Millipedes (Diplopoda) play a fundamental role in terrestrial ecosystems as detritivores and soil forming organisms (Golovatch & Kime, 2009; Joly et al., 2020), and are often adapted to certain microclimates, and biotic and abiotic factors (David, 2015). For example, the hygrophilous colobognathan millipedes, which contain the millipede species with the highest number of legs (Marek et al. 2021), are often found in moist habitats and are vulnerable to desiccation (e.g. Cook & Loomis, 1928; Wegensteiner, 1982; Golovatch et al., 2015). The Colobognatha comprise the four taxa Platydesmida, Siphonophorida, Polyzoniida and Siphonocryptida (Blanke & Wesener, 2014), and are probably one of the least studied and most neglected millipede groups with regard to their taxonomy (Brewer et al., 2012; Read & Enghoff, 2018), biology and morphology (Manton, 1961; Hoffman, 1980; Read & Enghoff, 2009; Shorter et al., 2018). They share highly derived mouthparts, which are siphon-like in some groups (Manton, 1961) and used to feed on liquid food (Moritz et al., 2021, 2022). Colobognatha might feed on fungi, algae, or bacterially degraded substances (Lewis, 1984; Dunger, 1993; Macias et al., 2019). The biology of this group is largely unknown, and observations of living animals are rare, as many species have not been collected nor observed since their initial discovery. However, some colobognathan species are reported to exhibit subsocial or colonial behavior like brood care, even male brood care, and the formation of aggregations with individuals of overlapping generations (e.g. Murakami, 1962a; Gardner, 1974; Enghoff, 1984; Lewis, 1984; Wong et al., 2020). Only few studies focus on the postembryonic development of Colobognatha (Murakami 1962a, 1962b, 1963; David & Couret, 1983; Enghoff et al., 1993; Wong et al., 2020), but available data suggest that the Colobognatha differ from most other Diplopoda by hatching with more than three leg pairs (Enghoff et al., 1993; Blanke & Wesener, 2014). Postembryonic stadia in millipedes extend between two molts (Enghoff et al., 1993; Minelli, 2015), and in rare cases the color pattern of a millipede can reflect the course of anamorphosis; i.e. the number of tergites added between subsequent stadia (Enghoff, 2011). Studies on the anamorphosis (the addition of segments and legs) of millipedes do not only give an insight into their biology but also allow conclusions concerning the evolution and phylogeny of millipedes (Enghoff et al., 1993).

Siphonocryptid millipedes are considered to represent an independent order (Shelley, 2003), but were previously treated as a family or suborder of the Polyzoniida (e.g. Hoffman, 1977; Enghoff & Golovatch, 1995; Arndt et al., 2008). The group comprises 7 known species (Enghoff et al., 2015) in two genera (Hirudicryptus Enghoff & Golovatch, 1995, Siphonocryptus Pocock, 1894), which makes it the second most species-poor millipede order after the Siphoniulida (Sierwald & Bond, 2007; Liu et al. 2017). Despite their few species, Siphonocryptida show a very unusual distribution pattern. They occur in Sumatra (Siphonocryptus compactus Pocock, 1894), Malaysia (Siphonocryptus latior Enghoff & Golovatch, 1995; Siphonocryptus zigzag Enghoff, 2010), Taiwan (Hirudicryptus taiwanensis Korsós, Enghoff & Chang, 2008), Nepal (Hirudicryptus quintelementum Korsós, Geoffroy & Mauriès, 2009), Georgia and Russia (Hirudicryptus abchasicus Golovatch, Evsyukow & Reip, 2015), and on the Macaronesian islands (Hirudicryptus canariensis (Loksa, 1967)). Siphonocryptida probably had a wider distribution in the past and the today observed relict distribution might be due to extinction linked to changes in climate and vegetation (Korsós et al., 2009; Golovatch et al., 2015). The phylogenetic position of the Siphonocryptida within the Colobognatha remains unresolved (e.g. Blanke & Wesener, 2014), not the least because not a single genetic sequence is available, and their biology and internal anatomy remains largely terra incognita. The siphonocryptid species Hirudicryptus canariensis, which is endemic to Macaronesia, measures up to 10 mm in length and 1 mm in width and has a flattened body with lateral paraterga and a minute cone shaped head. The species was re-described in detail by Enghoff & Golovatch (1995) and the first data on its post-embryonic development was presented by Enghoff et al. (1993) and Enghoff & Golovatch (1995).

Hirudicryptus canariensis is restricted to the volcanic islands Madeira, La Gomera and Tenerife (Loksa, 1967; Enghoff, 1992; Enghoff & Golovatch, 1995; Vicente & Enghoff, 1999), where it can only be found in the Laurisilva, evergreen laurel subtropical cloud-forests. Around 20 million years ago laurel forests could also be found in Southern Europe and Northern Africa. However, the laurel forest largely disappeared in the Mediterranean due to climatic changes in the Quaternary period, and it only persisted on the Macaronesian islands as a relic (Morales et al., 1996; Fernández-Palacios et al., 2011). On Tenerife, laurel forests occur in the Anaga Mountains in the east, the Monte del Agua in the west, and in small patches between those areas in the north of the island (de Nascimento et al., 2009; González-Montelongo & Pérez-Vargas, 2019). The evergreen Laurisilva on Tenerife is formed by ca. 20 tree species, of which many are Canarian endemics, and which vary in their composition according to the prevalent environmental conditions (Morales et al., 1996). For example, the heath Erica platycodon (Webb & Berthel.) Rivas-Mart. & al. is part of the ridge-crest evergreen laurel forest ecosystem, which occurs at year-round cloudy and windy ridges with the highest precipitation (del Arco Aguilar et al., 2010).

Here we present observations of the biology, habitat, and postembryonic development of Hirudicryptus canariensis in the Anaga-mountains of Tenerife (Canary Islands, Spain). Furthermore, we provide the first barcode data (CO1) for the order Siphonocryptida, which will aid in future identification.

Collecting and observation

Hirudicryptus canariensis was observed and collected on Tenerife (Canary Islands, Spain) in the laurel forest (Laurisilva) of the Anaga mountains in October 2019 at 9 localities ( Table 1, Fig. 1a). The GPS-coordinates and elevation of each locality was recorded with the mobile-app maps.me (accessible via https://maps.me/). For the publication localities were numbered from west to east, and locality numbers differ from those originally used when collecting. A distribution map (Fig. 1a) was generated in Q-GIS 3.8 Zanzibar (QGIS Development Team 2019). Living specimens were observed and photographed in the field with an Olympus TG-5 digital camera equipped with an Olympus LG-1 light guide. Specimens from each locality were fixed in 95% ethanol and investigated with a Discovery.V12 stereomicroscope (Zeiss) to count the number of tergites and leg pairs, and to note the absence or presence of gonopods. This data has been combined with available data from the literature (Enghoff et al., 1993) to a total of 296 specimens (Table 2). Specimens with less than 15 tergites were counted as juveniles of uncertain sex, specimens with 15 tergites or more were counted as females if gonopods were absent, or as males if gonopods werepresent (see below). Images of the different stadia were taken with a Keyence VHX2000. Furthermore, the tergite number and color pattern of 107 living specimens was recorded from photographs taken in the field. We did not combine the data from collected animals and animals photographed in the field to avoid artificial duplication of data, as some of the photographed animals might have been collected afterwards. As discussed by Enghoff et al. (1993: 107) the Colobognatha do not have completely fused body-rings and the identification of segment or diplosegments can be problematic as the number of dorsal and ventral elements might not correspond to each other. Therefore, we rather refer to the number of tergites and leg-pairs added and to ‘podous’ and ‘apodous tergites’ respectively, assuming a legless collum followed by three haplosegments (tergites with a single leg-pair) and a variable number of diplosegments (tergites with two leg-pairs) as done by Enghoff et al. (1993: 127-130) for Colobognatha. All data this study is based on, including additional photographs and tables, are deposited on Zenodo (https://doi.org/10.5281/zenodo.6353889).

Scanning electron microscopy (SEM)

Scanning electron microscopy (SEM) was used to examine the number of ommatidia and antennomeres, as well as sexual characters (pseudopenes and gonopods). For SEM specimens were dehydrated in an ascending ethanol series and critical point dried with a Leica EM CPD 300. The specimens were mounted to SEM-stubs using conductive tape and sputtered with gold (ca 35 nm) using the Cressington Sputter Coater 108auto. SEM-images were obtained with a Zeiss Sigma 300 VP scanning electron microscope at the ZFMK. Here the number of antennomeres includes the apical disc (carrying the sensory cones) which is traditionally interpreted as 8^th antennomere (e.g. Verhoeff, 1928: 788; Enghoff, 1990). For a discussion see Koch (2015).

Barcoding (CO1)

DNA was extracted from two individual, one from locality 4 (ZFMK-MYR08837; ON007316) and one from locality 9 (ZFMK-MYR08838; ON007317). Following Spanish ABS bylaw 124/2017 of 24 February (Real Decreto 124/2017, de 24 de febrero, relativo al acceso a los recursos genéticos procedentes de taxones silvestres y al control de la utilización) access authorization for the utilization of genetic resources is not needed for exclusively taxonomic purposes (Capitoli I, Articolo 3.2) as defined in Capitoli I, Articolo 2.3 (https://www.miteco.gob.es/en/biodiversidad/temas/recursos-geneticos/protocolo-de-nagoya/FAQ_Autorizaciones.aspx). For DNA extraction, muscle tissue from legs and body segments was used. A fragment of the cytochrome c oxidase subunit I (CO1) mitochondrial gene was amplified with the primers HCO2198-JJ and LCO1490-JJ (Astrin & Stüben, 2008) and sequenced as outlined in previous studies (e.g. Wesener, 2015). BLAST searches (Altschul et al., 1990) were used to check the sequences for contaminations. Sequences were translated into amino acids to rule out the accidental amplification of pseudogenes. New sequences were uploaded to Genbank (see Table 4).Sequences were aligned by hand in Bioedit (Hall 1999). Distance analysis was performed in Mega 6 (Tamura et al., 2013) using the uncorrected p-distance model for a total of two sequences with 683 positions. Variation among sites was modeled with gamma distribution with shape parameter = 1. Included were the 1st+2nd+3rd codon positions.

Habitat and aggregations

We found Hirudicryptus canariensis exclusively on the heath Erica platycodon in the Anaga mountains at elevations between 630 and 870 m (Table 1, Fig. 1a). The animals occurred only in humid places on the lower part of the trees, on or under its bark, or under moss growing on its trunk (Fig. 1b, c). Those habitats were also inhabited by julidan millipedes (Julidae and Blaniulidae), Collembola, and nymphs of Aneurus cf. tagasastei Enderlein, 1931 (Heteroptera: Aradidae). Sifted soil samples taken adjacent to the trees and sampled with Winkler extraction did not include any specimens of H. canariensis.

Hirudicryptus canariensis mostly occurred in accumulations of juveniles in different stadia and adults (Fig. 2a–c), although at some localities only single (5) or few individuals (6, 7, 9), mainly adults, have been found. Larger individuals were often encountered coiled, with their head and collum stuck underneath the lateral paraterga of their mid-body (Fig. 2d, e). At several localities (1, 2, 3, 4 & 8) H. canariensis was found in large numbers (Fig. 2a, b), in some spots counting several hundred individuals (e.g. 2 & 4). Overall, most of the individuals found were juveniles of undetermined sex (63%), while males (19%) and females (18%) were nearly equally abundant. The highest proportion of juveniles of undetermined sex (90%) was found at locality 1 (Fig. 1a).

Disturbance of aggregations led to a flight response by adults and juveniles, but no defense fluid was secreted. Removal of juveniles did not lead to any response by mature individuals in their surroundings. TW observed in March 2012 an adult individual of uncertain sex coiling around an egg clutch at locality 7.

In one juvenile specimen with six tergites a fungus was observed growing on the basal article of one antenna (Fig. 3).

Postembryonic development

The specimens with the lowest number of tergites had six tergites and seven leg pairs and lacked apodous tergites and coxal sacks (Fig. 4a); those were the most abundant individuals (Table 2). Individuals with six tergites had a single pair of ommatidia, and five antennomeres (Fig. 5a) including the apical disc. Ozopores were present on tergites 5 and 6 (Fig. 4a). The next larger individuals counted ten tergites and 13 leg pairs (Table 2).

Individuals with more than six tergites had a single apodous tergite, coxal sacks from leg-pair 3 onwards (Fig. 4b-d), and two pairs of ommatidia (Fig. 5b-d). The number of antennomeres (including apical disc) increased stepwise from five antennomeres in individuals with six tergites (Fig. 5a), via seven observed in individuals with eleven and 16 tergites (Fig. 5b, c), to a maximum of eight (Fig. 5d) observed in an individuals with 25 tergites (Fig. 5d).

The specimens were white or pale brownish (Figs 2c, 6). In individuals with more than six tergites, a shift from darker to paler colored tergites could be observed. This color pattern was best visible in smaller living individuals (Fig. 6) and less distinct or absent in specimens stored in ethanol (Fig. 7) and in larger individuals. A shift from darker to paler coloration occurred behind tergite six (Figs 6, 7a-d) and behind tergite 10–14 (Figs 6, 7e-h) as well as in some individuals behind tergite 14–22 (Fig. 6). In few individuals such a pattern could also be observed on following tergites (Table 3). Larger individuals of H. canariensis were generally darker and did not show such a clear pattern. The color pattern can be used to infer the stadia (see discussion).

Modifications of the coxae of the second leg pair (which carry the gonopores) in both sexes, and of leg pair 9 and 10 (gonopods) in males, could be observed earliest in specimens with 15 tergites and 2 color shifts (stadium III). These structures were fully developed in specimens with 3 color shifts (stadium IV). No changes could be observed in larger individuals (Fig. 8). As the smallest individuals showing sexual modifications counted at least 15 tergites (Table 2) and showed 2 color shifts, all individuals with less than 15 tergites and 2 color shifts were counted as juveniles of undetermined sex, while all individuals with at least 15 tergites and two color shift were counted as males if gonopods were present, or as females if gonopods were absent. In females with two color shifts (stadium III) the second coxae were enlarged and carried a mesal fold (Fig. 8a). In females with 3 (stadium IV) and more color shifts the operculum, and the inner and outer valves of the vulvae were fully developed, and the second sternite showed a median fold (Fig. 8b). In males with two color shifts (stadium III) the developing pseudopenes (gonoapophysis) on the coxae of leg pair 2 were cylindrical and blunt (Fig. 8c), while the fully developed pseudopenes of subsequent stadia were conical and carried an apical incision (Fig. 8d). The immature gonopods (Fig. 8e) mainly differed from the fully developed gonopods (Fig. 8f) by the shape and size of the flagellate extensions on the apicalpodomeres. The conspicuous lobe observed on fully developed anterior gonopods, extending beyond the surrounding setae (Fig. 8f), was in the immature gonopod represented by an inconspicuous extension, which was less than 1/3 of the length of the neighboring setae (Fig. 8e). The apical podomere of the posterior gonopod, which extended into a long flagellum and surpasses the anterior gonopod when fully developed (Fig. 8f), was in the immature gonopods a short and thin extension, which had ca. the length of podomere 5 (Fig. 8e). The largest male had 41 tergites and 75 leg pairs (Fig. 7i). The largest individual found was a female with 54 tergites and 101 leg pairs (Figs 6, 7j). The number of leg pairs was always uneven (Table 2).

Barcodes and genetic analyses

The barcoded CO1 fragments (Table 4) included 683 positions. The genetic pairwise distance between the two individuals from locality 4 and 9 was 0.5%

Habitat and distribution

Hirudicryptus canariensis seems to be associated to the plant Erica platycodon [Erica scoparia subsp. platycodon (Webb & Berthel.) A.Hansen & G.Kunkel] and humid conditions, as it has only been found on those trees and not in its surrounding in this study. For H. canariensis from La Gomera and Madeira we lack information on which plant they occurred (Enghoff & Golovatch, 1995). Erica platycodon has the two subspecies Erica platycodon subsp. platycodon, endemic to the Canary Islands, and Erica platycodon subsp. maderincola (D.C.McClint.) Rivas Mart. & al., endemic to Madeira (e.g. Rivas-Martinez et al., 1993; Arozena et al., 2008; Fagúndez et al., 2009; POWO, 2022). The close affinity of H. canariensis to Erica platycodon might be related to the structure of the bark of Erica platycodon, which easily peels off and creates abundant crevices. With its small (up to 10 mm in length, ca 1 mm in width) and dorso-ventrally flattened body, H. canariensis might be adapted toa subcortical life, as small and flattened bodies are typical for subcortical millipedes (Golovatch & Kime. 2009). However, other siphonocryptids (Hirudicryptus abchasicus, H. taiwanensis, H. quintumelementum, Siphonocryptus zigzag and S. latior) with a similar habitus are known to inhabit soil and leaf litter (Enghoff & Golovatch, 1995; Korsós et al., 2008, 2009; Enghoff, 2010; Golovatch et al., 2015; Zuev, 2017). On the Canary Islands, Erica platycodon is mainly found in the ridge-crest evergreen laurel forests, which occurs at year-round cloudy and windy ridges, where the highest precipitation occurs (del Arco Aguilar et al., 2010). Therefore, the presence of H. canariensis on Erica platycodon might be related to favorable conditions, with high humidity due to precipitation of mist, where E. platycodon growth. The distribution of E. platycodon (Government of the Canary Islands) matches (with the exception of its presence on El Hierro, where H. canariensis is currently unknown) the knowndistribution of H. canariensis on Tenerife, La Gomera and Madeira (Loksa, 1967; Enghoff, 1992; Enghoff & Golovatch, 1995; Vicente & Enghoff, 1999). These laurel forests are a tertiary relict of the subtropical North Thetian forest (del Arco Aguilar et al., 2010) and harbor a great biodiversity with high levels of endemism (e.g. Machado, 1976; Betzin et al., 2016). The absence of H. canariensis on El Hierro is either due to a lack of collecting efforts, or more likely the relatively young age of El Hierro (1.1 Ma) compared to Tenerife (11.5 Ma), La Gomera (12 Ma) (Fernández-Palacios & Whittaker, 2008) and Madeira (5 Ma) (Geldmacher et al., 2000, 2005), and the low dispersal ability of H. canariensis compared to the wind-dispersed heathers.

The patchy distribution of the Siphonocryptida is probably the result of extinction events of a once wider Palearctic distribution related to changes in climate and vegetation, as already pointed out by Korsós et al. (2008, 2009) and Golovatch et al. (2015). Thus, the laurel forests largely disappeared in Europe and only persist in relic habitats on Madeira and the Canary Islands (Fernández-Palacios et al., 2011) alongside H. canariensis.

Sociality and aggregations

We suggest that Hirudicryptus canariensis is social (Costa, 2018) or subsocial (Choe & Crespi, 1997) with parental brood care (see also Blanke & Wesener, 2014; their character 52) and forms aggregations. Yip & Rayor (2014) subdivide subsociality, defining egg-guarding alone as ‘transient subsociality’ in contrast to ‘prolonged subsociality’, which is defined as the guarding of juveniles or larvae over some period of time. As only eggs are guarded and disturbance of aggregations of juvenile and mature individuals did not lead to any protective response by the latter, we consider H. canariensis as ‘transient subsocial’. Hirudicryptus canariensis forms aggregations with large numbers of individuals of different stadia, where it is not clear whether this is the result of preferable conditions and availability of food or indicates sociality (i.e. attraction and interaction to each other). Such aggregations of H. canariensis have not been reported before and juveniles have been collected singly (Enghoff & Golovatch, 1995). There are reports on social and colonial behavior in other Colobognatha (e.g. Enghoff et al., 2015):Aggregation behavior of individuals of overlapping generations, as well as paternal brood care has been reported for the platydesmids Brachycybe lecontii Wood, 1864 (Gardner, 1974; Wong et al., 2020), Brachycybe nodulosa (Verhoeff, 1935) (Murakami, 1962a; Kudo et al., 2011) and Yamasinaium noduligerum Verhoeff, 1939 (Kudo et al., 2009), while maternal brood care has been reported for the platydesmids Dolistenus savii Fanzago, 1874 (Silvestri, 1950) and Dolistenus humicolus Verhoeff 1926 (Minelli, 1981), as well as for the polyzoniids Orsiboe ichigomensis Attems, 1909 (Murakami, 1962a) and Polyzonium germanicum Brandt, 1837 (Rimsky-Korsakow, 1895; Verhoeff, 1928). Therefore, parental care could be a common feature of the Colobognatha but needs further investigations of more taxa. So far brood care has not been reported for Siphonophorida. As in the platydesmidan Brachycybe nodulosa (Murakami, 1962a), the ratio of females to males in aggregations of Hirudicryptus canariensis is almost 1:1 (Fig. 1a), while a female biased sex ratio has been reported for Brachycybe lecontii (Wong et al. 2020) and Pseudodesmus sp. (Lewis 1984).

Post embryonic development – stadium I

Based on the available material we suggest that Hirudicryptus canariensis hatches with six tergites and seven leg pairs (= stadium I). Enghoff et al. (1993) and Enghoff & Golovatch (1995) also report the smallest individuals to have six tergites and seven leg pairs but suspect that these might represent stadium II. We argue that these individuals represent stadium I, because no smaller individuals could be found despite extensive sampling, and more importantly, stadium I individuals of other colobognathan taxa (Polyzoniida and Platydesmida) also have six tergites. However, the number of legs (Fig. 8) varies, with six tergites and four leg pairs in Polyzonium germanicum and Brachycybe nodulosa (Rimsky-Korsakow, 1895; Murakami, 1962b; David & Couret, 1983, 1985; Enghoff et al., 1993), and six tergites and five leg pairs in Brachycybe lecontii (Wong et al., 2020) (Fig. 8). Stadium II already has eight or more tergites in Polyzoniida (David & Couret, 1983; Enghoff et al.,1993) and Platydesmida (Murakami, 1963; Wong et al., 2020). We suggest that in stadium I the ultimate tergite, which is apodous in Polyzoniida and Platydesmida (Murakami, 1962b; David & Couret, 1983; Wong et al., 2020), carries two leg pairs in H. canariensis, while only in the following stadia the ultimate tergite is apodous, as is the case in the remaining Colobognatha. Thus, podous tergites develop without an apodous precursor between stadium I and stadium II. Other characters supporting that H. canariensis with six tergites represent stadium I are (1) the presence of a single pair of ommatidia, (2) antennae consisting of only five antennomeres (including the apical disc), and (3) the absence of coxal sacks. (1) A single pair of ommatidia is also present in stadium I of Polyzonium germancium. (2) Only five antennomeres are present in stadium I of P. germanicum (Rimsky-Korsakow, 1895), Brachycybe nodulosa (Murakami, 1962b) and Brachycybe lecontii (Wong et al., 2020: fig. 5),while the following stadia carry more ommatidia (if present) and antennomeres (seven antennomeres in stadium II, eight antennomeres in following stadia (Wong et al., 2020: fig. 5; including apical disc), as is also the case in H. canariensis. (3) Coxal sacks, only present in H. canariensis with more than six tergites, first appear in Brachycybe lecontii in individuals of stadium II (Wong et al., 2020), while for other Colobognatha this state has not been documented.

Post embryonic development – subsequent stadia

In juveniles and younger mature individuals of Hirudicryptus canariensis, the more or less regular color pattern corresponds to the total tergite numbers commonly observed in smaller individuals. We suggest that the color pattern can be used to trace the postembryonic development of the earliest stadia, as might be the case in several millipedes like some Platydesmida and the syphonocryptid Siphonocryptus zigzag Enghoff, 2010, which show a trans-segmental color pattern (Enghoff, 2011). In H. canariensis we observed rather a change in intensity than a well-developed trans-segmental pattern. The anterior six tergites (corresponding to stadium I) are often darker than the following tergites, and juveniles with 6 tergites were common. Subsequent shifts from darker to paler tergites could be observed in several individuals behind tergite 10–14, 14–22 and 18–28, which corresponds to the total numbers of tergites observed in some individuals. Based on the color pattern and the tergite numbers observed, we conclude that the number of color shifts +1 gives the number of the stadium, and that stadium I has 6 tergites and 7 leg pairs, stadium II has 10–14 tergites and 13–21 leg pairs, stadium III has 14–22 tergites and 21–37 leg pairs, and stadium IV has 18–28 tergites and 29–49 leg pairs. This is largely in congruence with the postembryonic development known from the platydesmidBrachycybe nodulosa, in which stadium I has 6 tergites, stadium II has 10–14 tergites, stadium 3 has 15–21 tergites and stadium IV has 21–29 tergites (Murakami, 1963: fig. 1). In H. canariensis the number of tergites and legs added during molting is irregular and can overlap between individuals of different stadia as is the case in Polyzonium germanicum (David & Couret, 1983, 1984) and Brachycybe (Murakami, 1963; Wong et al., 2020), and therefore is not reliable for the determination of stadia. In contrast to that, Hirudisoma latum (Ribaut, 1908) has a fixed number of tergites and leg pairs per stadium (Brolemann, 1935; Enghoff et al., 1993). In H. canariensis there is a strict correspondence between the always odd number of leg pairs (l) and the number of tergites (t), and it can be calculated with the formula l = 3 + 2(t – 5) (the number of haplosegments with a single leg pair is 3, the number of diplosegments with two leg-pairs is t – 5, as the collum and the ultimate diplosegment do not carry legs, and the 3 haplosegments have to be excluded).This formula applies for all stadia except for stadium I, which lacks an apodous tergite. In Polyzonium germanicum (David & Couret, 1983; Enghoff et al., 1993) and Brachycybe (Murakami, 1963; Wong et al., 2020), individuals with the same number of tergites can have a variable numbers of leg pairs, which is always even in P. germanicum (David & Couret, 1983). In other Colobognatha there can be several apodous ultimate tergites (e.g. Murakami (1963) and Wong et al. (2020) for Platydesmida, David & Couret (1984) for Polyzoniida). In contrast there is always only a single apodous ultimate tergite in H. canariensis. This means that in H. canariensis several podous tergites can develop de novo without an apodous precursor (or from a single apodous tergite), while in Polyzonium germanicum the number of new podous tergites always corresponds to the number of apodous tergites in the previous stadium (David & Couret 1983: table 5). Like the remaining Colobognatha H. canariensis develops by euanamorphosis; i.e. segments are added during each molt throughout life (Enghoff et al., 1993; Minelli, 2015).

We suggest that in H. canariensis the sexual characters of both sexes (vulvae, penes, gonopods) develop gradually via two molts, with not fully developed structures in stadium III (with 2 color shifts), and fully developed structures in stadium IV (with 3 color shifts). In subsequent stadia the external structure of the sexual organs does not change, which is of importance for taxonomic studies, which largely rely on gonopods for species identification. The smallest individuals of H. canariensis showing sexual modifications (including immature gonopods) observed in this study had 15 tergites, while the smallest individual reported by Enghoff et al. (1993) and Enghoff & Golovatch (1995) had 16 tergites. Our observations correspond to the first appearance of immature gonopods in Polyzonium germanicum with 15–21 tergites (David & Couret, 1983), and in Hirudisoma latum with 15 tergites (Brolemann, 1935; therein 16 including preanal ring), although these are considered as stadium IV. In H. canariensis, fully developed gonopods could be observed in stadium IV, while these appear in stadium V in P. germanicum (David & Couret, 1983) and H. latum (Brolemann, 1935). In the platydesmidan Brachycybe nodulosa males and females can be distinguished from stadia III onwards, suggesting the presence of leg-modifications in this stadium (Murakami, 1963: table 3). In B. lecontii immature gonopods first appear in individuals with 24 tergites, corresponding to stadium IV, and are fully developed in individuals with at least 35 tergites, corresponding to stadia V–VII (Wong et al. 2020). As the phylogeny of the Colobognatha remains unresolved (e.g. Blanke & Wesener 2014) and the development is only known for very few colobognathan species it remains unknown how these different developmental patterns evolved.

Fungi

The fungus growing on the antennae of a juvenile (stadia I) of H. canariensis does not belong to the Laboulbeniales, which are commonly reported from millipedes (e.g. Santamaria et al. 2014, 2016; Enghoff & Santamaria 2015), but is one of the less known non-Laboulbenialean fungi associated to millipedes as reported by Enghoff & Reboleira (2017). To our knowledge, this is the first parasitic fungus reported from a colobognathan millipede.

Barcodes and genetic analyses

The genetic distance of 0.5% between the two analyzed specimens of Hirudicryptus canariensis is relatively low, as for Glomeris marginata (Villers, 1789) intraspecific genetic distances of up to 4.7% are reported (Reip & Wesener, 2018). This high similarity in H. canariensis is most likely due to the geographical proximity of the populations with a distance of 7.3 km. Generally, the genetic distance increases with geographical distance, although in some cases Glomeris marginata showed 0.0% genetic p-distance between individuals sampled more than 1000 km apart, while larger genetic p-distances of 3% occurred in closer proximity of 9 km (Reip & Wesener, 2018). The moderate genetic distance of 0.5% in the analyzed CO1 fragment supports the view that H. canariensis is indeed native to Tenerife and not introduced from another locality. This is further supported by the very specific habitat of the species. Invasive species often show a low genetic diversity due to the founder effect or a higher genetic distance than native populations when multiple introductions occur (e.g. Tsutsui et al., 2000; Allendorf & Lundquist, 2003; Peacock et al., 2009; Estoup & Guillemaud, 2010). Genetic analyses could be used in future studies for a better understanding of the biogeography of Hirudicryptus canariensis, as the species is distributed on the volcanic Canary Islands Tenerife and La Gomera, and on the ca. 450 km distant Madeira.