Leaf and root litter species identity influences bacterial community 1 composition in short-term litter decomposition

6 Background: Microorganisms play a crucial role in litter decomposition in terrestrial ecosystems. 7 However, it remains unclear, which effects of leaf litter and root species on bacterial community 8 composition and diversity after one year's decomposition. 9 Methods: The leaf and fine roots litters of Robinia pseudoacacia , Quercus acutissima , Pinus 10 tabulaeformis and Pinus densiflora , which are the dominant afforestation species in Mount Tai, were 11 analysed using the Nylon litterbag method and Illumina Miseq high-throughput sequencing for the 12 amplification of bacterial 16S rRNA V4-V5. We measured the remaining litter mass and the bacterial 13 community composition and assessed the effects of leaf and root litter species on the bacterial 14 community after one-year decomposition periods. 15 Results: (1) The remaining masses of leaf and fine roots litters of the four plant species were 16 significantly influenced by organ type and species. The remaining mass of fine root litter was smaller 17 than that of leaf litter for broad-leaved species, and the opposite result was found for coniferous species. (2) The observed species Chao1 and phylogenetic diversity values were significantly lower for leaf litters than for fine root litter. The community richness index was positively correlated with the C content, C:N and lignin content and negatively correlated with N:P, N content and P content. The 21 bacterial community structure differed significantly among leaf and root litter decomposition for the 22 four species ( p <0.05). The bacterial community structure in leaf litter was most highly correlated with 23 the initial N content and N:P. The bacterial community structure in fine roots was most highly 24 correlated with the lignin content. (3) The bacterial phyla Bacteroidetes , Acidobacteria and 25 Gemmatimonadetes were significantly affected by litter and species type, and the relative abundances 26 of Firmicutes and Chloroflexi were only affected by litter type. The relative abundances of 27 Acidobacteria , Firmicutes and Chloroflexi in fine root litter were higher than those in leaf litter, while 28 the opposite result was found for Bacteroidetes . The bacterial genera Burkholderia-Paraburkholderia , 29 Sphingomonas and Mucilaginibacter were affected by litter type ( p <0.05). The relative abundance of 30 Burkholderia-Paraburkholderia in fine root litter was higher than that in leaf litter, while the opposite 31 result was found for Bradyrhizobium , Sphingomonas and Mucilaginibacter . Pearson correlation 32 analysis showed that the average relative abundance of the dominant phyla and genera was affected by 33 the initial litter properties, especially for Bacteroides , Acidobacteria , Burkholderia , and Sphingomonas . 34 Conclusions: Litter type, interaction between litter type and species were important than species in 35 shaping the bacterial diversity and community composition in decomposing litter. And this were 36 affected by initial chemical properties of the litter.

2 content, C:N and lignin content and negatively correlated with N:P, N content and P content. The 21 bacterial community structure differed significantly among leaf and root litter decomposition for the 22 four species (p<0.05). The bacterial community structure in leaf litter was most highly correlated with 23 the initial N content and N:P. The bacterial community structure in fine roots was most highly 24 correlated with the lignin content.  shown that bacteria are more resilient than fungi during later periods of litter decomposition (Wardle et 85 al. 2004;Chapman and Koch 2007). In view of the importance of the soil bacterial community 86 structure and diversity in ecosystems, these topics have received increasing attention from researchers 87 (Prescott and Grayston 2013;Guo et al. 2018). Therefore, we analysed the effects of leaf and root litter 88 species on the bacterial diversity and community composition in decomposing litter for four dominant 89 afforestation tree species in Mount Tai using high-throughput sequencing technology, which will 90 provide more comprehensive and complete information on the microbial community structure at a fine 91 resolution (Hong et al. 2015;Sauvadet et al. 2019). We also clarified the effects of microbial activities 92 and initial chemical properties on leaf and root litter and their decomposition, which provides a 93 theoretical basis for the microbe-driven mechanism of litter decomposition. We hypothesized that (1) 94 leaf and root litter species strongly influence the bacterial diversity and community composition of the 95 litter during decomposition, and that these effects influence the decomposition rate. (2) There was a 96 significant correlation between bacterial diversity and community composition and the initial chemical 97 properties of the litter.

117
Litter was collected in pure stands of RP, QA, PD and PT. At the beginning of October 2015, when 118 most of the litter fall occurred, fresh and intact leaf litter was directly collected from the forest floor, 119 air-dried for 10 d and stored for a week at room temperature (15-25℃). Fine root decomposition was 120 carried out using live roots with diameters less than 2 mm because it was difficult to separate fresh 121 roots from those already having decomposed for a period. In October 2015, fine roots (≤2 mm in 122 diameter) were excavated using shovels from the topsoil (0-20 cm depth) of pure stands where most 123 fine roots occur. Roots were transported to the laboratory, and the surface soil was removed by washing 124 in tap water and then in deionized water. To calculate the air-drying/oven-drying ratio of the 7 decomposition substrate, a small portion of the sample was oven-dried at 65℃ to a constant weight.

126
Then, we determined the carbon content (C), nitrogen (N), phosphorus (P), and lignin content in the 127 initial litter.

128
A field experiment was conducted using the litter bag method. Air-dried litter samples (4 g for fine 129 roots and 6 g for leaf litter) were enclosed in litter bags (15 × 15 cm) made of 1-mm nylon mesh.

130
Subsamples of the initial litter were oven-dried (65℃ for 48 h) to calculate the correction factor for 131 converting the air-dried mass to the water-free dry mass. In July 2016, the litter bags were placed in six 132 blocks using a randomized complete block design. Each block included all eight treatments, for a total 133 of 48 samples. The size of the blocks was 10 m × 10 m with 5 m × 5 m isolation zones between blocks.

134
Litter bags were placed in the forest-free area. Litter bags with leaf litter were pinned to the ground 135 surface to prevent movement by wind using U-shaped nails. Litter bags with root litter were inserted  After determining the dry weight, samples were ground to pass through a 1 mm mesh, and then 145 the total C and N contents in the litter were determined by an elemental analyser (ECS4010, Costech, 8 Italy). The total P contents were analysed by a continuous flow analyser (PROXIMA, Alliance, France).

147
We used ultraviolet spectrophotometry colorimetry to determine the lignin contents (Iiyama and Wallis 148 et al. 1988) and determined the ash content by igniting the oven-dried material for 6 hours at 600℃ in a 149 muffle furnace to the correct dry weight (Gupta and Singh 1981). cycling conditions were as follows: an initial denaturation at 98℃ for 1 min, followed by 30 cycles of 161 denaturation at 98℃ for 10 s, annealing at 50℃ for 30 s, and extension at 72℃ for 30 s, followed by 162 72℃ for 5 min. The same volume of 1X loading buffer (containing SYBR green) was mixed with the 163 PCR products, and the mixture was submitted to electrophoresis in a 2% agarose gel. Samples with 164 bright bands between 400 and 450 bp were chosen for downstream analyses. PCR products were mixed 165 in equal density ratios. Then, the mixed PCR products were purified with the Gene JET Gel Extraction 166 Kit (Thermo Scientific). For the generation of sequencing libraries, the NEB Next ® Ultra TM DNA 9 Library Prep Kit for Illumina (NEB, USA) was used, and the index codes were added under the 168 guidance of the manufacturer's recommendations. A Qubit ® 2.0 Fluorometer (Thermo Scientific) and 169 Agilent Bioanalyser 2100 system were used to assess the quality of the library. Library sequencing was 170 implemented on an Illumina HiSeq platform, and 250 bp/300 bp paired-end reads were generated.

190
The initial litter chemistry was controlled by leaf and root litter of different species (Table S1). In 191 addition, there were significant differences in initial chemistry between leaf litter and root litter of the 192 same species, especially for the N and P contents, C: N and N:P (Table 2). For the broad-leaved species 193 (RP and QA), the P content in the leaf litter was lower than that in the fine roots, but opposite results 194 were observed for coniferous species (PD and PT). Interestingly, there were significant differences 195 between leaf and root litter for QA, and the litter quality of the fine roots was significantly higher than 196 that of the leaves (i.e., fine roots had a higher N content and lower C: N). In addition, significant 197 differences were found for the leaf or root of different species (Table 2). Among the four fine root 11 mass percentage than those of coniferous species. The decomposition rates of the fine roots ranked as 210 follows: RP>QA>PT>PD. For leaf litter, the decomposition of RP was fastest, while the decomposition 211 of QA was slowest. The decomposition rates of the fine roots ranked as follows: RP>PT>PD>QA ( Fig.   212 2). In addition, there was a marked positive correlation between the remaining mass and the initial C 213 content and C:N (p<0.01), but the remaining mass and N content and N:P showed a significantly 214 negative correlation (p<0.01, Table 5).

216
The bacterial alpha-diversity indices of the litter were significantly different and were affected by 217 litter type, species and their interactions, except for the Shannon-Wiener index (Table 3, S2). All 218 sample coverage values were higher than 96%, suggesting that the sequence data well reflected the 219 microbial community composition. After one year's decomposition, the observed species Chao1 and 220 phylogenetic diversity (PD) values for the fine roots were higher than those for the leaf litter (Table 3).

221
In addition, these indices were significantly lower for the fine roots of broad-leaved species than for 222 those of coniferous species (p<0.05); however, opposite results were found for the leaf litter (Table 3, 12 showed a significantly positive correlation with the C:N and lignin content. The PD index was 231 significantly positively correlated with the C content, C:N and lignin content and was significantly 232 negatively correlated with the N:P, N content and P content (Table 4).

244
The bacterial phyla Bacteroidetes, Acidobacteria and Gemmatimonadetes were significantly 245 affected by leaf and root litter species (Table S3, p<0.05). The relative abundance of Bacteroidetes and 246 Acidobacteria demonstrated a significant response to different leaf and root litter species (Table S3).

247
The relative abundance of Bacteroidetes in fine root litter was lower than that in leaf litter, while the 248 opposite result was found for Acidobacteria, especially for QA (Fig. 4A, 4B). The relative abundance 249 of Gemmatimonadetes in RP leaf litter was significantly higher than that in the other three leaf litters 250 (p<0.05), but there was no obvious difference among the four fine root treatments (Fig. 4C). The 251 13 relative abundance of Firmicutes and Chloroflexi were only correlated with the leaf and root litter 252 (Table S3). The relative abundance of Firmicutes in fine root litter was higher than that in leaf litter, but 253 the difference was not significant (Fig. 4D). There was significant difference between the PT leaf and 254 root litter for the abundance of Chloroflexi (p<0.05, Fig. 4E).

255
The bacterial genera Burkholderia-Paraburkholderia and Mucilaginibacter were significantly 256 affected by litter type and species (Table S4, (Table S4). The relative abundance of 259 Burkholderia-Paraburkholderia in fine root litter was higher than that in leaf litter, while the opposite 260 result was found for Bradyrhizobium, Sphingomonas and Mucilaginibacter (Fig. 4F, 4G, 4H and 4J).

261
The relative abundance of Rhizomicrobium was significantly higher for QA and PT than that for RP and 262 PD (Fig. 4I).

263
There was no significant correlation between the relative abundances of Proteobacteria, 264 Cyanobacteria, Verrucomicrobia, Firmicutes and Actinobacteria and the initial litter chemistry (p>0.05, 265 Table 5). The relative abundance of Bacteroidetes had a significantly positive correlation with N 266 content and N:P (p<0.05) and a significantly negative correlation with lignin content (p<0.01). A 267 significantly positive correlation was observed between the relative abundance of Acidobacteria and 268 the initial lignin content (p<0.01), but a significantly negative correlation was observed with the N 269 content and N:P (p<0.05). A significantly positive correlation was observed between the relative 270 abundance of Planctomycetes with C:N and remaining mass (p<0.05), but a significantly negative 271 correlation was observed with N content and N:P (p<0.05). The relative abundance of 272 14 Gemmatimonadetes had a significantly negative correlation with P and C content (p<0.05). The relative 273 abundance of Chloroflexi had a significantly positive correlation with C: N (p<0.05, Table 5).

274
Correlation analysis indicated that the relative abundances of Burkholderia-Paraburkholderia and 275 Rhizomicrobium had a significantly negative correlation with the initial N content and N:P (p<0.05, 276 Table 6). There was no significant correlation between the relative abundance of Bradyrhizobium,  Table 6). A significantly positive correlation was observed between the relative abundance of 282 Chitinophaga and the N content (p<0.05) but a significantly negative correlation was observed with the 283 lignin content (p<0.05, Table 6).

285
The NMDS analysis of the bacterial community structure showed that different treatments were 286 clearly distributed in different quadrants, indicating a significant difference in the bacterial community 287 structure (Fig. 5). The results from the ANOSIM nonparametric test also showed that the bacterial 288 community structure in leaf litter was significantly different from that in fine roots (R=0.5208; p=0.03).

289
The redundancy analysis (RDA) of the bacterial community structure and the initial litter chemistry 290 showed that the initial N:P had the greatest impact on the bacterial community structure, followed by 291 the lignin content and N content (Fig. 6). The bacterial community structure in leaf litter was most 292 highly correlated with the initial N content and N:P. The bacterial community structure in fine roots 293 15 was most highly correlated with the lignin content (Fig. 6).

295
Effect of litter type and species on the bacterial diversity and decomposition rate

296
We found that the bacterial diversity was affected by litter type and species, and the leaf litter 297 bacterial diversity of coniferous species was lower than that of broad-leaved species (Table 3, (Table S1). In addition, there were significant differences in the bacterial 305 community structure between leaf and root litter (Fig. 5). The difference of micro-environment in leaf 306 and root litter decomposition significantly affect the microbial community. The higher humidity of the 307 soil environment was beneficial to microbial growth (Banerjee et al. 2016). These may be important 308 reasons for the significant differences in the bacterial community structure and decomposition rates 309 between the leaf and root litter (Fig. 2, 5).

310
Decomposer activity and the litter decomposition rate are highly dependent on litter quality 311 (Zhang et al. 2016;Lin et al. 2019). There were obvious differences in the remaining mass between leaf 312 and root litter from four dominant afforestation species in Mount Tai (Fig. 2). The remaining mass was 313 significantly positively correlated with the initial C content and C: N and was extremely negatively 314 16 correlated with the N content and N:P (Table 5) ). However, we found that there was no significant correlation between the relative 337 abundances of Proteobacteria and Actinobacteria and the initial litter chemistry (Table 5). In addition, 338 leaf and root litter species had no effect on the relative abundances of Proteobacteria and 339 Actinobacteria (Table S3). A possible explanation of these results is the "functional breadth 340 hypothesis", i.e., the ability of soil biota to efficiently decompose all litter types at the same time 341 (Keiser et al. 2014;Fanin et al. 2016). Here, we found no significant difference in the relative 342 abundances of Proteobacteria and Actinobacteria among all litters (Table 5, Fig. 4), suggesting that the 343 decomposer community had a broad functional ability to decompose various litter types (Lin et al.

344
2019). We found that the relative abundance of Bacteroidetes in fine roots was lower than that in leaf 345 litter, especially for QA and PD (Fig. 4A), and had a significantly negative correlation with the initial 346 lignin content (Table 5) Proteobacteria, but root addition promoted Actinobacteria. Interestingly, the results were opposite for 349 Acidobacteria, with the relative abundance being higher in fine root litter than in leaf litter, especially 350 for QA (Fig. 4B). Moreover, there was a significantly positive correlation between the relative 351 abundance of Acidobacteria and the initial lignin content (Table 5) (Yamada et al. 2009). We found that the relative abundances of Firmicutes and Chloroflexi were 370 affected by the leaf and root litter (Table 5). One possible explanation for this finding is that the initial 371 litter chemistry and physical positions of the leaf and root were different (Table 2).

373
At the genus level, the relative abundances of Burkholderia-Paraburkholderia (6.0%), 374 Sphingomonas (3.7%), Bradyrhizobium (3.2%) and Rhizomicrobium (3.0%) were higher than those of 375 other genera and were affected by litter type and species (Fig. 3, S4). Burkholderia-Paraburkholderia 376 and Rhizomicrobium, which were reported to participate in N cycling, are members of denitrifier and 377 19 N2 fixation taxa and require a high N availability (Cheng et al. 2017;Nie et al. 2018 Burkholderia-Paraburkholderia and Rhizomicrobium and the initial litter N content and N:P (Table 6).

380
One possible explanation for this finding is that a high initial N content in the litter could increase N 381 release and then decrease N availability after litter decomposition (Mooshammer et al. 2012). In this 382 study, the relative abundance of Sphingomonas, which had a positive correlation with the initial N 383 content and N:P but a negative correlation with the C: N and initial lignin content (