Alpine lichen diversity in an isolated sky island in the Colorado Plateau, USA—Insight from an integrative biodiversity inventory

Abstract Lichens are major components of high altitude/latitude ecosystems. However, accurately characterizing their biodiversity is challenging because these regions and habitats are often underexplored, there are numerous poorly known taxonomic groups, and morphological variation in extreme environments can yield conflicting interpretations. Using an iterative taxonomic approach based on over 800 specimens and incorporating both traditional morphology‐based identifications and information from the standard fungal DNA barcoding marker, we compiled a voucher‐based inventory of biodiversity of lichen‐forming fungi in a geographically limited and vulnerable alpine community in an isolated sky island in the Colorado Plateau, USA—the La Sal Mountains. We used the newly proposed Assemble Species by Automatic Partitioning (ASAP) approach to empirically delimit candidate species‐level lineages from family‐level multiple sequence alignments. Specimens comprising DNA‐based candidate species were evaluated using traditional taxonomically diagnostic phenotypic characters to identify specimens to integrative species hypotheses and link these, where possible, to currently described species. Despite the limited alpine habitat (ca. 3,250 ha), we document the most diverse alpine lichen community known to date from the southern Rocky Mountains, with up to 240 candidate species/species‐level lineages of lichen‐forming fungi. 139 species were inferred using integrative taxonomy, plus an additional 52 candidate species within 29 different putative species complexes. Over 68% of sequences could not be assigned to species‐level rank with statistical confidence, corroborating the limited utility of current sequence repositories for species‐level DNA barcoding of lichen‐forming fungi. By integrating vouchered specimens, DNA sequence data, and photographic documentation, we provide an important baseline of lichen‐forming fungal diversity for the limited alpine habitat in the Colorado Plateau. These data provide an important resource for subsequent research in the ecology and evolution of lichens alpine habitats, including DNA barcodes for most putative species/species‐level lineages occurring in the La Sal Mountains, and vouchered collections representing any potentially undescribed species that can be used for future taxonomic studies.

Integrating various lines of evidence in biodiversity inventories, for example, traditional taxonomic approaches with DNA sequence data, can provide improved perspectives into biodiversity surveys (Cao et al., 2016;Sheth et al., 2017). A phylogenetically informed reinterpretation of morphological characters can strengthen taxonomic conclusions and provide direction for nominal taxa in need of revision to more accurately portray evolutionary histories (Hutsemékers et al., 2012). In many groups, presently described species represent only a fraction of the estimated overall diversity (Guiry, 2012;Hawksworth & Lücking, 2017;Pons et al., 2006). Given the limited taxonomic expertise for many groups and the meticulous nature of classical monographic research, integrative approaches can help remedy the current "taxonomic impediment" problem (Dayrat, 2005;Vinarski, 2020) by facilitating the discovery and taxonomic description of novel taxa (Cao et al., 2016).
Alpine, arctic, and Antarctic habitats worldwide support specialized biological communities that are adapted to harsh environmental conditions (Billings, 1974;Chapin & Körner, 1994), and important components of alpine-adapted communities are often poorly known (Pereira et al., 2012). In these ecosystems, abiotic factors, especially climate, dominate biotic interactions, make them particularly vulnerable to changing climate (Cannone et al., 2007). Detecting changes in occurrence, distribution, or abundance of alpine species is based on knowledge of which species occur in specific locations, information that is conspicuously absent for many organismal groups.
In the intermountain region of western North America, expansions and contractions of species' ranges have proceeded through local movements along elevation gradients to and from scattered high-elevation patches of habitat throughout the Pleistocene (Guralnick, 2007;Jiménez-Moreno & Anderson, 2013) Here, isolated alpine habitats are geographically subdivided among different mountain ranges, for example, "sky islands," harboring unique biodiversity due to a variety of factors spanning multiple spatial and temporal scales (Knowles, 2000;Marx et al., 2017). Climate-driven distributional shifts lead to complex patterns of diversification and demography in alpine specialists (Galbreath et al., 2009). In the alpine zone of the Southern Rocky Mountains, vascular plants have been systematically inventoried over the past three decades, demonstrating rich plant communities harboring substantial endemism, ca. 10% (Fowler et al., 2014). However, only a limited number of studies investigating lichen diversity in the same region are available (Table 1).
In many alpine habitats, lichen communities are diverse and found in high relative abundance (Bruun et al., 2006;Imshaug, 1957). Alpine lichens, including those occurring on rock, soil, and/or alpine turf, play important ecological roles, ranging from nutrient cycling to habitat and food sources to soil stabilization (Asplund & Wardle, 2017).
Despite the high diversity and abundance of lichens in alpine ecosystems, many of these ecosystems are sensitive to environmental disturbances, including climatic shifts and changes in land management strategies (Cornelissen et al., 2001;Geml et al., 2010;Kranner et al., 2008;St. Clair et al., 2007). Therefore, monitoring alpine lichen communities can provide crucial insight into the biological impacts of climate change in some of the most vulnerable ecosystems.
However, the diversity and distributions of many components of alpine lichen communities have been incompletely characterized, emphasizing the need to efficiently generate crucial baseline assessments. Recent, collaborative taxonomic efforts have further highlighted the incredible lichen diversity in high altitude/latitude Sal Mountains, and vouchered collections representing any potentially undescribed species that can be used for future taxonomic studies.

K E Y W O R D S
ASAP, DNA barcoding, La Sal Mountains, Rocky Mountains, vouchered collections ecosystems (McCune et al., 2020;Nimis et al., 2018;Spribille et al., 2010Spribille et al., , 2020. These studies also reveal that a significant proportion of this diversity has not yet received formal taxonomic recognition.
For example, recent lichen diversity inventories in two Alaskan national parks revealed that ca. 10% of the sampled lichens could not be assigned to a known species (Spribille et al., 2010(Spribille et al., , 2020.
Effective strategies for using molecular sequence data to aid in the identification of fungi continue to be developed (Abarenkov et al., 2018;Lücking et al., 2020). Coupling these strategies with phenotype-based data will likely facilitate more effective cataloging the global fungal diversity and provide novel insight into ecological and evolutionary processes (Sattler et al., 2007;Struck et al., 2018). Dispersal of alpine and arctic lichens has received considerable attention in recent years. Frequent long-distance dispersal has been documented for a number of alpine and arctic lichens (Fernández-Mendoza & Printzen, 2013;Garrido-Benavent et al., 2021;Geml et al., 2010;Onuţ-Brännström et al., 2017) and a "mountain hopping" mechanism of dispersal explains, in part, the broad distribution of many alpine lichens (Garrido-Benavent & Pérez-Ortega, 2017). Therefore, the Rocky Mountains in North America play a fundamental role in understanding the processes that influence the distribution of alpine lichen communities (Garrido-Benavent & Pérez-Ortega, 2017;Hale et al., 2019;Weber, 2003). To investigate how this might look in practice, here we attempt to characterize lichen-forming fungal species diversity in the alpine zone of an isolated sky island in the Southern Rocky Mountains using an integrative taxonomic approach-the La Sal Mountains (hereafter the "La Sals") in the Colorado Plateau in eastern Utah. This insular range is surrounded by semiarid, lowelevation, canyon dissected terrain (Figure 1

| Study site, sampling, and morphology-based identifications
The La Sals comprise three distinct groups-the "North," "Middle," and "South" groups (File S1), which were formed in the Laramide orogeny during the Oligocene when intrusive, dioritic magmas uplifted overlying sedimentary rocks (Hunt & Waters, 1958;Nelson et al., 1992). Our sampling focused exclusively on habitats above

| DNA extraction, amplification, and sequencing
We attempted to generate molecular sequence data for the my- (e and f) distinct soil-dwelling lichen communities Beads (GE Healthcare, Pittsburgh, PA, United States), with cycling parameters following a 66-56°C touchdown reaction (Lindblom & Ekman, 2006

| DNA-based inference of mycobiont candidate species and phylogenetic reconstructions
All sequences generated for the study were subjected to an initial "blastn" search against GenBank's nucleotide collection (Altschul et al., 1990) to confirm family-level membership. Exploratory multiple sequence alignments (MSAs) of all ITS sequences generated for this study resulted in unreliable alignments due to the high variability of the ITS region at deeper fungal taxonomic levels. Therefore, all subsequent MSAs generated here were performed at the mycobiont family-level. Family-level MSAs were generated using the program MAFFT v7 (Katoh & Toh, 2008;Rozewicki et al., 2017). We implemented the G-INS-i alignment algorithm and "1PAM/K = 2" scoring matrix, with an offset value of 0.1, the "unalignlevel" = 0.4, and the remaining parameters were set to default values.
To circumscribe candidate mycobiont species from the familylevel ITS MSAs, we used Assemble Species by Automatic Partitioning (ASAP; Puillandre et al.). ASAP is a recently developed method that circumscribes species partitions using an implementation of a hierarchal clustering algorithm based on pairwise genetic distances from single-locus sequence alignments (Puillandre et al., 2021). The pairwise genetic distances are used to build a list of partitions ranked by a score, which is computed using the probabilities of groups to be panmictic species. ASAP, therefore, provides an objective approach to circumscribe relevant species hypotheses as a first step in the process of integrative taxonomy.
Each family-level ITS MSA was analyzed under a maximum likelihood (ML) criterion as implemented in IQ-TREE v2 (Nguyen et al., 2014), with 1,000 ultrafast bootstrap replicates (Hoang et al., 2017), with the best-fitting substitution model for the entire ITS region selected using ModelFinder (Kalyaanamoorthy et al., 2017).
Trees were visualized using FigTree v1.4.4 (Rambaut, 2008). Species partitions inferred from the family-level ASAP analyses were compared to phylogenetic reconstructions to determine reasonable DNA-based candidate species (DNA-CS) using the criterion of reciprocal monophyly in DNA-CSs, in addition to qualitative assessments of branch lengths and lichen morphology (see below).

| Integrative taxonomy
Incorporating phenotypic data in the assessment of DNA-CSs through integrative taxonomy provides critical information for evaluating species-level diversity in lichen-forming fungi (Lücking et al., 2020). Phenotypic traits of all vouchered specimens were examined in light of the DNA-CSs delimited using ASAP and the phylogenetic inferences. Diagnostic features from relevant taxonomic keys and monographs, as well as a variety of online resources, were considered. As needed, thin-layer chromatography (Culberson, 1969;Orange et al., 2001) was used to aid with specimen identifications.
To characterize how DNA-CSs compared with phenotypically circumscribed species and currently available sequence data in GenBank, each DNA-CS was categorized within the following categories: "match"-ITS sequences ≥98% similar to sequences from the same taxon currently available in GenBank; "species complex"-ITS sequences ≥97% similar to morphologically similar species but represented by multiple candidate species; "affinity"-sequences representing candidate species not recovered as monophyletic, represented by multiple candidate species; "mismatch"-ITS sequences <95% similar to sequences from the same taxon currently available in GenBank; and "no comparison"-sequences from identified species were not available in GenBank. We note that currently available ITS sequence data in GenBank represents only a small portion of extant fungal species, and many of the sequences are incorrectly identified to species level (Nilsson et al., 2006). Furthermore, pairwise similarity-based approaches, such as BLAST, can provide misleading perspectives into taxonomic assignment and relationships (Lücking et al., 2020). Hence, we used the Protax-fungi pipeline for taxonomic placement using ITS sequences, as implemented on the PlutoF platform and using the UNITE database (Abarenkov et al., 2010(Abarenkov et al., , 2018. Protax-fungi provides statistical assessment of taxonomic assignment precision, from species to phylum ranks, here choosing a plausible classification value of 0.05.

| Integrative taxonomy
Our integrative specimen identification approach (phenotype + sequence data) resulted in a total of 189 species (Table 2). In many cases, morphologically similar specimens were recovered in divergent, wellsupported phylogenetic lineages; and from an integrative perspective, these were combined into a single species (e.g., taxa in Candelariaceae and Megasporaceae; File S3). In other cases, traditionally accepted species known to display considerable morphological variable were recovered in divergent, well-supported phylogenetic lineages, and these were also combined into a single "integrative" species.
Over half (54%) of all candidate species inferred in this study were ≥98% similar to sequences representing the same taxa and presently available on GenBank (searched 15 December 2020). Over a third of all candidate species appear to belong to species complexes of morphologically similar taxa with at least 97% genetic similarity, representing 29 putative species complexes (Appendix S1; File S3). In contrast, nearly a third (31%) of all candidate species had no match on GenBank (sequences from identified species not presently available in GenBank) or their sequences from identified species were highly dissimilar from sequences from the same taxon on GenBank. Of the newly generated sequences that were <95% similar to sequences presently available in public databases, most belonged to members of Lecanoraceae, Megasporaceae, Psoraceae, Rhizocarpaceae, and Verrucariaceae (Table S1). Approximately 5% of new sequences were <90% similar sequences presently available on GenBank. In the Protax-fungi analysis, over 68% of sequences could not be assigned to species-level rank with statistical confidence, and nearly 8% of sequences were not assigned any taxonomic rank (File S4).

| D ISCUSS I ON
Despite the limited alpine habitat on the insular La Sals in the Colorado Plateau, USA, we document the most diverse alpine lichen communities known to date from the southern Rocky Mountains (Table 1), with at least 189 documented species of lichen-forming fungi (Appendix S1). The actual number of species in the alpine habitat in the La Sals is likely higher. Our integrative data (DNA sequence data + phenotype-based inference) suggest that a substantial number of nominal lichens occurring in the La Sals are comprised of multiple candidate species-level lineages (DNA-CS; Appendix S1; File S3). Including DNA-CS within these species complexes leads to a greater than 25% increase in species-level diversity, with at least 52 additional candidate species (Appendix S1). Furthermore, additional surveys, including alternative sampling strategies, would likely result in additional species not documented here (e.g., Vondrák et al., 2016). A preliminary checklist, with accompanying taxonomic notes, is reported in companion paper (Leavitt et al., in preparation), and below, we discuss the implications of our integrative inventory approach in better understanding alpine lichen diversity.  (Lücking et al., 2020;McCune et al., 2020;Spribille et al., 2020).
Similarly, DNA barcoding approaches are confounded by inherent limitations with the standard fungal DNA barcoding marker, the ITS, coupled with the lack of comprehensive DNA reference libraries for effective taxonomic assignment (Lücking et al., 2020;Nilsson et al., 2019). While the standard barcode marker for fungi can diagnose distinct species-level lineages in many cases (Schoch et al., 2012), variation in rDNA can in some species complexes provides biased perspectives, potentially over-splitting natural specieslevel groups. Intraspecific and intragenomic variation in this repeat region is not well known in fungi (Lofgren et al., 2019). In some lichen-forming fungi, for example, the Rhizoplaca melanophthalma aggregate, minimal intragenomic variation was observed (Bradshaw Note: The first column lists the family and number of sampled thalli represented by ITS sequence data in parentheses; the second column reports the number of species (average number of sequences/species) inferred using the newly proposed Assemble Species by Automatic Partitioning (ASAP) approach to empirically delimit candidate species-level lineages from family-level multiple sequence alignments; the third column reports the number of candidate species based on combining information from the ASAP partitions and phylogenetic reconstructions; and the fourth column reports integrative species, combining information from the DNA-based candidate species with morphological data.
TA B L E 2 Summary of lichen-forming fungal species diversity collected in the La Sal Mountains in eastern Utah, USA et al., 2020) and the ITS region successfully diagnoses the majority of species in this complex (Leavitt et al., 2013). However, based on genomic data from members of the R. melanophthalma aggregate, some highly divergent ITS sequences inferred as distinct candidate species in single-locus species delimitation analyses likely belong to a single species (Bradshaw et al., 2020;Keuler et al., 2020). Inferences from single-locus species delimitation analyses, such as those performed in this study, are inherently limited (Fujita et al., 2012), and most DNA-based species hypotheses will likely need to be investigated on a case-by-case basis.
Limitations with currently available reference libraries for sequence comparison were manifested in the low proportion successful taxonomic assignment of DNA-CS at the species level (File S4), with over 68% of sequences that could not be assigned to specieslevel rank with statistical confidence. Confounding the poor success in DNA-based taxonomic assignment, nearly a third of all morphologically identified species were not represented by sequence data in public repositories, and in other cases, sequence data from unidentifiable alpine lichens did not closely match to currently available sequence data (Appendix S1). The results support the perspective that substantially increasing the number of sequences based on verified material will be essential for efficient DNA barcode identification (Lücking et al., 2020;Nilsson et al., 2019). In general, for biodiversity surveys where comprehensive taxonomic treatments are impractical, best practices for reporting uncertain identifications are not well established. However, with recent improvements in how unclassifiable species hypotheses are handled in the UNITE database, these "dark" taxa can now be integrated with the taxonomic backbone of the Global Biodiversity Information Facility and an unlimited number of parallel taxonomic classification systems are supported (Nilsson et al., 2019). Without high-quality sequence databases that are thoroughly curated by taxonomists and systematists, integrative biodiversity inventories of lichen-forming fungi will remain laborand cost-intensive (Begerow et al., 2010).
The results of this study expand novel sequence data into publicly available repositories, providing the first ITS sequences for many species and candidate species-level lineages. Furthermore, we link these sequence data to digital imagery from vouchered specimens (File S1), in addition to the physical vouchered collections (Appendix S1). We documented unexpected ITS sequence variation in multiple nominal species occurring in the La Sals (Appendix S1; File S3), and the species complexes inferred here provide important direction for identifying lineages that require additional research.
By providing publicly available ITS sequence data for the majority of specimens collected for this study, our results can be easily integrated in future research using the formal barcoding marker for fungi (Schoch et al., 2012). Rather than relying exclusively on phenotypebased identifications that may be biased in comparison with other studies (Brunialti et al., 2012(Brunialti et al., , 2019Giordani et al., 2009), these ITS data can be directly integrated into a wide range of future studies using the standard DNA barcoding marker for fungi. Our molecularbased approach for initial species delimitation using ASAP provided only an initial perspective into diversity, providing important direction for future taxonomic research. With ongoing taxonomic revisions of lichen-forming fungi, including the description of new taxa, the sequence data generated here will facilitate more accurate reassignment of specimens from the La Sals to the appropriate taxonomic group.
While the high levels of diversity documented in this study are striking, we predict that other alpine habits in the southern Rocky Mountains may have similar levels of diversity. Small crustose lichens on rocks and soil are common in most alpine regions but are easily overlooked in vegetation surveys (Ahti & Oksanen, 1990). Even when recognized, these crustose lichens may not be documented because of difficulties with identification due to a lack of diagnostic features or environmental modifications in extreme habitats (Kappen, 1973;McMullin et al., 2020). Here, we aimed to overcome these challenges by integrating vouchered specimens (permanently The opportunistic sampling approach implemented in this study was intended to capture the broad range of species diversity in alpine habitat on the La Sals, rather than provide quantitative insight into distribution patterns or site-specific species richness (McCune & Lesica, 1992;McMullin et al., 2010;Newmaster et al., 2005).
Although different habitats throughout the alpine zone in the La Sals, for example, talus slopes, alpine turf, and late snowmelt areas, support distinct lichen communities, we cannot make robust comparisons among these different communities given the limitations with present sampling. Future ecological sampling approaches will be essential to characterizing factors influencing lichen community assembly and monitoring disturbances and ecological changes. For example, mountain goats (Oreamnos americanus) were released in 2013 in the La Sal Mountains with notable, site-specific impact on alpine communities, including lichens . The long-term impact of the large ungulates on alpine communities will require long-term monitoring. Coupling environmental sampling approaches with the DNA reference sequences provided here will facilitate more efficient sampling strategies for DNA-based monitoring of ecological changes in alpine lichen communities in the La Sals.
Both amplicon-based and whole-genome shotgun metagenomic approaches using environmental samples have been shown to capture higher levels of diversity than traditional inventory strategies (Keepers et al., 2019;Wright et al., 2019), the trade-offs among cost, speed, accuracy, and precision of metagenomic approaches must be carefully considered (Lücking et al., 2020).  (Imshaug, 1957), alpine-specific macrolichens, for example, cetrarioid species, Thamnolia subuliformis (Ehrh.) W. L. Culb., etc., were notably absent from alpine habitats in the La Sals, except for a single observation of Evernia divaricata (L.) Ach. Our hope is that results from this study will provide further impetus to explore questions relating to the origin and stability of alpine lichen communities.

ACK N OWLED G M ENTS
We acknowledge support from Canyonlands Natural History

CO N FLI C T O F I NTE R E S T
The authors declare no conflicting interests.