Experimental microcosms consisting of three compartments were constructed using polypropylene plastic in the greenhouse of Forestry College of western campus in Guizhou university, Guiyang, China (106°22′E, 29°49′N, 1120 m above the sea level). The thickness of the material was 2 mm (Figure 1). Each microcosm was composed of two planting compartments on opposite sides and a competition compartment in the center. Five circular holes of 5mm diameter were drilled in the baffle plate between the planting compartments (Pc) and competition compartment (Cc). Depending on the treatment, nylon meshes of 20 μm or 0.45 μm were attached to either side of the baffle plate to form an air gap to prevent the flow of nutrients among compartments. The 20 μm nylon mesh can allow mycelium to pass through but not for plant roots, while 0.45 μm nylon mesh can neither allow mycelium nor roots to pass through (Yang et al., 2017). The size of each of the three compartments was about 10 cm × 10 cm × 10 cm (length × width × height).

The experiment was conducted using the mycorrhizal fungus treatments with or without C. etunicatum (purchased from the Institute of Nutrition Resources, Beijing Academy of Agricultural and Forestry Sciences, BGA0046) and the nutrient utilization treatments for competition compartment through mycorrhizal networks interconnecting two planting compartments, were constructed using meshes of 20 μm for mycelium connectivity and of 0.45 μm for no mycelium connectivity. The mycorrhizal fungus treatments included the inoculation with AM fungi (M⁺), and the control without AM fungi (M⁻); The nutrient utilization treatments were the following: (1) the common nutrient competition (common utilization, Cu) treatment in which the competition compartment was allowed to be connected to the planting compartments of E. adenophorum and A. annua via a mycorrhizal network (20 μm of double-nylon meshes were used on both sides of two baffle plates, allowing mycelium to grow freely from both side compartments into the competition compartment); (2) the single nutrient utilization (Su) treatment in which the competition compartment could be connected by mycelium to one planting compartment of E. adenophorum or A. annua only (20 μm of double-nylon mesh used on the one baffle plate, with 0.45 μm mesh on the other plate, allowing AM mycelium to pass through from one planting compartment to the competition compartment, but preventing in-growth from the other planting compartment); (3) the nutrient non-utilization (Nu) as a control (0.45 μm nylon meshes against the two baffle plates on both sides, to prevent the growth of AM mycelium into competition compartments). The mycelium and spore existed in competition compartment soil by checking under Cu and Su of M⁺ treatments, which showing that mycelium passed through planting compartment to competition compartment through mycelial expansion, and formed interconnective mycorrhizal networks between Pc and Ac.

A 2.5 kg mixture containing limestone soil and sand (3:1 by volume) was placed into each compartment; this mixture had first been sterilized at 0.14 Mpa at 126°C for 1 h. The soil, which had been collected from a typical karst habitat close to Guiyang city, had pH 7.45, total nitrogen (N) concentration 2.27 g kg^–1, available N 127.48 mg kg^–1, total phosphorus(P) 0.90 g kg^–1, available P 11.48 mg kg^–1, total potassium (K) 4.99 g kg^–1, available K 287.30 mg kg^–1, by the measurement refering to Tan (2005). The seeds of E. adenophorum and A. annua were collected from a karst mountain experiencing severe soil degradation in Guanling county, Guizhou province, China. Five seeds of each species were planted into the soil in their respective planting compartments and two plants per compartment were kept after germination. 50g of inoculum was added after seeding, consisting of spores, mycelium and plant roots. The inoculum had been propagated with Trifolium repens for 4 months which had been grown in limestone soil by the pre-sterilized treatment at 0.14 Mpa, at 126°C for 1 h before inoculation with C. etunicatum. Each treatment was repeated six times. Thus, the overall factorial experimental design contained three nutrient access treatments times two mycorrhizal treatments (M⁺, M⁻) times six replicates as 36 samples. From germination, the experimental plants were cultured in a greenhouse for 3 months and were arranged in a randomized block design in western campus of Guizhou university, then were harvested for measurement.

The determination of mycorrhizal colonization rate adopted methods described by He et al. (2012). The plant biomass of E. adenophorum and A. annua was determined by weighing root, stem and leaf material after drying at 80°C for 24 h. The determinations of N and P concentration of plant tissues followed the traditional Kjeldahl method and the Molybdenum-antimony anti-colorimetric method (Bao, 2000). The N acquisition and P acquisition were calculated by multiplying each plant nutrient concentration by each plant biomass.

Statistical analyses were performed using the SPSS 23.0 software. All of the data were tested for normality and homogeneity of variance before analysis. First, two- or three-way or three ANOVAs were applied to test for the overall effects of plant species (E. adenophorum, A. annua) mycorrhizal fungus (M⁺, M⁻), nutrient utilization treatments (Cu, Su, and Nu) and their interactions on plant biomass, N acquisition and P acquisition and N/P ratio, respectively. Then the single factor differences between Cu, Su, and Nu on biomass, N and P acquisition and the N/P ratio, respectively, were analyzed by the least significant difference (LSD) test, and a t-test was applied to compare the differences in N acquisition and P acquisition and the N/P ratio between treatments of M⁺ and M⁻. All results graphics were produced using the Origin 2018 software.

The mycorrhizal colonization rates of E. adenophorum were significantly higher than those of A. annua under Cu, Su and Nu treatments (Table 1). For the mycorrhizal colonization rates of E. adenophorum and A. annua, there were non-significant differences among the Cu, Su, and Nu treatments. Both species had the same pattern of mycorrhizal colonization rate as follows: Su > Cu > Nu.

For E. adenophorum seedlings, the mycorrhizal fungus treatments significantly affected the individual, root and leaf biomass but did not affect stem biomass (Table 2). Significantly larger biomass in M⁺ than in M⁻ was measured for whole, plant individuals under Nu, for root biomass under Cu, Su, and Nu and for leaf biomass under Su, while an opposite significant response (M⁺ < M⁻) was seen for stem biomass under Su (Figures 2A–D). The nutrient utilization treatments significantly affected stem biomass (Table 2); the stem biomass of Su was significantly lower than that of Cu and Nu under M⁺ treatment (Figure 2C). The interaction of M × U significantly affected stem biomass but did not significantly affect root, leaf or individual biomass (Table 2). On the other hand, for A. annua seedlings, the mycorrhizal fungus treatments significantly affected the stem biomass but did not significantly affect individual plant, root or leaf biomass (Table 2). A significant M⁺ > M⁻ effect was present in individual and stem biomass under Su treatment. Adversely (M⁺ < M⁻) affected individual biomass and leaf biomass under Cu treatment (Figures 2A,C,D). Nutrient utilization treatments significantly affected leaf biomass (Table 2). Under M⁺ treatment, the stem biomass under Cu and Su was significantly greater than under Nu treatment, and under M⁻ treatment, the individual and stem biomass under Cu and Su was also significantly greater than under Nu treatment, the root and leaf showed no significant difference among Cu, Su, and Nu treatments in this study (Figures 2A–D). The interaction of M × U did not significantly affect individual, root, stem and leaf biomass of A. annua (Table 2). Besides, the mycorrhizal fungus treatments significantly affected the individual biomass. However, the plant species and nutrient utilization were significantly affected seedling accumulations via three-way ANOVAs analysis, including their interactions (Table 6). Overall, the results indicated that AM fungi differentially increased the biomass accumulation of the invasive plant E. adenophorum and native plant A. annua, even though an adverse non-significant trend (M⁺ < M⁻) was observed in individual biomass of A. annua under the Cu treatment. After inoculation with AM fungi, the biomass of A. annua was greater than that of E. adenophorum under Su treatment. However, E. adenophorum biomass was even larger in comparison to A. annua under the Cu and Nu treatments. This result indicates that the biomass of E. adenophorum was higher than that of A. annua when E. adenophorum competed with A. annua for nutrients, but compared to E. adenophorum, when native A. annua utilized nutrients singly via its mycelium, it accumulated more biomass than the invasive E. adenophorum.

For the invasive plant E. adenophorum, the mycorrhizal fungus treatments significantly affected root N acquisition and did not significantly affect stem, leaf or individual N acquisition (Table 3). A significant M⁺ > M⁻ effect was observed in root N acquisition under Cu, Su, and Nu treatments, but a significant M⁺ < M⁻ effect was observed in stem N acquisition under the Su treatment (Figures 3B,C). The nutrient utilization treatments had a significant effect on root, leaf and individual N acquisition, but did not significantly affect stem N acquisition (Table 3). Stem N acquisition was observed as Cu > Su > Nu under the M⁺ treatment, while the N acquisition of individual, root and leaf was observed as Su > Cu > Nu under M⁺ and M⁻ treatments (Figures 3A–D). The interaction of M × U significantly affected stem N acquisition but did not significantly affect N acquisition of root, leaf or individual (Table 3). For the native plant A. annua, there were no significant effects of the mycorrhizal fungus treatments on N acquisition of root, stem, leaf or individual plant (Table 3). A significant difference between M⁺ and M⁻ treatments was observed in individual and stem N acquisition under Su, while M⁺ and M⁻ did not significantly affect root or leaf N acquisition under Cu, Su and Nu treatments (Figures 3A–D). The nutrient utilization treatments significantly affected stem, leaf and individual N acquisition, but the effect on root N acquisition was non-significant (Table 3). Under the M⁺ treatment, the root N acquisition was observed as Cu > Su > Nu, the individual, stem and leaf N acquisition was observed as Su > Cu > Nu; Under M⁻ treatment, the stem N acquisition was observed as Cu > Su > Nu, the individual, root and leaf N acquisition was observed as Su > Cu > Nu (Figures 3A–D). The interaction of M × U did not significantly affect N acquisition of root, stem, leaf or individual plant (Table 3).

For individual plants, under M⁻ treatment, the N acquisition of A. annua was higher than that of E. adenophorum under Su and Cu treatments. In addition, under M⁺ treatment, N acquisition of A. annua was also greater than E. adenophorum under Su but that of A. annua was lower than E. adenophorum under Cu treatment (Figure 3A). Furthermore, the mycorrhizal fungus treatments and the nutrient utilization treatments significantly affected individual plants N acquisition but not for plant species and their interactions via further three-way ANOVAs analysis (Table 6). This result suggests that AM fungi differentially enhanced the N acquisition for the invasive E. adenophorum and the native A. annua, even though a non-significant adverse effect (M⁺ < M⁻) was observed in individual N acquisition for the native A. annua under Cu treatment. After inoculation with AM fungi, when the native A. annua absorbed nutrients from the competitive compartment through it mycelium singly under Su treatment, the N acquisition was higher than that of E. adenophorum, but when both species had access to the competitive compartment through mycelium, the N acquisition of A. annua was lower than that of E. adenophorum.

For E. adenophorum, the mycorrhizal fungus treatments significantly affected P acquisition of root, stem, leaf and individual plant (Table 4). A significant difference between M⁺ and M⁻ was observed in individual, root, and leaf P acquisition under Cu, Su, and Nu treatments, and in stem P acquisition under Cu and Nu treatments (Figures 4A–D). The nutrient utilization treatments significantly affected stem, leaf and individual P acquisition (Table 4). Under M⁺ treatment, the individual and stem P acquisition under Cu was significantly higher than under Su and Nu treatment, and leaf P acquisition under Cu and Su was significantly greater than under Nu treatment; Under M⁻ treatment, the stem P acquisition under Cu and Su treatments was significantly lower than Nu, and leaf P acquisition under Su was significantly higher than Su and Nu treatments (Figures 4A,C,D). The interaction of M × U significantly affected P acquisition of stem and leaf (Table 4). For A. annua, the mycorrhizal fungus treatments significantly affected P acquisition of stem, leaf and individual plant (Table 4). A significant difference between M⁺ and M⁻ was observed in individual and stem P acquisition under the Su treatment (Figures 2A,C). The nutrient utilization treatments significantly affected leaf and individual P acquisition (Table 4). A significant Su > Nu was observed in individual P acquisition under the M⁺ and M⁻ treatments, and stem P acquisition in Cu and Su was significantly higher than in Nu under M⁺ and M⁻ treatments, and leaf P acquisition in Cu and Su was significantly higher than in Nu under the M⁻ treatment (Figures 4A,C,D). The interaction of M × U did not significantly affect P acquisition in root, stem, leaf or individual plant (Table 4).

For individual plants, under M⁻ treatment, the P acquisition of A. annua was higher than E. adenophorum under Su and Cu treatments, but the P acquisition of A. annua was lower than E. adenophorum under Nu treatment. In addition, under M⁺ treatment, A. annua was also greater than E. adenophorum under Su but was lower than E. adenophorum under Cu and Nu treatments (Figure 4A). Besides, the mycorrhizal fungus treatments and the nutrient utilization treatments significantly affected individual plants P acquisition, as well as the interaction of S × M, S × U, and S × M × U for plants P acquisition, but not for plant species (Table 6). It showed AM plays a crucial role in P utilization of two species, even though biomass was not affected significantly by species. The intrinsic differences were evident in the ability to accumulate biomass for these two species. These results illustrate that AM fungi significantly increased P acquisition of the invasive E. adenophorum, and differentially enhanced the P acquisition of the native plant A. annua. Moreover, when E. adenophorum competed with A. annua for nutrients (under Cu condition), the AM fungi enabled the invasive plant E. adenophorum to accumulate more P than the native plant A. annua.

Firstly, for E. adenophorum seedlings, the mycorrhizal fungus treatments significantly affected root, stem, leaf and individual plant N/P ratio (Table 5). A significant M⁺ < M⁻ effect was observed in individual plant, root, and leaf N/P ratio under Cu, Su, and Nu treatments, and in stem N/P ratio under Cu and Su treatments (Figures 5A–D). The nutrient utilization treatments significantly affected root, stem, leaf and individual N/P ratio (Table 5). The root N/P ratio was observed as Cu > Su > Nu under M⁻ treatment, but it was observed as Su > Cu > Nu under the M⁺ treatment, and the N/P ratio of individual, stem and leaf was observed as Su > Cu > Nu under M⁺ and M⁻ treatments (Figures 5A–D). The interaction of M × U significantly affected stem, leaf and individual plant N/P ratio (Table 5). Secondly, for A. annua seedlings, the mycorrhizal fungus treatments was significantly affected stem and individual plant N/P ratio, but did not significantly affect root or leaf N/P ratio (Table 5). A significant M⁺ < M⁻ effect was observed in individual N/P ratio under Cu, Su and Nu treatments, and in root N/P ratio under the Su treatment, and in stem N/P ratio under the Cu treatment (Figures 5A–C). The nutrient utilization treatments significantly affected stem and individual plant N/P ratio (Table 5). The N/P ratio of individual plants was observed as Su > Cu > Nu under M⁺ and M⁻ treatments, and the stem N/P ratio under Cu treatment was significantly greater than Su and Nu under M⁻ treatment (Figures 5A,C). The interaction of M × U significantly affected the stem N/P ratio (Table 5). Finally, for individual plant N/P ratio, under M⁻ treatment, the N/P ratio of invasive E. adenophorum was significantly higher than that of native A. annua under Su treatment; Under M⁺ treatment, the individual plant N/P ratio of invasive E. adenophorum was significantly lower than that of native A. annua under Cu and Nu treatments (Figure 5A). The mycorrhizal fungus treatments significantly affected individual plants N/P ratio, but not for plant species and mycorrhizal fungus through further three-way ANOVAs analysis (Table 6). Overall, AM fungi significantly reduced the N/P ratio of both the invasive plant E. adenophorum and the native plant A. annua.

AM fungi differentially increased the biomass, and the acquisitions of N and P for the invasive E. adenophorum and native plant A. annua in this experiment (Figures 2A, 3A, 4A). Previous studies demonstrated that soil microbes affected nutrient transformation and plant growth (Lugtenberg et al., 2013; Leff et al., 2015), for instance, AM fungi facilitated plants’ uptake of mineral nutrients by extending the spatial absorption scale and thereby promoting plant growth (Lin et al., 2017; Wang W. X. et al., 2017). He et al. (2007) revealed that the AM fungi promoted the accumulation of biomass, N and P of the host plants in karst soil. In this study our results broadly illustrate that AM fungi increases biomass, N and P acquisition through expanding the surface area of absorbed nutrients by mycelium in both a native and invasive plant species. Plant resource allocation patterns can reflect the adaptation of plant growth and development to environmental conditions (Aoyagi and Akimoto, 2010). In our study, both plant species accumulated more biomass, N and P in the stem and leaf than root (Figures 2–4), which indicating that both plants preferentially allocate resources to photosynthetic organ leaves and stems, and thereby producing more photosynthetic products for their growth (Saxena and Ramakrishnan, 2010). He and Han (2010) showed that the N/P ratio of plant leaves can be used to assess the nutrient status of soil. Plants with an N/P ratio that is less than 14 are thought to be mainly limited in growth by N, while plants with an N/P ratio greater than 16 are mainly limited by P, and plants with a ratio between 14 and 16 are co-limited by both N and P (Koerselman and Meuleman, 1996). In this study, the N/P ratio of the invasive E. adenophorum decreased from above 16 to below 14 via AM fungi under Cu and Su treatments (Figure 5D). This indicates that the growth of the invasive plant E. adenophorum may be limited by P in karst areas of southwest China, and that AM fungi alleviates this limitation. AM fungi have been shown to promote P uptake by plants, resulting in P no longer being the plant’s main limiting factor (Yang et al., 2017). This phenomenon is of great significance to plant growth in the low P areas of karst regions. In addition, Cu treatment symbolizing mycorrhizal networks connecting both species reduced the N/P ratio of invasive and native species in karst soil in this experiment (Figure 5A), and this is consistent with our hypothesis (H1). Invasive E. adenophorum had stronger P absorption than A. annua via interconnected mycorrhizal networks in this study (Figure 4A). Therefore, the N/P ratio of the invasive E. adenophorum was lower than that of the native A. annua. Plants in karst areas are characterized by low levels of P (Wen et al., 2018), and AM fungi can promote absorption by host plants of P in the soil (Zhang et al., 2018). Additionally, Hu et al. (2016) argued that typical alien invasive plants have higher P uptake in southwestern China, which is consistent with the results of this study. This means that E. adenophorum would be more invasive due to successful competition for nutrients in the fragile lower-phosphorus karst habitats.

When invasive plants competed with native plants, AM fungi tend to have a positive impact on invasive plants (Neuenkamp et al., 2019). Yu et al. (2014) argued that AM fungi enhanced the competitiveness of invasive E. adenophorum against native species. Higher mycorrhizal colonization rate gives a competitive advantage to invader over native species in nutrient resources (Zhang et al., 2017), which supporting that E. adenophorum had greater competitive advantage due to higher mycorrhizal colonization rate, when comparing with A. annua by Table 1. Our data indicated that when the invasive E. adenophorum competed with native A. annua for nutrients in karst areas, the AM fungi conferred the invasive plant E. adenophorum a competitive advantage in obtaining nutrients as compared to the native plant A. annua, and E. adenophorum had higher mycorrhizal colonization rate than A. annua. This was consistent with our hypothesis (H2). In the karst regions of southwest China, the AM fungi composition differed significantly between vegetation types (Liang et al., 2016), and the invasion of alien plants will shift the species composition and diversity of AM fungi, change the preference of AM fungi to the original host, and establish a symbiotic relationship favorable to the invasive plant (Vogelsang and Bever, 2009). Additionally, AM fungi can facilitate the invasion of alien species by decreasing the competitiveness of the native plants (Yang et al., 2014), and promoting the growth of invasive species in a nutrient-poor environment (Chen Q. et al., 2019), which might contribute to their successful invasion into karst habitats. Moreover, AM fungi conferred invasive plants with greater advantage in competition with native plants, and AM fungi had a negative impact on the growth of native plants (Batten et al., 2008; Zubek et al., 2016). Fellbaum et al. (2014) showed that mycorrhizal networks of AM fungi in the soil simultaneously provide multiple host plants with nutrients, and AM fungi can preferentially allocate nutrients to high-quality hosts. Yuan et al. (2014) argued that invasive plant may produce substances in the soil that inhibit native plants. It may explain our results that interconnected mycorrhizal network may give priority to nutrients to the invasive plant E. adenophorum, while biomass and N acquisition of the native plant A. annua decreased via an interconnected mycorrhizal network (Figures 2A, 3A). AM fungi only had a very significant effect on the root biomass accumulation and N and P acquisition of the invasive plant E. adenophorum, but had no any significant effect on the root of native plant A. annua (Tables 2–4). The root system of E. adenophorum has strong vegetative reproductive ability, and can occupy the soil habitat quickly, and its root system can enrich AM fungi in the rhizosphere and break the reciprocal symbiotic relationship between AM fungi and native plants (Jiang et al., 2014). Additionally, Hawkes et al. (2006) argued that invasive plants influence the availability of the mycorrhizal fungi network to native plants in soil via earlier root activity compared to native plants. Therefore, in the karst habitats of southwest China, earlier root activity of E. adenophorum compared to the native plant A. annua may be one of the important reasons for the former’s invasiveness. In summary, the invasive E. adenophorum can successfully invade in the karst soil, possibly through a competitive advantage in nutrient uptake in comparison to native plants, and the effect of AM fungi on the invasiveness of E. adenophorum may be mediated by earlier root activity.