Effects of Chicken Farming on Soil Phosphorus Availability and Associated Microbial Properties in Lei Bamboo (Phyllostachys praecox) Forest Ecosystems

Background: Bamboo-chicken farming (BCF) is a popular bamboo complex management model in Southeast Asia owing to its high economic benets. However, the effects of BCF on phosphorus (P) availability and the associated microbial communities in soil remain poorly understood. In this study, we compared the soil properties, P fractions, phosphatase activities, and bacterial community compositions in the surface soil (0–20 cm) of a typical bamboo (Phyllostachys praecox)-chicken farming system under different grazing densities (represented as distances of 5, 15, 25, and 35 m from the henhouse, respectively). The variables were also compared with the soil measurements from an adjacent pure bamboo forest without chicken framing (control site). Results: We observed a signicant increase in soil pH, cation exchange capacity (CEC), total N (TN), total P (TP), and available potassium (AK) with increasing grazing density, while soil organic carbon (SOC) showed no signicant difference between the sites. The total P accumulation of the soil was also more rapid than that of SOC and TN with increasing grazing density. Labile P and moderately labile P dominated the soil P accumulation under BCF. In particular, Resin-Pi (labile P), NaHCO 3 -Pi (labile P), and 1 M HCl-Pi (moderately labile P) increased by 100–233%, 83–183% and 414–1314%, respectively, compared with the control values. In contrast, the contribution of labile or moderately labile organic phosphorus to the total phosphorus (Pt) content decreased signicantly with increasing grazing density from 38.54% (control) to 17.65% (5-m site). Phosphatase activity also increased with increasing grazing density, which suggests that BCF effectively promoted the mineralization of soil Po. A redundancy analysis showed that the changes in bacterial community structure were closely related to Resin-Pi and 1 M HCl-Pi (r 2 = 0.938 and 0.958, respectively). The relative abundances of the phosphobacteria Flavobacterium, Pseudomonas, Streptomyces, and Arthobacter increased with increasing grazing density,


Background
Bamboo is an important component of tropical and subtropical forest ecosystems and is also a widely exploited natural resource in southern China (Phimmachanh et al., 2015;FAO, 2010). The bamboo forest industry has a long history in China, and the planting of pure forests is still the main management Microorganisms are involved in several key biogeochemical processes of P cycling, including P uptake, release, and redistribution (Richardson & Simpson, 2011;Pistocchia et al., 2018;. Therefore, studying the changes in microbial community structure under BCF conditions, especially Prelated microorganisms, can further our understanding of the dominant mechanisms underlying the changes in soil P fractions. Furthermore, changes in microbial communities can affect soil enzymes and their activities, as enzymes in soil are mainly secreted by microorganisms (Tarafdar & Claassen, 1988; Waring et al., 2014). Acid phosphatase (ACP) and alkaline phosphatase (ALP) are non-speci c enzymes that promote the bioavailability of phosphorus in soil by catalyzing the hydrolysis of ester-phosphate bonds of monoesters in orthophosphate (other than phytate) (Fraser et al., 2015;Nannipieri et al., 2010). Moreover, the ACP and ALP activities in soil have been measured to assess the conversion process of Po to Pi, as they are considered to play a major role in this process (Tarafdar & Jungk, 1987). Therefore, combining the study of the microbially mediated Pi supply pathway with P fraction change mechanisms can improve our understanding of P biogeochemical cycles and bioavailability under BCF conditions.
In this study, we aimed to determine the impact of chicken farming on soil P availability and microbial community structures in Lei bamboo (Phyllostachys praecox) forests. We assessed the biogeochemical changes under different grazing intensities to elucidate the relationships between the P fractions and bacterial community and to evaluate the sustainability of BCF systems. Speci cally, we tested the following hypotheses: (1) chicken farming largely affects soil P status-especially the P fractions; (2) grazing density signi cantly affects the distribution of organic and inorganic P fractions; and (3) chicken farming signi cantly increases soil P availability and accelerates soil phosphatase activity, which consequently impacts the bacterial community structure in soil.

Experimental site
The study site is located in Jingshan Town (N 30°24′, E 119°52′, 125 m ASL), which is northwest of Hangzhou in Zhejiang Province, China ( Figure 1). The region has a mid-latitude subtropical monsoon climate, with an average annual rainfall and temperature of approximately 1454 mm and 17.8 ℃, respectively. The minimum and maximum temperatures are 2 ℃ in January and 39 ℃ in July, respectively. The experiment site is a Lei bamboo shoot production area, and its soil is classi ed as Ferralic Cambisol (FAO, 2006). The region is located on the sunny side of the hill with a slope of 10-15 °. The canopy closure is 80-90%, and no other plants grow on the surface. The plot has undergone shortterm intensive management for two years. The BCF system has been operating for six years, and only selective logging is conducted to maintain an appropriate canopy density. The BCF area is approximately 15 ha and contains 1500-1800 chickens per hectare.

Experimental design and soil sampling
In this study, we characterized the changes in grazing intensity as the differences in grazing distance from the hen house, which is based on the chickens' habits. This technique is a common method for estimating plant and animal density and abundance (Stumpp et al., 2005;Manthey & Peper, 2010). A total of four sites (5,15,25, and 35 m distance from the hen house) and one control site (CK; > 60 m from the hen house) were assessed in the same manner. A pure bamboo forest located > 60 m from the hen house was selected as the control site as its environmental conditions and initial soil properties were similar to those of the BCF forest.
For all sites, ve 2 × 2 m sampling plots were randomly selected in an S shape along the slope direction.
Three samples from the same soil layer (0 -20 cm) of each sampling plot were collected in mid-May 2019 and carefully mixed to form ve composite samples. The composite samples were then transferred into a sterile sealed bag, placed on ice, and transported to the laboratory for pretreatment. After the samples were thoroughly mixed and hand-sieved through a 2-mm sieve, a portion of each fresh soil sample was stored at 4 ℃ and -80 ℃ for further analysis. The remaining portion of the soil sample was air-dried and stored for determining soil properties, P fractions, and phosphatase activity.

Soil basic properties
Soil pH was measured from a soil suspension (1:2.5, w/v) using a pH meter (PHS-3E; REX, China). The soil cation exchange capacity (CEC) was determined via the ammonium acetate method (Chapman, 1965). Brie y, 2 g of air-dried soil was placed in a 100 ml centrifuge tube and then centrifuged at 4000 rpm for 5 min with 50 ml of 1 mol·L -1 NH 4 OAc solution. The supernatant was then discarded, and the above process was repeated 3-4 times. The soil sample was then rinsed 4-5 times with 25 ml 95% ethanol. After discarding the ethanol solution, 1 g solid magnesium chloride and 2 ml liquid para n were added to the soil, and the NH 4 + content was determined using a Kjeldahl nitrogen analyzer (UDK159; Velp Scienti ca; Italy). Soil organic carbon (SOC) was measured using a TOC analyzer (Multi N/C 3100; Analytik Jena, Germany), and soil alkali-hydrolysable nitrogen (AN) was determined following the alkalidiffusion method (Bremner et al., 1996). Soil available potassium (AK) was extracted with 1 mol·L -1 CH 3 COONH 4 at pH 7 and measured by ame photometry (FP6410 INESA, China).

Fractionation procedure for soil P
The continuous extraction method proposed by Hedley et al. (1982) and modi ed by Tissen and Moir (1993) was used to fractionate the soil P in this study. Brie y, the P fractions were extracted from 0.5 g air-dried soil in a 50 ml centrifuge tube via the following steps: (1) two 9 × 62 mm resin strips and 30 ml distilled water were added to the centrifuge tubes, and the phosphorus in the resin strips was extracted with 0.5 M HCl after 16 h of shaking at 160 rpm (Resin-Pi); (2) after removing the aqueous solution, 30 ml 0.5 M NaHCO 3 at pH 8.5 was added and shaken for 16 h to extract NaHCO 3 -P; (3) 30 ml 0.1 M NaOH was added and shaken for 16 h to extract NaOH-P; (4) 1 M HCl-Pi was extracted by adding 30 ml 1 M HCl to the centrifuge tube and shaken for 16 h; (5) the soil residue was further extracted with 15 ml concentrated HCl at 80 ℃ (conc. HCl-P); and nally, (6) P was extracted by boiling the soil residue in 8 ml concentrated H 2 SO 4 and 10 drops of HClO 4 (Residual-P). The concentrations of Pi in the different extracts (Resin-Pi, 1 M HCl-Pi, and conc. HCl-P) were determined via the method described by Murphy and Riley (Murphy & Riley, 1962). The extracts that were obtained using NaHCO 3 and NaOH were rst acidi ed with 6 ml and 1.6 ml 0.9 M H 2 SO 4 , respectively, before determining the Pi concentrations via the method by Murphy and Riley. The total P concentrations of the different extracts (NaHCO 3 -P, NaOH-P, and conc. HCl-P) were determined by the ammonium persulfate digestion method. Finally, the Po concentration of the different extracts (NaHCO 3 -P, NaOH-P, and conc. HCl-P) was calculated as the difference between total P and Pi.
2.5 ACP and ALP activities ACP (EC 3.1.3.2) and ALP (EC 3.1.3.1) activities were determined following a modi ed method as described by Tabtabai and Bremner (1969). Brie y, 0.5 ml toluene was added to 1 g of air-dried soil and then shaken for 15 min. A modi ed universal buffer (pH 6.5 for ACP and pH 11.0 for ALP) was then added, and the sample was subsequently incubated at 37 ℃ for 24 h. The formation of p-nitrophenol was determined at 400 nm after the termination of the enzymatic reaction. ACP and ALP activities were expressed as µmol pNP g -1 soil d -1 .
2.6 Soil DNA extraction, PCR, and high-throughput sequencing The Fast DNA ® Spin Kit for Soil (MP Biomedicals, U.S.A) was used to extract the soil DNA according to the manufacturer's instructions. We used 1% agarose gel electrophoresis to detect the DNA extraction quality and NanoDrop2000 to determine the DNA concentration and purity. Primers 338F (5'-ACTCCTACGGGAGGCAGCAG-3') and 806R (5'-GGACTACHVGGGTWTCTAAT-3') were used to amplify the V3-V4 hypervariable region sequences of the bacterial 16S rRNA gene . Polymerase chain reaction (PCR) ampli cation, library construction, and Illumina Miseq sequencing were performed by Shanghai Majorbio Bio-pharm Technology Co., Ltd.

Bioinformatic analysis
Trimmomatic v 0.32 was used to perform quality control on the original sequencing sequence (Lohse et al., 2012). The quality control sequence was performed double-end splicing using FLASH v 1.2.11 (Magoč et al., 2011). Chimeras 16S sequences were removed, and high-quality nucleic acid sequences were clustered into operational taxonomic units (OTUs) based on 97% similarity using UPARSE v 7.1 . The alignment threshold was 70%, and each sequence was classi ed and annotated using the RDP classi er (http://rdp.cme.msu.edu/). The phosphobacterial species were identi ed by database screening.

Statistical analysis
We used the SPSS statistical software package version 21.0 for Windows to perform the statistical analyses in this study. A one-way analysis of variance (ANOVA) and Duncan's test were used to identify the signi cant differences (p < 0.05) between the soil samples. The alpha diversity (Shannon, Chao1, and Observed species index) of the soil bacterial communities was calculated using the Vegan package of R (ver. 3.6.3). To determine the differences in soil bacterial communities between the different sites, we used QIIME to conduct a principal coordinate analysis (PCoA) based on Unifrac distance. We conducted a redundancy analysis (RDA) on soil environmental factors and bacterial communities, which was visualized by the Vegan and ggplot2 packages of R. Finally, a Pearson's correlation analysis was used to test the correlation between soil phosphobacteria and P fractions.

Soil properties
The BCF soil showed signi cantly different chemical properties to that of the control soil (Table 1). Soil pH, CEC, SOC, TN, AN, and AK were signi cantly higher in BCF (p < 0.05) than in the control. The soil pH, CEC, TN, AN, and AK were highest at the 5-m site; however, we observed no signi cant difference in SOC across the different sites (p > 0.05). The soil total P (Pt) content in the BCF was signi cantly higher than that of the control (p < 0.05) and increased with increasing grazing density. Pi accounted for approximately 60-82% of Pt and also showed similar variability to that of Pt. The Pi/Pt value signi cantly increased with increasing grazing density (p < 0.05). Compared with the control, the BCF soil had a signi cantly higher total Po content but lower Po/Pt values (p < 0.05).

Soil P fractions and phosphatase activity
Our results showed that the BCF system signi cantly affected the relative content of each P fraction ( Figure 2). The relative content of labile P was approximately 26-28% of Pt. In comparison, the relative content of moderately labile P was 52-64% of Pt, which gradually increased with increasing grazing density. Furthermore, NaHCO 3 -Pi contributed the largest proportion to labile P in the soil, accounting for approximately 44 -50%. Moreover, the relative content of Resin-Pi gradually increased with increasing grazing density, while the proportion of NaHCO 3 -Po in labile P gradually decreased. Chicken farming had altered the composition of moderately labile P. The proportion of NaOH-extracted P in moderately labile P decreased gradually with increasing grazing density from 86% to 40%. Correspondingly, the proportion of 1 M HCl-Pi increased with decreasing grazing distance; the proportion at 5 m was higher than those of the other sites and the control group by ~1.36-fold and 13.54-fold, respectively. Compared with the control, chicken farming reduced the relative content of sparingly labile P, but the variability in its composition was limited (the relative content of conc. HCl-Pi varied within the range of 59-68% across all sites).
Labile P, moderately labile P, and sparingly labile P increased signi cantly with increasing grazing density (p < 0.05) (  Chicken farming affected the activity of ACP and ALP in soil (Figure 3). The activity of soil ACP was signi cantly higher in BCF than in the control (p < 0.05) (Figure 3a), but no signi cant differences were observed between the different sites (p > 0.05). The highest ALP activity was observed at the 5-m site (9.97 µmol pNP g -1 soil d -1 ), which was signi cantly higher than the other sites and the control (p < 0.05) (Figure 3b). Compared with the control, the ALP activity at 15, 25, and 35 m signi cantly increased by 66.43%, 26.96%, and 69.61%, respectively (p < 0.05).

Soil bacterial community structure
Except for 25 m, we observed no signi cant difference between the alpha diversity index of the different sites and the control (p > 0.05) ( Figure S1). We used a PCoA to compare the bacterial communities between the sites and the control based on unweighted and weighted Unifrac distances ( Figure S2). The unweighted PCoA clearly separated the BCF and control bacterial communities ( Figure S2a). Moreover, the 15-, 25-, and 35-m sites were tightly clustered, whereas the 5-m site notably deviated from the group. The sample aggregation in the weighted analysis was similar to that in the unweighted analysis. The combined axis of PCo1, PCo2, and PCo3 accounted for 28.94% and 74.81% of the total change in the unweighted and weighted PCoA, respectively. The ANOSIM results also revealed signi cant differences in the bacterial community composition between the different sites (p < 0.05).
In all the soil samples, the dominant bacteria phyla with average relative abundances of > 1% were Proteobacteria (38.91%), Acidobacteria (16.27%), Actinobacteria (10.61%), Bacteroidetes (7.89%), Chloro exi (7.05%), TM7 (6.11%), Gemmatimonadetes (3.71%), AD3 (1.67%), Firmicutes (1.08%), and Verrucomicrobia (1.07%) ( Figure S3). We performed a one-way ANOVA to compare the individual taxa at the phylum level ( Figure 4). Compared with the control, the relative abundance of Proteobacteria and TM7 had signi cantly increased in BCF (p < 0.05), except at the 5-m site. The relative abundances of Acidobacteria and Bacteroidetes in the 15-, 25-, and 35-m sites did not signi cantly differ from that of the control (p > 0.05). However, their relative abundances in the 5-m site were signi cantly different to those of the other sites and the control (p < 0.05). The relative abundance of Gemmatimonadetes was signi cantly higher at the 5-m and 15-m sites compared with the 25-m and 35-m sites and the control (p < 0.05). Soils under BCF conditions had a lower relative abundance of AD3 compared with the control soil (p < 0.05). Furthermore, the relative abundance of Firmicutes was similar in the BCF sites and the control, except at 5 m (p > 0.05).

Correlating P fractions with the bacterial community
The results of the Pearson correlation analysis inferred the relationship between soil parameters and the dominant bacteria phyla ( Figure S4). Among the different soil chemical properties, Proteobacteria and The RDA showed that the soil chemical properties and P fractions explained 44.79% of the variations in soil bacterial community structure ( Figure 6). According to the Monte Carlo permutation test (permutation = 999), the Pi fractions, Po fractions (except conc. HCl-P), soil properties, and phosphatase were signi cantly correlated with the structure of bacterial communities. In Figure 6, a longer arrow represents a stronger relationship between the soil variables and bacterial community composition. Therefore, according to the variance of r 2 , Pi fractions had a stronger correlation with bacterial community changes than Po fractions (Table S1). Furthermore, among all soil parameters, AK and pH showed the strongest correlation with the changes in bacterial community (r 2 = 0.963 and 0.962, respectively), followed by 1 M HCl-Pi and Resin-Pi in the Pi fractions (r 2 = 0.958 and 0.938, respectively).

Effect of BCF on P fractions
Labile Pi (Resin-Pi and NaHCO 3 -Pi) is the main source of P for plant growth and can be directly absorbed and utilized by plants . In our study, we found that the soil labile Pi content in BCF increased signi cantly compared with the control and also increased with higher grazing density. This trend may be attributed to the soil adsorption of P from manure accumulation (Neufeldt et al., 2000). The increase in P content from manure is more signi cant for labile Pi than other P fractions (Maranguit et Marcos et al. (2019) found that grazing can alter soil properties by affecting litter decomposition, manure deposition, and soil compaction, thereby altering soil microbial compositions. In this study, we found that BCF increased the relative abundance of Proteobacteria, likely owing to an increase in soil available nutrients and carbon (Goldfarb et al., 2011;Fierer & Jackson, 2006;Fierer et al., 2007). In addition, previous studies have shown that most Proteobacteria and Actinobacteria are oxygen-dependent (Emerson et al., 2010;Hamamura et al., 2006). Therefore, relatively high grazing density can lead to soil compaction, which decreases the soil oxygen content (Martıńez & Zinck, 2004;Shah et al., 2017); this may explain the relatively low abundance of Proteobacteria and Actinobacteria at the 5-m site ( Figure 4). Furthermore, Wang et al. (2016) found that the relative abundance of Actinobacteria correlated with the soil C:N ratio. In our study, the TN content increased with increasing grazing density, but SOC showed no signi cant change, indicating a gradual decrease in the C:N ratio. This may be another reason for the signi cantly lower Actinobacteria abundance at the 5-m site. Bacteroides was more abundant in the BCF soils, especially at 5 m, which may be owing to the high abundance of Bacteroides in the digestive system of chickens under BCF conditions . Bacteroides is also involved in the degradation of complex carbohydrates and has been signi cantly positively correlated with soil respiration (Thomas et al., 2011;Wang et al., 2016). Therefore, the gradual increase in Bacteroides with increasing grazing intensity may aggravate the loss of SOC. Costello (2007) found that members of AD3

Effect of BCF on the P-associated bacterial community structures
have oligotrophic or micronutrient-requiring characteristics. Thus, our results show that chicken farming signi cantly reduces the relative abundance of AD3, which suggests that BCFs may be able to alleviate soil nutrient de ciency.
In this study, we observed signi cant differences in the relative abundance of phosphobacteria (e.g., Flavobacterium, Pseudomonas, Streptomyces) between the BCF and control soils. Flavobacterium is a relatively common bacteria (Soltani et al., 2010) that stimulates plant growth and promotes the developmental plasticity of plant roots through auxin production (Verbon & Liberman, 2016;Zhao et al., 2019). Flavobacterium is also a highly active phosphobacteria that dissolves phosphate by producing organic acids (Nahas, 1996;Cunningham & Kuiack, 1992). We found that the relative abundance of Flavobacterium was signi cantly related to Pi (Resin-Pi, NaHCO 3 -Pi, and 1 M HCl-Pi) ( Figure 5), which suggests that the bacteria enhances soil P bioavailability. Moreover, the signi cant enrichment of Pseudomonas and Arthrobacter in BCF soils may help to mobilize sparingly labile P, as their relative abundances were signi cantly positively correlated with recalcitrant and labile P. The signi cant positive correlation between Streptomyces and the Pi fraction and the lower labile Po fractions (such as NaOH-Po and conc. HCl-Po) suggests that the bacteria was involved in the mineralization of lower labile Po as well as the microbial immobilization of Pi to lower labile Po. Similar results were observed by Battni et al. (2017;. In our study, the signi cant negative correlation between Burkholderia and the P fractions may been linked to the accumulation of chicken manure, which alkalizes the originally acidic soil. Although chicken manure increases the soil P content, a more alkaline soil is known to inhibit the activity of Burkholderia (Pereira et al., 2013;Stopnisek et al., 2014). Bacillus is one of the most important phosphobacteria (Hayat et al., 2010) and has strong resistance to hostile environments (Zhang, 1990). In this study, the relative abundance of Bacillus did not differ signi cantly between the different sites and the control and was not signi cantly correlated with soil properties. This suggests that the soil environment in the BCF sites did not in uence Bacillus activity, which supports its high resistance to changing environments.

Effect of BCF on phosphorus availability and sustainable land use
BCF signi cantly enhanced the adsorption capacity for cations (CEC) and improved the fertilizer retention capacity in soil (Ge et al., 2014). Additionally, the increased soil CEC under higher grazing densities may be due to the ner soil texture (Elliott et al., 1986;Teague et al., 2011). We observed no signi cant difference in SOC between the different grazing densities, which may be due to a number of factors: 1) higher grazing densities may damage the soil structure, which increases the loss of water-soluble organic carbon ( . Therefore, the lower C:N and C:P ratios and the higher labile and moderately labile P contents under higher grazing densities indicate a possible risk of P leaching. In our study, the gradual decrease in the relative content of Po, the increased phosphatase activity, and the increased abundance of phosphobacteria with increasing grazing density suggest that chicken farming promotes the mineralization of Po to Pi. However, further long experiments in this ecosystem are required to determine whether the decrease in Po will eventually lead to P de ciency and become a limiting factor in the P cycle. Among the different soil properties, pH had the highest impact on bacterial community structure, likely owing to the narrow pH range for optimal bacterial growth (Ragot et Figure 5); this likely further in uenced the overall soil microbial composition.

Conclusion
We found that chicken farming in bamboo forests signi cantly altered both the soil P distribution and the soil bacterial community structure. Labile and moderately labile P signi cantly increased with increasing grazing intensity, which suggests that chicken farming effectively enhances soil P bioavailability.
Furthermore, grazing intensity could cause the redistribution of soil P and promote the mineralization of Po, as evidenced by an increase in the relative contents of labile and moderately labile P, a decrease in the relative content of Po, an increase in phosphatase activity, and changes in phosphobacteria abundance.
Our results suggest that phosphobacteria regulated the soil P cycle, mineralized lower labile Po (Streptomyces), and converted insoluble phosphate to soluble phosphate (Pseudomonas and Arthrobacter). Overall, our ndings indicate that BCF can alleviate P de ciency in subtropical forests and increase the soil P supply capacity and potential. However, high grazing density or long-term BCF practices can lead to soil nutrient imbalances and P leaching, which impacts the sustainability of BCF systems. Declarations Z.Z. and X.Z. conceived the study. X.G., S.L., X.Z. and C.Y. performed the eldwork. X.G., F.B and S.L. measured the soil samples and collaborated with data analysis. X.G. drafted the manuscript and performed statistical analyses. All authors read and approved the nal manuscript. Availability of data and materials

Abbreviations
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.
117. Zhu, Y. (2012). Study on economic e ciency of forest-chicken compound management patterns in the lower Yellow River. Chinese Agricultural Science Bulletin, 28, 70-73.  Percentage of each phosphorus (P) fraction in labile P (a), moderately labile P (b), sparingly labile P (c), and total P (d) at the different grazing sites.

Figure 3
Acid phosphatase (a) and alkaline phosphatase (b) activity in the bamboo forest soil under different grazing distances from the hen house. The error bars indicate the standard error of the mean (n = 5).
According to Duncan's test, the different letters indicate signi cant differences at p < 0.05.

Figure 4
Differences in the dominant phyla (average relative abundance > 1%) of soil bacteria at different distances from the hen house.

Figure 5
Pearson correlation analysis between soil phosphorus fractions and the phosphobacteria genera. Figure 6 Redundancy analysis of the selected soil properties, phosphorus fractions, and phosphatase for bacterial community structures.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Supplementarymaterial.docx