The cow manure was dried and drained, the moisture content of which was controlled to below 80%, and then different botanical oil meals were added (soybean meal, rape meal, peanut bran, and sesame meal). The ratio of cow manure to different plant oil meals was 7:3 (mass ratio), and a small amount of organic fertilizer fermentation Effective Microorganisms bacteria was added. The water content was controlled at 70%, the compost was mixed and stirred evenly, pile fermentation, until the temperature reached above 60°C, and was mixed after 48 h. After seven days for one pile and four times of piling, the cow manure and botanical oil meal were fully decomposed and the cow manure–botanical oil meal organic fertilizer was prepared.

The test area was located in the tobacco-producing area of Jingxi City, Guangxi Province (105°56’E, 22°51’N). A single-factor completely random block design was used for a total of six treatments (CK: no fertilization; NF: administration of 1500 kg/hm² of cow manure; NFC: application of 1500 kg/hm² of cow manure and rapeseed bran organic fertilizer; NFD: application of 1500 kg/hm² of cow manure and soyabean meal organic fertilizer; NFHS: application of 1,500 kg/hm² of cow manure and peanut bran organic fertilizer; NFZM: application of 1,500 kg/hm² of cow manure and sesame meal organic fertilizer). The main nutrient contents of various organic fertilizers are shown in Supplementary Table 1. Each process had three replicates for a total of 18 blocks. Before transplanting, the fertilizer was used as the base fertilizer. After the fertilizer was spread, the land was ploughed to mix the fertilizer evenly with the soil, and then the K326 tobacco seedlings with vigorous growth and developed root systems were transplanted to the field. At the time of tobacco harvesting, 10 representative tobacco plants were randomly taken from each plot, and the plant height, leaf number, maximum leaf length, and maximum leaf width were measured according to the tobacco agronomic trait survey method (YC/T 142-1998). The tobacco leaves were harvested, then baked in the baking room, and the flue-cured tobacco yield was calculated after the baking was completed. The tobacco was graded according to GB/T5606.1-2004.

Two months after tobacco transplanting, each treatment was sampled using a five-point sampling method. First, the soil surface leaves, gravel, and other debris were swept away, and then soil samples at 5–20 cm were collected using a soil drill with a diameter of 3.5 cm, and about 2 kg of soil was taken from each community. Each soil sample was divided into two parts. One part (0.5 kg) was stored at −80°C for soil microbial analysis and the other part (1.5 kg) was used for the analysis of soil physicochemical properties and the detection of soil enzyme activity.

Soil ammonium nitrogen and nitrate nitrogen were extracted using a potassium chloride solution. Ammonium nitrogen can react with phenol to form blue indigo phenol dye, nitrate nitrogen can react with hydrochloride N(1-naphthyl)-ethylenediamine to form red dye, and the two have maximum absorption at 630 and 543 nm, respectively. This method can be used to determine the content of soil ammonium nitrogen and nitrate nitrogen (Page, 1983). The available phosphorus content in soil was extracted using sodium bicarbonate extractant, and the content of available phosphorus was measured using the continuous flow analysis of air fragments (Olsen, 1954). Soil available potassium was leached in a neutral 1 mol/L ammonium acetate solution and determined using a flame photometer (Model 410. Sherwood scientific Ltd Cambridge, UK). The determination of soil organic matter content followed the potassium dichromate oxidation-volumetric method (Franz et al., 1996). Soil pH was measured using a pH meter (China Shanghai Sanxin Instrument Co., Ltd.) in a 1:2.5 (w/v) soil solution. Electrical conductivity (EC) was measured with a 1:2.5 soil/water suspension using an MP513 conductivity meter (Shanghai Sanxin Instrument Co., Ltd., China). Ten grams of fresh soil was taken and dried at 105°C for 12 h, and then the soil moisture content was measured using the drying weighing difference method.

Soil alkaline phosphatase (S-AKP) activity was determined via diasodium phosphate colorimetric chromometry using the Soil Alkaline Phosphatase Activity Assay Kit (Liang et al., 2017). Soil catalase activity (S-CAT) was determined based on ultraviolet absorption using the Soil Catalase Activity Assay Kit (Geng et al., 2017). Soil peroxidase activity (S-POD) was determined by catalyzing the oxidation of organic matter to quinone using the Soil Peroxidase Activity Assay Kit (Tian and Shi, 2014). The absorbance values of alkaline phosphatase (AKP), catalase (CAT), and peroxidase (POD) at 660, 240, and 430 nm, respectively, were measured using an ultraviolet-visible spectrophotometer (UV-5100B) to reflect their respective enzyme activities. All kits were purchased from Beijing Box Shenggong Technology Co., Ltd., and the ultraviolet-visible spectrophotometer was produced by Shanghai Yuanxi Instrument Co., Ltd.

Total genomic DNA was extracted from each soil sample using the DNeasy PowerSoil Kit (100) isolation kit (QIAGEN Co. Ltd., Germany). The quality and concentration of extracted DNA were determined using 2% agarose gel electrophoresis and nanodroplets NanoPhotometer N50 Touch (Implen Gmbh Co. Ltd., Germany).

The V3–V4 region of the bacterial 16SrRNA gene was amplified using the universal primers 338F (5′-ACTCCTACGGGAGCAGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) (Nan et al., 2016). The internal transcribed spacer (ITS) of the fungus was amplified using the universal primers ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2R (5′-GCTGCGTTCTCCATCGATGC-3′) (Rachel et al., 2013). The PCR reaction consisted of 2 μL 10 × buffer, 2 μL 2.5 mM dNTPs, 0.8 μL forward primer (5 μM), 0.8 μL reverse primer (5 μM), 0.4 μL polymerase, 0.2 μL Bovine serum albumin, 10 ng template DNA, and finally the addition of ddH₂O to reach a total volume of 20 μL. The procedure for the PCR reaction was as follows: predenaturation at 95°C for 3 min; then denaturation at 95°C for 30 s and annealing at 55°C for 30 s; followed by extension for 45 s at 72°C, and these three steps were repeated for a total of 35 cycles; followed by single extension at 72°C for 10 min; and ending at 4°C. The PCR products were sequenced on the Illumina MiSeq platform of Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China).

The off-machine raw data was spliced using FLASH (v1.2.11). QIIME (Quantitative Insights Into Microbial Ecology, v1.9.1) was used for quality control of the original sequences. Using Uparse (v11), the obtained high-quality sequences were clustered into operational taxonomic units (OTUs) ≥ a similarity threshold of 97% (Edgar et al., 2011). The α diversity of fungal communities was analyzed based on the Chao and Shannon indices using the Mothur (v.1.30.0) software (Schloss et al., 2009). Redundancy analysis (RDA) was performed using the R (v.3.3.1) software package “Vegan,” and primary coordinate analysis (PCoA) was performed via direct mapping based on Euclidean distances. Python (v.2.7.0) software was used to analyze the relative abundances of species. Linear discriminant analysis effect size (LEFSe) analysis was used to linearly discriminate samples according to different classification methods (Segata et al., 2011). Using linear discriminant analysis (LDA), species with significant differences were found in the sample classification. The OTU abundance table of soil bacteria was normalized using PICRUSt1, and then Kyoto Encyclopedia of Genes and Genomes (KEGG) function-annotated OTUs based on the greengene ID corresponding to each OTU. According to FUNGuild, the functional classification of soil fungi and the abundance information of each functional classification in different samples were obtained. The pheatmappackage and R vegan package were used to generate heatmaps and statistical correlations (Kolde, 2015). Statistical analysis was conducted using IMB SPSS Statistics 25^th Edition (IBM, USA) and supplementary calculations were performed in Microsoft Excel 2016. Using Duncan’s new complex range method, one-way analysis of variance (ANOVA) was used to analyze changes in soil physicochemical properties and enzyme activity. R software was used to plot a visualization of environmental factor correlation analysis. OriginPro^®2018 (OriginLab Corp., USA) and Adobe Illustrator CC2019 (Adobe Systems Inc., USA) software were used for drawing.

Supplementary Table 2 shows the effects of the single application of cow manure and the mixed application of cow manure with different botanical oil meals on tobacco growth indicators. For the maximum blade width, the NFHS treatment was the largest and significantly higher compared to the NF, NFD, and NFZM treatments. While the NF treatment maximum blade width was the smallest, there was no significant difference between it and NFC, NFD, and NFZM treatment.

Supplementary Table 3 shows the effects of the single application of cow manure and the mixed application of cow manure with different botanical oil meals on the yield and quality of tobacco. The results showed that the mixed application of cow manure and different botanical oil meals could increase the yield of tobacco, and the largest yield was found in the NFHS treatment (NFC: 14.29%; NFD: 11.94%; NFHS: 28.29%; and NFZM: 21.34%). All fertilization treatments could increase the proportion of fine tobacco, and the largest proportion was found in the NFD treatment (NFC: 20.03%; NFD: 41.21%; NFHS: 10.24%; and NFZM: 34.53%). There was no significant difference in average price among all treatments, and the largest difference was 1.82 yuan/kg in the NFC treatment. The NFHS treatment had the highest output value, mainly because its output was much higher than those of the other treatments.

Compared with CK, the content of NH₄⁺-N in the soil in the NF treatment was the highest, and significantly higher than that in the NFC and NFD treatments. The content of soil NO₃^–-N was the highest in the NFHS treatment and the lowest in the NFZM treatment, and there were significant differences between the two and the CK, NF, and NFD treatments. The soil available-P content in the NFD and NFHS treatments was the highest, which was significantly higher than that in the CK and other treatments. The content of soil available-K was the highest in the NFD treatment, and there were significant differences among all fertilization treatments except for the NF and NFZM treatments. The soil organic matter content in CK was the lowest and was significantly lower than that in the NFC, NFD, NFHS, and NFZM treatments, but there was no significant difference in soil organic matter content among the four treatments. The soil pH content in the NFZM treatment was the highest, and there was no significant difference between the other four treatments and the control treatment, which were all significantly lower than that in the NFZM treatment. The EC of NFC and NFD was the highest and was significantly higher than that of the other four treatments, and the EC of the NFZM treatment was the lowest. Compared with the NF treatment, the available-K and EC of the NFC treatment were significantly increased; the available-K, available-P, and EC in the NFD treatment were significantly increased; and the available-P, available-K, and NO₃^–-N were significantly increased in the NFHS treatment. The pH of the NFZM treatment was significantly increased (Table 1).

Compared with the CK, although there was no significant difference between the S-POD activities of NFZM and CK, the other treatments increased the activities of S-CAT, S-POD, and S-AKP. In general, the improvement of S-AKP activities was the largest (NF: 5.93 times; NFC: 5.67 times; NFD: 3.38 times; NFHS: 4.27 times; and NFZM: 5.71 times). Compared with cow manure alone, the NFC, NFD, and NFZM treatments significantly reduced the activity of S-CAT (Figure 1A); the NFD treatment significantly increased the activity of S-POD (Figure 1B); and the NFD treatment significantly reduced the activity of S-AKP (Figure 1C).

In the soil bacterial community, compared with CK, the Chao index of each fertilization treatment was extremely significantly higher than CK, and the increase in the NFC treatment was the most obvious (Figure 2A). The Shannon index values of soil bacteria in each fertilization treatment were significantly higher than that of CK (Figure 2C). In the soil fungal community, compared with CK, the Chao index of each fertilization treatment was extremely significantly higher than that of CK (Figure 2B). However, in the Shannon index of soil fungi, there was no significant difference between each fertilization treatment and CK (Figure 2D). By comparing the NF treatment with the fertilizer treatment with the addition of different botanical oil meals, it was found that there were no significant differences in the Chao index and Shannon index between the NF treatment and the NFC, NFD, NFHS, and NFZM treatments in soil bacterial communities. In the soil fungal community, compared with NF treatment, the Chao index of the NFZM treatment was significantly increased, the Shannon index of the NFHS treatment was significantly decreased, and both the Chao index and Shannon index of the NFD treatment were significantly decreased.

The results of PCoA showed that no treatment (CK) and the single application of cow manure (NF), cow manure combined with rapeseed meal (NFC), cow manure combined with soybean meal (NFD), cow manure combined with peanut meal (NFHS), and cow manure combined with sesame meal (NFZM) could be significantly separated based on the PC1 axis. The contribution rates of PC1 and PC2 to the species composition difference between treatments were 75.06%/6.1% (bacteria) and 43.65%/15.14% (fungi), respectively (Figures 3A, B). In the soil fungal community, the fertilization treatments and CK could be significantly distinguished on the PC1 axis. The fertilization treatments, although were close on the PC1 axis, could be distinguished on the PC2 axis; this indicated that the application of cow manure and cow manure mixed with different botanical oil meals could change the soil fungal community structure. In the soil bacterial community, CK and each fertilization treatment could be significantly distinguished by the PC1 axis. Different from fungal communities, the NF and NFC treatments of soil bacterial communities were similar on the PC2 axis, indicating that soil bacterial communities applied with cow manure and cow manure mixed with botanical oil meal were similar.

Figure 4 shows the composition in soil microorganisms at the phylum level under different treatments. In the soil bacterial community, the top five relative abundance were Actinobacteriota (11.65, 32.07, 33.88, 30.72, 30.61, and 35.17% in CK, NF, NFC, NFD, NFHS, and NFZM, respectively); Chloroflexi (31.44, 17.55, 17.03, 17.86, 16.71, and 13.68%, respectively); Proteobacteria (15.05, 17.00, 17.70, 17.68, 18.71, and 21.16%, respectively); Acidobacteriota (16.48, 14.13, 11.63, 14.21, 12.29, and 9.84%, respectively); and Firmicutes (2.80, 7.42, 7.03, 7.29, 9.25, and 7.97, respectively). Notably, the relative abundance of GAL15 in CK was much higher than that in other treatments (9.93%), which was 486.5–983% higher than that in other treatments (Figure 4A). In the soil fungal community, the dominant phyla with the highest relative abundances were Ascomycota (the relative abundances in CK, NF, NFC, NFD, NFHS, and NFZM were 78.60, 64.44, 71.69, 73.08, 65.98, and 74.08%, respectively); Mortierellomycota (3.49, 22.87, 12.81, 18.24, 20.71, and 15.15%, respectively); Basidiomycota (13.49, 3.47, 8.52, 2.91, 2.02, and 3.32%, respectively); and Olpidiomycota (0, 1.06, 0.11, 0.36, 0.26, and 0.33%, respectively). There was a small amount of accumulation in all treatments, but not in CK (Figure 4B).

According to the results of the effects of different treatments on the soil physicochemical properties and tobacco yield and quality, the NFD, NFHS, and NFZM treatments as well as the NF treatment were selected for LEFSe multilevel species difference analysis. The classification level ranged from phylum to genus, and the LDA threshold was 3.

In the soil bacterial community, Bacillus, Fictibacillus, and Micromonosporaceae were significantly enriched in the NF treatment compared with the NFD treatment, while Sphingononas and Streptomyces were significantly enriched in the NFD treatment (Figure 5A). Compared with the NFHS treatment, at the phylum level, Methylomirabilota was significantly enriched in the NF treatment, while Firmicutes, Bacteroidota, and Pseudolabrys were significantly enriched in the NFHS treatment (Figure 5C). Compared with the NFZM treatment, at the phylum level Chloroflexi and Planctomycetota were significantly enriched in the NF treatment group, while the phyla Bacteroidota and Actinobacteriota as well as the genera Spingononas and Streptomyces were significantly enriched in the NFZM treatment (Figure 5E). In the soil fungal community, compared with the NFD treatment, Mortierellomycota at the phylum level and Cordana, Schizothecium, Venturiocistella, Sterigmatomyces, and Trichosporon at the genus level were significantly enriched in the NF treatment. The phylum Ascomycota and the genera Conocybe and Subulecystidium were significantly enriched in the NFD treatment (Figure 5B). Compared with the NFHS treatment, the Chytridiomycota phylum and the Schizothecium, Acremonium, Basidiomycota Emericellopsis, Metarhizium, Lecanicillium, Cephalotrichum, Gonytrichum, Gonytrichum, Aspergillus, Psilocybe, and Olpidium genera were significantly enriched in the NF treatment. The Chytridiomycota phylum and the Ustilaginoidea, Myrothecium, Westerdykella, Phoma, Penicillum, and Talaromyces genera were significantly enriched in the NFHS treatment (Figure 5D). Compared with the NFZM treatment, the Mortierellomycota phylum and the Cordana, Sanocladium, Emerlopsis, Clonostachys, Neocosmospora, Metarhizium, Lecythophora, Pyrenochaetopsis, Aspergillus, and Psilocybe genera were significantly enriched in the NF treatment, and the Ascomycota and Chytridiomycota phyla, as well as the Chaetomium, Zopfiella, Dichotomopilus, Gibellulopsis, and Penicillium genera, were significantly enriched in the NFZM treatment (Figure 5F).

This work studied the relationship between the soil physicochemical properties, enzyme activities, and microbial community on the genus level using RDA analysis. The results showed that the interpretative values of the RDA1 and RDA2 axes for soil bacteria and fungi were 77.17%/5.32% and 50.97%/8.50%, respectively (Figures 6A, B). In the soil bacterial community, AKP had the greatest influence on the soil bacterial community, while NO₃^–-N had the least. Except for the negative correlation between NH₄⁺-N and NO₃^–-N, other soil physicochemical properties and enzyme activities were positively correlated. Overall, the microorganisms in the NF, NFC, NFD, and NFHS treatments were positively correlated with soil physicochemical properties and enzyme activities. The NFZM treatment was negatively correlated with NH₄⁺-N and positively correlated with other soil physicochemical properties and enzyme activities. In the soil fungal community, AKP had the greatest influence on soil fungi, while NO₃^–-N had the least influence. Among all soil physicochemical properties and enzyme activities, NO₃^–-N was only positively correlated with POD and EC, available-P was only negatively correlated with NO₃^–-N, POD, and EC, and other soil physicochemical properties were positively correlated with enzyme activities. In each fertilization treatment, CK was positively correlated with NO₃^–-N, while the soil fungal communities in the NFC, NFD, NFHS, and NFZM treatments were positively correlated with available-P, AKP, CAT, NH₄⁺-N, PH, available-K, and organic matter, and negatively correlated with EC, POD, and NO₃^–-N.

This study assessed the correlations between soil physicochemical properties, enzyme activities, and microorganism using correlation heatmaps (Figures 6C, D), which showed the top 30 species in total abundance at the genus level. In the soil bacterial community, Intrasporangium, Micromonospora, Streptomyces, Gaiella, Mycobacterium, Bacillus, and Defluviicoccus showed significant positive correlations with AKP, while norank_f_norank_o_norank_c_TK10, norank_f_norank_o_norank_c_AD3, norank_f_Xanthobacteraceae, and norank_f_norank_o_IMCC26256 showed significant negative correlations with AKP. In addition, it was found that the correlation heatmap could further explain the results of the RDA analysis. AKP, available-K, and EC were in a straight line in the RDA analysis, and their colors were similar in the correlation heatmap. However, because EC had the shortest length in the RDA analysis, its significance in the correlation heat map was not as obvious as those of AKP and available-K. In the soil fungal community, Acremonium, Plectosphaerella, Gibellulopsis, Penicillium, Ramophialophora, and Trichosporon showed significant positive correlations with AKP, while Membranomyces and Saccharomyces showed significant negative correlations with AKP. It was worth noting that Plectosphaerella was significantly positively correlated with five soil physicochemical properties and enzyme activities, while Membranomyces was significantly negatively correlated with six soil physicochemical properties and enzyme activities, which are key components of the soil fungal community.

The functions of soil bacterial communities in KEGG metabolic pathways were analyzed using PICRUSt prediction. The five functions related to microbial metabolism in KEGG were metabolism (51.70–53.46%), genetic information (15.03–16.36%), environmental information processing (13.35–14.17%), cellular processes (3.33–3.72%), and organismal systems (0.74–0.87%) (Figure 7A). At KEGG level 2, the top three functions were membrane transport (14.87–16.02%), amino acid metabolism (14.25–15.16%), and carbohydrate metabolism (14.61–14.74%) (Figure 7B). Xenobiotics biodegradation and metabolism as well as glycan biosynthesis and metabolism were also worth attention. Compared with CK, the increase (24.32–31.32%) and decrease (24.84–28.52%) of the two fertilization treatment treatments were the largest. Therefore, at KEGG Level 3, this work focused on the changes of specific metabolic pathway functional genes involved in membrane transport, amino acid metabolism, carbohydrate metabolism, xenobiotics biodegradation and metabolism, and glycan biosynthesis and metabolism.

In membrane transport (Figure 7C), the abundance of functional genes of the phosphotransferase system (PTS) was the highest. Compared with the CK, the NF, NFC, NFD, NFHS, NFZM treatments were significantly higher. For the functional genes of ABC transports, those in the fertilization treatments were significantly lower than those in CK, while there was no significant difference in bacterial secretion between the fertilization treatments and CK. For amino acid metabolism (Figure 7D), compared with CK, the relative abundances of functional genes related to tryptophan metabolism, lysine metabolism and valine, leucine, and isoleucine degradation were significantly increased in all treatments, among which the NFZM treatment was the highest, but there was no significant difference between treatments. Compared with CK, the relative abundances of functional genes in histidine metabolism; lysine biosynthesis; valine, leucine, and isoleucine biosynthesis; cysteine and methionine metabolism; and alanine, aspartate, and glutamate metabolism were significantly decreased in all treatments, and there was no significant difference between treatments. In carbohydrate metabolism (Figure 7E), compared with CK, the relative abundances of functional genes in pyruvate metabolism, butanoate metabolism, propanoate metabolism, and glyoxylate and dicarboxylate metabolism in all treatments were significantly increased, and there was no significant difference between treatments. However, the relative abundances of functional genes in amino sugar and nucleotide sugar metabolism, starch and sucrose metabolism, and C5-branched dibasic acid metabolism were significantly decreased, and there was no significant difference between treatments. In xenobiotics biodegradation and metabolism (Figures 7F, G), the relative abundances of functional genes of the metabolic pathways in each fertilization treatment were significantly higher compared to CK, among which the relative abundance of the functional genes of the top five metabolic pathways were aminobenzoate degradation, caprolactam degradation, naphthalene degradation, chloroalkane and chloroalkene degradation, and toluene degradation. Importantly, compared with CK, the relative abundances of functional genes for ethylbenzene degradation and polycyclic aromatic hydrocarbon degradation in the NFZM treatment were significantly increased. In glycan biosynthesis and metabolism (Figure 7H), compared with CK, the relative abundances of functional genes of peptidoglycan biosynthesis, lipopolysaccharide biosynthesis, other glycans degradation, glycosphingolipid biosynthesis-ganglio series, and glycosaminoglycan degradation were significantly decreased. For N-glycan biosynthesis, only the NFZM treatment was significantly decreased compared with CK.

FUNGuild was used to predict the function of soil fungal communities (Figure 8). Regardless of CK or each fertilization treatment, the relative abundances of undefined saprotrophs, endophyte- litter saprotroph-soil saprotroph-undefined saprotrophs, and plant pathogens accounted for more than 50%, indicating that they constituted the majority of the soil fungal community. Compared with CK, the relative abundances of undefined saprotrophs in each fertilization treatment decreased, while the relative abundances of endophyte- litter saprotroph-soil saprotroph-undefined saprotrophs increased. Notably, there were different functional communities in each treatment. The relative abundances of animal pathogens in the NFC treatment was higher than that in the other groups. In the NFD treatment, the relative abundance of animal pathogen-dung saprotroph-endophyte-epiphyte-plant saprotroph-wood saprotrophs was higher than that in other treatments. In the NFHS treatment, the relative abundance of animal pathogen-endophyte-lichen parasite-plant pathogen-wood saprotrophs was higher than that in other treatments. The relative abundance of plant pathogens in the NFZM treatment was higher than that in other treatments.

The crop growth, yield, and quality are the most direct reflections of the crop growth environment. The results showed that compared with other fertilization treatments, the combined application of cow manure and peanut bran had the highest tobacco agronomic traits and yield. Compared with other organic fertilizers, peanut bran had a lower carbon: nitrogen ratio, a faster decomposition rate after application in soil, and a faster fertilizer effect. This treatment could provide full nutrition to tobacco in the seedling stage, which helped the roots absorb the nutrients more quickly. The nutrients then spread to all parts of the plant. The findings of this study were consistent with previous results. Roeswitawati et al. (2021) found that peanut bran and biological mud had significant effects on the growth and yield of sweet corn, with 15 tons/ha of peanut bran and 125 ml/L of biological mud having the best effect. The average price of flue-cured tobacco was the highest after the mixed application of cow manure and rapeseed oil meal, which indicated that rapeseed oil meal could improve the quality of flue-cured tobacco.

The physicochemical properties of soil are the key factors that determine soil fertility. The long-term application of chemical fertilizers or the continuous application of the same manure would cause the unpredictable consumption of soil nutrients, which seriously damages soil productivity. The results showed that the mixed application of cow manure with different botanical oil meals had different changes in soil available-P, available-K, EC, NO₃^–-N, and pH: rapeseed oil meal significantly increased the effect of EC, soybean meal significantly increased the effect of available-P and available-K, peanut bran significantly increased the effect of NO₃^–-N, and sesame meal significantly increased the effect of pH. The soil available-P, available-K, EC, NO₃^–-N, and pH interacted with each other to some extent. Soil with neutral pH can reduce the fixation effect of P and improve the availability of soil P (Muyang et al., 2016). For soil with high organic matter content, the mineralization of organic matter can provide part of the required inorganic phosphorus, which weakens the effect of fixing phosphorus and increases the content of soil available-P. The NFHS treatment had the lowest NH₄⁺-N content and the highest NO₃^–-N content. NH₄⁺-N and NO₃^–-N are converted into each other through nitrification and denitrification. NO₃^–-N is easily dissolved and has high activity in the soil, which can quickly provide nitrogen nutrition for crops; however, it is also easy to lose. This characteristic meets the growth demand for tobacco; tobacco grows rapidly and can turn senescence and mature at an appropriate time (Ghiasi et al., 2021).

Soil enzyme activity reflects the ability of soil to perform catalytic substance conversion. The activity mainly derived from the secretions of soil animals, plants, and microorganisms as well as the decomposition of various residues. To some extent, soil enzyme activity participate in the occurrence and development of soil and the formation and evolution of soil fertility (Li et al., 2023). Yang et al. (2022) found that deep tillage accompanied by the application of organic fertilizer increased the activities of sucrase, cellulase, and urease in wheat fields; furthermore, Liu et al. showed that the application of organic fertilizer, chemical fertilizer (nitrogen fertilizer), and organic fertilizer combined with chemical fertilizer increased the activities of sucrase in soil by 124.1, 80.9, and 145.6%, respectively, while urease and alkaline phosphatase were also increased to varying degrees (Liu et al., 2021). These findings were consistent with the results of the present study. The application of cow manure and mixed application of cow manure and botanical oil meal could improve soil enzyme activity. The effects of adding different botanical oil meals on soil enzyme activity were different. After the addition of soybean meal, S-POD activity was significantly increased compared with cow manure applied alone, but S-AKP activity was significantly decreased. This might be because the N element in the soybean meal was higher than that in the other botanical oil meals, while the P element content was low. Therefore, soil microbial reactions related to the metabolism of N were enhanced, while the number of reactions associated with P decreased.

For soil bacterial communities, the Chao index and Shannon index were significantly increased by the single application of cow manure and the combined application of cow manure and botanical oil meal. For soil fungal communities, the Chao index was increased by the single application of cow manure and the combined application of cow manure and botanical oil meal. Thuy et al. found that buffalo manure compost and vermicompost had positive effects on the activity of culturable bacteria and the bacterial Shannon index in soil (Thuy et al., 2013). Lupwayi et al. (2005) found that the Shannon index of soil bacteria was not significantly affected by cow manure and pig manure in the first two years, but significantly increased in the third year, and these results were consistent with the present study. Compared with the single application of cow manure, the addition of different botanical oil meals had no effects on the Chao index and Shannon index of soil bacteria, but for soil fungi, the addition of soybean meal significantly reduced the Chao index and Shannon index. The β diversity of microbial communities in each treatment was significantly different from that in CK. However, compared with cow manure application alone, although there were differences with the addition of different botanical oil meals, the difference was not as large as that in CK, which was verified based on the results of soil microbial α diversity.

In the soil bacterial community, Actinobacteriota, Chloroflexi, and Proteobacteria accounted for more than 50% of soil bacteria and were the dominant phyla. Actinobacteriota includes important saprophytic bacteria that can promote the decay of plant and animal remains in soil and play a role in the nitrogen cycle in soil (Hazarika et al., 2022). Chloroflexi is a class of photoautotrophic bacteria known for their special 3-HP fixed CO₂, making these bacteria of great significance to the entire ecosystem as primary producers (Narsing et al., 2022). Proteobacteria is the largest and most diverse phylum of bacteria and is of extensive importance in terms of phylogeny, ecology, and pathogenicity (Feng L. et al., 2023). Ascomycota occupies the greatest abundance in the soil fungal community. Studies have found that Ascomycota exists in soils all over the world and contains the pioneer decomposer species, whose main function is to ensure soil ecological security (Eleonora et al., 2019). The present study found that the abundance of Mortierellomycota in NF, NFC, NFD, NFHS and NFZM treatments was greater than that in CK. Studies have found that as a saprophytic fungus, Mortierellomycota has a high content in soil rich in organic matter and is a key microorganism in soil carbon and nutrient conversion (Shi et al., 2020). This was also confirmed by the results of the present study, as both the single application of cow manure and the combined application of cow manure with botanical oil meal increased the soil organic matter content.

LEFSe analysis revealed the microbial differences between cow manure and soybean meal, peanut bran, and sesame meal mixed application, and the single application of cow manure. The results showed that the combined application of cow manure with soybean meal, peanut bran, and sesame meal could increase the relative abundances of functional bacteria in soil compared with the single application of cow manure, and this effect was more obvious in the soil fungal community. In the soil bacterial community, subgroup_7 was enriched with the application of the three botanical oil meals. Wang et al. found that the relative abundance of subgroup_7 was significantly positively correlated with the 1000-grain weight, grain number per panicle, and yield of rice, and was a functional flora related to crop yield (Wang et al., 2021). As a saprophytic bacterium, Spingomonas paucimobilis UT26 can simultaneously degrade methyl parathion and γ-HCH (Lan et al., 2014). Kruczyńska et al. reported that Bacteroidota can become an important indicator of soil quality in the study of soil biodegradation because the decrease of the abundance of these beneficial microorganisms may be related to the decrease of soil quality and fertility, and ultimately affect crop yield (Kruczyńska et al., 2023). In the soil fungal community, the three kinds of botanical oil meal were enriched in Chaetomium and Penicillium. Chaetomium can be used as a biocontrol bacterium. One Chaetomium species, C. globosum Cg-6, has a 60% inhibition potential against Rhizoctonia solani, and can increase the tuber germination rate (94%) (Arunkumar et al., 2021). Penicillium is a kind of phosphorus solution fungus that can be used as a bacteriological agent mixed with phosphate fertilizer and can transform phosphorus that is difficult to be absorbed and utilized by plants into a form that can be absorbed and utilized. This function is indispensable for the material cycle of the soil ecosystem (Chen et al., 2023).

AKP is the most influential factor for both soil bacteria and soil fungi. The level of S-AKP activity directly affects the decomposition and transformation of organic phosphorus in soil and its bioavailability and is an indicator to evaluate the direction and intensity of soil phosphorus biotransformation (Hu et al., 2015). Studies have shown that adding 0.01–3.0 mg/kg cadmium significantly improves the activity of S-AKP, and the soil bacterial community also responds to the addition of cadmium (Han et al., 2019). Chen et al. (2019). found that the long-term application of nitrogen fertilizer and pig manure decreased the activity of S-AKP and the abundances of phoD genes. Intrasporangium and Micromonospora are the soil bacteria most closely related to AKP. Studies have shown that there are freezing protein-coding sequences in Intrasporangium, and this strain has strong adaptability to harsh environments (Thomas et al., 2018). Karthikeyan et al. (2022) found that Micromonospora, as a biological control agent, can inhibit Fusarium wilt and promote the growth of host plants when it is colonized in the rhizosphere of Casuarina. We found that Acremonium and Gibellulopsis were the soil fungi with the strongest correlation with AKP. Khan et al. (2021) isolated the endophytic fungus Acremonium sp. Ld-03 from the bulb of Tripterygium wilfordii. It was used against Fusarium oxysporum, Botrytis cinerea, Botryosphaeria dothidea, and Fusarium fujikuroi. Acremonium sp. Ld-03 has certain inhibitory effects, and its secondary metabolites contain indoleacetic acid, which has a significant growth-promoting effect on leek. Studies have shown that CEF08111, a species of Gibellulopsis, can effectively control cotton verticillium wilt and induce a defense response in cotton plants (Feng Z. et al., 2023).

For the functional prediction of the soil bacterial community, the relative abundance of Metabolism at LEVEL 1 was the largest, accounting for more than 50%, and the top three relative abundances at LEVEL 2 were membrane transport, amino acid metabolism, and carbohydrate metab olism, which was consistent with previous results (Huang et al., 2021). In membrane transport, the relative abundance of the functional genes of the phosphotransferase system was the largest, and the fertilization treatments had significantly higher abundances than CK, which also confirmed the previous results. AKP had the strongest correlation with soil microorganisms. Amino acids are an important component of soil organic nitrogen that can be used by some bacteria in the soil as precursors to synthesize plant growth regulators in the process of physiological metabolism to stimulate or promote plant growth and development (Tan et al., 2022). Tryptophan is an important precursor substance for auxin synthesis in plants, and its enhanced metabolism can regulate crop root morphology and promote root growth (Tashkandi et al., 2022). Enhanced lysine metabolism can promote chlorophyll synthesis, regulate stomatal opening, and promote leaf photosynthesis, which is conducive to crop yield increase (Fitriana et al., 2022). Carbohydrates are the most active organic component in soil, as well as one of the easily degraded components. There are interactions between carbohydrates, clay, and various ions that can stabilize the soil structure, and they also has the certain instruction to the change of soil microorganisms. In carbohydrate metabolism, pyruvate plays an important pivotal role by connecting glucose, fatty acid, and amino acid metabolism through acetyl CoA. Studies have shown that pyruvate metabolism is enhanced in non-rhizosphere and rhizosphere soils contaminated by heavy metals, which may be a signal of resistance to adversity (Zhang M. et al., 2022). In xenobiotics biodegradation and metabolism, the relative abundances of 11 functional genes were increased, and most of these heterologous metabolites were polluted by macromolecular organic compounds (such as organophosphorus, organochlorine, and aromatic hydrocarbons). Studies have shown that components including amino acids, microorganisms, and isoenzymes are combined with fertilizers and biological agents to produce products that degrade agricultural residues. This can achieve multi-purpose utility, such as the degradation of agricultural residues, yield increase, and improved stress resistance (Fei et al., 2021). In glycan biosynthesis and metabolism, peptidoglycan is the basic component of the bacterial cell wall and strengthens the cell wall, and lipopolysaccharide is the basic component of gram-negative bacterial cell walls. The relative abundances of the functional genes of biosynthesis of both decreased. It was speculated that with the application of plant-derived organic fertilizer, the physical properties of soil were improved, and the soil was not compact and had good air permeability. Bacteria no longer needed a hardened cell wall to survive.

In the function prediction of soil fungi, it was found that the relative abundance of saprophytes accounted for the majority in all treatments. Such organisms feed on dead organic matter and can decompose dead and weakened organic matter into simple substances for easy absorption and recycling by plants, which is beneficial to soil nutrient activation and crop growth and extremely important to soil ecosystems (Mahmud et al., 2017). In addition, it was found that the soil fungal communities exhibited the special functions of plant endophytes, plant saprophytes, and wood saprophytes due to the application of different botanical oil meals. Endophytic fungi live in plants, and this special environment makes it easier for these fungi to play a role in biocontrol compared to other saprophytes and epiphytes. Zhang L. et al. (2022) isolated 144 endophytes from different tissues of medicinal plants, among which strain 8ZJF-21 showed broad-spectrum antifungal activity and promoted plant growth by producing a variety of extracellular enzymes and metabolites. Wood saprophytes are regarded as forest cleaners. These fungi can decompose dead branches and fallen leaves and return them to nature, thereby participating in the material cycle and promoting the natural metabolism of forest trees to maintain ecological balance (Schwarze et al., 2022).