Sympolymnia, a new genus of Neotropical ant-like spider, with description of two new species and indirect evidence for transformational mimicry (Araneae, Salticidae, Simonellini)

Sympolymnia, a new genus of myrmecomorph jumping spider belonging to the tribe Simonellini Peckham, Peckham & Wheeler, 1889, is described. It comprises five species: the type species, Sympolymnia lucasi (Taczanowski, 1871), comb. nov., Sympolymnia lauretta (Peckham & Peckham, 1892), comb. nov., Sympolymnia edwardsi (Cutler, 1985), comb. nov. and Sympolymnia shinahota sp. nov. and S. cutleri sp. nov. Sympolymnia lauretta (Peckham & Peckham, 1892) is recorded from Bolivia for the first time. Ontogenetic shifts of ant-resemblance are observed: Juveniles of S. cutleri sp. nov. and S. lauretta mimic black ants of the genus Crematogaster Lund, 1831, but those of S. shinahota sp. nov. most closely resemble Pseudomyrmex ethicus (Forel, 1911). Adults of S. cutleri sp. nov., S. lauretta and S. shinahota sp. nov. resemble the ant Camponotus sanctaefidei Dalla Torre, 1892 and orange adults of S. shinahota sp. nov. are putative mimics of Camponotus latangulus Roger, 1863.


Introduction
Ant-resembling spiders have fascinated many naturalists over centuries and are a promising group to study mimicry and evolution through natural selection (Nelson and Jackson 2012;Ceccarelli 2013). The morphological resemblance to ants (or myrmecomorphy) occurs in many spider families (Cushing 1997), but is particularly common in jumping spiders (Maddison 2015;Maddison and Szűts 2019). The ant-like appearance has reached an extreme in Synemosyna Hentz, 1832 (Peckham and Peckham 1892;Oliveira 1988), a genus of the Neotropical tribe Simonellini Peckham, Peckham & Wheeler, 1889. Several species of Synemosyna were cited for their strong resemblance to members of the ant genus Pseudomyrmex Lund, 1831 (see Cushing 1997 for a review). The similarity was attributed to a conspicuous-ly elongated body with short legs and locomotory movement with short, rapid lunges (Oliveira 1988). Additionally, these forms have a deeply-constricted abdomen, which imitates the separation between the postpetiole and the gaster of Pseudomyrmex ants (Oliveira 1988).
The genus Synemosyna was established by Hentz (1846) to accommodate S. formica Hentz, 1846. Three species were subsequently described in this genus Peckham 1892, 1894). Galiano (1966) revised Synemosyna and synonymised Simonella Peckham & Peckham, 1885 with the former, which was accepted by subsequent authors (Cutler 1981(Cutler , 1985(Cutler , 1993Cutler and Müller 1991;Makhan 2006) and presently Synemosyna comprises 20 species (WSC 2020). Cutler (1985) emphasised the presence of an ovate abdomen in Synemosyna lauretta and S. lucasi (besides the large spermathecae and simple copulatory tubes) and described Synemosyna edwardsi which also possessed these characters. Those three species differ from the elongated and slender species of Synemosyna that resemble Pseudomyrmex ants and are possibly mimics of ants of the genus Crematogaster Lund, 1831 (Cutler 1985). In the present contribution, we revise the taxonomy of these species and report the results of a survey in four forest ecoregions in Bolivia, including the description of two new species, a new country record and indirect evidence of transformational ant mimicry.

Methods
The sampling of spiders and ants was conducted in 10 locations in primary forest in four forest ecoregions of Bolivia (Amazon, Yungas, Chiquitano and Bolivian Tucuman forest, according to the ecoregion classification by Navarro and Ferreira 2011) (Figs 1, 2). Spiders and ants were collected with a beating tray. Photographs of live spiders and their habitats were taken with a Panasonic Lumix GX-80 system camera fitted with macro lenses. Colour was described from photos of live specimens. Spiders were euthanised with ethyl acetate and stored in 80% ethanol. Preserved specimens were ex-amined under both dissecting microscopes and a compound microscope with reflected light and identified using original and re-descriptions and keys (e.g. Peckham and Peckham 1892;Galiano 1966Galiano , 1967Mello-Leitão 1933;Taczanowski 1871;Cutler 1985). Photos of preserved specimens were taken with the Visionary Digital Passport II Imaging system at the Zoological Museum, University of Hamburg, Germany.
Female genitalia were dissected as in Levi (1965), examined after digestion in ~15% sodium hydroxide (NaOH) solution and clarified in clove oil to examine the internal structures. Temporary preparations were observed and photographed by GR using a Leica DM500 compound microscope and a Leica M60 stereomicroscope. Structures were sketched on incident light photograph models using a computer system for drawing and treatment of the image (Wacom digitiser tablet with GIMP, free software). All measurements, which were obtained with an ocular micrometer, are given in millimetres. Body length (BL) measurement refers to the distance from the anterior margin of the carapace to the posterior margin of the opisthosoma. Morphological terms and description formats follow the main recent studies on similar jumping spiders (Ruiz and Maddison 2015).  Navarro and Ferreira (2011), map produced with QGIS (version 2.14.3, http://www.qgis.org/en/site/).

Ecoregion distribution
Distributional records without coordinates were georeferenced via the gazetteers GeoLocator (http://tools. freeside.sk/geolocator/geolocator.html) and GeoNames (http://www.geonames.org/). The ecoregion affinities of the species at a continental level were investigated by visualising the coordinates and shapefiles of the regionalisation of Neotropical ecoregions by Olson et al. (2011) by using the geographic information system, QGIS (version 2.14.3, http://www.qgis.org/en/site/). For Bolivian ecoregions, the shapefile from Navarro and Ferreira (2011) was used, as the results were more consistent than those obtained from Olson et al. (2011). Geographic coordinates are shown in decimal degrees with reference datum WGS84 and elevation in metres above sea level (m a.s.l.).

Ant mimicry
In this study, an indirect, correlative method is employed to support mimicry, without studying the impact of receiver responses on mimic fitness. While correlations do not imply causality, correlative approaches are useful for investigating putative cases of resemblance between taxa and extrapolating the consequences of mimicry beyond a single, well-studied population (de Jager and Anderson 2019). To illustrate adaptive divergence between mimetic and non-mimetic phenotypes, we identified derived traits in the mimic that likely evolved in association with species-specific models and receivers.
For the analysis of ant resemblance, we considered all ants that were collected in the surveyed locations and were about the same body length of the spiders. The similarity was analysed, based on a qualitative, descriptive assessment of integument colour, shine, development (e.g. appressed, erected, short, long) and colour of hairs and shape of body parts (e.g. abdomen shape: fusiform or ovate; apically pointed or rounded).
Etymology. The specific epithet, Sympolymnia, is a combination of "sym", meaning "with" in Greek and "Polymnia", one of the nine Muses of Greek mythology, daughter of Zeus and Mnemosyne and the protector of the divine hymns and mimic arts. The gender of the name is feminine.
Remarks. Cylistella has a rounded, beetle-like habitus without constrictions and is the morphologically most distinct group within this tribe. Sympolymnia gen. nov. is possibly most related to Synemosyna (Table 1). This relationship is indicated by the presence of a constriction between the cephalic and thoracic parts and a single female genital opening. Additionally, in Synemosyna aurantiaca and Synemosyna formica, the embolus of the male palp is elongated and winds around the retro-dorsal surface of the dorsum of the cymbium, similar to species of Sympolymnia gen. nov. However, all species of Sympolymnia gen. nov. are distinguished from Synemosyna by consistent differences in genitalic and somatic characters (Table 1), requiring an adequate generic treatment (see also Ruiz and Bustamante 2016;Kanesharatnam and Benjamin 2018;Rubio et al. 2020 for other, recent generic taxonomy in Salticidae).  (Cutler, 1985), comb. nov.
Diagnosis. Thoracic part narrower than cephalic part in dorsal view, distinctly concave posteriorly in lateral view (Fig. 6A); abdomen at most with slight dorsal constriction; bulb of male palp about 65% of cymbium length, tibial apophysis in retro-lateral view with relatively broad base (see Galiano 1966 for line drawings), apex with spine-like process; epigyne opening small (width: ~ 0.07 mm), transversely elliptical.
Variation. Smaller juveniles (BL ≤ 3.1 mm) of S. lauretta had a shiny black body with pointed abdominal apex (Fig. 8I). Larger juveniles (BL 3.2-3.9 mm) had a bright and pointed abdomen, but a black, matt carapace. The body of the adults (BL ≥ 3.9 mm) was completely matt, and the abdomen apically more rounded (Figs 3B, 8K).
Geographical and ecological distribution (Fig.  7). Sympolymnia lauretta was reported from Brazil (Peckham and Peckham 1892;Galiano 1966;Podgaiski et al. 2007;Rodrigues et al. 2016), Argentina (Galiano 1966) and Peru (Cutler 1985). This species is predominately found in seasonal forests south of 15°S. According to the biogeographic regionalisation by Olson et al. (2011), the records of this species refer to Araucaria moist forest, Cerrado, Alto Parana Atlantic forest and Serra do Mar coastal forests. In the present study, S. lauretta was observed in Tucuman (Tarija Department: Arambulo) and Chiquitano forests (Santa Cruz Department: Bermejo, Santa Rosa de la Mina, Santiago de Chiquitos). The records of the present study are the first for Bolivia. In almost all primary forest locations, S. lauretta was obtained from isolated smaller trees and larger bushes overgrown with climbing plants (Fig. 2C, D), which are typically found in early successional forest and along forest edges. Sympolymnia lauretta was not collected in closed, moist forests in the study area. Cutler and Edwards (2002) recorded S. cf. lauretta from Trinidad Island (Lesser Antilles). The taxonomic status of this population remains to be determined.  (Figs 3C, 4B, 8A-C); male palp ( Fig. 9C-F) with small cymbium, bulb about 80% of cymbium length, tibial apophysis in retro-lateral view narrow, apex evenly tapering (Fig. 9C, D); copulatory opening in female very small and quadrangular (width: ~ 0.05 mm) (Fig. 5C).
Comparisons. The male palp of S. shinahota sp. nov. (Fig. 9C-F) resembles that of S. lucasi. However, the latter is distinguished by a male palp tibial apophysis with a spine-like apex (evenly tapering in S. shinahota sp. nov.), the thoracic part narrower than the cephalic region in dorsal view and distinctly concave posteriorly in lateral view and the epigyne opening transversely elliptical. In the other two species with known males, the tibial apophysis of the male palp is either broader (S. lauretta) or narrower (S. edwardsi) than in S. shinahota sp. nov. In all four congeners, the abdomen is not or only indistinctly constricted anteriorly.
Description. Female holotype (Figs 3C, 4B, 5A-D). Total length: 4.25 mm. Carapace length: 2.27 mm; width: 0.90 mm. Integument slightly shiny, orange yellowish with dark bands around eyes of last three rows, dorsum with sparse, simple, moderately long, whitish-yellow setae, denser and longer on anterior half of cephalic area and posterior half of abdomen; without pubescence. Carapace slender and elongated, cephalic portion a little longer than wide (width: 0.85 mm), as wide as widest thoracic part, smooth, marked constriction (width 0.52 mm) between cephalic and thoracic part, two translucent white areas at each side of constriction, separated by narrow black area (imitating part of femora I), constriction followed by globular, somewhat wrinkled knob, which is terminated behind by short pedicle which is more slender than the anterior constriction, evenly tapering when seen in lateral view and concave in dorsal view. Eyes arranged in four rows, quadrangle formed by the second and fourth rows of eyes wider than long, anterior eyes large, in contact, occupying entire front of vertical inclination of face; second pair placed on back behind eyes of first pair, but a little further from each other than distance between external borders of first ones, directed sideways; posterior eyes larger, separated by the same distance as those of second row, eyes of the third pair halfway between second and fourth. Chelicerae light brown, with three teeth on promargin and two on retromargin. Anterior half of sternum yellow and posterior half light brown. Abdomen length: 1.80 mm; width: 1.10 mm, of same length as carapace, broader, commencing by short pedicle that appears to constitute prolongation of that of thoracic part, dorsally completely covered by scutum, integument smooth, constriction in anterior portion. Slender and comparably long legs, in order 4, 3, 1, 2, third and fourth pair stouter, light brown. Epigyne (Fig. 5A-C): epigynal plate forming part of the epigastric sclerite, copulatory opening in very small and quadrangular (width: ~ 0.05 mm), copulatory ducts long, starting in a chamber, forming a spiral with one loop and entering the spermathecae posteriorly; spermatheca lung-shaped; copulatory ducts anterior to spermathecae.   Male paratype (Fig. 9). Total length: 3.81 mm. Carapace length: 1.94 mm; width: 0.74 mm. Cephalic portion a little longer than wide (width: 0.71 mm), thoracic constriction 0.50 mm wide. General form of carapace, chelicerae and sternum as female, colouration dark brown. Abdomen length: 1.70 mm; width: 0.83 mm, covered dorsally and completely by a scutum, with a conspicuous constriction in the anterior portion; colouration dark brown. Palp (Fig. 9C-F): small cymbium, bulb about 80% of cymbium length, irregularly shaped spherical, embolus long, arising at the basal side of the bulb, without complete circular revolution and is accommodated on a slight retro-lateral concavity of cymbium, lacking pars pendula, tibial apophysis tooth-like, narrow, moderately tapering, tip directed forward.

Variation.
Juvenile females had a shiny, dark brown-blackish body surface with dark orange to light brown cephalic part and a pointed abdomen (Fig. 8A). The constriction in the proximal half of the abdomen was marked by a relatively broad, light transverse band. Adult females had a matt, blackish body surface with a dark orange or completely black cephalic part, the bright band in the abdominal constriction indistinct or absent (Fig. 8B, C). The holotype (BL 4.25 mm), collected in Cafetal, Buena Vista (Santa Cruz Department), had a matt, orange yellowish body except for some darker patches around the second and third eyes and the abdominal apex (Figs 3C, 4B). The abdominal constriction becomes less pronounced with increasing body length.
Etymology. The specific epithet, shinahota, refers to a place with many ants or an ant nest in the Yuracaré language (Querejazu 2005), spoken by the Yuracaré people living along the Chapare River in the Amazon Basin of Bolivia.
Geographical and ecological distribution (Fig. 7). Sympolymnia shinahota sp. nov. is known from Brazil, Amazonas State, Manaus and from the Bolivian Departments of Cochabamba (Villa Tunari) and Santa Cruz (Buena Vista). The collection locations of this species were situated in moist Amazonas forest regions, including Uatuma-Trombetas moist forest (Amazonas State, Manaus) and pre-Andean Southwest Amazon rainforest (Villa Tunari, Cochabamba Dept. and Buena Vista, Santa Cruz Dept.). However, in all areas that were surveyed in the present study, S. shinahota sp. nov. was exclusively collected in early successional forests in large tree-fall gaps or secondary forest from isolated, small trees that were densely overgrown with climbing plants, several metres away from the edge of primary forest. (Maddison 2018) suggest that this species also occurs in the Amazon rainforest in Ecuador (Yasuni National Park). Light variants were also reported in S. lauretta and S. lucasi (Galiano 1967). However, these reports refer to specimens that were stored in alcohol or formaldehyde for a longer time and may have faded due to the preservative (Galiano 1967). This study reports for the first time an Diagnosis. Sympolymnia cutleri sp. nov. and S. lauretta are indistinguishable in their somatic characters. Sympolymnia cutleri sp. nov. can be separated from all congeners by an epigyne with small, semi-circular opening (Figs 5F, G, 10D, E) (opening of epigyne very large and longitudinal-elliptical in S. lauretta).

Remarks. Photographs of two individuals
Description. Female holotype. Total length: 4.20 mm. Carapace length: 2.10 mm; width: 0.81 mm. Integument slightly shiny, dark brown, blackish, dorsum with sparse, simple, moderately long, whitish setae, denser and longer on anterior half of cephalic area and posterior half of abdomen. Carapace slender and elongated, cephalic portion as long as wide (width: 0.81 mm), as wide as widest thoracic part, smooth, marked constriction (width 0.62 mm) between cephalic and thoracic part, two translucent areas at each side of constriction, separated by narrow dark area (imitating part of femora II), constriction followed by globular, somewhat wrinkled knob, which is terminated behind by short pedicle which is more slender than the anterior constriction, evenly tapering when seen in lateral view and concave in dorsal view. Eyes arranged in four rows, quadrangle formed by the second and fourth rows of eyes wider than long, anterior eyes large, in contact, occupying entire front of vertical inclination of face; second pair placed on back behind eyes of first pair, but a little further from each other than distance between external borders of first ones, directed sideways; posterior eyes larger, separated by same distance as those of second row, eyes of the third pair halfway between second and fourth. Chelicerae light brown, with five teeth on promargin and three on retromargin. Anterior half of sternum pale yellow and posterior part dark brown, blackish. Abdomen length: 2.00 mm; width: 1.10 mm, of same length as carapace, broader, commencing by a short pedicle that appears to constitute a prolongation of that of thoracic part, covered dorsally and completely by a scutum, without constriction, smooth. Slender and comparably long legs, in the order 4, 3, 1, 2, first pair pale (leg I with dark longitudinal bands on anterior and posterior sides), third and fourth pair stouter, dark brown. Epigyne (Fig. 5F, G): epigynal plate forming part of the epigastric sclerite, with a small semi-circular opening, wider than long (width: ~ 0.17 mm); copulatory ducts starting in a small chamber (hard to see), forming a spiral and entering the spermathecae posteriorly; spermatheca lung-shaped; copulatory duct anterior to spermatheca, between them. Male unknown.
Variation. One immature (shiny surface, pointed abdomen) and five female adults (matt surface, rounded abdomen) (Fig. 8L) were collected, showing the same ontogenetic shift in integument shine and abdomen shape as observed in the other congeners.
Etymology. The specific epithet, cutleri, is a patronym in honour of Bruce E. Cutler in recognition of his contributions to the taxonomy of Simonellini.
Geographical and ecological distribution (Fig. 7). Sympolymnia cutleri sp. nov. is exclusively known from the type location in Bolivian Yungas forest.

Species richness and ecoregion distribution
This study provides the first records for the tribe Simonellini for the Departments of Cochabamba, Santa Cruz and Tarija and the ecoregions Chiquitano forest and Bolivian Tucuman forest. The presence of previouslyunrecorded or unknown species was not surprising, as the Bolivian spider fauna is little known (Cutler 1981;Perger and Perger 2017;Perger andRubio 2018, 2020), which is consistent with the sampling effort reported for other invertebrate groups (Perger andSantos-Silva 2010, 2018;Perger and Grossi 2013;Perger 2015;Perger and Guerra 2016). With three species, Bolivia and Brazil have the highest species richness of Sympolymnia amongst all of the Neotropical countries.
The distribution of Sympolymnia spp. appears to correspond to the delineation of humid Andes forest, Amazon rain forests and sub-humid, semi-deciduous forests (including Chiquitano forest, Bolivian Tucuman forest, Cerrado and Atlantic forest) (Fig. 7). This pattern sug- gests that the significant shift in seasonality and related conditions triggered diversification in Sympolymnia. For example, the Chiquitano forest ecoregion is distinguished by a pronounced seasonality and lower annual precipitation than Amazon forests (Ibisch and Mérida 2003). Accordingly, an essential factor contributing to the high species richness in Bolivia might be the meeting of several biogeographic realms and their corresponding faunal elements. High diversity in insect groups (Pearson et al. 1999;Kitching et al. 2001;Wappes et al. 2011) and Castianeirinae spiders (Perger and Perger 2017) in Bolivia was explained by high species turnover between a large number of ecoregions. Nevertheless, the comparably high richness of Bolivian species of Sympolymnia is likely partly attributed to low sampling efforts in Brazil, Ecuador, Peru, Venezuela and Colombia. Further taxonomic work and sampling campaigns are needed to clarify the species richness patterns of Sympolymnia.

Ant mimicry
The consideration of all ant species that were collected in the microhabitats of Sympolymnia species allowed a robust preliminary assessment of potential ant models. We identified at least four potential ant model species in three genera and subfamilies (Table 2). These species were amongst the most abundant ants in all microhabitats in which those resembling Sympolymnia spp. were collected (a subjective estimation).
Strong indirect support for ant mimicry was provided by: 1) species-specific similarity involving morphological aspects (Table 2, Fig. 8); 2) sympatry: both spiders and ants were found in the same microhabitats; and 3) mimic less abundant than the ant model. Alternative processes, such as convergent evolution, exploitation of perceptual bias, developmental or phylogenetic constraints, spatial autocorrelation, crypsis, or random matching (de Jager and Anderson 2019), were unlikely to be responsible for the resemblance between ants and spiders in the present case.
All suggested ant models are known for being well defended. The stinger in Crematogaster species is well developed, but the venom is applied topically by wiping on a victim instead of injecting it inside the body (Buren 1959). Camponotus species do not possess stingers, but are well protected by their powerful mandibles and the release of defensive chemical compounds (Fisher and Cover 2007).
Mimicry complexes involving adult polymorphic jumping spiders were described for Synemosyna aurantiaca (Mello-Leitão, 1917) (reviewed by Cushing 1997) and Myrmarachne in Australia (Pekár et al. 2017). Gilbert (2005) proposed that polymorphism in a mimetic species increases the protection against predation because it reduces the number of mimics per model. When a morph increases too much in frequency within the habitat, it may lose its mimetic protection and be exposed to greater predation (Gilbert 2005).
The present study is the first describing putative mimicry complexes involving ontogenetic shifts of ant resemblance (transformational mimicry) in Simonellini. It may be hypothesised that transformational mimicry commonly occurs amongst ant-mimicking spiders, assuming that juvenile spiders face at least the same level of predation pressure as adult spiders and considering that ant models have castes occupying a discrete mode in the size-frequency distribution. However, the genus Sympolymnia appears to include species with and without transformational mimicry, possibly depending on the maximum attainable body size or the presence of suitable ant models.
The reported individuals of S. lucasi (female holotype 3.66 mm; male allotype 3.3 mm, juveniles unknown) have about the same body size as Crematogaster-resembling juveniles of S. lauretta and S. cutleri sp. nov. Considering the small size, it is likely that S. lucasi lacks transformational mimicry and mimics exclusively Crematogaster ants. In the Bolivian species of Sympolymnia, larger body size and the lack of large Crematogaster species likely favoured the resemblance to different ant models.
Transformational mimicry involving Crematogaster and Camponotus ants was proposed for myrmecomorphic sac spiders of the genus Myrmecium Latreille, 1824 in the Amazon forest (Oliveira 1988). Amongst ant-mimicking jumping spiders, transformational mimicry was observed in several African (Edmunds 1978) and Australian (Ceccarelli 2010) species of Myrmarachne MacLeay, 1839 and may occur in the majority of Myrmarachne species (Wanless 1978). Amongst Neotropical jumping spiders, transformational mimicry is only known for Zuniga magna Peckham & Peckham, 1892. The juveniles of Z. magna resemble Camponotus ants, while the adult males imitate Pseudomyrmex gracilis (Fabricius, 1804) and the females Neoponera villosa (Fabricius, 1804) (Oliveira 1988). The limited knowledge of transformational mimicry in myrmecomorphic spiders, in general, is likely explained by the fact that many species are known only from adult individuals.

Conclusion and outlook
The presence of Sympolymnia throughout South America and the sympatry with Synemosyna species in many locations (e.g. Galiano 1966;Podgaiski et al. 2007;Rodrigues et al. 2016) suggests an early split and the subsequent radiation of the two lineages, accompanied by the selection for Pseudomyrmecinae-resemblance in adult Synemosyna and Myrmicinae-, Formicinae-and Dolichoderinae-resemblance in adult Sympolymnia.
In addition to similarity, sympatry of models and mimics is considered a critical factor in the selection for mimicry (de Jager and Anderson 2019). A more detailed analysis of cooccurrence patterns is needed to investigate the importance of polychromatism, transformational mimicry and mimicry complexes for ant mimicry in Sympolymnia. Additionally, the influence of macro-environmental conditions should be considered, as the distributional patterns of Sympolymnia species suggest that such factors may have contributed to generating divergent selection pressures in this genus.