A new species and the phylogeny of the South American genus Gromphas Brullé, 1837 (Coleoptera: Scarabaeidae: Scarabaeinae: Phanaeini)

ABSTRACT Gromphas jardim Cupello & Vaz-de-Mello sp. nov. is described from Bolivia and Central Brazil and is endemic to the biogeographic province of Rondônia. It resembles Gromphas amazonica and Gromphas inermis, but is distinguished primarily by the pattern of pronotal granulation and the form of the apical tubercle of male protibiae, which is modified in a tapered spur. Based on 31 morphological characters, a phylogenetic analysis returned a single most parsimonious tree where Gromphas is monophyletic and has the following internal topology: ((G. aeruginosa + G. lemoinei) (G. dichroa (G. inermis (G. amazonica + G. jardim)))).


Introduction
The South American dung beetle genus Gromphas Brullé, 1837 was the focus of a recent taxonomic revision presented by us Vaz-de-Mello 2013, 2014). In those works, we recognized Gromphas as a member of the subtribe Gromphadina of Phanaeini along with Oruscatus Bates, 1870 and composed by five species: Gromphas lemoinei Waterhouse, 1891, from the Llanos of Venezuela and Colombia, Gromphas aeruginosa (Perty, 1830) and Gromphas amazonica Bates, 1870, from the Amazon, Gromphas inermis Harold, 1869, from the Atlantic Forest and open environments (Cerrado, Chaco and Pampa) of the Southern Cone, and Gromphas dichroa Blanchard, 1846, from the Chaco and Pampa. All these species were redescribed and illustrated, and their taxonomy and nomenclature problems were discussed.
Yet an important question remained open regarding a specimen collected in the municipality of Cáceres, in the Alto Pantanal region of the state of Mato Grosso, central Brazil, and identified by us as a female of G. amazonica in Cupello and Vaz-de-Mello (2013, p. 463). On that occasion, we stated that Cáceres was an unexpected location for that species and distant more than 1000 km from the second southernmost point known for G. amazonica, Contamana, Peru. We also pointed out that, despite sharing several unique characteristics with the other examined specimens of G. amazonica (as, for example, the distinctly carinulate 1. Pronotum granulated and with mid-longitudinal line of smooth and glossy tegument at centre (clearly seen in {fig. 3}). Pronotal prominence formed by a pair of horizontal projections preceded by a pair of smooth concavities ({figs 1-4, 7-9}); in very small specimens, pronotal prominence absent and concavities reduced to two impressions in anterior margin of pronotum ({figs 5, 10}). Protibial spur only slightly curved apically ({figs 22, 46}). Genae and frons with smooth tegument adjacent to eyes; remaining tegument of genae and frons with squamose granulation (  Apical tubercle of male protibia modified in a short and tapered spur lacking any row of setae in its dorsal surface; apical tuft of setae independent of this spur (Figure 2A,B). Ventral carina of protibia simple in both sexes ( Figure 2A). Pronotal granulation, in lateral view, reaching posterior margin of pronotum ( Figure 3A). Cephalic projection narrow, with apex much narrower than distance between apices of apical lobes of clypeus ( Figure 1F).  Figure 1A-D, F). Aedeagus extracted and genital capsule glued in a triangular label and internal sac placed in a microvial with glycerine, all pinned with the holotype.   COCHABAMBA: Territory of the Yuracaré people ("Juacares Indians"), north side of the Cordillera de Cochabamba ("Cortillera de Cochabamba"), without date and collector (probably collected by Alcide d'Orbigny in 1832; see comments below) -1 male (BMNH).

Etymology
The specific name, a noun in apposition, is a patronym honouring Arlindo da Silva Jardim , Brazilian aviator and grandfather of the first author. Having grown up in the small, rural village of Dom Viçoso, Minas Gerais, Arlindo Jardim achieved his childhood dream and flew professionally worldwide for over four decades. He will remain as a source of inspiration for MC.

Description
Colour. Anterior region of clypeus black; remainder of head and pronotum with dark olive green and copper metallic reflections. Elytra, metasternum, ventral surface of legs and pygidium dark olive green with metallic sheen and silky appearance. Ventrites entirely black or black with week metallic green reflections. Head. Margin of clypeus with four lobes ( Figure 1F) and distinctly upturned. Genae and frons completely granulate, including region adjacent to eyes ( Figure 1F). Cephalic projection a raised carina with converging sides and emarginate apex in major specimens ( Figure 1G); apex narrower than distance between apices of apical lobes of clypeus ( Figure 1F).
Thorax. Pronotum convex; lateral region with dense granulation reaching the posterior margin ( Figure 3A), density of granulation decreasing posteromedially; posteromedian region smooth or with strongly effaced granulation ( Figures 1A, E, 4); posterior fossae apparent only as two very shallow and sometimes only weakly indicated impressions removed from the pronotal posterior margin ( Figures 1E, 4). Posterior margin of pronotum rounded. Mesosternum with dense pilosity. Metasternum with fine and sparse punctation at centre. Anteromedian angle of metasternum convex and with globose apex; area in front of anteromedian angle with evident setae.
Legs. Protibia slightly narrower in males than in females ( Figure 2); in ventral view, longitudinal carina simple in both sexes ( Figure 2A). Space between protibial lateral teeth deeper in males than in females ( Figure 2). Protibial spur with apex strongly expanded and curved downward ( Figure 2). Inner apical angle of protibia with a tuft of setae longer and denser in males than in females; in males, tubercle of inner apical angle developed as a short and tapered spur independent of apical tuft of setae ( Figure 2). Apical protarsomere with a long, distal spiniform prolongation. Mesotarsi and metatarsi with apical tarsomere slightly curved at apex. Metatibia very broad and robust. Metatibial spur with apex distinctly curved. Elytra. Striae very fine and, especially striae 1-4, carinulate from base to half or apical two-thirds of elytra. Sutural margin glossy and only sparsely punctate; basal half of sutural margin with sheen extending laterally onto first or second interstria.
Abdomen. Pygidium lacking basal margin and with irregular sculpture. Groove of propygidium extending to base of pygidium. Abdominal sternites microsculptured and sparsely punctuated.
Aedeagus. Apex of phallobase, in ventral view, with membranous area expanded triangularly in the middle ( Figure 1C). Medial sclerite only slightly curved, almost flat.

Intraspecific variation and taxonomic discussion
At a first glance, G. jardim resembles superficially G. amazonica and, to a lesser degree, G. inermis, and, in fact, has been confused with these species both in collections and recent publications. Deposited at the BMNH, the oldest specimen known to us bears four labels with different identifications (Figure 4): an older, which by the calligraphy we assign to Charles O. Waterhouse, former curator of entomology at the BMNH, has written 'Gormphas amazonicus Bates', while the other two more modern labels identify that specimen, respectively, as G. amazonica and 'G. lacordairei Brullé, 1834ʹ, an unavailable name referring to G. lacordairii Burmeister, 1874, junior synonym of G. inermis (see more in Cupello and Vaz-de-Mello 2013). The fourth label has handwritten the word 'Coproides', but the remaining information is effaced and completely unreadable. d'Olsoufieff (1924) examined a specimen of G. amazonica in the Muséum national d'Histoire naturelle, Paris, labelled 'coproides Dej. Cayenne (coll. Mniszech)' and probably the unavailable name 'coproides' was used before the description of G. amazonica by Bates (1870) as a name in litteris to refer to this species. The three specimens found in OUMNH, in turn, including the holotype, are part of a large series of dung beetles recently collected in Bolivia and the basis for the works of Hamel-Leigue et al. (2006, 2009; they were identified and illustrated in these publications as 'G. lacordairei Brullé'. The geographical distribution and probably the other information present for 'G. lacordairei Brullé' in Hamel-Leigue et al. (2006, 2009 have mixed data belonging in fact to G. jardim and G. inermis. Similarly, as said in the Introduction of the present work, in Cupello and Vaz-de-Mello (2013), we provisionally identified the specimen (now paratype) from Cáceres as a G. amazonica. Now, in possession of a greater number of specimens, the differences between G. jardim, G. inermis and G. amazonica became much clearer.
Gromphas jardim shares only with G. amazonica, G. inermis and G. dichroa the characters: genae and frons granulated adjacent to eyes ( Figure 1F), absence of pronotal prominence, protibiae narrower in males than in females (Figure 2), and protibial spur expanded at apex ( Figure 2); only with G. amazonica and G. inermis, G. jardim shares the character margin of clypeus with four lobes ( Figure 1F). Probably this last characteristic, which is an apomorphy shared by them (see the phylogenetic analysis below), was the main cause for the past misidentifications.
Yet G. jardim is easily differentiated from G. inermis by having metatibial spur distinctly curved apically (straight in G. inermis), posterior margin of pronotum rounded (projected at middle in G. inermis), elytral striae carinulate (simple in G. inermis), and metasternum and sutural margin of elytra with fine and sparse punctation (dense punctation in G. inermis); furthermore, pronotal hump and sutural margin of elytra raised are present in major specimens of G. inermis but absent in G. jardim ( Figure 1B). On the other hand, the medial sclerite of the internal sac of G. jardim is very similar to that of G. inermis and no significant difference between them was found ({fig. 59}).
From G. amazonica, G. jardim is differentiated most easily by the shape of the apical tubercle of male protibia, which, although much more developed in G. jardim than the tiny and almost imperceptible tubercle of the other four species of Gromphas, is still much smaller than that of G. amazonica; in G. jardim, the tubercle has the shape of a tapered spur and is separated from the apical tuft of setae, which rests adjacent to the spur (Figure 2A,B); in G. amazonica, the spur is long, laterally flattened and curved and has the tuft of setae on its dorsal surface as a row of setae. The shape of the cephalic projection of G. jardim is similar to that of G. inermis, i.e. it is narrower than the distance between the apices of the apical lobes of clypeus ( Figure 1F), while that of G. amazonica has the equivalent width of that distance. Other differences between G. jardim and G. amazonica are: the colour, which is dark olive green and has metallic reflections in G. jardim, but black, dark blue, dark green or reddish-brown and never has metallic reflections in G. amazonica; and the pronotal granulation, which penetrates more the posterior portion of the pronotum and, in lateral view, reaches the posterior margin in G. jardim ( Figure 3A), whereas in G. amazonica the granulation is restricted to the anterior portion of the pronotum and never reaches the posterior margin ( Figure 3B). The form of the granules of the head and pronotum is also distinct between the two species, being wider and flattened in G. amazonica and more rounded and smaller in G. jardim (this second form is very similar to that of G. inermis). Finally, the longitudinal carina of the ventral surface of protibia is simple in both sexes of G. jardim, resembling G. aeruginosa and G. lemoinei, but is distinct to that of G. amazonica, G. inermis and G. dichroa, which, in males, has a row of tubercles on its basal half and, in females, is simple. The constant presence of the posterior pronotal fossae in G. jardim also distinguishes this species from G. inermis and G. amazonica, in which these fossae are usually absent.
The spiniform projection at the apex of apical protarsomere was not observed in one of the three females of G. jardim examined by us, and we believe that this is due to the wear, as happens in some G. amazonica, the only other species of Gromphas that has this kind of apical protarsomere (Cupello and Vaz-de-Mello 2013). On the other hand, the nature of the posterior pronotal fossae varies: in the two males observed, the fossae are clearly marked and easily visible to the naked eye, whereas those of the three females are much less marked and almost imperceptible. Whether this is a case of individual or sexual variation is difficult to say until the examination of a larger number of specimens of both sexes.

Comments
While the holotype and the three female paratypes of G. jardim were collected in the 21st century and have label information detailed enough to permit an easy understanding of their origin, the male paratype is much older and has a puzzling history. Only one of the five labels attached to this specimen before our work has information about its provenance (Figure 4). This label is circular and has 'Bolivia' written on one side and '46/76ʹ, on the other side. According to Max Barclay (pers. comm.), '1846-76 refers to a collection, all with the same data, acquired in 1846 and including 325 Coleoptera and 250 Lepidoptera', and these collecting data are 'Territory of Juacares Indians (north side of the Cortillera (sic) de Cochabamba)'. We believe that 'Juacares Indians' refers to the Yuracaré, an indigenous people resident on the north side of the Cordillera de Cochabamba, in the department of Cochabamba, Bolivia. To our knowledge, the only European naturalist who crossed this remote region before 1846 was the French zoologist and explorer Alcide d 'Orbigny (1802'Orbigny ( -1857, who visited a Yuracaré village on May 28 1832 and stayed there for 4 days (Papavero 1971). Indeed, d'Orbigny described and illustrated in detail this people in his great work Voyage dans l'Amérique Méridionale (d 'Orbigny 1835'Orbigny -1847. So we believe he was the probable collector of the male paratype of G. jardim. This finding is also interesting because, if correct, it indicates that not all insects collected by Alcide d'Orbigny are deposited in the Muséum National d'Histoire Naturelle, Paris, France, as suggested by Horn and Kahle (1936) and Evenhuis (1997).

Bionomics
The label data indicate that the holotype and the two paratypes from Beni, Bolivia, were collected in traps baited with human faeces in open habitats. These specimens also had some unidentified phoretic mites attached to their legs, especially to the metatarsi. The recorded months for G. jardim are May and October.

Comments
Although only nine specimens of G. dichroa were examined in Cupello and Vaz-de-Mello (2013), we recognized a wide colour variation in this species. Now, having examined the MACN collection, we found other seven specimens collected between 1926 and 1927 in Santo Tomé, Corrientes, Argentina, a locality already recorded for this species. This series is very interesting because it shows for just one locality much of the colour variation described by us earlier, including specimens that are entirely green, with green pronotum and dark blue elytra, and with pronotum metallic green on sides and along posterior margin and greenish-red in centre and dark blue elytra. This finding supports our hypothesis that colour varies individually within at least some populations, and it is not necessarily a matter of geographic variation nor evidence for recognizing more than one taxa. Yet, as pointed out by one of the anonymous reviewers of this work, 'the fact that the colour is subject to wide variation in one of the studied populations does not implicate that this variation necessarily occurs in the other populations of this species.
There is the probability that some populations bear just a single or two colours'. We agree that this is a possible scenario and so more collecting is needed for a fuller understanding of colour variation in G. dichroa.
The female from the Hermann Burmeister collection has one of its tarsi complete (i.e. it is not broken as in all the other females examined) and it is identical to the tarsus of G. inermis: the apical protarsomere is short and lacks an apical spiniform prolongation. inermis is present. The occurrence of this species in the Argentine province of Mendoza, as first reported here, is the most western record known for G. inermis ( Figure 5). During our examination of specimens for the revision of Gromphas, we found a female from Artigas, Uruguay, that is very interesting by being matte black and having the pronotal surface completely irregular; nonetheless, taking into account all the other characteristics, it fits in G. inermis, and so we identified that specimen with this name in Cupello and Vaz-de-Mello (2013). Since then, we examined the Hermann Burmeister collection, housed in MACN, and there we found two other Uruguayan females (from 'Banda Orient.') with a very similar colour and irregular pronotum that were identified by Burmeister with the unavailable name 'Gromphas rugicollis Nob.'. For now, we consider these differences as intraspecific variations of G. inermis (possibly teratological specimens); we observed black specimens of G. inermis from other regions and, in fact, black and brown colorations seems to be related to teneral specimens, which would also explain the irregular surface of pronotum. All the other Uruguayan specimens examined by us are typical G. inermis.

Phylogenetic analysis
The phylogenetic analysis performed here had three main goals: to test the monophyly of Gromphas, to know the phylogenetic relationship of its species, and, therewith, to raise hypotheses about the transformation of their characters through the evolutionary process. In recent works (Philips et al. 2004;Cupello and Vaz-de-Mello 2013), it was well demonstrated that Gromphas is a member of the tribe Phanaeini and closely related to Oruscatus; together, these two genera form the monophyletic subtribe Gromphadina. The monotypic genus Bolbites Harold, 1868, previously considered to be related to this clade, was transferred to Phanaeina by us (Cupello and Vaz-de-Mello 2013) based primarily on the results of the phylogenetic analysis of Philips et al. (2004;morphological data) and of Ocampo and Hawks (2006;molecular data), besides other published information about the nesting behaviour of these three genera (e.g. Cabrera-Walsh and Gandolfo 1996;Halffter and Edmonds 1982;Sánchez and Genise 2008). In Phanaeina, Bolbites is sister of the lineage comprising the remaining genera of the subtribe (Philips et al. 2004;Ocampo and Hawks 2006). Taking these information into account, we selected as outgroups one of the two species of the undoubtedly monophyletic Oruscatus, Oruscatus davus (Erichson, 1847), and one Phanaeina, Bolbites onitoides Harold, 1868, the latter used to root the tree and selected among the Phanaeina by its close morphological similarity with Gromphadina.
Based on 31 informative morphological characters (see below and Table 1), an exhaustive search (implicit enumeration) was conducted using the program TNT (Goloboff et al. 2008) with Fitch parsimony and equally weighted characters, which resulted in a single tree (Figure 6) of length 46, consistency index 78 and retention index 77, both indices calculated in WinClada (Nixon 1999-2002. Supports of clades were calculated in TNT using both Bootstrap and decay index (Bremmer support) values, this last index based on a search of the 88 suboptimal trees with up to 10 steps longer than the optimal tree. The most parsimonious tree was edited in WinClada, where unambiguous characters were plotted. The ambiguous characters, on the other hand, were individually optimized based on scenarios considered more likely by us and discussed throughout the text below.

The monophyly of Gromphas
In our analysis, the monophyly of Gromphas was strongly supported by eight uncontroverted synapomorphies: outer sides of pterothorax (mesepisternum, metepisternum and sides of metasternum) entirely glabrous (char. 11-1; Figures 1B, 7A), metepisternal tab absent (12-1; Figure 7A), sides of anteromedian angle of metasternum with granulose tegument (13-1); {figs 49-52}), centre of metasternum densely punctate (15-1), inner apical angle of male protibiae with a tiny tubercle (21-1; {fig. 23b}), elytral striae without basal fossae (24-1), epipleura strongly curved and wide from base of elytron to metacoxa (28-1), and ventrobasal margin of phallobase with median incision (30-1; {figs 53-56}). The high support values (bootstrap 99, decay index 7) also give us great confidence in the monophyly of Gromphas. Edmonds (1972) and Philips et al. (2004) stated that the metepisternal tab was absent in Gromphas and Oruscatus; the first author considered the presence the metepisternal tab as one of the defining characteristics of the phanaeines when these two genera are excluded. However, as we observed in Cupello and Vaz-de-Mello (2013), the two species of Oruscatus have, in fact, a short tab in their metepisterna covering the margin of the elytral epipleura, which has an evident depression at this point to receive the tab, in the same way as in Phanaeina ( Figure 7B). We observed this same short metepisternal tab and the respective epipleural depression in a series of other dung beetle genera, including species of Copris Geoffroy, 1762, Dichotomius Hope, 1838 ( Figure 7D), Chalcocopris Burmeister, 1846, Homocopris Burmeister, 1846, Ontherus Erichson, 1847, Canthidium Erichson, 1847, and Isocopris Pereira & Martínez, 1960, in the tribe Coprini, Helictopleurus d'Orbigny, 1915, Liatongus Reitter, 1893and Euoniticellus Janssens, 1953, in Oniticellini, and Bubas Mulsant, 1842 in Eucraniini, the putative sister tribe of Phanaeini (Philips et al. 2004;Tarasov and Génier 2015), we have not seen this structure. Nonetheless, the metepisternal tab seen in Phanaeina is unique among the dung beetles by its great development, being much more curved and longer than in any other group observed by us ( Figure 7C). Therefore, we consider that this unique form, rather than its mere presence, is one of the defining synapomorphies of Phanaeina. In its turn, Gromphas is exclusive among the Phanaeini in that it has no trace of a metepisternal tab ( Figure 7A). The epipleura, however, has a perceptible vestigial depression at the same point as in Oruscatus and the other Phanaeini, indicating again that a metepisternal tab could have been present in the ancestors of the Gromphas. Edmonds (1972, pp. 814-815) hypothesized that the metepisternal tab anchors the elytra, maintaining 'close elytral appression along the elytral suture by restricting lateral slippage of the elytra'. Giving assistance during digging, probably this is the same role that the metepisternal tab has in those other dung beetles, which are fossorial and paracoprid.
The presence of only four protarsomeres, rather than five, is also unique to Gromphas, at least among the Phanaeini. In this tribe, protarsi, if present, are found only in females, but 'are always reduced in size and clawless' (Edmonds 1972, p. 770). In some groups, protarsi are completely absent. Bolbites and Oruscatus are included in this latter case and, in consequence, it is difficult to say if the condition seen in Gromphas (i.e. protarsi with four tarsomeres) was the ancestral condition of Gromphadina and then Oruscatus took a step further and lost completely the protarsi, or if the condition in the last common ancestor was the five-articulated protarsi and then the reduction occurred in Oruscatus and Gromphas independently. So, in the first case, protarsi with four tarsomeres would be a synapomorphy of Gromphadina, whereas in the latter case it would be a synapomorphy of Gromphas. Given these difficulties in the ambiguity of this character, we chose not to include it in the analysis.
One of the most interesting features of this clade is the reduced sexual dimorphism. In G. aeruginosa and G. lemoinei, the ventral carina of protibiae is simple in both sexes ({fig. 22}), whereas it is ornamented with tubercles or teeth in Oruscatus ( Figure 8B), Bolbites ( Figure 8C) and in three of the other four Gromphas (G. jardim is the exception) ( Figure 8A). In the other Phanaeina, this carina is also simple in both sexes. Other reductions in sexual dimorphism are the protibiae very broad and the protibial tuft of setae of same length in both sexes. In the other Gromphas (and Oruscatus and Bolbites), males have narrower protibiae and the tuft of setae is much longer and denser in males than in females (Figure 2, {fig. 23}). The presence of a pronotal prominence is also a remarkable characteristic of this clade, being absent in the other Gromphadina; nonetheless, as is typical for these two species, there is no sexual difference in this feature. In Phanaeina, in general, males have very elaborate ornamentation on head and pronotum, and it has been known since Darwin (1859Darwin ( , 1871 that these structures play a key role in the sexual access to females by males and, therefore, are under constant pressure from sexual selection (see, for example, Otronen 1988;Rasmussen 1994;Escobar 2003;Emlen and Philips 2006;Rowland and Emlen 2009). But what would be the reason for these structures being as developed in females as in males in this clade? A few other groups of Phanaeini also have females with ornamentation greatly developed, as in males, including those of the group lancifer of subgenus Megaphanaeus d 'Olsoufieff, 1924, of Coprophanaeus d'Olsoufieff, 1924, and, to a lesser extent, the group faunus of Sulcophanaeus d'Olsoufieff, 1924, groups that include some of the largest dung beetles of the New World (Edmonds 2000;Edmonds and Zidek 2010). Otronen (1988) supposed that females of Coprophanaeus (M.) ensifer (Germar, 1821) need a developed armature to face combat against other females and so obtain enough food for their large larvae; larvae of smaller dung beetles, on the other hand, do not need large amounts of food and therefore females of these species do not need to face such fierce disputes. Not having such a large size, females of G. aeruginosa and G. lemoinei are unique in possessing well-developed ornamentation. Only with a greater knowledge of their biology will we be able to answer this question more firmly, but it is possible that some idiosyncrasy in their behaviour leads females to fight battles and so be in need of such armament.
The second main lineage within Gromphas, clade B, is supported by five synapomorphies (three uncontroverted, two ambiguous): cephalic projection as a carina with base wider than apex (char. 5-1; ambiguous; {figs 31-33}), pronotum globular (10-1), ventral carina of male protibiae with a row of tubercles on basal half (17-2; ambiguous; {fig. 23b}), protibial spur strongly expanded at apex (18-1; {fig. 23}), and, on basal third or basal half, sheen and punctation of sutural margin of elytra extend onto first or second interstriae (27-1; {figs 13-15}). The values of support for this clade, however, are the lowest in our analysis (bootstrap 55, decay index 2). One interesting synapomorphy of this lineage is the form of the ornamentation of the ventral carina of male protibiae. In Bolbites, this carina has few tiny, irregular teeth and one large central tooth ( Figure 8C); in Oruscatus, this ornamentation is modified into a single central tooth without any accessory teeth or tubercles (Figure 8B). In G. dichroa, G. inermis and G. amazonica, the carina has a row of tiny, regular tubercles, with no apparent difference between these species ({fig. 23b}); in G. jardim, as a controverted autapomorphy (char. 16-1, homoplastic with clade A), this ornamentation is lost and the carina is simple and continuous (Figure 2A).
The next clade, clade C, is supported by five synapomorphies (two uncontroverted, three ambiguous): apical margin of clypeus evidently upturned (1-1), lateral margin of clypeus lobate (3-1; Figure 1F), cephalic carina emarginate apically (6-1; ambiguous; Figure 7), pygidium without basal margin (29-1; ambiguous and homoplastic with clade A; {fig. 35}), and medial sclerite of internal sac wider and only slightly curved (31-1; ambiguous) ({figs 59, 60}). The support values give us confidence in this hypothesis (bootstrap 76, decay index 2). This lineage shares with G. aeruginosa and G. lemoinei one important condition, the basal margin of pygidium completely absent ({fig. 36}). Gromphas dichroa, in its turn, possesses the basal margin complete as Bolbites and the other Phanaeina ({fig. 35}). The two species of Oruscatus apparently have an intermediate condition: the basal margin is present, but in the majority of specimens it is interrupted in the middle by the sulcus of propygidium. So, albeit complete in G. dichroa, we believe that there is a general tendency in Gromphadina for the loss of this basal margin, perhaps linked to the great development of the propygidium (which has the length of the pygidium in Gromphadina, and is shorter than the pygidium in Phanaeina), and, therefore, we judge that the most likely scenario was the parallel and independent loss in both lineages rather than a unique loss in the ancestral Gromphas and a new origin in G. dichroa. The form of the medial sclerite of the internal sac of this lineage is very different to that of clade A and G. dichroa and, at same time, homogeneous between its species ({figs 59, 60}). As this sclerite is absent both in Bolbites and Oruscatus, it was not possible to properly polarize this character and, given the topology of the tree, it remained ambiguous in our analysis.
The last clade includes G. amazonica and G. jardim, sharing four apomorphies (two uncontroverted and two controverted): centre of metasternum with sparse punctation (15-2; homoplastic with G. aeruginosa), tubercle of apical inner angle of protibiae modified in a strong spur (22-1; Figure 2, {fig. 21b}), apical protarsomere with spiniform projection at apex (23-1; {fig. 45}), and elytral striae carinulate (25-1; homoplastic with clade G. aeruginosa + G. lemoinei;{fig. 38}). The support values are also robust (bootstrap 78, decay index 3). The most remarkable synapomorphy of this lineage is the protibial tubercle developed into a strong spur, which has no parallel within the genus; the great differences between the two species regarding the form of this spur are discussed in the description of G. jardim above.

Phylogenetic conclusion
Through this phylogenetic analysis, we observed that features related to tegument [e.g. punctation (chars. 17 and 26) or carinulae of elytral striae (char. 25)], which are of a great importance for species identification, were the most error-prone characters for the construction of primary homologies and, so, led to more homoplasies (see Nixon and Carpenter 2012 on homoplasy as error). The differences in pronotal granulation were also very difficult to codify and to include in our analysis because of their gradual variation between species. In fact, there is no pair of species of Gromphas with the same pattern of pronotal granulation. For the same reason, codifying the differences in the form of the anteromedian angle of metasternum was virtually impossible. Nonetheless, in this latter case, there is a clear homogeneity among the forms founded in G. amazonica, G. jardim and G. inermis ({figs 50, 51}), on one side, and in G. aeruginosa and G. lemoinei ({fig. 52}), on the other. The form seen in G. dichroa, which is high, narrow and truncate apically, is unique in the genus ({fig. 49}). Finally, the usual apomorphic absence of posterior pronotal fossae in G. inermis and G. amazonica appeared ambiguous in our analysis, with two possible scenarios: the loss of the fossae in the ancestral of clade C and its subsequent reappearance in G. jardim, or independent losses in G. inermis and G. amazonica. In this case, we have no opinion on which scenario would be more likely and therefore we have not plotted this character on the tree.