Plant–soil microbe feedbacks drive seedling establishment during secondary forest succession: the ‘successional stage hypothesis’

INTRODUCTION

Forests are increasingly under threat due to logging, fire, severe drought and other natural disturbances, which may compromise their ability to conserve biodiversity and regulate climate systems (Powers and Marin-spiotta 2017). Extensive deforestation and forest degradation will leave behind a complex landscape consisting of a matrix of forest patches under different levels of succession (Quesada et al. 2009). Traditionally, vegetation succession is considered an orderly process with directional change in species composition across variable spatial and temporal scales (Odum 1969). After anthropogenic interference or natural disturbance, the secondary succession of forest ecosystem generally follows the sequence from grassland, shrubland to forest (each successional stage includes pioneer and non-pioneer plant species). Unlike the primary succession that leaves little or no biological legacy on barren landscape, secondary succession following deforestation generally leaves legacies in soil such as native seed banks and propagules of herbs and woody species (Lemenih and Teketay 2006; Meers et al. 2012). However, a key question that remains undetermined is why does the seedling recruitment of herb, shrub and tree follow a sequence rather than emerging simultaneously during secondary succession? Understanding the mechanisms governing forest successional pathways is critical for the development of successful forest restoration and conservation strategies worldwide (Poorter et al. 2019).

FACTORS AND MECHANISMS DRIVING VEGETATION SUCCESSION

Starting from the first views of Clements (1916), predicting forest succession has been a key challenge for scientists for more than a century. Ecologists have long focussed on clarifying the patterns of succession and factors that drive secondary successional processes. Successional pathways could be affected by various parameters, such as soil resources, physical site conditions, disturbance regimes, plant species life-history traits and vegetation–climate interactions (Caplat and Anand 2009; Cattelino et al. 1979; Powers and Marin-spiotta 2017; Taylor et al. 2020). Additionally, some important hypotheses were also proposed to elucidate the succession phenomena: (i) direct regeneration hypothesis (Hanes 1971)—tree communities will regenerate to similar pre-disturbance composition levels shortly after disturbance, depending on the regeneration strategies of component species and the nature of disturbance regimes; (ii) the intermediate disturbance hypothesis (Connell and Slatyer 1977)—intermediate disturbances increase the abundance of rare species and result in higher local species diversity; (iii) the resource ratio hypothesis (Tilman 1985)—nutrient poor or coarse textured soils favour the species that can tolerate limited resources and (iv) the resilience hypothesis (Johnstone et al. 2010)—young, regenerating forests are more vulnerable to changes in succession governed by climate change, since young forest stands are more exposed (i.e. without overstory buffering capacity) relative to mature stands.

Notably, intra- and interspecific neighbourhood interactions among adult plants can conjointly contribute to species replacement during forest succession. Specifically, the biomass production of early-successional herb is significantly reduced in conspecific soil, and a negative effect arises due to the selective increase in host-specific soil pathogens (Kardol et al. 2007; van de Voorde et al. 2011), thereby impeding the competitive advantages of herbs and allowing shrubs to establish in the habitats. Early-successional herbs are generally light-demanding species, with high resource consumption rates that make them highly sensitive to competition by neighbours (i.e. poor competitive-response ability) (Gόmez-Aparicio 2009). In contrast, adult shrubs have larger competitive advantages for aboveground resources (e.g. light) over herbs due to their larger individual size, and once they establish in the habits, they easily exclude herbs, leading to successional development. Similarly, trees are usually late-successional species with a conservative use of resources (i.e. large competitive-response ability), which are strong competitors for aboveground resources relative to shrubs.

The previous theoretical frameworks of forest succession focussed mainly on soil abiotic conditions, climatic factors, neighbourhood interactions or competitions among adult plants, but there has been little scientific integration of the roles of soil biotic components in driving the seedling emergence of later-successional plants. It is known that plant species replacement during succession can be roughly divided into two steps, including seedling establishment and individual interactions among adult plants. The initial step is characterized by the successful seed germination and seedling establishment in soil (i.e. early stage of plant life cycle), which is a critical process for successful species replacement and depends largely on soil conditions (Lett and Dorrepaal 2018); and the second step is characterized by the development of seedling into adult plant and replacement of pre-existing species by the competitive adult plant. Unlike the mature-phase plants with size-asymmetric competitive advantages, the small-size shrub and tree seedlings are not able to compete with abundant pre-existing adult plants for above- and belowground resources. Hence, over the course of secondary succession, we should focus more on the effects of pre-existing adult plants on the seedling establishment of subsequent plants from the perspective of soil biotic factors instead of neighbourhood resource competition among adult plants. To date, no single model or hypothesis can accurately simulate or predict all aspects of forest succession and debate remains (Taylor et al. 2020). Given that the small-size seedlings can be strongly affected by various soil microbes (Igwe and Vannette 2019), incorporating soil microbe-related mechanistic details into successional theory may improve our understanding of seedling establishment at the initial step.

ROLES OF SOIL MICROBES IN PLANT–SOIL FEEDBACK AND VEGETATION SUCCESSION

Many classical concepts regarding ecological theory (e.g. forest succession) have been developed without explicit consideration of soil microbes, which represent the bulk of the functional and phylogenetic diversity of the earth (Barberan et al. 2014). In recent decades, the roles of soil microbes in forest ecology have gained increased attention due to the extensive associations between roots and mycorrhizal fungi as well as the exponential increase of forest diseases worldwide (Soudzilovskaia et al. 2019). Plant pathogens can negatively affect plant growth, survival and/or reproduction, which include pathogenic fungi, bacteria, viruses and oomycetes (Bever et al. 2015). Mycorrhizal fungi benefit plants by enhancing their nutrient access and stress tolerance (Rasmann et al. 2017). Soil pathogens and mycorrhizal fungi are commonly considered the key determinants of plant–soil feedback (PSF) (Bagchi et al. 2014; Tedersoo et al. 2020). Changes in soil properties that are caused by plants, which in turn affect plant performance are termed PSF (Bever et al. 1997; van der Putten et al. 2013). Positive PSF value commonly indicates better plant performance in conspecific soil relative to heterospecific soil, while negative PSF makes soil less suitable for conspecifics. Different plant species can directly induce specific changes in soil biotic characteristics via litter and rhizosphere effects, leading to species-specific PSF effects that differ among plant species or functional groups (Freschet et al. 2013; Hannula et al. 2021; Sun et al. 2022; Zhang et al. 2016). In addition, plant-mediated changes in microclimate and soil abiotic properties can also result in variable indirect effects on soil microbial communities as well as interspecific PSF processes (Ke et al. 2015). Since various soil pathogenic and mycorrhizal types have distinct evolutionary histories, anatomies and functional significance, these antagonists and mutualists may differentially affect the fitness of individual plants, interspecific interactions and hence PSF (Tedersoo et al. 2020).

A large number of literatures focussed on the consequences of microbially mediated PSF effects for plant diversity and species coexistence (Crawford et al. 2019; Johnson et al. 2012; Lekberg et al. 2018; Pan et al. 2021). Currently, increasing studies indicate a potential role of microbe-induced PSF changes in driving species replacement and grassland succession (Kuťáková et al. 2020; Zhang et al. 2021). Specifically, early-successional herbs experienced strong negative PSFs regulated by soil pathogens, which facilitated their replacement by later-successional species with greater tolerance to pathogens (Kardol et al. 2007; van der Putten 2003). Mid-successional herbaceous species had neutral or negative PSFs (Kardol et al. 2006; Zhang et al. 2016), and the root addition of conspecifics and root decomposition by saprotrophic organisms had neutral or negative effects on mid-successional grasses in ex-arable fields (Zhang et al. 2016). Late-successional plants experienced neutral or positive PSF and performed best in late-successional soil with abundant beneficial mycorrhizal fungi (Koziol and Bever 2015, 2019), which slowed succession and resulted in temporal stability in grassland and old-field (Kardol et al. 2006; Revilla et al. 2013). Generally, PSF involves short-term processes regulated by soil microbes (e.g. mycorrhizal fungi, soil-borne pathogens and nitrogen-fixing bacteria), and long-term processes mediated by saprotrophic organisms that affect plant growth by influencing nutrient availability through litter decomposition and interaction with root exudates (Poorter et al. 2012; Wardle et al. 2004; Zhang et al. 2016).

It has been concluded that the PSF strength experienced by a plant species was positively correlated with that species’ successional stage, and the changes in soil microbial communities (e.g. soil pathogens and mutualists) could explain the shifts in plant species abundances during secondary succession in tallgrass prairie (Bauer et al. 2015). Soil microbes play distinct roles in each successional stage, and interspecific plant–soil microbe feedback effects may change with time to favour grassland successional development (Bauer et al. 2015; Kardol et al. 2006; Koziol and Bever 2019). While the microbe-regulated PSF effects were mainly observed during grassland and old-field succession, it remains largely unclear how the plant–soil microbe feedbacks of pre-existing plants drive the seedling establishment of subsequent plants during secondary forest succession.

IMPORTANCE OF SOIL MICROBES IN EXPLAINING FOREST SUCCESSION

The previous work enhanced our understanding of PSF effects on old-field and grassland succession by examining specific microbial functional groups and comparing herbaceous adult plant biomass in conspecific soil with heterospecific soil, and provided evidences that the accumulation of pathogens and mutualists in conspecific soils generally had negative, neutral and positive effects on early-, mid- and late-successional species, respectively (Kardol et al. 2006, 2007; Koziol and Bever 2015, 2019). However, the microbe-mediated mechanisms driving secondary forest succession are still unclear, and the PSF effects among adult herbs regulating old-field and grassland succession cannot adequately predict the initial seedling establishment of forest mid- and late-successional species. For example, at the initial stage of forest mid-successional seedling recruitment i.e. expected to experience a nearly neutral PSF effect, question remains regarding why the few small-size shrub seedlings can successfully colonize in the presence of large amounts of adult herbs. Limiting fast-growing herb performance via negative PSF (induced by pathogens) is likely to introduce other competitive herbs and lead to herbaceous species coexistence and stability of grassland, which may not be enough to maintain the continuous growth and survival of slow-growing shrub seedling. Adult shrub can certainly exhibit competitive advantage for aboveground resource over herb due to its larger size. However, small shrub seedling with no size-asymmetric aboveground competitive advantage is not able to colonize as fast as herb and exhibits lower competitive ability for belowground resources due to its allocation patterns (e.g. lower root:shoot ratio, higher inversion in unproductive tissues such as stems) (Köchy and Wilson 2000). Secondly, pioneer plant species (e.g. pioneer shrubs and trees) are found to be more poorly defended against soil-borne pathogens than non-pioneer species (Domínguez-Begines et al. 2020; Pizano et al. 2014). If pathogens play a major role in seedling establishment, the non-pioneer shrub or tree species would preferentially colonize instead of pioneer species at mid- and late-successional stages, disagreeing with the successional pathway. Finally, given that late-successional tree seedling is expected to perform worse in mid-successional soil with fewer beneficial symbionts (e.g. ectomycorrhizal [ECM] fungi) than in its conspecific soil (Pfennigwerth et al. 2018), the most pressing question is how the initial establishment of tree seedling succeeds without strong positive PSF from co-occurring conspecifics. Recently, a dynamical modelling framework revealed that the accumulation of a specific microbial group (e.g. mutualist) could be a necessary but insufficient condition for generating successional dynamics (Jiang et al. 2020). Since most plant species simultaneously interact with various soil microbial groups, it is important to consider the net effects of soil total microbial communities to unravel the biological mechanisms driving initial seedling establishment during forest succession.

POTENTIAL EVIDENCES FOR THE HYPOTHESIS DURING SECONDARY FOREST SUCCESSION

Pre-existing vegetation can have great impacts on the establishment of later species during secondary forest succession. For example, a meta-analysis focussed on published studies conducted under natural field conditions in degraded terrestrial ecosystems, and suggested that herb species generally exerted weak negative effects (including abiotic and biotic effects) on the growth and survival of shrubs (Gόmez-Aparicio 2009). Since herb species compete efficiently for soil nutrients and largely suppress shrub seedling performance, we assume that soil biotic factors (e.g. microbial communities) conditioned by herbs may generate positive feedback effect and contribute to the successful establishment of shrub seedlings usually observed in grasslands. As evidenced by the publications in Table 1, a recent study revealed that the compositions of soil functional fungal groups (e.g. ericoid and arbuscular mycorrhizal [AM] fungi) did not change across the succession from pure grassland to young and mature shrublands in the Pyrenees (Grau et al. 2019). It suggested that the beneficial fungal OTUs (operational taxonomic units) that were common in shrublands were already present in grassland, possibly as propagules, such as spores or explorative mycelia. In contrast, significant compositional changes in plant pathogenic fungi were observed along the succession (Grau et al. 2019), indicating that the fungal pathogens in grassland soil exhibited host specificity and may not infect the shrub species in shrublands. Our field investigations provided evidence that pathogenic fungi (e.g. Peroneutypa, Gibellulopsis and Cladosporium) persisted in grassland soils and herb roots, whereas these pathogens were not detected in shrub roots and probably did not infect shrubs (Zhang et al. 2022, 2023). Further, our greenhouse experiment confirmed that soil microbes from three herbaceous species (Poa annua, Koeleria macrantha and Anemone rivularis) had positive feedbacks on the seedling growth of shrub species (Berberis sichuanica) compared with sterilized soil, and herb soils shared similar AM fungal species (e.g. Glomus indicum, Paraglomus laccatum and Septoglomus constrictum) with shrub seedling (data not published). In summary, soil beneficial fungal groups from herbaceous species in grassland rather than pathogens may drive secondary succession by facilitating the expansion and dominance of shrubs (García de León et al. 2016).

Dissimilarly, herbs were found to have a significant negative neighbour effect on tree seedling emergence and growth (Gόmez-Aparicio 2009). It has been reported in greenhouse experiments that live soils from grasslands had strong negative impacts on pioneer and non-pioneer tree seedlings, and soil sterilization had positive influences on seedling growth, suggesting that the effects of antagonistic microbes were greater than mutualistic ones in grassland soils (Pizano et al. 2014). Liang et al. (2022) also found that the soils conditioned by herbs significantly lowed the seedling biomass of pioneer tree species, due to the increased accumulation of fungal phylum Blastocladiomycota that are considered pathogens of vascular plants. Actually, forest disturbances leave soil biotic legacies that influence the recovery processes of the post-disturbance ecosystem (e.g. early-successional grassland) (Jõgiste et al. 2017). The abundances of mycorrhizas and ECM fungal diversity declined with increased fire and logging intensity in forestland, and severe forest disturbance with high tree mortality can kill most of the existing ECM community, therefore leaving few soil legacies of ECM fungal communities that may negatively affect subsequent tree establishment (Dahlberg et al. 2001). Meanwhile, soil saprotrophic phosphorus-solubilizing fungi (e.g. Penicillium spp.) were less abundant in open land sites after forest disturbance than mature and regenerating forests, and facultative pathogenic fungi (e.g. Fusarium) were found to make up a large proportion of the soil fungal community in deforested sites, which may constrain seedling recruitment and reduce the rate of secondary forest succession (Shi et al. 2019). Recent work also showed that the active functional fungal community in grassland soil shifted from pathogenic fungi to beneficial fungi over the course of succession (Hannula et al. 2017). Several pathogenic fungal species (e.g. Peroneutypa scoparia, Gibellulopsis nigrescens and Cladosporium pseudocladospori) were found in early-successional grassland soils (Zhang et al. 2022, 2023), which could infect the roots of late-successional tree seedling (Picea asperata). Hence, the larger accumulation of pathogens for tree diseases relative to beneficial mycorrhizal fungi after deforestation is likely to result in net negative microbial feedback effect and inhibit tree seedling performance in early-successional grassland.

Shrub can be used as nurse plant to facilitate tree seedling recruitment, which is usually attributed to their benefits of proximity and buffering harsh abiotic conditions (e.g. microclimate amelioration, increased soil fertility and protection against herbivory) (Padilla and Pugnaire 2006). In addition to the positive abiotic effect, it has been reported in a subtropical forest that soil inoculum from mid-successional species can increase AM fungal colonization and the competitive advantage of late-successional species (Liao et al. 2018). The soils conditioned by shrubs favour more ericoid and ECM fungi than non-conditioned soils (Pfennigwerth et al. 2018). Shrub root systems support fungal taxa (dark septate endophytes, ericoid and ECM fungi) compatible with adjacent tree seedlings that establish after wildfire (Hewitt et al. 2017), and the ECM fungi associated with established shrubs are essential in facilitating subsequent tree seedling establishment (Nara and Hogetsu 2004). The increasing abundance and diversity of mycorrhizal fungi can protect their host plants from soil-borne pathogens and environmental stress, reduce the detrimental effects of allelochemicals and improve overall nutrient acquisition by plants (Tedersoo et al. 2020). Further, Liang et al. (2022) performed a greenhouse PSF experiment and confirmed that tree seedling (Betula platyphylla) grew better in the soils conditioned by mid-successional shrubs compared with early-successional herbs, which was ascribed to the accumulation of the plant growth-promoting bacterial phylum Armatimonadetes and fungal genus Acremonium. The bacterial and fungal communities were the key explanatory factors for tree biomass variation. A recent study also reported that more abundant ECM fungi, chemoheterotrophs, sulfate-respiration bacteria and bacterial taxa involved in aromatic compound degradation were found in shrubland soil than in grassland soil (Zhang et al. 2023). These soil functional microbes can promote organic compound degradation and increase soil nutrient content. Moreover, the soil biotic legacies (e.g. pathogens) from deforestation will decline in strength over time as a result of decomposition of dead trees and woody debris (Jõgiste et al. 2017; Köster et al. 2009, 2015). The mid-successional shrubland soils may contain few species-specific pathogens for trees due to weak biotic legacy from pre-disturbance forest and lack of adult host trees. Thus, we speculate that the modification of soil microbes by shrubs would enhance the initial colonization of tree seedlings.

SUCCESSIONAL STAGE HYPOTHESIS

Based on the above discussion and potential evidences, we propose a ‘successional stage hypothesis’ (Fig. 1) to systematically explain why the seedling recruitment of shrub and tree species follows a successional sequence rather than emerging simultaneously in spite of the presence of their seed banks following deforestation. Dissimilar to the previous work comparing the plant performance in conspecific live soil with heterospecific live soil, this hypothesis emphasizes the net plant–soil microbe feedback effect (seedling growth with versus without soil microbes). We hypothesize that the direction and strength of microbial feedback effects are dependent on successional stage. On the one hand, due to the differential roles of soil microbes in successional stages to accelerate the succession process, the direction of feedback effect may be expected to vary from positive to negative. Pre-existing adult herbs and soil microbial legacies from disturbed forests would directly (via litter and residue decomposition and rhizosphere effects) and indirectly (via changes in microclimate and soil abiotic properties) affect the soil total microbial communities (including mycorrhizal fungi, saprotrophic fungi, pathogens, plant growth-promoting bacteria, etc.), which promote the initial seedling recruitment of subsequent shrub species (mid-successional stage), but inhibit the performance of tree seedling (late-successional stage). The soil microbes conditioned by pre-existing mid-successional shrubs would promote the seedling recruitment of subsequent late-successional trees. On the other hand, the positive feedback strength may decline with the increase in temporal interval between two plant species from different successional stages, and the negative feedback strength may increase with increasing temporal interval. For example, the strength of positive feedback effect from herbaceous species would be stronger for pioneer shrub than non-pioneer shrub to favour the prior colonization of pioneer shrub, while the negative feedback strength for pioneer tree would be weaker than that for non-pioneer tree. This hypothesis may also hold true for the positive feedback effect of soil microbes (including the overall net effect of the combination of microbial biomass, abundance, composition, diversity, enzyme activity, etc.) conditioned by shrub species on pioneer and non-pioneer tree seedlings.

It should be noted that our hypothesis mainly focuses on the strong feedback effect of pre-existing adult plant on subsequent plant seedling recruitment. The initial stage of seedling establishment may generate weak or negligible reverse feedback effect on the pre-existing adult plant, due to the very limited amount of litter and root exudate produced by the small seedling. We do not consider the role of soil nutrients or the resource competition among plant species with different life forms during secondary succession, because the decomposition of abundant plant residues after forest disturbances can leave enough soil nutrient legacies that may support seedling growth (Veldkamp et al. 2020).

TESTING THE SUCCESSIONAL STAGE HYPOTHESIS IN A SUBALPINE FOREST

We conducted a greenhouse experiment to test the hypothesis and make it more convinced (Zhao et al. 2023). Along the successional sequence in a subalpine forest, seedlings of four tree species (white birch, red birch, dragon spruce and Faxon fir) were transplanted in pots with soil inocula from rooting zones of early-successional herbs, mid-successional shrubs and late-successional trees. Soil mixtures in pots for seedling growth were prepared by mixing sterilized background soil (93% of volume) with rooting zone soil (7% of volume), which could ensure that plant performance would be mainly affected by biotic characteristics of these inocula instead of abiotic effects (Teste et al. 2017). We calculated the biotic feedback index (log₁₀ ratio of seedling total biomass in inoculated live soil and sterilized background soil) (Semchenko et al. 2018). Putative ecological guilds were assigned to fungal sequences using the FUNGuild database (Nguyen et al. 2016). Obtained results showed that early-successional soils contained more pathogenic fungi (e.g. Bionectriaceae, Botryosphaeriaceae and Nectriaceae) than mid- and late-successional soils, which exerted negative feedbacks and inhibited tree seedling growth. These pathogens can cause canker diseases or branch dieback for woody trees such as Pinus, Pterocarya, Eucalyptus and Guapira species (Burgess et al. 2019; McDermott-Kubeczko et al. 2020; Spear and Broders 2021). Additionally, soil biota from mid-successional stage generated positive feedbacks and facilitated tree performance, while fungal species (e.g. Sphaerosporella brunnea and Wilcoxina rehmii) within the dominant family Pyronemataceae were strongly enhanced in mid-successional soils. These ECM fungi are known to form ectomycorrhizas with Picea spp., hydrolyse complex compounds (cellulose, oil and pectins) and increase the resistance of spruce seedlings to root fungal pathogens (Perry et al. 2007; Peterson 2012). Further result showed that soil microbes from herbs had positive feedbacks on the seedling growth of shrub species (B. sichuanica) (data not published). The above findings highlighted the roles of plant–soil microbe feedbacks during secondary forest succession and provided direct evidences for our hypothesis.

Moreover, we examined the effects of soil microbial legacies of herbs and shrubs on the root traits of coniferous tree seedlings (Wang et al. 2023). Seedling root growth and ECM colonization on root tips were facilitated when grown in the soil microbial legacy of shrubs rather than herbs. Higher root biomass was significantly correlated with the increase of ECM fungal colonization percentage as well as the decrease of soil inorganic nitrogen content. The soil legacy from shrubs shared the same ECM taxa at the genus level (e.g. Wilcoxina, Hebeloma, Sphaerosporella, Tuber and Trichophaea) as those colonized on tree seedling roots. The above results provided evidences that the soil fungi from mid-successional shrub could enhance ECM colonization and nutrient uptake by seedling root and thus facilitate tree growth.

OUTLOOK

In forest restoration activities, the direct introduction of target tree species into early-successional habitat is often accompanied by high ecological and economic costs. To overcome the problem of strong inhibition of tree species by herbs as well as the large positive effects of shrubs on trees, a two-phase restoration strategy (shrubs are first planted into the herb layer, and tree species are later introduced under the shrub cover) should be considered a promising strategy for restoring forests (Gόmez-Aparicio 2009). Our hypothesis focuses on the feedback effects of pre-existing adult plants on subsequent plant seedling growth, and provides a possible explanation for the soil microbe-related PSF mechanisms underlying the two-phase restoration strategy. Well-designed microbial inoculation, pot culture and field experiments along a successional chronosequence will allow for testing this hypothesis and reverse effects using next-generation sequencing and genomic analyses.

The factors regulating feedback direction and strength, which vary from positive to negative among successional stages, may depend on soil microbial characteristics, plant growth form and functional trait and forest type. Various forest types contain plant species of different life-history traits, and experience different disturbance regimes (e.g. windstorm, insects, disease, fire or logging) under varied soil and climatic conditions (e.g. soil nutrient, light and water availability) (Poorter et al. 2019; Pulsford et al. 2016; Taylor and Chen 2011). Although we list some evidences for the hypothesis in several subtropical and subalpine forests, other forest types may follow dissimilar successional processes that could be governed by quite different biotic drivers and mechanisms. Whether the successional stage hypothesis is always true in different forest types or climate zones is still in doubt, and it requires further evaluations using different plant species across various forests.

It has been established that forest succession is highly unpredictable and stochastic, and the context-dependent factors can modulate forest successional trajectory (Norden et al. 2015; Poorter et al. 2019; Powers and Marin-spiotta 2017). For example, the forest succession in Dinghushan Biosphere Reserve of subtropical China can be roughly divided into three stages, which disagree with the sequence from grassland, shrubland to forest. The early, middle and late stages are characterized by the dominances of fast-growing conifers and shrubs, a mixture of conifers and heliophytic broadleaf trees, and slow-growing broadleaf trees, respectively (Liao et al. 2018). Due to the distinct successional pathway that may exist in a specific forest ecosystem, the applicability of the hypothesis needs to be further tested.

Soil microbial communities can vary more quickly and be more sensitive to environmental change at short time scale than plants. The strength of plant–soil microbe interactions may therefore vary temporally, due to the changes in soil microbial community composition with increasing time of soil conditioning (Lepinay et al. 2018). Actually, most of the previous PSF studies were performed via short-term glasshouse experiments and terminated at the same time for all species (Cheeke et al. 2019; Koziol and Bever 2019). The temporal variation in microbe-mediated PSF was rarely quantified and its impacts on plant performance and succession remain largely unknown. Hence, our hypothesis should be tested using multiple sampling and repeated molecular analysis of soil microbial communities to explore the potential consequences of temporally varying PSF for forest succession. Additionally, knowledge about the temporal changes in the proportional contribution of each microbial component to successional stage needs to be improved. It requires a broader taxonomic survey of plants being influenced by microbial communities and a deeper understanding of the temporal patterns of causal mechanisms through which the specific microbial taxa directly or indirectly affect plant performance.

Funding

This work was supported by the National Natural Science Foundation of China (41930645 and 31870607), Youth Innovation Promotion Association of the Chinese Academy of Sciences (2019363) and National Key Research and Development Program of China (2017YFC0505000).

Acknowledgements

We thank Prof. Leho Tedersoo (Natural History Museum of Estonia), Xiao-Hu Wang and Yong-Ping Kou (Chengdu Institute of Biology) for providing valuable comments on the manuscript.

Conflict of interest statement

The authors declare that they have no conflict of interest.

REFERENCES