Albinaria candida (Pfeiffer, 1850)

Relationships between species of the A. candida View in CoL group and the A. cretensis group from western Crete

An ML tree based on concatenated sequences of 2050 loci shows that neither the species of the A. candida group nor the species of the A. cretensis group from western Crete form monophyletic units ( Fig. 3 View Figure 3 ). The deepest split separated the populations from Gramvousa Peninsula and neighbouring sites (populations 1–4) plus A. xanthostoma from Rodopos Peninsula (populations 5–7) from all other species of the A. candida group and the A. cretensis group. In the populations from Imeri Gramvousa Island ( Fig. 1B View Figure 1 ; Fig. 2 View Figure 2 : population 1), the western part of Gramvousa Peninsula ( Fig. 1C, D View Figure 1 ; Fig. 2 View Figure 2 : population 2), and the adjacent part of the west coast ( Fig. 1F View Figure 1 ; Fig. 2 View Figure 2 : population 4), anterior palatal folds are frequently developed; therefore, they were included in the A. candida group as A. loosjesi , whereas the population from the east side of Gramvousa Peninsula ( Fig. 1E View Figure 1 : population 3) lacks anterior palatal folds and was included in A. tenuicostata ( A. cretensis group) by Nordsieck (1999, 2004, 2017). However, the A. tenuicostata populations from central and southern western Crete ( Fig. 1G View Figure 1 ; Fig. 2 View Figure 2 : populations 30–33) form a distinct clade ( Fig. 3 View Figure 3 ). The form from eastern Gramvousa Peninsula was originally described as Clausilia grabusana Boettger, 1883 . We will use the name Albinaria grabusana in the following for the species from Gramvousa Peninsula and neighbouring sites because this name has priority over A. loosjesi Nordsieck, 1977 .

Albinaria tenuicostata + Albinaria species 1 (in the sense of Bamberger et al. 2022) represent the sister group of all species of the A. candida group and the A. cretensis group from western Crete except A. xanthostoma and A. grabusana ( Fig. 3 View Figure 3 ). The remaining taxa of the A. candida group from western Crete, A. candida and A. amalthea , form a monophyletic group with Albinaria byzantina (Charpentier, 1852) of the A. cretensis group. Albinaria amalthea , usually characterized by anterior palatal folds, is the sister species of A. byzantina without anterior palatal folds. A population from the Mavros Gorge near Orthouni without anterior palatal folds, which was classified as A. byzantina by Bamberger et al. (2022), turned out to belong to A. amalthea . Albinaria candida , which lives syntopically with A. byzantina at some sites, is the sister group of these two species ( Fig. 3 View Figure 3 ). Albinaria sublamellosa ( Boettger, 1883) represents the sister group of all remaining species of the A. cretensis group. Albinaria eburnea (Pfeiffer, 1854) , which vicariates with A. sublamellosa in the Samaria Gorge, is the sister group of the remaining species. Albinaria cretensis from Akrotiri Peninsula on the north coast of Crete is the sister group of Albinaria sphakiota (Maltzan, 1887) from the south-west coast of Crete. The remaining populations form a complex including Albinaria troglodytes (Schmidt, 1868) , Albinaria virginea (Pfeiffer, 1846) , and some populations from westernmost Crete, which were previously also included in A. virginea but are deeply separated from this species, representing a distinct cryptic species referred to as Albinaria species 2 by Bamberger et al. (2022).

The Neighbor-Net based on an SNP data set shows that the western Cretan Albinaria species represent an almost star-like radiation, in which even the more distantly related species are able to hybridize ( Fig. 4 View Figure 4 ). It visualizes, for example, a conflict in the data with regard to the relationships between A. eburnea , A. byzantina , and A. amalthea . The ML tree based on the concatenated sequence data set shows the common root of A. eburnea spratiana and A. eburnea samariae ( Fig. 3 View Figure 3 ). In contrast, A. eburnea spratiana from the north side of the Lefka Mountains has tighter connections with A. byzantina in the Neighbor-Net ( Fig. 4 View Figure 4 ), with which it shows admixture ( Fig. 5 View Figure 5 ), whereas A. eburnea samariae from the south slope of the Lefka Mountains has tighter connections with A. sublamellosa , with which it hybridizes in the Samaria Gorge ( Fig. 5 View Figure 5 ).

Population genetic structure

ADMIXTURE analyses of the species previously included in the A. candida group and some of the species of the A. cretensis group potentially involved in hybridization with species of the A. candida group were based on one random SNP each from 864 loci. Cross-validation error was lowest between K = 9 and K = 15, with an absolute total minimum at K = 14 (Supporting Information, Fig. S2 View Figure 2 ). Results between K = 9 and K = 15 are presented in Figure 5 View Figure 5 .

The ancestral populations delimited with K = 9 correspond to A. grabusana (populations 1–4), A. xanthostoma (populations 5–7), A. amalthea (populations 8–13), A. candida (populations 14–17), A. byzantina (populations 18–20), A. eburnea (populations 21–23), A. sublamellosa (western populations 24–26), A. sublamellosa (eastern populations 27–29), and A. tenuicostata (populations 30–33). In the following, we summarize which clusters are split with increasing K. We do not mention which clusters fuse again at some values of K. With K = 10, the southernmost population of A. grabusana ( A. loosjesi sigridae Nordsieck, 2017 ) was separated from the other populations of A. grabusana . With K = 11, the populations of A. amalthea were divided into two clusters corresponding to A. amalthea unipalatalis Nordsieck, 2017 (populations 8 and 9) and A. amalthea bipalatalis (Martens, 1878) (populations 12 and 13), while a population of A. a. amalthea (Westerlund, 1878) (population 11) and a population without anterior palatal folds (population 10) are a mixture of both clusters. With K = 12, the population of A. grabusana from eastern Gramvousa Peninsula (population 3) was separated from the populations from western Gramvousa Peninsula (populations 1 and 2), and A. sublamellosa schultesi Wiese, 1988 from Agia Roumeli (population 26) was separated from the A. sublamellosa populations from the Samaria Gorge (populations 24 and 25). With K = 13, A. eburnea spratiana Nordsieck, 2017 (population 21) was separated from A. eburnea samariae Nordsieck, 2004 (populations 22 and 23), and the northernmost population of A. tenuicostata from Mili (population 30) was separated from the southern population of A. tenuicostata (populations 31–33). With K = 14, the populations of A. candida from Akrotiri Peninsula (populations 14 and 15) were separated from the A. candida population from Drapanos Peninsula and the probably introduced population from near Sougia at the south coast (populations 16 and 17), and one of the A. byzantina populations is separated from the others. With K = 15, the populations of A. a. amalthea and A. amalthea ssp. (populations 10 and 11) became a separate cluster, and the populations of A. sublamellosa from the Samaria Gorge (populations 24 and 25) were separated.

Notable admixture (population average> 10%; Fig. 5 View Figure 5 ) between taxa recognized as species was found in some populations. Albinaria amalthea population 13 shows admixture from its sister species, A. byzantina , for K = 9, 10, and 12 (average 31% for K = 12), but not for values of K at which A. amalthea is split into several clusters. For K = 12, A. amalthea population 12 also shows admixture from A. byzantina (average 11%). Albinaria eburnea population 21 shows admixture from A. byzantina for K = 9–11 and 15 (average 18% for K = 9). For K = 14, there is even a separate cluster including A. byzantina population 20 and 49% of the genome of A. eburnea population 21. Albinaria eburnea population 23 shows admixture from A. sublamellosa for K = 12 and 15 (average 22% for K = 12). Most of the genome of A. sublamellosa populations 24 and 25 is supposedly derived from A. eburnea (average 76%), but only for K = 14 (for other values of K, only a much smaller part of the genome of population 25 is derived from A. eburnea ).

Species validation by isolation-by-distance tests

Isolation-by-distance tests ( Fig. 6 View Figure 6 ; Table 2 View Table 2 ) showed that the differentiation between A. grabusana (including A. loosjesi ) and A. tenuicostata ( Fig. 6A View Figure 6 ) and the differentiation between A. grabusana and A. xanthostoma ( Fig. 6B View Figure 6 ) and A. amalthea ( Fig. 6C View Figure 6 ) cannot be explained by IBD. We therefore consider A. grabusana to be a separate species. Likewise, the differentiation between A. amalthea and A. xanthostoma ( Fig. 6D View Figure 6 ) and between A. amalthea and A. candida ( Fig. 6E View Figure 6 ) cannot be explained by IBD, and we also consider them to be distinct species. Because the western populations of A. amalthea (populations 8 and 9), which were classified as a distinct subspecies, A. amalthea unipalatalis , by Nordsieck (2017), were classified as a distinct cluster in the ADMIXTURE analyses with K = 11, 13, and 14 ( Fig. 5 View Figure 5 ), we tested whether their differentiation from the other A. amalthea populations (A. a. amalthea , A. amalthea ssp., and A. amalthea bipalatalis ) can be explained by IBD ( Fig. 6F View Figure 6 ). Because this hypothesis could not be rejected, we consider the subgroups of A. amalthea identified in the ADMIXTURE analysis ( Fig. 5 View Figure 5 ) to be conspecific. Because there is admixture between A. byzantina and A. amalthea (in population 12, average 11% for K = 12; Fig. 5 View Figure 5 ), we tested whether their differentiation can be explained by IBD ( Fig. 6G View Figure 6 ). This hypothesis has been rejected, corroborating the species status of the two taxa. Because there is also admixture between A. byzantina and A. eburnea (in A. eburnea population 21, average 18% for K = 9; Fig. 6 View Figure 6 ), we tested whether their differentiation can be explained by IBD ( Fig. 6H View Figure 6 ). The results were ambiguous. The differentiation between the two taxa cannot be explained by IBD within A. byzantina but can be explained by IBD within A. eburnea . There is also admixture between A. eburnea and A. sublamellosa ( A. sublamellosa populations 24 and 25 would even be classified as A. eburnea with K = 14, but not with other values of K; Fig. 5 View Figure 5 ). However, their differentiation cannot be explained by IBD ( Fig. 6I View Figure 6 ) if all A. eburnea and A. sublamellosa populations are considered, meaning that the species status of the two taxa is corroborated.

Test for introgression

The data set for ABBA-BABA tests included taxa that showed admixture ( Fig. 5 View Figure 5 ) and data conflict in the Neighbor-Net ( Fig. 4 View Figure 4 ), namely A. amalthea , A. byzantina , A. eburnea , and A. sublamellosa . Tests were based on 50 209 SNPs (Supporting Information, File S1) of populations of these species. Test results are summarized in Table 3 View Table 3 .

Test 1, based on the topology ( A. corrugata ( A. byzantina ( A. eburnea spratiana , A. eburnea samariae ))), indicated a significant introgression between A. byzantina and A. eburnea spratiana at the north side of the Lefka Mountains ( Table 3 View Table 3 ).

Test 2a, based on the topology ( A. corrugata ( A. sublamellosa ( A. eburnea samariae , eburnea spratiana ))), indicated significant introgression between A. sublamellosa and A. eburnea samariae in the lower part of the Samaria Gorge ( Table 3 View Table 3 ). If population 22 of A. eburnea samariae from the upper end of the Samaria Gorge, which is far from A. sublamellosa populations and does not show admixture ( Fig. 5 View Figure 5 ), was added (test 2b), the test for introgression remained significant ( Table 3 View Table 3 ).

Considering the varying proportions of admixture between the different A. amalthea populations with A. byzantina ( Fig. 5 View Figure 5 ), we conducted three tests for introgression ( Table 3 View Table 3 ). Test 3a indicated significant introgression between the easternmost population of A. amalthea (population 13), which showed the highest proportion of admixture ( Fig. 5 View Figure 5 ), and A. byzantina . If population 12, which showed much less admixture ( Fig. 5 View Figure 5 ), was added, only a moderately significant signal of introgression remained (test 3b). The test was not significant if all A. a. amalthea + A. amalthea ssp. + A. a. bipalatalis populations (populations 10–13) were included (test 3c).