How habitat loss and fragmentation affects biodiversity and ecosystem functions is an important topic in ecological research (Lindenmayer and Fischer, 2007; Haddad et al., 2015; Si et al., 2016). Empirical studies in macro ecosystem commonly emphasize that habitat loss has triggered negative effects on biodiversity, while the effects of habitat fragmentation per se (hereafter termed habitat subdivision) are still large uncertainties (Haddad et al., 2015; May et al., 2019). Habitat subdivision is a landscape-level reconfiguring that subdividing a single large of habitat into several smaller areas involves multidimensional effects on biodiversity (Ewers and Didham, 2006; Lindenmayer and Fischer, 2007). The most intuitive effects are derived from increase of habitat patches, decrease of habitat area, changes of isolation, and even novel ecological boundaries (Ewers and Didham, 2006; Haddad et al., 2017; Bueno and Peres, 2019). It is still controversial to clarify the impacts of habitat subdivision due to its divergent changes in configuration (Villard and Metzger, 2014). The habitat amount hypothesis holds that the effect of habitat subdivision is consistent with the effect of habitat amount changes (Haddad et al., 2017). However, increased researches have demonstrated that the impacts of habitat area reduction and isolation changes caused by habitat subdivision on biodiversity cannot be replaced by the impacts of habitat amounts (Haddad et al., 2017; Deane et al., 2022). Therefore, habitat subdivision induced changes of geometric properties may strongly influence biodiversity and ecosystem functions directly and indirectly (Deane et al., 2022). The geometric effects of habitat subdivision on soil microbial biomass and diversity are especially unclear due to the substantial mismatch between habitat subdivision scale and microbial habitat area or diffusion capacity (Baas-Becking, 1934; Finlay, 2002). Furthermore, few studies have focused on the responses of soil microbial biomass and diversity to long-term (e.g., more than 50 years) habitat subdivision (Peay et al., 2007; Vannette et al., 2016), despite the important contribution of soil microorganisms to ecosystem functions (van der Heijden et al., 2007; Delgado-Baquerizo et al., 2016).

Habitat subdivision may alter soil microbial biomass and diversity at a regional scale by changing patch sizes and isolation, and soil abiotic and biotic factors (Peay et al., 2007; Vannette et al., 2016; Fanin et al., 2018; Li et al., 2020b). It is generally considered that the distribution pattern of microorganisms is completely different from that of macroorganisms due to the high species number, high diffusion ability, and small size of microorganisms (Baas-Becking, 1934; Finlay, 2002; Si et al., 2016; Zeng et al., 2019). Nonetheless, studies have increasingly indicated that unequal-sized patches and diffusion limitation caused by habitat subdivision are main factors affecting soil microbial diversity (Peay et al., 2007; Vannette et al., 2016; Li et al., 2020b). A study conducted in a lava-fragmented landscape, for example, showed that root-associated fungal communities were affected by fragmented area (Vannette et al., 2016), which was consistent with the theory concerning the relationship between species and area relationship based on macroorganisms (Macarthur and Wilson, 1963, 2016). Researchers have also suggested that changes in soil microbial biomass and diversity after habitat subdivision could be explained by other theories, such as island biogeographical theory (Macarthur and Wilson, 1963, 2016) and the regional similarity hypothesis (Mouquet and Loreau, 2002), which states that community changes result from the interaction between dispersal and species competition. In addition, the changes in the soil environment and biotic interactions caused by habitat subdivision can also greatly affect soil microbial diversity (van der Heijden et al., 2007; Fierer, 2017; Li et al., 2020b). The area of fragmented habitat, for example, can affect soil bacterial diversity through soil moisture (Li et al., 2020b) and can affect root-associated fungal communities through their host plants (Vannette et al., 2016). Landscape ecologists have therefore suggested that future habitat fragmentation studies should assess the effects of unequal-size patches and isolation (Wilson et al., 2016).

Explaining the effects of habitat subdivision on species and communities is difficult because of the ecosystem differences in connectivity, habitat area, and heterogeneity (Si et al., 2016; Wilson et al., 2016; Zeng et al., 2019). For example, habitat subdivision studies often involve hidden uncertainties related to incomplete diffusion barriers (e.g., corridors) or large-scale matrix changes (e.g., fire and logging); such uncertainties make it difficult to generalize about the effects of habitat subdivision on biodiversity (Wardle et al., 1997; Haddad et al., 2015; Wilson et al., 2016; Kardol et al., 2018). Simplified experimental studies, such as mesocosm or microcosm experiments (Åström and Bengtsson, 2011; Delgado-Baquerizo et al., 2018), cannot provide a complete understanding of the natural response of biodiversity to habitat subdivision. As an alternative to mesocosm and microcosm experiments, islands provide a more natural system for studying basic problems of ecology and evolution (Wardle et al., 1997; Losos and Ricklefs, 2009; Kardol et al., 2018), and land-bridge islands in particular are ideal model systems for testing the consequences of habitat subdivision (Si et al., 2016; Wilson et al., 2016; Zeng et al., 2019). In the case of land-bridge islands, the fragmented habitats are surrounded by water with a high-contrast heterogeneity, which severely limits the dispersal of certain taxa (Wilson et al., 2016). On the other hand, land-bridge island system is also a natural laboratory for verifying theory of island biogeography (Wilson et al., 2016; Li et al., 2020b). However, only a few studies have explored the distribution patterns of soil microorganisms among islands (Li et al., 2020b). Larger islands may lead to higher soil microbial diversity due to higher soil environmental heterogeneity, however, there is still doubt whether landscape-scale habitat heterogeneity can match the scale threshold of soil microorganisms (Bru et al., 2011; Fierer, 2017).

In the current study, we used mainland-island system to explore two questions: (1) examine effects and process of unequal-sized patches and isolation after habitat subdivision on soil microbial biodiversity in fragmented islands via comparing soil microbial biodiversity and composition among mainland fragmented into small islands, mainland fragmented in to medium islands, and mainland fragmented in to large islands; (2) verify island biogeographic theory and heterogeneity hypothesis via finding major factors that affect soil microbial diversity and composition among islands. The islands, which are located in Jiangxi Province, China, were created in 1972 when Zhelin Reservoir was established (Supplementary Figure S1). In 2020, after the islands had experienced nearly 50 years of habitat subdivision, we assessed the soil bacterial and fungal communities and soil abiotic variables on 61 land-bridge islands and 9 adjacent mainland sites. We categorized the 61 land-bridge islands into three types based on land area (small, medium, and large), and we also characterized the attributes of each island (e.g., distance to mainland, perimeter to area ratio, and island shape index). We hypothesize that: (1) soil microbial biomass and diversity would be lowest when mainland fragmented into small-sized islands because small patches may be too small to sustain a local population (Debinski and Holt, 2000); (2) soil microbial biomass and diversity may be not suitable for island biogeography theory but support other theories such as habitat heterogeneity hypothesis because of the mismatch between fragmented scale and micro-habitat scale (Sonnier et al., 2014) or dispersal-competition tradeoff (Mouquet and Loreau, 2002).

We found that both soil bacterial and fungal biomass were higher on islands than that on mainland, but the degree of changes depended on types of subdivision and their associated habitat heterogeneity, which consistent with our first hypothesis. Across three types of islands, area- or distance-induced increase in soil pH consistently enhanced bacterial biomass. The strong and positive associations between soil microbial biomass and soil pH were well reported at different ecosystems due to the narrow pH niche for soil microbial communities (Rousk et al., 2009; Chen et al., 2015). The greatest change in bacterial biomass of large subdivision was attributed to combined effects of soil pH and soil nutrient quality which indicated that higher habitat heterogeneity in larger island may play an important role in bacterial biomass (Li et al., 2020b). In contrast, soil fungal biomass was promoted by distance-induced decrease in soil moisture because the higher soil moisture in the subtropical forests might inhibit the growth of soil fungi (Cruz-Paredes et al., 2021). In addition, habitat subdivision consistently decreased B:F ratio on each type of islands, suggesting that soils on mainland are dominated by bacterial-based energy channels while soils on each of the three types of islands are dominated by fungal-based energy channels. The shift from bacterial-based to fungal-based energy channels resulting from habitat subdivision could result in decreases in soil nutrient losses and CO₂ release, which in turn could cause the fragmented islands to be more resistance to climate changes (de Vries et al., 2013; Chen et al., 2016).

In contrast to our first hypothesis, only on the medium-sized islands, soil bacterial and fungal richness were higher than that on the mainland, which was somewhat consistent with the intermediate disturbance hypothesis (Connell, 1978; Huston, 1979). That is to say, changes in bacterial and fungal richness were mainly derived from extrinsic process and associated intrinsic competition (Raynaud and Nunan, 2014). For example, the increase in bacterial richness was attributed to that lower moisture could promote competition ability of subdominant species (Carson et al., 2010); the increase in fungal richness was attributed to that higher soil pH could benefit ectomycorrhizal fungi growth and reproduction (Wardle and Lindahl, 2014). The lack differences in microbial richness between mainland and small or large islands could be due to the offset effects among various external environment (Raynaud and Nunan, 2014). The strong differences in microbial biomass between mainland and islands were highly associated changes in habitat heterogeneity resulting from changes in island geometrical properties (Yu et al., 2012). Normally, microbial local interactions are likely to be obscured by relatively large samples which encompass high environmental heterogeneity (Raynaud and Nunan, 2014). The consistent and positive relationships between microbial richness and dissimilarity in any subdivision cases showed that the subdivision process may encounter strong microbial interactions. Bacterial dissimilarity was lowest on medium-sized islands could perhaps be due to the reduced fitness of the original dominant, mainland bacterial taxa on medium-sized islands and to the increased fitness of previously subdominant competitors that were favored by medium-sized islands (Chesson, 2000; Mouquet and Loreau, 2002). The lower sensitive of soil fungal dissimilarity to habitat subdivision than bacterial fungal similarity could be due to the fact that soil fungi have mycelial networks and are more resistant than bacteria to environmental disturbances (Rousk et al., 2010; Wang et al., 2020). The strong negative relationships between microbial richness and dissimilarity without reference to island size suggests that these fragmented habitats may still be experiencing the replacement of the previously most competitive taxa by other competitors (Chesson, 2000; Mouquet and Loreau, 2002).

Our permutational multivariate analysis of variance analysis showed that both soil bacterial and fungal community composition were significantly changed by long-term habitat subdivision. From small to large islands, the relative abundance decreased for the dominant bacterial phylum Actinobacteriota but increased for the dominant bacterial phylum Proteobacteriota, although both phyla were previously characterized as copiotrophic based on classical life strategy theory (Fierer et al., 2007). The opposite relationships of the relative abundance of Actinobacteriota and Proteobacteriota with soil nutrients (SOC and TN) on the subtropical land-bridge islands indicates that these taxa still had different preferences for substrates and nutrients, although previous studies showed that the relative abundance of both phyla were positively correlated with soil nutrients (Lauber et al., 2008; Fierer et al., 2011; Laliberté et al., 2017). We also found that the relative abundance of most bacterial phyla and fungal classes were related to soil moisture and soil pH on the subtropical land-bridge islands, which is consistent with previous studies (Rousk et al., 2010; Chen et al., 2016, 2019; Wu et al., 2020). Our research showed that changes in the relationships between the relative abundance of several fungal classes after fragmentation with soil properties (such as the negative relationships between Agaricomycetes and soil pH and soil moisture, and the positive relationships between Eurotiomycetes and soil pH and total N) were also consistent with other studies (Hartmann et al., 2017; Li et al., 2020a). In addition, based on the reduction in the relative abundance of Mortierellomycetes, we infer that fungi are more closely related than bacteria to vegetation (Li et al., 2020a). Differences in microbial signatures between mainland and island also derived from soil properties, especially soil moisture and soil pH (Supplementary Figures S4, S5). For example, bacterial genus Roseiarcus and Kitasatospora were preferred on mainland due to their negative relationships with soil pH (Kämpfer, 2006; Rezaei et al., 2011; Kulichevskaya et al., 2014; Riahi et al., 2022), and fungal genus Mortierella was preferred on mainland due to its close interaction with plant roots (Toju and Sato, 2018). Overall, we found that habitat fragmentation greatly altered the composition of soil bacterial and fungal communities for each island size, and that the soil and plant properties associated with the habitat fragment-induced changes in community composition differed for bacteria and fungi.

Our ordinary least squares regression and structural equation model found only the relationship between bacterial richness and distance to mainland was consistent with species-distance relationships, while microbial richness was completely inconsistent with species-area relationships (MacArthur and Wilson, 2016). In contrast, both microbial biomass and diversity of soil bacteria or fungi were mainly associated with soil abiotic variables. Soil pH was still strongly associated with soil microbial biomass and richness (Rousk et al., 2009), although there were anti-directional responses in bacterial biomass and richness to soil pH might be due to the disproportionate contributions of dominant taxa to biomass than subdominant taxa (Bastida et al., 2021). We found soil fungal biomass decreased with increased soil moisture and increased with soil substrate quantity. The increase of soil carbon and nutrient content could provide more substrates for fungi growth (Bastida et al., 2021), however, negative effects of soil moisture on soil fungal biomass indicated that higher water content may inhibit their mobility to capture soil nutrients (Manzoni et al., 2012). Partly consistent with another land-bridge study (Li et al., 2020b), our results showed the island area had less effects on soil bacterial and fungal richness than distance to the mainland across the three types of islands, and this was especially the case for bacterial richness. Generalized additive model analysis showed that there were unimodal relationships between distance to mainland and bacterial richness across the three types of islands. Based on the coexistence theory of metacommunities (Mouquet and Loreau, 2002; Loreau et al., 2003) and as previously noted the intermediate disturbance hypothesis (Connell, 1978; Huston, 1979), the highest biodiversity may occur at a medium dispersal level because both species homogenization caused by high dispersal and competitive exclusion caused by low dispersal would reduce diversity. Our results also showed that soil bacterial or fungal dissimilarity were mainly associated with their richness and distance to mainland across the three types of islands. The robust negative relationships between microbial richness and dissimilarity across the three types of islands confirmed that species coexistence was inhibited when the community became more dissimilar (Mouquet and Loreau, 2002). Furthermore, we found that the variables associated with soil fungal richness or dissimilarity had less explanatory power than the variables associated with soil bacteria richness perhaps because some functional groups of soil fungi (e.g., ectomycorrhizal fungi) are more closely related than soil bacteria to the changes in vegetation composition (Peay et al., 2007). Across the three types of islands, we found that the richness of soil microorganisms was mainly associated with island area and distance to the mainland while soil microbial dissimilarity was mainly associated with microbial richness and distance to the mainland.

Using 61 subtropical land-bridge islands with a 50-year history of subdivision and 9 adjacent mainland sites as a model system, we highlight several findings: Firstly, soil bacterial and fungal biomass in different-sized subdivision were higher than that in mainland, while increased the soil bacterial and fungal richness only on the medium-sized islands. These changes were due to subdivision-specified habitat heterogeneity, especial for soil moisture and soil pH. Secondly, soil microbial dissimilarities were robustly and negatively associated with microbial richness suggests that these fragmented habitats may still be experiencing the replacement of the previously most competitive taxa by other competitors. Finally, soil microbial distributions are not consistent with species-area relationships, but determined by habitat heterogeneity and tradeoff between competition and dispersal.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm.nih.gov/, PRJNA898793. The bacterial and fungal ASV table and relevant environmental variables is available on GitHub at https://github.com/DimaChen-ctgu/Island-2020.

All authors contributed to the article and approved the submitted version, and due care has been taken to ensure the integrity of the work. The accompanying manuscript constitutes original unpublished work and is not under consideration for publication elsewhere.

This study was supported by the National Natural Science Foundation of China (42177272), and the Youth Innovation Promotion Association of the Chinese Academy of Sciences (2015061).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb.2022.1063340/full#supplementary-material