Root associated fungi were isolated from the host plants following the protocol described by Hambleton and Currah (1997) with some modifications. Root systems were first thoroughly washed in tap water to remove soil, debris, foreign roots, and ericaceous roots not physically connected to the host plant. Healthy, fine hair roots (as described by Vohník, 2020) were excised from the root system using scissors. The hair roots were divided into two subsamples, which were subjected to a weak or a stringent surface sterilization protocol. Two surface sterilization protocols were used, as Smith and Read (2008) indicate that even short duration chemical surface sterilization can kill some fungi growing within plant cells. Half of the fine roots were subjected to one minute surface sterilization in 1% NaOCl, while the remaining were subjected to a 3-minute surface sterilization regime, as recommended by Hambleton and Currah (1997). For each surface sterilization treatment, roots were then cut into 1-3 mm pieces using sterile forceps and a scalpel, and 8-10 fragments were placed on two different isolation media 100 ug/mL tetracycline: Oatmeal Agar (OA) and Modified Melin-Norkrans (MMN). The resulting four isolation plates from each plant were stored in a dark climate chamber for 30 days at 14°C. Hyphae emerging from the root fragments were then aseptically transferred to OA or MMN to generate pure cultures of individual fungal species. Subcultures were incubated for 60 days at 14°C in a dark climate chamber. Those subcultures that successfully grew were first grouped by locality and then further grouped into morphotypes within each locality based on colony morphology (margin, texture, colony color, secreted pigments etc.). We avoided creating morphotypes spanning localities and refrained from grouping cultures with any ambiguities. We sampled material for both DNA extraction and permanent storage at -80 degrees in 20% glycerol (from fungal material from each of the cultures using a clean sterile pipette tip. The voucher material for permanent storage is kept within the facilities at the University Museum of Bergen.

The fungal material was crushed in CTAB buffer for 2 minutes at 30 Hz using a Qiagen Retch tissue lyser. The DNA was then extracted using a CTAB approach (Murray and Thompson, 1980) with modifications (Gardes and Bruns, 1993). DNA isolates were cleaned using a column cleanup kit (VWR pregold Cyclepure) and sent to Barcode of Life initiative (BOLD) (Ratnasingham and Herbert, 2007) for sequencing the internal transcribed spacer region (ITS), which is the chosen barcoding marker for fungi (Schoch et al., 2012). The sequences were generated by using the primer pair ITS1-F and ITS4 (White et al., 1990; Gardes and Bruns, 1993) and were sequenced bidirectionally. All sequences are stored within the BOLD system database (Ratnasingham and Hebert, 2013) including a reference picture, which are freely available. The sequences can be accessed by searching by either the project code “FINCH”, or individually by sequence ID ( Table S1 ). We then annotated our sequences using the UNITE database as a reference (Nilsson et al., 2018; Kõljalg et al., 2020), and all sequences identified to a unique species hypothesis in the UNITE database (using a 95% identity threshold as described by U’ren et al., 2009) were treated as belonging to a single OTU ( Table S1 ). Sequences were assigned membership to functional guilds using the FunGuildR package for R based on the FUNguild database (Nguyen et al., 2016) and each sequence’s best BLAST match to the UNITE database with percent identity >95%.

All statistical tests and figures were generated in the R statistical environment (R Core Team, 2018). The iNEXT package (Hsieh et al., 2016) was used to generate alpha diversity estimates for i) the site level based on number of cultures and number of plants sampled and ii) the host level based on both number of cultures and number of sites sampled ( Table 1 ). In addition, gamma diversity was estimated for the coastal heathland landscape as a whole, based on both number of cultures and number of sites. Sorensen’s beta diversity was calculated pairwise between each of the sampled root systems using the betapart package (Baselga and Orme, 2012), and subsequently partitioned into its nestedness and turnover components. For each of these three measures, we examined i) the distance decay relationship with the geographic distance between sites and ii) the differences between the categories: between hosts-within sites, between hosts-between sites, and within host - between sites. Distance decay relationships were tested using linear models, as the data were normally distributed according to Shapiro-Wilkes tests, while between host and between site differences were not-normally distributed and assessed using a Kruskal-Wallis test with a post-hoc Dunn test to identify differences between groups. Patterns in beta diversity were also explored in an NMDS ordination of those root systems for which >7 cultures were generated. The ordination structure was calculated using the metaMDS function in the vegan package (Oksanen et al., 2020) on a Bray-Curtis dissimilarity matrix with two dimensions and a maximum of 500 random starts.

The observed dominance of Ascomycota within the coastal heathland system is consistent with previous investigations of the heathland mycoflora (Anderson et al., 2003; Bougoure et al., 2007). We frequently encountered Phialocephala species within the root systems and this genus is a commonly isolated dark septate endophyte of ericaceous hosts (Grünig et al., 2008; Jumpponen and Trappe, 2008), as are the Cryptosporiopsis anamorphs of Pezicula (Sigler et al., 2005; Yang et al., 2018). Phialocephala fortinii was frequently recovered across hosts and sites and this species is known to increase host above- and below-ground biomass, as well as host N- and P-content (Newsham, 2011). Given the prevalence of this species complex and its impacts on host growth and reproduction, P. fortinii is likely a vital species in terms of host persistence and function in coastal heathlands. Several well-known ERM species (Perotto et al., 1996; Monreal et al., 2000; Smith and Read, 2008) were also isolated, including members of the Hyaloscyphaceae, Meliniomyces, Oidiodendron and Rhyzoscyphus. Mycena was the most common basidiomycete, and this genus has recently been demonstrated to have functional lability with the capacity to colonize living plant roots (Thoen et al., 2020). Notably, we did not recover Sebacinales within the host mycoflora. These taxa are frequent, culturable associates of ericaceous plant roots (Selosse et al., 2007; Vohnik et al., 2016; Weiß et al., 2016) and have been reported as “ERM like” (Selosse et al., 2007; Vohnik et al., 2016). The taxonomic entities within our study are mostly known from Norway in only sporadic reports both in GBIF and NBIC, however, they are frequently reported in culture- and molecular-based surveys of soil and root fungal biodiversity (Vrålstad et al., 2002a; Vrålstad et al., 2002b; Blaalid et al., 2012; Blaalid et al., 2013). This would suggest that the biodiversity in the ericaceous root-associated fungal community is underreported, rather than novel for science.

Estimates of the total ericaceous root-associated fungal diversity in Norwegian coastal heathlands are as high as 231 OTUs. However, it must be noted that due to methodological bias, this estimate is unlikely to reflect the true total fungal diversity associated with ericaceous roots in this habitat. Numerous studies have documented a taxonomically diverse array of fungi that are not readily cultured and rarely or never detected in culture-based studies (Arnold et al., 2007; Higgins et al., 2011; Høyer and Hodkinson, 2021). Metabarcoding based culture-independent studies of ericaceous hosts (Vaccinium calycinum: Leopold et al., 2021, Cassiope tetragona: Lorberau et al., 2017) report more than 600 OTUs per host, indicating that the true root-associated fungal diversity in Norwegian coastal heathlands is likely much higher than what is reported here.

The ericoid root-associated communities in coastal heathlands exhibited both low nestedness and high rates of turnover between sites. This high degree of between-site heterogeneity is consistent with findings from other inventories of heathland fungal communities (de la Pena et al., 2012; Macia-Vicente and Popa, 2022). As observed by Macia-Vicente and Popa (2022), there was no discernible distance decay relationship in community composition, and the Norwegian heathland communities each had a distinct mycofloral “fingerprint” that likely reflects local stochastic processes that create a high degree of endemism within ericaceous root fungi. Coastal heathland habitats are formed across centuries of anthropogenic activity (Hjelle et al., 2010; Kaaland, 2012), including prescribed burning, turf cutting, and grazing, all of which shape the habitat’s vegetation (Vandvik et al., 2005; Måren and Vandvik, 2009; Velle and Vandvik, 2014; Bargmann et al., 2015). The unique land use history of each coastal heathland likely shapes the characteristic fungal community found in each location as well (Hazard et al., 2014). For example, grazing intensity and grazer identity can impact both soil fungal community diversity and biomass (van der Heyde et al., 2017; Eldridge and Delgado-Baquerizo, 2018; Xun et al., 2018; Dudinszky et al., 2019). However, the direct effects on plant associated mycoflora of the low to moderate grazing pressure that most coastal heathlands experience remains largely unknown. However, studies of ectomycorrhizal fungi suggest that factors other than grazing may be more influential in shaping the plant-associated mycoflora (Dudinszky et al., 2019; Tervonen et al., 2019). For example, coastal heathlands are subjected to periodic burning, and fire is known to affect microbial community composition (Hewitt et al., 2016). Within coastal heathlands, fire disturbance has historically occurred on a 20–30-year cycle (Kaaland, 2012) and these fires reduce nitrogen and phosphorus levels in heathland soils (Green et al., 2013), which in turn is expected to affect fungal community dynamics. Still, little is known about how the mycoflora is influenced by such short cyclic fire events. Studies from natural pyrophilic ecosystems have emphasized the predictable effect of fire on fungal community dynamics (Hopkins et al., 2021), and frequent wildfires are known to promote fire specialists (Hansen et al., 2019). However, coastal heathlands are relatively “young fire landscapes” and it remains to be elucidated the impact of this frequent fire cycle on the mycoflora. The current study (n=8) lacks the needed statistical power to infer the specific roles of land use, including grazing and burning, on shaping fungal communities. Nevertheless, we argue that the highly heterogenous fungal communities we observe in Norway’s coastal heathlands are likely a product of the unique historic land use practices at each locality (Hazard et al., 2014; Tervonen et al., 2019).

Despite identifying a unique mycoflora at each coastal heathland site, we failed to detect signs of host-specificity among the ericoid root-associated fungi surveyed here. The plants investigated hosted similar numbers of fungal species irrespective of host identity, and many of the fungal species were found in association with multiple ericaceous hosts. Previous comparisons of culture-dependent and culture-independent methods for characterizing ericaceous root-associated fungal communities indicate that while the two different methods may recover different species, they detect the same biological patterns (Walker et al., 2011). This suggests the absence of host specificity we observe here is not simply a product of bias towards only those fungi in the root systems that are culturable. Little community divergence between ERM host taxa has been demonstrated in some systems (Kjoller et al., 2010; Walker et al., 2011), but distinct communities partitioned by host identity have been recovered in others (Bougoure et al., 2007; Ishida and Nordin, 2010). We argue that the high degree of versatility in life strategies exhibited by ERM fungi (Perotto et al., 2018) combined with resource partitioning theory (Bruns, 1995) are factors explaining the high diversity and heterogeneity of fungal communities both between hosts and between sites in coastal heathlands.

Fungal conservation is typically under-prioritized (Heilmann-Clausen et al., 2015; Cao et al., 2021; Gonçalves et al., 2021), and few frameworks exist for the management and monitoring of not only individual fungal species, but fungal communities as a whole. Management authorities frequently use alpha and gamma diversity metrics to identify and prioritize areas for conservation, focusing on biodiversity and species richness both at the local and landscape levels. However, there is an increasing body of work that suggests beta diversity metrics can also be useful parameters to assess drivers of community assembly (Anderson et al., 2011; Chiantore et al., 2018) and function as a tool in conservation (Bush et al., 2016; Socolar et al., 2016). In order for fungi to be included in holistic management strategies that focus on total biodiversity, we therefore require well developed protocols for assessing fungal biodiversity that generate robust alpha and beta diversity estimates. Here, standardized monitoring programs will yield important data for capturing such diversity metrics Haase et al., 2018). Despite generating more than 100 cultures for many of the sites examined here, species accumulation curves did not approach saturation for the individual localities nor the total gamma diversity of the coastal heathland landscape. Extrapolated alpha diversity estimates suggest that on a locality level, further recovery of biodiversity would most efficiently be made by increasing the number of plants sampled per locality rather than increasing the number of cultures generated per root system. On a landscape level, gamma diversity estimates would be improved by increasing the number of sites sampled, rather than by increasing the number of plants sampled. Based on these results, we recommend that when assessing biodiversity in coastal heathlands, sampling strategies should be designed to maximize the number of hosts sampled at individual sites in order to provide robust alpha diversity estimates, and to maximize the number of localities sampled for providing the best possible estimate of gamma diversity at the landscape level. This study provides base knowledge for how sampling strategies can be designed to provide the reliable biodiversity estimates that are needed to allow the incorporation of fungal biodiversity into conservation strategies.

Throughout much of Europe, coastal heathlands are already conserved areas owing to their cultural and historical importance. We argue that protection of these areas for their societal value has the knock-on effect of preserving the unique mycofloral biodiversity present in each coastal heathland. We observe high heterogeneity and substantial species turnover between individual coastal heathlands that was unrelated to geographic distance and speculate this may reflect the unique land-use history at each locality. Thus, protecting individual coastal heathland areas indirectly conserves a unique suite of fungi at each site. Such “In-situ” habitat conservation approaches have been suggested for fungi already in the 1990s (Hawksworth, 1991), and coastal heathlands are a prime example of how this concept can function. Notably, focus on heathland restoration has increased in recent years (de la Pena et al., 2012; van der Bij et al., 2018; Radujković et al., 2020). Such restoration will have knock-on effects of conserving such frequently overlooked organismal groups within these habitats, including fungi. Moreover, the success of coastal heathland restoration, as indicated by vegetation establishment, is greatly improved by reinoculation of the belowground community which effectively functions as a conservation measure for fungal biodiversity (van der Bij et al., 2018). However, restoration and conservation of coastal heathlands requires follow-up with a combination of low-intensity measures such as mowing, grazing, burning and regular shrub clearance to maintain the coastal heathland landscape (Blindow et al., 2017; Walmsley et al., 2021), all of which are expected to benefit those species adapted to a semi-natural system, including fungi.

Plant-soil feedback loops have been established as drivers of plant diversity, abundance, and succession (Bardgett and van der Putten, 2014) and heathland restoration experiments have underpinned the importance of plant–fungal interactions to achieve restoration success (de la Pena et al., 2012; van der Bij et al., 2018; Radujković et al., 2020). We argue that in ecosystems like coastal heathlands where fungi play significant ecological roles with effects on multiple trophic levels, holistic conservation and restoration approaches that include action plans for fungi would be beneficial.

A broad diversity of root-associated fungi was isolated from ericaceous hosts in coastal heathlands, representing primarily Ascomycete endophyte, saprophyte, and ERM fungi. We observed little host specificity among the fungi isolated, as has previously been observed in these communities. The fungal communities at different coastal heathland localities were highly heterogeneous, exhibiting low nestedness and high species turnover, with each locality bearing a unique mycofloral ‘fingerprint’ which we hypothesize may be linked to the unique land-use history at each site, although further research is needed to confirm this. Based on our own attempts to quantify the alpha and gamma diversity in Norwegian coastal heathlands, we recommend that in order to provide robust estimates of alpha and gamma diversity that can be used for management purposes, sampling strategies should aim to maximize the number of plant hosts investigated to improve alpha diversity estimates, and to maximize the number of sites sampled to improve gamma diversity estimates. Finally, existing and future conservation and restoration efforts targeting European coastal heathlands are expected to have knock-on effects, facilitating conservation of fungal biodiversity.