The diseased leaf samples of weed plants were collected from agricultural areas of Qinghai region, China. The processes of fungi separation and purification were based on the protocols mentioned by Fang (1998) [11]. Briefly, the infected sections of the tissues were cut into 1.0 × 1.0 mm fragments. These tissue sections were soaked in 3% sodium hypochlorite solution for 3 min for surface disinfection, and then rinsed 3 times with sterile water. The redundant water was removed with sterile absorbent paper towels. Each fragment was placed on potato dextrose agar (PDA) medium and cultured in a dark incubator at 25˚C. Four tissue blocks were placed on each plate. Mycelia appeared on the media surface 4 days after inoculation. The agar blocks were excised from the mycelial edges in the plates and transferred to new PDA plates for obtaining pure cultures of the fungus. The strains of fungus were coded and stored at 4˚C for future testing.

Agar blocks from a 7-day-old fungi mycelia colony were transferred into a 500 mL flask with 100 mL of potato dextrose (PD) broth. Cultures were incubated at 25˚C with shaking at 180 rpm for 7 days. The culture solution of fungi strains was passed through the microporous membrane (Φ = 0.45 μm) under vacuum to obtain the culture filtrate.

Twenty-five seeds of A. fatua and B. juncea were placed on a filter paper in each of four Petri dishes each treatment, respectively. 3.0 mL of culture filtrate of each fungus was added to each of four Petri dishes consider as replicates. Petri dishes arranged with the same number of seeds were treated with 3.0 mL culture liquid and distilled water as control. Dishes were incubated at 25˚C with photoperiod of 12 h. The count was made on the 5^th day. The inhibition of seeds germination was measured as the percentage reduction of plumule and radicle length and germination rate in the treatment versus the germination growth in the control with sterile water. The herbicidal activities of the fungal strains were measured by the following formula (1):

Germinationinhibition ( % ) = Quantityofseedgerminationinthecontrol − Quantityofseedgerminationinthetreatment Quantityofseedgerminationinthecontrol × 100 (1)

Herbicidal activities of the culture filtrate were determined by leaf punctured assay. Detached leaves of weeds plant grown pierced using a sterile needle. Droplets (10 µL) of each filtrate were applied to each of these positions (2 punctures per leaf, 3 leaves per filtrate). After drip application, the leaves were kept at 25˚C in the sterilized moist chambers with photoperiod of 12 h. The sterile water was used to serve as a control. The diameter of lesions and symptoms was observed after 72 h.

Thirty milli-liter of culture filtrate was applied on 3 - 4 leaf stage weed plants. The inoculated weed plants were placed into the greenhouse at 25˚C with photoperiod of 12 h. Each treatment was repeated four times, and the sterile PD inoculated plants were used as controls. The following characteristics were measured: disease incidence, disease index, plant height (cm), and fresh weight (g). Disease incidence and index was calculated according to the previous report (Frans et al., 1987) [12]. The fresh weight was calculated for each replication using the formula (2):

Thefreshweightcontrolefficiency ( % ) = Freshweightofcontrol − Freshweightoftreated Freshweightofcontrol × 100 (2)

To evaluate the safety effect of fungi to crops, we selected five main cultivated crops, i.e. wheat, barley, oil rapeseed, pea and broad-bean. A volume of 30 mL of culture filtrate were sprayed on a pot of 3-4 leaf stage of the test crops. The same volume of sterile PD broth was used as control. Crop safety determination of each replicate was evaluated by measuring the growth inhibition rate and fresh weight inhibition rate 1 month later.

A volume of 50 mL of GD-2 and PA-2 culture filtrate was applied on both weeds, respectively. Control was performed using sterile water. Control and treated weed plants were kept in chambers at 25˚C under a 12 h photoperiod followed by 10 h in the dark. Analyzed sample were gathered for each treatment at 24, 48, 72, 96 120, and 144 h from the beginning of the exposure. Each treatment consisted of three replicates.

Superoxide-dismutase (SOD) activity was measured by the means of Ahn and associates (2005) [13]. Peroxidase (POD) activity was determined with the guaiacol method (Mäkinen and Tenovuo, 1982) [14]. According to the method of Hegab and associates (2013) [15], catalase (CAT) activity was determined to be a decrease in absorbance at 240 nm due to consumption of H₂O₂. Ascorbate-peroxidase (APX) activity was determined using the procedure of Nakano and Asada (1981) [16]. Polyphenol oxidase (PPO) activity was assayed according to Inoue and associates (Inoue and Izumi, 2020) [17]. The activity of phenylalanine ammonia lyase (PAL) was determined according to the change in absorbance of trans-cinnamic acid at 290 nm based on the description of Hegaband associates (2013) [15].

Chlorophyll content in leaves was determined as described previously (Jiang, et al., 2008) [18]. The level of lipid peroxidation was determined based on the Malondialdehyde (MDA) content, which was measured by using the 2-thiobarbituric acid (TBA) method (Mutlu, 2011) [19]. Membrane permeability was determined as described by Hassan et al. (2014) [20]. Soluble sugar content was measured by the report describedby Gao and associates (Gao et al., 2019) [21]. Soluble protein content was determined by Coomassie Brilliant blue G-250 method (Sedmak and Grossberg, 1977) [22].

GD-2 and PA-2 were identified by morphological and conservative gene cloning analysis. Morphological characteristics were based on the protocols described by Wei (1979) [23]. DNA was extracted from GD-2 and PA-2 mycelia by using Ezup kit (Sangon Biotech (Shanghai) Co., Ltd., China). The internal transcribed spacer (ITS) region of ribosomal RNA gene was amplified using the fungal universal primers ITS4 (5’-TCCTCCGCTTATTGATATGC-3’) and ITS5 (5’-GGAAGTAAAAGTCGTAACAAGG-3’). The PCR amplification system (50 μL) consisted of 4.0 μL dNTP mixture (2.5 mmol∙L⁻¹ each of dATP, dTTP, dCTP and dGTP), 1.0 μL each of ITS4 (20 μmol∙L⁻¹) and ITS5 (20 μmol∙L⁻¹), 5.0 μL of 10× Taq reaction buffer, 1.0 μL Taq DNA polymerase (2.5 U∙μL⁻¹), 2.0 μL template DNA, and 36.0 μL of dd H₂O. PCR conditions were 94˚C for 5 min, followed by 35 cycles of 94˚C for 1 min, 51˚C for 1 min, and 72˚C for 1 min; and a final extension at 72˚C for 10 min. After amplification, products were separated by electrophoresis on 1.0% agarose gel for 1 h, and then the gel was stained with ethidium bromide solution (0.5 μg∙mL⁻¹) for 15 min. The PCR products were imaged by a Bio-Rad GelDoc 2000 imaging system, and sequenced by the Sangon Biotech (Shanghai) Co., Ltd., China. The DNA sequences were recalled by the software Chromas version 2.6.6 (Technelysium Pty Ltd., Australia). The homology of the ITS sequences was searched in the NCBI fungal DNA databases. The phylogenetic tree of GD-2 and PA-2 and related fungal species was constructed for their ITS sequences by the neighbor-joining method using MEGA 5.0 (Tamura et al. 2011) [24] and the strength of the branches of the tree was judged with the bootstrapping analysis method at 1000 replicates.

Preliminary experimental data and charts are processed in Microsoft Excel. Data difference significance was analyzed by one-way ANOVA (P < 0.05) in SPSS 20 software and labeled by the alphabetic method.

In this collection, a total of 68 fungi were isolated from different weed plants in eight locations of Qinghai area. A list of fungal isolates and their sources is given in Table 1, in which fungi of different genera are isolated from different plants.

To screening efficient fungi in control A. fatua and B. juncea, 68 strains from eight collection areas were rated for their potential as biocontrol agents against these two weeds by using seed germination assessment. Most fungal treatments affected the weed seeds germination with the range of from 18% to 88% (Figure 1). Mean inhibition rate of A. fatua was 35.2%, while B. juncea the average inhibition rate was 36.4%. Germination inhibition rate of GD-2, GD-5, XN-7, HZ-1, HZ-31, HL-1, and PA-2 ranged from 63% to 82% for A. fatua, 55% - 80% for B. juncea. All treatments reduced A. fatua plumule lengths by 18.5% - 87.1% and radicle lengths 20.9% - 89.8% by of the control (Figure 2). Plumule and radicle length with PA-2 treatment were reduced by 87.1% and 89.8% for both parameters, respectively, compared with control. Reductions with GD-2 treatment were by 86.7% and 89.1% for both parameters, respectively. GD-5, XN-7, HZ-1, HZ-31, and HL-1 have also presented higher reductions on plumule and radicle length of A. fatua. Likewise, above treatments were the most toxic to B. juncea and also caused the higher reductions in seed growth (Figure 3). These seven strains GD-2, GD-5, XN-7, HZ-31, HL-1, HZ-1 and PA-2 which showed higher herbicidal activity to these two weeds in vitro were selected for next experiments.

Effect on A. fatua leaves was more marked when the culture filtrate from GD-2 was applied (Figure 4). Maximum damage on B. juncea leaves was obviously observed with culture filtrate from PA-2. For GD-2 treatment, the lesions infected 80% of the leaf area. The symptoms were the yellow macula and halos of chlorosis on the surface of A. fatua leaves for 3 days, and formed dead zone at 5 days. For PA-2 treatment, the lesions affected 92% of the foliar area. The symptoms were brown spots and rot area on the surface of B. juncea leaves at 3 days and spots gradually joined into patches at 5 days.

The highest disease incidence (86%) of A. fatua was reach when the culture filtrate from GD-2 was sprayed (Table 2). Approximately 21% reduction in the height was observed for this treatment compared with the control. Significant differences were observed for the fresh weight, which ranged from 18.54 to 34.86 g.

Disease incidence of B. juncea ranged from 56% to 89% and was increased with PA-2 treatment by 89% compared with the control (Table 3). For this treatment, plant height was reduced by 36.55% and fresh weight by 33.76% of the control, respectively.

Growth of oil rapeseed was most reduced with HZ-31 treatment and by 30.5% of the control and followed by pea by 22.8% (Figure 5(a)). However, growth of the same crop was no significant inhibited by GD-2, GD-5, and XN-7 treatments. On the whole, culture filtrates of HZ-31, HL-1, HZ-1, and PA-2 have greater effects

on crop growth than that of GD-2, GD-5, and XN-7. Reduction in fresh weight ranged between 7.4% and 24.6% of the control (Figure 5(b)). The maximum fresh weight of oil rapeseed inhibition rate (24.6%) was reached when the culture filtrate from HZ-31 was sprayed. HL-1 culture filtrate reduced fresh weight of oil rapeseed by 22.3% and HZ-1 to oil rapeseed (21.3%). In short, HZ-31, HL-1, HZ-1, and PA-2 were the most toxic to all tested crops, resulting in a greater fresh reductions, while GD-2, GD-5, and XN-7 were least toxic to all tested crops, especially broad-bean, pea and oil rapeseed.

The activity of SOD reached its maximum in the treatment with both strains 96 h after inoculation (Figure 6(a)). PPO activity in the treatment with both strains was significantly lower in comparison to control throughout the experiment

(Figure 6(b)). Comparison of CAT, POD, and PAL activity at different time points indicated that the treatments had lower levels compared to control (Figures 6(c)-(e)). The highest APX activity was observed after 144 h of treatment when it was 102% and 167% above control, respectively (Figure 6(f)).

Generally, the chlorophyll content of both weeds decreased gradually with increase in treatment time (Table 4). Treatments resulted in a significant decrease of chlorophyll contents compared to control within 144 h of application. The MDA content gradually increased with increase in processing time and was reached at maximum at 144 h (Figure 7(a)). There was no significant difference in MDA content between treatment and control in the early growth stages of both weeds. However, the MDA content in treatment and control was significantly difference after 48 hours of weeds treated. For relative conductivity, they were increased by using the culture filtrate and reached the highest at the 96 h (Figure 7(b)). As the treatment time increases, the soluble sugar content of each treatment increases accordingly (Figure 8(a)). There was no significant difference

in soluble sugar content in all treatments during the early growth stages of weeds. However, after 96 hours of weed growth, the soluble sugar content in treatments was significantly higher than those in the control. The soluble protein content first showed an increase followed by a decreasing pattern (Figure 8(b)).

The centre of the GD-2 colony is slightly protruded, and the surface is cotton flocculent; the mycelium grows densely, and the surface of the substrate is amaranth, followed by an earthy yellow, the conidia are fusarium-shaped, slightly curved, tapering at both ends, 3 - 7 segregates, most of which are 3 segregated, with a size of (24.0 - 70.0) µm × (3.5 - 4.0) µm.

The colony of PA-2 on PDA plate was dense, dirty white and viscous at the beginning and dark green around it until black, with the same positive and negative color. The edges are irregular, with obvious roots, folds, leather-like, growing close to the medium; the mycelia are black-brown, septate, rosary-like, sparsely branched and conidia are conidial; the mycelia in the base are developed, and the aerial hyphae is undeveloped and not easy to pick.

538 bp and 560 bp fragment of the ITS region from the genomic DNA of GD-2 and PA-2 was amplified with the fungal ITS primers, respectively. The phylogenetic tree constructed with the ITS sequences from GD-2, other strains of F. avenaceum and other related fungal species showed that GD-2 belongs to a clade with three strains of F. avenaceum (KY272780.1, KY272764.1, EU255801.1) supported by a 100% bootstrap value. Similarly, PA-2 belongs to a clade with seven strains of A. pullulans supported by a 100% bootstrap value. Based on the morphological characteristics and ITS sequences, the strain GD-2 and PA-2 was identified as F. avenaceum and A. pullulans, respectively (Figure 9).