Phylogenetic relationships of Kimberleytrachianext section
Our phylogeny (Figs 2-3) supports the basal position of a clade including K. aequum Köhler, 2011, K. nelsonensis n. sp. and K. serrata n. sp. These species exhibit a rather flat shell (H/D<0.52) with periostracal projections on the entire surface (Köhler, 2011b: fig. 218; Figs 4C-D, 6D-F, 7A-C in Appendix). Although most Kimberleytrachia species exhibit a moderately elevated shell (H/D>0.52), a flat shell is characteristic for Torresitrachia and all other closely related genera (except Succochlea n. gen. and Rhagada Albers, 1860 [in Martens and Albers, 1860]). Therefore, a flat shell probably represents an ancestral character state in Kimberleytrachia. The penial anatomy of K. aequum and K. nelsonensis n. sp. is assumed to be largely plesiomorphic in possessing all typical features of the genus, such as a well-developed pad-like structure and regular, well-spaced and smooth epiphallic and penial longitudinal pilasters and transverse lamellae (Köhler, 2011b: fig. 220). The inner penial wall sculpture of most other species appears to be characterised by increased complexity (sinuous lamellae as in K. jacksonensis n. sp., presence of pustulation as in K. deflecta, fusion of elements such as in K. somniator Köhler, 2011) or by the loss of features (K. alphacentauri Köhler, 2011).
Kimberleytrachia and Succochlea n. gen. have strikingly similar genital anatomies including a corresponding layout of penis and epiphallus. Despite the remarkable complexity, our phylogeny suggests that these anatomical configurations likely are the result of parallel evolution.
In regard to the delimitation of species, the present study confirms conclusions of previous revisions of north-western Australian camaenid gastropods, such as Exiligada Iredale, 1939 (Criscione et al., 2012), Australocosmica Köhler, 2011 (Köhler, 2011a; Criscione and Köhler, 2013c), Mesodontrachia Solem, 1985 and related camaenids from the Northern Territory (Criscione and Köhler, 2013a), Nanotrachia Köhler and Criscione, 2013, Setobaudinia (Criscione and Köhler, 2013b), Retroterra, Baudinella and Molema Köhler, 2011 (Criscione and Köhler, 2014a) that camaenid species are most reliably identified by a combination of morphological (shell, genital anatomy) and molecular evidence.
By evaluating the combined differentiation in shell, penial morphology and mitochondrial sequences, seventeen morphologically and phylogenetically distinct clusters have been differentiated. These represent ten already known and seven yet undescribed species of Kimberleytrachia (Figs 2-3). One additional species is placed within the new genus Succochlea gen. n. for its morphological and phylogenetic distinctiveness from all other species. Refer below for complete taxonomic descriptions. All these species are well-differentiated by means of their shell and/or genital anatomy.
Two species, K. umbonis (Solem, 1979) and K. leopardus n. sp., revealed considerable intraspecific genetic structuring. Each species includes two distinct mtDNA clades (Figs 2-3). In both cases, the genetic divergence is associated with small geographic distances between populations and morphological homogeneity. It is therefore not deemed to justify formal taxonomic acts.
Generally, the amounts of genetic differentiation between species were comparable with interspecific distances observed in other north-western Australian camaenids, such as >3% (COI and 16S) in Exiligada >4% (16S, Criscione et al., 2012), Rhagada >4% (16S) and >7% (COI) (Johnson et al., 2012), Setobaudinia >6% (COI) and >9% (16S) (Criscione and Köhler, 2013b), Amplirhagada Iredale, 1933 >8% (COI) and >10% (16S) (Köhler and Johnson, 2012), Nanotrachia >5% (COI) and >3% (16S) (Köhler and Criscione, 2013), Baudinella, >3% (COI) and >4% (16S) Retroterra (Criscione and Köhler, 2014a). In Kimberleytrachia there has been a slight overlap between intra- and interspecific distances as was occasionally also observed in other genera.
Shells are of limited value for the discrimination of Kimberleytrachia species confirming a more general statement by Solem (1981a) on the diagnostic value of shells in camaenids from the Kimberley. The shells of most species are within the same size range of 18 to 25 mm in diameter, except K. alphacentauri and K. deflecta having a significantly smaller shell (Köhler, 2011: Table 11). Most shells of Kimberleytrachia species share a subglobose shape but some are discoidal (K. aequum; Köhler, 2011b: fig. 218; K. serrata Fig. 4D) or exhibit a very low spire (K. chartacea Köhler, 2011; Köhler, 2011b: fig. 208, K. leopardus, Fig. 4B and K. nelsonensis, Fig. 4C).
The amount of differentiation in the genital characteristics is frequently not sufficient to satisfactorily delimitate species: The general configuration of the inner penial wall (with proximal lamellae and pad and more than one distal pilaster) is shared by most species, except K. alphacentauri having a distinctive sculpture with a single pilaster and lamellae extending distally and almost reaching the gonopore (Köhler, 2011b: fig. 217). There is a species-specific variation in number and relative dimension of sculptural elements, but the divergence is too subtle to be useful as a reliable taxonomic marker. This observation conflicts with findings of several other studies on NW Australian camaenid genera, where the genital anatomy was recognised as the most informative and convenient source of information for identifying species (Solem, 1979, 1981a, 1981b, 1985, 1988, 1997; Köhler, 2010a, 2010b, 2011b, 2011c, Criscione and Köhler, 2013a). However, a similarly conserved genital anatomy has been observed in Torresitrachia (Solem, 1979, 1985; Köhler, 2011b).
Kimberleytrachia species were generally found to be allopatric. Only on Boongaree Island three species were found to occur in sympatry (marked with ‘S’ in Fig. 1): K. aequum, K. alphacentauri and K. canopi Köhler, 2011. These species differ markedly in shell size and shape as well as in their genital anatomy – in particular the sculpture of their inner epiphallic and penial walls (Köhler, 2011b). They are not closely related and their sister species are not found on Boongaree Island. Hence, they must have originated from independent colonisers of this island.
For Australian camaenids reinforcement has been hypothesized to explain patterns where morphological divergence in sympatric species has been more pronounced than usual. In Amplirhagada, differences in the genital anatomy between sympatric species are often particularly conspicuous and many of the species with the most highly derived penial anatomy consistently occur in sympatry with at least a second congener (Köhler, 2011b). Correspondingly, the distinct morphology of these three species from Bongaree Island may be indicative of ecological niche partitioning (Chiba, 1999, 2002; Solem, 1981a; Cameron, 1992), enabling them to persist in sympatry.
Patterns of diversification and distribution
Over many millions of years, the monsoon climate typical of northern Australia has carved the ancient Kimberley sandstones into a deeply dissected landscape with barren plateaus and deep gorges. With its soils being depauperate in nutrients and having a low capacity for storing water, the Kimberley boasts a varied patchwork of vegetation types. Vine thickets, a form of rainforest, thrive only in sheltered places while more exposed areas are covered with grassland, bushland or savannah.
The patchily distributed vine thickets are the preferred habitat for many camaenid groups for their moister and more buffered microclimate. One of the factors believed to contribute to the high diversity of camaenid land snails in the Kimberley is the multiplicity of isolated vine thicket/rainforest patches across the region (Solem, 1991). Created by climate fluctuations throughout the Quaternary and Mid to Late Tertiary (Bowler, 1982), these patches have essentially been acting as habitat refugia whereby gene flow between them has been restricted by intervening unsuitable habitat. Because camaenid land snails are deemed to be rather poor dispersers, the patchy distribution of rainforest species has resulted in allopatric patterns of speciation and endemism throughout the region (Cameron, 1992). This evolutionary scenario has specifically been proposed to explain the multitude of narrowly endemic species of Amplirhagada found throughout the Kimberley (Solem, 1981; Köhler, 2011b).
The basal splits in the topology of mtDNA trees of Kimberleytrachia are shallow (Figs 2-3). This may be indicative of rather simultaneous lineage differentiation throughout the entire range of the genus. The actual evolutionary rates of Kimberleytrachia are unknown. If accelerated evolutionary rates of up to 10% per million years like those reported from several pulmonates (Thomaz et al., 1996; Chiba, 1999; Thacker and Hadfield, 2000; Watanabe and Chiba, 2001; Pinceel et al., 2005) were considered, then the average genetic divergence observed between species of Kimberleytrachia could indicate that the initial diversification within Kimberleytrachia may have occurred about 1 Million years ago (Ma). A more moderate evolutionary rate, however, would proportionally place this event further back in time. Consequently, the initial diversification within Kimberleytrachia can cautiously be postulated to have occurred sometime around the Pliocene-Pleistocene boundary, between roughly 1.5 and 3.0 Ma. This period of time was characterized by severe aridification across all major Australian biomes (McLaren and Wallace, 2010) leaving the Kimberley rainforests particularly patchy. Such dynamic landscape changes have generally had a profound effect on the evolution of the Australian biota (e.g. Pepper et al., 2011), including camaenid land snails.
Kimberleytrachia is closely associated with rainforests and vine thickets and species of this genus were exclusively found in vine thickets. Most species occur in the wettest parts of the Kimberley, the Prince Regent Reserve and adjacent areas, which currently receive 1,200 mm or more of average annual rainfall and which have a current average rainforest cover of more than 5% of surface area (for maps of rainfall and rainforest cover see Kimber et al., 1991). The coastal regions with a more extensive rainforest cover boast the largest number of species while only three species, K. crawfordi (Solem, 1979), K. setosa and K. umbonis, occur further inland and further south in regions that receive at least 800 mm rainfall and have between 1 and 5% rainforest cover altogether. For this intimate connection of snails with rainforest habitats, it appears plausible to postulate that the fragmentation of rainforests during the Plio-Pleistocene may have played a significant role in initiating the diversification of distinct lineages.
Available distributional data suggests that species in the wetter parts of the Kimberley mainland have comparatively smaller distributional ranges than the species in the drier parts. This observation seems counter-intuitive because one would expect that denser rainforest cover and wetter climate facilitate dispersal between habitat islands. Accordingly, species distributions should be larger in wetter than in drier parts where rainforest patches are more scarcely distributed. However, in correspondence with the patterns in Kimberleytrachia, it has been demonstrated for other camaenids in north-western Australia that both the species richness per area and the species turnover between adjacent areas are decreasing with increased aridity (Solem and McKenzie, 1991; Köhler, 2011b; Gibson and Köhler, 2012). Drier areas support fewer species but these species usually have wider distributions. This phenomenon has been explained with the ecological function of rock habitat as litho-refugia. Couper and Hoskin (2008) firstly suggested for lizards that sheltered rock habitats provide environmental conditions similar to rainforests permitting the persistence of rainforest lineages even in areas where rainforests have disappeared. Such litho-refugia have clearly been important for the persistence for several camaenid lineages in drier parts of the Kimberley and even have facilitated their radiation into semi-arid environments, such as on the limestone outcrops of the Victoria Bonaparte bioregion of the East Kimberley (Criscione et al., 2012; Köhler and Criscione, 2013). Accordingly, the survival of K. crawfordi, K. setosa and K. umbonis in inland areas may not exclusively rely on the presence of rainforests. If these species were not as strictly associated with rainforests as their counterparts in wetter areas, then the patchiness of rainforests would be less of an obstacle to their dispersal throughout the landscape; hence their wider distributions. However, the environmental envelope of Kimberleytrachia does apparently not permit their survival under more xeric conditions typical of the northern, southern and interior parts of Kimberley, regions that receive less than 800 mm of average annual rainfall.
The three species occurring outside of the >1,200 mm annual rainfall area (K. crafordi, K. umbonis, K. setosa) are member of the same clade together with a fourth species (K. leopardus). The diversification within this lineage might have been triggered by adaptation to more xeric environments.
Several Kimberleytrachia species are endemic to offshore islands. However, the phylogenetic data suggests that they have differentiated roughly at the same time as the mainland species. During the Pleistocene, the islands inhabited by Kimberleytrachia were repeatedly connected with the mainland due to fluctuating sea levels. Although creating a potential for dispersal between islands and with the mainland, this phenomenon did not influence the patterns of distribution and lineage differentiation on the islands. Similar conclusions were drawn for species of Amplirhagada inhabiting the Kimberley islands (Johnson et al., 2010; Köhler and Johnson, 2012).