Drivers of community composition
Although no consistent evidence of deterministic succession in endophyte communities was observed; there was some evidence of deterministic processes shaping the community assembly of fungal endophyte communities. Host identity had the largest impact on endophyte community composition (Figure xx, Table xx). Host identity alone explained ~23% of the total ~29% variation in turnover of endophyte composition explained by the final model. Host identity is known to be a major determinant of the composition of foliar fungal endophyte communities (Christian et al. , 2016; Davidet al. , 2016; Vincent et al. , 2016; Liu et al. , 2019; Yao et al. , 2019). The specificity of endophytes to a particular host or group of taxonomically related hosts ranges from extremely narrow to relatively wide (Arnold and Lutzoni, 2007; Põlmeet al. , 2018). This may explain why, even though host identity was the important factor structuring foliar fungal endophyte communities, the identity of only two host species was retained in the final model, with C. inerme consistently being removed as one of the variables which explained the least turnover in endophyte community composition.
Geographic distance between samples was the only significant variable consistently retained in every GDM (i.e. the full dataset and each of the three hosts individually) (Figure 2 & Figure 3)). This indicates that the similarity in composition of foliar fungal endophytes decays with increasing distance, a pattern commonly observed in ecological communities (Nekola and White, 1999; Soininen et al. , 2007). Indeed, others have shown that foliar fungal endophytes are dispersal limited over distances from 10km to >100km (David et al. , 2016; Koide et al. , 2017; Oono et al. , 2017). However, here we show that even over a shorter spatial distance, i.e. <1500m, there is consistent evidence for distance decay driving turnover in endophyte community composition.
The GDMs performed for each host individually, showed that the factors responsible for the turnover in foliar fungal endophyte community composition within a host is variable as is the rate and magnitude of this turnover (Figure 3, Supporting information Table S5-S7). For example, distance to the indigenous forest, the difference in minimum temperature and geographic separation were the most important variables explaining the turnover in foliar fungal endophyte communities forE. crispa, C. inerme and S. chirindensis respectively, holding all other variables constant in their individual models. Others have shown how the distance to a potential inoculum source, in our case a large indigenous forest, can be important for causing compositional differences in fungal communities (Glassman et al. , 2017). It has also been shown how abiotic conditions, like temperature differences, can shape endophyte community composition due to differences in the physiological performance of endophyte species in dealing with the prevailing conditions (Arnold and Herre, 2003; Peay et al. , 2016; Unterseher et al. , 2016). Such findings highlight how both deterministic (i.e. temperature differences) and stochastic (i.e. distance to an inoculum source or geographic separation) forces can be important factors causing turnover in foliar fungal endophyte communities in the same system but for different hosts.