The female genitalic region of Petrobiellus akkesiensis is quite typical for Machilidae. The coxites IX and styli IX (excluding the apical spine) are nearly equal in length (Fig. 1A). The ovipositor, composed of paired gonapophyses VIII (anterior gonapophyses) and gonapophyses IX (posterior gonapophyses), is long, slender, and subparallel, extending beyond the apex of the styli IX by about twice their length (Fig. 1A, D). The terminal setae at the tips of the gonapophyses VIII and IX are long and simple (Fig. 1C, F). The chaetotaxy of the gonapophyseal articles is uniform (Fig. 1B, C, E, F).

The male genitalic region of P. akkesiensis is highly specialized. Paramere VIII is absent. The coxites IX and styli IX are remarkably robust, with the styli IX (excluding the apical spine) measuring about half the length of the coxites IX (Fig. 2A). The coxites IX are mesally concave, and both the penis and parameres IX are exposed between the two coxites (Fig. 2A). A row of strong setae occurs near the posterior margin of the coxites IX (Fig. 2B, C). The styli IX are dorsomesally concave, and the concave surface is densely furnished with approximately 20 dark brown, strong spines (Fig. 2B–D; see fig. 2I in Mtow and Machida 2024). The penis is long, extending beyond the distal end of the coxites IX by 40% of the coxite length (Fig. 2A–C), composed of two articles. The proximal article, accounting for approximately 80% of the penis length, is straight, whereas the distal article is dorsally curved, terminating in a wide genital opening (phallotreme) (Fig. 2E, F). Short setae are densely distributed on the dorsal surface of the posterior part of the proximal article and the entire distal article (Fig. 2F, G). The parameres IX, located ventral to the penis, are slender and composed of I + 5–6 articles: the proximalmost article, representing 60–70% of the total paramere length (Fig. 2A, E). Tubular setae (grooved setae, as described by Sturm 1991) are observed on neither the penis nor the parameres IX (Fig. 2E–H).

In both male and female Petrobiellus, a pair of ventral sacs is present on the abdominal segments I, VI, and VII, and two pairs are present on each of the abdominal segments II to V.

We conducted 19 mating experiments with Petrobiellus akkesiensis. In 9 of these no sexual interactions occurred, whereas in the remaining 10 experiments the pairs displayed mating behavior. Of the 10 experiments involving mating interactions, 6 resulted in successful matings. In 1 of the 4 unsuccessful cases, the male failed to grasp the ovipositor; in the other 3 cases, the females unilaterally terminated genital contact. The 6 successfully completed matings are summarized in Table 1.

The mating experiment begins when a male and female P. akkesiensis are placed into the arena (glass Petri dish or plastic case). As soon as the male recognizes the female, he rushes toward the female and positions himself either to her left or right side, placing his maxillary palps on her thorax (cf. Fig. 3A). In 8 of the 10 experiments, the male stood right of the female; in the remaining two, left of her. After securing his position, the male brings his forelegs and head (particularly the clypeal region) close to the female’s body, continuing to use his maxillary palps for contact. He then begins vibrating the female, simultaneously twisting his body and pressing his postabdomen against hers. This vibration continues throughout the pairing, with occasional pauses, and may persist even after successful spermatophore transfer or after genital contact between the pair has ended.

Between 10 sec and 2 min after the pair assumes the abovementioned formation (n = 9), the male protrudes his penis at a right angle from between the coxites IX on both sides to locate the female’s ovipositor. He then grasps the basal part of the ovipositor with the hook-like tip of the penis (Fig. 3B; File S1 [Movie at 02:09], File S2 [Movie at 00:27], File S3 [Movie at 01:10]). The genital opening at the distal end of the penis is hidden behind the ovipositor. The male’s styli IX also assist in holding the ovipositor. While probing for the ovipositor with the penis, the male simultaneously prepares the styli IX for holding: the stylus on the side closest to the female is bent forward, whereas the opposite stylus extends backward (e.g., Fig. 4A; File S3 [Movie at 00:55]). Once the penis grips the ovipositor, the stylus on the side nearest to the female (i.e., the right stylus IX when the male is on the left side of the female, and vice versa), which is bent forward, presses down on the middle portion of the ovipositor from above with the dorsal surface of its proximal region. Meanwhile, the opposite stylus (i.e., the left stylus IX when the male is on the left side of the female, and vice versa), which is stretched backward, pushes up on the ovipositor’s middle region from below using its dorsomesal concave area, lined with numerous strong spines. Thus, the ovipositor is secured at three points: at its base by the penis and at its middle by the styli IX on both sides. This holding occurs quickly: approximately 2 min after the male first captured the female, as shown in Fig. 3A–C (Case I in Table 1), or in < 30 sec, as shown in Fig. 3D (Case V in Table 1). The parameres do not appear to participate in mating; they are positioned ventral to the penis, although they are occasionally observed to separate from it (File S3 [Movie at 01:10–01:16]).

As soon as the ovipositor holding is established, the male’s ventral sacs begin to swell, progressing from the posterior to the anterior abdominal segments: first, a pair of ventral sacs VII and VI expands (Figs 3D, 4B; File S2 [Movie at 00:46], File S3 [Movie at 01:16]), followed by two pairs on each of the abdominal segments V and IV, and finally one of the two pairs on the abdominal segment III enlarges (Fig. 4C; File S3 [Movie at 01:44]). Simultaneously, an opaque spermatophore is discharged onto the dorsal side of the ovipositor from the genital opening at the tip of the penis (Fig. 4C; File S3 [Movie at 01:50]; cf. Fig. 4D), rapidly expanding to its maximum size, approximately 0.7 mm in diameter (approximate volume: 0.18 mm³), within a few minutes (Fig. 4D). Once the spermatophore reaches its full size, the swollen ventral sacs begin to shrink in reverse order, from the anterior to the posterior abdominal segments (Fig. 4E), and eventually retract beneath the coxites (Fig. 4F; File S3 [Movie at 08:24–08:42]). Briefly thereafter, presumably due to the uptake of sperm fluid (through the gap between gonapophyses) by the ovipositor, the spermatophore gradually diminishes in size (Fig. 4G). It is not completely absorbed, and its final size may remain as large as that shown in Fig. 4I. This marks the completion of spermatophore transfer.

After spermatophore transfer is complete, the male’s styli IX soon release the ovipositor (Fig. 4H), but it takes longer for the penis to disengage (File S1 [Movie at 09:58], File S2 [Movie at 25:36], File S3 [Movie at 13:52]). Even after the penis has released the ovipositor, the male continues to hold and vibrate the female for a short while before they eventually separate (Fig. 3C; File S1 [Movie at 12:45], File S2 [Movie at 25:48]). The entire mating behavior of P. akkesiensis, from the moment the male holds the female to their final separation, lasts approximately 10–30 min (n = 6, Table 1).

Among the specimens preserved in alcohol, we found a female with a spermatophore attached to the dorsal side of the basal region of her ovipositor. The spermatophore had a conical shape tapering posteriorly, with a base approximately 0.2 mm in diameter and a height approximately 0.4 mm (approximate volume: 0.005 mm³; Fig. 5A–C). The size of the spermatophore was nearly the same as those reduced to their minimum volume (cf. Fig. 4I). The posteriorly pointed tip of the spermatophore was inserted between the left and right gonapophyses of the ovipositor (Fig. 5C, D). Close examination of the spermatophore surface revealed numerous winding, thread-like structures, likely sperm, embedded in but partially exposed on the surface (Fig. 5E).

Kaplin (1985), who proposed a three-subfamily system of Machilidae, placed Ditrigoniophthalmus, which is often considered to be a paleoform of Machilidae, under Petrobiellinae, and some researchers have followed this (Zhang et al. 2018; Montagna 2020). However, this genus cannot be included in Petrobiellinae because of significant differences in diagnostic features, such as the number of abdominal styli (cf. Kaplin 1979, 2000). Other authors therefore regarded Petrobiellinae as a monogeneric subfamily consisting solely of the genus Petrobiellus (cf. Sturm and Bach de Roca 1993; Sturm and Machida 2001). In fact, after excluding Ditrigoniophthalmus from Petrobiellinae, Kaplin (2000) placed it in a newly established subfamily, Ditrigoniophthalminae, created exclusively for this genus. Petrobiellinae has generally been considered a derived lineage within Machilidae, along with Petrobiinae (Sturm and Bach de Roca 1993; Sturm and Machida 2001). However, Kaplin (1985) assigned Petrobiellinae an ancestral position within the family, and recent total evidence analyses indicate a close relationship of Petrobiellinae and Ditrigoniophthalmus (Zhang et al. 2018; Montagna 2020; cf. Matushkina and Klass 2020). Furthermore, recent molecular phylogenetic studies and total evidence analyses have indicated a close relationship between Petrobiellinae and Meinertellidae (Ma et al. 2015; Zhang et al. 2018; Montagna 2020; Guan et al. 2021), although morphological evidence to support this affinity is lacking (Sturm and Bach de Roca 1993; Sturm and Machida 2001; Klass and Matushkina 2018). Two species of Petrobiellus described by Ma et al. (2015), which are critical to the proposed relationship between Petrobiellinae and Meinertellidae, merit careful reconsideration. Ma et al. (2015) identified jumping bristletails collected from inland Yunnan, southern China (near Laos), as Petrobiellus, and named them Petrobiellus bannaensis and Petrobiellus puerensis as new species. However, as Kaplin (2020) noted, these descriptions violate Article 13.1.1 of the International Commission on Zoological Nomenclature (1999), as they are based only on molecular data without morphological descriptions. As such, these two scientific names are invalid. Moreover, the purported distribution of these two “Petrobiellus” species is problematic: they were described from inland China, whereas Petrobiellus is known only from coastal regions of the Far East (see below). Given that the molecular phylogenetics aligns these two species to the family Meinertellidae, the validity of their assignment to Petrobiellus is questionable. Notably, the subtropical and tropical rainforests near the collection sites in China are likely (if not exclusively) inhabited by meinertellid genera such as Machilellus Silvestri, 1911, Machilontus Silvestri, 1912, and Megalopsobius Silvestri, 1912. Additionally, several key diagnostic features of Petrobiellus, such as the absence of scales on the head, cephalic appendages including antennae, and legs (see below), are also shared by Meinertellidae. Therefore, the assignment of these two inland Chinese species to Petrobiellus is doubtful, as is the inferred relationship between Petrobiellinae and Meinertellidae, which was based on molecular phylogenetic analysis treating these species as representatives of Petrobiellus.

Petrobiellinae is a monogeneric subfamily represented by Petrobiellus. The genus was established by Silvestri (1943) based on several diagnostic features, the major ones being 1) the absence of scales on all parts of the antennae, mandibles, maxillae, labium, legs, and styli; 2) dumbbell-shaped lateral ocelli; 3) coxal stylets on the mid- and hindlegs; and 4) one pair of ventral sacs on the segments I, VI, and VII and two pairs on each of the segments II to V. To date, several species have been recorded from the coastal regions of the Far East, specifically Japan and Russia: Petrobiellus takunagae from Shirahama (Wakayama, Japan), Shimoda (Shizuoka, Japan), and Sado Island (Niigata, Japan) (Silvestri 1943; Machida 2020; Mtow 2021); Petrobiellus akkesiensis from Akkeshi and Noboribetsu (Hokkaido, Japan) (Uchida 1949; Mtow and Machida 2024; herein); Petrobiellus curvistylis Uchida, 1954 from Hachijo Island (Tokyo, Japan) (Uchida 1954); Petrobiellus kusakini Kaplin, 1980 from Simushir Island (Chishima Islands, Japan) (Kaplin 1980); and Petrobiellus sachalinensis Kaplin, 2020 from Sakhalin (Russia) (Kaplin 2020). Machida (2020) reported that P. curvistylis from Hachijo Island is likely a synonym of P. takunagae, which is widely distributed along the coastal regions of the Okinawa Islands to Honshu. Uchida (1954) also suggested this possibility in the original description of P. curvistylis. If this synonymy is correct, the genus Petrobiellus currently comprises four species.

Males and females are known in P. curvistylis, which was described based on two males, two females, and three juveniles, but only females are known for P. kusakini and P. sachalinensis. P. takunagae and P. akkesiensis were also initially described based only on female specimens, but males were eventually discovered, although the sex ratio in their populations is heavily biased toward females. As a result, the understanding of male morphology in this genus has progressively advanced (for P. takunagae: Sturm and Bach de Roca 1993; Sturm and Machida 2001; Klass and Matushkina 2018; for P. akkesiensis: Mtow and Machida 2024; herein).

The general features of the genitalic region observed in Petrobiellus akkesiensis were consistent with those reported for other Petrobiellus species. The genitalic characteristics of Petrobiellus are summarized below.

Female. (cf. Silvestri 1943; Uchida 1954; Sturm and Bach de Roca 1993; Klass and Matushkina 2012; Kaplin 2020; Mtow and Machida 2024; herein). The female genitalic region is normal. The coxites IX and styli IX (excluding the terminal setae) are approximately equal in length. The ovipositor is long, slender, and subparallel. The chaetotaxy of the ovipositor is simple, with long terminal setae at the tips of both gonapophyses.

Male. (cf. Uchida 1954; Klass and Matushkina 2018; Matushkina and Klass 2020; Mtow and Machida 2024; herein). The male genitalic region is highly specialized. The coxites IX and styli IX are remarkably robust, with the styli IX measuring approximately half the length of the coxites IX. The coxites IX are mesally concave, and the penis is almost fully exposed between the coxites IX on both sides. A row of strong setae is present near the posterior end of the coxites IX. The styli IX are dorsomesally concave, and their concave surfaces are densely covered with strong spines. The penis consists of a long, straight proximal article and a short distal article, the latter curving dorsally and terminating in a wide genital opening. Short setae are densely distributed on the dorsal surface of the posterior part of the proximal article and on the distal article. The parameres VIII are absent. The parameres IX are slender, located ventrally to the proximal article of the penis, slightly exceeding its length, and composed of I + 3–6 articles: I + 3 in Petrobiellus takunagae (Sturm and Bach de Roca 1993; Klass and Matushkina 2018) and I + 5–6 in P. akkesiensis (Mtow and Machida 2024; herein). Uchida (1954) did not describe the articulation of the parameres in Petrobiellus curvistylis, likely due to limited observational detail. Both the penis and parameres IX lack tubular setae (referred to as grooved setae by Sturm 1991).

The ovipositor of Petrobiellus is slender and subparallel, and its chaetotaxy is regular, without setae/spines specialized for digging. It can be categorized as the primary type according to the classical classification (cf. Wygodzinsky 1941; Janetschek 1991). Sturm and Bach de Roca (1993), with further reference to Janetschek (1991), argued that the classification of the ovipositor into primary and secondary types does not reflect phylogeny, and they proposed a new system that distinguishes four types: 1) the primary type is found in Charimachilis, Mesomachilis, and some other genera of Machilinae; the ovipositor is short, not surpassing the tip of the styli IX, and is generally thickened distally; the tip of the gonapophyses VIII may be pointed, rounded, or have a pointed chitinous projection; 2) the secondary type is usually found in Machilinae and rarely in Petrobiinae and Meinertellidae; the ovipositor is also short, not surpassing the tip of the styli IX, and is generally thickened distally, but the gonapophyses VIII bear strong digging bristles and always lack chitinous teeth on the distal articles, and their tips are never pointed; 3) the tertiary type is common in Machilinae, Petrobiinae, and Meinertellidae; the ovipositor is long, usually surpassing the tip of the styli IX, and is not thickened distally; the chaetotaxy is regular and lacks specialized setae on the gonapophyseal articles; and 4) the quaternary type is known only from some genera of Meinertellidae; it resembles the tertiary type but shows an interrupted pattern in chaetotaxy on the distal articles of the gonapophyses, with setation lacking at intervals. Sturm and Bach de Roca (1993) stated that the primary type is clearly the ancestral form, and the quaternary type is the most derived. Although they did not discuss the phylogenetic implications of the secondary and tertiary types, these two are the most widely distributed among the four. The ovipositor of Petrobiellus is classified as the tertiary type.

As summarized by Sturm and Machida (2001), spermatophore transfer in Archaeognatha is classified into three modes (see 4.3.3.). The first mode, “indirect transfer of spermatophore(s) deposited on carrier thread” (mode 1), is known from Machilinae of Machilidae and Pedetontus in Petrobiinae of Machilidae, and considered the ancestral mode and the groundplan of spermatophore transfer in Archaeognatha. The other two modes are derived and involve no thread-spinning: one is the “direct transfer of spermatophore by deposition on ovipositor” (mode 2), known from Petrobius of Petrobiinae; the other is the “indirect transfer of spermatophore deposited on stalk” (mode 3), known exclusively in Meinertellidae. The mating behavior of Petrobiellus, described in 3.2. and discussed in 4.3.2., represents a unique mode that cannot be classified within any of the three previously known categories. In this newly observed behavior, the spermatophore is discharged directly onto the ovipositor, which is grasped by the penis and held by the styli IX, and transferred directly to the female. During mating, the male of Petrobiellus protrudes his penis at a right angle from between the coxites IX and grasps the basal region of the ovipositor with the distally-curved penis. The effective grasping of the ovipositor is facilitated by the suitable length of the penis, the dorsally curved, hook-like shape of its distal article, and the dense setation on the dorsal surfaces of the posterior part of the proximal article and the distal article. The styli IX on both sides also assist in holding the ovipositor in coordination with the penis. The concaved dorsomesal surfaces of the styli, furnished with many strong spines, are well-suited for this role. The coxites IX are mesally concave, creating a wide space between them, which allows the penis greater freedom of movement. Additionally, the row of strong setae near the distal ends of the coxites IX may function to prevent the ovipositor from slipping forward. Thus, the unique and specialized features of the male genitalic region in Petrobiellus are clearly associated with this specialized mode of mating and can be regarded as remarkable autapomorphies of the genus.

Klass and Matushkina (2018) argued that mating involving the spinning of threads, which is most commonly observed in Archaeognatha, the sistergroup of Dicondylia, and exclusively in all Zygentoma, the sistergroup of Pterygota, represents the groundplan of Ectognatha. They focused on the distribution pattern of tubular setae, which are specialized for spinning threads, in Archaeognatha and Zygentoma. These setae are present only on the parameres in archaeognathans that perform mode 1 mating (see Klass and Matushkina 2018); on both the parameres and the penis in Lepisma saccharinum Linnaeus, 1758 [see Sturm 1956; Matushkina and Klass 2020; Sturm (1997) mentioned their presence only on the parameres] and only on the penis in Thermobia domestica (Packard, 1873) and Tricholepidion gertschi Wygodzinsky, 1961 (see Sturm 1987, 1997; Matushkina and Klass 2020). Based on this, Klass and Matushkina (2018) inferred that the groundplan of Ectognatha may have included tubular setae on both the parameres and the penis. In this context, Mesomachilis, sometimes referred to as a paleoform of Archaeognatha, is particularly noteworthy. Mesomachilis is composed of two subgenera: Mesomachilis Sturm, 1991 and Raromachilis Sturm, 1991. In the former, tubular setae are present on both the parameres and the penis, whereas in the latter, they are found only on the parameres (Sturm 1991).

Petrobiellus possesses parameres only on abdominal segment IX, while lacking those of the abdominal segment VIII. The parameres of Petrobiellus are thin and have not been observed to actively participate in mating; they hug to the ventral side of the penis, although they are occasionally found to diverge from it. However, Sturm and Machida (2001) and Klass and Matushkina (2018) inferred that the parameres of P. takunagae, which bear numerous setae on their surfaces, may function as sensors for detecting the positional relationship between the penis and the ovipositor. Sturm and Bach de Roca (1993), Sturm and Machida (2001), and Matushkina and Klass (2020) proposed that 1) the presence of both parameres VIII and IX, 2) articulation of the distal article, and 3) possession of tubular setae represent the groundplan features in Archaeognatha; conversely, 1) the reduction of the parameres VIII (and even the parameres IX), 2) obliteration of articulation in the distal article, and 3) loss of tubular setae are considered derived features. Thus, the lacking of the parameres VIII and the absence of tubular setae in the parameres IX of Petrobiellus can be understood as derived conditions.