Two-way ANOVA was performed to test for effects of AMF and biochar addition on N₂O emission, gene abundance, root traits and soil properties (P ≤0.05) at low or high P supply. Post-hoc Tukey HSD tests were performed to determine significant differences. If the two-way interaction AMF× BC was significant, the complete data was shown; if not, only the significant main effects were shown. Significant differences between the AMF and no AMF inoculation at each P supply or between BC and no BC addition at each P supply were based on the t-test (P ≤0.05). Pearson’s correlation analysis was used to test the relationships among the N₂O emission, abundance of genes, root traits and soil properties. The random forest model in R studio software was used to analyze the comprehensive effects of genes, root traits and soil properties on the soil N₂O emissions.

Two-way ANOVA was performed to test for effects of P and biochar addition on yield, N₂O emission and gene abundance (P ≤0.05) in the field experiment. Post-hoc Tukey HSD tests were performed to determine significant differences. Pearson’s correlation analysis was used to test the relationships among the yield, N₂O emission and gene abundance.

The interaction AMF × BC was not significant for shoot biomass and shoot N and P contents ( Table 1 ). The shoot biomass was influenced significantly by mycorrhiza at both P rates, but the biochar main effect was a significant source of variation for shoot biomass at P300 only ( Table 1 and Figure 1 ). The mycorrhizal main effect was a significant source of variation for shoot N content and shoot P content at both P40 and P300 ( Table 1 and Figure 1 ). In the presence of mycorrhiza, shoot biomass and shoot N content were significantly lower than in the non-mycorrhizal treatment regardless of the P rate. By contrast, AMF addition was associated with a decrease in shoot P content at P40 and an increase at P300 ( Figure 1 ). Compared to the low P supply, the high P significantly promoted shoot growth (by 16-34%), shoot N content (by 33-47%) and shoot P content (by 145-218%) (calculated from Figure 1 ). In the field experiment, the interaction BC×P was significant for wheat yield ( Table 2 ). Compare to no P addition treatment, wheat yield increased significantly (by 33% and 41%, respectively) in P addition treatments.

Root biomass and root tissue density were significantly influenced by the interaction AMF×BC at P40 and P300 ( Table 1 ). At the low P supply, compared to the control, root biomass decreased significantly (by 30% to 48%) in the other treatments ( Figures 2A, B ). Regarding root tissue density, contrasting results were obtained at two P supplies. Compared to the other treatments, root tissue density was more than 2-fold lower in the combined AMF and BC treatment at P40 and in the control at P300 ( Figure 2B ). Root tissue density had a significantly negative correlation with root diameter in both P40 and P300 ( Table S2 ).

The AMF×BC interaction was a significant source of variation for total root length and specific root length at P40 only ( Table 1 ). In the combined AMF and BC treatment, total root length was 36% lower, and specific root length was 8.4% higher, than the control at P40 ( Figure 2C ). At the high P supply, both main effects significantly influenced total root length and specific root length ( Table 1 ). The total root length and specific root length decreased significantly (by 32% and 65%, respectively) with mycorrhizal inoculation. Biochar addition decreased total root length by 40% and specific root length by 22% ( Figures 2C, D ).

Leaf Mn concentration was significantly influenced by the interaction AMF×BC at P300 only ( Table 1 ). Compared to the control, Mn concentration was decreased significantly in the combined AMF and BC treatment at P300 ( Figure 2E ). At the low P supply, both main effects significantly influenced Mn concentration ( Figure 2E ) that increased by 10% with biochar addition and decreased by 17% with mycorrhizal inoculation.

The mycorrhizal colonization was significantly influenced by the AMF × BC interaction ( Table 1 ). No root colonization was detected in the treatments without AMF addition. AMF colonization tended to be higher at P40 than P300. In the given P treatment ( Figure 2F ) biochar addition increased AMF colonization by 27-33% compared to the AMF treatment ( Figure 2F ).

The soil N₂O emissions was significantly influenced by the AMF×BC interaction at P40 only ( Table 1 ). At the low P supply, compared to the control, the soil N₂O emissions decreased significantly (by 88%, 76% and 93%) in, respectively, the AMF, BC and the combined AMF and BC treatments ( Figure 3A ).

The biochar main effect was a significant source of variation for soil N₂O emissions at P300. Compared to control, the soil N₂O emissions decreased significantly (by 30% and 24%, respectively) with biochar addition and mycorrhizal inoculation. Compared to the low P supply, the high P significantly suppressed soil N₂O emissions by 50-95% (calculated from Figure 3A ).

In the field experiment, the soil N₂O emissions was significantly influenced by P and biochar addition ( Table 2 ). Compared to no biochar and no P environments, biochar and P addition reduced soil N₂O emissions by 32% and 39%, respectively ( Figure 3B ).

The AMF×BC interaction was a significant source of variation for soil NO₃ ^--N at P40 only ( Table 1 ). The mycorrhizal main effect was a significant source of variation for soil NO₃ ^--N at P300. At the low P supply, the soil NO₃ ^–N was 48% higher with mycorrhizal inoculation than the control, but decreased by 27% in the presence of mycorrhiza compared to the non-mycorrhizal treatment at P300 ( Figure S1A ).

The interaction AMF×BC interaction was not significant for the soil NH_4⁺-N ( Table 1 ). The soil NH₄ ⁺-N was significantly influenced by mycorrhiza and biochar addition at both P rates. Biochar addition was associated with an increase in the soil NH_4⁺-N at P40 and P300. Mycorrhizal inoculation was associated an increase at P40 and a decrease at P300 ( Figure S1B ).

The abundance of nirS gene was significantly influenced by mycorrhizal addition at P40 only, but the biochar main effect was a significant source of variation for nirS gene abundance at both P rates ( Table 1 ). At the low P supply, nirS gene abundance decreased significantly (by 25%-26%) with biochar addition or mycorrhizal inoculation, but abundance of nirS gene increased by 19% with biochar addition at P300 compared to the no biochar treatment ( Figure 4A ). The AMF×BC interaction was a significant source of variation for abundance of nirK gene at P300 only ( Table 1 ). Compared to the control at P300, the nirK gene copy number decreased significantly (by 35% to 59%) in the other treatments ( Figure 4B ). In the field experiment, P addition decreased significantly the abundance of nirS and nirK genes by 17% and 45% compared to no P addition, respectively ( Table 2 ).

The nosZ gene abundance was significantly influenced by the AMF×BC interaction at P40 only, whereas at 300 P both main effects significantly influenced nosZ genes ( Table 1 ). Compared to the control at P40, the nosZ gene abundance increased significantly (by 84% to 96%) in the other treatments. at the high P supply, the number of nosZ gene copies increased significantly (by 37% and 65% with biochar addition or mycorrhizal inoculation, respectively) ( Figure 4C ).

The nosZ/(nirK+nirS) ratio was significantly influenced by the interaction AMF×BC at P40 and P300 ( Table 1 ). Compared to the control, the nosZ/(nirK+nirS) ratio increased significantly (by 115% to 256% at P40 and 78% to 102% at P300) in the other treatments ( Figure 4D ). The nosZ/(nirK+nirS) ratio at high P level was 4.06 to 7.56 folds higher than at low P level (calculated from Figure 4 ).

In the field experiment, the nosZ gene and the nosZ/(nirK+nirS) ratio were significantly influenced by the BC×P interaction. Compared to the BC0P0 treatment, the nosZ gene and nosZ/(nirK+nirS) ratio increased significantly by 43% and 226% in the BC+P+ treatment. The nosZ gene and nosZ/(nirK+nirS) ratio increased significantly by 35% and 172% with P addition compared to no P addition treatment ( Table 2 ).

At the low P supply, N₂O emission was correlated positively with shoot growth, root growth and nirS gene abundance. The nosZ gene copy number, nosZ/(nirK+nirS) ratio, soil NO₃ ^–N, NH₄ ⁺-N, SOC and AMF colonization were correlated negatively with N₂O emission. At the high P supply, the nosZ abundance and nosZ/(nirK+nirS) ratio were correlated negatively with N₂O emission, whereas shoot growth, root morphology, nirK copy number and soil NO₃ ^–N were correlated positively with N₂O emission ( Figure S2A ). In the field experiment, N₂O emission was correlated positively with nirK and nirS genes, and was correlated negatively with wheat yield, nosZ and nosZ/(nirK+nirS) ratio ( Figure S2B ).

When the data from the two P treatments of greenhouse experiment were combined across 25 variables, the abundances of nirk and nirS genes were correlated positively with N₂O emission, but nosZ/(nirk+nirS) ratio showed a negative correlation with N₂O emission. Root growth was correlated with N₂O positively, and shoot growth and AMF colonization were correlated negatively. The mature leaf Mn content was correlated positively with N₂O. Additionally, Olsen P and soil NH₄ ⁺-N showed a negative correlation with N₂O ( Figure 5A ).

Most important variables to influence soil N₂O emissions based on the random forest model were ranked in order of importance ( Figure 5B ). The nosZ/(nirK+nirS) ratio, soil NH₄ ⁺-N, nirS abundance and mature leaf Mn concentration were the significant factors associated with N₂O emission. Shoot P and N content, root/shoot ratio, and nosZ copy numbers were the additional significant factors influencing N₂O emission ( Figure 5B ).

​In summary, the addition of biochar and/or AMF had a negative effect on the total root length for both P40 and P300. The AMF addition decreased leaf Mn concentration, whereas biochar addition increased leaf Mn concentration at P40, suggesting biochar and AMF addition promote P uptake. Compared with P40, the high P significantly promoted wheat growth (16-34%), nutrient content (33-218%) and yield (33-41%), but inhibited soil N₂O emissions (32-95%). However, biochar and/or AMF addition reduced soil N₂O emissions by 24-93% and 32%, respectively, in greenhouse and field experiments. This decrease was associated mainly with the diminished abundance of nirK and nirS (17-59%) and the increased nosZ (35-65%) in the greenhouse and field experiments, respectively.

In the present study, the results showed that the wheat root and shoot biomass growth and yield were all higher at the high than low P supply regardless of the biochar or AMF additions ( Figure 1 ), suggesting the inhibiting effect of low P on plant growth. In general, AMF has a negative effect on plant growth in P-rich soils, whereas it promotes nutrient acquisition in the low-P soils (Johnson et al., 2015). However, in the greenhouse experiment, AMF additions (AMF added alone or in combination with BC) reduced biomass and nutrient content at both low and high P rates ( Figure 1 and Table 1 ). Compared to the control, AMF addition significantly decreased shoot biomass and P content at the low and high P supply ( Figure 1 ). AMF inoculation could inhibit the root nutrients absorption in the high nutrient environment, while maintaining the contribution of AMF absorption pathway, finally, this negative effect influenced the crop growth (Smith et al., 2011). The root length decreased significantly by mycorrhizal inoculation in the present study ( Figure 2 ), hinting a trade-off between AMF and root length in absorption of available P (Li et al., 2019; Zhang et al., 2023). We also found that leaf Mn concentration was significantly reduced by AMF inoculation as well, especially in low P environment ( Figure 2E ). Leaf Mn concentration has been proposed as a signature for the strength of carboxylates exudation in the rhizosphere (Lambers et al., 2021). It means that mycorrhizal inoculation reduced the rhizosphere carboxylate releasing of wheat root. Similar trend was also found in the previous work on Kennedia species that the AMF inoculation reduced exudation of organic carboxylates up to 50% (Ryan et al., 2012). However, the reduction in the release of carboxylate may decrease the ability of plants to utilize insoluble phosphorus (Lambers et al., 2008). As a thin-root species, wheat mainly depends more on the root system in P acquisition compared with thick-root species (Wen et al., 2019). Therefore, AMF inoculation may suppress wheat growth through inhibiting the root growth and exudation, and then bring a negative effect on wheat growth.

In the different P environments, adding biochar alone had no significant effect on shoot N and P content in wheat shoots in the greenhouse ( Figure 1 ), but biochar and P addition could promote wheat yield in the field ( Table 2 ). Biochar addition could significantly reduce root biomass, total root length ( Figures 2A, B ), which suggested that biochar addition can diminish carbon partitioning to roots. Our findings in previous field experiment also indicated that biochar addition decreased assimilate partitioning to rice roots at elongation stage (Liu et al., 2021). That might have been attributed to biochar addition improving root activity (Cao et al., 2019). For instance, our results showed that biochar increasing leaf Mn concentration ( Figure 2E ). This indicated that the addition of biochar could increase the release of carboxylate. We also found that biochar addition can enhance the AMF colonization ( Figure 2F ), which may be caused by biochar can change soil nutrient availability (such as P available, Figure S1 ) and be a refuge for colonizing fungi (Warnock et al., 2007).

Meanwhile, our results showed a negative relationship between root tissue density and root diameter in both low and high P environments ( Table S2 ). This may be because the increase in root diameter was mainly driven by the high thickness or proportion of root cortex (Kong et al., 2019). Due to the cortex density was significantly lower than stele density in root structure, suggesting that an increase of cortex thickness meant a decrease in root tissue density (Kong et al., 2019). Meanwhile, the higher proportion or thickness of cortex can provide more niche space for AMF colonization and facilitate higher mycorrhizal colonization intensity (Kong et al., 2017). In the present study, AMF colonization was the highest, while RTD decreased under AMF and biochar combined treatment at the low P supply ( Figures 2B, F ), reflecting a integrated variation of root morphology and AMF in plant P-acquisition strategy.

In both greenhouse and field experiments, N₂O emission flux was much higher at the low P supply compared to the high P supply ( Figure 3 ), which was consistent with the results reported elsewhere (Shen et al., 2021; Shen and Zhu, 2022). The suppression of soil N₂O emissions by P addition can partly be attributed to the improved shoot N content and thus decreased availability of N substrates (Xiao et al., 2022a). In the present study, the shoot N content and yield at the high P supply was indeed significantly higher at high compared with low P supply ( Figure 1 ), providing a support for the above viewpoint.

In addition, compared to low P supply, the abundance of nirK and nirS genes associated with N₂O production decreased by 17% to 59%, but the nosZ genes associated with N₂O consumption (N₂O→N₂) in denitrification were increased by 34% to 1340% at the high P supply, which result in the nosZ/(nirK+nirS) ratio at the high P supply was 2.72 to 7.56 times greater than that at the low P supply ( Figure 4 and Table 2 ), suggesting a strong influence of P addition on N cycling genes involved in denitrification (Xiao et al., 2022b). In the present study, among the top five factors affecting the N₂O emission, nosZ/(nirK+nirS), nirS, leaf Mn concentration and stem P content were closely related to soil Olsen P across all the treatments ( Figure 5B ). Therefore, these findings suggest that the variation of soil P environment dominated N₂O emission by altering proportion of functional microbes in the total microbiome, nutrient acquisition strategy, and plant growth.

As Figure 3 shown, AMF addition significantly decreased N₂O emissions, especially at the low P supply. There was no evidence of a positive effect of AMF on crop growth in the present study ( Figure 1 ); therefore, the suppressed effect of AMF on N₂O emission may be not via promoting plant N retention and reducing availability of soil N for N₂O production. Nevertheless, we found that AMF decreased the copy number of nirK genes and increased nosZ genes, thus reducing N₂O emission by regulating the denitrification process ( Figure 4 ). Zhao et al. (2021) and Storer et al. (2018) both reported that AMF could directly influence denitrification and reduce N₂O emission, which was consistent with the results in this study.

Biochar application had no significant positive effect on the shoot N and P content regardless of the P supply in greenhouse experiment ( Table 1 ). However, a decrease in N₂O emission by biochar addition was likely associated with a decreased copy number of nirS genes at the low P supply and nirK genes at the high P supply, and with the increased abundance of the nosZ gene to modify the denitrification process ( Figure 4 ). For instance, N₂O emission had a negative correlation with nosZ gene and positive correlation with nirK genes in the field experiment ( Figure S2B ), which was consistent with our previous research in the field (Liu et al., 2020). In addition, biochar application could significantly reduce P availability in soil, especially in low P environment, which might bring about P deficiency stress, and then reduced activity of denitrifying microorganisms led to decreased soil N₂O emissions (Wang et al., 2022).

Because biochar enhanced AMF colonization ( Figure 2F ), AMF and biochar showed a significant interaction on N₂O emission at the low P supply ( Table 1 ). The inhibition effect of AMF plus biochar application on N₂O emission was stronger than that of biochar application alone ( Figure 3A ). This may be attributed mainly to a decreased copy number of nirS genes in the either AMF or BC treatment. In summary, the effect of AMF and biochar on suppressing N₂O emission was mainly via regulating denitrification process rather than reducing substrates for denitrification by promoting plant growth.