Response of soil bacteria on habitat-specialization and abundance gradient to different afforestation types

Studies involving response of subgroups of soil microorganisms to forest change, especially comparative studies on habitat-specialization and abundance gradient were still lack. In this study, we analyzed the response of soil bacterial diversity and structure to afforestation types and its relationship to environment of Fanggan ecological restoration area under the classification of subgroups by habitat-specialization and abundance gradient based on abundance ratio respectively. The results were: (1) On the habitat-specialization gradient, the variation of OTUs species number and abundance was consistent and positively correlated with habitat-specialization; on the abundance gradient, the variation was opposite and OTUs species number was negatively correlated with abundance gradient; (2) The distribution frequency of each subgroup on both gradients was the highest in broad-leaved forests, but the abundance was the opposite. The distribution frequency of the same stand showed no difference among habitat-specialization subgroups, but the abundant subgroup in broad-leaved forests was the highest among the abundance subgroups; (3) α-diversity was positively correlated with habitat-specialization but negatively with abundance, with the highest mostly in broad-leaved and mixed forests; (4) Community structure among stands on habitat-specialization gradient showed no significant difference, but that of rare subgroup between broad-leaved forests and other stands significantly differed. Plant diversity and vegetation composition correlated stronger with community structure than spatial distance and soil physicochemical properties on both gradients. Our results provided a new perspective for revealing the effects of afforestation types on soil bacteria from the comparison of habitat specialization and abundance gradient.

Forest ecosystems played a crucial role in generating primary productivity, regulating nutrient biogeochemical cycles and providing a variety of ecological services 1 .But with the rapid development of economy and society, a large number of natural forests have been deforested to meet the demand for wood, which has been seriously damaged the forest ecosystems 2,3 .According to historical records, the primary forests in Shandong were almost cut down in the Qin and Han dynasties, and the secondary forests were also destroyed repeatedly.Therefore, many areas of Shandong province had been bare rocks, and some had been artificial or secondary forests 4 .Recent studies have shown that the barren hills in the northern Shandong central mountain area of began to suffering increasing forest cutting and land reclamation activities in the fifteenth century 5 .During the sixteenth century to the eighteenth century, the local virgin forests were seriously destroyed or even cleared away, left barren mountains 5 .To alleviate the crisis of the deforestation of natural forest, plantations have been rapidly rised and related researches have attracted the attention of researchers.However, plantations in Shandong central mountain area began late, in the 1960s.Due to poor soil nutrients, conifers such as Platycladus orientalis and Pinus tabulaeformis were mainly planted 6 .
Soil microorganisms are key components that link the nutrient composition and cycle between above and below ground of forest ecosystem 7,8 .In recent years, interests in the factors that shape and control soil microbial community composition and diversity have been spring up 9 , especially since the rise of high-throughput sequencing.Factors such as vegetation type, soil physicochemical properties, plant diversity and spatial distance

Study site
Fanggan ecological restoration area, locate in the northern part of Shandong Central Mountain area (117°24′45″-117°28′5″E, 36°24′23″-36°26′44″N), was mountainous and hilly terrain with the highest altitude 860 m and more than 30 hilltops above 400 m.The climate type of the restoration area is continental monsoon climate with distinct features of four seasons and the same period of rain and heat.The average annual temperature was 12.4 °C, the lowest extreme temperature was − 22.5 °C and the average annual precipitation was more than 830 mm, concentrate in July to September.The soil type in the restoration area was mountain brown soil, and the average thickness of litter was 4 cm.The vertical structure of vegetation community was obvious, but the species were relatively simple.The tree layer mainly included Pinus densiflora, P. tabulaeformis, Populus davidiana, Robinia pseudoacacia and Diospyros lotus, the shrub layer was mainly consisted of Vitex negundo var.heterophyll, Ziziphus jujuba var.spinosa, Grewia biloba and the regeneration seedlings of the tree layer, the herbaceous layer was consisted mainly of some species of Gramineae, Cyperaceae and Compositae.The plantations in the restoration area were all planted between 1975 and 1985, with relatively consistent forest age.

Vegetation survey and soil sampling
Through literature review and field investigation, forests of the restoration were mainly distributed in the altitude range of 300-750 m, so the vegetation community investigation was carried out in dense forest canopy region of the altitude range.Because the original planting area of different forest types was quite different, this study allocated the quadrat quantity according to the actual area proportion of each forest type.A total of 20 quadrats including 17 quadrats of 3 forest types of plantations, namely, 3 coniferous forests (CF), 3 coniferous broadleaved mixed forests (MF) and 11 broadleaved forests (BF) as well as 3 shrubs quadrats as comparison were set up.In order to reduce the influence of topographic factors, each quadrat contained a consistent amplitude of topographic variation as far as possible, and the geographical coordinates, elevation, slope direction and slope have been recorded (Table 1).The area of tree layer quadrat was 10 m × 10 m, individuals with DBH ≥ 3 cm were measured per tree, and the species name, coverage, quantity, base diameter and height of shrubs as well as species name, height and coverage of herbs were also recorded.The survey area of each shrub quadrats was 5 m × 5 m, and the statistics of herb layer were carried out.The α-diversity indexes calculated of vegetation community includes Shannon-winner index (Eq.1), Simpson diversity index (Eq.2), Shannon evenness index (Eq.3), Simpson evenness index (Eq.4), Pielou evenness index (Eq.5) and species richness.0-10 cm topsoil and 20-30 cm subsoil were collected in each quadrat according to the five-point sampling method with sampling points arranged at the center and four corners (3 m from the quadrat's vertex) of the quadrat 35,42 .The samples from same layer at 5 points in each quadrat were thoroughly mixed and divided into two portions.So, there were totally 80 soil sample, for one part was stored in liquid nitrogen for molecular biology extraction and analysis, and the other was stored at room temperature and brought back to the laboratory for physicochemical analysis as soon as possible.
( In the above equations, m represents the number of species in the quadrat, N represents the total number of individuals of all species, and Pi represents the important value of species i in the quadrat.The diversity index of a stand is equal to the average value of the index for the various sides of the stand.

Determination of soil physicochemical properties
Soil moisture and dry matter content were determined by gravimetric method 43 , pH was determined by potentiometric method using 2.5:1 water-soil ratio.Organic carbon content was determined by potassium dichromate oxidation-spectrophotometry 44 , and available phosphorus was determined by molybdenum-antimony anti-color spectrophotometry 45 .We used potassium chloride solution extraction-spectrophotometry and ultraviolet spectropotometry to determine ammonium nitrogen (NH 4 + -N) and nitrate nitrogen (NO 3 − -N) respectively.The content of nitrite nitrogen (NO 2 − -N) was determined by naphthalene ethylenediamine hydrochloride color spectrophotometry 46,47 .The model of the spectrophotometer is T6 UV-visible spectrophotometer produced by Beijing Puxi General Instrument Co., LTD.The pH meter model is SEvenExcellence S400-Basic made by Mettler Toledo Corporation.

High throughput sequencing of soil microorganisms
Soil bacterial DNA was extracted by TGuide S96 magnetic bead method and 16S rDNA fragments of DNA were amplified by primers 515F (50-GTG YCA GCMGCC GCG GTAA-3) and 926R (50 -CCG YCA ATTYMTTT RAG TTT-30).The PCR products were purified by the VAHTS-TM DNA Clean Beads method and quantified.Unique barcode was used to separate each sample to prevent cross-contamination.Sequencing and data processing (1) Filter of raw data quality Trimmomatic 48 was used to filter dual-end sequencing files.Parameter setting: Window size was set as 50 bp.The reads will be cut from the start of the window once average Q-score within the window is lower than 20.(2) Identification and removal of primer sequences Cutadapt (Version 1.9.1) was used to identify primer sequences according to the parameters allowing the maximum error ratio of 20% and the minimum coverage of 80% 49 .(3) Double-ended reads splicing USEARCH (Version 10) was used for sample double-ends reads stitching according to the minimum overlap length of 10 bp, the minimum similarity of 90% allowed in overlap area and maximum error base number of 5 bp 50 .(4) Removal of chimera The criteria of chimera are as follows: divide query sequence into chunks without overlap and compare them with the database; The best match of each chunk in the database is selected, and the two best parent sequences are finally selected.The sequence to be detected was compared with the two parents.If a sequence of the two parents has greater than 80% similarity to the Query sequence, the query was judged to be a chimera.Use UCHIME (Version 8.1) to remove chimeras 51 .
(5) OTUs clustering and species annotation Cluster analysis was performed using UPARSE at the 97% similarity level, and the usearch command was used to remove affine sequences and singleton OTUs in the process.Each clustered OTUs was annotated into seven subgroupomic classes of boundary, phylum, order, family, genus and species by the sine method through the 16 s in silva or the ITS database in Unite.The community abundance tables for each rank were obtained.The present study was carried out mainly on the OTUs subgroupomic level and the following analysis were all based on the OTUs abundance table.

Classification on abundance gradient
In this study, to estimate the relative abundance of abundant and rare subgroups the thresholds were defined as 1 and 0.01% of the total sequence reads 52,53 .Further to subdivide bacterial communities, the six categories were defined with reference to recent publications 54,55 : (1) permanent abundant subgroup (AAT): relative abundance great than 1% in all samples, (2) permanent rare subgroup (ART): relative abundance less than 0.01% in all samples, (3) conditional medium abundance subgroup (MT): relative abundance between 0.01 and 1% in all samples, (4) conditionally abundant subgroup (CAT): relative abundance great than 0.01% in all samples and 1% in some samples, (5) conditionally rare subgroup (CRT): relative abundance less than 0.01% in some samples but never great than 1% in any sample, (6) conditionally rare and abundant subgroup (CRAT): relative abundance ranging from 0.01% to 1%.The above AAT and CAT were classified as abundant subgroup, ART and CRT as rare subgroup, and MT and CRAT as medium abundance subgroup.www.nature.com/scientificreports/Classification on habitat-specialization gradients Levins niche width (Eq.6) was calculated using the 'spaa' package of R v 4.0.0 56 .The frequency of occurrence of each OTU was randomized and rearranged 1000 times using the replacement method of the 'EcolUtilis' package, respectively.Then the zero distribution of the niche width index was calculated for each OTU.Based on the consensus that habitat-generalized subgroup owe a wider niche width than the habitat-specialized subgroup 40 , OTUs were classified into a habitat-generalized or a specialized group based on whether the frequency of observed occurrence exceeded the upper 95% confidence interval or fell below the lower 5% confidence interval, and the OTUs were distributed to the habitat-neutral group if the observed niche width was within in the 5% to 95% confidence interval range 18 .

Bacterial community diversity calculation
Based on the OTUs abundance matrix, the Shannon-Winner diversity indexes and OTUs species richness of each habitat-specialization subgroups and abundant subgroups were calculated.The non-metric multidimensional scaling (NMDS) function of 'vegan' package was used to analyze differences in bacterial subgroup community structure between afforestation types, and the significance of structural differences was tested by Permanova test (by ANOSIM function in vegan package of R software, permutations = 999, bray).Mantel test was carried out to analyze the relationship between community structure and environmental factors such as tree species composition, plant diversity, physicochemical properties and spatial distance.The spatial distance was calculated by the three-dimensional relationship of latitude coordinates, longitude coordinates and altitude.RDA function was taken to analyze the effects of tree composition, tree diversity and soil physicochemical properties on microbial subgroup community structure.The significance of RDA model was tested by ANOVA and the Monte Carlo permutation test was taken to analyze the significance of the influencing factors of each explanatory variable in the RDA model (by anova.ccaand envfit function in vegan package of R software, permutations = 999).Plotting was done using the R 4.0.0 basic program package and 'ggplot2' package.The significance of soil physicochemical properties, niche width, abundance and α-diversity indexes among forest types were analyzed by one-way ANOVA test (SPSS 19.0).

Results and analysis
A total of 2,506,761 sequences were tested from all soil samples, and were divided into 1830 OTUs.The sequences number of 20 topsoil samples ranged from 46,595 to 73,284, and the OTUs ranged from 940 to 1579.Sequences number of 20 subsoil samples ranged from 47,705 to 70,743, and OTUs ranged from 1181 to 1568.On the habitatspecialization gradient, the OTUs numbers of habitat-generalized subgroup accounted for only 5.19%, and the abundance ratio was only 1.70%.The habitat-specialized subgroup accounted for 65.41%, and the abundance ratio was also as high as 83.23% in topsoil.The OTUs number of the habitat-generalized subgroup accounted for only 5.79% and the abundance ratio was only 1.83%, while the OTUs number of habitat-specialized subgroup was as high as 62.79% and the abundance ratio was as high as 81.88% in subsoil.On the abundance gradient, the OTUs numbers of abundant subgroup in the topsoil accounted for only 3.20%, but the abundance ratio was as high as 47.38%.OTUs numbers of rare subgroup accounted for 93.82%, but the abundance ratio was only 41.21%.The OTUs numbers of abundant subgroup in subsoil only accounted for 1.87%, but the abundance ratio was as high as 36.55%.The OTUs numbers of rare subgroup accounted for as high as 94.84%, but the abundance ratio was only 47.22%.Soil physicochemical features were shown in Table 2.

Effects of afforestation types on the distribution frequency and abundance of soil bacteria along habitat-specialization and abundance gradients
Distribution frequency of soil bacteria was presented by the niche width.Results showed the distribution frequency of each habitat-specialization subgroup in topsoil and subsoil was significantly the highest in broad-leaved forests (P < 0.01), and even the frequency of habitat-neutral and specialized subgroups in coniferous forests (6) and mixed forests was slightly lower than in the shrub (Fig. 1).The distribution frequency of each abundance subgroup in topsoil and subsoil was also increased obviously in broad-leaved forests, especially for abundant subgroups (P < 0.01).The niche width of the abundant subgroup in broad-leaved forests also showed higher than other subgroups (Fig. 2).On the habitat-specialization gradient, the abundance of each subgroup in topsoil has increased with habitatspecialization.The abundance of the same habitat-specialization subgroup in the topsoil and subsoil showed consistent with the stand changes.The abundance of habitat-generalized and neutral subgroups was higher in shrub and mixed forests than in the coniferous forests and broad-leaved forests and that of habitat-specialized subgroup showed the highest in the shrub and the lowest in the broad-leaved forests, and significantly differ (P < 0.05) (Fig. 3).On the abundance gradient, the abundance of abundant and medium abundance subgroups in the topsoil and subsoil both significantly decreased, whereas that of rare subgroup increased in broad-leaved forests.This indicated the soil conditions of broad-leaved forests were more comfortable for most species of the rare subgroup.The abundance of abundant subgroup of both soil layer in coniferous forests and mixed forests were significantly higher than in shrub and broad-leaved forests.The abundance of medium abundance subgroup    Shannon-Winner index of habitat-generalized subgroup of each afforestation stands in both topsoil and subsoil was significantly higher than that of shrub in all tree forests (P < 0.05) (Fig. 5(TG)(SG)), but only the index of habitat-specialized and neutral subgroups in subsoil of broad-leaved forests significantly higher than other stands (P < 0.05) (Fig. 5(TS)(SS)), and the habitat-neutral subgroup of topsoil were almost unaffected by afforestation (Fig. 5(TN)).The species richness of habitat-specialized subgroup in topsoil was significantly higher in broad-leaved forests than in other stands (P < 0.01) (Fig. 5(TN)), while that of habitat-generalized and neutral subgroups was highest in mixed forests (Fig. 5(TG)(TN)).The species richness of each habitat-specialization subgroups in subsoil was higher in broad-leaved forests and mixed forests than in coniferous forests and shrub (Fig. 5(SG)(SN)(SS)).The diversity of rare subgroup in both soil layer was higher than that of abundant and medium abundance subgroups.Shannon-Winner index of abundant and rare subgroups in topsoil and subsoil was higher in broad-leaved forests and mixed forests than in coniferous forests and shrub (Fig. 6(TA)(SA)(TR) (SR)).Species richness of topsoil rare subgroup was significantly higher in broad-leaved forests than in other stands, and that of rare subgroup in subsoil was significantly higher in broad-leaved forests and mixed forests than in coniferous forests and shrub (P < 0.05).Shannon-diversity index and species richness were consistent among forest types in both topsoil and subsoil (Fig. 6(TM)(SM)).

Responses and driving factors of soil bacterial composition and structure to afforestation types along habitat-specialization and abundance gradients
For OTUs with abundance greater than 1%, species number of habitat-generalized subgroup was significantly higher than that of habitat-neutral and specialized subgroups, and the abundance distribution was more uniform in the topsoil (Fig. 7a, b, c).Among all forest types, the distribution uniformity of abundance ratio was the best in broad-leaved forests.The abundance ratio of top 3 OTUs in topsoil habitat-generalized and neutral subgroups were the highest in the coniferous forests (Fig. 7a, b), while the highest of specialized subgroup was in the shrub (Fig. 7a).The abundance ratio of each OTUs of all habitat-specialization subgroups in broad-leaved forests was the lowest, which was consistent with its higher distribution uniformity.The abundance ratio of top 3 OTUs of habitat-generalized subgroup was the highest in shrub, and the highest of the habitat-neutral and specialized subgroups was in the coniferous forests (Fig. 8a, b, c).Among the top three abundant subgroups of topsoil, the highest abundance ratio was found in the coniferous forests (Fig. 9a), the highest abundance of medium abundance subgroup was found in the mixed forests and the highest in shrub of rare subgroup (Fig. 9b), and the highest abundance was found in the shrub layer in the rare groups (Fig. 9c).In subsoil, the highest proportion of OTUs of abundant subgroup was found in mixed forests (Fig. 10a), and that of medium abundance and rare subgroups was found in shrub (Fig. 10b, c).
NMDS analysis was carried out on the community structural difference among afforestation types along habitat-specialization and abundance gradient.The 2D figures showed that the coordinate points of the rare subgroup in both layers of the broad-leaved forests were significantly separated from other stands, that is, the coordinate points of the broad-leaved forests were distributed at the upper right position of the figure, while that of other stands distributed at the lower left.The coordinate points of community structure of each habitat-specialization subgroup could not be distinguished among each stand, and the same for abundant and medium abundance subgroups (Figs.11, 12).However, in the right half of all NMDS maps, only coordinate points of broad-leaved forests were distributed, indicating that in some broad-leaved forests, bacterial community structure had been significantly different from that of other stands (Fig. 11).Further ANOSIM analysis showed that afforestation affected non-significantly on the community structure of each habitat-specialization subgroup in both topsoil and subsoil, but significantly affected the community structure of the rare subgroup in topsoil.
The environmental affection on the community structure among afforestation types of each subgroup have also been analyzed.Mantel analysis showed that plant diversity and vegetation species composition rather than soil chemical properties and spatial distance significantly affected bacterial community structure among habitatspecialization and abundant subgroups.Plant diversity significantly affected the structure of each habitat-specialization subgroup among afforestation types in both layers, and the species composition of tree layer significantly affected the structure of each habitat-specialization subgroup in topsoil and habitat-specialized subgroup in subsoil.The species composition of shrub layer significantly affected the structure of habitat-specialized subgroup in topsoil (Table 3).The species composition of tree layer significantly affected the structure of all abundance subgroups except abundant subgroup of subsoil.Plant diversity and species composition of shrub layer significantly affected the structure of medium abundance and rare subgroups in topsoil (Table 4).

Effect of species composition
On the habitat-specialization gradient, the RDA model of whole species composition of tree layer significantly explained the community structure of each habitat-specialization subgroup in both topsoil and subsoil.But the change of important value of any individual tree species did not show significant effect in the topsoil.The importance values of Robinia pseudoacacia and Morus alba significantly affected the community structure of the habitat-specialized subgroup in subsoil (Table S1, Fig. S1).The RDA model of species composition of shrub layer  significantly explained the community structure of habitat-generalized and neutral subgroups in the subsoil.Individual shrub species composition did not significantly affect the community structure of habitat-generalized subgroup in the topsoil.The significant impact factors for habitat-neutral subgroup in topsoil was Platycladus orientalis, Robinia pseudoacacia and Morus mongolica, for habitat-specialized subgroup in topsoil was Vitex negundo var.heterophylla, for habitat-generalized subgroup in subsoil was Vitex negundo var.heterophylla and Vitex negundo var.heterophylla and for habitat-neutral and specialized subgroups in subsoil was Robinia pseudoacacia and Morus mongolica (Table S2, Fig. S2).
On the abundance gradient, the RDA models of overall species composition of tree layer for explaining the community structure of medium abundance and rare subgroups of topsoil as well as abundant and rare subgroups of subsoil were significant.In the topsoil, the significant impact species for abundant subgroup were of Quercus acutissima and Pinus densiflora, for medium abundance subgroup was Populus davidiana, Diospyros lotus and for rare subgroup was Robinia pseudoacacia and Populus davidiana.In the subsoil, significant impact factors of abundant subgroup were Platycladus orientalis, Robinia pseudoacacia, Morus alba and Ziziphus mauritiana, and for medium abundance subgroup were Robinia pseudoacacia and Morus alba (Table S3, Fig. S3).The overall species composition of shrub layer could not significantly explain the changes of community structure of each abundance subgroup.Monte Carlo analysis also showed that only Vitex negundo var.heterophylla, Vitex negundo var.heterophylla and Morus mongolica significantly affected the community structure of a few subgroups (Table S4, Fig. S4).

Effect of plant diversity
The RDA model of plant diversity feature to explain the community structure of each habitat-specialization subgroup in the topsoil and subsoil was significant.Monte Carlo test results has also showed that all the diversity indices significantly affected the structure of each subgroup in the topsoil, and the evenness indices significantly affected the structure of each subgroup in the subsoil (Table S5, Fig. S5).For each abundance subgroup, all diversity indexes have significantly affected the structure changes of rare subgroup in topsoil and medium abundance subgroups in subsoil.Shannon-Winner diversity index, Simpson diversity index and Pielou evenness index had significant effects on the structure of other subgroups in topsoil and subsoil (Table S6, Fig. S6).

Effect of soil chemical properties
The RDA model of soil physicochemical properties to explain the community structure of habitat-neutral and specialized subgroups in topsoil and habitat-generalized subgroup in subsoil was significant.However, Monte Carlo test showed that the significant factors affecting the structure of each subgroup were different (Table S7).According to the RDA 2D figure, the differences of each habitat-specialization subgroup in the topsoil has been more affected by soil physicochemical properties than that in the subsoil.The coordinates community structure of the habitat-generalized and neutral subgroups of broad-leaved forests in topsoil directed the same as the increase of pH, available P and ammonia nitrogen, indicating that these physicochemical properties have significantly www.nature.com/scientificreports/changed the structure of the two subgroups of in the topsoil of broad-leaved forests, while the coordinate points of the habitat-specialized subgroup in subsoil of broad-leaved forests were contrary to the changes of ammonia nitrogen and pH, indicating that ammonia nitrogen and pH also significantly affected the structure habitatspecialized subgroup in subsoil.No significant difference has been proved in the structure of other subgroups among stands (Fig. S7).For each abundance subgroup, soil physicochemical properties significantly explained the changes of community structure of rare subgroup in topsoil.Monte Carlo test showed the structure of abundant subgroup significantly correlated with the content of available P and nitrate nitrogen, and that of medium abundance and rare subgroup significantly correlated with the content of ammonia nitrogen and pH.In the subsoil, the structure of abundant subgroup was significantly correlated with ammonia nitrogen content, and that of medium abundance subgroup was significantly correlated with dry matter content and ammonia nitrogen content, as well as the structure of rare subgroup was significantly correlated with dry matter, organic carbon, nitrate nitrogen content and pH (Table S8, Fig. S8).

Discussion
Response of abundance and distribution frequency of soil bacteria to afforestation along habitat-specialization and abundance gradients Research results have shown that the consistency of variation of species number and abundance of soil bacterial OTUs were significantly different on the two gradients.At the habitat-specialization gradient, the species number and abundance ratio changed consistently and showed the highest of habitat-specialized subgroup while the lowest of habitat-generalized subgroup.This was consistent with the research of Estavillo et al. about forest loss and biodiversity threshold 57 .However, at the abundance gradient, the abundant subgroup owns a small number of OTUs species and a high abundance ratio, while the rare subgroup showed the opposite.Some studies also showed that the total number of sequences of abundant subgroup was dominant while the species number of rare subgroup was absolutely dominant 18 .
The distribution frequency of soil bacterial OTUs was characterized by Levins niche width.The higher the value is, the more evenly distributed the individuals are in each resource state, that is, the higher the degree of niche differentiation and the better the interspecies co-existence.However, a lower value means that the distribution of individuals in each resource state is different and the co-existence between species is bad.Our results showed that the response of distribution frequency to stand change was highly consistent between the two subgroups classification gradients.The broad-leaved forests significantly promoted the coexistence and niche differentiation of each habitat-specialization subgroup, and most positively promoted the distribution frequency of abundant subgroup on the abundance gradient.The niche width of each habitat-specialization subgroup in the same forest type were similar, but that of the abundant subgroup were higher than the medium abundance and rare subgroups, especially in the broad-leaved forests.This was because abundant subgroup existed more abundant-dominant OTUs than rare subgroup, allowing them more competitive and thus adapt quickly to changing environments 54 .
The response of bacterial OTUs abundance to stand changes also differ between the two gradients.The abundance of each habitat-specialization subgroup was the lowest in the broad-leaved forests, which could be interpreted as that the soil environment of broad-leaved forests was conducive to niche differentiation among species and reduced the occurrence probability of a few high-abundance species 58 .The abundance of abundant and medium abundance subgroups in broad-leaved forests also decreased, but the abundance of rare subgroup increased, which could be interpreted as the decline in abundance of abundant subgroup enhanced the abundance of rare subgroup and occupied new niches 59,60 .

Response of α diversity characteristics to afforestation along habitat-specialization and abundance gradients
The α diversity of habitat-specialized subgroup and rare subgroup was the highest on their respective gradients.This was consistent with the conclusion reported by Wu et al. that the diversity and species richness of rare subgroup in topsoil and subsoil were significantly higher than those of abundant and medium abundance subgroups 58 .In general, the diversity index of each subgroup in broadleaved forests was higher than that of other stands 61 , and the response of diversity to stand change on habitat-specialization gradient was stronger than that on abundance gradient, of which the habitat-specialized subgroup responded the strongest.A number of studies have shown that bacterial habitat specialists were more susceptible to extinction than bacterial habitat generalists when habitat conditions changed 62,63 , and this could legitimately interpret the fact that bacterial α-diversity of habitat-specialization gradient more strongly responded to stand changes than the abundance gradient.The abundance gradient only represents the proportion of individual number of species, and the response of species and individual numbers of each subgroup to stand change were similar and intersecting, so the response of α diversity to stand change on the abundance gradient was weak.

Responses of soil bacterial composition and community structure to afforestation along habitat-specialization and abundance gradients
The results showed that the response of the composition and distribution features of bacteria to stand change was different between two classification gradients.In the broadleaved forests, the distribution uniformity of each habitat-specialization subgroup was the highest.However, on the abundance gradient, the most uniform distribution of OTUs in the broad-leaved forests was only in the abundant and rare subgroup of topsoil and rare subgroup of subsoil.This was consistent with the fact that rare subgroup owns a large number of species but a low abundance, which resulted in a more even distribution of abundance than the abundant and medium abundance subgroups 64,65 .In addition, the most uniform distribution in broad-leaved forests and the worst distribution uniformity in shrub and coniferous forests, was consistent with the conclusion on the increasing complexity of soil bacterial network from temperate coniferous forests succession to broad-leaved forests in Liupanshan Mountain 66 , which promoted co-existence and niche differentiation by reducing intraspecific and interspecific competition of bacteria 67 .Coordinate points of community structure of each habitat-specialization subgroup of some broad-leaved forests were obviously seperated from that of other stands.These indicated the soil environmental changes in broad-leaved forests lead to significant changes in soil bacterial community structure on the gradient of habitat-specialization.This conclusion was consistent with that the distribution frequency of each habitat-specialization subgroup in broad-leaved forests significantly higher than other stands thus the community structure must be obviously different from that of other stands in which a few OTUs dominated.In terms of abundance gradient, ANOSIM analysis showed that the differences among each forest type only significantly affected the community structure of rare subgroup in the topsoil, because their niche width was narrow and more sensitive to environmental changes 68 , while the abundant subgroup tended to be more resilient and robust to environmental changes.The conclusion of this study provides a scientific reference and a new perspective for the comparative study of soil microbial response to stand change from the two gradients of soil microbial habitat specialization and abundance. https://doi.org/10.1038/s41598-023-44468-x

Figure 1 .
Figure 1.Effects of afforestation types on niche width of soil bacteria of different habitat-specialization subgroups (S: Shrub; CF: Coniferous forests; MF: Mixed coniferous and broad-leaved forest; BF: Broad-leaved forests; * indicates significant difference at the level of 0.01).

Figure 2 .
Figure 2. Effects of afforestation types on niche width of soil bacteria of different abundance subgroups (S: Shrub; CF: Coniferous forests; MF: Mixed coniferous and broad-leaved forest; BF: Broad-leaved forests; * indicates significant difference at the level of 0.01).

Figure 3 .
Figure 3. Effects of afforestation types on the abundance of soil bacteria of different habitat-specialization subgroups (S: Shrub; CF: Coniferous forests; MF: Mixed coniferous and broad-leaved forest; BF: Broad-leaved forests; Different lowercase letters indicate that the index of the same soil layer is significantly different at the level of 0.05).

Figure 4 .
Figure 4. Effects of afforestation types on the abundance of soil bacteria of different abundance subgroups (S: Shrub; CF: Coniferous forests; MF: Mixed coniferous and broad-leaved forest; BF: Broad-leaved forests; Different lowercase letters indicate that the index of the same soil layer is significantly different at the level of 0.05).

Figure 5 .
Figure 5. Effects of afforestation types on α diversity of soil bacteria of different habitat-specialization subgroups (S: Shrub; CF: Coniferous forests; MF: Mixed coniferous and broad-leaved forest; BF: Broad-leaved forests; TG: Habitat-generalized subgroup of topsoil; TN: Habitat-neutral subgroup of topsoil; TS: Habitatspecialized subgroup of topsoil; SG: Habitat-generalized subgroup of subsoil; SN: Habitat-neutral subgroup of subsoil; SS: Habitat-specialized subgroup of subsoil; Different lowercase letters indicate that the index of the same soil layer is significantly different at the level of 0.05).

Figure 6 .Figure 7 .
Figure 6.Effects of afforestation types on α diversity of soil bacteria of different abundance subgroups (S: Shrub; CF: Coniferous forests; MF: Mixed coniferous and broad-leaved forest; BF: Broad-leaved forests TA: Abundant subgroup of topsoil; TM: Medium abundance subgroup of topsoil; TR: Rare subgroup of topsoil; SA: Abundance subgroup of subsoil; SM: Medium abundance subgroup of subsoil; SR: Rare subgroup of subsoil; Different lowercase letters indicate that the index of the same soil layer is significantly different at the level of 0.05).

Figure 12 .
Figure 12.NMDS 2D figure of effects of afforestation types of bacterial community structure of each abundance subgroup (○ represents shrub; △ represents coniferous forests; + represents mixed coniferous and broad-leaved forests; × represents broad-leaved forests).

Table 1 .
Community information of shrub and afforestation forests.

2 ij Table 2 .
Soil physicochemical properties of each afforestation types.Different lowercase letters indicate that the index of the same soil layer is significantly different at the level of 0.05.

Table 3 .
Mantel test of effect of environmental factors on soil bacterial community structure of each habitatspecialization subgroup.TG habitat-generalized subgroup of topsoil, TN habitat-neutral subgroup of topsoil, TS habitat-specialized subgroup of topsoil, SG habitat-generalized subgroup of subsoil, SN habitat-neutral subgroup of subsoil, SS habitat-specialized subgroup of subsoil.Bold indicates P < 0.05.

Table 4 .
Mantel test of effect of environmental factors on soil bacterial community structure of each abundance subgroup.TA abundant subgroup of topsoil, TM medium abundance subgroup of topsoil, TR rare subgroup of topsoil, SA abundance subgroup of subsoil, SM medium abundance subgroup of subsoil, SR rare subgroup of subsoil.Bold indicates P < 0.05.