The Hercules pseudoscorpions from Madagascar: A systematic study of Feaellidae (Pseudoscorpiones: Feaelloidea) highlights regional endemism and diversity in one of the “hottest” biodiversity hotspots

pseudoscorpions from Madagascar: A systematic study of Feaellidae (Pseudoscorpiones: Feaelloidea) highlights regional endemism and diversity in one of the "hottest" biodiversity hotspots. Arthropod Systematics & Phylogeny


Introduction
Madagascar is one of the world's priority conservation hotspots and amongst the most biodiverse countries with extreme levels of endemism and high species diversity across rich vegetation zones, including tropical rainforest in the east, spiny desert in the south and dry deciduous rainforest in the west (Mittermeier et al. 2004). Sadly, more than 90% of original habitats are already lost and major habitat types are irreversibly damaged (Ganzhorn et al. 2008). Madagascar's high levels of biodiversity and endemism are correlated with landscape heterogeneity, bioclimatic zonation and a long history of isolation that includes rifting from West Gondwana (Africa and South America) beginning as early as 166 Ma and ending by 116 Ma, and from India-Seychelles 85-90 Ma, during the break-up of Gondwana (Ali and Aitchison 2008;Vences et al. 2009). Ecological disparities between the arid west and humid east, montane and lowland rainforest refugia, and habitats separated by canyon and riverine barriers have driven adaptive trait diversification among the island's fauna and flora, allowing species to occupy new niches as they emerged (Wood et al. 2015). In addition to ongoing deforestation and overharvesting, anthropogenic climate change is increasingly threatening the remaining natural habitats and accelerating biodiversity loss in this natural laboratory (Morelli et al. 2020).
In the past two decades, there has been major progress in documenting the Malagasy invertebrate fauna at the taxonomic level, including many arachnid lineages (e.g., Griswold et al. 2012;Saucedo et al. 2015;Wood and Scharff 2018). Several studies have addressed patterns and causes of invertebrate speciation (e.g., Miraldo et al. 2011;Wesener et al. 2011;Agnarsson et al. 2015), but no data are presently available for pseudoscorpions, a mesodiverse arachnid lineage, following the more diverse spiders and mites, which are found in all ground habitats on Madagascar (WPC 2022). No detailed taxonomic studies have ever been conducted on Malagasy pseudoscorpions and 15 species are presently recognized in eight families -a relatively small number compared to regions such as Kenya (133 species) or South Africa (ca. 140 species). In a brief contribution, Heurtault (1986) listed eight of the presently 25 recognized pseudoscorpion families on Madagascar and identified twelve putatively endemic genera in these families but also wrote that "Les Pseudoscorpions (…) ne présentent pas de cas "d'explosion" ou de "pulverisation" spécifique" ("The pseudoscorpions do not represent a case of explosive speciation") in contrast to many other faunal groups. She was also amongst the first to note the presence of the pseudoscorpion family Feaellidae Ellingsen, 1906 on Madagascar and mapped a record from Toliara Province but did not describe any species, however.
Feaellid pseudoscorpions have a unique morphology that includes a dorsoventrally compressed body, a thick and granulate cuticle, the presence of 2-6 lobes on the anterior margin of the carapace, and raptorial pedipalps with heavily armed and robust pedipalpal femora, resulting in a somewhat hulky appearance, hence the name "Hercules pseudoscorpions" that is introduced here to symbolize this morphology for non-arachnologists. Together with their sister-family, the Pseudogarypidae Chamberlin, 1923, this family constitutes one of the two basal suborders of pseudoscorpions, Atoposphyronida Harvey 2019, which is characterized by the unique presence of carapaceal lobes, spiracles near sternites IV and V, a heavily sclerotized anal plate, and the absence of venom glands in the chelal fingers (Harvey 1992;Benavides et al. 2019). Atoposphyronida may have originated in the Permian or even Carboniferous (Benavides et al. 2019). Today, the Feaellidae are restricted to the Mediterranean and tropical biomes of former Gondwanan landmasses ( Fig. 1), although compression and amber fossils from the Triassic and Paleogene of Europe highlight a once wide distribution (Henderickx andBoone 2014, Kolesnikov et al. 2022). The taxonomy of Feaellidae is still in flux and there is no recent generic revision but six genera and subgenera are presently recognized in two extant subfamilies. Cybellinae Judson, 2017 includes a single genus, Cybella Judson, 2017 with four species from subterranean habitats in the Southeast Asian countries of Cambodia, Malaysia and Vietnam (Judson 2017; Har-  vey 2018). The second subfamily, Feaellinae Ellingsen, 1906, includes the monotypic Iporangella Harvey et al., 2016 from the Brazilian Atlantic rainforests, andFeaella Ellingsen, 1906, which is divided into three subgenera based on the number of carapaceal lobes: Feaella (Feaella) Beier, 1955 with three species from tropical Africa (Guinea-Bisseau, Côte d'Ivoire, Congo and Kenya) having six carapaceal lobes; Feaella (Tetrafeaella) Murthy and Ananthkarishnan, 1977 with twelve Recent species from southern Africa, India, Sri Lanka, the Seychelles, Maldives, Western Australia and a Baltic amber fossil (Harvey et al. 2016a(Harvey et al. , 2016bNovák et al. 2020;Harvey 2022); and Feaella (Difeaella) Beier, 1966 with a single species from Kruger National Park, South Africa (Beier 1966). Although all species in the subgenus Feaella (Tetrafeaella) have four carapaceal lobes, it is likely that this subgenus is polyphyletic as it includes species from diverse climates (Mediterranean, tropical) and habitats (e.g., rocky outcrops in Western Australia, savannah and coastal habitats in continental Africa ;Beier 1955;Harvey et al. 2016a), and with diverse morphologies.
Before the onset of phylogenetics, most feaellid species were considered widespread, e.g., Feaella (Feaella) mirabilis Ellingsen, 1906 across western and central Africa (Heurtault-Rossi and Jézéquel 1965; or Feaella (Tetrafeaella) indica (Chamberlin, 1931) across Bangladesh, India and Sri Lanka (Chamberlin 1931;Batuwita and Benjamin 2014). However, recent molecular studies have indicated extreme local endemism in a morphologically cryptic fauna with slow rates of evolution, implying that many species hypotheses will have to be revised (Harvey et al. 2016a;Novák et al. 2020). In contrast to the historical literature, Feaellidae may in fact turn out to be a highly suitable group for vicariance biogeography because they seem to disperse poorly and occur almost exclusively on continental landmasses despite a record from the Maldives that might indicate trans-oceanic dispersal (Novák et al. 2020). Certainly, the present distribution is indicative of diversification from continental drift and the Late Cretaceous (Cenomanian: 99 Ma), Burmese amber fossil, Protofeaella peetersae Henderickx and Boone, 2014 shares prime synapomorphies of feaellids, providing unequivocal evidence for an ancient radiation beyond age estimates provided by molecular clock analyses (Henderickx and Boone 2016).
The presence of feaellids in Madagascar went largely unnoticed even in the arachnological community apart from sporadic records (Vachon 1960;Heurtault 1986) but comprehensive biodiversity surveys conducted by Brian L. Fisher and his team have established records of Hercules pseudoscorpions from all of Madagascar's drier vegetation zones but not from eastern forests. Unlike other regions of the world where feaellids are highly elusive and rarely collected, Hercules pseudoscorpions are amongst the more abundant pseudoscorpions in sclerophyll and savannah habitats and have been frequently collected in pitfall traps, often in large numbers. However, occurrences seem to be localized and strictly tied to specific microhabitats or localities. Following our detailed study, the Malagasy fauna also falls into distinct morphological clades that show strict regional zonation: a northern clade (in the former Antsiranana Province), a western clade (in the former Mahajanga and Toliara Provinces), and a southern clade (in the former Toliara Province) (Fig. 2). These three clades share the presence of four carapaceal lobes and hence resemble Feaella (Tetrafeaella) rather than Feaella (Feaella) or Feaella (Difeaella). However, considering that each of these three clades present unique diagnostic characters, are genetically well-differentiated and correspond to specific regions of Madagascar, they are each recognized here as new genera.
This study aims to: (1) document the Malagasy fauna of Hercules pseudoscorpions both at the generic and species level across dry and subarid bioclimatic zones in Madagascar; (2) provide the first comprehensive monograph of a Malagasy pseudoscorpion lineage to date; (3) use taxonomic data to discuss endemism and distributional patterns with regards to hotspot biogeography of other invertebrates; and (4) discuss potential reasons for the unprecedented levels of feaellid diversity in Madagascar.

Morphology
All specimens were collected between 2002-2006 and are deposited in the California Academy of Sciences (CAS), San Francisco and the Museum of Nature Hamburg -Zoology (formerly Zoological Museum Hamburg (ZMH)). Specimens sequenced as outgroups are lodged in the Western Australian Museum (WAM), Perth. Most Malagasy specimens were collected by Dr. Brian Fisher during extensive surveys of Malagasy invertebrates and preserved in 75% ethanol. Specimens were sexed, identified and sorted using a Leica M125C stereomicroscope. Dissected parts were kept in microvials. Measurements (in mm) were taken with a Leica M205A stereomicroscope and Leica Application Suite X Version 3.0.1. Digital images were taken with a custom-made BK Plus Lab System by Dun, Inc. using a Canon EOS 7D Mark II camera, an attached 5×-magnification microscopic lens in a P-51 CamLift-System installation controlled by P-51 Camlift Controller ver. 2.8.0.0 and by Capture One ver. 9.3. Scanning electron images were taken from temporarily mounted specimens using a Hitachi TM4000Plus scanning electron micrograph. Since many specimens were coated with a thick layer of dirt or particles, ultrasound cleaning was used to improve imaging results, however, success was limited. Drawings were made by hand using original images that were checked against primary specimens after illustration. Maps were created using QGIS Version 3.0 (https://www.qgis.org) and coordinates taken from original labels. Images, trees and maps were edited with Adobe Photoshop Version CC 2017.

DNA Extraction, Sequencing and Taxon Sampling
DNA extraction and Sanger sequencing was trialed for representative specimens (one per locality) but because all specimens were collected almost two decades ago, amplification of mitochondrial or ribosomal markers failed. Still, we were able to amplify a fragment of the nuclear protein-coding Histone 3a locus (H3), using primers and protocols from a previous study (Harms et al. 2019). Amplification of this short gene fragment (324 bp in total) was successful for eighteen Malagasy samples, representing eighteen localities (Table 1). Two additional Feaella (Tetrafeaella) specimens were also sequenced, including Feaella (T.) cf. anderseni (Harvey, 1989)

Phylogenetic Analyses
The H3 dataset comprising the eighteen Malagasy feaellids, two Feaella species and eleven outgroup taxa for a total of 31 terminals, was aligned using MAFFT version 7 online server (https://mafft.cbrc.jp/alignment/ server), applying the G-INS-i method of alignment, which assumes global homology (Katoh et al. 2002, Katoh and Toh 2008, Katoh and Standley 2013 Mean uncorrected pairwise (p) genetic distances were calculated among all feaellid taxa using MEGA v. 11 (Tamura et al. 2021).

Phylogenetic Analyses
Maximum likelihood analysis of the H3 dataset recovered Feaellidae as monophyletic with T. cf. mucronata and T. cf. anderseni forming a clade, sister to the Malagasy feaellids, which in turn, formed three geographically delimited clades (Fig. 3)  95% for the central, southern and northern clades, respectively. Uncorrected p-distances between Malagasy terminals ranged from 0% (MGF 034 and MGF 033) to 15.2% (between Sept Lacs and Forêt de Bekaraoka). Mean uncorrected p-distances between the Malagasy species and two Feaella species was 15.5% and among the Malagasy clades were as follows: central-southern clades (11.4%), northern-central clades (12.0%) and northern-southern clades (13.1%). Mean uncorrected p-distances within each clade were 5% for the northern and southern clades, and 6% for the central clade. Subdivision into three geographically delimited clades is also mirrored by morphological differences, primarily morphometrics and chaetotaxy of the pedipalps, which are commonly used in pseudoscorpion taxonomy for species delimitation. Based on results of the phylogenetic analyses, genetic distances and supporting morphological data, these three geographically-delimited Malagasy clades are recognized as genera below.

Subfamily Feaellinae Ellingsen, 1906
All genera and species described below can be placed in the subfamily Feaellinae based on the presence of pleural platelets, the absence of the cheliceral rallum and an unmodified coxa III. Cybellinae lack pleural platelets, coxa III is modified, and the rallum comprises two blades (Judson 2017;Harvey 2018).
Note that there are significant morphological differences between the species included in the three different subgenera of Feaella which are retained here for practical reasons but almost certainly polyphyletic and in need of revision. The status and composition of Feaella (Tetrafeaella) is problematic due to many poorly described species, including the type species Feaella (T.) indica Chamberlin, 1931, and the knowledge that the Australian fauna is misplaced in the genus (Harvey et al. 2016a).
The newly defined genera also differ from all previously described genera of Feaellinae and can be differentiated using characters in the key below:  Diagnosis. Antsirananaella gen. nov. differs from other Malagasy genera, Toliaranella gen. nov. and Mahajanganella gen. nov., by the following characters: the pres-ence of five specialized setae on the retrolateral face of the movable chelal finger that are arranged in a row between trichobothrium b and the terminal chelal teeth (arranged in a group in Toliaranella gen. nov. and Mahajanganella gen. nov.); the presence of five terminal teeth (including one large tooth) on the fixed chelal finger and seven on the movable chelal finger (number of terminal teeth varying from 7-9 teeth in fixed and movable fingers in other Malagasy genera); anteromedial lobes of carapace closer to each other and slightly longer than anterolateral lobes (all four lobes equidistant in other Malagasy genera); and larger body size (1.80-2.17/2.17-2.47 ♂/♀ in Antsirananaella gen. nov. and ca. 1.56-1.89/1.96-2.33 ♂/♀ in all other genera). Antsirananaella gen. nov. differs from Cybella (Southeast Asia) by having platelets on the pleural membrane (absent in Cybella), from Iporangella (Brazil) by the presence of specialized setae on the movable chelal finger (absent in Iporangella), from Feaella (Difeaella) (South Africa) and Feaella (tropical Africa) by the presence of four anterior carapaceal lobes (two and six, respectively), from Feaella (Tetrafeaella) in conti-nental Africa by having a less pronounced depression on the base of coxa I and on top of coxa II (distinctly more pronounced in Feaella (T.) cf. mucronata in Figs 25-27), and from the Australian species presently attributed to Feaella (Tetrafeaella) by having fewer coxal spines (one pair versus three or four in the Australian taxa).

Key to genera and subgenera of Feaellinae
Etymology. This genus is named after the former Antsiranana Province, the area where the specimens were found. The gender is feminine.  Description. The following description is based on examination of all species in Antsiranana. Typical feaellid habitus with a spherical abdomen, short and robust pedipalps, four prominent carapaceal lobes and dark reddish-brown body colour. -Carapace: (Figs 5A, 7A, 8A, 9A, 10A): With four distinctly pointed anterior lobes; anteriolateral lobes slightly broader and wider than anteriomedial ones; anteriomedial lobes distinctly longer and closer to each other than anteriolateral ones; two pairs of eyes with lenses, equal in size, second pair partly covered by cuticula; four prominent posterior lobes (pm, pl in Figs 7A, 8A, 9A, 10A) and two longitudinal furrows more distally (af, pf in Figs 7A, 8A, 9A, 10A); with two medio-lateral mounds; and two postero-lateral processes. -Pedipalp: With a distinct conical protuberance on trochanter (Figs 4A, B, 7D, 8D, 9D, 10D); femur broad and with one prolateral triangular process plus a retrolateral hump, patella cone-shaped. Chelal hand very small and with one large medial tooth at the base of each finger. Fixed finger with 9 trichobothria including dt, movable finger with 4 trichobothria (Figs 6A, B, C, 7C, 8C, 9C, 10C). Movable finger with 5 specialized setae on ventral face, arranged in a transverse row between b and terminal teeth (Fig. 6A). Chelal teeth large and retrorse, arranged in three rows on both chelal fingers. Terminal teeth situated in compact groups facing medial, four equally sized teeth and one larger tooth on fixed finger, seven equally sized teeth on movable finger (Figs 6A, B, C, 7C, 8C, 9C, 10C). With 2 sensory setae (dt in Figs 6B, 7C, 8C, 9C, 10C) on dorsal tip of fixed finger. -Chelicera: Palm with five long and several short setae; is and ls close; sbs proximal on fixed finger; movable finger with 1 subdistal seta (gs); es on palm close to base of movable finger; galea absent; spinneret conical; serrula exterior with 16-20 blades; no rallum could be found in any of the Malagasy species because of dirt; movable finger short and without teeth. coxae II with irregular shaped spines various in number framing depression of coxa I (Fig. 5B); all coxae reaching towards midline and with coxa IV bigger in size than coxa I-III; cuticle within depression strongly granulate.
-Legs (Figs 4B, 7B, 8B, 9B, 10B): Trochanter I and II rather circular, trochanter III and IV rather elliptical; femur I and II slightly longer than patella I and II; femur III and IV shorter than patella III and IV; all tarsi long and slender, without specialized tactical trichobothria; subterminal tarsi with two curved and smooth claws; all setae acuminate; arolium much shorter than claws and with fimbriate distal margin; claws divided. -Abdomen (Fig. 4A, B, C): Paired tergites and sternites medially divided; anal plate strongly sclerotized and with a cir-
Etymology. This species is a patronym for the senior author's family that include parents Sylvia and Günther Lorenz and brother Henrik. Trichobothrial pattern: esb and est in the proximal half of the retrolateral face; ib, isb and ist situated basally as a slightly curved line, isb and ib closer to each other than isb and ist; eb and it situated sub-distally and very close to each other, with it more distal and more medial than eb; et situated distally and approximately the same distance to dt than to eb; dt situated very distal in a plain pit; st situated sub-basally at the same sagittal level as esb; st and est also on the same level as ist; t distinctly closer to sb than to b; sb distinctly closer to t than to st. Chelal fixed finger with 10 teeth in the OR, 20 in the MR and 15 in the IR; movable finger with 11 teeth in the OR, 14-15 teeth in the MR, and 14 teeth in the IR. -Dimensions (mm): Holotype ♂: Body length 1.94; abdomen length 1.33; abdomen width 1.12 (without pleura), 1.24 (with pleura); carapace length 0.58; carapace width 0.42. Pedipalp: trochanter 0.22; femur length 0.54; femur width 0.32; patella 0.46; chela (without pedicel) 0.54; hand length (without pedicel) 0.12; width 0.14; movable finger length 0.42. Leg I: trochanter 0.11; femur 0.19; patella 0.17; tibia 0.17; tarsus 0.27. Leg IV: trochanter 0.17; femur 0.13; patella 0.28; tibia 0.29; tarsus 0.35. Allotype ♀: Same as male except body length 2.26; abdomen length 1.59, width 1.36 (without pleura), 1.49 (with pleura); carapace length 0.66, width 0.49. Pedipalp: trochanter 0.30, femur length 0.75, width 0.44, patella length 0.61, chela (without pedicel) 0.64, hand length (without pedicel) 0.14, width 0.18. Leg I: trochanter 0.14, femur 0.25, pa- GenBank Code. OP589964. The species differs from all congeners by more than 6.9% in the H3 dataset.
Habitat. Dry deciduous forest between the coast and a rising limestone plateau.  Etymology. This species is named after my former Latin teacher, Gerhard Faulstich, who might be one of the wisest men ML has ever known.
Habitat. Tropical dry forest at 550 m altitude. All specimens were found in sifted litter (leaf mold, rotten wood).
Distribution. Presently known from three localities: type locality (BLF10116) and two additional localities (BLF9556 and BLF9872) in the Sava Region (formerly Antsiranana Province).
Etymology. The specific epithet is a patronym in honour of Marla Elisa Nibasumba, the senior author's godchild.

Description.
The following description is based on holotype and allotype. -Carapace (Fig. 10A): 1.43-1.49 (♂), 1.30-1.46 (♀) times longer than broad. -Pedipalps (Fig. 10C): Trichobothrial pattern: sb on the movable finger and est on the fixed finger on the same sagittal level. et situated in the distal third between dt and eb on the fixed finger. Chelal fixed finger with 11 teeth in the OR, 21 teeth in the MR, and 12 teeth in the IR; movable finger with 11 teeth in the OR, 13 teeth in the MR, and 13 teeth in the IR.  Diagnosis. Mahajanganella gen. nov. is morphologically unique by having the following characters, which differ from other feaelloid members in Madagascar: 5 specialized setae on the retrolateral face of the movable chelal finger arranged in a group between trichobothrium b and terminal teeth (arranged in a row in Antsirananaella gen. nov.); smaller overall size than Antsirananaella gen. nov.; stronger granulate cuticle than Antsirananaella gen. nov.; all four anterior lobes of carapace with the same distance to each other (smaller distance between anteriomedial lobes than to anteriolateral ones in Antsirananaella gen. nov.). Like Antsirananaella gen. nov. and Toliaranella gen. nov., it differs from Cybella by having platelets on the pleural membrane (absent in Cybella), from Iporangella by the presence of specialized setae on the movable chelal finger (absent in Iporangella), from Feaella (Difeaella) and Feaella (Feaella) by the presence of four anterior carapaceal lobes (two and six, respectively), from Feaella (Tetrafeaella) in continental Africa by having a less pronounced depression on the base of coxa I and on top of coxa II (distinctly more pronounced in Fe- aella (T.) cf. mucronata), and from the Australian species presently attributed to Feaella (Tetrafeaella) by having fewer coxal spines (one pair versus three or four in the Australian taxa).
Etymology. The genus is named after the former Mahajanga Province, where many of these feaellids occur. The gender is feminine.

Description.
The following description is based on holotype and allotype of M. heraclis sp. nov. -Carapace (Figs 12A, 14A, 15A, 16A): Moderate granulate and all four anterior lobes with same distance to each other. -Pedipalp (Figs 11A, B, 13A, B, C, 14C, D, 15C, D, 16C, D): Trichobothrial pattern: esb between st and ist on the sagittal level with ist slightly more distal and st slightly more proximal than esb; t distinctly closer to sb than to b; sb distinctly closer to t than to st; it on the fixed finger between eb and et distinctly closer to eb than to et. Movable finger with five specialized setae on the ventral face, arranged in a group and slightly more distally situated than b (Fig. 13A). Terminal teeth varying in number, both in fixed and the movable fingers each with 7-8 terminal teeth (one large tooth included on fixed finger). -Cheli- cera: Most specimens were too dirty to get secure information about characters. Coxal region (Fig. 12B, C), legs (Figs 11B, 14B, 15B, 16B), abdomen (Fig. 11A, B, C) and genital region (Fig. 12B) with the same characters as Antsirananaella gen. nov.
Etymology. This species is named after the divine hero Herakles (lat. Hercules) who, in Greek mythology, is known for his power and courage. He is often portrayed with a mace, which resembles the strong pedipalps of this species. It is to be treated as a noun in apposition.
Habitat. The specimens were found in sifted litter of tropical dry forest at an elevation of 135 m.
Distribution. Presently known only from the type locality (BLF3599) and from three additional locations (BLF3571, BLF6506 and BLF6446) in the Boeny Region (formerly Mahajanga Province).

Diagnosis.
Very similar to the type species but remarkably more granulated cuticle on the carapace, especially above the posterior pair of eyes; anterior lobes on the carapace are not pointed but rounded at the tip.
Etymology. This species is named after the Mexican artist Frida Kahlo de Rivera ( † 1954) whose unmistakable character were her striking eyebrows, which she included in many of her self-portraits. The species is reminiscent of her because of the strongly granulated cuticle above the second pair of eyes, which resemble 'eyebrows'.

Description.
The following description is based on holotype and allotype. -Carapace (Fig. 15A): Cuticle strongly covered with granulate especially above the second eye pair which makes it look like an additional lateral pair of mounds behind the anterior margin; anterior lobes on carapace not pointed but rounded at the tips; 1.26-1.42 (♂), 1.18-1.34 (♀) times longer than broad.
GenBank Code. OP589969. The species differs from all congeners by more than 6.5% in the H3 dataset.  Etymology. This species is a patronym honoring Arnold Schwarzenegger, a famous former bodybuilder from Austria, known as an actor in the movie "Terminator", former governor of California and now supporting conservation programs.

Description.
The following description is based on holotype and allotype. -Carapace (Fig. 16A): 1.47 (♂), 1.38-1.43 (♀) times longer than broad. -Pedipalps (Fig. 16C, D): see trichobothrial pattern in genus description. Chelal fixed finger with 8 teeth in the OR, 13-17 in the MR and 9-12 in the IR; movable finger with 8-9 teeth in the OR, 11-13 in the MR and 9-12 in the IR; terminal teeth build a group of 6 equally sized and 1 larger tooth on the fixed finger and 8 equally sized teeth on the movable finger.

Diagnosis. Toliaranella gen. nov. differs from all other
Malagasy feaellids by the presence of a perforated organ above the coxal spines and smallest body size of any Malagasy groups. Like Antsirananaella gen. nov. and Mahajanganella gen. nov., it differs from Cybella by having platelets on the pleural membrane (absent in Cybella), from Iporangella by the presence of specialized setae on the movable chelal finger (absent in Iporangella), from Feaella (Difeaella) and Feaella (Feaella) by the presence of four anterior carapaceal lobes (two and six, respectively), from Feaella (Tetrafeaella) in continental Africa by having a less pronounced depression on the base of coxa I and on top of coxa II (distinctly more pronounced in Feaella (T.) cf. mucronata) and from the Australian species presently attributed to Feaella (Tetrafeaella) by having fewer coxal spines (one pair versus three or four in the Australian taxa).
Etymology. This genus is named after the former Toliara Province, where all specimens were collected. The gender is feminine.
Etymology. This species is named after the latin 'pumilus, -a, -um' which means "dwarf". The specific epithet references the small size of all specimens.
Description. The following description is based on holotype and allotype. -Carapace (Fig. 20A): 1.29-1.44 (♂), 1.24-1.47 (♀) times longer than broad. -Pedipalp (Fig. 20C, D): Femora very robust (1.55-1.69 (♂), 1.43-1.80 (♀) times longer than broad; trichobothrial pattern: same as in Mahajanganella gen. nov., except: est on fixed finger on same sagittal level as sb on movable finger. Chelal fixed finger with 9-10 teeth in the OR, 12-15 in the MR and 8-12 in the IR; movable finger with 6-10 teeth in the OR, 9-11 in the MR and 9-10 in the IR; fixed finger with 6 equally sized and one larger terminal tooth, movable finger with 7 equally sized terminal teeth. -Chelicera, legs, abdomen and genital region: With same characters as all members before.   Habitat. Specimens were found in sifted litter of tropical dry forest (leaf mold, rotten wood) at an elevation of 375 m.

Etymology. This species is a patronym honoring Brian
Fisher who collected many Malagasy feaellids.
Description. The following description is based on holo type and allotype. -Carapace (Fig. 21A): 1.25-1.39 (♂), 1.27-1.39 (♀) times longer than broad. -Pedipalps (Fig. 21C, D): See trichobothrial pattern in description of T. griswoldi sp. nov. Chelal fixed finger with 10 teeth in the OR, 13-16 in the MR and 10-13 teeth in the IR; movable finger with 8-10 teeth in the OR, 9-12 in the MR and 8-11 teeth in the IR. Fixed finger with 6-7 equally sized and one large terminal tooth, movable finger with 7 terminal teeth equal in size. Etymology. This species is a patronym honoring Charles Griswold who collected many Malagasy feaellids.
Description. The following description is based on holotype and allotype. -Carapace (Fig. 22A): Very compact, heavily granulate, 1.20-1.31 (♂), 1.18-1.31 (♀) times longer than broad. -Pedipalps (Fig. 22C,  D): Similar trichobothrial pattern as T. pumila sp. nov. but the following difference: est on fixed finger situated between sb and t of the movable finger on sagittal level. Chelal fixed finger with 9-10 chelal teeth in the OR, 12-14 in the MR and 9-11 in the IR; movable finger with 6-8 in the OR, 7-11 in the MR and 7-8 in the IR.  GenBank Code. OP589959, OP589960. The species differs from all congeners by more than 3.7% in the H3 dataset.
Habitat. Specimens were found in sifted litter of spiny forest thicket and tropical dry forest at an elevation of 150-300 m.
Distribution. Currently known from the type locality (BLF4810) and two additional localities of (BLF4984 and BLF4815) in the Anosy Region (former Toliara Province) in southern Madagascar.

Toliaranella mahnerti sp. nov.
Diagnosis. Very similar to T. pumila sp. nov. but larger Habitat. Specimens were found in leaf litter in spiny forest thicket, degraded and undegraded gallery forest, and in disturbed and undisturbed riparian forest; elevation 20-160 m. All close to a river stream.  Etymology. This species is named after the type locality, which is the southernmost point of Madagascar.
GenBank Code. OP589961. The species differs from all congeners by more than 3.3% in the H3 dataset.
Habitat. Specimens found in sifted litter from spiny forest thicket very close to the coastline at an elevation of 160 m.
Distribution. Known from the two southernmost localities (type locality and BLF5500) among all Malagasy feaellids in the Androy Region (formerly Toliara Province).

Biogeography
The only available studies specifically addressing Malagasy pseudoscorpions are short notes (Vachon 1960, Heur tault 1986 pointing out that all families present, including Feaellidae, do not seem to have undergone excessive in-situ speciation and that many genera are shared with continental Africa, southern India, Australia or Southeast Asia (Heurtault 1986). The present contribution is the first detailed study on a pseudoscorpion lineage in Madagascar and we point out that, after detailed examination, it is evident that at least the Feaellidae of Madagascar are a highly endemic fauna at both the generic and species level. We argue that this is most likely true for other pseudoscorpion families occurring on Madagascar but none of these have been revised taxonomically. Although there appear to be no families restricted to the island, endemism might in fact be substantial at the genus and species level and initial assessment of low endemism may result from the general problems of pseudoscorpion taxonomy, such as cryptic morphologies that change little over time, perhaps because of niche conservatism and retainment of ancestral ecological characters in soil-dwelling fauna (Wiens and Graham 2005). Undoubtedly detailed morphological studies in conjunction with genetic data will be necessary to fully unravel the evolutionary history of the Malagasy pseudoscorpion fauna.
Feaellids are an ancient lineage with a fossil origin that dates back to the Late Triassic and Late Cretaceous (Henderickx and Boone 2016;Harms and Dunlop 2017;Kolesnikov et al. 2022) and molecular clock estimates suggest the lineage originated in the Paleozoic (Benavides et al. 2019). No molecular phylogeny currently exists for the family Feaellidae but it can be hypothesized that the origins of Malagasy Hercules pseudoscorpions is Gondwanan, similar to other ancient invertebrate lineages in soil habitats such as pill millipedes (Wesener et al. 2010). Under a continental vicariance scenario, we predict this fauna to be closely allied with that of the Seychelles and southern India, followed by southern Africa, and then tropical Africa (Yoder and Nowak 2006) -a scenario that could be tested in a molecular phylogenetic framework because feaellids occur in all three of these regions and at least those from India, Maldives and the Seychelles are morphologically quite similar (e.g., Mahnert 1978;Novák et al. 2020) but differ from those in Brazil and Australia (e.g., Harvey et al. 2016a, b) and also Southeast Asia (Judson 2017). We hypothesize further that a dated molecular phylogeny should reflect continental vicariance to some degree, although we cannot rule out the possibility that some of the new genera described herein are older. Not only are feaellids an ancient Pangaean lineage, but the molecular data presented here and elsewhere (Harvey et al. 2016a) clearly refute a historical concept based on morphology that considers many feaellid species to be widespread. In line with recent molecular studies on other soil-dwelling pseudoscorpions (e.g., Cosgrove et al. 2016;Harms 2018;Harms et al. 2019), the case is rather the opposite and feaellids appear to show extreme microendemism, habitat restriction and pronounced population structuring despite the absence of morphological disparity, perhaps comparable to some mygalomorph spiders that typically thrive in localized populations with limited dispersal between them (Greenberg et al. 2021) or mite harvestmen occurring also in soil habitats (De Bivord and Giribet 2010; Jay et al. 2016). Vicariance at various spatial scales (continental, regional and local) and different time periods seems to be a major factor that needs to be tested, although at least some feaellids occur in coastal habitats (Beier 1955) and might disperse, as recently suggested for a species found in the Maldives (Novák et al. 2020). Overall, our data are congruent with a hypothesis of pronounced endemism at all spatial scales, vicariance as a common case for diversification, and small species ranges in specific habitats in the presence of relative morphological stasis that may be caused by niche conservatism. Feaellids therefore emerge as a new model group to test hypotheses pertaining to vicariance biogeography. Vences et al. (2009) reviewed possible diversification mechanisms for Madagascar and distinguished between ecographical constraints (meaning adaptive radiation in response to climatic shifts), speciation in rainforest refugia, montane refugia, riverine barriers, and river catchments as testable scenarios. Many studies on invertebrates have emphasized the importance of mountain or rainforest refugia (Wesener et al. 2011;Wood et al. 2015) and pointed out that in-situ speciation in such refugia, together with time and adaptive niche radiation, has been fundamentally important in diversification. Many taxonomic revisions have also focused on taxa that are exclusively or predominantly found in the high-altitude forests of the east (Griswold 1997;Griswold et al. 2012;Wood and Scharff 2018) and documented substantial microendemism. For more arid habitats, a study by Wesener (2009) found that endemism in pill millipedes from dry forests was not as pronounced as that of species occurring in mesic habitats. Comprehensive taxonomic revisions on goblin spiders (Family Oonopidae Simon, 1890) have also documented major endemism at the generic level for the Malagasy fauna (Álvarez-Pardilla et al. 2012;Saucedo et al. 2015), but wide species distributions in arid habitats that roughly correspond with the generic distributions in our dataset. Perhaps the highest similarity is with the distribution of the goblin spider genus Opopaea Simon, 1891, where species endemism is high even in arid habitats (Andriamalala and Hormiga 2013). So, what is driving the high diversity of feaellids in Madagascar? Adaptive radiation in response to climatic zonation is a likely hypothesis for the three feaellid genera described in our paper. Antsiarananaella gen. nov. is clearly restricted to the dry deciduous savannah forests of the north, a biome that is well known as a center of endemism in both vertebrates and invertebrates (e.g., Brown et al. 2016). The Irodo and Loky Rivers seem to be major barriers here that define species ranges to some extent and divide the ranges of A. faulstichi sp. nov. and A. leniae sp. nov. and A. marlae sp. nov. although forest cover and topography also seem to be important. The distribution of Mahajanganella gen. nov. encompasses the dry deciduous forests of the western Malagasy biome and species seem a little more widespread and river drainages do not seem as important here, although our species hypotheses are primarily based on morphology and we cannot rule out the possibility that more widespread species such as M. heraclis sp. nov. might show significant genetic structuring or cryptic speciation. Toliaranella gen. nov. is restricted to the subarid biome of the south, which is primarily covered by arid spiny bush vegetation. Our maps and genetic data (whenever available) clearly show that river drainages can define species ranges, such as with T. mahnerti sp. nov. that is found in the Fiherenana and Onilahy drainages but nowhere else. The Mandrare drainage also has a distinct species (T. fisheri sp. nov.) although some species such as T. meridionalis sp. nov., are found in coastal drainages of the Cape St. Marie Special Reserve in the south. It appears that ecographical constraints are effective at the generic level in Malagasy Hercules pseudoscorpions, whereas river catchments and sclerophyll forest refugia offer a testable framework to derive species limits in this fauna. It might also be pointed out that feaellids are absent from two of the major Malagasy biomes which is the central highlands and the evergreen forests to the east. These are also the regions that receive the highest annual rainfall with >1500 ml per year, whereas areas occupied by feaellids altogether receive less than 1000 ml. It appears that Malagasy feaellids are absent from rainforest habitats, unlike the fauna of central and western Africa or species recorded from Brazil that have been collected from rainforest habitats (e.g., Harvey et al. 2016b) or Southeast Asia where the only records are from caves (Judson 2017;Harvey 2018). Ecologically this fauna seems to be more closely aligned with that of eastern Africa where feaellids have been collected from dry leaf litter in savannah or near-coastal habitats (Beier 1966;Mahnert 1982;Henderickx 2009). Interestingly, some other pseudoscorpion lineages in Madagascar show exactly the opposite pattern, such as the family Pseudotyrannochthoniidae Beier, 1932 that seems to be restricted to the central highlands and the evergreen forests according to survey records. Revising these taxonomically could complement our understanding of the Malagasy pseudoscorpion fauna and help generate a more comprehensive picture of pseudoscorpion diversification patterns on the island.

Conservation
Madagascar is amongst the "hottest" biodiversity hotspots and the conservation of the remaining natural or semi-natural habitats is crucial to ensure lineage survival under intense anthropogenic pressure. It is therefore relevant to note that none of the species described in this study come from strictly protected areas or reserves such as Tsingy de Bemaha Reserve, Betampona Reserve, Tsaratanana Reserve, or Zahamena Reserve, but that some species occur in areas that have gained some conservation status after the specimens were collected: Ankarafantsika National Park ( . We emphasize here that many other species with narrow distributional ranges occur in habitats that do not have any conservation status at present, such as T. fisheri sp. nov. Conservation incentives should target these species and their habitats whenever possible.

Systematics & Taxonomy
In this paper, we describe three new genera of Hercules pseudoscorpions, each restricted to a specific biome and climate of Madagascar, and each of these containing multiple species with short ranges. Aside from biogeographical and conservation aspects, our study also highlights the need to revise generic concepts in the family Feaellidae, specifically the genus Feaella in the Afrotropics that is presently divided into the subgenera: Feaella (Feaella), Feaella (Tetrafeaella) and Feaella (Difeaella) based on the number of anterior lobes of the carapace ranging from two to six (Beier 1966;Mahnert 1978). The most com-mon combination is four lobes as in Feaella (Tetrafeaella) and this condition is shared by all Malagasy species although these species are dissimilar from the type species, Feaella (Tetrafeaella) cf. mucronata (Figs 25,26,27) from Amanzimtoti in South Africa, both morphologically and genetically. Species from Australia, the Seychelles, the Maldives and the Indian subcontinent are also presently classified into the subgenus Tetrafeaella, which makes little sense biogeographically and it is evident that the distinguishing character (number of lobes) is of limited phylogenetic relevance.
It is not the aim of this study to test the generic limits within Feaellidae. However, a phylogenetic analysis with a detailed morphological investigation, including scanning electron microscopy and perhaps even micro-computed tomography, should be undertaken to resolve the phenetic classification. Our study may provide an avenue insofar as we define new genera and have identified morphological characters that had been overlooked or not given widespread attention, such as the sensory apparatus and setae, chelal tooth morphology, and cuticle sculpturing, which is an obvious feature in all feaellids that generally have a thick and armed body cuticle. Focusing on minute details unfortunately seems to be important in a morphologically highly conserved pseudoscorpion lineage where species are similar across continents despite significant genetic divergences (e.g., Harvey 2016a). Analyzing the causes for such a conserved morphology and the resulting morphological crypsis of lineages, both at the species and generic level, might be an interesting task for the future, not only for Feaellidae, but also for some other pseudoscorpion lineages because strong selective pressures might exist limiting morphological change over millions of years and across continents. Alternatively, this pattern may also represent an exceptional case of ecological niche conservation reflected by morphological stasis. Since feaellids are an ancient invertebrate lineage with possible origins in the Carboniferous (Benavides et al. 2019), the low morphological diversity might be a result of the latter, but virtually nothing is known about feaellid biology and ecology and that discussion is left for the future.