Influence of endemic versus cosmopolitan species on the local assembly of ectomycorrhizal fungal communities

An enduring question of community ecology is how remarkably speciose communities can co-exist, particularly in light of competitive exclusion (Hutchinson, 1961). Ectomycorrhizal fungal (EMF) communities are no exception to this paradox; in a survey of old-growth Pseudotsuga menziesii stands, Kranabetter et al. (2018) reported a metacommunity (γ diversity or regional species pool) of c. 360 EMF species on mesotrophic sites across a single watershed. Efforts to experimentally test EMF species interactions and community assembly can be insightful, but also hampered by the sheer number of taxa and difficulties in culturing an obligatory symbiosis under laboratory conditions (Kennedy, 2010; Branco, 2019). Similarly, the quantification of key morphological traits related to species distribution patterns, such as spore size and shape, can be instructive but inconclusive (Halbwachs et al., 2015; Kilvin, 2020). Given the challenges involved in an inductive approach, it may be advisable to employ perspectives from broader theoretical models to better understand the assembly of this ecologically important group. In this Letter we describe how distinctions in geographical range for EMF species (regional to continental) can be used to better understand spatial processes governing local community assembly. A useful conceptual model for EMF communities should foremost explain how large numbers of species can co-exist in a uniform environment while in competition for a limited resource (Palmer et al., 2003). Earlier niche models focused on competitive exclusion, where a co-existence equilibrium was only thought possible if species were less similar than expected by chance (Diamond, 1975). Modern niche models, however, consider factors beyond consumable resources, such as space, to play a significant role in determining community structure. An example of this is the patch-dynamic model, which predicts that the total number of species can be increased by permitting the co-existence of functionally similar species through a trade-off between colonization and competition ability (Tilman, 1994; Leibold et al., 2004). This proposed trade-off between colonization and competition strategies has been examined successfully at a fine scale for select EMF species (Kennedy et al., 2011; Moeller et al., 2016; Smith et al., 2018), but extrapolating these processes to the diverse EMF metacommunities of natural forest landscapes remains challenging (Pickles et al., 2012; Bahram et al., 2016). With regards to colonization capacity, the dispersal of free-living eukaryotes has been suggested to follow a random, uniform probability distribution (Finlay, 2002), and, moreover, long-distance dispersal has been shown to play a role in the global distribution of some fungal species (Moncalvo & Buchanan, 2008; Harrower et al., 2015). Considering the propensity for widespread dispersal, we might expect most EMF species to exhibit a large geographic range. Nonetheless, the improved availability of a unified genomic database for EMF species (Nilsson et al., 2019) has revealed a substantial cohort of endemic species within EMF communities (Talbot et al., 2014; Kranabetter et al., 2018). The pervasiveness of endemic species, as opposed to more cosmopolitan members, underscores the nonrandom dispersal behaviour among EMF taxa and, consequently, we propose that endemism is relevant to models of colonization-competition trade-offs. Specifically, we expect a cosmopolitan group to reflect a more uniform distribution over space, indicative of an effective long-distance colonization strategy, whereas an endemic group would display higher species turnover over the same distances, indicating a higher degree of patchiness predicted by a competitive strategy (Durett & Levin, 1998). To address this hypothesis, we re-examined the EMF communities of late-seral Pseudotsuga menziesii stands in dry maritime forests of Vancouver Island (British Columbia), as described in Kranabetter et al. (2018), ranging from 0.05 to almost 20 km apart. The 11 study plots (0.15 ha in size) were selected for their uniformity in stand composition, soil fertility and elevation so as to isolate spatial effects on community assembly. A total of 238 species from 50 genera were recorded, dominated by many Cortinarius, Piloderma, Ramaria, Russula and Tricholoma species typical of well-drained soils with modest nitrogen availability (see Kranabetter et al. (2018) for further details on the EMF communities, soil chemistry, and forest/vegetation composition). Endemics were defined as taxa where all accession records have been located only in western North America (usually British Columbia through to northern California) using a 99% Species Hypothesis (SH) threshold value in Unite (v.08; Nilsson et al., 2019). Cosmopolitan species under the same 99% threshold were distributed much further abroad (e.g. eastern North America, Europe or Asia). Western North America was considered an appropriate range for endemics due to geographic barriers (e.g. Pacific Ocean, Rocky Mountains), and because the number of SH accession records usually precluded any finer resolution in geographic distribution. A threshold value of 99% was chosen as the most precise for delineating an SH geographic range using the entire internal transcribed spacer (ITS) region, particularly for diverse genera such as Cortinarius (Garnica et al., 2016). We used Mantel tests to quantify the distance decay and β diversity of endemic and cosmopolitan subcommunities. The randomness of residual plots was used to assess the appropriateness of fitting a linear or segmented regression model to the distance-decay relationship. We calculated and compared (Welch t-test) three types of pairwise dissimilarity matrices for each subcommunity based on species incidence: the Sorensen (βSOR), representing total β diversity, and its additive components of turnover (βSIM) and nestedness (βNES) (Baselga, 2010; Supporting Information Methods S1). Species turnover reflects the degree to which species are replaced from one community to the next, without necessarily a net change in plot-level (α) diversity. Dissimilarity driven by nestedness, by contrast, implies that species-poor sites are subsets of species-rich sites. Quantifying these effects can provide further insight into community assembly since dominance by either component can produce identical Sorensen dissimilarity metrics. The proportion of endemic to cosmopolitan species in the metacommunity was well balanced overall (122 endemic species vs 116 cosmopolitan) and among plots (average of 41.6 (SE 1.79) and 43.0 (SE 1.91), respectively) (Fig. 1). Despite the similarity in α diversity, our findings suggest some significant differences in the spatial patterns between endemic and cosmopolitan subcommunities. The endemic group had significantly higher average Sorensen dissimilarity than the cosmopolitan group, and a less significant Mantel correlation across interplot distance (Table 1; Fig. 2a). A post hoc curve fitting test, however, suggests the endemics had a segmented, nonlinear relationship (P = 0.001, R2 = 0.27) that was not evident for the cosmopolitans (P = 0.97) (Fig. 2a). The break point in Sorensen dissimilarity for the endemic group was 4.3 ± 0.8 km, after which distance decay was elevated over cosmopolitans and did not vary with interplot distance (Fig. 2a). Neither subcommunity displayed a significant Mantel correlation in species turnover (βSIM) but the endemic group had a significantly higher overall rate of turnover than cosmopolitans (Table 1). The post hoc curve fitting indicated the endemic group had a nonlinear relationship in βSIM as well (P = 0.002, R2 = 0.25), in contrast to cosmopolitans (P = 0.48), albeit because of a small number of plots < 2 km apart (Fig. 2b). Lastly, we found a significant Mantel correlation in nestedness (βNES) for only the cosmopolitan group, but overall this component contributed a minor amount to β diversity for both subcommunities (Table 1; Fig. 2c). Greater rates of endemic species turnover, especially beyond c. 4 km spacing, emphasizes the discrete, infrequent distribution of many endemic taxa. The evidence for a reduction in βSOR at less than c. 4 km, while too fine to affect the Mantel calculation, would be consistent with similar nonlinear distance-decay patterns at local scales (Bahram et al., 2013). This fine-scale clustering suggests a more prevalent strategy of competition rather than colonization for the endemic subcommunity (Durett & Levin, 1998; Leibold et al., 2004; Chase & Myers, 2011). Patchy distributions may result from limited spore dispersal by endemic species and a greater reliance instead on vegetative spread to colonize new roots and persist at microsites over time. Similar colonization patterns have been repeatedly revealed in studies of green-tree retention and forest edges, where the greater diversity of EMF communities on seedlings is associated with root and hyphal contact with mature trees (Hagerman et al., 1999; Kranabetter, 2000; Cline et al., 2005; Dickie & Reich, 2005). We also expect that many endemics of this landscape are mid- and late-seral stage dependent, which typically colonize forests only after canopy closure (e.g. certain species of Ramaria, Amanita, Tricholoma, Russula; Kranabetter et al., 2005), rather than multi-stage fungi (present in all forest age classes), thereby mirroring short- and long-distance colonization strategies during EMF succession (Redecker et al., 2001; Twieg et al., 2007; Kyaschenko et al., 2017). Such a temporal relationship would be in agreement with stochastic spatial competition models, which predict early-seral systems to be dominated by colonizer (primarily cosmopolitan) species, and then, towards later stages of succession, displaced by competitor (often endemic) species if population density is sufficient (Durett & Levin, 1998). Although some endemic EMF species are undoubtedly specific to Pseudotsuga, many of these taxa have been found with other tree hosts (e.g. Tsuga, Pinus and Abies) in western North America, suggesting that host-specificity is unlikely to be the sole constraint to a more widespread distribution. Endemism is generally driven by either the downsizing of a species range due to historical events, or speciation after immigration, both of which are affected by environmental history as well as biological traits (Bruchmann & Hobohm, 2014). The speciation rate hypothesis claims that the isolation associated with lower dispersal behaviour can decrease gene flow and increase speciation, which could generate a negative relationship between recent endemism and dispersal (Papadopoulou et al., 2008). Increased speciation promoted by a localized dispersal strategy may then, to some degree, suggest specialization to the local ecosystem (Buchi & Vuilleumier, 2014), perhaps as a trade-off for the reductions in geographic range (sensu Moeller & Peay, 2016; Pellissier, 2015). Meeds (2020), for example, found that coastal endemics of British Columbia, such as Lactarius luculentus (SH1679693.08FU), had greatly enhanced exoenzyme production on phosphorus deficient soils. Other west coast endemics, including Lactarius cf subviscidus (SH1679722.08FU), also display superior uptake capacity of NH4+ on highly productive sites (Kranabetter et al., 2015). The spatial relationships between endemic and cosmopolitan EMF taxa explored in this study may thus tie into a larger inquiry regarding how evolutionary and ecological processes affect the considerable species richness found at these temperate latitudes (Kennedy et al., 2012; Sanchez-Ramirez et al., 2015). In comparison to endemics, the cosmopolitan group displayed significantly lower species turnover across the watershed, with only a subtle increase in distance decay due to a small nested effect (net change in dissimilarity of 0.05). This trend might be explained by the slightly greater disparity in cosmopolitan species richness among plots compared to endemics, particularly the communities at Second and Contour creeks (Fig. 1). Given the uniformity in plot conditions across the watershed, we suspect the more frequent co-occurrence of cosmopolitan species at close distances, with some minor loss of incidence at further spacing, was not due to any particular environmental filter but instead a result of stochastic (neutral) processes such as priority effects (Kennedy, 2010). In either case, the 'core' group of frequent and often abundant cosmopolitan species found in these coastal stands of Vancouver Island, such as Cenococcum geophilum (SH1830453.08FU), Hebeloma cf velutipes (SH1733362.08FU), Laccaria bicolor (SH1719822.08FU) and Piloderma cf olivaceum (SH1709468.08FU), have been accessioned repeatedly over large expanses of global forests, underscoring both their superior dispersal abilities and possibly a wider, more generalist niche than most endemics. In conclusion, we compared patterns in β diversity between the endemic and cosmopolitan taxa that co-exist in these diverse, late-seral EMF communities. The significantly higher rate of species turnover, along with a nonlinear correlation in dissimilarity across space, indicated more localized dispersal and patchy distribution of endemic EMF species. These patterns illustrate how endemism at a continental scale can, in turn, be reflected by limits in colonization and greater species turnover at a local scale. Given the difficulty in obtaining life history traits for all EMF species within such diverse metacommunities, the ability to readily separate EMF communities into endemic and cosmopolitan subgroups using global genomic databases offers an expedient method to explore local assembly processes. Thanks to Barbara Hawkins (University of Victoria) and Peter Ott (B.C. Ministry of Forests) for support and helpful discussions. Funding for this project was provided by the British Columbia Ministry of Forests, Lands, Natural Resource Operations and Rural Development. MCM and JMK conceived study. JMK contributed community surveys, while MCM undertook data analysis and wrote the manuscript with contributions from JMK. Methods S1 Dissimilarity and distance decay calculations. Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.