*2.4. Phenotypic Assessment*

In order to test the hypothesis that phylogenetically defined taxa in *Cora* are phenotypically undifferentiated and may reflect taxonomic inflation, we analyzed 87 formally described and sequenced species of *Cora* [33], plus two outgroup taxa in the genus *Corella* (Table S2). We established a one-sequence-per-taxon ITS tree, selecting the type sequence for each species when available and using the alignment from Lücking et al. [33] as a template (Table S2). After deleting the non-relevant sequences and keeping only the 89 target sequences (File S10), the alignment was subjected to RAxML 8.2.0 [58] analysis on a local CPU to reconstruct the best-scoring maximum likelihood tree under the universal GTRGAMMA model; bootstrapping was performed with 1000 pseudo-replicates.

For the 87 selected *Cora* species (and the two *Corella*), we assembled a matrix for a total of 20 characters, including one ecological character (preferred substrate; row 1), 11 morphological, anatomical, and chemical characters (rows 2–12, based on largely on previous studies [32,33,40], onward simply referred as "phenotypical characters"), and eight chorological characters, defined as biogeographic regions (rows 13–20; Table 1; scores on Table S3). We then defined phenotypes by combining the character scores for the 11 phenotypical characters (columns 2–12 in Table S3) into a character state string for each species. Phenotypes were considered identical if they agreed in the complete string between two species. The number of phenotype characters was kept intentionally low and limited to those that were constant within a lineage to avoid false positives in terms of different phenotypes (the number of potential phenotypes increases exponentially with the number of characters). Our aim was to test if even a very low number of characters would result in a sizable number of distinct phenotypes. A high level of cryptic speciation (or potential taxonomic inflation) could be expected if the number of distinct phenotypes resulting from the combination of these 11 characters was substantially lower than the number of lineages distinguished (87).

For each possible pair of species, we computed molecular phylogenetic distance using the Kimura 2-parameter model implemented in BioEdit 7 through DNADIST 3.5 [68,69]. In parallel, we computed phenotypic distance by calculating total character state difference over all 11 phenotypical characters (columns 2–12 in Table S3).

To assess potentially cryptic speciation, we defined four types of crypticity (Table 2). To avoid confusion with the use of the terms "cryptic" and "crypsis" in zoology, here we propose to add the prefix "phylo-" before cryptic when referring to a phylogeny-based phenetic context. We defined "same" phenotype as total phenotype character distance = 0 (identical) and "similar" phenotype as total phenotype character distance = 1 (low). Taxa differing in more than one character or with a total character distance >1 were considered distinct.

**Table 1.** Ecological (1), phenotypical (11), and chorological (8) characters and states scored for the selected 87 species in the genus *Cora* (plus two *Corella*). Only the 11 phenotypical characters (Size through Bleeding pigment) were utilized for generating the character strings.


a Soil means growing directly on bare soil, while ground means growing between terrestrial vegetation, e.g., over grasses. b Sutures refer to the lines apparent between two lobes, typically appearing as if the lobes have been sown together—*short*: only present along a small part of the lobes (see Figures 3I,K and 4G in [33]); *long*: present along most of the lobes (see Figures 5F and 7D,N in [33]). c Color when fresh or rewetted, not dried specimen. d Most samples are glabrous, but those with trichomes can be further divided—*felty* (rare, see Figure 8K,L in [33]): with short hairs formed by single hyphae; *setose* (most common type, see Figures 3L, 5O and 7K,L in [33]): with distinct hairs composed of at least partly agglutinate hyphae; *strigose* (few species, see Figures 4D,E and 10E,F in [33]): with conspicuous, long trichomes always composed of agglutinate hyphae.

In addition to the maximum likelihood phylogenetic tree based on the ITS barcoding locus, we also computed a cladogram based on all 20 characters using PAUP 4.0b10 [75,76]. To test the hypothesis that tree structure based on phenotype and distribution was significantly different from random (i.e., there was correlated structure in the ecological, phenotypical, and chorological data), we computed 100 random trees in PAUP based on the taxon set to simulate random distribution of character states relative to tree topology. For each tree, we calculated the following five indices based on phenotype character state distribution: parsimony tree length (TL; in steps), consistency index (CI), retention index (RI), rescaled retention index (RC), and homoplasy index (HI). The first and last indices (TL, HI) are proportional to noise in the phenotype data and inversely proportional to structure, whereas the other indices (CI, RI, RC) behave the opposite way.


**Table 2.** Definition of different types of crypticity within an evolutionary context and the corresponding terms used in other studies [34].
