Morphological identification

The greenhouse specimens are up to 3.3 mm high with up to 5.5 whorls, the whorls are well rounded with a moderately impressed suture. The umbilicus is small in juveniles to nearly fully grown adults, but significantly opens up in fully developed specimens. A sharp and slightly elevated parietal lamella is present throughout growth, and there is never a columellar lamella; the columella may be slightly curved (Fig. 1).

Open in a separate window

Figure 1

Tornatellides cf. boeningi, Gothenburg Botanical Garden.

(A) Fully grown specimen, 3.3 mm height, GNM 308512, (B) nearly fully grown specimen, 3.0 mm height, SMNS-ZI0144090, (C) subadult specimen, 2.1 mm height, GNM 308493; scale bar = 1 mm (Photo credit: Ira Richling).

Identification of species of Tornatellides Pilsbry, 1910, especially of unknown origin, is difficult, because of the absence of a modern revision and the rather subtle conchological differences. The detailed anatomical studies for the whole family Achatinellidae by Cooke & Kondo (1961) primarily focused on classification and differentiation above species level. Nonetheless, they dissected ten different species of Tornatellides from five different “groups” (see below) and did not indicate any anatomical features enabling species identification. Furthermore, preliminary molecular studies so far presented only at conferences (e.g. Yeung et al., 2018) on Hawaiian representatives—the Hawaiian Islands being one major centre of diversification—even suggest a higher diversity and limitations in the conchological discrimination of the species. Given this situation, and although more than 100 years old, the revision of Pilsbry & Cooke (1914–1916) in the Manual of Conchology seems still the most reliable and comprehensive basic source of systematic information combined with the updates published by Cooke & Kondo (1961, especially p. 247).

Following the identification keys by Pilsbry & Cooke, we concluded that the greenhouse species could either belong to the “simplex-group” or possibly the “perkinsi-group”, the latter, however, being ruled out by the defining presence of clearly developed columellar lamellae in juveniles which our species is missing. The “simplex-group” of Pilsbry & Cooke (1914–1916: 195–215) includes representatives as geographically diverse as from the islands of Japan, Taiwan, Polynesia, New Zealand, Galapagos and Hawaii summing to 21 species and two subspecies. With corrections and additions made or summarized by Cooke & Kondo (1961) the now called “oblongus-group” (T. simplex (Pease, 1865) is regarded as a synonym of T. oblongus (Anton, 1839)) contains 22 species including 12 Hawaiian representatives.

Of those potential candidates the following seven non-Hawaiian species were excluded for the reasons given in parentheses: T. inexpectatus (Pilsbry, 1901) (species status questioned by Pilsbry & Cooke (1914–1916) and subsequently treated as a subspecies of T. boeningi, e.g. Takara, 1954), T. tryoni Pilsbry & Cooke, 1915 (different shape, less strongly developed parietal lamella), T. perforatus (Liardet, 1876) (regarded as Lamellidea pusilla (Gould, 1847) by Cooke & Kondo, 1961), T. subperforatus (Suter, 1909) (fewer whorls, more rapidly expanding, columellar lamella developed in juveniles), T. inconspicuus (Brazier, 1872) (regarded as belonging to Tornatellinops Pilsbry & Cooke, 1915 by Cooke & Kondo, 1961 and conchologically different: “columella twisted and entering spirally” as stated in the orginal description by Brazier, 1872 and confirmed with material from Lord Howe Island by Iredale, 1844: 308, pl. 18, fig. 7; currently synonymised with Tornatellinops jacksonensis (Cox, 1864) by Stanisic et al., 2018), T. clarionensis Dall, 1926 (p. 485, pl. 35, fig. 9: unusually large, whorls less convex, parietal lamella hardly developed), T. mexicana Dall, 1926 (p. 484–485, pl. 35, fig. 8: clearly distinct by two columellar lamellae), leaving T. boeningi, T. oblongus (Anton, 1839) and T. chathamensis (Dall, 1892) as possible identifications.

Among the Hawaiian species, the key in Pilsbry & Cooke (1914–1916: 202–203) leads to either T. procerulus (Ancey, 1904) or T. kilauea Pilsbry & Cooke, 1915 (our specimens’ diameter is about 58–59% of the height). However, it may be seriously questioned whether the first option (diameter less than 50% of the length) leading to T. konaensis Cooke & Pilsbry, 1915/T. kahoolavensis Cooke & Pilsbry, 1915 is correct because measurements of the respective figures lead to ratios of 55% or 58% (as for procerulus). In any way, T. konaensis and T. kahoolavensis are ruled out by size or shape of the umbilicus. T. procerulus (different shape of the base, higher aperture and higher ultimate whorl) and T. kilauea (too broad and rapidly expanding) also do not match our species. In addition, all Hawaiian species appear to have a straighter outline or less convex whorls (see also figures in Severns (2011)).

From the three remaining non-Hawaiian candidates, the Galapagos species T. chathamensis was illustrated with its syntype and recent material by Miquel & Herrera (2014: 110–112, figs. 5–7). Considering size (taken from the image, 3.1 mm high) and number of whorls the syntype seems fully grown. Despite a certain similarity, compared to our larger specimen with the same number of whorls, it has a smaller umbilicus and mainly lacks the specific widening of the last whorl (compare Figs. 1A and ​and1B),1B), so we regard it as different. In addition, Cooke & Kondo (1961) describe “one or two moderately to well-developed columellar folds in juvenile specimens” for material from Galapagos in the Bishop Museum collection, providing another argument against this identification.

A comparison with T. oblongus is complicated by the wide distribution of the species ranging throughout almost all suitable southeast Pacific islands from the Society Islands in the west, reaching Pitcairn in the southeast and up to the Marquesas in the north, going along with a considerable variation in shell shape and the development of the parietal lamella (Cooke & Kondo, 1961). As a result, we regard it impossible to distinguish our species morphologically from T. oblongus.

T. boeningi was described from Taiwan and Yen (1939) reported additional specimens from Hongkong, and with T. [b.] inexpectatus regarded as subspecies or even synonym it is known also from the adjacent Ryukyu Islands and the Izu Islands south of Honshu. Furthermore, it occurs on the Ogasawara Islands as reported by Wada, Kawakami & Chiba (2012) who observed its survival of the passage through a bird’s digestive system thus suggesting a bird-mediated dispersal mode. Conchological comparison to images of the lectotype (Yen, 1939: fig. 26 and Fig. 2) and a Japanese specimen figured by Sasaki (2008: fig. 15) shows great resemblance to our species; the original description gives dimensions of 3.25 mm height and 1.88 mm diameter which falls well within the range of the material from the greenhouse.

Open in a separate window

Figure 2

Tornatellina boeningi Schmacker & Boettger, 1891 (now Tornatellides boeningi), lectotype, SMF 42552, 3.3 mm height, Tamsui, Taiwan; scale bar = 1 mm (Photo credit: Sigrid Hof).

Given the choice with no clear conchological preference to either of the two candidate species T. oblongus and T. boeningi, we tentatively identify our material as T. cf. boeningi accounting for probabilities: An origin in southeast Asia seems much more likely than in very remote Oceania.

Regarding the type material of T. boeningi there is some confusion that should be clarified. Our conclusion is that Yen (1939) correctly designated the lectotype from the original material and that the type locality is Tamsui: While the species was explicitly described from material originating from northern Taiwan [Formosa] (Tamsui) (other than most material dealt with by Schmacker & Boettger (1891) in the same paper which came from Takao, southern Taiwan), Yen (1939) gave as type locality “Takao, Formosa”, both under the species account as well as in the figure caption of the lectotype. A request at the SMF confirmed that the original label by Boettger states “Takao, S. Formosa”, but Zilch as curator at the time added a label with “N-Formosa: Tamsui”. The SMF material originates from collection O. Boettger, ex Schmacker, so the other potential error that Yen selected from non-type material is ruled out. The copy of Yen in the SMF library is annotated by Zilch stating that Boettger’s label is surely erroneous because the species was only mentioned from Tamsui. Thus the lectotype selection by Yen (1939) was based on the correct, but obviously erroneously labelled material—not unlikely given the overall assemblage of material.

Molecular identification

Being aware of the remaining uncertainty in the conchological identification of the Tornatellides species we tested molecular identification as well.

For two specimens we obtained a partial COI sequence of 658 base pairs (bp), both representing the same haplotype. For 16S we generated a single partial sequence of 420 bp (Table 1).

Searches in GenBank taxonomy and by the BLAST^® tool resulted in a total of six partial COI-sequences (two Tornatellides, three Elasmias and one Tornatellaria species) leading to an alignment of 654 bp and 122 partial 16S sequences (one Hawaiian Tornatellides and otherwise 37 different haplotypes of T. boeningi) aligned to a 417 bp dataset including gaps. For geographic origin and GenBank-accession-numbers see Table 1.

This extremely limited number of available sequences, however including one “target” species, partially with missing species identifications, did not allow meaningful further analyses other than calculating pairwise p-distances as a first test of identity.

For neither COI nor 16S sequences did the comparison result in a match. Minimum p-distance in COI was 5.4% to an unidentified Tornatellides species from Molokai, but even an Elasmias species from Maui only showed 5.7% difference (Table 2). Minimum p-distance for 16S was expectedly lower (i.e. sequences more similar), but surprisingly the difference to the unidentified Hawaiian Islands representative was lower than to our candidate taxon from the conchological analyses when all its haplotypes were treated as a group (Table 3). However, the 37 Tornatellides boeningi haplotypes showed a considerable within-group divergence of 1.1% and in a pair-wise comparison of all different haplotypes, two Hahajima-haplotypes showed less divergence (1.68% and 1.69% respectively) than the Hawaiian specimen and other Hahajima-haplotypes.

The interpretation of the “Tornatellides boeningi” sequences from Hahajima is further complicated by the fact, that for the Ogasawara Islands, Pilsbry & Cooke (1914–1916) described the endemic species T. tryoni (see also above, conchologically ruled out). So far only Wada & Chiba (2011)—based on unpublished COI and 16S sequence data—speculate that “Tornatellides tryoni of Ogasawara is phylogenetically closest to Tornatellides boeningi, a seashore species of Japan and Ryukyu Islands, and they are most likely identical species”. The identification as “boeningi” for the published sequences (Wada, Kawakami & Chiba, 2012) is obviously based on that side remark in Wada & Chiba (2011). Since the type locality of T. boeningi is located on northern Taiwan, the potential synonymy of the two species still remains to be proven.

In conclusion, available sequence data neither enable a meaningful identification nor even help to narrow down a potential source area. This is mainly due to the nearly complete absence of accessible comparative data so urgently needed especially in the identification of introduced species.