DISCUSSION
Although the magnitude of global warming in the tropics is lower than that in high latitudes (Deutsch et al. 2008; IPCC 2013), warming is likely to have more severe deleterious impacts on species in the tropics than those in the temperate and boreal zones because of low tolerance of tropical species to temperature fluctuations (Ghalamboret al. 2006; Deutsch et al. 2008; Wright et al.2009; Khaliq et al. 2014; Chan et al. 2016). Rising temperatures in the tropics are directly responsible for tree growth reduction (Feeley et al. 2007), increased tree mortality (Aleixoet al. 2019), and species range shift (Khaliq et al. 2014; Chan et al. 2016). Besides these direct effects, global warming could also have many as consequential indirect effects (Liu & He 2019; Pugnaire et al. 2019), of which the warming effect on PSF is critical not only to maintaining tropical tree diversity (Bell et al. 2006; Mangan et al. 2010; Bagchi et al. 2014; Ecket al. 2019; Schroeder et al. 2020) but also to regulating soil carbon cycling (Nottingham et al. 2020). However, it has not been well understood how warming may affect PSF and how that could in turn affect tropical biodiversity. Our study showed that warming decreased the hazard ratio of seedling mortality in the rhizobia-associated O. semicastrata but had no effect on that of the EcM fungi-associated C. patelliormis (Figure 2). Such differential effects of elevated temperature on the survival of heterospecific seedlings is new to the literature and could have profound implications to understanding climate impact on tropical tree diversity and shed mechanistic light on the debate of biodiversity change (Vellend et al. 2013; Gonzalez et al. 2016). Biodiversity change driven by changes in land use and climate does not have to involve the net loss (or gain) of species; change occurs if the abundances of constituent species shift (Dornelas et al. 2014). The different mortalities observed for the rhizobia-associated and the EcM fungi-associated trees afford an example showing that global warming could facilitate the survival of species with certain biological traits while impede those without, thus leading to change in tropical tree composition under climate change (Johnson et al. 2018). This finding adds an explanation why change in community composition has been so widely observed and become a major pattern of biodiversity change (Dornelas et al. 2014; Hillebrand et al. 2018).
Although the individual roles of mycorrhizal fungi and pathogenic fungi in regulating plant diversity have well been appreciated (van der Heijden et al. 1998; Bell et al. 2006; Liu et al.2012; Corrales et al. 2018), increasing evidence has shown what matters to biodiversity maintenance is the net (interactive) effect of beneficial and antagonistic fungi (Merges et al. 2018; Liu & He 2019; Liang et al. 2020; Schroeder et al. 2020; Germain & Lutz 2021). Our study shows that elevated temperature decreased the relative abundance of plant-pathogenic fungi (Table 1) while increased the abundance of EcM fungi (Table 2). This temperature mediated tradeoff between the two fungal guilds would inevitably alter the net outcome of PSF, depending on the relative magnitude between the positive PSF induced by beneficial microbes (e.g., EcM fungi) versus the negative PSF induced by antagonistic microbes (e.g., plant-pathogenic fungi) (Tables 1 and 2). This is consistent with previous observations that the decreased abundance of plant-pathogenic fungi and the increased abundance of EcM fungi could alleviate the negative PSF (Corraleset al. 2018; Chen et al. 2019; Liu & He 2019). The warming weakened negative PSF observed in this study is not surprising because the summer temperature of 27.5 °C in our study area (Xu et al. 2015) is already higher than the optimal temperature (20-25 °C) for pathogen reproduction, and an increase in temperature of 0.89 °C (Figure 1c) would make the temperature suboptimal to pathogens. The effect of climate change on fungi is particularly pronounced because fungi have been observed to be especially sensitive to temperature change (Alsteret al. 2018). As such, a weakened negative PSF (the JC effect) under even moderate warming would likely result in a decline in tropical tree species diversity.
The rhizobia-associated O. semicastrata , carrying coevolved host-specific soil-borne pathogens (Li et al. 2009), experienced a weakened negative PSF under warming due to the decreased abundance of plant-pathogenic fungi and the increased abundance of EcM fungi. However, different from the sensitiveness of O. semicastrata to the differential changes of the two fungal guilds under warming, seedling mortality of the EcM fungi-associated C. patelliormis , immune to soil-borne pathogens (Figure 2b), was unaffected by warming. This is possibly because EcM fungi can form Hartig nets around the roots of their host species that develops a line of defense against soil-borne pathogens (Marx 1972; Bennett et al. 2017; Segnitz et al.2020; Tedersoo et al. 2020). Once Hartig nets have been formed around the roots (to prevent pathogens’ attack), the EcM fungi-associated tree species may no longer be sensitive to the change in load levels of plant-pathogenic and EcM fungi.
We would like to note two possible limitations in our study. One is that the functional classification of fungi by the FUNGuild database is considered coarse (Nilsson et al. 2019; Guo et al. 2020) although the system is widely used to classify fungal guilds (Hannulaet al. 2017; Mommer et al. 2018; Chen et al. 2019; Vetrovsky et al. 2019; Delgado-Baquerizo et al. 2020). By quantifying the responses of plant-pathogenic fungi and EcM to warming, we attributed the decreased seedling mortality of O.semicastrata (Figure 2a) to the decreased abundance of plant-pathogenic fungi (Table 1) as well as the increased abundance of EcM fungi under warming. We suggest future studies to collect data to further ascertain their specificity of the 421 FUNGuild classified pathogens to seedlings of O. semicastrata , e.g., using fungal species-specific markers (Romero et al. 2021).
The second possible limitation is that the warming simulation of OTCs also dried soils (Figure S2), a well-recognized side effect of OTC experiments (Hollister & Webber 2000). The good news is that these dual effects of OTCs happen to follow the typical climate pattern in the tropics, i.e., reduced precipitation (or increased drought) and increased temperature (Malhi et al. 2008; Bachelot et al.2020). Although no correlation between the change in OTC soil moisture and the soil temperature was detected in our study (Figure S2), changes in pathogen-induced seedling mortality under warming could be magnified because of the high moisture preference of plant-pathogenic fungi (Swinfield et al. 2012; Liu & He 2019; Romero et al.2021). Moreover, EcM fungi are known to improve drought tolerance of host plants (Wang et al. 2021), and thus would increase seedling survival under drought conditions (Brunner et al. 2015).
In conclusion, our three-year OTC warming experiment shows that the effects of global warming on seedling performance of tropical trees is not simply positive or negative, but depends on the net changes in the PSF. Our results indicate that warming weakens the negative PSF on rhizobia-associated host tree species owing to the reduced relative abundance of plant-pathogenic fungi and the increased relative abundance of EcM fungi, while there is no detectable benefit for EcM fungi-associated tree species. The differential warming effects on seedling mortality of tree species with different microbial associations could have a profound impact on tree species composition in tropical forests. Our study predicts that the weakened negative PSF could undermine the role of the JC effect in maintaining tree species diversity in the tropics.