The study areas covered a longitude range of 115°–129°E and latitude range 39°–54°N, and included a warm, a moderate, and a cold temperate zone, with altitudes ranging from 786 m to 932 m. The average annual precipitation ranges from 383.53 mm to 836.23 mm. Three sampling areas were identified in each zone of low, middle, and high latitude. The survey was mainly conducted in national nature reserves or scientific research bases, away from habitation, and mature forests with good preservation and strong regional representation were selected. The specific areas of low latitude to high latitude included: Long-term Biodiversity Monitoring Plot of Dongling Mountain in Beijing (LL): The experimental samples were located in Beijing Forest Ecosystem 20 hm² Research Station, Chinese Academy of Sciences, five 20 m × 20 m standard quadrates were selected, and each quadrate was 100 m apart, the vegetation type of the sample plot is Quercus liaotungensis forest; Changbai Mountain National Nature Reserve of Jilin Province (ML): The experimental samples were located in the forest monitoring plots of 25 hm² in Changbai Mountain, five 20 m × 20 m standard quadrates were selected, and each quadrate was 100 m apart, the vegetation type of the sample plot are Quercus mongolica and Pinus koraiensis mixed forest, and Huzhong National Nature Reserve in Daxing’anling, Heilongjiang Province (HL): the experimental samples were located in the forest monitoring plots of 25 hm² in Daxing’anling, five 20 m × 20 m standard quadrates were selected, and each quadrate was 100 m apart, the vegetation type of the sample plot is Larix gmelinii forest. A summary of these three research areas is shown in Table 1.

From each of the three areas, forest soil samples were acquired in July 2020. For this, standard experimental plots were set up in the core areas of the national nature reserves (see Table 1). Within each forest type, five sampling plots were established, and within each plot soil samples (0–20 cm) were collected using an 8 cm diameter soil auger from 15–20 locations along an S-shaped path. Before sampling, superficial debris (e.g., leaves and dry vegetation) was removed. The collected soil samples were transferred to sterile seal bags and stored in an icebox. Following transport to the laboratory, the soil was immediately sieved (2 mm mesh) to remove stones and plant material. One portion of the samples was air-dried to conduct physical and chemical analyses, whereas the remainder was kept at −80°C until required for microbial analysis.

A soil-water (1:2.5 w/v) suspension was shaken for 30 min prior to measuring the pH with a pH meter. The soil total N (TN) and organic C (SOC) were quantified using an Elemental Analyzer (Elementar, Langenselbold, Germany). Available nitrogen (AN) was sequentially digested in H₂SO₄–HClO₄, 0.5 M NaHCO₃, and 2.0 M KCl, and then examined with a continuous flow analysis system (SKALAR SAN++, the Netherlands). After wet digestion with HClO₄–H₂SO₄, total phosphorus (TP) was determined spectrophotometrically using standard procedures.

Total DNA of soil microorganisms was extracted using the QIAGEN DNeasy PowerSoil Pro Kit according to the manufacturer’s instructions. The V3, V4 region of the bacterial 16S rRNA region was amplified using universal primers 338 F and 806 R. Each sample was tagged with a different six bp barcode present at the 5′end of the primer. The PCR amplification protocol consisted of an initial denaturation stage at 94°C for 10 min, followed by 30 cycles of 90°C for 60 s, 55°C for 60 s, and 72°C for 60 s, and final extension at 72°C for 10 min. The product was purified using the QIAquick PCR Purification Kit (QIAGEN). The PCR products of different samples were sequenced by BMC Biotechnology Co., Ltd., using an Illumina MiSeq sequencer.

The raw sequences were processed using QIIME1. Forward and reverse reads were merged using FLASH software. ‘Trim’ was used to remove low-quality sequences shorter than 150 bp, reads with a mean quality score below 20, and sequences containing ambiguous nucleotides. Chimeric sequences were identified and removed with the uchime algorithm. A similarity of 97% was applied to divide the reads into operational taxonomic units (OTU) through UPARSE (V.7.0.1090). Operational taxonomic identities were determined using QIIME by executing the BLAST algorithm against sequences in the SILVA database (https://www.arb-silva). Diversity indices Chao1, ACE, Shannon and Simpson were calculated using ‘alpha_diversity.py’ to reflect the a diversity of the bacterial community, and ‘beta_diversity.py’ was used to calculate the Bray-Curtis distance matrix to reflect the ß diversity.

Differences in soil physicochemical properties between the three treatments were analyzed using one-way ANOVA at a 0.05 significance level coupled with Duncan’s tests, using SPSS 23.0 (IBM SPSS Statistics for Windows). OTU- level a diversity indices, such as Chao1 index, ACE index, Shannon index and Simpson index, were calculated using the OTU table in QIIME1. Venn diagrams and heatmap representations of the relative abundances of bacterial OTUs among soil samples were generated using the “vegan” package in R. Non-metric multidimensional scaling (NMDS) was conducted based on the Bray—Curtis dissimilarity at the OTU level, also using “vegan” (Frey et al., 2021). Redundancy analysis (RDA), implemented in “vegan”, was used to generate compositional profiles, agglomerating the OTUs to the phylum level and genus level. Indicator species analysis based on the OTU level was performed as outlined in Rime et al. (2015) using “vegan” and the “labdsv” (version 1.2–2) R package.

The determined chemical properties of the soil collected at three different latitudes is summarized in Table 2. The mean annual temperature (MAT) decreased with increasing latitude in the three locations. The pH and content of SOC, TN, AN and AP differed as well. The SOC content was lower and the soil pH was higher at the lowest latitude but they did not differ for the other two locations. Contents of TN, AN and AP values were lower for the highest latitude than for the other two locations.

Analysis of 16S rDNA gene sequences in the 15 samples revealed information on the structures of the bacterial communities. The vast majority of OTUs were detected at all three latitudes (n=1,267), but unique and partially shared OTUs were recognized as well, as visualized in a Venn diagram (Figure 1). There were 1,472 identified bacterial OTUs recognized in HL, 1,591 in ML and 1,507 in LL, and these soils contained 44, 11 and 34 unique OTUs, respectively (Figure 1). The fewest number of OTUs were exclusively shared by LL and HL (n=27), whose locations were most distant, and the most OTUs were exclusively shared by ML and LL (n=179).

As shown in Table 3, the bacterial a diversity of the forest soils at different latitudes varied significantly. HL produced the lowest values for the ACE, Chao 1, Shannon and Simpson index and these were all significantly lower compared to LL (p<0.05). The highest values for Shannon and Simpson index were observed in LL with 8.85 and 0.9948, respectively, followed by ML and HL. The higher ACE and Chao one index of ML was not significant different compared to HL or LL (p>0.05) (Table 3).

NMDS plots based on confidence intervals (p<0.05) were produced (Figure 2A) that showed an absence of overlap between the different sites, which illustrated that the soil bacterial communities (stress = 0.0225) were clearly different in the three areas. This was confirmed by PERMANOVA analysis, whose findings are shown in a box plot (Figure 2B) (significant difference between groups, R ² = 0.743, p<0.01).

At the phylum level (Figure 3), the soil bacterial communities of the LL, ML, and HL were dominated by Acidobacteria (35%, 43%, and 43%, respectively) and Proteobacteria (37%, 35%, 39%). Much lower levels were reported for Rokubacteria (8.6%, 7%, 0.5%), Actinobacteria (7%, 4%, 6%), and Gemmatimonadetes (4.9%, 2.6%, 1.8%), and even lower abundance was noted for other phyla (Figure 3). There were differences between the samples taken at the different latitudes, notably for lower levels of Acidobacteria in LL (p<0.001), but also for the other phyla shown in the figure. Proteobacteria members were lowest in soil collected in ML(p<0.01). The relative abundances of Gemmatimonadetes, Rokubacteria, Chloroflexi and Nitrospirae produced decreasing trends (of variable strengths) with an increase of latitude (Figure 3).

At the genus level, the soil of HL and LL was dominated by Candidatus_Solibacter (4.1%, and 5.0%, respectively) and Bryobacter (4.5% and 3.8%), while these genera were present at significantly lower (p<0.001) levels in ML (Figure 4). Roseiarcus (4.5%) and Granulicella (4.3%) were particularly abundant in HL but undetected in ML. Conversely, Sphingomonas and MND1 were dominant in ML, while Lactococcus was particularly abundant in LL (p<0.001). Thus, Bryobacter and Roseiarcus were the most abundant genera in the HL where Nitrospira and MND1 were the least abundant. Sphingomonas was the most abundant genus in ML, where Roseiarcus and Granulicella were least abundant. Lastly, Candidatus_Solibacter was the most abundant genus in LL while Granulicella was present at lowest abundance here (Figure 4).

A heatmap analysis was performed to compare over- and under-abundance of the 24 most abundant bacterial genera in the five replicates for each latitude (Figure 5). This revealed co-enrichment patterns, as indicated that the genera Elin6067, MND2, Sphingomonas and Dongia in LL, Candidatus_Udaeobacter, GAS113, Reyranella and Candidatus-Koribacter in ML, Burkhoderia, Rosearcus, Granulicella and Occallatibacter in HL, (Figure 5).

A LEfSe analysis was conducted to determine the classified bacterial taxa with significant relative abundance differences among the different soil samples. Significant differences were observed among 16 taxa in the three latitudes, as indicated by LDA effect size scores of >4.5 (Figure 6). These are shown to the right of the cladogram in the figure.

Pearson correlation analysis was used to detect correlations between soil bacteria a-diversity and soil chemical factors. The results (Table 4) indicated that the ACE and Chao1 diversity indices correlated with TN, AN and TP (p<0.05), while the Shannon diversity was correlated with MAT and pH (p<0.05). The Simpson diversity did not correlate with any of the determined soil chemical properties (Table 4).

An RDA analysis indicated that soil chemical properties were important factors driving the structural and compositional changes of the soil bacterial communities (Figure 7A). The cumulative variations in the first and second RDA axes were 42.95% and 13.40%, respectively. Soil pH, MAT, SOC and TN were the major factors explaining the composition differences of the soil bacterial community. Further, through the correlation heat map analysis found the soil pH produced significant positive correlations with Rokubacteria, Nitrospirae and Chloroflexi, and negative correlations with Acidobacteria. Members of Rokubacteria correlated positively with all soil parameters except for SOC, while that soil parameter produced significant negative correlations with Bacteroides, Proteobacteria and Actinobacteria. Lastly, Cloroflexi correlated positively with AN and Nitrospirae with MAT (Figure 7B).

Most available studies have suggested presence of specific soil microbial communities at specific latitudes (Zhou et al., 2016; Tian et al., 2018; Zheng et al., 2021). The present study shows that the bacterial Shannon index of forest soil decreased gradually with increasing latitude, and the difference between highest and lowest latitude was significant (Table 3). There are two possible reasons for this. On the one hand, higher temperatures at lower latitudes promote the metabolism and population growth of soil bacteria, and accelerating litter decomposition and improving nutrient availability (Anderson et al., 2009; Zhou et al., 2020). On the other hand, lower pH at high latitudes acidifies forest soils, leading to an increase in acidophilus, which may reduce the number of other species and thus reduce soil bacterial diversity. The mid-latitude soil exhibited the highest bacterial ACE and Chao1 index, though this was not significantly different from low-latitude. Variations in ACE and Chao1 indices may be attributed to the effect of soil physicochemical factors, as levels of TN, AN and TP correlated significantly with soil ACE and Chao1 indices. This indicates that changes in latitudes can affect soil physicochemical factors and therefore affect the abundance of bacteria.

Explaining how biodiversity changes with spatial, temporal, and environmental gradients and its main drivers remains a central focus on biodiversity science. In recent years, research was performed to identify the factors that influence and drive the spatial distribution of microorganisms. Most studies have identified soil pH as the main factor driving soil microorganisms diversity and spatial distribution (Fiere and Jackson, 2006; Griffiths et al., 2011; Tripathi et al., 2018; Chen et al., 2022). In this study, we found that pH directly affected the diversity of soil bacterial communities in different latitudes and produced significant correlations with four phyla, suggesting that pH is a major factor driving alpha diversity in forest soil bacterial communities. This is consistent with results from the literature. Likewise, Bahram et al. (2018) found that pH caused niche differentiation of soil bacteria and fungi on a global scale and identfied certain trends of composition and diversity of soil bacterial communities with pH. However, according to a recent study, on a continental scale temperature regulates the diversity of microorganisms in forest soils (Zhou et al., 2016). Our study found that the soil bacterial alpha-diversity decreased with increasing latitude, with a significant correlation between temperature and soil bacterial Shannon diversity (Table 3), indicating that soil bacteria had strong latitudinal diversity patterns related to temperature in the temperate forests of China. Temperature can affect soil bacterial diversity through various mechanisms. First of all, the most important direct mechanism is that a higher temperature will increase the soil bacteria’s metabolic rates and growth (Wang et al., 2009; Segura et al., 2015). Secondly, higher temperatures promotes plant diversity and productivity, which in turn provides substrates and nutrients for bacterial growth, thus improving soil bacterial diversity (Lange et al., 2015; Deng et al., 2020). Thirdly, higher temperatures support more effective litter decomposition and promote nutrient release, increasing nutrient availability that allows higher bacterial diversity (Zhou et al., 2012; Gao et al., 2022). Finally, the temperature may interact with soil pH, soil nutrients and other environmental factors to regulate the change of soil bacterial diversity (Shen et al., 2013; Delgado-Baquerizo and Eldridge, 2019).

Our results identifiud the main bacterial phyla at different latitudes as Acidobacteria and Proteobacteria, followed by Rokubacteria, Actinobacteria and Gemmatimonadetes. Furthermore, we found that increasing latitude significantly decreased the abundance of Acidobacteria. Members of that phylum have a strong sensibility to the environment, such as vegetation type and soil pH, and these can affect the diversity and community composition of soil Acidobacteria. In both HL and ML, the soil pH was under 5.0, Acidobacteria in these soils can degrade cellulose under microaerobic and anoxic conditions, and the refractory needles in coniferous forests increased, which promoted the expansion of Acidobacteria populations. Previous studies have shown that Proteobacteria and Actinobacteria are mainly involved in the decomposition of organic matter, and the relative abundance of Proteobacteria is positively correlated with soil carbon content (Li et al., 2018; Li et al., 2022; Zhang et al., 2022). Indeed, both the SOC content and the relative abundance of Proteobacteria was highest in HL, which was consistent with previous results. Most Actinomycetes are aerobic, grow at an optimum temperature of 28°C–30 °C, and are abundant in neutral or slightly alkaline soils. The relative abundance of Actinobacteria was highest in LL, this is mainly because the low latitudes provides suitable temperature and soil pH for the growth of Actinobacteria. Increasing latitude brought about a reduction in the relative abundances of Gemmatimonadetes, Rokubacteria, Chloroflexi and Nitrospirae, which was somewhat inconsistent with published observations that described revegetation resulted in initial increases and then decreases of Gemmatimonadetes, Chloroflexi, and Nitrospirae (Zhou et al., 2016; Tian et al., 2018). Further research is needed to address the reasons for this discrepancy, which may be attributed to the different vegetation types and different latitudinal gradients.

At the genus level, Candidatus_Solibacter and Bryobacter were highly abundant in two or three of the investigated soils, while Roseiarcus, Granulicella and Lactococcus were abundant in one soil type (Figure 4), which was consistent with other studies (Challacombe et al., 2017; Zhang et al., 2021; Kim et al., 2021). Members of the most abundant Candidatus_Solibacter and Bryobacter genera are involved in decomposing organic matter and use this as a carbon source. Interestingly, their relative abundance was significantly lower in ML. This may be due to the acidity of soil in high latitude coniferous forest, which provides a suitable environment for their growth, while low latitude broad-leaved forest provides sufficient carbon sources. The high relative abundance of Roseiarcus in the high-latitude coniferous forest soil of HL was consistent with previous results (Zhang et al., 2014; Rime et al., 2015; Du et al., 2016). The effects of latitude on the diversity and abundance of bacteria have been linked to the quantity and characteristics of the organic matter derived from plant litter, as it constitutes an important nutrient source for soil microorganism growth (Wagai et al., 2011).

The main drivers of soil bacterial composition can result in inconsistent responses at different study scales. Many studies showed that difference of soil factors at local and regional scales had a major impact on microorganisms communities, while climate factors and geography plays a more important role at larger scales (Tripathi et al., 2012; Zhang et al., 2020). In the present study, the soil bacterial community structure was affected by multiple environmental factors and was not determined by a single factor, which was similar to other results (Griffithset al., 2011; Delgado-Baquerizo and Eldridge, 2019). According to Pearson correlation analyses, the 10 most abundant bacterial phyla were mainly influenced by pH, MAT, SOC and TN: Proteobacteria, Acidobacteria, Chloroflexi, Rokubacteria and Nitrospirae all produced significant, though variable correlations with one or more of these soil factors (Figure 7B). Although temperature is considered a major factor regulating soil bacterial composition at large scales, it only signficiantly affected abundance of Rokubacteria and Nitrospirae in our study, wherea pH played an important role in the distribution of soil bacterial core species in the different latitudes. This may be because pH can affect specific communities, such as increasing the relative abundance of Acidobacteria, while decreasing the relative abundance of actinomycetes, thus affecting the spatial distribution of the communities (Xia et al., 2016; Tian et al., 2021). In addition, soil nutrient factors such as SOC and TN are also closely related to bacterial composition, suggesting these too are main drivers of soil bacteria at different latitudes. Thus, there are no single climatic factors or soil physical and chemical properties, but more likely multiple factors interact with each other to regulate changes in soil bacterial composition. Therefore in the future, environmental factors shoud more comprehensively be considered in combination to explore their interactions and respective contributions, it can better reveal the driving mechanism of soil microbial distribution and stability on a large scale.