Sylvicanthon Halffter & Martínez, 1977

Cupello, Mario & Vaz-De, Fernando Z., 2018, A monographic revision of the Neotropical dung beetle genus Sylvicanthon Halffter & Martínez, 1977 (Coleoptera: Scarabaeidae: Scarabaeinae: Deltochilini), including a reappraisal of the taxonomic history of ‘ Canthon sensu lato’, European Journal of Taxonomy 467, pp. 1-205 : 16-59

Genus Sylvicanthon Halffter & Martínez, 1977

Sylvicanthon Halffter & Martínez, 1977: 36 , 45, 61–63.

Sylvicanthon – Halffter & Edmonds 1982: 139. — Martínez 1987: 47. — Kohlmann & Halffter 1990: 8. — Hanski & Cambefort 1991: 472. — Medina & Lopera-Toro 2000: 301, 311. — Vaz-de-Mello 2000: 186, 195. — Escobar 2000a: 200, 210. — Medina et al. 2001: 133. — Solís & Kohlmann 2002: 2. — Halffter 2003: 31. — Medina et al. 2003: 25, 29–30, 36, 38–41, 45, 59, 62. — Durães et al. 2005: 724. — Scheffler 2005: 14. — Hamel-Leigue et al. 2006: 3. — Medina & Pulido 2009: 56. — Scholtz et al. 2009: 567. — Carvajal et al. 2011: 99, 117, 316. — Vaz-de-Mello et al. 2011a: 6, 11, 19, 26, 33, 41. — Krajcik 2012: 63 (as junior synonym of Canthon ). — Solís & Kohlmann 2012: 3 (as junior synonym of Canthon ). — Boilly & Vaz-de-Mello 2013: 107. — Medina et al. 2013: 468, 473. — Noriega et al. 2015: 101. — Feer 2015: 8. — Pacheco et al. 2016: 145. — Tarasov & Dimitrov 2016: 15. — Chamorro et al. 2018: 76, 98. — Espinoza & Noriega 2018: 147, fig. 2.

Silvicanthon [sic] – Noriega 2004: 39.

Type species

Canthon candezei Harold, 1869 , by original designation ( Halffter & Martínez 1977: 62).

Diagnosis

The following combination of characters is found uniquely in Sylvicanthon : small- to medium-sized species usually with dark colouration (black, dark blue, dark green or purple, but also light green and coppery) and, sometimes, with metallic sheen, and general body shape oval, flat and compact; clypeus with two small, contiguous at base or only slightly separated apical teeth, never with emargination between them (group candezei , 14 species, Fig. 6 B–G), or clypeus with four large teeth, but also without emargination between them (group enkerlini , a single species, Fig. 14 ); external margin of genae with an evident denticle adjacent to the junction with clypeus; dorsal surface of eyes large (width about one fifth of that of interocular space); pronotum with lateral edges forming a strong medial angle and without prescutellar depression; anterior region of hypomera strongly excavated and delimited posteriorly by a complete hypomeral carina ( Fig. 35 ); external margin of anterior region of hypomera with a very small tubercle or simple, without tubercle; mesoventrite horizontal and very short (Fig. 54); mesocoxae with

very broad external margin ( Figs 21–22 ); tarsal claws with a very rudimentary basal angle ( Fig. 10 ); protibiae with two or three teeth and with internal margin expanded or straight ( Figs 33–42 ); second metatarsomere equal to or longer than basal tarsomere; metafemur margined anteriorly on its ventral face (posterior margin present or not) ( Figs 46 –53, 114–115); elytra without scutellar depression and with very narrow, usually ill-delimited striae of variable number; pygidium and propygidium separated by a very fine, medially angulate carina.

Etymology

Masculine, from the Latin word ‘ silva ’, meaning forest ( Brown 1956), and the Greek ‘ canthon ’, an insect name ( Harold 1869b; Pereira & Martínez 1959). Although this was not explicitly stated by them, it is likely that Halffter & Martínez (1977) chose this name inspired by the fact that species of Sylvicanthon are typical inhabitants of New World tropical rainforests.

Redescription

HEAD. Clypeus with two tiny apical, obtuse or slightly acute teeth without emargination between them ( Fig. 6 B–G; group candezei ) or with four large, acute teeth also without emargination between them ( Fig. 6A ; group enkerlini ); a single row of setae covering teeth’s base (except in S. securus , which has an individual row of setae for each tooth); apical margin of clypeus clearly or only slightly bent upwards. Genae with strong or rudimentary tooth after junction with clypeus ( Fig. 6 ). Clypeo-genal suture complete, fine and well impressed; fronto-clypeal suture indistinct in the middle and present only on the sides. Dorsal portion of eyes large – interocular distance four times the largest width of eyes – and without internal margin. Posterior edge of head with or without a margin between the eyes. Tegument of dorsal region of head varying from smooth or with evident micropunctation to strong three-dimensional alveolar microsculpture obliterating micropunctation. Maxilary and labial palps with three palpomeres. Labium distinctly ‘U’-emarginate on its anterior edge ( Fig. 8 ). Suture between mentum and submentum rounded ( Fig. 8b ) or in a deep ‘Y’ shape ( Fig. 8a ).

THORAX. Pronotum margined only anteriorly and laterally, with lateral edges strongly angulated in the middle; posterior edge unmargined and, in most of the species, with a fine transverse line at the centre (usually, extending up to second elytral stria); tegument ranging from smooth, bright and with strong micropunctation to with strong three-dimensional alveolar microsculpture, diffuse shinning and obliterate micropunctation; lateral fossae and prescutellar depression absent. Hypomera divised into two parts by a complete transversal hypomeral carina; anterior part strongly excavated and delimited posteriorly by a low vertical wall; tegument with strong rivose microsculpture, glabrous at centre ( Fig. 35 A–B) or covered by long and dense erect yellow setae ( Fig. 35 C–D); external edge, in general, with a tiny tubercle or, sometimes, simple, without tubercle; posterior part with tegument strongly microsculptured; entirely glabrous (group candezei ) or with a few long, individual setae aligned longitudinally close to external edge ( Fig. 9D ; group enkerlini ). Mesepimeron, metepisternum and prosternum glabrous and with microsculptured tegument. Mesoventrite very short and horizontal ( Fig. 14A ); in the middle, with a triangular projection towards body’s anterior region which has a very shallow fovea in its centre (in general, projection covered by prosternum in pinned specimens); tegument with micropunctation and with a transverse row of very short setae. Meso-metaventrite margin very fine and straight. Metaventrite very wide between mesocoxae and slightly convex; tegument with strong rivose microsculpture on the sides and on anterior region between mesocoxae; at centre, tegument ranging from microsculpture absent to well-marked micropunctation to strong three-dimensional alveolar microsculpture with discrete micropunctation.

LEGS. Tarsal claws of all legs slightly angulate at base (more clearly so in protarsi, Fig. 5A ). Profemora with tegument always covered by microsculpture (rivose or alveolar); anterior margin, at apex of profemora, smooth ( Fig. 9A ; group candezei ) or with a row of denticules in both sexes ( Fig. 9B ; group enkerlini ). Protibiae truncate at apex, narrow or wide, and with internal margin straight or strongly expanded at its apical half; with two or three teeth on external margin varying in size from large, wide, and well separated from one another to small, narrow, and restricted to apical third of protibiae; external edge of protibiae covered by smaller denticules, including area between larger teeth ( Fig. 11 ). Dorsal face of protibiae with four longitudinal lines: the most external one very fine, entirely glabrous and, in general, indistinct at the apical expansion; the second most internal one covered by a row of short setae only at its apex; the third one covered by longer setae from its base to apex; the fourth line (the most external one) strongly carinate, entirely glabrous and interrupted at the basal lateral teeth. Ventral surface of protibiae with two longitudinal lines, one parallel to the internal edge and the other at the middle, both interrupted by setae; the latter line keeled at its basal half. Protarsi with five tarsomeres; basal and apical tarsomeres at least as long as tarsomeres II–IV combined. Mesocoxal cavities with wide external margin ( Fig. 7 A–B) and narrow internal margin. Meso- and metatrochanters with a fine tuft of long yellowish setae turned back at their posterior edges. Meso- and metafemora very flattened. Mesofemora margined anteriorly at its basal two thirds; non-margined area with a row of very short setae; posterior margin absent. Metafemora with ventral face always margined anteriorly ( Fig. 13 ), posterior margin present only in the aequinoctialis subgroup (Fig. 114–115); apical half of anterior edge covered by row of setae; micropunctation present throughout the tegument; at base, with or without coarse elongate punctation. Metatibiae curved, metatibial spur straight. Meso- and metatarsi flattened. Metatarsomeres II and V subequal in length and longer than the other tarsomeres; metatarsomere IV shorter than the others; meso- and metatarsi with a continuous row of setae throughout its internal edge.

ELYTRA. Scutellar depression absent. With nine to seven visible striae; from elytral suture, striae progressively more effaced; humeral carina present ( Fig. 12C ) or not. Tegument of interstriae ranging from smooth with evident micropunctation to with strong three-dimensional alveolar microsculpture

obliterating micropunctation. Epipleural carina well marked; epipleura slightly sinuous, almost straight, in lateral view, and with tegument with strong rivose microsculpture.

ABDOMEN. Six visible abdominal segments (ventrites); tegument of ventrites I–V with strong rivose microsculpture; ventrite VI with very diffuse rivose microsculpture at middle and more clearly marked on the sides; micropunctation absent or very subtle. Pygidium curvilinear, subtriangular and at least slightly convex in both sexes; entirely margined, apical margin much wider than lateral margin in both sexes (usually wider in males); margin between pygidium and propygidium arched; tegument with alveolar microsculpture always present, but ranging from strong to diffuse.

AEDEAGUS. Parameres shorter or as long as phallobase and very variable in shape, with external faces symmetrical or asymmetrical; when asymmetrical, external face of right paramere flat and external face of left paramere concave. In lateral view, parameres with or without ventral keel or notch and with simple or bifurcate apices. Basal piece of phallobase always with depressed dorsal area.

SEXUAL DIMORPHISM. Males: Protibial spur ( Fig. 15 ) narrow or wide, apex with a profound notch or straight with two lateral spines of unequal length. Ventrite VI strongly narrowed at middle by emargination on its posterior edge ( Fig. 14D ); anterior margin covered by weak medial expansion of ventrite V, or ventrite V without medial expansion. Abdomen without lateral foveae. Females: Protibial spur spiniform ( Fig. 15F ) (except in S. proseni , whose spur is bifid, Fig. 15H ). Ventrite VI very wide at middle, posterior edge straight, without emargination ( Fig. 14 B–C, E); anterior margin distinctly covered by medial expansion on the posterior margin of ventrite V, or ventrite V without medial expansion. Abdomen with or without three pairs of lateral foveae between ventrites I–II, II–III and III–IV (Fig. 74–77).

Taxonomic history of Sylvicanthon

The first species of what was to become Sylvicanthon was described by Harold (1868a) in his landmark revision of the genus Canthon , namely C. aequinoctialis Harold, 1868 , from “ Columbien, Neu-Granada ”. In the following year, Harold (1869a) described two other new species for the genus, including Canthon candezei Harold, 1869 , from “ Tapajos ”, Pará, Brazil, another species currently included in Sylvicanthon .

The next author to describe species of Sylvicanthon was Schmidt (1920), in his first paper on Canthon . There, he presented Canthon furvus , from “ Peru, Bolivien ”; C. securus , from “S urinam ”; and C. obscurus and C. foveiventris , both from “ Espirito Santo ”. After having described them, Schmidt (1920) asserted that they were close to C. candezei and that, besides colour, those five species were similar in their oval body shape and by the presence of protibiae with a truncate apex and small, apical lateral teeth, clypeus with only two tiny teeth, large eyes, hypomeron with a transverse carina (wrongly cited as “ Prosternum ”, common misunderstanding among old works on Scarabaeinae , as firstly noted by Halffter 1958, 1961, but who, in turn, misinterpreted them as being the propleura) and metafemora with anterior margin. Finally, Schmidt (1920) proposed a key to differentiate those five species. It is worth noting that this group and its unique character combination was an early draft of what would eventually be proposed as the genus Sylvicanthon .

As explained above, in his second work on the genus, Schmidt (1922) updated Harold’s key to the species of Canthon and, there, he grouped C. aequinoctialis with C. candezei and the four species he had described in 1920 under the following steps: 2. Metatibiae with a single spur; 3. Clypeus with teeth; 10. Dorsum, excepting scutellar impression, without elevations or depressions; 26. Large eyes, i.e., space between eyes only five times larger than the diameter of each eye; 27. Clypeus with two teeth; 30. Pygidium with an angulate margin; 31. Protibiae with truncate apex, prosternum [sic] with a complete transverse carina and with external margin with a nodule or a very weak tooth. In the following couplet, he grouped C. obscurus and C. aequinoctialis by the presence of a humeral carina, on the one side, and C. furvus , C. securus , C. foveiventris and C. candezei by its absence, on the other; next, the presence of a protibial internal expansion joins C. furvus with C. securus , while its absence groups C. candezei and C. foveiventris . The steps leading to the group of five species that was to form Sylvicanthon , as well as those differentiating them, remained almost unchanged in the updated version of Schmidt’s key by Balthasar (1939).

Already during that ‘Latin-American phase’ of taxonomic activity on ‘ Canthon sensu lato’ exposed above, Glaphyrocanthon , a taxon described as an independent genus, but lowered to subgenus of Canthon by Halffter & Martínez (1977), shares much of its taxonomic history with Sylvicanthon . Martínez (1948a) described the genus to accommodate only two Venezuelan species, but the following year ( Martínez 1949a), he added two new species from Bolivia: G. bridarollii Martínez, 1949 , from the province of Chapare, department of Cochabamba, and G. proseni Martínez, 1949 , from the province of Nor Yungas, department of La Paz. Regarding the first species, Martínez (1949a) stated it appeared to be closely related to Canthon foveiventris , a species that he believed should be transferred to Glaphyrocanthon (“ creo que deberá ser incluída en Glaphyrocanthon ”), while he asserted that G. proseni was close to Canthon aequinoctialis , another species that he supposed to be assignable to Glaphyrocanthon . Those transfers were formally proposed by Pereira & Martínez (1956), who, among several other transfers, positioned in Glaphyrocanthon the four species that Schmidt (1920, 1922) had grouped with C. foveiventris and C. aequinoctialis (i.e., C. candezei , C. obscurus , C. furvus , and C. securus ) and presented an identification key to them. Soon after, Pereira & Martínez (1960) described Glaphyrocanthon cobosi , from the Cordillera province, Santa Cruz department, Bolivia, a species that was thought to be related to G. obscurus .

Martínez et al. (1964), in a review of Glaphyrocanthon , recognized four species assemblages in the genus: the variabilis , juvencus and aequinoctialis species groups, of the nominotypical subgenus, and the subgenus Coprocanthon Martínez, 1950 . The aequinoctialis group corresponded to the assemblage composed of C. candezei and allied species proposed by Schmidt (1920, 1922), and differed from other Glaphyrocanthon in having the unique combination of large eyes and ventral surface of metafemora with a fine anterior margin. According to Martínez et al. (1964), that group was composed of G. aequinoctialis , G. proseni , G. candezei , G. obscurus , G. foveiventris , G. securus , G. furvus , G. bridarollii , G. cobosi and their new species G. enkerlini , from São Luis, Maranhão, Brazil. Three years later, Martínez & Pereira (1967) described a last species for the group, G. machadoi , an inhabitant of the ‘brejos de altitude’ (Caatinga moist-forest enclaves) of Serra Negra, Pernambuco, Brazil, and which was said to be closely related to G. foveiventris and G. bridarollii .

Halffter & Martínez (1977) then made major changes in the classification of ‘ Canthon sensu lato’. In their work, Glaphyrocanthon , Coprocanthon , and Geocanthon Pereira & Martínez, 1956 were synonymized (the first name remaining valid) and the taxon was lowered to subgenus status under Canthon . The former aequinoctialis group, in particular, was totally dismembered and none of its species remained in Glaphyrocanthon : G. enkerlini was transferred to the Canthon subgenus Francmonrosia mainly because of its denticulate anterior edge of profemora ( Fig. 9B ) and protibiae with strong internal expansion ( Fig. 11A ), while G. aequinoctialis and G. proseni were synonymized and the species transferred to Canthon s. str. (both acts without any formal justification), where it remained isolated as the single member of the aequinoctialis lineage (“ línea aequinoctialis ”).

Five species of the former aequinoctialis group, however, G. candezei , G. furvus , G. bridarollii , G. machadoi and G. cobosi (the latter considered by them a junior synonym of Canthon xanthopus Blanchard, 1846 , name of unknown application by entomologists for more than a hundred years; see discussion below), remained grouped and were transferred to the new genus Sylvicanthon . This latter genus was said by Halffter & Martínez (1977) to be related to Pseudocanthon and Vulcanocanthon Pereira & Martínez, 1960 , sharing with them the presumed absence of a fine margin on the posterior edge of their head. Those authors also presented a discussion on the differences between Sylvicanthon and C. aequinoctialis and the reason why, according to them, this species should not be part of that genus. Lastly, Halffter & Martínez (1977) also judged that the remaining three species of the former aequinoctialis group ( C. obscurus , C. foveiventris and C. securus ), judging from Schmidt’s descriptions, would “very likely” be close to the species included by them in Sylvicanthon , but no formal transfer was proposed.

As the main goal of Halffter & Martínez (1997) was to reevaluate the supraspecific classification of ‘ Canthon sensu lato’, and not to make a detailed study of its alpha taxonomy, questions such as species redescriptions and distributions, as well as a new identification key, were not addressed by them. Therefore, as no other work has so far reevaluated the taxonomy of Sylvicanthon , our knowledge about its species was extremely fragmentary and sometimes simply incorrect.

As addressed several times throughout this work, the geographical distribution attributed to some species of Sylvicanthon was largely wrong ( S. candezei is the best example), while the majority of the species were known only from their type locality (e.g., S. securus , S. obscurus , S. furvus and S. enkerlini ). Besides, the species delimitation and identification were problematic, with as many as six different species identified in publications and collections under the same name (e.g., S. bridarollii , S. edmondsi sp. nov., S. attenboroughi sp. nov., S. genieri sp. nov., S. seag sp. nov. and S. candezei , all of them mingled under the name S. bridarollii or S. candezei ) and with wrong synonymies (between the species now known as Canthon xanthopus and C. cobosi , and S. aequinoctialis and S. proseni ). Finally, even the definition of Sylvicanthon had serious flaws as a consequence of character states believed to be universal in the genus not being found in some of its species (the most obvious being the absence of a fine margin on the posterior edge of the head), the inclusion of some species that clearly do not belong to the genus ( C. xanthopus , C. machadoi and C. cobosi ) and, at the same time, the exclusion of some species that are close to those belonging to Sylvicanthon ( S. aequinoctialis comb. nov., S. proseni and S. enkerlini ). The only modifications proposed to the taxonomy of Sylvicanthon since Halffter & Martínez (1977) were the “informal transfers” (or, as put by Sikes & Barclay 2017, “accidental” transfers; i.e., species cited in Sylvicanthon for the first time without the status of new combinations explicitly recognized) of Canthon foveiventris (by Vaz-de-Mello & Louzada 1997) and C. obscurus (by Vaz-de-Mello 2000), following the conjectures of Halffter & Martínez (1977) that those species belonged to Sylvicanthon .

Phylogenetic relationships of Sylvicanthon

Halffter & Martínez (1977) included Sylvicanthon among the Canthonina, considered by them as one of the six subtribes of Scarabaeini (see Halffter & Martínez (1966) for a deeper discussion about the positioning of Canthonina). In more modern classifications (e.g., Smith 2006; Scholtz et al. 2009; Bouchard et al. 2011; Tarasov & Génier 2015; Tarasov & Dimitrov 2016), this taxon is raised to the tribe category and renamed to its senior synonym Deltochilini (see Bouchard et al. 2011). Scarabaeini , in turn, is now composed of only a handful of African genera (see Forgie et al. 2005, 2006; Scholtz et al. 2009). Until very recently, Deltochilini comprised over 120 genera and 800 species that despite being distributed all over the world had their global distribution largely centred in the Gondwanian continents (Scholtz 2009; cited as Canthonini ). Nevertheless, despite being widely considered as valid, successive phylogenetic analyses –being based either on morphological (e.g., Philips et al. 2004; Vaz-de-Mello 2007b; Tarasov & Génier 2015) or on molecular evidence (e.g., Ocampo & Hawks 2006; Monaghan et al. 2007; Sole & Scholtz 2010; Mlambo et al. 2013; Tarasov & Dimitrov 2016) – have shown that the tribe Deltochilini is not monophyletic, being instead composed of a large number of non-closely related lineages ( Tarasov & Génier 2015; Tarasov & Dimitrov 2016; see also the extensive discussion about the phylogeny of Canthonini presented by Scholtz 2009). However, several of the New World taxa of Deltochilini , including the type genus Deltochilum , form together a recurring monophyletic group in most of those analyses.

In the studies of both Ocampo & Hawks (2006) and Monaghan et al. (2007), all the New World Deltochlini included in their analyses formed a single clade (species of the genera Deltochilum , Scybalophagus Martínez, 1953 , Malagoniella Martínez, 1961 , Megathopa Eschscholtz, 1822 , Eudinopus Burmeister, 1840 and Canthon , in Ocampo & Hawks (2006), and of Canthon , Scybalocanthon , Scybalophagus , Hansreia , Deltochilum , Eudinopus and Megathoposoma Balthasar, 1939 , in Monaghan et al. (2007)). Tarasov & Génier (2015), who included a much broader sample of genera, also recovered an exclusive New World Deltochilini clade, which had as members the genera Anomiopus Westwood, 1842 , Scatonomus , Scybalocanthon , Tetraechma , Hansreia , Canthon , Scybalophagus , Deltochilum , Malagoniella , Megathoposoma and Sylvicanthon , and was referred to as Deltochilini sensu stricto by the authors. Other American genera of Deltochilini , however, appeared very distant from this clade, such as Paracanthon , Zonocopris , Canthochilum and Cryptocanthon . Previously, with a smaller number of genera, Vaz-de-Mello (2007b) had already obtained a very similar result to that of Tarasov & Génier (2015) concerning the American Deltochilini , with a clade composed of Canthon , Sylvicanthon , Scatonomus and Anomiopus , while genera like Zonocopris , Canthonella , Paracryptocanthon and Cryptocanthon appeared scattered among different branches of the tree. Finally, the molecular phylogeny of Tarasov & Dimitrov (2016) also recovered a monophyletic group composed exclusively of New Wold Deltochilini genera. Based both on the topology of the molecular ( Tarasov & Dimitrov 2016) and morphological ( Tarasov & Génier 2015) trees and on some diagnostic apormorphies found by Tarasov & Génier (2015), Tarasov & Dimitrov (2016) formally redefined Deltochilini to include only 22 American dung beetle genera, Sylvicanthon included among them; the other groups previously included in Deltochilini were regarded as incertae sedis in Scarabaeinae .

Despite the previously described similarities, the four analyses that included species of Sylvicanthon have found very distinct relationships for the genus among the Deltochilini . In the tree conceived by Medina et al. (2003), which had the goal of studying the phylogenetic relationships of what we are denominating as the ‘ Canthon sensu lato’, ‘ Sylvicanthon bridarollii ’ 13 appeared forming a clade with Canthon (Glaphyrocanthon) politus Harold, 1868 , species widely distributed in Central America and northern South America ( Vulcano & Pereira 1964) . That clade, in turn, appeared as a sister to the topology ( Anisocanthon villosus ( C. (G.) luteicollis ( C. (G.) femoralis ( C. (G.) rubrescens + C. (G.) angustatus )))), of which all the analysed Glaphyrocanthon and Anisocanthon were part. That group as a whole, called “node I” by Medina et al. (2003), was supported by three synapomorphies: anterior part of hypomeron (erroneously cited as “proepimeron”) deeply excavated and glabrous, and posterior part of hypomeron (erroneously cited as “prosternon”) glabrous in the area close to the border with pronotum. Interestingly enough, two of those three character states are variable among species of Sylvicanthon : the pilosity of the anterior part of hypomeron varies extensively in density from almost totally absent (e.g., S. candezei and S. seag sp. nov., Fig. 35A ) to present with long and dense setae (e.g., S. bridarollii , Fig. 35 C–D), while the posterior part of hypomeron is indeed glabrous in the candezei group, but has an ill-delimited row of setae parallel to the margin with the dorsal portion of pronotum in S. enkerlini , the single species of its group ( Fig. 9D ).

The other two phylogenetic analyses that included Sylvicanthon species – Vaz-de-Mello (2007b) and Tarasov & Génier (2015) – were not aimed at studying the particular relationships between the genera of ‘ Canthon sensu lato’ nor even that of the American Deltochilini , but, rather, they studied the phylogeny of the Ateuchini ( Vaz-de-Mello 2007b) and the evolutionary history of the major lineages in Scarabaeinae ( Tarasov & Génier 2015) . As a natural consequence, their results are less informative as to the Sylvicanthon relationships than those of Medina et al. (2003).

As we have seen, much remains to be studied about the relationships between Sylvicanthon and the other groups of ‘ Canthon sensu lato’. To reach a sound understanding of this issue, a greater number of species of Sylvicanthon , of different groups and subgroups, should be included in future phylogenetic analyses, as well as different representatives of Deltochilini , especially those taxa more similar in morphology to Sylvicanthon . Based on the comparative studies performed for this work (including several species still undescribed), the tentative hypothesis that we propose for testing by formal phylogenetic analyses is the following: among the Deltochilini sensu Tarasov & Dimitrov (2016) , Glaphyrocanthon as considered today (i.e., sensu Halffter & Martínez 1977) should represent a large paraphylum composed of non-closely related lineages 14. Within that paraphylum, a number of smaller genera with a uniform morphology – e.g., Sylvicanthon and Pseudocanthon – should be nested. These genera could form either a single clade within Glaphyrocanthon , or (which we believe is most likely) represent distinct lineages with independent origins within ‘ Glaphyrocanthon ’. If this latter hypothesis is retrieved in phylogenetic analyses, Glaphyrocanthon will need to be divided once again into different genera as it was in the past ( Coprocanthon and Geocanthon , former genus/subgenus, were synonymized with Glaphyrocanthon by Halffter & Martínez 1977). Then, its limits will possibly be restricted to those of what Martínez et al. (1964) considered to be the variabilis species group of Glaphyrocanthon . In fact, this hypothesis has several similarities to the results of Medina et al. (2003), who obtained Glaphyrocanthon paraphyletic with one species of both Sylvicanthon and Anisocanthon nested within it; the major difference is that we do not include Anisocanthon among the possible members of this great ‘clade Glaphyrocanthon ’. It is also interesting to point out that in the phenogram presented by Kohlmann & Halffter (1990) for the American ‘Canthonina’, Sylvicanthon and Glaphyrocanthon appeared grouped by about 50% of similitude and isolated from the remainder ‘ Canthon sensu lato’, so revealing a close morphological connection between these two taxa.

A second hypothesis that has already been raised (FZVM, personal observations; Olivier Montreuil, personal communication to MC in June 2014) says that Sylvicanthon and some other ‘ Canthon sensu lato’ genera (especially Pseudocanthon and Canthonella ) would be close to a group of genera of Gondwanian distribution formally placed in Deltochilini that are sometimes separated in a tribe by their own, Epilissini (e.g., d’Olsoufieff 1947; Lebis 1953; Montreuil 2008, 2010, 2011; Montreuil & Thery 2011, 2016; Montreuil & Viljanen 2011; Montreuil et al. 2014; Vaz-de-Mello et al. 2011 b). Montreuil (2010) revalidated Epilissini from the synonymy with Canthonini (which, in turn, is today a junior synonym of Deltochilini ) based on two main morphological features presented by those genera: protibiae truncate at apex and hypomeron strongly excavated anteriorly. These two characteristics, in fact, are seen in Sylvicanthon and in several other Neotropical deltochiline genera, including Pseudocanthon and some Glaphyrocanthon , hence indicating a possible close relationship between them and the Epilissini sensu Montreuil (2010). The overall shape of the meso- and metatarsi, which are strongly flattened (“foliaceous”) and with tarsomeri slightly rectangular, is also very similar between Sylvicanthon and the Epilissini .

Although both hypotheses are not mutually exclusive – the ‘clade Glaphyrocanthon ’ could be part of Epilissini , which would make Deltochilini sensu stricto as defined by Tarasov & Dimitrov (2016) polyphyletic – none of the phylogenetic analyses published so far supports a close relationship between either Sylvicanthon or Glaphyrocanthon with Epilissini nor that this tribe is monophyletic. In Tarasov & Génier (2015), for example, genera of Epilissini appeared scattered throughout the phylogeny. In that analysis, Canthonella , Ochicanthon Vaz-de-Mello, 1999 and Epactoides d’Olsoufieff, 1947 (all of them included in Epilissini ) were clustered with Tanzanolus Scholtz & Howden, 1987 from Africa and which was not included among the Epilissini by Montreuil (2010), forming the sister clade of almost all the remainder Scarabaeinae . Bohepilissus Paulian, 1975 and Tesserodoniella Vaz-de-Mello & Halffter, 2006, in turn, the other two genera of Epilissini sensu Montreuil (2010) included in the analysis, were separated one from the other in distinct branches of the tree. In the molecular analysis of Monaghan et al. (2007) the scenario was no different: genera included in Epilissini such as Arachnodes Westwood, 1847 , Aleiantus d’Olsoufieff, 1947 and Paronthobium Paulian, 1984 did not show any close relationship 15. This panorama is repeated in every single phylogenetic analysis so far published.

Lastly, a third phylogenetic hypothesis also involving Sylvicanthon – which is inconsistent with the ‘clade Glaphyrocanthon ’ hypothesis championed by us – is based mainly on the morphology of the tarsal claws. In the taxonomic discussion presented by Halffter & Martínez (1977) to distinguish Sylvicanthon among the Deltochilini , the authors stated that their new genus differed from Pseudocanthon by the form of their claws, which are simple in Sylvicanthon , while having a distinct basal angle in Pseudocanthon . However, we observed during our studies that this statement is not correct. All the species of Sylvicanthon have, in fact, a small angulosity at the base of their claws (more easily seen on protarsi, even though present in all legs) ( Fig. 5A ) which, despite the distinct contrast with the strong angulosity of Pseudocanthon ( Fig. 5B ), we interpret to be homologous to it and different only in size.

It is worth noting that a similar angulosity is present in an even larger size and moved towards the centre of the claw in other former deltochiline genera and, hence, in a more obvious way than in both Pseudocanthon and Sylvicanthon . Those genera are Canthonella ( Fig. 5C ) and Ipselissus d’Olsoufieff, 1935 from the New World; Temnoplectron Westwood, 1841 and Sauvagesinella Paulian, 1934 from Australasia; Lepanus Balthasar, 1966 ( Fig. 5D ) from the islands of Java, New Guinea and Australia; Oficanthon Paulian, 1985 ( Fig. 5E ) from New Guinea; and, to a lesser degree, Nesovinsonia Martínez & Pereira, 1959 from the island of Mauritius ( Martínez & Pereira 1959; Halffter & Martínez 1967; Matthews 1974; Paulian 1985). Halffter & Martínez (1967) raised the hypothesis – which would be later favoured also by Matthews (1974) – that the presence of tarsal claws with a basal angle would be homologous among at least some of these taxa. Therefore, that feature would be evidence for the existence of an ancient lineage of a wide Gondwanian distribution represented nowadays by those relict genera.

Although not included in their discussion, Sylvicanthon and Pseudocanthon share all the main characteristics listed by Halffter & Martínez (1967) to support that hypothesis. Even though one could argue that the claw angulosity of Sylvicanthon is basal (rather than medial) and much shorter, it is possible to observe a clear, gradual transition between the extreme form seen, for example, in Canthonella ( Fig. 5C ), passing through the less-developed and closer-to-base forms of Oficanthon ( Fig. 5E ) and Pseudocanthon ( Fig. 5B ), and, finally, to Sylvicanthon ( Fig. 5A ), so indicating the possible homology between these structures. Lepanus alone shows this wide variation, as seen, for instance, in two Australian species described by Matthews & Weir (2002): L. loftyensis , with a long tooth displaced from the base of the claw, and S. penelopae , with a basal angulosity similar to that of Sylvicanthon ( Fig. 5D ). Interestingly enough, all these genera with toothed claws have the hypomeron strongly excavated anteriorly, one of the characteristics highlighted by Montreuil (2010) in his definition of the tribe Epilissini . Therefore, within the context of Montreuil’s hypothesis, we could see the group proposed by Halffter & Martínez (1967) and Matthews (1974) as one of the internal branches of Epilissini and, in this way, several different lineages of this tribe would have been spread independently across Gondwana. Matthews (1974), for instance, supposed another Gondwanian lineage represented in the Americas by Cryptocanthon and Canthochilum , and by Tesserodon Hope, 1937 and Ignambia Heller, 1916 in Australasia, a hypothesis that was later expanded and discussed in further detail by Vaz-de-Mello & Halffter (2006).

Refinement of the phylogenetic placement of Sylvicanthon should be a task for future works. The three hypotheses discussed here – ‘clade Glaphyrocanthon ’, tribe Epilissini , and the ‘tarsal claw’ hypothesis – are not entirely incompatible, but the scenario endorsed by us is that the first hypothesis is most likely correct and that the ‘clade Glaphyrocanthon ’ is part of the Deltochilini sensu Tarasov & Dimitrov (2016) . As the alpha taxonomy of Glaphyrocanthon is yet largely unknown and several species still await description, it is firstly necessary that this genus undergo a broad taxonomic revision before formal phylogenetic analyses are performed.

Limits of Sylvicanthon and morphological comparison with allied genera

As herein delimited, the genus Sylvicanthon is composed of a cohesive and easily identifiable group of species. The diagnosis offered above is very similar to the definition given by Schmidt (1920) to the group formed by Canthon candezei and his four new species (see ‘Taxonomic history’ above) and to that of the aequinoctialis group of Glaphyrocanthon sensu Martínez et al. (1964) . Characteristics such as the presence of large eyes, metafemora with anterior margin, complete hypomeral carina, and excavated hypomeron had already been used by different authors for the delimitation of those groups, as well as by Halffter & Martínez (1977) in the original description of Sylvicanthon . These latter authors, however, considered the supposed absence of a fine margin on the posterior edge of the head as one of the features that would most readily diagnose Sylvicanthon . Nonetheless, we concluded instead that this character is neither suitable to mark the limits of Sylvicanthon nor to indicate its phylogenetic placement for two main reasons. Firstly, as recognized even by Halffter & Martínez (1977) themselves, head without posterior margin is a condition seen not only among the Sylvicanthon , but also in a number of other New World deltochiline genera, such as Pseudocanthon , Vulcanocanthon and some species of Cryptocanthon and Anomiopus . On the other hand, contrary to what was said by Halffter & Martínez (1977), the posterior margin is not absent in not all of the species originally included in Sylvicanthon by them. All the specimens of Canthon machadoi and C. xanthopus possess a complete posterior margin, whereas this characteristic is variable in S. furvus and S. bridarollii , from completely absent to marked only very finely and being interrupted. Sylvicanthon candezei is the only species originally included in Sylvicanthon by Halffter & Martínez (1977) of which we have not observed any specimen with at least a vague trace of a margin on the posterior edge of the head. This margin is absent or vestigial also in seven of the nine new species herein described or transferred to Sylvicanthon ( S. foveiventris , S. genieri sp. nov., S. seag sp. nov., S. edmondsi sp. nov., S. attenboroughi sp. nov., S. obscurus , and S. monnei sp. nov.), while in two other species ( S. enkerlini and S. aequinoctialis ) this margin is always present and usually complete. In S. proseni , all the variation seen in the genus can be found: in a same population, specimens range from having a complete margin to those with a totally absent margin. Although there is a clear tendency for the posterior margin to be absent or atrophied in Sylvicanthon , this is a variable characteristic with very little taxonomic value or phylogenetic signal.

Genera with which Sylvicanthon can be sometimes confused are Pseudocanthon , Glaphyrocanthon and Canthonella , all of which share the excavated hypomeron, which is transversally divided by a complete hypomeral carina (incomplete in several species of Glaphyrocanthon ). With Pseudocanthon , in particular, Sylvicanthon shares large eyes and the absence of prescutellar and scutellar impressions, but they can be easily distinguished from one another by the shape of the pronotum (with lateral edge forming a medial angle in Sylvicanthon , and lateral edges straight and subparallel in Pseudocanthon ), shape of protibiae (with three large teeth and distinctly expanded at middle in Pseudocanthon , and with internal expansion or not and with two or three small apical teeth ( candezei group) or three large teeth distributed through the apical half of protibiae ( enkerlini group) in Sylvicanthon ), and elytral pilosity (glabrous in Sylvicanthon , and covered by minute setae in Pseudocanthon ), as well as the overall body shape, which is much shorter and flatter in Sylvicanthon than in Pseudocanthon . Number and shape of clypeal teeth also vary between the two taxa: in Pseudocanthon , there are always four acute teeth, the middle two larger than the lateral ones, and with base covered by a single row of short setae extending almost to genae. In Sylvicanthon , in turn, these characters vary widely between the two species groups. In the candezei group, there are two small teeth and the row of setae (which can be single or divided into two smaller fragments at the base of each tooth) hardly extends beyond the base of the teeth ( Fig. 6 B– G). In S. enkerlini , the single species of its group, an intermediate condition is seen: there are four large teeth similar to those of Pseudocanthon , but the row of setae is limited to the base of the teeth as in the other species of Sylvicanthon ( Fig. 6A ).

Canthonella , in turn, is readily differentiated from Sylvicanthon by tarsal claws with a strong basal tooth (while this tooth is very small in Sylvicanthon ), shape of clypeal teeth and presence of a trochantofemoral pit on the front legs, while this latter feature is absent in Sylvicanthon (Vaz-de-Mello et al. 2011). Glaphyrocanthon , as already discussed, is probably paraphyletic and, hence, some of its species are more closely related to Sylvicanthon (and, consequently, share some apomorphies with this genus) than to other of their congeneric species. The set of species that today forms Glaphyrocanthon is similar to Sylvicanthon , among other characteristics cited throughout this work, by the shape of pronotum, whose lateral edges have a medial angulosity. Even so, Glaphyrocanthon can be easily differentiated from Sylvicanthon by having profemora lacking both the anterior and the posterior margins on its ventral surface (anterior margin always present in Sylvicanthon , posterior margin present or not), eyes with a very narrow dorsal portion (wide in Sylvicanthon ) and protibiae obliquely truncate at apex (rectly truncate in Sylvicanthon , Fig. 11 ).

Species composition

As defined in the present work, Sylvicanthon has 15 species divided into two groups: the enkerlini group, with a single species, S. enkerlini , and the candezei group, with five subgroups: the candezei subgroup, with S. candezei , S. foveiventris and S. genieri sp. nov.; the aequinoctialis subgroup, with S. aequinoctialis and S. proseni ; the furvus subgroup, with S. obscurus , S. furvus , S. monnei sp. nov. and S. mayri sp. nov.; the bridarollii subgroup, with S. bridarollii , S. attenboroughi sp. nov., S. edmondsi sp. nov. and S. seag sp. nov.; and the securus subgroup, with the sole S. securus . This new composition differs from that of Halffter & Martínez (1977) in three main aspects: 1) two species originally included in Sylvicanthon (the former S. machadoi and S. xanthopus , along with a former junior synonym of the latter, Glaphyrocanthon cobosi , here revalidated) are removed from Sylvicanthon and transferred to Canthon ; 2) six species previously placed in Canthon are transferred to Sylvicanthon : C. aequinoctialis , positioned in Canthon s. str. by Halffter & Martínez (1977), and its former junior synonym, Glaphyrocanthon proseni , which is here revalidated; Glaphyrocanthon enkerlini , placed in Canthon (Francmonrosia) by Halffter & Martínez (1977); and C. securus , C. foveiventris , and C. obscurus , species mentioned as possible Sylvicanthon by Halffter & Martínez (1977), but never formally transferred (but see ‘Taxonomic history’ above for more details about previous informal transfers of the latter two species); and 3) six new species are herein delimitated, described, and named for the genus. Furthermore, the recognition of two groups and five subgroups of species in Sylvicanthon is also new. In the following paragraphs, we discuss the proposed changes.

Although the results of this work undoubtedly show that Canthon xanthopus , Glaphyrocanthon cobosi and G. machadoi (hereafter cited in their original combinations) do not belong to Sylvicanthon , the genus to which they should be transferred is not as clear. The transfer of C. xanthopus is the least problematic and made with the most confidence. Among the evidence indicanting that this species does not belong to Sylvicanthon are a) shape of clypeal teeth (large, acute and very apart from each other in C. xanthopus ( Fig. 45 ), and small, usually obtuse and contiguous or very close in Sylvicanthon , Fig. 6 B– G); b) size of eyes (small eyes of about one-eighth of the width of the interocular space in C. xanthopus , and large eyes of about one-fifth of the width of the interocular space in Sylvicanthon ); c) shape of protibiae (with large teeth and apex occupied by the apical tooth in C. xanthopus , and small teeth and truncate apex in Sylvicanthon , Fig. 11 ); d) shape of hypomeron (only slightly depressed anteriorly and with hypomeral carina incomplete in C. xanthopus , and deeply excavated and with complete hypomeral carina in Sylvicanthon , Fig. 35 ); and e) external margin of mesocoxae (narrow in C. xanthopus , see Fig. 7C , but very wide in Sylvicanthon , see Fig. 7 A–B), among several other characters. At the same time, this species conforms with the definition (certainly artificial) of Canthon given by Halffter & Martínez (1977). Therefore, we propose that C. xanthopus be returned to its original combination (see below for more details on the relationship of this species with other Canthon ).

Finding a proper generic placement for Glaphyrocanthon machadoi and G. cobosi is a more complicated task given the peculiar combination of characters that these two closely related species possess. Among the main differences with Sylvicanthon are a) clypeal teeth very separated from one another (teeth contiguous or very close in Sylvicanthon ); b) dorsal portion of eyes narrow and with a fine internal margin (large unmargined eyes in Sylvicanthon ); and c) external margin of anterior part of hypomeron strongly sinuous and with a strong tooth (straight or only slightly sinuous and usually with a very tiny tubercle in place of the tooth in Sylvicanthon ). On the other hand, these two species have their general morphology very similar to that of Glaphyrocanthon , especially concerning the shape of the head (clypeus and eyes) and the excavated hypomeron. Even so, the presence of an anterior margin on the metafemora of these two species differentiates them from Glaphyrocanthon , whose species (at least the described ones) have unmargined metafemora 16. Using the keys of Halffter & Martínez (1977) and Vazde-Mello et al. (2011), these species would be identified as Canthon . Based on that, and knowing that Canthon is highly artificial and comprises a large number of incertae sedis species of ‘ Canthon sensu lato’ (i.e., it is a ‘wastebasket’ taxon), we propose the temporary transfer of G. cobosi and G. machadoi to this genus. We believe, however, that when future works will have dedicated themselves to revise both Glaphyrocanthon and Canthon and given a new definition to these genera, those two species should be repositioned in some genus other than Canthon ; possibly, a new one (see discussion on the revalidation of Canthon cobosi from the synonymy with C. xanthopus in the section dedicated to the former species).

Six other species formely positioned in Canthon by Halffter & Martínez (1977) are here incorporated into Sylvicanthon .As previously mentioned, Halffter & Martínez (1977) synonymized Canthon aequinoctialis and Glaphyrocanthon proseni , stating that the species, which was considered by previous authors to be close to those that today form Sylvicanthon (e.g., Schmidt 1922; Balthasar 1939; Martínez et al. 1964), would be isolated in its own lineage among the Canthon s. str. and, hence, had no intimate relationship with Sylvicanthon . In the discussion of the latter genus, Halffter & Martínez (1977) stated that Canthon aequinoctialis indeed had several similarities with Sylvicanthon , but they were nevertheless different in the following characters present allegedly only in the former species: a) posterior edge of head margined between eyes; b) mentum with anterior edge emarginate in ‘U’; c) suture between submentum and gula ‘Y’-shaped; d) protibial teeth of medium size; e) meso- and metatarsomeri I of length subequal to that of meso- and metatarsomeri II; and f) mesocoxae with very narrow external margin.

Save the one concerning character (c), all the other arguments are refuted in the present work and, based on the description and diagnosis given above, Canthon aequinoctialis and Glaphyrocanthon proseni are transferred to Sylvicanthon . Characteristic (a) was discussed above and we saw that, contrary to what was written by Halffter & Martínez (1977), the presence or absence of a margin on the posterior edge of head is highly variable among the Sylvicanthon (and presumably also among other genera such as Pseudocanthon ), both intra- and interspecifically, and, hence, it is not a good indicator of generic limits. Besides, several specimens of the now Sylvicanthon proseni lack this margin, although the majority of the specimens show at least some vague traces of it.

Regarding character (b), it was seen that the so-called ‘U’-emarginate mentum in S. aequinoctialis and S. proseni is at most only slightly different from the shape seen in the other Sylvicanthon ( Fig. 8 ), the same occurring to character (f) ( Fig. 7 B–C). As for characteristic (d), the protibial teeth in those two species are indeed different from the remaining Sylvicanthon of the candezei group ( Fig. 11B ), but it seems to be an intermediate condition between that seen in S. enkerlini ( Fig. 11A ) and other related groups such as Glaphyrocanthon and the other species of the candezei group. Therefore, we can see these differences as a transformation series from the more general condition observed in Glaphyrocanthon and S. enkerlini , where the protibial teeth are large and well separated from one another, passing through the state present in S. aequinoctialis and S. proseni , whose teeth are of medium size, to the pattern found among the remaining Sylvicanthon , which have small teeth limited to the apical region of the protibiae (members of the candezei subgroup took a further step in this sequence and lost the basal tooth, keeping only the two apical ones).

Finally, characteristic (c) is indeed unique to the aequinoctialis subgroup: while the suture between the submentum and gula has a ‘Y’ shape (shape highlighted by a row of long setae covering it, Fig. 8B ) in the members of this subgroup, in all the other species of Sylvicanthon this suture is rounded and may or may not be covered by a row of setae ( Fig. 8A ). Nonetheless, as previously noted by Halffter & Martínez (1977), the shape of this suture is extremely variable in Canthon and, therefore, it should not be taken as evidence of phylogenetic affinity between its species and the aequinoctialis subgroup. In this way, the ‘Y’-shaped suture is here interpreted as a synapomorphy of the aequinoctialis subgroup, a modification of the rounded shape found among the other Sylvicanthon and supposedly present in the ancestor of this genus.

In addition, Canthon securus , C. foveiventris and C. obscurus – all species that were considered to be closely related to today’s Sylvicanthon by previous authors (e.g., Schmidt 1920, 1922; Balthasar 1939; Martínez et al. 1964; Halffter & Martínez 1977) – are here at last transferred to the latter genus. These three species show all the characteristics present in the description and diagnosis given above and are intimately related to the other members of the candezei group, with which they are commonly confused both in collections and in publications. Halffter & Martínez (1977) stated that, very likely, these three species would have to be transferred to Sylvicanthon , but they did not argue why they themselves did not make these transfers (probably because they had not seen the type specimens, though). Although those transfers had never been proposed in a formal manner until now (i.e., explicitly stating they were new combinations and giving the arguments for the new taxonomic acts), subsequent authors started citing two of them – S. foveiventris and S. obscurus – in combination with Sylvicanthon (e.g., Vaz-de- Mello & Louzada 1997; Vaz-de-Mello 2000; Hernández 2002; Durães et al. 2005; Falqueto et al. 2005; Almeida & Louzada 2009; Hernández & Vaz-de-Mello 2009; Hernández et al. 2011; Larsen 2012; Culot et al. 2013), including important online biodiversity databases such as the Encyclopedia of Life and the Catalogue of Life.

Finally, the last transfer to Sylvicanthon presented herein is that of Glaphyrocanthon enkerlini , considered part of the aequinoctialis species group of Glaphyrocanthon by Martínez et al. (1964) – and thus close to today’s Sylvicanthon – but placed in Canthon (Francmonrosia) by Halffter & Martínez (1977). These latter authors justified their transfer of G. enkerlini to Francmonrosia with the following characteristics: a) profemora with denticulation on its anterior margin; b) protibiae with internal expansion and c) with truncate apex. The latter two characteristics are, in fact, also present in Sylvicanthon and, thus, can be equally used as an argument to place G. enkerlini in this latter genus (protibiae with internal expansion is also seen in Glaphyrocanthon ). Profemora with denticulation on its anterior margin, however, are not seen in any other Sylvicanthon and are indeed a characteristic present in all Francmonrosia (as currently defined, at least), which could be seen to support Halffter & Martínez’s hypothesis. However, the denticulation in G. enkerlini is very distinct from the one shown by the great majority of the Francmonrosia : in species of the latter subgenus, teeth tend to be large, few in number, and present only in males, while they are small and, together, give a serrate appearance to the profemora of both sexes in G. enkerlini ( Fig. 9B ). Canthon (F.) carbonarius Harold, 1868 , however, is unique in its subgenus in being similar to G. enkerlini concerning this characteristic and, indeed, it was cited by Halffter & Martínez (1977) as the closest relative of G. enkerlini .

Despite the similarities between G. enkerlini and C. carbonarius , in particular, and the putative affinity of G. enkerlini with Francmonrosia , in general, several observations were made in the present work that, in our opinion, refute Halffter and Martínez’s classification and simultaneously give support to the proposed placement of G. enkerlini in Sylvicanthon : 1) shape of metaventrite, which is swollen on its anteromedial region and has a strong transverse depression on its anterior region in Francmonrosia , while it is entirely flat and has an area of rivose microsculpture typical of Sylvicanthon on its anterior part in G. enkerlini ; 2) size of dorsal portion of eyes, which is wide in G. enkerlini and Sylvicanthon ( Fig. 6 ), while it is narrow in Francmonrosia ; 3) external margin of mesocoxae, which is broad in G. enkerlini and Sylvicanthon ( Fig. 7 A–B), but narrow in Francmonrosia ( Fig. 7C ); 4) presence of scutellar depression in Francmonrosia , while it is absent both in G. enkerlini and in Sylvicanthon ; 5) the relative width between pronotum and elytra, the former being wider than, or subequal to, the latter in Francmonrosia , while elytra are distinctly wider than pronotum in Sylvicanthon and G. enkerlini ; and 6) shape of pronotum, which is swollen, strongly convex and, in lateral view, much higher than the elytra in Francmonrosia ( Fig. 10A ), while it is flat and at the same level as the elytra both in Sylvicanthon and in G. enkerlini ( Fig. 10 B–C). Other characteristics differing G. enkerlini from Francmonrosia , on the one hand, and approximating it to Sylvicanthon , on the other, are size and overall body shape ( Francmonrosia includes species much larger and more robust than Sylvicanthon ). The margin of the posterior edge of head, notwithstanding the problems discussed above, also gives some indications of a closer relationship between G. enkerlini and Sylvicanthon than of that species with Francmonrosia . In this latter subgenus, this margin is always complete and well marked, while in G. enkerlini it is fine and, sometimes, almost imperceptible, in the same way as seen in the majority of the Sylvicanthon . Due to all these arguments, G. enkerlini is herein transferred from Canthon (Francmonrosia) to Sylvicanthon .

However, there are clear differences between S. enkerlini and the other Sylvicanthon . The most obvious of them is the shape of clypeus, structure that in S. enkerlini possesses four large, acute teeth ( Fig. 6A ), while all the remaining species have only one pair of small, obtuse or slightly acute teeth ( Fig. 6 B–G). Furthermore, the shape of protibiae, as already discussed, is very different between S. enkerlini and the other Sylvicanthon ; the condition seen in that species is very similar, however, to that found in many Glaphyrocanthon (see drawings in Rivera-Cervantes & Halffter (1999), for example). Presence of denticulation on the anterior margin of profemora ( Fig. 9B ) and pilosity on the posterior part of hypomeron ( Fig. 9D ) are also conditions seen in Sylvicanthon exclusively in S. enkerlini . Therefore, together with characteristics in habitat preferences (see details further in the text in the sections ‘Biogeography’ and ‘Natural history’), the differences between S. enkerlini and the remaining Sylvicanthon are used here to divide the genus into two species groups: the enkerlini group, with the single S. enkerlini , and the candezei group, with all the other 14 species of the genus.

It is interesting to note that S. enkerlini has an ‘intermediate’ overall morphology between the Sylvicanthon of the candezei group and Glaphyrocanthon , presenting characteristics interpreted here as either apomorphies of Sylvicantahon (such as metafemora with anterior margin ( Fig. 21B ), eyes with wide dorsal portion ( Fig. 6A ), mesocoxae with wide external margin ( Fig. 7 A–B), hypomeron strongly excavated anteriorly strongly excavated ( Fig. 9D ), protibiae truncate at apex ( Fig. 11A ) and absence of scutellar and prescutellar depressions) or plesiomorphies shared with Glaphyrocanthon that are not present in other Sylvicanthon (e.g., the overall shape of protibiae ( Fig. 11A ), including large teeth and the expansion on their internal margin, and pilosity on the posterior part of hypomeron ( Fig. 9D )). Besides, S. enkerlini shows some unique apomorphies within the genus, such as denticulate profemora ( Fig. 9B ). Therefore, this new vision of Sylvicanthon presented herein brings with it intrinsic hypotheses about the morphological evolution and evolutionary relationships within the genus that should be tested in future phylogenetic analyses.

The candezei group, in turn, has as unique, easily observable features protibial teeth of smaller size ( Fig. 11 B–J) and, especially, a clypeus with only two small apical teeth ( Fig. 6 B–G). The 14 species are divided into five cohesive subgroups of species. The candezei subgroup, with three species, is immediately separated from the remaining Sylvicanthon by the presence of only two protibial teeth ( Fig. 11J ), while all the other members of the genus have three ( Fig. 11 A–I). The aequinoctialis subgroup, with only two species, is distinct from the remaining species in the shape of its protibiae (wider and with more robust teeth than the other Sylvicanthon of the candezei group, Fig. 11B ), presence of metafemora with posterior margin ( Fig. 31 ), suture between submentum and gula ‘Y’-shaped ( Fig. 8B ), and constant presence of humeral carina ( Fig. 12C ; in S. obscurus , this carina is only present in some specimens). The four members of the furvus subgroup are differentiated from the other Sylvicanthon in having, in combination, protibiae with a clear internal expansion ( Fig. 11 D–E), abdomen with three pairs of lateral foveae in females ( Fig. 16 A–C; see discussion in the next section) and parameres bifurcate at the apex ( Fig. 19 B–E), while the four members of the bridarollii subgroup possess protibiae with straight internal margin ( Fig. 11 H–I) (except S. bridarollii , Fig. 11 F–G) and abdomen lacking lateral foveae in both sexes. The remaining subgroup only includes S. securus , a species that has ambiguous relationships with two of the other four subgroups. On the one hand, S. securus is close to the furvus subgroup in view of the strong internal expansion on its protibiae (the strongest protibial expansion in any species of the candezei group, Fig. 11C ), but it is distinct from its species in having simple parameres ( Fig. 19A ) and because its females do not bear abdominal foveae, characters that, in turn, are typical of the species of the bridarollii subgroup.

Although this classification in subgroups has the aim of representing monophyletic groups, future formal phylogenetic analyses should test this scheme, present a clearer positioning for S. securus , and evaluate with more robust data the complex evolution of the characters used to delimit the subgroups (especially the evolution of the abdominal foveae, which, apparently, had multiple origins in different subgroups; see discussion below). Table 2 summarizes the characteristics used in this classification of the species of Sylvicanthon .

Comparative morphology of species of Sylvicanthon

The differences between the species of Sylvicanthon , especially among members of a same subgroup, are usually very subtle and lie in a combination of characters related to colour, microsculpture of the tegument and shape of parameres. In this section, we present a detailed comparison between the species of Sylvicanthon in relation to key structures both to the delimitation and to the identification of its species.

DORSAL COLOURATION. Excluding S. genieri sp. nov., which has a coppery colouration throughout the body, the other species of Sylvicanthon possess darker shades such as green ( S. securus , S. candezei , S. mayri sp. nov., S. monnei sp. nov. and some specimens of S. seag sp. nov.), dark blue (some specimens of S. seag sp. nov. and in S. edmondsi sp. nov.) or black with bluish or violet reflexes ( S. aequinoctialis , S. proseni and S. enkerlini ). The more common pattern, however, is the bicolour one, in which pronotum and head are purple and elytra are black, dark blue or dark green, as seen in S. foveiventris , S. bridarollii , S. attenborough sp. nov., S. obscurus and some specimens of S. furvus and S. seag sp. nov. In S. obscurus and S. seag sp. nov., the colouration pattern varies geographically: in S. obscurus , populations in Espírito Santo ( Brazil) show a gradient where individuals may have head and pronotum green or yellow, at the one extreme, or purple, at the other, while elytra are always dark green or dark blue. In the Brazilian northeast, however, only specimens close to the ‘purple-head-and-pronotum extreme’ were observed (see more details in the discussion of this species). In S. seag sp. nov., the colouration also varies in a north-south axis: populations close to the banks of the Amazon river (e.g., Manaus and Belém) and those in the state of Maranhão show the bicolour pattern described above, while individuals from the Guianas, Venezuela and Trinidad possess a purple head, and green or blue pronotum and elytra (see detailed description in the discussion of this species).

Teneral specimens of all the species possess colouration distinct from the one described above. They are clearer and, most of the time, brownish with weak bluish and greenish reflexions. As this pattern is common to all the species, only the mature colouration is described and discussed throughout this work. Therefore, it is important to keep always in mind that occasional teneral specimens will not fit what is presented in each species’ description.

SCULPTURE OF THE TEGUMENT. The pattern of the tegument sculpturing on head, pronotum, elytra, and metaventrite was one of the main sources of evidence for the species delimitation in Sylvicanthon . The terminology for the description of the microsculpture has been employed in a very inconsistent and imprecise way in the literature concerning ‘ Canthon sensu lato’. Therefore, it is necessary to clarify the meaning of the terms used throughout this work.

When the surface of the tegument does not show any visible microsculpture, it is denominated smooth, even though micropunctation (i.e., small, individual non-sericeous perforations of the tegument) may be present. Microsculpture (i.e., the pattern of impressions or elevations on the surface of the body seen under microscopic lens; modified from the definition of ‘sculpture’ by Torre-Bueno 1989), in turn, can be present among Sylvicanthon in two general ways: the first one, when the tegument surface is covered by a net of tiny, well-delimited polygons 17 contiguous to one another, is denominated alveolar microsculpture (following Krell 1994 and Harris 1979) ( Fig. 12A ). The second one, where microsculpture does not show a well-defined regular pattern and is, instead, formed of a conjunct of tiny lines running in a nonparallel way, is denominated rivose microsculpture (following Harris 1979) ( Fig. 12B ). The alveolar pattern may have a three-dimensional appearance, being clearly present in the tegument’s most external layer, or may have a flatter aspect, when it seems to be internal to the tegument, and more externally covered by a translucid, smooth layer (i.e., an external layer without microsculpture). In the second case, the alveolar microsculpture is always more difficult to observe and, in some occasions, only some vague traces can be seen, when they are called ‘diffuse microsculpture’. Krell (1994) discussed in detail several aspects of the evolution, physiology and structure of the insect alveolar microsculpture and how this feature can serve to delimitate species.

All the species of Sylvicanthon have a very complex microsculpture pattern, which can show either only a little variation among conspecific individuals (e.g., in the four species of the furvus subgroup) or a remarkable intrapopulational (e.g., S. proseni ) or geographical (e.g., S. seag sp. nov. and S. candezei ) variation. The meaning of those different levels of intraspecific variation is, for now, unknown to us. Some areas of the tegument, however, have a similar sculpturing in all the individuals of Sylvicanthon : hypomeron (both the anterior and the posterior parts), mesepimera, metepisterna, prosternum, epipleura, and the external sides and antero-medial region of metaventrite are always covered by strong rivose microsculpture ( Fig. 12B ). It is actually on the tegument of the centre of the pronotum and elytra where lies the greater part of the variation.

The pronotum of S. enkerlini , S. foveiventris , S. edmondsi sp. nov., S. securus and the species of the furvus subgroup, as well as in the northern populations of S. candezei , is smooth and has evident micropunctation; in S. aequinoctialis , S. bridarollii , S. seag sp. nov., S. attenboroughi sp. nov. and in the southern populations of S. candezei , the alveolar microsculpture on the pronotum may be present, but it is always flat and micropunctation is evident; S. genieri sp. nov. is the only species where the pronotum always presents a strong alveolar microsculpture at the centre with the micropunctation obliterated by it, while S. proseni shows all the variation discussed above. In turn, elytra of S. foveiventris , S. securus , S. monnei sp. nov. and individuals from northern populations of S. candezei and S. seag sp. nov. are smooth and have dense micropunctation; those of S. aequinoctialis have a very subtle alveolar microsculpture, while those of S. mayri sp. nov. have a diffuse microsculpture, being the micropunctation evident in both species; and the elytra of the southern populations of S. candezei and S. seag sp. nov. and those of S. genieri sp. nov., S. proseni , S. bridarollii , S. edmondsi sp. nov., S. attenboroughi sp. nov. (majority of the specimens) and S. furvus show an evident alveolar microsculpture obliterating the micropunctation. Finally, a unique pattern is seen on the elytra of S. obscurus and S. enkerlini : the tegument of the centre of the elytra is smooth and has a dense micropunctation, while that of the sides and apex has a strong alveolar microsculpture obliterating micropunctation.

TRANSVERSE LINE ON THE POSTERIOR EDGE OF PRONOTUM. Schmidt (1920, 1922) assigned the presence of a fine transverse line at the centre of the posterior edge of the pronotum only to S. furvus , which is one of the features used by him to differentiate this species from S. obscurus in his key. In fact, this fine, sulcate line is absent in S. obscurus , as well as in S. foveiventris , but it is present in S. candezei , S. securus and S. aequinoctialis , species studied by Schmidt (1920, 1922), as well as in S. enkerlini , S. genieri sp. nov., S. proseni , S. seag sp. nov., S. edmondsi sp. nov., S. attenboroughi sp. nov., S. monnei sp. nov.,

S. mayri sp. nov. and the majority of the specimens of S. bridarollii (see discussion on the geographical variation on this character in S. bridarollii in this species’ section). Therefore, in the same way as for other characteristics thought earlier to be exclusive to only one species in the genus (see below the discussion about the coarse punctation on metafemora and the three pairs of female abdominal foveae), the transverse line of pronotum is actually a feature widely present among the Sylvicanthon .

COARSE PUNCTATION AT THE BASE OF METAFEMORA. In four species of Sylvicanthon , S. foveiventris ( Fig. 13A ), S. furvus ( Fig. 13E ), S. monnei sp. nov. ( Fig. 13C ) and individuals from southern populations of S. bridarollii ( Fig. 13G ), the ventral surface at the base of metafemora has a group of elongate and narrow impressions that is here denominated coarse punctation of the base of metafemora. Schmidt (1920, 1922) was the first author to observe this characteristic, saying it was present in S. foveiventris . Martínez (1949 a), in the original description of S. bridarollii , observed the same coarse marks in this species and, indeed, he said his new taxon was close to S. foveiventris . In the present work, besides having observed for the first time this feature in S. furvus and S. obscurus , and describing it for the new species S. monnei sp. nov., we also noticed a curious geographical variation in S. bridarollii ( Fig. 36 ): individuals from populations from central Bolivia, in places such as Cochabamba and Santa Cruz, have a very evident coarse punctation, whereas farther north in Bolivia and in southern Peru, this punctation is sparser and ill marked. Northwards in Peru and in Ecuador and Colombia, this punctation becomes increasingly thinner (almost undistinguishable from the micropunctation of the rest of the tegument) and, in the northernmost populations, it is completely absent. This clinal variation is very interesting because it shows that the coarse punctation is likely to be homologous to the micropunctation of the remainder of the tegument (which is present in all the other Sylvicanthon ). Since species in three different subgroups show this kind of punctation, it is possible that it has evolved at least three times in Sylvicanthon . A final species, S. obscurus , is noteworthy in presenting, apparently, an intermediate condition: in the entire surface of its metafemora, the punctation is thicker and more evident than the micropunctation seen in other species (e.g., S. securus , S. edmondsi sp. nov., etc; or in the apical region of S. foveiventris , S. furvus , S. monnei sp. nov. and S. bridarollii ), but they are not as elongated and deep as in the species with coarse punctation ( Fig. 13B ).

In the same way as for the abdominal foveae (see below), it is reasonable to suppose that the coarse punctation of the mefemora has a role in the chemical communication (through pheromones) between individuals of a same species. In the four species that possess it, the density of this punctation can vary intrapopulationally, but, with the sole exception of a specimen of S. foveiventris collected in Itamonte (Minas Gerais, Brazil), which has only vague vestiges of it, all the remaining specimens studied clearly show the coarse marks at the base of their metafemora. Other members of ‘ Canthon sensu lato’ that have similar coarse punctuation are two of the species of Canthon (Peltecanthon) Pereira, 1953 , namely C. staigi ( Pereira, 1953) and C. sulcatus Castelnau, 1840 (see Halffter & Martínez 1967: figs 9–10).

ELYTRAL STRIAE. Halffter & Martínez (1977) and Tarasov & Génier (2015) stated that Sylvicanthon have nine elytral striae and, indeed, this is probably the ancestral condition in the genus. In all the species, counting the number of striae is usually a difficult task, since, save the three or four more internal ones, they are very effaced, fine and discontinuous. As an effect, the maximum number of observable striae vary among the species: in S. proseni , S. obscurus and the members of the bridarollii subgroup the maximum number is nine; in S. candezei , S. genieri sp. nov. and S. aequinoctialis it is eight; and in S. foveiventris , S. securus , S. furvus , S. monnei sp. nov. and S. mayri sp. nov., seven. Another important feature to stress related to the elytral striae is the presence of a fine carina on the humeral region of the eighth stria; because of its position, this carina is hereafter called the humeral carina ( Fig. 12C ). Three species of Sylvicanthon possess it: S. aequinoctialis , S. proseni and S. obscurus . In the latter species, only two-fifths of the individuals studied show the humeral carina; in the other specimens, the eighth stria is flat and very fine in the same way as the other internal striae (see more details in the discussion of this species). On the other hand, all the specimens of S. aequinoctialis and S. proseni evidently show this carina, although in the first species it tends to be longer than in the second. Other groups of Neotropical Deltochilini also show a humeral carina, including species of Canthon and Scybalocanthon (e.g., see the discussion on Canthon xanthopus below).

SEXUAL DIMORPHISM. As commonly observed in Deltochilini , few secondary sexual characteristics are evident among the species of Sylvicanthon , a fact that turns the sexing of specimens into a difficult task. Being telecoprids, they have not developed the horns and other armaments that evolved among paracoprid dung beetles and which are used in combats between males inside tunnels and other closed spaces (see Emlen & Philips (2006) for a detailed discussion on the relation between horns and paracoprid behaviour). In this way, sexual dimorphism is limited in Sylvicanthon to more discreet features, such as the shape of the sixth ventrite ( Fig. 14 B–E) and that of the protibial spur ( Fig. 15 ).

In all Sylvicanthon , the fit between the apex of the pygidium and the sixth ventrite is distinct between males and females: in males, ventrite six has an emargination on its posterior edge that allows the pygidium to extend itself further into the ventral side of the abdomen ( Fig. 14D ); in females, there is no evident emargination and the pygidium does not extend towards the ventral side of the abdomen ( Fig. 14 B–C, E). Thus, it is possible to differentiate both sexes observing that the sixth ventrite of males is narrowed at the middle, whereas it is wide in females. Some species, however, exhibit modifications on the posterior edge of ventrite five that makes the observation of this difference difficult. Mainly in females of S. candezei , S. seag sp. nov., S. edmondsi sp. nov., S. attenboroughi sp. nov. and both sexes of S. securus , ventrite five has a weak medial expansion on its posterior edge covering the anterior region of ventrite six and thus giving a narrowed appearance to the latter ( Fig. 14E ). In males of those species, ventrite five also presents this medial flange and, in general (except in S. securus ), it can be as or even more developed than in females ( Fig. 14D ).

Another sexual dimorphism found among the species of Sylvicanthon lies in the shape of the protibial spur: in females, this spur is spiniform and simple, without any bifurcation or apical expansion ( Fig. 15F ); the only exceptions are the females of S. proseni , whose spur is apically bifurcate ( Fig. 15H ). In sharp contrast to this uniform female shape, males of Sylvicanthon show wide interspecific variation in the morphology of their protibial spurs, with a unique pattern found in each species (excepting the group formed by S. furvus , S. monnei sp. nov. and S. mayri sp. nov., which shows the same shape in all the three species). In S. enkerlini , the spur is long, narrow and bifurcate at the apex, with the external branch longer than the internal ( Fig. 15A ). In the candezei subgroup, the spur is broad and bifid, with the external branch always longer than the internal one: in S. foveiventris , the internal branch is very broad and only slightly shorter than the external one, which is spiniform ( Fig. 15B ); in S. candezei , the internal branch is only slightly indicated and the separation between it and the external branch is horizontal ( Fig. 15C ); in S. genieri sp. nov., an intermediate condition is seen ( Fig. 15D ). In the aequinoctialis subgroup, the spur is also broad and bifid: in S. aequinoctialis , the internal branch is only slightly indicated ( Fig. 15E ), while it is much more developed in S. proseni ( Fig. 15G ); the female spur of S. proseni differs from males in being more narrowed and having the separation between external and internal branches in a closer angle (in ‘V’, Fig. 15H ), whereas in males it is much broader (in ‘U’). In the bridarollii subgroup we see the highest level of interspecific variation: in S. bridarollii ( Fig. 15I ), the spur is narrowed and the internal branch is only slightly developed; in S. seag sp. nov., the spur is broad and the internal branch is also short ( Fig. 15J ) (in a single specimen from French Guiana, one of the spurs has a small tubercle in the space between the two branches); in S. edmondsi sp. nov., the two branches are very developed (the external one slightly longer) ( Fig. 15K ); and, lastly, in S. attenboroughi sp. nov., the spur has a narrow base, the apical region is broad and the internal branch is only slightly developed, clearly shorter than the external one ( Fig. 15L ). In the furvus subgroup, a shape is observed in S. obscurus , whose internal branch is broad and much shorter than the external one, from which it is separated by an open angle ( Fig. 15M ), and the one observed in the other three species, where the external and the internal branches are subequal in length and separated by a very close angle ( Fig. 15N ). Finally, the spur of S. securus differs from all the other species in having the internal branch, which is very broad, longer than the external one, which is spiniform ( Fig. 15O ). Such sheer diversity in the shape of the male protibial spur must have been the fruit of an intense pressure from sexual selection, although the role played by the protibial spurs in the life of the Sylvicanthon is still unknown.

A third sexually dimorphic characteristic is restricted to only five species in the genus: S. foveiventris and members of the furvus subgroup. Females of these five species possess three pairs of foveae located on the sides of the abdomen at the sutures between ventrites I–II, II–III and III–IV ( Fig. 16 ). It is interesting to note that the presence of these foveae among the species and sexes in Sylvicanthon was matter of a wide confusion in the literature.

Schmidt (1920) included this characteristic in the description of S. foveiventris (and, probably, it was thanks to this feature that the species was named as such), but he did not make any mention whether these foveae were present in both sexes or whether they were unique to just one of them. It is curious to note, actually, that this was the only one of the four species of Sylvicanthon described by Schmidt (1920) without the sexual dimorphism discussed by him. For S. obscurus and S. furvus , although he had described sexual dimorphism related to the protibial spur, Schmidt (1920) did not mention the presence of abdominal foveae. Two years later, Schmidt (1922) erred again by assigning the presence of abdominal foveae to both sexes of S. foveiventris , an opinion that was repeated by most of the subsequent authors ( Pereira & Martínez 1956; Martínez et al. 1964). The exception was Balthasar (1939), who, in turn, was also wrong in suggesting that foveae might be a male characteristic in S. foveiventris (“nur beim ♂?”, p. 189); once more, nothing was mentioned about the presence of this characteristic in both S. furvus and S. obscurus . As demonstrated in the present work, not only are the abdominal foveae present in other species than S. foveiventris , but in all these species they are restricted to females, being the easiest and the most reliable way to separate the sexes.

Little variation exists between the abdominal foveae of S. foveiventris ( Fig. 13E , 16D ), S. furvus ( Fig. 16A ) and S. monnei sp. nov. In S. obscurus , there is a row of long setae covering the anterior margin of each fovea, which, therefore, stand out from the remaining tegument of the abdomen ( Fig. 16C ). In the two females of S. mayri sp. nov. studied, in contrast, the abdominal foveae are clearly more narrowed and superficial than those seen in the other four species ( Fig. 16B ; see more details in the discussion of S. mayri sp. nov.).

In the same way as for the coarse punctation at the base of metafemora, it is here hypothesized that the abdominal foveae should also play a role in the intraspecific chemical communication, and, in this particular case, as they are secondary sexual characteristics, probably related to the sexual behaviour. The presence of tegumentary exocrine glands secreting semiochemicals is known for several groups of Scarabaeinae (e.g., Tribe 1975; Pluot-Sigwalt 1982; Houston 1986; Burger et al. 1990, 1995 a, 1995 b, 2002, 2008). Pluot-Sigwalt (1982), having studied several dung beetle genera, found an intricate distribution pattern of exocrine glands in the abdominal ventrites. In some cases, both males and females had these glands (although there was a remarkable sexual dimorphism in their numbers and dispositions), while in other cases they were restricted to just one sex. Interestingly enough, there was a clear distinction between telecoprid and paracoprid groups, with a much larger number of genera of the first group – to which Sylvicanthon belongs – presenting glands than of the second ( Pluot-Sigwalt 1982).

The row of setae seen exclusively in the foveae of S. obscurus ( Fig. 16C ) may work as a kind of net to capture and concentrate close to the female’s body the pheromones released into the air by the abdominal glands. Houston (1986), for instance, described two exocrine gland complexes associated to a tuft of setae present in the front legs of some species of Onitis Fabricius, 1789 . Following his hypothesis, the pheromone secreted by one group of glands present only in males could work as a trail marker and, in this way, it could indicate to females the path to the dung mass or to the nest entrance. Therefore, the tuft of setae associated to exocrine glands would serve to deposit the pheromone to the substrate while the path is being marked. Although in S. obscurus these glands and associated setae are exclusive to females, not males, it is still possible that some similar behaviour occurs in this species. Studies on the reproductive habits of the Sylvicanthon species are necessary in order to know the actual function of the abdominal foveae in this group.

AEDEAGUS. As discussed above, males of Sylvicanthon do not differ from the majority of the other telecoprid dung beetles in not having horns or any other kinds of armaments. In these groups, it is in the shape of the parameres and pieces of the internal sac and their counterpart in the female sexual organ that the sexual selection should act in a more intense way (see for a discussion on the evolution and anatomy of the copulatory organs of a model species of dung beetle: Werner & Simmons 2008; House & Simmons 2003; Tarasov & Solodovnikov 2011). This fact is reflected in the large variety of forms of aedeagi seen in Sylvicanthon , where even closely related species with a very similar external morphology may have entirely distinct parameres. The morphology of the female genitalia was not studied for this work, but, as seen in other groups of Scarabaeinae that had it examined (e.g., Zunino 1971, 1972, 1975, 1976, 1978, 2012; Zunino & Halffter 1988; Marchisio & Zunino 2012), it probably varies in a similar degree as the aedeagus. As emphasized by Zunino (1987), Méndez & Córdoba-Aguilar (2004) and Ah-King et al. (2014), among several other authors, there is no reason to suppose that there is a higher morphological diversity among the male genitalia than among the female ones, since these structures evolve jointly in a complex interaction of evolutionary pressures derived from female choice and intra- and intersexual conflicts (see also Simmons 2014).

It is in the candezei and bridarollii subgroups where we can find the greatest diversity in paramere shapes. In the first subgroup, the parameres may be straight and simple (i.e., without any notch or keel in the ventral region; S. foveiventris , Fig. 17B ), may have a strong ventral keel and a short notch on its posterior region ( S. genieri sp. nov., Fig. 17D ), or may bear a deep ventral notch without any keel ( S. candezei , Fig. 17C ). In the bridarollii subgroup, in turn, the parameres may be simple ( S. bridarollii , Fig. 18A ), with a short ventral keel ( S. edmondsi sp. nov., Fig. 18C and S . attenboroughi sp. nov., Fig. 18D ) or be highly modified with a ventral keel so deep that almost divide the parameres into two halves ( S. seag sp. nov., Fig. 18B ). With the exception of S. foveiventris and S. bridarollii , all the other species have sharply asymmetric parameres, with external faces differing from one another by being either concave or flat.

In the furvus subgroup, whose parameres are apically bifurcate and have no ventral keels or notches, the difference between the species is subtler in three of the four species. In S. furvus ( Fig. 19C ) and S. mayri sp. nov. ( Figs 19E , 44 A–B), the inferior branch of the bifurcation is distinctly projected and divergent from the superior branch, whereas in S. monnei sp. nov. ( Figs 19D , 44 C–D) the inferior branch is only slightly projected and is parallel to the superior branch. In those three species, both branches are connected by a fine membrane, but in the fourth member of their subgroup, S. obscurus , both branches are free and the superior branch projects itself much stronger than the inferior one, which has an acuminate apex ( Fig. 19B ). In the aequinoctialis subgroup, different from the other subgroups discussed before, there is no difference between the species ( Fig. 17 E–F); in both, the parameres are simple and have a broad depressed area at the apex which extends posteriorly farther in the left paramere than in the right. Finally, S. securus ( Fig. 19A ) and S. enkerlini ( Fig. 17A ) have elongate and simple parameres, without any ventral keel or notch.

Biogeography

Sylvicanthon are largely distributed in the Neotropical region as defined by Morrone (2014, 2015 b), occurring from Honduras to the Amazon Basin, the Atlantic Forest in NE and SE Brazil, and in the transition zone between the latter two biomes and the Brazilian Cerrado and Caatinga ( Figs 20 , 22 ). Among the areas into which Morrone (2014) divided the Neotropical region, Sylvicanthon is present in two of the three subregions (Brazilian and Chacoan), in all of the six dominions and in 24 of the 53 provinces. Four species ( S. genieri sp. nov., S. bridarollii , S. edmondsi sp. nov. and S. furvus ) are also present in the province of Paramo, in the South American Transitional Zone between the Neotropical and the Andean regions.

The candezei species group and S. enkerlini are very distinct with respect to the environments in which they live ( Fig. 20 ). Species of the first group are distributed in the four great areas of tropical rainforests in the Neotropical region – the Central American tropical forests, the Choco, the Amazon forest and the Atlantic Forest –, where temperature and humidity are high, there is a dense and continuous canopy and the sunlight incidence on the surface of the soil, in the leaf litter in the understory and in the lower tree layers is very low. Thanks to these factors, mammalian dung, upon which Sylvicanthon feed, remains fresh and available for consumption and nidification for a longer time in those habitats. In contrast, S. enkerlini occurs in the transitional zones between Amazonia, Cerrado, Caatinga and the Atlantic Forest, in more open, shorter and drier semidecidual forests, such as the Mata dos Cocais, in Maranhão, the tableland forests (from the Portuguese ‘ matas de tabuleiro ’) in the coast of Ceará, and dryer forested areas typical of Cerrado and Caatinga in the interior of the Brazilian states of Piauí, Bahia, and Minas Gerais ( Figs 20 , 22 ; see more details on the distribution of S. enkerlini in the ‘Natural history’ section for this species).

Martínez et al. (1964) raised the hypothesis that Sylvicanthon (cited as “ grupo aequinoctialis ” of Glaphyrocanthon ) originated in Amazonia and, from there, it dispersed to the other areas where it is present today. The greatest species diversity in the genus is indeed located in the Western Amazonia, in cloud forest areas close to the slopes of the Andes in Bolivia, Peru, Ecuador and Colombia, where S. genieri sp. nov., S. proseni , S. bridarollii , S. edmondsi sp. nov., S. attenboroughi sp. nov., S. furvus and S. mayri sp. nov. occur. The distribution of three of these species – S. proseni ( Fig. 30 ), S. bridarollii and S. attenboroughi sp. nov. ( Fig. 34 ) – extends eastwards into the Amazon Basin, but the latter two are absent on the left banks of the Amazon river. In the latter region, it occurs S. securus ( Fig. 41 ) and S. seag sp. nov. ( Fig. 34 ), whose distribution extends farther north into the Guianas and, in the case of the latter species, also towards Venezuela and the island of Trinidad, the only insular record for the genus. Two further species occur in the Amazon region: S. monnei sp. nov., which is distributed throughout the forests of northern Mato Grosso and southern Pará, a dryer Amazon area with a more pronounced seasonality where rainy and dry seasons are sharply demarcated ( Fig. 41 ), and S. candezei , which exhibits a very peculiar bow-shaped distribution from the mouth of the Tapajós River through the dry forests of the transitional region bordering the Cerrado in southeastern Pará and goes southwards into southern Amazon and its transition zone with the Pantanal region ( Fig. 24 ). Together, therefore, eleven of the fifteen species of Sylvicanthon are typical elements of the Amazon forest. Following the hypothesis of Martínez et al. (1964), which is championed here, at least three distinct dispersal waves departed from the Amazonian centre of origin: the ancestors of S. foveiventris ( candezei subgroup) and S. obscurus ( furvus subgroup) towards the Atlantic forest, and those of S. aequinoctialis into Central America ( Fig. 20 ).

The dispersal of Sylvicanthon aequinoctialis into the forests of Central America probably occurred only after the uplift of the Isthmus of Panama. The chronology about when this landbridge connecting North and South Americas was formed enough to allow the passage of fauna is yet a matter of great debate. Some authors argue for an older formation, where the vast marine passage connecting the Caribbean Sea and the Pacific Ocean, the so-called Central American Seaway, was closed in the Miocene, some 15 million years ago, and, thereafter, the right conditions allowing the biotic movement between the Americas were fully present (e.g., Montes et al. 2012 a, 2012 b, 2015; Jaramillo et al. 2017). Other authors, instead, argue that, between 15 million and about three or four million years ago, Central America and the northwest coast of South America were indeed close (about 200 km away from one another), but yet separated by a very deep sea, which prevented faunistic interchange until the final completion of the Isthmus of Panama about 3 million years ago (e.g., Coates & Stallard 2013; O’Dea et al. 2016). This latter case would be an analogous situation to the current Wallace and Lydekker lines, on the limits, respectively, of the Sunda and Sahul continental shelves, in Indonesia, which mark the separation between the faunas of the Oriental and Australasian biogeographical regions ( Coates & Stallard 2013).

Curiously enough, studies on the dung beetle dispersal from South to North America argue that this movement occurred in two independent waves, one during the Miocene and another during the Plio– Pleistocene ( Kohlmann & Halffter 1988, 1990; Edmonds 1994; Price 2009; Halffter & Morrone 2017). Not coincidently, these two migratory waves correspond exactly to those two distinct phases in the geological history of the Isthmus of Panama as theorized by Coates & Stallard (2013) and O’Dea et al. (2016) (but see the replies by Jaramillo et al. 2017 and Molnar 2017). The first migratory wave would have advanced through a chain of several small islands in the then-existing strait between Central America and South America’s northwest coast, while the second wave would have occurred during the gradual advance of the South American tropical forest into Central America after the uplift of the Isthmus of Panama was completed. Being a tropical forest dweller, S. aequinoctialis should have invaded Central America from the Choco forest in northern South America during this second migratory wave ( Fig. 20 ), in the same way as done by Glaphyrocanthon and allied groups ( Kohlmann & Halffter 1990).

Together, those migratory waves were part of the so-called Great American Biotic Interchange of the Neogene. Several groups of dung beetles of a clear South American origin, such as Phanaeus MacLeay, 1819 , Coprophanaeus d’Olsoufieff, 1924 , Dichotomius Hope, 1838 , Ontherus Erichson, 1847 , Canthidium , Ateuchus , Uroxys , Deltochilum , Canthon and Scybalocanthon , trod this very same path from the southern continent towards Central America and, in the most successful cases, to North America , places where some of these groups have undergone a remarkable adaptive radiation. The reverse path, that is, from North to South America , was traveled by a much smaller number of lineages, including Onthophagus Latreille, 1802 and Copris Geoffroy, 1762 . It is interesting to note that this was exactly the opposite migration pattern of the mammals, whose primary direction was southwards and which were the focus of much of the studies on the Great American Interchange (e.g., Simpson 1980); in this context, it is especially important to remember that mammals are the main group of vertebrates which dung beetles depend on for food. Birds, on the other hand, migrated mainly from south to the north ( Weir et al. 2009), in the same way as it seems to have occurred with some scarab groups (e.g., Dynastes Kirby, 1825 ; Huang 2016) and dung beetles (personal observations 18). We agree that future studies on the participation of the dung beetles and other insect taxa in the Great American Interchange will shed light on the ecological factors driving the distinct patterns seen in different groups of organisms. But most importantly, the dung beetle northwards migration pattern shows that simplistic hypotheses championing a ‘more competitive nature’ of North American elements fail to explain the whole biotic movement, to say the least.

The other two species of the candezei group, occuring outside Amazonia, are S. foveiventris and S. obscurus , both inhabitants of the Atlantic Forest ( Figs 24 , 41 ). Although the South American Dry Diagonal formed by the Caatinga, Cerrado and Chaco separates the humid forests of Amazonia and the Atlantic Forest, both biomes have never been completely isolated from one another and, in fact, they have a long and complex history of interactions throughout the Cenozoic ( Costa 2003; Batalha-Filho et al. 2013; Daniel & Vaz-de-Mello 2016; Ledo & Colli 2017; and references cited therein). Studies based mainly on the distribution and phylogeny of land vertebrates showed that there was a great mixture of fauna between those two areas, especially during intervals of more humid climate, when gallery forests and patches of humid forests penetrated more strongly into the interior of the dry regions of central South America and in this way served as bridges connecting the two biomes. Batalha-Filho et al. (2013), for example, found that there were two great bird interchange movements between Amazonia and the Atlantic Forest during the Neogene, the oldest having occurred between southwestern Amazonia and southern Atlantic Forest during the Miocene (24 to 5 million years ago) and the most recent one during the Plio–Pleistocene (the last 5 million years) between northeastern Amazonia and the northern portion of the Atlantic Forest. In this way, the phylogeny of South American animals shows several cases where sister species or sister groups occur separately, one in the Amazon, the other in the Atlantic Forest.

If Sylvicanthon and other Deltochilini took part in episodic migratory waves as those of the example discussed above, or if they had a history of a continuous communication between the Amazonia and the Atlantic Forest, is still a question for further research. What is clear, nonetheless, is that at least two lineages – the one which gave origin to S. obscurus and the other which originated S. foveiventris – arrived independently at the Atlantic Forest from the Amazon region. As argued by Costa (2003), the dispersal history of any animal group between these two biomes should have been extremely complex and deeply influenced by local events that varied over time, scale and mainly in the effect that they had in different groups of organisms, as shifts in river courses, climate changes or refuge formations. Therefore, hardly a single phenomenon or pattern will be found to explain the whole movement of fauna between the Atlantic Forest and the Amazon which Sylvicanthon species could fit in. In order to understand in deep detail the biogeography of Sylvicanthon , it is necessary that we first study empirically the phylogenetic relationships between its species.

It is also interesting to note here that each of the polytypic subgroups of Sylvicanthon show both a distribution and a phylogenetic pattern that could lead an evolutionary biologist to classify them as superspecies, that is “monophyletic group[s] of allopatric or nearly allopatric taxa that are known or believed to have evolved to the species level” ( Amadon 1966, 1968; see also Mayr 1931, 1942, 1963, 1970; Mayr & Diamond 2001; Mayr et al. 1953; Haffer 1986; Mallet 2007). This observation may indicate that the species of at least some of the subgroups represent very recent events of allopatric speciation, since even in cases where the former geographic barrier seems to have already collapsed, the allospecies (i.e., each constituent species of a superspecies) remain largely allopatric or parapatric in relation to its sister or other closely-related species.

For instance, in the bridarollii subgroup, the distribution of S. bridarollii and S. attenboroughi sp. nov. are parapatric, with a clear overlap in southern Peru and northwestern Brazil, but neither penetrates completely into the range of the other, with S. bridarollii being the only species occurring both south to Bolivia and north to Colombia, while S. attenboroughi sp. nov. is the only one to occur farther east in Brazil ( Fig. 34 ). Sylvicanthon edmondsi sp. nov., on the other hand, is found in sympatry with S. bridarollii throughout its distribution range, although no possible case of hybridization has been found, indicating thereby that a complete reproductive isolation must already have been developed, which in turn made sympatry possible without the merging of both species. The fourth member of this subgroup/superspecies, S. seag sp. nov., is the only one completely allopatric in relation to the other three species, being separated from them by the Amazon River.

The case of the candezei subgroup is probably distinct, since the three allospecies are distributed very apart from each other, one in the Atlantic Forest ( S. foveiventris ), the second in the southeastern Amazonia ( S. candezei ), and the third on the slopes of the Andes ( S. genieri sp. nov.) ( Fig. 24 ). This may indicate instead a relict distribution pattern of a subgroup that was once widely distributed in the Amazon Basin and, perhaps, even in the Atlantic Forest. The distribution of the species in the furvus subgroup ( Fig 41 ), in turn, is so poorly known that it is difficult to make any generalization. Given the very suble morphological differences between its three Amazonian allospecies ( S. furvus , S. monnei sp. nov. and S. mayri sp. nov.), however, we believe this subgroup, in the same way as discussed above for the bridarollii subgroup, should have experienced a series of recent speciation events. As for the aequinoctialis subgroup, the continuing uplift of the Colombian Andes, with the subsequent isolation of two independent demes on either side of the mountain range, should have been the vicariance event responsible for the division of the ancestral species that gave origin to S. aequinoctialis and S. proseni ( Fig. 30 ).

Although the concept of superspecies has been mostly applied to vertebrate taxa (particularly birds), some authors such as Huang (2017), dealing with the genus Dynastes , and Mayr (1963: 501), referring to the Mycotrupes LeConte, 1866 studied by Olson et al. (1954), have already used it to refer to monophyletic groups of allopatric species in Scarabaeoidea . Although one could argue that the adoption of this term would only add more uncertainty to an already rather turbulent epistemological context (see the discussion on the species concept above), the first author of this monograph believes that its synthesis of both biogeographical (allopatry) and phylogenetic (monophyly) information is of great value for evolutionary studies, particularly for those dealing with allopatric speciation processes (i.e., the formation of geographic barriers to the gene flow, the development of reproductive isolation and the multiplication of species). Therefore, he encourages a more widespread application of the superspecies concept in biogeographical and taxonomic works dealing with scarab beetles. See Amadon (1966, 1968) for a more detailed defence of the value of the term superspecies.

Natural history

Literature and label information make it clear that Sylvicanthon species are primarily coprophagous, consuming human faeces and other primate, pig and cow dung, although there are also some records of specimens feeding on carcases. All the species are nocturnal, as can be inferred from their large eyes with smooth corneas ( Caveney & McIntyre 1981; McIntyre & Caveney 1998) and from their usual dark colouration ( Hernández 2002; Feer & Pincebourde 2005). It is curious to note that, in this genus context, some species can be classified as eurytopic, such as S. proseni , S. aequinoctialis , S. bridarollii and S. seag sp. nov., as they live in a wide altitudinal gradient, have a vast distribution and are very abundant (usually, they are among the most abundant species in the dung beetle communities of which they are part), while other species, thanks to their rarity or environmental specificity (or both factors), are clearly stenotopic, like S. securus , S. foveiventris and, especially, S. furvus , S. monnei sp. nov. and S. mayri sp. nov. Sylvicanthon foveiventris , for instance, although not rare, is found only in forest areas higher than 600 m, whereas S. securus , despite having a wide distribution in the north of the Amazon region, is a very rare species. In fact, this relationship between specialist and less-abundant species in contraposition to more generalist and more abundant ones is clear when we compare the relative abundance of S. seag sp. nov. and S. securus in places where those two species occur in sympatry: the relation can vary from three up to 65 S. seag sp. nov. for each S. securus in the same area, according to the data gathered for the present work (see more details in the discussion of S. securus ). Nothing is known about the nesting behaviour of Sylvicanthon , although it is reasonable to assume they belong to Pattern IV as defined by Halffter & Edmonds (1982), where most of the American Deltochilini are classified to.

Identification key to the species of Sylvicanthon Halffter & Martínez, 1977

1. Clypeus with four teeth ( Fig. 6A ). Anterior margin of profemora with denticulation at its apical half ( Fig. 9B ). Protibiae with three large and widely separated teeth and with a strong expansion on its internal edge ( Fig. 11A ). Hypomera with posterior part with about 5 long setae forming a longitudinal row close to external edge ( Fig. 9D ). Ecotone between the Amazon rainforest, Cerrado, and the Atlantic Forest in Brazil (Maranhão, Piauí, Ceará, Bahia, and Minas Gerais) .......................... enkerlini group: Sylvicanthon enkerlini ( Martínez et al., 1964) comb. nov.

– Clypeus with only two small apical teeth ( Fig. 6 B–G). Anterior margin of profemora without denticulation ( Fig. 9A ). Protibiae with two or three small or medium-sized teeth; if teeth are large, internal edge always straight; if small, internal edge straight or expanded ( Figs 11 C–J). Hypomera with posterior part glabrous. Humid tropical forest in Central America, Amazonia and Atlantic Forest ...................................................................................................................... candezei group: 2

2. Protibiae with two small teeth and internal edge always straight ( Fig. 11J ) ... candezei subgroup: 3. – Protibiae with three small or large teeth and internal edge straight or expanded ( Fig. 11 C–I) ........ 5

3. Head and pronotum purplish and elytra green or dark blue ( Fig. 23A ). Pronotum and elytra with no microsculpture at centre. Metafemora with coarse elongate punctation at base ( Fig. 13A ). Females with three pairs of lateral foveae on abdomen between ventrites I–II, II–III, and III–IV ( Fig. 14B , 16D ). Parameres simple, without ventral keel or notch ( Fig. 17B ). Forests above 600 m in southeastern Brazil.................................................... Sylvicanthon foveiventris ( Schmidt, 1920) .

– Different colour pattern. Centre of pronotum and elytra with strong microsculpture ( S. genieri sp. nov.) or with smooth or even absent microsculpture ( S. candezei ). Metafemora without coarse punctation at base. Abdomen of both sexes without lateral foveae. Parameres with ventral keel and/ or notch ( Fig. 17 C–D). Amazonia .................................................................................................... 4

4. Dorsal colouration of the body dark green ( Fig. 25A ). Centre of pronotum, elytra, and pygidium without microsculpture (northern populations) or with flat alveolar microsculpture weakly marked (southern populations); in both cases, micropunctation abundant (denser on pronotum than on elytra). Posterior edge of ventrite V, in males, with a weak flange covering anterior edge of ventrite VI ( Fig. 14D ); in females, posterior edge of ventrite V with a strong medial expansion over ventrite VI ( Fig. 14E ). Parameres with strong ventral notch and without ventral keel ( Fig. 17C ). Humid tropical forests from the mouth of the Tapajós River down to the semideciduous forests of southern and southeastern Amazonia in Brazil (Pará and Mato Grosso) .......................................................... ............................................................................................... Sylvicanthon candezei (Harold, 1869) .

– Dorsal colouration bright coppery (occasionally, with greenish reflections on head and on the sides of elytra) ( Fig. 27A ). Centre of pronotum, elytra, and pygidium with strong alveolar microsculpture obliterating micropunctation. Posterior edge of ventrite V, in males, without medial flange (rarely, with a very weak trace of flange over anterior edge of ventrite VI); in females, posterior edge of ventrite V with weak medial flange over ventrite VI. Parameres with short ventral notch and strong ventral keel ( Fig. 17D ). Cloud forests of western Amazonia and slopes of the Andes in Ecuador and Peru...................................................................................................... Sylvicanthon genieri sp. nov.

5. Suture between submentum and gula Y-shaped ( Fig. 8B ). Protibiae with internal edge straight and medium- or large-sized teeth at their apical half ( Fig. 11B ). Ventral face of metafemora with posterior margin ( Fig. 31 ). Humeral carina always presente ( Fig. 12C ). Parameres with depressed apical area ( Fig. 17 E–F); without ventral keel or notch. Central America and Amazon Basin ....................................................................................................... aequinoctialis subgroup: 6.

– Suture between submentum and gula rounded ( Fig. 8A ). Protibiae with internal edge straight or expanded and with small and narrow teeth at their apical third ( Fig. 11 C–I). Ventral face of metafemora without posterior margin ( Fig. 13 ). Humeral carina usually absent (present only in some specimens of S. obscurus ). Parameres without depressed apical area; with or without ventral keel and notch. Amazon Basin and Atlantic Forest .................................................................................. 7

6. Dorsum, ventral surface of metafemora, and pygidium with bright and lustrous appearance. Centre of pronotum and elytra without microsculpture or with very subtle flat alveolar microsculpture. Ventral surface of metafemora and pygidium with very fine three-dimensional alveolar microsculpture (i.e., with tiny alveoli). Posterior edge of head with margin between eyes always complete. Posterior margin of metafemora extending from apex to at least the height of trochanter (usually going beyond it) ( Fig. 31A ). Protibial spur of females spiniform ( Fig. 15G ). From Honduras to northern Colombia.................................................... Sylvicanthon aequinoctialis (Harold, 1868) comb. nov.

– Dorsum, ventral surface of metafemora, and pygidium with diffuse shine and with a more matte appearance. Centre of pronotum, ventral surface of metafemora and pygidium with very strong three-dimensional alveolar microsculpture; micropunctation ranging from very dense to absent. Posterior edge of head with margin between eyes complete, incomplete, or absent. Posterior margin of metafemora not reaching trochanter (usually extending little beyond the apical half of metafemur) ( Fig. 31B ). Protibial spur of female bifid ( Fig. 15H ). Amazonia ( Colombia, Ecuador, Brazil, Peru, and Bolivia) ............................................ Sylvicanthon proseni (Martínez, 1949) stat. et comb. nov.

7. Protibiae with internal margin straight ( Fig. 11 H–I) (moderately expanded in northern populations of S. bridarollii ( Fig. 11G ) and only slightly expanded in southern populations of this species ( Fig. 11F ); in these cases, pronotum with alveolar microsculpture). Lateral portions of metaventrite completely glabrous. Abdomen of both sexes without lateral fovea. Amazonia ............... bridarollii subgroup: 8

– Protibiae with internal edge clearly expanded ( Fig. 11 C–E). Pronotum without alveolar microsculpture. Sides of metaventrite with some few setae near metacoxae ( Fig. 7B ). Abdomen of females with ( Fig. 16 A–C) or without lateral foveae (absent only in S. securus ) Amazonia and Atlantic Forest .................................................................................................................................11

8. Hypomeral cavity covered at centre by long yellowish erect setae ( Fig. 25 C–D). Protibiae with internal edge moderately ( Fig. 11G ; northern populations) or only slightly ( Fig. 11F ; southern populations)expanded.Metafemora with( Fig. 13G )or without( Fig. 13H ) coarse elongate punctation at base. Metaventrite covered at centre by strong three-dimensional alveolar microsculpture. Pygidium with three-dimensional alveolar microsculpture strongly marked and with very subtle, almost imperceptible punctation. Parameres symmetrical (both with external face flat) and simple, without ventral keel or notch ( Fig. 18A ). Western Amazonia in Colombia, Ecuador, Brazil (Acre, Rondônia), Peru and Bolivia.......................................... Sylvicanthon bridarollii (Martínez, 1949) .

– Hypomeral cavity glabrous at centre ( Fig. 35 A–B; rarely with one or two very short setae); setae, if present, restricted to anterior and posterior regions of the cavity. Protibiae with internal edge straight ( Fig. 11 H–I). Metafemora without coarse elongate punctation at base (except in rare especimens of S. attenboroughi sp. nov.) ( Fig. 13F ). Metaventrite with very fine alveolar microsculpture at centre and progressively more diffuse towards posterior region. Pygidium with or without alveolar microsculpture strongly marked and with micropunctation of variable density, but always evident. Parameres asymmetrical (external face of left paramere excavated and external face of right paramere flat) and with ventral keel ( Fig. 18 B–D). Amazonia ........................................................................ 9

9. Anterior edge of ventrite VI of females distinctly covered by medial flange of posterior edge of ventrite V. Parameres with strong ventral notch and with ventral carina strongly projected; apical half of parameres squared ( Fig. 18B ). Northern Amazonia, in Trinidad, Venezuela, Guyana, Suriname, French Guiana and Brazil; except the region east of its mouth, always limited to the left banks of the Amazon River .......................................................................................... Sylvicanthon seag sp. nov.

– Anterior edge of ventrite VI of females only subtly covered by weak medial expansion of posterior edge of ventrite V. Parameres with elongate appearance and shorter ventral keel ( Fig. 18 C–D). Western and southern Amazonia ..................................................................................................... 10

10. Head dark purple, pronotum with strong greenish or bluish shine at centre and purplish on sides, elytra dark blue or purple, and meso- and metafemora orangish-brown or yellowish ( Fig. 38 A–B). Dorsal surface of head with very subtle, almost imperceptible micropunctation. Northwestern Amazonia, mainly in Sub-Andean areas in Colombia, Ecuador and Peru......... Sylvicanthon edmondsi sp. nov.

– Head and pronotum entirely very dark purple (almost black), without central greenish or bluish central spot on pronotum; elytra dark green or dark blue; meso- and metafemora orangish-brown, reddish-brown or dark brown ( Fig. 38 C–D). Dorsal surface of head with micropunctation evident on posterior region of clypeus and mainly on frons. Southern Amazonia, on the right margin of the Amazon River, in Brazil and Peru........................................... Sylvicanthon attenboroughi sp. nov.

11. Each clypeal tooth with base covered by a separate row of setae. Elytra with dense micropunctation and without trace of alveolar microsculpture. Internal edge of protibiae strongly expanded ( Fig. 11C ). Abdomen of both sexes without lateral foveae. Parameres simple, without apical bifurcation (Fig. 88). Northern Amazonia, in Suriname, French Guiana and Brazil................................................................... Sylvicanthon securus ( Schmidt, 1920) comb. nov.

– Pair of clypeal teeth with base covered by a single row of setae. Elytral sculpture variable, alveolar microsculpture present or not. Internal margin of protibiae not as strongly expanded as in S. securus ( Fig. 11D–E ). Females with three pairs of foveae at the sides of the abdomen between ventrites I–II, II–III, and III–IV ( Fig. 16 A–C). Parameres bifurcate at apex ( Fig. 19B–E ). Southern and western Amazon and the Atlantic Forest ......................................................................... furvus subgroup: 12

12. Elytra with strong micropunctation and without microsculpture at centre and with strong threedimensional alveolar microsculpture on sides and apex. Humeral carina present in about two-fifths of the specimens ( Fig. 12C ). Metafemora without coarse elongate punctation at base ( Fig. 13B ). Lateral foveae of abdomen of females covered by row of long setae ( Fig. 16C ). Parameres with branches of apical bifurcation free, with no membrane connecting them ( Fig. 19B ). Northern Atlantic Forest, from Alagoas to Espírito Santo ( Brazil) ................ Sylvicanthon obscurus ( Schmidt, 1920) .

– Elytra with different sculpture pattern; either entirely microsculptured or entirely smooth. Humeral carina always absent. Metafemora with coarse elongate punctation ( Fig. 13C, E ) (except S. mayri sp. nov., Fig. 13D ; see comments above). Lateral foveae of female abdomen glabrous ( Fig. 16 A– B). Parameres with branches of apical bifurcation connected by a fine membrane ( Fig. 19 C–E). Amazonia ........................................................................................................................................ 13

13. Head with diffuse shine and strong alveolar microsculpture covering the entire tegument. Meso- and metafemora with ventral surface completely covered by rivose microsculpture, with no smooth areas and with strong three-dimensional alveolar microsculpture. Elytra with diffuse shine and entirely covered by strong three-dimensional alveolar microsculpture. Pygidium with diffuse alveolar microsculpture at centre and apex, and with strong rivose microsculpture at base. Eastern slopes of the Andes in Peru and Bolivia............................................... Sylvicanthon furvus ( Schmidt, 1920) .

– Head shiny and with weak alveolar microsculpture, which, in some areas, is totally absent. Meso- and metafemora with ventral surface almost entirely smooth and with evident micropunctation, except on anterior apical area with rivose microsculpture. Elytra shiny, entirely smooth or with very diffuse microsculpture and ill-delimited alveoli. Pygidium completely smooth at centre and with rivose microsculpture occasionally present on sides of base. In general, in lower Amazon areas; with no records from the Peruvian and Bolivian Andes ......................................................................... 14

14. Elytra smooth, with no trace of microsculpture throughout its surface. Metafemora with coarse elongate punctation at base ( Fig. 13C ). Abdominal foveae of females always well marked and deep ( Fig. 16A ). Parameres with inferior branch of apical bifurcation weakly projected or straight, without posterior excavation ( Fig. 44C ). Southern Brazilian Amazonia (Pará and Mato Grosso)................................................................................................. Sylvicanthon monnei sp. nov.

– Elytra with diffuse microsculpture, which is difficult to see and has ill-defined alveoli. Metafemora with coarse punctation modified in fine, simple points at centre of base ( Fig. 13D ; but see discussion above). Abdominal foveae of females very shallow (especially between ventrites III–IV) ( Fig. 16B ). Parameres with inferior branch of apical bifurcation well projected and largely divergent from superior branch, with strong posterior excavation ( Fig. 44A ). Western Amazonia in Colombia, Brazil (Amazonas and Acre) and Peru................................................... Sylvicanthon mayri sp. nov.

Figures

Fig. 6. Head, dorsal view. A. Sylvicanthon enkerlini (Martínez et al., 1964) comb. nov. B. S. candezei (Harold, 1869). C. S. genieri sp. nov. D. S. proseni (Martínez, 1949) stat. et comb. nov. E. S. seag sp. nov. F. S. securus (Schmidt, 1920) comb. nov. G. S. furvus (Schmidt, 1920).

Fig. 14. A. Ventral view of the anterior region of the pterothorax of S. proseni (Martínez, 1949) stat. et comb. nov. Abbreviations: MsEs = mesoventrite; SEs = meso-metaventrite suture; MsC = mesocoxa; RAMtEs = anterior region of metaventrite. B–E. Sexual dimorphism in the abdominal ventrites of Sylvicanthon Halffter & Martínez, 1977. B. S. foveiventris (Schmidt, 1920), ♀. Note the three lateral foveae between ventrites I–II, II–III, and III–IV and the broad ventrite VI at the middle, without emargination on its posterior edge and without its anterior edge covered by a medial flange of ventrite V. C. S. proseni stat. et comb. nov., ♀, ventrite VI broad. D. S. candezei (Harold, 1869), ♂. Note how the ventrite VI is narrow at the middle due to the emargination on its porterior edge and to the medial flage of the posterior edge of the ventrite V. E. S. candezei (Harold, 1869), ♀. Despite its posterior edge not being emarginate, its anterior edge is covered by a well-developed medial flange of ventrite V, which gives a narrowed aspect to the centre of ventrite VI.

Fig. 35. Variation on the pilosity at the centre of the hypomeral cavity among members of the bridarollii subgroup. A. Sylvicanthon seag sp. nov. B. S. attenboroughi sp. nov. C–D. S. bridarollii (Martínez, 1949). Note the first two species have the hypomeral cavity entirely glabrous or with setae limited to its periphery (centre always glabrous), whereas S. bridarollii possesses long and dense setae throughout the tegument of the hypomeral cavity.

Fig. 21. Sylvicanthon enkerlini (Martínez et al., 1964) comb. nov. A. Dorsal view. B. Ventral view. C. Holotype and its labels.

Fig. 22. Distribution of Sylvicanthon enkerlini (Martínez et al., 1964) comb. nov. Note that the species occupies transition zones between the Amazonia, Cerrado, Caatinga and Atlantic Forest biomes, regions much dryer and more open than the humid forests occupied by the species of the candezei group.

Fig. 10. Lateral view. A. Canthon (Francmonrosia) sp. B. Sylvicanthon enkerlini (Martínez et al., 1964) comb. nov. C. S. candezei (Harold, 1869).

Fig. 33. Sylvicanthon bridarollii (Martínez, 1949) type material. A. Holotype and its labels. B. Allotype and its labels.

Fig. 34. Distribution of the four species of the bridarollii subgroup.

Fig. 36. Clinal morphological variation in S. bridarollii (Martínez, 1949). Note that, from south to north, the colouration of metafemora gradually fades from dark brown to reddish brown and orange with a light-brown base in populations of northern Peru, Ecuador, and Colombia, and that the internal margin of protibiae (indicated by the red arrows) becomes progressively much subtler. Other features varying along this north-south cline are the presence of coarse elongate punctures at the base of metafemora (present in southern populations and absent in the northern ones) and of a fine transverse line on the posterior edge of pronotum (absent in the southern populations). See the text for more details.

Fig. 37. Sylvicanthon seag sp. nov. A. Dorsal view of a bicolour specimen (present only on the banks of the Amazon River and in the Brazilian state of Maranhão). B. Dorsal view of a bluish specimen. C. Dorsal view of a greenish specimen. D. Ventral view of a greenish specimen.

Fig. 38. A–B. Sylvicanthon edmondsi sp. nov. A. Dorsal view. B. Ventral view. C–D. Sylvicanthon attenboroughi sp. nov. C. Dorsal view. D. Ventral view.

Fig. 39. Sylvicanthon securus (Schmidt, 1920) comb. nov. A. Dorsal view. B. Ventral view.

Fig. 40. Type material of three species of Sylvicanthon described by Schmidt (1920). A. S. secures (Schmidt, 1920) comb. nov. a. Lectotype. b. Lectotype’s labels. B. S. obscurus (Schmidt, 1920). a. Lectotype. b. Lectotype’s labels. c. Paralectotype 1. d. Paralectotype 3. e. Paralectotype 9. C. S. furvus (Schmidt, 1920). a. Lectotype. b. Lectotype’s labels. c. Paralectotype 4. d. Paralectotype 6. e. Paralectotype 3. f. Paralectotype 1.

Fig. 41. Distribution of Sylvicanthon securus (Schmidt, 1920) comb. nov. and the four species of the furvus subgroup.

Fig. 42. Sylvicanthon obscurus (Schmidt, 1920). A. Dorsal view of the purplish form. B. Dorsal view of the yellowish form. C. Ventral view.

Fig. 46. A. Canthon cobosi (Pereira & Martínez, 1960) stat. et comb. nov., holotype. B–C. Canthon machadoi (Martínez & Pereira, 1967) comb. nov. B. Paratype 2. C. Paratype 4.

Fig. 8. Head, ventral view. A. Sylvicanthon bridarollii (Martínez, 1949). B. S. proseni (Martínez, 1949) stat. et comb. nov. See the differences in the shape of the suture between submentum and gula in these two species (red arrows), which is rounded in S. bridarollii and most of the other species of Sylvicanthon, while it has a ‘Y’-shape in the two representants of the aequinoctialis subgroup.

Fig. 9. A. Sylvicanthon bridarollii (Martínez, 1949), profemur. Note that, as occurs in the other species of the candezei group, there is no denticulation on the anterior edge, which is, by contrast, completely smooth. B. S. enkerlini (Martínez et al., 1964) comb. nov., profemur.Arrow indicates the denticulation of anterior edge, a condition present in both sexes and unique in the genus. C. S. enkerlini, metepisternum. Arrow indicates the small tubercle at the suture between metepisternum and the lateral region of metaventrite. D. S. enkerlini, hypomeron. Notice the row of long setae present on the lateral of posterior region, a condition seen uniquely in this species in Sylvicanthon.

Fig. 5. Subdentate tarsal claws in different groups of dung beetles. The presence of this form of tarsal claw was seen as evidence for a close phylogenetic relationship between several groups of Scarabaeinae distributed over the continents of former Gondwana. A. Protarsal claws of Sylvicanthon proseni (Martínez, 1949) stat. et comb. nov. Red arrow indicates the basal agularity of the claw. B. Pseudocanthon vitraci (Fleutiaux & Sallé, 1889) (modified from Matthews 1966). C. Canthonella parva (Chapin, 1930) (modified from Halffter & Martínez 1967). D. Lepanus penelopae Matthews & Weir, 2002 (left) and L. loftyensis Matthews & Weir, 2002 (right) (modified from Matthews & Weir 2002). E. Oficanthon mirabilis Paulian, 1985 (modified from Paulian 1985).

Fig. 11. Morphological diversity in the protibiae of Sylvicanthon Halffter & Martínez, 1977. A. S. enkerlini (Martínez et al., 1964) comb. nov. B. S. aequinoctialis (Harold, 1869) comb. nov. C. S. securus (Schmidt, 1920) comb. nov. D. S. obscurus (Schmidt, 1920). E. S. furvus (Schmidt, 1920). F. S. bridarollii (Martínez, 1949) (southern populations). G. S. bridarollii (northern populations). H. S. attenboroughi sp. nov. I. S. edmondsi sp. nov. J. S. candezei (Harold, 1869). Note the variation related to the presence and degree of development of an expansion on the internal edge of protibiae and to the number, size and position of the external teeth.

Fig. 7. Pterothorax, ventrolateral views. A. Sylvicanthon aequinoctialis (Harold, 1868) comb. nov. B. S. obscurus (Schmidt, 1920). C. Canthon xanthopus Blanchard, 1846. Abbreviations: MsC = mesocoxa; MMsC = external margin of mesocoxa; CMtE = centre of metaventrite; LMtE = lateral region of metaventrite; MtC = metacoxa.

Fig. 13. Morphological variation in the metafemora of Sylvicanthon Halffter & Martínez, 1977. A. S. foveiventris (Schmidt, 1920). Note the presence of coarse elongate punctures at the base of the metafemur (arrow) and the tegument without microsculpture. B. S. obscurus (Schmidt, 1920). Observe the tegument entirely covered by coarse non-elongate punctures. C. S. monnei sp. nov. D. S. mayri sp. nov. E. S. furvus (Schmidt, 1920). F. S. edmondsi sp. nov. G. S. bridarollii (Martínez, 1949) (southern populations). H. S. bridarollii (Martínez, 1949) (northern populations). Note the differences in colouration and in the presence of coarse punctation at the base of metafemur in the two ends of the clinal variation observed in S. bridarollii.

Fig. 12. Tegument variation in Sylvicanthon Halffter & Martínez, 1977. A. Tridimensional alveolar microsculpture of the elytra in S. proseni (Martínez, 1949) stat. et comb. nov.. B. Rivose microsculpture on the external side of metaventrite in S. obscurus (Schmidt, 1920). C. S. proseni stat. et comb. nov., humeral carina, feature present also in S. aequinoctialis (Harold, 1869) and in about two-fifths of the studied specimens of S. obscurus (Schmidt, 1920).

Fig. 15. Morphological variation and sexual dimorphism in the protibial spur of Sylvicanthon. A. S. enkerlini (Martínez et al., 1964) comb. nov., ♂. B. S. foveiventris (Schmidt, 1920), ♂. C. S. candezei (Harold, 1869), ♂. D. S. genieri sp. nov., ♂. E. S. aequinoctialis (Harold, 1868), ♂. F. S. aequinoctialis comb. nov., ♀. G. S. proseni (Martínez, 1949) stat. et comb. nov., ♂. H. S. proseni stat. et comb. nov., ♀. I. S. bridarollii (Martínez, 1949), ♂. J. S. seag sp. nov., ♂. K. S. edmondsi sp. nov.,♂. L. S. attenboroughi sp. nov., ♂. M. S. obscurus (Schmidt, 1920), ♂. N. S. furvus (Schmidt, 1920), ♂. O. S. securus (Schmidt, 1920) comb. nov., ♂.

Fig. 45. Canthon xanthopus Blanchard, 1846. A. Lectotype and its labels. B. An ordinary specimen.

Fig. 31. Metafemora of the two species of the aequinoctialis subgroup. A. Sylvicanthon aequinoctialis (Harold, 1868) comb. nov. B. S. proseni (Martínez, 1949) stat. et comb. nov. Arrow indicates the basal end of the posterior margin of the ventral face of metafemora (note that it is longer in S. aequinoctialis comb. nov. than in S. proseni stat. et comb. nov. and may exceed the trochanter in the first species).

Fig. 16. Abdominal lateral foveae of females (arrows). A. Sylvicanthon furvus (Schmidt, 1920). B. S. mayri sp. nov. C. S. obscurus (Schmidt, 1920). D. S. foveiventris (Schmidt, 1920).

Fig. 19. Aedeagus morphological diversity in Sylvicanthon. A. S. securus (Schmidt, 1920) comb. nov. B. S. obscurus (Schmidt, 1920). C. S. furvus (Schmidt, 1920). D. S. monnei sp. nov. E. S. mayri sp. nov.

Fig. 17. Aedeagus morphological diversity in Sylvicanthon. A. S. enkerlini (Martínez et al., 1964) comb. nov. B. S. foveiventris (Schmidt, 1920). C. S. candezei (Harold, 1869) (arrows points to the ventral notch). D. S. genieri sp. nov. (idem). E. S. aequinoctialis (Harold, 1868). F. S. proseni (Martínez, 1949) stat. et comb. nov.

Fig. 18. Aedeagus morphological diversity in the bridarollii subgroup.A. S. bridarollii (Martínez, 1949). B. S. seag sp. nov. (arrow points to the ventral notch). C. S. edmondsi sp. nov. (idem). D. S. attenboroughi sp. nov. (idem).

Fig. 44. Differences on the parameres of Sylvicanthon mayri sp. nov. and S. monnei sp. nov. (grayish zones represent original membraneous areas). A–B. S. mayri sp. nov. C–D. S. monnei sp. nov. Note that both branches of the apical bifurcation of the parameres of S. mayri sp. nov. are much more divergent than those of S. monnei sp. nov., which makes the internal angle between them more open in the first species (~110º) than in the second (~78º). In the same way, as the inferior branch is much more projected in S. mayri sp. nov. than in S. monnei sp. nov., the angle between the posterior region of that branch and the rest of the paramere is more open in S. monnei sp. nov. (~147º) than in S. mayri sp. nov. (~137º), species that seems to have a strong excavation at this point of the paramere (indicated by red arrow in A).

Fig. 20. Combined distribution of the species of Sylvicanthon in the Neotropical region. Note that the genus occupies the four great areas of tropical humid forest of the American continent: Central American forests, the Choco of the South American northwest, the Amazon Rainforest and the Atlantic Forest. Following the hypothesis of Martínez et al. (1964) that the Amazon Basin was the ancestral area of the Sylvicanthon of the candezei group, three lineages have independently dispersed from there (arrows): that of the ancestors of S. aequinoctialis (Harold, 1868), which, from the Choco, invaded Central America at the pace of the tropical forests’ gradual northwards advance, occurred during the Plio-Pleistocene, and those of the ancestors of respectively S. obscurus (Schmidt, 1920) and S. foveiventris (Schmidt, 1920), which reached the Atlantic Forest through the corridors of humid forests that bridged that biome to the Amazon forest during the wettest periods of the Neogene (see a more detailed discussion in the text).

Fig. 30. Distribution of the two species of the aequinoctialis subgroup.

Fig. 24. Distribution of the three species of the candezei subgroup.

Fig. 23. Sylvicanthon foveiventris (Schmidt, 1920). A. Dorsal view. B. Ventral view. C. S. foveiventris type material: a. Lectotype. b. Lectotype’s labels. c. Paralectotype 1’s label. d. Paralectotype 6’s labels.

Fig. 25. Sylvicanthon candezei (Harold, 1869). A. Dorsal view. B.Ventral view.

Fig. 27. Sylvicanthon genieri sp. nov. A. Dorsal view. B. Ventral view.