In 2020, I received papered females of S. bouyeri from southeastern Tibet. One ♀ had been collected at the type-locality Bomê County, 2111 m, on 23 June, and 2 ♀♀ had been collected in Nyingchi City, 2052 m, on 22 June. From all of these, only one egg was obtained, off-white in color and not hatching during that year. I opened it on 05 April 2021; there was no embryo inside. During my expedition to the southern border of Tibet in the summer of 2021, 2 ♂♂ and 1 ♀ of S. bouyeri were captured during the nights of 26–29 June, at 2134 m in Mêdog County. Unfortunately, this female also laid only a single ovum in the paper triangle envelope, and it was lost on the following ecological survey. On 17 June 2022, I went back to the place, 12 ♀♀ and 3 ♂♂ were collected during the next 10 days. The females finally oviposited in a closed cylindrical net-cage (1.8–2.2 mm mesh), which was hung outdoors away from the vegetation in Mêdog. These caged eggs were later collected into a gauze bag (single mesh opening width: ca. 0.1 mm) and driven back to the research site in Kunming City, Yunnan, at an elevation of 1940 m. The eggs were placed indoors in a gauze cage (0.3 mm mesh) on a piece of mesh suspended ca. 1 cm above a moisture-saturated cotton pad, and sprayed with water every day. The resulting air humidity surrounding the eggs was ca. 80–95% RH at a temperature of ca. 17–21°C.

Newly hatched caterpillars were offered a large variety of host plants. A total of 35 species within 26 families of plants were chosen for the tests to larvae (sections 3.2.1–3.2.26), including 11 families commonly fed by African saturniids, and several families used to breed Asian species in the subfamilies Saturniinae, Salassinae and Agliinae. Also plants from a few other families found in the natural habitat of S. bouyeri were offered. All plants were collected in northern Kunming at elevations of ca. 1800–2000 m, using only fresh stems and leaves without pesticide contamination. Newly hatched larvae were offered plants the same day and sealed with just one plant species each in a plastic zip lock bag for observation. Feeding behaviour was recorded after 24 hours, and larvae that rejected their plant were switched to another plant species for testing. If feeding was confirmed, the lower parts of the plant’s stems were inserted into a bottle filled with water (gaps in the bottleneck were closed with tissue paper). Plants with larvae were placed in a separate cage to continue observations indoors or outdoors.

After feeding had ended (after the liquid defecation), larvae were placed in a container with 9–10 cm of loose and uniform peat soil at the bottom for pupation (peat had been produced naturally from Heilongjiang Province of China; ca. 0.35–0.47 g/cm³ with moisture and holes, pH 6.28–6.51, 20.8–24.3°C and 52.6–61.4% RH). The soil surface was covered with ca. 1–2 cm of Sphagnum moss (Sphagnaceae) and several fallen walnut leaves.

This work supplements here a circular sequence of the complete mitochondrial genome of S. bouyeri [GenBank: OR754233], its specimen data and high-throughput files have been uploaded to NCBI [BioProject: PRJNA905660].

DNA sequence data were generated from a fresh L₄ caterpillar of S. bouyeri [BioSample: SAMN31802199], which was killed and preserved in 95% (± 5%) ethanol at –18°C. DNA was extracted with chloroform-isoamyl alcohol and fragmented by ultrasonication. DNA library preparation (Illumina DNA Prep #1000000025416) included end repairing, 3’-end adenylation and ligation of sequencing adapter. Size selection was carried out on an agarose gel prior to amplification by PCR. After clean up and check, qualified libraries were sequenced on an ILLUMINA NovaSeq 6000. The resulting data [SRA: SRR22414765] were de novo assembled with SPAdes 3.15.4 (Prjibelski et al. 2020) using default parameters. The assembled mitochondrial genome had a length of 15,281 bp and was annotated using MITOS2 (Donath et al. 2019). Annotations were visualised as a mitochondrial genome map with OGDRAW (Greiner et al. 2019) and re-drawn as a linear plot in PHOTOSHOP BETA 25.0 (Fig. 1).

Phylogenetic trees (Fig. 15) were estimated from all codon positions of the 13 protein-coding mitochondrial genes [PCGs], namely COI, COII, ATP8, ATP6, COIII, ND3, ND5, ND4, ND4L, ND6, Cytb, ND1 and ND2. These PCGs were extracted from the complete mitochondrial genome of S. bouyeri and an additional 17 species with publicly available data (Jiang et al. 2009; Chen et al. 2014; Langley et al. 2020; Nethavhani et al. 2022; Kim et al. 2022). These additional sequences were from Asian-African Saturniidae, and Bombycidae as an outgroup (Table S6).

MAFFT V7 (Katoh et al. 2019) was used with default parameters to align the sequences. Maximum likelihood (ML) analyses with 1,000 bootstrap replicates were carried out in MEGA X (Kumar et al. 2018) using the recommended model GTR + G + I model, and ML analyses with 10,000 ultrafast bootstrap replicates (Hoang et al. 2018) in IQ-TREE 1.6.12 (Nguyen et al. 2015) using the best-fit model GTR + F + R3 (Kalyaanamoorthy et al. 2017). The IQ-TREE result (Fig. 15) was prepared in AFFINITY PHOTO 1.10.5.

Homology of morphological structures, chaetotaxal terminology and its abbreviations follow Liu (2023): L_1–6 = 1^st–6^th larval instars; T_1–3 = 1^st–3^rd thoracic segments; ­A_1–10 = 1^st–10^th abdominal segments; O = ocellar; SO = subocellar; F = frontal; AF = adfrontal; C = clypeal; G = genal; A = anterior; V = ventral; SV = subventral; L = lateral; M = medial; D = dorsal; SD = subdorsal; XD = tactile dorsal; MD = microdorsal; P = parietal; S = stemma; BaS = sensillum basiconicum; ChS = sensillum chaeticum; TrS = sensillum trichodeum; PlS = sensillum placodeum; CaS = sensillum campaniformium; DiS = sensillum digitiformium; StS = sensillum styloconicum; SP = sensory pore; -I: uni-setal, 1 seta; -II = bi-setal type, 2 setae; -III = multi-setal, more than 2 setae.

In the sections 3.1.1–3.1.7, the lengths, widths, heights, and quantitative statistics (e.g., the quantity of setae and crochets) are based on single individual observations, unless otherwise specified. The widths of head capsules (for each sample, it is the distance between the pair of S6) are derived from the optimal host plant from the tests (see section 4.2).

Host plant preference was systematically tested by restricting batches of larvae to just one plant at a time, and observations are presented by plant family in the following sub-sections. To reduce the risk of larval infectious diseases for larvae indoors, the rearing density had to be reduced. Therefore, some of the larvae from plants discussed in sub-sections 3.2.4, 3.2.5, 3.2.10, 3.2.14, 3.2.22 and 3.2.24 were placed onto wild trees. This also provided an opportunity to collect potential parasitoids in the wild of Yunnan. However, all of these caterpillars were missing within a week, most probably preyed upon or having left the plants by themselves.

Plants in the genus Zanthoxylum were found in the habitat of S. bouyeri in Mêdog, therefore, Zanthoxylum armatum was tested as a host for ten larvae. Only minimal bite marks were observed after two days, and 4 larvae had starved to death on the 3^rd day. The plant was finally considered to be rejected, and the remaining larvae were transferred to other plants for testing.

Mêdog is located on the southern side of the Himalayas, and was the main area where S. bouyeri was collected. At this locality, these moths were observed under lamps at altitudes of 1953–2150 m in valleys (Fig. 10). While no traps had been set at lower elevations, no adults of the species were trapped between 2845 m and 3188 m in late June 2021.

During surveys at 1300–2400 m, local Lhoba people confirmed that there was snow cover every year in December and January. A total of 6,707 outdoor climatic data points was collected in Mêdog at 2134 m during 19–26 June 2022 (Table S2). Temperatures ranged from 14.3 to 27°C (average: 18.4°C), while relative humidity ranged from 54.2 to 98.7% (average: 89.4% RH).

All data of times mentioned in the sections 3.4.1–3.4.4 are based on UTC+8.

Unhatched ova of S. bouyeri were stored in a plastic zip lock bag at room temperature (ca. 19–24°C) from 25 Jul. to 3 Sep. 2022. A dead minute wasp found in this bag is the only egg parasitoid recorded for the genus Sinobirma so far:

Single ♀ wasp (Fig. 9C): Head, thora, abdominal segments and coxae mainly shiny black; femur, trochanter, tibia and tarsus brown, mostly on the medial areas. Antenna near black, except for the anterior parts of the scape and pedicel, which are dark brown. Antennal flagellum 9-segmented, with the apical 5 flagellomeres forming an enlarged club. Median ocellus located anterior to lateral ocelli, the latter close to inner orbits. Forewing with clear costal, postmarginal and stigmal veins, but only the costal vein is visible in the marginal area of the hindwing.

The topologies of the ML trees calculated in both MEGA X and IQ-TREE were exactly the same. The topological relationship is (((Saturniini + Attacini) + (Micragonini + Eochroini)) + (Urotini + Bunaeini)) + outgroup (Fig. 15). All nodes within tribes have ultrafast bootstrap values of 100%, while nodes between tribes are only weakly supported with 80–90%.

At present, the putative sister genera of Sinobirma still lack microscopic morphological observations on ova for comparison with Sinobirma. The egg chorion of Pseudantheraea discrepans (Butler, 1878) shows a fine reticulation at 24x magnification (Seydel 1939), while Bouyer et al. (2004) described the ova of Pseudantheraea impe­rator Rougeot, 1962 as more than twice the size of those of P. discrepans, and figured the ova of the latter in color. Although images of the eggs of S. malaisei (Rougerie et al. 2012) and Tagoropsis flavinata (Walker, 1865) (Lampe 2010: 41) were published, all of them looked off-white in color and the shapes are very close to S. bouyeri, based on macroscopic views.

Similarly to eggs, very little is known about the early larval instars of the relatives of Sinobirma. The L₁ of S. malaisei appear almost indistinguishable from those of S. bouyeri as illustrated in Rougerie et al. (2012), except maybe by the absence of gray spots around scoli D of T₂–A₉ in the former (Rougerie et al. 2012). In contrast, the light gray spots are more conspicuous between pairs of scoli D-III in most examined L₁ caterpillars of S. bouyeri, though nearly impossible to observe in some specimens (e.g., Fig. 12C). Both species have very large and shiny prothoracic shields during L₁.

Rougerie and Estradel (2008) provided an important larval observation involving species within several putative or potential sister genera of Sinobirma, including Maltagorea vulpina (Butler, 1882), Maltagorea fusicolor (Mabille, 1879), Usta angulata Rothschild, 1895 and Parusta thelxinoe Fawcett, 1915. The L₁ of these four species were described as having 2 primary setae borne on each XD, which is a trait also clearly shown in each larval instar of S. bouyeri. Rougerie and Estradel (2008) also mentioned that the setal quantities of scoli SV of T_1–3 of M. fusicolor L₁ are “4-3-3”, which is the same as for the L_1–5 of S. bouyeri. Chalazae/Scoli SV were reported for T₁–A₉ of the larvae of Tagoropsis genoviefae Rougeot, 1950, P. discrepans and M. fusicolor (Rougerie and Estradel 2008), and these examples appear similar to S. bouyeri.

Lampe (2010: 41) and Staude et al. (2016: S37–S38) figured photographs of the larvae of T. flavinata. The L₁ larval black prothoracic shields and head capsule, and the yellow ground color of T₁–A₁₀ of S. bouyeri look very similar to the L₁ of T. flavinata. More importantly, the scoli D of T₂–A₉ appear to be morphologically similar in both species, being of an asterisk-like type, especially in the mature instar (L₅). They also share high bases that bear spiny setae outwardly elongated toward multiple directions. This shared character was also illustrated for the scoli D of T. genoviefae by Rougeot (1955: 40).

In the L₅ of T. flavinata, the sclerotized parts (mainly the head capsule, prothoracic shield, anal shield, legs T_1–3, lateral plates of prolegs A_3–6 and A₁₀, bases of scoli SV, L, SD, D and XD) are of similar colors near red-orange. In contrast, the same areas of fresh L₅ of S. bouyeri are all bright yellow just after ecdysis (Fig. 7I), but then turn into shiny black as the larva hardens. This suggests that the integumentary systems of the genera Tagoropsis and Sinobirma, especially their sclerites may have comparable biochemical characteristics.

The largest segments are always A_3–4, for S. bouyeri L_2–5, making the larva as a whole shaped cylindrical but slightly tend to fusiform. As mentioned above, when the larvae were in a physically relaxed state, the dorsal area of the junction zone between A₁/A₂ appeared to be the most sunken area in lateral view, especially in L_2–5 (Fig. 7C, E, G, J). A similar sunken junction zone between A₁/A₂ appears to be present in T. flavinata L_1–5. Based on illustrations in Lampe (2010) and Staude et al. (2016), mature larvae of Pseudaphelia apollinaris (Boisduval, 1847), Urota sinope (Westwood, 1849), Usta wallengrenii (C. & R. Felder, 1859) and Usta terpsichore (Maassen, 1885) also appear to have obvious depression of the junction zone between A₁/A₂ in lateral view. In addition, the macroscopic forms of T. flavinata L_1–5 also show intumescent A_3–4 similar to S. bouyeri.

The bases of the paired scoli D-III are medially fused on A₈ in S. bouyeri. This character is also present in T. flavinata as described by Packard (1914: 166) and in Tagoropsis hanningtoni (Butler, 1883) as figured by Seydel (1939). Furthermore, the original description (Rougeot 1950) of the larva of T. genoviefae explicitly noted the same fusion on dorsal A₈.

One noteworthy feature of S. bouyeri is the flattened A₁₀ in lateral view of L_1–5, with the posterior margins of the pair of lateral plates of the anal prolegs usually combined to form a minor arc outline in dorsal view. This means that the opening angle between the two anal prolegs is very flexible, whereas it is usually in a horizontal state (nearly straight line). This structure is presumably suitable for crawling and clustering on the smooth surfaces of the undersides of leaves.

Furthermore, the triangular anal shields of L_4–5 of S. bouyeri are all drawn out to form a pointed posterior tip, a feature shared with the mature larvae of T. genoviefae (Rougeot 1962: 70), T. hanningtoni (Lampe 2010: 42) and T. flavinata. The anal shields of these three species are triangular in shape, with a drawn-out posterior tip of relatively small size, while U. sinope (Staude et al. 2016: S37–S40) and M. auricolor (Rougerie 2005: fig. 449; Basquin 2015) appear to have a more reduced posterior tip of their A₁₀. A comparable structure is also present in P. apollinaris (Staude et al. 2016: S37–S38) and P. flavivitta (Walker, 1862) (Lampe 2010: 43), with their anal shields drawn out to form a very long spine pointing posteriad, more pronounced than in S. bouyeri. Oberprieler (1997) considered these similar anal shields potentially a distinctive feature shared by the group comprising Pselaphelia, Pseudaphelia, Tagoropsis and Urota.

For S. bouyeri, each scolus D-II on the anal shield bears 2 primary setae during L_1–5. A similar trait had been known to occur in some other members of the tribe Urotini, but also in other African tribes (Rougerie and Estradel 2008).

The pupal shell of S. bouyeri has three pairs of dorsal thoracic tubercles of large sizes, and a very long, spiny cremaster on the tip of A₁₀. Seydel (1939) described the pupa of T. hanningtoni as having a total of six tubercles borne on the dorsal area of thoracic segments and the pupa ending with a pointed cremaster, very much like the pupal painting of T. flavinata illustrated by Cooper and Cooper (2002: 49). The green pupa of P. discrepans figured by Rougeot (1955: Pl. 1), Gardiner (1982: Pl. 20) and Lampe (2010: 40) shows the same distribution of dorsal thoracic tubercles as S. bouyeri, and the former species has an elongated cremaster with hooks. The pupa of M. auricolor (Rougerie 2005: fig. 513) appears to have a similar spined tip on A₁₀, but attenuated to being relatively shorter than in S. bouyeri. Also M. fusicolor has an extended cremaster that was observed to have two terminal hooks (Bouvier 1936: 39–40: “Tagoropsis subocellata”). In fact, Oberprieler (1997) mentioned that both genera Tagoropsis and Pseudantheraea, as well as Pselaphelia, Pseudaphelia and Urota, all have comparable pupae based on their elongated cremaster. Furthermore, each of the spiracles of T₁ and A_2–8 of S. bouyeri is protected within a cavity under an elliptic annular cap. Rougerie and Estradel (2008) observed similar characters for the pupal spiracles of M. auricolor, Maltagorea monsarrati (Griveaud, 1968), M. fusicolor, T. genoviefae and P. discrepans.

While multiple potential sister genera share the aforementioned morphological features with S. bouyeri, differences in structural detail exist between them. For examples, images in the above literature show that scoli D of T₂–A₉ appear spine-like and strongly elongated in the mature larva of P. discrepans, with lateral margins of the anal shield curved. Larvae of M. auricolor and similarly M. fusicolor (illustrated by Paulian 1953 as “Copaxa subocellata”) have more reduced scoli D and SD on T₂–A₉ than S. bouyeri. Moreover, M. auricolor has only one pair of dorsal thoracic tubercles borne on pupal T₃, a feature that differs from Tagoropsis, Pseudantheraea and Sinobirma, which have a total of three pairs on T_1–3, respectively.

The trait of medially fused scoli D on A₈ has been known to be shared by larvae of Tagoropsis spp., S. malaisei and S. bouyeri, but M. auricolor and P. discrepans hatch with a pair of scoli D on A₈ that are completely separated from each other. Rougeot (1950) had suggested this to be an important character to distinguish the genera Tagoropsis and Pseudantheraea. Moreover, the mature larvae of P. apollinaris and P. flavivitta have reduced scoli D and SD, appear to have a triangular prothoracic shield with a drawn-out anterior tip, and extended lateral margins that form a wider shape in dorsal view. All of these features make these larvae resemble the Asian genus Salassa Moore, 1859, whose larvae are well-known examples of leaf-mimicing. However, the study by Rougerie (2005: 293) supports the hypothesis of evolutionary convergence between the two African genera and the Asian Salassa.

Mature larvae of U. wallengrenii and U. terpsichore do not have an elongated posterior tip of the anal shield, and the pupa of the former species has a posterior elevated crest in annular shape on A_4–6. These characteristics also appear in the pupa of U. angulata (Rougerie and Estradel 2008). Nässig et al. (2015) suggested that the genus Usta may have a more distant relationship to the core group containing Sinobirma, while Rougerie et al. (preprint) assigned both Parusta and Usta to the tribe Eochroini with strong support. This is contrary to Pinhey (1956: 26), who considered the genus Parusta to be potentially closely related to Pseudaphelia. The genera Eudaemonia and Antistathmoptera are not considered in this section due to obvious adult differences to Sinobirma (e.g., Oberprieler 1997).

Although the color combinations of the final larval instar of the above related African genera do not have similar patterns to S. bouyeri, coincidentally, the colors of the mature larva of the Australian Opodiphthera astrophela (Walker, 1855) are similar, especially in the painting in Scott (1890). Similarly, mature larvae of Saturnia centralis Naumann and Löffler, 2005 are also superficially similar to S. bouyeri (Liu and Peigler 2021), because integuments of both species are black grounds studded with green patches and stripes. They also share obvious aggregation behavior during larval instars. If Felix Bryk had encountered the mature larva of Sinobirma 80 years ago, he might have felt certain of the “close relationship” of Sinobirma to these other genera. However, such color patterns (black or dark brown grounds, with vivid green stripes or patches on especially the lateral areas) occur in different genera or even subfamilies of Saturniidae, representing most probably examples of convergent evolution.

Despite all the similarities and differences, further data from African species would be needed to identify which characters are informative to reconstruct a morphology-based phylogeny. At present, our knowledge of the above genera is incomplete, and especially detailed morphological and biological work involving ova, larvae and pupae are rare. Nevertheless, Nässig and Oberprieler (1994) and Rougerie et al. (preprint) postulated that Tagoropsis, Pseudantheraea, Maltagorea and Sinobirma form a closely related group with similar morphology and genes. The preimaginal stages of S. bouyeri further support this hypothesis. Out of these genera, the larvae of Sinobirma are morphologically most similar to Tagoropsis, especially due to their highly similar scoli D (form and fusion) and anal shield. Likewise, pupae of the two genera are more similar to each other than compared to other genera. Although Cooper (2002: xiv) explicitly classified these four genera in the new tribe “Tagoropsini” and excluded Urota, the larval characteristics combined with adult morphology indicate that Urota is more likely to be the sister taxon of the above core group (“Tagoropsini”). Unlike Tagoropsis, Pseudantheraea, Maltagorea and Sinobirma, the larva of U. sinope looks stubbier, rather than an elongated cylinder, which is caused by the middorsal areas of T₂ and T₃ being swollen and elevated. Furthermore, the scoli D of T₂–A₉ of U. sinope are noticeably reduced to rounded, pimple-like structures.

Exploring the plants that Sinobirma accepted or rejected can provide insights into its ecological niche. Previously, only Rougerie et al. (2012) reported that S. malaisei L₁ failed to rear on Prunus cerasifera (Rosaceae). In addition, Rougerie (2003) mentioned that at the biotope of S. malaisei in western Yunnan, the vegetation including many flowering Castanopsis (Fagaceae).

This study tested host plant acceptance by S. bouyeri systematically. Plants completely rejected by the larvae of S. bouyeri came from the families Ericaceae, Malvaceae, Lauraceae, Betulaceae, Magnoliaceae, Pinaceae, Vitaceae and Santalaceae. Similarly, there were minute bite marks, but the larvae did not continue to feed and even starved to death when attempted to rear on Meliaceae, Rutaceae and Theaceae.

Altingiaceae seems to be a special case where within a group some larvae clearly accepted but some totally rejected the offered host plant, thereby separating two dinstict larval clusters in the rearing container. Ultimately, none of the larvae developed on this host. Similarly, S. bouyeri rejected two but fed one species within the family Fabaceae, but the accepted plant caused the caterpillars to die consecutively over several days. The same acceptance yet mortality pattern occurred also with Anacardiaceae, Fagaceae, Nyssaceae, Phyllanthaceae and Oleaceae. Even more problematic was the rearing on Sapindaceae and Ulmaceae, which resulted in all individuals of each larval group to die rapidly the same day.

Regrettably, no valid data were available for the group testing on Symplocaceae and it remains unknown whether the larvae of S. bouyeri can molt to L₂ or even complete the whole larval stage. This family of plants is rarely reported as a host plant of saturniids, e.g., A. atlas was reported to feed on Symplocos paniculata in India (Rondot 1887: 68: “Symplocos crataegoides”), and Symplocos tinctoria is a hostplant for Callosamia promethea (Drury, 1773) in South Carolina (Ferguson 1972: 236–237).

Of all the tested host plants, only 5 families of plants were able to support the development of S. bouyeri larvae from L₁ molting into L₂. The least suitable plant was J. regia, as all larvae died soon after the first ecdysis. This result is consistent with the mortality of S. malaisei observed by Rougerie et al. 2012. Larvae feeding on L. indica had shown signs that this plant wasn’t the right host, i.e., the slow growth of L₂ larvae and increased size differences between individuals. Yet, it cannot be ruled out fully that some individuals of S. bouyeri feeding on this plant would have survived to pupation. Salix spp. were observed to be cultivated along the country roads in Nyingchi, and there were native plants of this genus growing in the habitat of S. bouyeri in Mêdog. However, this plant didn’t appear to be a correct host in the natural environments and was also unsuitable for captive rearing, although a fair number of willow-feeding larvae can survive to maturity. Larval feeding on the family Rosaceae was equivocal, with individuals on wild rose revealing continuous mortality over multiple days during L_1–2, while the group fed on P. cerasoides completed all larval instars.

Of all the plants tested, C. nepalensis in the family Coriariaceae is the best host plant for S. bouyeri. There were no larval losses from hatching to the last larva burrowing into the soil, and the developmental duration was almost a month shorter than that of larvae fed on P. cerasoides (Table S1). As a shrub (Fig. 9A), C. nepalensis is widely distributed in the humid regions of southern provinces of China and the Sub-Himalayas (Min and Brach 2008). Host testing showed that most of the plants rejected by S. bouyeri were tall trees in nature. The dark integument of the mature larvae and its feeding habit at night indicate that S. bouyeri inhabits the understory, and the true host plants in this biotope may be shrubs and vines. C. nepalensis would fit such requirements.

Of the genera that are closely related to Sinobirma, Tagoropsis occurs only on the mainland of Sub-Saharan Africa and was long known to feed on Sapindaceae. An early report by Schultze (1914) noted that T. flavinata fed on Allophylus africanus in the wild in Cameroon. In South Africa, larvae of T. flavinata were recorded as feeding on the same plant (Platt 1921: “Schmidelia africana”; Pinhey 1972: 54: “Allophylus melanocarpus”). More recently, Allophylus natalensis and Allophylus dregeanus were listed as host plants of T. flavinata (Cooper and Cooper 2002: 48; Staude et al. 2016: S37). Working in then French Congo, Seydel (1939) published a record of A. africanus used to rear T. hanningtoni, for which Crotch (1956: 148) also suggested “might be tried on maple and sycamore”. Santin (2004) listed Malus (Rosaceae) for T. hanningtoni, while Meister (2011: 131) added Bauhinia (Fabaceae) for the same moth, with the records of A. africanus for T. genoviefae, T. hecqui Bouyer, 1989, T. rougeoti Fletcher, 1968 and T. sabulosa Rothschild, 1907.

Pseudantheraea is a polyphagous genus occurring in central-western Africa, feeding on multiple botanical families in the wild. Poga oleosa (Anisophylleaceae) and Uapaca guineenis (Phyllanthaceae) were recorded as natural hosts of P. discrepans in Gabon, while the genus Terminalia (Combretaceae) was accepted in captivity (Rougeot 1949; Rougeot 1962: 75). Moreover, Gardiner (1982: 199) listed Mangifera indica (Anacardiaceae) for the same species. Salix caprea (Salicaceae) was observed to be accepted by P. discrepans (Bouyer et al. 2004). In the captivity in Germany, the larvae of P. discrepans were further listed as feeding on Caloncoba welwitschii (Achariaceae), Coelocaryon botryoides (Myristicaceae), Ricinodendron heudelotii (Euphorbiaceae), S. babylonica and Fagus (Fagaceae), whereas its sister species P. imperator possibly accepted the genera Fagus, Mangifera, Terminalia and Uapaca (Meister 2011: 130). P. discrepans has been reported as an edible insects in Angola (Lautenschläger et al. 2017) and Congo, and people in the latter country recently supplemented the list of accepted host plants with Entandrophragma candollei (Meliaceae), Spondias dulcis (Anacardiaceae), Spondias mombin (Anacardiaceae) and Staudtia kamerunensis (Myristicaceae) (Mabossy-Mobouna et al. 2022).

Maltagorea is restricted to Madagascar and lacks detailed preimaginal reports. Larvae of M. fusicolor have been known to feed in the wild on the tapia tree Uapaca bojeri (Phyllanthaceae) (Oberthür 1916: “Syntherata cambouei”, “Tapia Edulis (ou Chrysopia sp.?) [sic]”), Euphorbiaceae (Paulian 1953) and Agauria (Ericaceae) (Griveaud 1961: 26: “Tagoropsis subocellata madagascariensis”); Meister (2011: 129) also listed Aguiaria excelsa (Malvaceae) as a host plant. M. auricolor feeds naturally on U. bojeri (Basquin 2015).

As mentioned above, a further seven African genera Urota, Pseudaphelia, Pselaphelia, Eudaemonia, Antistathmoptera, Usta and Parusta have been considered as relatives of Sinobirma to varying degrees in different works. According to the comprehensive catalogues of the host plants of Saturniidae by Stone (1991), Rougerie (2005) and Meister (2011), the above African genera primarily exploit the families Meliaceae, Zingiberaceae, Lauraceae, Combretaceae, Anacardiaceae, Fabaceae, Myrtaceae and Burseraceae.

More rigorous comparative conclusions could not be drawn as it wasn’t possible to test the host plants of all of the above African Saturniidae. Furthermore, it is uncertain whether records in literature resulted in larvae reaching pupation or could even sustain multiple generations. However, polyphagy is here proposed to be a shared trait of the closely related genera Tagoropsis, Pseudantheraea, Maltagorea and Sinobirma.

The natural habitat of S. bouyeri in China is recorded here as the Nyingchi area of the Tibet Autonomous Region, a prefecture-level city whose territory contains both northern and southern sides of the eastern Himalayas and the Yarlung Tsangpo River [upper Brahmaputra]. Bomê [Pome] and Mêdog [Metok] are both counties that belong to Nyingchi, as the type locality. The average annual rainfall of the former region exceeds 800 mm, largely in spring (>300 mm) and summer (>300 mm), with an average temperature of 0–2°C in winter and 16–18°C in summer (Song and Wang 2013: 47, 61–62).

Zheng et al. (2018: 4) reported the average annual rainfall of Mêdog at 2000 m as more than 2200 mm, and average temperatures of 4–6°C in winter and 17–19°C in summer. The area is wet and cloudy, covered with subtropical evergreen broad-leaf forest, which is classified as the notophyllous type in Wolfe (1979: Pl. 1). In vegetation surveys of Tibet, this area was detailed and classified as “Mêdog district, eastern Himalayan tropical evergreen rainforest province, southeastern Asian tropical evergreen or seasonal rainforest region, tropical vegetation zone” by Zhang et al. (1988), with the eco-region named “eastern Himalayan broadleaf forests” of Indomalayan biomes (e.g., Hoekstra et al. 2010: 188–189).

The genera Tagoropsis, Pseudantheraea and Maltagorea are mainly forest and grassland dwellers (Fig. 17), with the predominant climates of their habitats being principally tropical rainforest, monsoon and savanna, but also small areas with oceanic and humid subtropical climates (Peel et al. 2007). In contrast, Sinobirma only adapted to the last type. Both Tagoropsis and Maltagorea tend to be distributed in the highlands of the western and central Somali Plate, while Sinobirma is a strictly montane genus which endemics to the northeastern Indian Plate. It suggests that relatively humid climates might have been necessary for the dispersal and distribution of the most recent common ancestor of this core group.

Based on the oviposition behavior described in section 3.4.1, females of S. bouyeri may prefer to oviposit in the wild onto surfaces facing the sky (such as the uppersides of leaves or branches), as well as into narrow crevices or pits (such as cracks of stems). This would provide the eggs more exposure to rain and warmth within the understory that’s low in sunlight.

Comparing observations made for the L_4–5S. bouyeri larval groups reared on C. nepalensis and P. cerasoides (section 3.4.3), it appears that larval cluster size and location on the plant have an influence on the gregarious behaviour. For smaller larvae that inhabited the leaf undersides during the day, the more larvae in a cluster, the more stable the rhythmical feeding-resting behavior was. In contrast, larger larvae that typically clustered on lower stems during the day exhibited the most stable gregarious behaviour. These results are based on only a relatively small number of individuals and observations, and more rigorous experiments are needed in the future to further determine the regulatory role of other factors like light and pheromones in the circadian rhythm of larvae.

Pupation of the mature larvae in captivity was described in detail in section 3.4.4. In nature, soil tends to be substantially more dense and to include more rock particles and plant roots. Therefore, most larval S. bouyeri might only burrow to a soil depth of less than 10 cm naturally, but otherwise similarly to the results in section 3.4.4.

Peigler (1999) suggested that pupae and cocoons formed below ground may escape fires. Located at the windward side of the eastern Himalayas, forest floors in Mêdog often experiences accumulation of debris caused by heavy summer and autumn rainfalls. In captivity, larvae had not been observed to reinforce their tunnels and pupal chambers with silk or other measures. Therefore, most pupae in the wild are likely to be completely buried below ground. The pupae of S. bouyeri have a long cremaster and spine-like abdominal and head tubercles, with the abdominal tubercles pointing latero-caudad. Presumably, waggling the flexible abdominal segments (A_5–7) results in rebuilding the pupal chamber and forward movement when the moth emerges. The elliptic annular cap of each spiracle may prevent clogging of the spiracles with soil, especially while pressing against soil for movement.

Rougerie (2003) provided a first biological description of the genus Sinobirma in western Yunnan: “Their flight times were remarkably constant, the females arriving at the sheet at about 21:00 and the males between 23:30 and 00:00 local time… S. malaisei arrived at the light in an erratic fashion, fluttering around on the ground before settling on the sheet or surrounding shrubs”. These observations are essentially consistent with the behavior of S. bouyeri.

Cooper and Cooper (2002: 48) described T. flavinata in Africa as a “rather weak flier”, whose females are readily attracted to car headlights. Likewise, both P. discrepans and P. imperator are attracted to lights during the night, and statistical data were recorded in detail by Bouyer er al. (2004). Maltagorea is also nocturnal, but the flight times of different species within the genus vary significantly (Basquin and Rougerie 2009).

The above publications indicate that some species of the African genera Tagoropsis, Pseudantheraea and Maltagorea have two (or more) flights per year. In contrast, Asian Sinobirma is strictly univoltine (see introduction), which is probably an adaptation to the colder Himalayas, but may also correspond to seasonal metabolic rhythms of their natural host plants.

Seydel (1939) described the larva of T. hanningtoni as being processionary like Thaumetopoeinae (Notodontidae) and pupating in soil without cocoon. Rougeot (1950) recorded the larvae of T. genoviefae as feeding on a shrub called “Evetom [sic]” during the night. He reported that when the larvae stopped feeding, they left the food plant and rested on the ground with help of a few silken threads for pupation, sheltered by leaves or plant fragments. Cooper and Cooper (2002: 48) noted ova of T. flavinata having been “laid in arching rows of eggs glued sideways on twigs in up to 7 tiers and the underside of leaves of the larval foodplant”. They also reported larvae being “highly gregarious at all instars, moving quickly in single file, head-to-tail, from one position to the next” and “Although the bristles have been said to sting, they have had no effect on us whatsoever... Larvae feed mainly at night spending the day resting in a clump, often low down, on a thick stem of the larval foodplant” and pupated “among leaf litter on the ground”. Nocturnal feeding by larval T. flavinata had also been noted already earlier by Bouvier (1928: 531).

P. discrepans has long been known for its peculiar green pupa hanging in a loose cocoon amongst the vegetation (e.g., Holland 1892: “Saturnia arnobia”). Bouyer et al. (2004) added that “Larval behavior was characterized by an intensively gregarious habit in all instars. Feeding and resting were nearly always carried out synchronously and in tactile contact. Another unexpected observation was that the larvae always left their feeding site in the late afternoon (around 17 h Central European Summer Time) for approximately one hour and congregated at the neck of the bottle in which the branches were kept”.

Oberthür (1916) reported that the pupa of M. fusicolor was surrounded by debris and rested between leaves, and Paulian (1953) added that it was naked, protected by loose silk to link together fragments of dry leaves. Bouvier (1936: 40) suggested that such an unformed cocoon revealed its relationship to the genus Pseudantheraea. Similarly, Rougerie (2005: fig. 513) figured the pupa of M. auricolor, also being protected in a loose cocoon, and Basquin (2015) stated that the species pupated in litter at the base of trees. In contrast, the genus Maltagorea still lacks any description of larval behavior.

Considering the above larval characteristics, Tagoropsis is the closest genus to Sinobirma at the behavioral level. Some species of the two genera are known to have similar larval circadian rhythm and naked, subterranean pupae.

Larval fluorescence has rarely been reported for Saturniidae. Besides S. bouyeri, Adès (2007) illustrated larval fluorescence of L₅ of Actias sinensis (Walker, 1855), Actias felicis (Oberthür, 1896) and hybrid A. sinensis ♂ × Actias dubernardi (Oberthür, 1897) ♀. Fluorescence occurred primarily at the bases of scoli, with excitation by UV light (peak wavelength 404 nm). Similarly, larvae of Syssphinx albolineata (Grote and Robinson, 1866) exhibited fluorescence under UV light, visible on scoli, prothoracic and anal shields, and especially the minute tubercles of the integument (Wagner and Nall 2022). Mature caterpillars of Citheronia regalis (Fabricius, 1793), Eacles imperialis (Drury, 1773), Antheraea polyphemus (Cramer, 1776), Antheraea compta Rothschild, 1899, Actias luna (Linnaeus, 1758) and Hyalophora cecropia (Linnaeus, 1758) have been shown to fluoresce in whole or in part under UV light (Moskowitz 2018; 2021; Liu 2023).

The fluorescent green stripes of S. bouyeri L₅ were conspicuous in daylight, which seemed to provide excellent camouflage in vegetation when larvae clustered at the inferior parts of the stems during the day — due to the rainy climate, Sub-Himalayan broad-leaved forest is extensively covered with moss, especially the understory.

Using a taxonomic key for the order Hymenoptera (Goulet and Huber 1993), the single saturniid parasitoid wasp in section 3.5 was recognised as belonging to the subfamily Telenominae (Platygastridae). More specifically, the specimen might belong to the genus Telenomus Haliday, 1833, which is well-known to parasitise Lepidoptera, and a few genera in this subfamily have been deployed as biocontrol agents (Austin et al. 2005). With parasitic behavior being unlikely to occur indoors (Kunming), this species is most likely a natural parasitoid of S. bouyeri in its Himalayan habitat.

Nethavhani et al. (2022) published a molecular phylogeny based on the 13 mitochondrial protein-coding genes. The molecular results of the present study detailed in section 3.6 strongly support congruent topologies of the Maximum Likelihood and Bayesian analyses, except for the addition of Urotini in this study (Fig. 16). The new phylogeny supports an African origin of Sinobirma, because the tribal clade represented by S. bouyeri is sister to Bunaeini — an entirely Afro-Madagascan taxon with high genetic diversity (Kitching et al. 2018). Unfortunately, the genera Tagoropsis, Pseudantheraea or Maltagorea have no complete mitochondrial genomes publicly available at present, and comparative analyses within this group aren’t possible yet.

Regier et al. (2008) published a Maximum Likelihood tree based on four protein-coding nuclear gene regions, which demonstrated that the tribe “Urotini” in the traditional sense was polyphyletic. with the “urotine” Tagoropsis closest to Bunaeini and the “urotine” Usta to Micragonini. Based on the analysis of ultraconserved elements [UCEs], Rougerie et al. (preprint) classified Usta as belonging to the re-instated tribe Eochroini, with a restricted, monophyletic tribe Urotini comprising only (((Pseudantheraea + Maltagorea) + Sinobirma) + (Tagoropsis + “Tagoropsiella Darge, 2008”)) + Urota.

However, the topologies in these papers (Fig. 16B, C) are congruent at the tribal level, with Bunaeini always closest to Urotini, and together sister to Micragonini + Eochroini, which in turn are sister to Saturniini + Attacini. The topologies based on mitochondrial data, both in Nethavhani et al. (2022) and this study, also support Bunaeini + Urotini, as well as the sistergroup relationships between each Micragonini + Eochroini and Saturniini + Attacini. However, both analyses of miochondrial data place Micragonini + Eochroini as sister to Saturniini + Attacini, albeit with low support, while analyses of nuclear data support Micragonini + Eochroini as sister to Bunaeini + Urotini.

At first glance, the inclusion of the Himalayan genus Sinobirma in the otherwise exclusively African tribe Urotini might seem at odds from a biogeographic perspective and require discussion. Following the phylogeny of Nässig and Oberprieler (1994) and considering the discussions in sections 4.1–4.6, the genera Tagoropsis, Pseudantheraea, Maltagorea and Sinobirma are here considered to form a monophyletic group. The taxonomic status of Tagoropsiella, which occurs only on the African mainland and is close to Tagoropsis (Table S5), is ambiguous, and the genus is possibly a subjective synonym of Tagoropsis. Whether or not to distinguish this taxon does not affect the biogeographic conclusions regarding Sinobirma, which is why Tagoropsiella is here regarded as a supplementary group close to Tagoropsis. The currrent distribution of the genus Sinobirma could be explained by the following hypotheses, which are based on the distributional conditions of a most recent common ancestor of this generic group. In contrast, their current geographic distributions is not necessarily indicative of evolutionary relationships:

Hypothesis I. A Gondwanan origin of the group’s ancestor, prior to India separating from Madagascar (land connection or dispersal across narrow straits still possible). After complete separation of India (straits too wide for dispersal), the ancestor present in India developed into a distinct lineage (Sinobirma) that arrived in Asia, whereas the individuals in Africa split into lineages on the mainland (Tagoropsis and Pseudantheraea) and Madagascar (Maltagorea).

Hypothesis II. After India had separated completely from Madagascar, the ancestor of Sinobirma originated from taxa in mainland Africa (Tagoropsis and Pseudantheraea) or Madagascar (Maltagorea), followed by expansion to Asia (Sinobirma) and extinction in Africa.

Hypothesis III. The common ancestor originated on the Eurasian Plate, or the insular India (completely separated from Madagascar), or formed after the two land masses collided, giving rise to the Asian lineage (Sinobirma) and dispersing to mainland Africa (Tagoropsis and Pseudantheraea) and finally Madagascar (Maltagorea).

Nässig and Oberprieler (1994) considered certain similarity between the genital structures of S. malaisei and M. auricolor a potential synapomorphy. The genital structures of Saturniidae are not strongly influenced by environmental factors (climates, vegetations, natural enemies, etc.), and Sinobirma might have reached Asia through northern Africa or insular India, while developing relatively little change in genital structures, as in some members of the Madagascan Maltagorea. Consequently, these similarities between the two genera are plesiomorphic traits retained by chance. This would fit to both hypotheses I and II.

Immature stages of the family Saturniidae make up the longest part of the lifecycle, and larval habitus and behavioral features are particularly influenced by evolutionary pressures linked to the biotopes they occur in. Firstly, the immatures of Pseudantheraea and Maltagorea each of features that are unique within the group of genera, while Tagoropsis and Sinobirma share fairly visible, similar characteristics, i.e., in morphology and biology of larvae and pupae. These shared features are unlikely to have evolved convergently, and the sharing of homologous characteristics indicates that the historical habitat of Sinobirma was probably more similar to the one of Tagoropsis.

Secondly, both Pseudantheraea and Tagoropsis occur on the African continent in adjacent or partially overlapping habitats (Fig. 17), yet, their dominating adaptive characteristics of immatures are divergent. With the evolutionary diversification of Maltagorea on the island of Madagascar in mind, it seems reasonable to assume a similar diversification for Sinobirma if it had drifted on an insular India. One would expect more autapomorphies in the immature stages, which is not case as described in this paper. Consequently, these immature characteristics are only in agreement with one evolutionary hypothesis, namely hypothesis II.

Considering the extensive distribution of Urotini and its sister tribe Bunaeini in Africa (see section 4.6), these African taxa and Sinobirma seem unlikely to be of shared Asian origin, rendering hypothesis III unrealistic. The Himalayas have been central to the evolution of most Asian saturniid genera/subgenera except Perisomena Walker, 1855, i.e., the genera Actias Leach, 1815, Antheraea Hübner, 1819, Telea Hübner, 1819, Antheraeopsis Wood-Mason, 1886, Cricula Walker, 1855, Lemaireia Nässig & Holloway, 1988, Rinaca Walker, 1855, Cachosaturnia Naumann, Löffler & Nässig, 2012, Saturnia Schrank, 1802, Neoris Moore, 1862, Loepa Moore, 1859, Attacus Linnaeus, 1767, Archaeoattacus Watson, 1914, Samia Hübner, 1819, Rhodinia Staudinger, 1892, Solus Watson, 1913, Salassa and Aglia Ochsenheimer, 1810. These genera have all a wider subtropical and tropical Asian, temperate palearctic or even New World distribution. In contrast, Sinobirma is a relict taxon with a distribution restricted to the northeastern corner of the Indian plate.