIAGEN Tissue Lyser II Retsch MM400, Düsseldorf, Germany) and record the exact weight of the tissue powder (around 2 mg).

Finally, we measured the percentage of total root AM fungal colonisation and specific AM fungal structures by staining roots with dye (Pelikan 4001 ink) using techniques modified from Philips and Hayman (1970). This allowed us to visualise colonisation of mycorrhizal structures within the roots. Once stained, root samples were immersed in a mixture of water, glycerin and lactic acid (v:v:v) and were inspected under a Leica S9i Microscope (55x magnification)(Leica Microsystems, Wetzlar, Germany). To measure the percentage of mycorrhizal colonisation, we used the modified gridline intersect method from Giovannetti and Mosse (1980). Root length colonisation (%) was calculated as a measure of all mycorrhizal structures present in the root, also using the equation presented in Giovannetti and Mosse (1980).

Statistical analysis

For non-temporally measured response variables (i.e., only at the end of the experiment), we used linear mixed-effects models to test the effects of plant density (as a continuous variable), AM fungi and drought treatment on plant responses, while using pot placement (i.e., two different tables used in each climate room-thus, four tables in total) as a random intercept (to account for any variability in light intensity among the four tables). For the temporal data collected for plant height, plant leaf production and chlorophyll content throughout the study, we used linear mixed-effect models: fixed effects in these models were the same as for the previous models, whereas the random effects were the week of measurement and pot placement (e.g., following the model structure of the lme4 package in R: biomass ~ plant density*AMF treatment*drought treatment + (1|random effect1) + (1|random effect2). Mixed-effects models were run using the lme4 package (Bates 2015) for R statistical software v4.0.3 (R Core Team 2020). The treatment effects in mixed models were evaluated with a Type III Analysis of Variance (ANOVA) with Satterthwaite’s method for the estimation of degrees of freedom, using the lmerTest package v4.0.3 (Kuznetsova et al. 2017). Model assumptions (e.g., homogeneity of variance and normality of residuals) were inspected visually for each linear model. To meet the model assumptions, some response variables were log-transformed (indicated in Table 1). We ran all mixed-models separately for C. jacea and C. stoebe, given that we expected the effects of all treatments to be general across the two species and to further reduce the complexity of models. Conditional R2 values were taken as the proportion of total variance explained through both fixed and random effects of the linear models and their statistical significance was obtained from the r2glmm package v4.0.3 (Nakagawa and Schielzeth 2012; Jaeger 2017). To explore the strength of relationships between two variables for understanding of potential mechanisms, we tested correlations between response variables such as, root N content and root mycorrhizal colonisation, using major axis regression models (RMA) with the lmodel2 package v4.0.3 (Legendre 2018). We further ran a multivariate statistical test (PERMANOVA) with 999 permutations via the adonis2 function and carried out a principle component analysis (PCA) using the vegan package v4.0.3 (Oksanen et al. 2020). The PCA allowed us to analyse variation in multiple plant trait responses including those of specific leaf area (SLA), leaf dry matter content (LDMC), leaf number, leaf chlorophyll content (measured as SPAD), root diameter and specific root length (SRL). For the PCA, we only chose two extremes of density treatments (density = 1 and density = 5) to understand how plant traits at these two ends may help explain post-drought recovery in two plants. We ran PCA in the vegan package (Oksanen et al. 2020) and used the scores of the first and second PCA axes (as they two explained most of the variation) to represent overall variation in response to drought and intraspecific competition.

Table 1 Results from linear mixed-effect models testing the effects of extreme drought and rewetting (DR), intraspecific competition intensity (DEN), and AM fungi presence (AM fungi) for C. jacea and C. stoebe (with table number as a random effect)
Full size table

A multivariate statistical test (PERMANOVA) was run using the adonis2 function in the vegan package. All (data) figures were created using the ggplot 2 package v4.0.3 (Wickham 2016).

Results

Plant biomass responses

Increasing plant density consistently decreased shoot and root biomass of individual plants in both C. jacea and C. stoebe (shoot: F1,111 = 173.49, P < 0.001 and F1,111 = 202.54, P < 0.001; root: F1,111 = 156.74, P < 0.001 and F1,111 = 162.56, P < 0.001, for both C. jacea and C. stoebe respectively; Table 1; Fig. 2), leading to a 76% decrease in total biomass (shoot + root) of plant individuals at the highest plant density in C. jacea and a 71% decrease in C. stoebe, compared to the lowest plant density treatments (F1,111 = 180.73, P < 0.001 for C. jacea and F1,111 = 191.20, P < 0.001 for C. stoebe; Table 1; Fig. 2). Increasing plant density further increased the root: shoot ratio of C. stoebe (F1,109 = 6.06, P < 0.05; Table 1; Fig. 2), while C. jacea was unaffected (F1,107 = 1.72, P = 0.19). Recovery from the extreme drought event exacerbated these negative plant density effects on plant biomass at the end of the experiment, specifically by reducing root biomass in C. jacea (significant interaction between drought and density, F1,109 = 11,28, P < 0.01) but not in C. stoebe (F1,110 = 0.63, P = 0.43; Table 1; Fig. 2).

The presence of AM fungal species increased shoot biomass of both plants (F1,110 = 17.71 and F1,108 = 28.16, P < 0.001, for C. jacea and C. stoebe respectively; Table 1; Fig. 2). These biomass responses to AM fungi shifted depending on the drought treatment and plant density. For example, C. jacea individuals subjected to extreme drought responded negatively to AM fungi resulting in decreased root and total biomass (Root: F1,111 = 6.61, P < 0.05; Total Biomass: F1,111 = 6.58, P < 0.05; Table 1; Fig. 2); whereas, in C. stoebe these conditions led to a shift in root: shoot ratio with reduced root biomass allocation (F1,107 = 4.26, P < 0.05; Fig. 2). At high plant density, C. jacea and C. stoebe plants also responded negatively to AM fungi with further reductions in their shoot biomass (F1,111 = 9.93, P < 0.01 for C. jacea; F1,110 = 14.82, P < 0.001 for C. stoebe; Table 1; Fig. 2). However, in C. stoebe individuals recovering from drought, the presence of AM fungi at high plant densities resulted in a significant increase in the root: shoot ratio (F1,105 = 4.96, P < 0.05; Table 1; Fig. 2).

Temporal plant responses

We found that extreme drought induced a complete mortality of the plants within three pots (no recovery after rewetting), all of which were C. jacea at the highest population density in our experiment, with two of them inoculated with AM fungi and one without. Apart from these plants, recovery was visible for most plants, with a 34% increase in height for C. jacea and a 5% increase for C. stoebe during the recovery period after rewetting of pots (Temporal height data: Supplementary Fig. 2; Temporal SPAD data: Supplementary Fig. 3).

Fig. 2
figure 2

Plant biomass responses of C. jacea (left: a, c and e) and C. stoebe (right: b, d and f) to drought, AM fungi and plant density. Average plant total biomass: a, b; Average aboveground biomass: c, d; Average belowground biomass: e, f; Root to shoot ratio. Raw data are shown as points, whereas dashed lines are based on linear regressions

Full size image

Plant trait responses

Leaf morphological traits

Leaf trait were lesser responsive to AM fungi or to drought treatment compared to plant density in both species (Table 1). Increasing plant density reduced leaf trait values in both C. jacea and C. stoebe, such as declines in LDMC in C. stoebe (F1,112 = 13.46, P < 0.001; Table 1 & Supplementary Fig. 4) and SLA in C. jacea (F1,108 = 8.34, P < 0.01; Table 1). Leaf trait responses to AM fungi in C. jacea were, however, dependent on other treatments, such as drought, in which AM fungi induced a greater decline in LDMC (F1,106 = 4.12, P < 0.05; Table 1 & Supplementary Fig. 4) and an increased SLA, although only in C. jacea (F1,108 = 5.75, P < 0.05; Table 1).

Root morphological traits

Root trait responses to the experimental treatments were species dependent and only evident in C. stoebe, not in C. jacea (Table 1; Fig. 3). In C. stoebe, the presence of AM fungi increased root diameter (F1,105 = 12.94, P <0.001; Table 1 & Fig. 3), however, when subjected to extreme drought, the presence of AM fungi decreased root diameter (F1,105 = 5.53, P <0.05; Table 1). By contrast, we found increase in root diameter among C. stoebe individuals after the extreme drought event in the presence of AM fungi when grown at high densities (F1,105 = 4.96, P <0.05; Table 1 & Fig. 3).

Fig. 3
figure 3

Plant average root diameter responses of C. jacea (a) and C. stoebe (b) and specific leaf area (log-transformed) of C. jacea (c) and C. stoebe (d) to drought, AM fungi and plant density. Drought treatments are indicated in grey, control indicated in green. Raw data are shown as points, whereas dashed lines are based on linear regressions

Full size image

Plant nutrient content

Centaurea stoebe plants showed a significant increase in leaf and root nitrogen (N) content to extreme drought (F1,110 = 4.23, P < 0.05; F1,110 = 8.73, P < 0.01, for leaf and root N, respectively). Increasing plant density had the opposite effect in C. stoebe resulting in declines in leaf and root N content (F1,112 = 5.45, P < 0.05; F1,112 = 28.64, P < 0.001, for leaf and root N, respectively; Table 1; Fig. 4) and subsequent increases in leaf and root C: N ratio (F1,111 = 7.10, P < 0.01; F1,111 = 30.48, P < 0.001, for leaf and root C: N, respectively; Fig. 4). Changes to these responses in combination with other treatment were not statistically significant (Table 1). In C. jacea, plants subjected to the extreme drought event only substantially increased their leaf and root N content the presence of AM fungi (leaf: F1,107 = 8.26, P < 0.01 and root: F1,111 = 7.58, P < 0.01) or when grown at high plant densities (leaf: F1,106 = 10.59, P < 0.01 and root: F1,109 = 7.61, P < 0.01; Table 1; Fig. 4).

Fig. 4
figure 4

Plant nutrient content responses of C. jacea (left) and C. stoebe (right) to drought, AM fungi presence and plant density. Drought treatment indicated in grey, control indicated in green. Leaf C: N ratio: a, b; Root C: N ratio: c, d. Raw data are shown as points, whereas dashed lines are based on linear regressions

Full size image

Mycorrhizal colonisation responses

Root colonisation in the AM fungal treatment averaged 13.7% (standard deviation (sd) = 7.9, min = 1.7%, max = 32.9%) and 21.3% (sd = 11.5, min = 0.6%, max = 54%) for C. jacea and C. stoebe, respectively, while all plants not grown in soil inoculated with AM fungi showed no root fungal colonisation. Root colonisation by AM fungi in C. jacea plants declined due to extreme drought or due to plant density (F1,47 = 8.96, P < 0.01 and F1,47 = 4.20, P < 0.05, for drought and plant density, respectively; Supplementary Fig. 5). Centaurea stoebe plants did not show any variation in their root mycorrhizal colonisation in response to either extreme drought or to increased plant density. We found no interactive effects of extreme drought and plant density on the root colonisation by AM fungi in both plants (Supplementary Fig. 5). Root colonisation by AM fungi associated with other response variables, but only in C. jacea. For instance, root colonisation was positively associated with root biomass (R2 = 0.12, P < 0.05) but negatively associated with SLA (R2 = 0.14, P < 0.01 and root N (R2 = 0.15, P < 0.05; Supplementary Fig. 6).

Principle component analysis

We found a significant variation in trait responses to density in both plant species (pseudo F1,63 = 13.34, P < 0.001 and pseudo F1,96 = 18.79, P < 0.001; for C. jacea and C. stoebe respectively, Fig. 5) as well as significant variation in trait responses to drought but only for C. jacea (pseudo F1,34 = 7.23, P < 0.001; Fig. 5). This suggests that trait responses in C. jacea were influenced by both drought and plant density treatments, while in C. stoebe, trait variability was primarily driven by plant densities (Fig. 5).

Fig. 5
figure 5

Principle component analysis (PCA) of plant responses to drought (red and orange colours = extreme drought, blue and azure = constant moisture) and intraspecific competition ((Highest Competition = 5 individuals competing, No Competition = 1 individual alone)) using plant trait variables. Arrows generated based on loading values

Full size image

Discussion

With increasing drought frequency and severity, it is important to investigate how biotic interactions influence plant drought tolerance and recovery, as this has important repercussions on the plant community composition and functioning (Walter 2018). In this study, we investigated how increasing plant densities (as a gradient of intraspecific plant competition) impacted the post-drought recovery of C. jacea and C. stoebe. We found that the biomass recovery of plants after rewetting following an extreme drought was constrained by intraspecific plant competition, although only true for the native resident C. jacea (Table 1; Fig. 2). For instance, increasing plant densities induced a strong negative effect on root biomass of C. jacea especially when subjected to drought, indicating how intraspecific interactions can modulate post-drought recovery (Table 1; Fig. 2). Moreover, density-constrained plant recovery of the resident native C. jacea was not ameliorated in the presence of AM fungi. Root trait responses, namely root diameter, provide insight into the underlying mechanisms contributing to the reduced drought recovery under high intraspecific competition within C. jacea compared to C. stoebe.

Intraspecific competition can intensify drought effects by restricting both the biomass accumulation in plants and their ability to spread in the soil, hindering their capacity to obtain sufficient water supply (Casper and Jackson 1997; Foxx and Fort 2019; Postma 2021; Rehling et al. 2021). Our results highlighted this, as one major difference between C. stoebe and C. jacea was that C. stoebe allocated substantially more biomass into root growth when grown in high densities at the expense of shoot growth (as indicated by increased root: shoot ratio; Table 1; Fig. 2). This shift in biomass allocation allows C. stoebe to still meet its nutrient and water requirements even in populations with high intraspecific competition. Such shifts in biomass allocation strategies at high plant densities have been reported previously (Ravenek et al. 2016; Rehling et al. 2021), and we suspect that this may have allowed C. stoebe to persist and recover after drought in their high density treatments (Table 1; Fig. 2).

We further suspect that C. stoebe utilized AM fungal symbiosis more effectively to recover after drought event compared to C. jacea, particularly at high intraspecific competition. For instance, in the presence of AM fungi alone, C. stoebe produced thicker roots (higher diameter), possibly to optimise the symbiosis with AM fungi as thicker roots are usually associated with high AM fungi colonisation (Table 1; Fig. 3; Wen et al. 2019; Bergmann et al. 2020). By contrast, in the presence of AM fungi and drought, C. stoebe produced thinner roots (low diameter) in soils. This effect of drought on root diameter of C. stoebe was, however, overturned at high plant density, where C. stoebe again had thicker roots (Table 1; Fig. 3). This demonstrates how plasticity in root trait responses allows C. stoebe to tolerate different biotic and abiotic stressors. Such plasticity in trait responses may enable range-expanding plants, such as C. stoebe to adapt to local conditions and thus promote their establishment even under drought stress (Usui et al. 2023).

The lack of root trait plasticity in C. jacea may have contributed to its greater vulnerability to extreme drought event in our study, particularly when intraspecific competition was high. This is further illustrated in our PCA, which showed variability in trait responses of C. jacea to both drought event and plant density, while trait variability in C. stoebe were mainly driven by plant densities (Fig. 5). These findings highlight the importance of a plants’ trait plasticity in adapting to changing environmental conditions and help us to better understand variability in plant recovery after extreme drought events (Berg and Ellers 2010; Thakur et al. 2022).

Despite some evidence for AM fungal-mediated benefits to C. stoebe after the drought in high intraspecific competition (Table 1), our results are less conclusive on the positive roles of AM fungi to foster plant recovery after extreme drought, at least with a single AM fungal species used in our study. Indeed, AM fungi has been well-studied for ameliorating the effects of drought on their host plants (Delavaux et al. 2017; Jayne and Quigley 2014). By contrast, in both plants we found indications that plants in symbiosis with AM fungi may have a disadvantage when subjected to adverse conditions, particularly high intraspecific competition. Both C. stoebe and C. jacea produced more aboveground biomass in AM fungal soils (Table 1; Fig. 1). However, for both plant species, these responses shifted from positive to negative in high plant density treatments, indicating that the negative impact of intraspecific competition topples any positive AM fungal effects (Table 1; Fig. 2). Such density-dependent reduction in AM fungal benefits for plants could be due to an increase in the cost: benefit ratio of mycorrhizal colonisation with plant roots, as competition for light increases with increasing plant density and photosynthetic ability declines (Koide and Dickie 2002). As such, plants may become more carbon limited than P or N limited and benefit less for mutualistic exchange of nutrients with AM fungi (Koide and Dickie 2002; Pérez and Urcelay 2009; Werner et al. 2018). Furthermore, this (plant) density dependent decline in mutualistic interactions between plants and AM fungi may have further implications on the ability of AM fungi to mitigate negative plant responses to droughts.

Although intraspecific competition strongly constrained the post-drought recovery of C. jacea (Table 1), their nutrient values, such as N content (leaf and root), showed a substantial increase after the drought event, particularly in the presence of AM fungi at high plant densities. Whether such shifts in N content were costly for their biomass recovery or if they would have fostered recovery, in the long run, remains to be tested. Both drought and plant density also negatively influenced mycorrhizal root colonisation in C. jacea (Supplementary Fig. 5). Reduced soil moisture can instigate the dieback of other beneficial microorganisms than AM fungi (Preece et al. 2019; Xu et al. 2020), and the accumulation of fresh litter (leaves and roots) from drought-stressed plants. Subsequent rewetting of the soil during the recovery period may have initiated rehydration and lysis of the dead microbial cells, as well as a boost in microbial activity (Borken and Matzner 2009; Brangari et al. 2021; Leitner et al. 2017). Rewetting of the soil can thereby create a temporary pulse of soil nitrogen due to increased accessibility of N through diffusive transportation and accelerated microbial activity and N mineralisation (Gao et al. 2020; Rennenberg et al. 2009). This increase in nutrient availability upon rewetting has been well established in other studies (Gao et al. 2020). Indeed, in C. stoebe, rewetting led to an increase in root nitrogen (Table 1; Fig. 4); however, in C. jacea, the benefits from rewetting appeared to be contingent on the extent of intraspecific competition and the presence of arbuscular mycorrhizal fungi (Table 1; Fig. 4; Supplementary Fig. 6).

In conclusion, we highlight how two closely related plants might have different strategies to recover from the drought when growing under intraspecific competition combined with interactions with mycorrhizal fungi. The results of our study are indeed limited in their application as we used only two plant species with a single species of AM fungi. Nevertheless, the AM fungi used in our study is one of the most studied mycorrhizal fungi, and both Centaurea species are commonly found in temperate grasslands, which makes our findings relevant for highlighting the importance of intraspecific root trait variation for understanding the recovery of grassland plants after extreme drought. Moreover, although our study found that AM fungi or intraspecific plant competition impacted the outcome of the drought treatment on both plant species, we did not differentiate if these plant responses were a direct response to the low soil moisture content or an indirect response via the shifts in soil microbial activity and nutrient availability during the rewetting period. We encourage future studies to consider indirect pathways, such as the co-response of other soil microorganisms in presence of AM fungi to be able to better explain how rewetting after the drought in presence of symbiotic and competitive interactions impact plant performance in grasslands.