Network complexity of rubber plantations is lower than tropical forests for soil bacteria but not fungi


 Soil microbial communities play a crucial role in ecosystem functioning. Past research has examined the effects of forest conversion on soil microbial composition and diversity, but it remains unknown how networks within these communities respond to forest conversion such as when tropical rainforest are replaced with rubber plantations. In this study, we used Illumina sequencing and metagenome shotgun sequencing to analyze bacterial and fungal community network structure in a large number of soil samples from tropical rainforest and rubber plantation sites in Hainan Island, China. Our results showed only a few shared network edges were observed in both bacterial and fungal communities, which indicates that forest conversion altered soil microbial network structure. We found a greater degree of network structure and a larger number of network edges among bacterial networks in samples from tropical rainforest compared to samples from rubber plantations. The difference was especially pronounced during the rainy season and indicates that rainforest bacterial networks were more complex than rubber plantation bacterial networks. However, rubber plantations soil fungal networks showed more higher links and higher network degree, suggesting that forest conversion does not reduce fungal network complexity. We found that some groups of Acidobacteria were keystone taxa in our tropical rainforest soils, while Actinobacteria were keystone taxa in rubber plantation soils. In addition, seasonal change had a strong effect on network degree, the complexity of soil bacterial and fungal network structure. In conclusion, forest conversion changed soil pH and other soil properties, such as available potassium (AK) and total nitrogen (TN), which resulted in changes in bacterial and fungal composition and network structure.


Introduction
Tropical rainforest have the highest biodiversity of any ecosystem and harbor more than 60% of all known plant and animal species [1]. However, over the past several decades, logging, mining, slash and burn agriculture have caused widespread deforestation and forest degradation. Of these,the conversion of forest to agriculture has caused the most forest loss [2].
Hainan is home to a large area of tropical rainforest rich in biodiversity. It is a part of the Indian-Malay rainforest system and the northern edge of the world's rainforest distribution. However, rubber plantations now account for almost a quarter of the total extent of vegetated areas on Hainan Island [3].
The soil microbiome is highly diverse and comprises up to one quarter of Earth's diversity [4]. Soil microbes play a critical role in the maintenance of soil quality and function, and they represent the majority of biodiversity in terrestrial ecosystems [5]. A number of studies have investigated the impact of the conversion of tropical forests to rubber plantations on soil microbial composition and diversity [3,[6][7][8][9][10][11][12]. Studies conducted in Indonesia [6], Malaysia [7] and South China [8][9][10] have found signi cant differences between rubber plantations and tropical forests, speci cally that the diversity of soil bacteria was higher in rubber plantations than in rainforest. Compared to primary forests, agricultural systems tend to have higher bacterial richness but lower fungal richness [7][8][13][14]. However, there are few studies on the effects of forest conversion on soil microbial network structure.
Network analysis is an increasingly popular tool for investigating microbial community structure, as it integrates multiple types of information and may represent systems-level behavior [15]. The soil microbial network is viewed as a critical indicator of soil health and quality [16]. Network analysis of taxon cooccurrence patterns provides new insight into the structure of complex microbial communities, insight that complements and expands on the information provided by the more standard suite of analytical approaches [17]. Previous work has shown that agricultural intensi cation can reduce microbial network complexity [18]. Logging alters soil fungal network in tropical rainforests, i.e., a better-organized fungal community in the select cut stands when compared with the primary stands [19]. Soil bacterial networks are less stable under drought than fungal networks [20]. Soil networks become more connected as ecological restoration progresses [21]. So far, very few studies have assessed the impact of forest conversion on soil microbial networks and it is still unclear whether forest conversion as well as seasonal change in uences the structure and complexity of microbial networks. Here we explored bacterial and fungal community network structure using Illumina sequencing based on samples collected from tropical rainforest and rubber plantations in Hainan Island, China. We aimed to test the hypothesis that (1) forest conversion alters microbial networks by altering microbial community composition [3] and that soil microbial activity is strongly in uenced by plant species [22]. (2) Soil microbial network structure in rainforest sites is more complex and stable than in rubber plantations because natural systems were more connected than arti cial systems [21]. By evaluating these hypotheses, we want to clarify the drivers and mechanisms that link forest conversion to differences in soil microbial network structure. This study will provide critical information for understanding and managing microbial communities in tropical forests of China and elsewhere.

Study site
This study was conducted on Hainan Island (18°10′-20°10′N and 108°37′-111°03′E), south China. The total area of Hainan Island is about 34,000 km 2 [23]. Hainan Island is the largest island within the Indo-Burma Biodiversity Hotspot in tropical Asia [24] and has a tropical monsoon climate. Hainan Island has a warm and humid climate all year round, with an average annual temperature of 22-26℃. The rainy season occurs from May to October, with a total precipitation of about 1500 mm, accounting for 70-90% of the total annual precipitation. Only 10-30% of the total annual precipitation falls within the dry season, from November to April. Rainfall is abundant, ranging from 1,000 mm to 2,600 mm yearly, with an average annual precipitation of 1,639 mm. The central part of Hainan Island is mountainous and contains old-growth tropical rainforests and monsoon forests. Rubber plantations are found on the plateaus surrounding the central mountainous zone.

Soil sampling
We selected ve rainforests as our study sites: Bangwang mountain, Diaoluo mountain, Wuzhi mountain, Yinge mountain and Jianfeng mountain. Five rubber plantations were selected in Haikou, Danzhou, Qiongzhong, Wanning and Ledong ( Figure S1). More information on the study sites is provided in Table   S1. For each site, thirteen soil samples were collected, thus there were a total of 130 samples collected between the rubber plantations and tropical rainforest per sampling interval. Soil sampling was performed in the rainy season (July) and dry season (January). Thus, there were a total of 260 soil samples (130 per forest type). Soil samples were divided into two parts: one was used to analyze soil water contents, soil pH, total nitrogen, total phosphorus (TP), total potassium (TK), nitrate nitrogen (NN), ammonium nitrogen (AN), available phosphorus (AP), potassium (AK). The other was used for DNA extraction. Soil properties were analyzed following the methods described in by Lan et al. [11]. Soil properties of the rubber plantation and rainforest sites are shown in Table S2.
Metagenomic shotgun sequencing libraries were prepared and then sequenced by Majorbio, Inc.
(Shanghai, China) using the Illumina HiSeq 2000 platform. The NR gene catalog was aligned against the Kyoto Encyclopedia of Genes and Genomes (KEGG) database using BLAST (Version 2.2.28+) and then functionally annotated using KOBAAS 2.0 according to previously described methods [28] Bioinformatics and data analysis Raw fastq les were demultiplexed and quality-ltered using QIIME [29] (version 1.17). Operational Taxonomic Units (OTUs) were clustered with a 97% similarity cut-off using UPARSE [30], and chimeric sequences were identi ed and removed using UCHIME. Using the RDP Classi er, the phylogenetic a liation of each 16S rRNA gene and ITS gene sequence was determined using a con dence threshold of 70% with the SILVA 16S rRNA database and UNITE database, respectively [31]. The relative abundance was determined for each taxon [32] using MOTHUR [33].

Statistical analysis
For the co-occurrence network analyses, we only focus on the top 300 most abundant OTUs of the two forest types. The networks of each habitat during each sampling period (tropical rainforest and rubber plantations in dry season and rainy season) were constructed with 65 samples each. Interactions consisted of Spearman's rank correlations and co-occurrence networks were constructed using only signi cant correlations of ρ> 0.6 (P < 0.01) [17], because this cutoff includes a range of interactions strengths [20]. The networks were then visualized in R using the igraph package. To reveal the distribution pattern of correlation coe cients, the frequency of the coe cients ofρ> 0.3 (P < 0.01) were plotted. The Network Analyzer tool in Cytoscape (version 3.4.0) was used to calculate network topology parameters including number of nodes, edges, degree, betweenness, closeness. In order to evaluate the network differences between tropical rain forest and rubber forest sites, Venn diagrams were plotted to reveal the number of shared edges and unique edges which were calculated using igraph. Keystone OTUs were selected on the basis of high network degree, high closeness centrality, and low betweenness centrality as de ned by Berry and Widder [34]. To evaluate the proportional in uence of each phylum on bacterial and fungal network structure, node degrees of each phylum were calculated and bar plots were created. Correlation coe cients between species and functions were calculated based on metagenomics data.
Here we used the top 50 most abundant species and top 50 KEGG functions (pathway level 3). Then the species and function correlation network was constructed on the Major bio cloud platform (https://cloud.majorbio.com/). To reveal the relationship between microbial taxon and environment variables, two-way correlation networks were also constructed on the Major bio cloud platform. The topological role of each node in a network was assessed by the Z i and P i values, where Z i represents the nodes connectivity within a module, and P i measures the degree of a node connected with other modules [35]. All species can be divided into four groups according to the simpli ed criteria [36], namely peripherals (Z i < 2.5 and P i < 0.62), connectors (P i > 0.62), module hubs (Z i > 2.5) and network hubs (Z i > 2.5 and P i > 0.62). The Z i and P i values were calculated using GIANT package of Cytoscape. The Z i -P i plot was created with ggplot2 in R.

Bacterial and fungal networks
Our results showed most nodes of bacterial networks ( Figure 1) and fungal networks ( Figure 2) varied with forest type in both the dry season and rainy season. For bacterial networks, there were 2559 and 2501 edges in tropical rainforest and rubber plantation in dry season respectively, but these two networks only shared 262 edges, accounting only about 10% the total edges ( Figure 3A-B). Similarly, these networks only shared 519 edges during the rainy season. For fungal networks, there were only 4 and 5 shared edges (i.e., no more than 5% of the total edges) in dry season and rainy season, respectively ( Figure 3C-D).
The number of edges of bacterial and fungal networks were almost equivalent during the dry season.
However, in rainy season, there were more edges in the bacterial network in tropical rainforest than in the rubber plantation (Table 1 & Figure 3B). For the network structure of the fungal community, more edges were observed in rubber plantations in rainy season (Table 1 & Figure 3D). Similarly, there were no signi cant differences in both bacterial and fungal network degree between tropical rainforest sites and rubber plantations in the dry season ( Figure 4A, C). In the rainy season, rainforest sites had higher bacterial network degree, while rubber plantations showed higher fungal network degree ( Figure 4B, D). For bacterial networks, more nodes (OTUs) with high degree (rubber plantation had 2 nodes with degree greater than 75, rainforest had 8 such nodes) of rainforest were observed during the rainy season ( Figure  S2B). For fungal networks, 15 nodes of higher degree (degree greater than 25) were observed in rubber plantations, however, only 7 such nodes existed for rainforest sites ( Figure S2D). These results indicate rubber plantation fungal network structure was more complex than tropical rainforest during the rainy season, but that the reverse was true for bacteria.
When considering the ratio of positive to negative correlation coe cients, more correlations (greater than 0.3, P < 0.05) were calculated, and the results showed that the negative correlations between bacterial and fungal OTUs of rubber plantations were consistently stronger than for tropical rainforest in both dry season and rainy season ( Figure 5).
For both the bacterial and fungal communities, neither tropical rainforest nor rubber plantation networks possessed module hubs and network hubs ( Figure S3-S4). For bacterial network, the majority of nodes in both the rubber plantation and tropical rainforest networks were connectors ( Figure S3). However, for fungal networks, the majority of the nodes in both rubber plantation and tropical rainforest networks were peripherals and connectors ( Figure S4). The ratio of peripherals and connectors of these two forest types was not different indicating the network structures of rainforest and rubber plantation were very similar as visualized in Figure 1 and Figure 2. However, the bacterial networks had more connectors than fungal networks, which suggests bacterial networks may contain more generalists than fungal networks do. This indicated bacterial network were more complex than fungal network, which can further con rmed by visualization of the network (Figure 1 and Figure 2).
For bacterial groups, members of the phyla Acidobateria, Planctomycetes and Verrucomicrobia showed higher degree in the tropical rainforest sites than in rubber plantations, suggesting that these taxa are strongly associated with the other members of the community in tropical rainforest ( Figure S5A).
Members of the phyla Actinobacteria showed higher degree in rubber plantations. Seasonal change also had effects on network degree for soil bacterial networks. For instance, Chloro ex had higher degree in rubber plantation in the dry season, but the opposite is true in the rainy season. For fungal networks, members of Basidiomycota showed higher degree in tropical rainforest sites duringin the dry season, however, Ascomycota showed higher degree in rubber plantations ( Figure S5C-D) during the rainy season.
We used total degree of each phylum to reveal the in uence of each phylum on network structure ( Figure   6). For bacteria, Proteobacteria, Actinobacteria and Acidobacteria had a large in uence on network structure ( Figure 6). Acidobacteria and Planctomycetes contributed more to rainforest networks than rubber plantation networks. However, Actinobacteria and Chloro ex showed the opposite. For fungi, Ascomycota and Basidiomycota had large in uence of network structure. Both Ascomycota and Basidiomycota had stronger in uence on rainforest networks than rubber plantation networks. The in uence of Ascomycota was stronger during the rainy season than in the dry season, indicating seasonal change also had impact on fungal community networks.

Keystone taxa
Keystone OTUs of the bacterial and fungal communities were selected on the basis of high degree, high closeness centrality, and low betweenness centrality. The results showed that forest conversion altered the keystone taxa of bacteria and fungi. The keystone taxa of bacteria were very different between rubber plantations and tropical rainforest sites in both the dry season and rainy season. For bacteria, there were more keystone taxa in tropical rainforest sites than in rubber plantations in both the dry season and rainy season indicating that the tropical rainforest networks had higher complexity. We found that some groups of Acidobacteria are keystone taxa in tropical rainforest sites but disappeared after forest conversion. There were more Actinobacteria bacteria in rubber plantations than in tropical rainforest sites (Table S3).
For fungi, more keystone taxa were observed in rubber plantations than in tropical rainforest sites during both the dry season and rainy season, indicating the rubber plantation networks were more complex. Most keystone taxa belong to Ascomycota suggesting member of this group are very import for network structure. In addition to forest conversion, seasonal changes also affect the keystone taxa of the fungal community network. There were more Basidiomycota OTUs in the dry season, but more Ascomycota in rainy season (Table S4).

Two-ways correlation networks
Two-way network analysis of the 50 most abundant species (metageomic data, the 50 most abundant species all belong to bacteria groups) and the 50 most abundant KEGG functions revealed that soil microbial community structure in at rainforests sites was more complex than rubber plantations ( Figure  7). Both rubber plantations and rainforest networks were more complex in the rainy season than in dry season. We also found that metabolism was the most important function in soil microbial network. Surprisingly, species of Actinobacteria negatively correlated with other species and function in rubber plantations (Figure 7). Two-ways correlation network analysis revealed the interaction between microbial composition and environmental variables. This analysis includes different environmental factors as nodes in the network, and the number of connections these nodes have indicates the number of OTUs that are impacted by that environmental factor (Figure 8). For bacteria, elevation had the highest network degree at 106, and was followed by AK (104), soil pH (86) and TK (9). In other words, elevations, AK, soil pH are all drivers of bacterial community composition. Soil pH negatively correlated with most bacterial Acidobcteria OTUs. For fungi, elevation had the highest network degree (61), followed by AK (51), longitude (15), and NN (11). AK positively correlated with most OTUs of Basidiomycota. Relationship between OTU abundance and soil pH revealed the soil pH negatively correlated with members of Acidobacteria, but positively correlated with members of Chloro exi and members of Ascomycota (Figure 9). AK positively correlated with members of Planctomycetes Verrucomicrobia and Basidiomycota, however negatively correlated with Chloro exi and Ascomycota.

Discussion
Forest conversion reduces soil bacterial network complex Land-use changes increasingly threaten biodiversity, particularly in tropical forests [37]. However, we still have little understanding of how soil networks responsd to forest conversion, such when rainforests are converted to rubber plantations. Our results showed that forest conversion had large effects on both soil bacterial and fungal networks. More edges (Table 1) and higher degree (Figure 4) of tropical rainforest bacterial networks were observed, especially during the rainy season, which indicates that the rainforest bacterial network was more complex than the rubber plantation network. This consistent with previous observations that soil bacterial networks were more complex in natural systems than in crop soil [38].
Further study showed that soil networks become more connected as nature restoration progresses [21]. The observed decrease in network complexity and cohesion supports the hypothesis that cropping may enhance the isolation of bacterial taxa [38], which results in lower connection of the network. In addition, at the microscale, the structure of tilled soils is more homogeneous, and the soil pores are less connected than in soils under without tillage [39], such as rainforest soil. In nature, soil ecosystems are highly heterogeneous since soil microbial biodiversity hot spots can form spatial and temporally within soil aggregates [40]. This spatial heterogeneity likely plays an important role for the interactions among microbes and the mechanisms by which more complex and diverse communities drive various nutrient cycling processes on small spatial scales [4].
A large number of studies employing microbial network analysis have enriched our understanding of microbial co-occurrence patterns in various soil ecosystems, however, very little is known of whether differences in the structure of microbial networks have consequences for microbiome functioning [4]. Our results demonstrated that more species related with metabolism in natural system than in the agricultural system, especially in the rainy season. This is in line with a previous study conducted in Sumatra, Indonesia, which found that the transformation of forest to rubber results in a 10-16% decrease in community metabolism [41]. Fewer interactions between microbial species (most of them are bacteria) and functions in rubber plantations demonstrated that forest conversion reduced soil bacterial network complex.
Forest conversion does not reduce soil fungal network complexity Surprisingly, rainforest bacterial networks were characterized by fewer edges (Table 1) and lower degree (Figure 4), which means that rubber plantation bacterial networks were more complex than the native forest. Although, our results were consistent with previous observations which found that fungal community networks were better organized disturbed forest compared to primary forest [19]. Banerjee et al. [18]'s observation showed that organic agricultural elds harbored much more complex fungal networks with many more keystone taxa than conventional managed elds. Forest conversion resulted in shifts in fungal composition from Basidiomycota to Ascomycota (Figure S7), as seen in previous investigations [3,11]. Previous work showed that Basidiomycota species show higher drought sensitivity than Ascomycota species [42], this would result in a shift in richness and abundance of Basidiomycota species ( Figure S6). Many Basidiomycota species are capable of long-distance dispersal [43][44], which may result in a decrease in fungal network. This possibly explained why Ascomycota OTUs contribute more to the network structure than Basidiomycota ( Figure 6). Overall, reduction in abundance and richness of Basidiomycota species led to an increase in fungal links in rubber plantations.

Forest conversion enhanced the stability of soil network
The positive to negative ratio of network links indicates the balance between facilitative and inhibitive relationships within a network [45]. Theoretical studies, for example, predict that ecological networks that consist of weak interactions are more stable than those with strong interactions [46][47], and that compartmentalization and presence of negative interactions increase the stability of networks under disturbances [47][48][49]. In our study, more negative correlations were detected in rubber plantation, indicating the network structure of rubber plantation soils was more stable than rainforest soils [20].

Driver of the network structure
Forest conversion results in the loss of plant diversity, plant biomass and increasing soil pH [8][9]. Rubber plantations had a signi cantly higher pH, which explains the relative decrease in the abundance of Acidobacteria [8]. Our results demonstrate that keystone taxa of soil microbes change after forest conversion (Table 1). We found that many OTUs of Acidobacteria t our criteria as keystone species for rainforest sites, which is consistent with previous ndings [50]. Unexpectedly, OTU11388 and OTU11373, both Acidobacteria, were observed in rainforest soils in both the dry and rainy seasons, indicating Acidobacteria were very important for rainforest soil bacterial networks ( Figure 6 and table S3). Higher AK concentration resulted in a higher abundance and more taxa of of Actinobacteria (Figure 9), which suggests that Actinobacteria contributed more in rubber plantation than in rainforest ( Figure 6). Indeed, forest conversion reduced the abundance of Actinobacteria OTUs ( Figure S7) Due to the human disturbance in rubber plantations, the soil will inevitably be slightly polluted with herbicides and domestic garbage. Previous study showed member of Actinobacteria were observed in contaminated soil [51].
Forest conversion also increases land use intensity [52], including the application of fertilizer and herbicide. Herbicide application also caused signi cant decreases in root colonization and spore biomass of arbuscular mycorrhizal fungi in tropical agriculture [53]. Soil nutrient concentration shows a decline around the roots of rubber plantations compared to those from rainforests [54]. Our observation is no exception, for instance, AK and TN concentration was signi cant lower in rubber plantation than in samples from rainforest sites ( Table 2). Higher concentration of AK reasonably explained the higher contribution of Basidiomycota on the network structure ( Figure 8B) due to AK positive association with Basidiomycota.
Spatiotemporal heterogeneity can be a major driver of the abundance and distribution of keystone taxa in soil which is a highly heterogeneous and multifaceted environment [55][56][57]. Seasonal variability determines the structural and compositional properties of microbiomes in an environment, and as such, a keystone species might be present only in a speci c season or time period [50]. It was interesting that more bacterial OTUs were identi ed as connectors during the rainy season than in the dry season. Connectors have been characterized as generalists [36], and generalists drive covariation among communities in a network [19]. Previous observation demonstrated that some keystone taxa that were found in the dry season disappeared during the rainy season [58]. Seasonal changes possibly explained the keystone taxa was observed in rainy season but not in dry season.

Conclusion
Our knowledge about land-use impacts on soil ecosystems is mostly limited to biodiversity and ecosystem functions, leaving uncertainty about how soil networks change after forest conversion. This study is the most comprehensive report on changes in network structure that occur when tropical rainforests are converted into rubber forest. Our study showed that forest conversion altered both bacterial and fungal soil networks, reduced bacterial network complexity and enhanced fungal network complexity, especially during the rainy season. One possible reason maybe that forest conversion changed soil pH and other soil properties, which altered bacterial composition and subsequent network structure. Our study demonstrates the impact of forest conversion for soil network structure, which has important implications for ecosystem functions and health of soil ecosystems in tropical regions.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Availability of data and material
The raw reads were deposited into the NCBI Sequence Read Archive (SRA) database (Accession Number: SRP108394, SRP278296, SRP278319).

Competing interests
The authors have no con icts of interest to declare. LGY, and WZX conceived of and designed the study; YC and SR conducted the sampling; CBQ, and ZXC performed analyses; LGY performed statistical analyses and wrote the manuscript, all authors edited the manuscript. Tables   Table 1 Topological