*3.6. Phenotype Assessment*

Total phylogenetic distance between possible pairs of species ranged between 0 and 120 substitutions, and total phenotype distance between 0 and 16 steps or character state differences (Figure 9). There was no discernible correlation between total phylogenetic and phenotype distance, i.e., phenotype distance did not depend on genetic distance. Mean total phenotype distance over all pairs of species was 7.18 (±2.01); if only pairs of species with a phylogenetic distance of 10 or less were considered, mean total phenotype distance was only marginally different (7.29 ± 2.11). Additionally, if only pairs of species representing sister clades were considered, no difference was detected in mean total phenotype distance (7.23 ± 3.19). Thus, on average, more closely related species were not more phenotypically similar to each other than more distantly related species.

**Figure 9.** Distance between possible pairs of species for ITS (0–120 bp substitutions), and phenotype (0–16 characters steps). Blue dots on the left side of the graph correspond to closely related species pairs only, orange dots to all other species pair comparisons.

Using only the 11 morphological, anatomical, and chemical characters (columns 2–12 in Table S3), we differentiated 82 distinct phenotypes for the 87 included species of *Cora* (excluding *Corella*). Three phenotypes were shared by two lineages each, and one phenotype was shared between three lineages (Figure 10). There was only one instance of two lineages or species that could be considered eu-(phylo-)cryptic, namely the related *C. squamiformis* and *C. terricoleslia* (Figure 10); both are terrestrial species known from the central Andes (Bolivia). However, the two are not sister species and so it remains unclear whether the congruen<sup>t</sup> phenotype is the result of symplesiomorphy (cryptic speciation) or homoplasy. There were two instances of kapo-(phylo-)cryptic ("near-cryptic") sister species (Figure 10), namely *C. minor* and *C. paraminor* (both known from Costa Rica), and *C. hirsuta* and *C. schizophylloides* (both known from Colombia). In both cases, the two sister species are phylogenetically distinct, with four positional ITS differences in *C. minor* vs. *C. paraminor* and 11 positional ITS differences in *C. hirsuta* vs. *C. schizophylloides* (File S10), and the phylogenetic separation is supported by a single character in each case (i.e., minor differences in the phenotype): short vs. no sutures in *C. minor* vs. *C. paraminor* (a rather subtle feature), and a strigose vs. setose lobe surface in *C. hirsuta* vs. *C. schizophylloides*; the latter two also differing in substrate preference (terrestrial vs. epiphytic). Furthermore, there was one case of two closely related, allo-(phylo-)cryptic (i.e., "semi-cryptic") species, namely *C. applanata* vs. *C. reticulifera*: whereas the first is known from the Andes (documented by 32 sequenced samples), the second appears to be restricted to (south-)eastern Brazil and Uruguay (documented by 66 sequenced samples; Figures 10 and S1, Table S4); both exhibit 16 positional ITS differences (File S10) and are, hence, clearly separated phylogenetically, but do not show any discernable phenotypical differences with respect to the tested characters. Finally, there were two instances of pseudo-(phylo-)cryptic species (i.e., the same phenotype having evolved in homoplasy in distantly related lineages), one encompassing three species (*C. campestris* vs. *C. caliginosa* vs. *C. terrestris* in blue) and the other one two species (*C. cuzcoensis* vs. *C. davibogotana* in green; Figure 10). Since only eu-, kapo-, and allo-(phylo-)cryptic lineages could be considered as potential "false positives" (i.e., taxa

interpreted as species that may in fact represent infraspecific lineages), the potential error of overestimation of the species richness estimate in this dataset, by using phylogeny as sole evidence (i.e., potential taxonomic overinflation), would only be four species, out of 87 (i.e., less than 5%).

**Figure 10.** Clade structure (87 *Cora* species plus two *Corella* outgroup) highlighting types of crypticity observed throughout the tree. The inset map highlights the allopatric distribution of *C. applanata* (northern Andes) vs. *C. reticulifera* (southeastern Brazil).

Our analysis of the TL, CI, RI, RC, and HI indices derived from phenotypic characters, comparing random, DNA-based, and phenotype-based trees, showed that trees based on phenotype alone were substantially more structured in terms of their phenotypic characters than DNA-based trees (Figure 11). DNA-based trees were also always closer to random trees than to phenotype-based trees in their phenotypic signal. This suggests a relatively high level of homoplasy for individual character states and a low level of character state intercorrelation. On the other hand, the mean pairwise character distance of over seven steps indicates that species generally have distinct morphologies, supporting their phylogenetic delimitation. Overall, closely related species usually look distinct and have different substrate ecologies, whereas distantly related species are more likely to appear similar.

**Figure 11.** Analyses showing TL, CI, RI, RC, and HI indices derived from phenotypic characters comparing random (Random), DNA-based (DNA), and phenotype-based (Phenotype) trees, assuming that DNA-based trees represent the underlying "true" phylogeny. For DNA-based trees, the corresponding index is always between that of random and phenotype-based trees, but in all cases closer to the random trees, indicating that phenotype structure correlates little with deeper nodes and mostly with terminal lineages.

Although there was a strong correlation between phylogenetically defined clades interpreted as species and their associated phenotypic characters, phenotype structure became generally diffuse at deeper nodes, as evidenced by the lack of correlation between phylogenetic and phenotypic distance. However, limited phylogenetic structure was still found for some characters at higher clade levels. Three small clades uniformly contained species with specific phenotype character states that may represent synapomorphies for these clades: one clade containing all species with a bleeding pigment formed upon rewetting previously dried collections (*C. rubrosanguinea* clade); one clade with species forming adnate, rounded, confluent hymenophores different from the concentrically shaped hymenophores in most other species (*C. garagoa* clade); and one clade with species colonizing naked soil with completely flattened, adnate thalli (*C. reticulifera-applanata* clade; Figure 10).
