A taxonomic study of the Brazilian turtle ants (Formicidae: Myrmicinae: Cephalotes) Oliveira, Aline Machado Powell, Scott Feitosa, Rodrigo Machado Revista Brasileira de Entomologia 2021 2021-09-13 65 3 e 20210028 1 52 5XK2K Linnaeus 1758 [83,199,1806,1825] Insecta Formicidae Cephalotes Animalia Hymenoptera 17 18 Arthropoda species atratus  ( Figs. 17, 18)  The atratusgroup was proposed for the first time by De Andrade and Baroni Urbani (1999). Before that, the species C. alfaroi, C. atratusand C. serraticepsbelonged tothe genus Cephalotes, while the species C.oculatus, C.opacusandC.placidus belongedto thegenusEucryptocerus. The species C. alfaroiwas not included in this study since it does not occur in Brazil. In the morphological phylogeny by De Andrade and Baroni Urbani (1999, see their Fig. 24), the Central American hamulus group is the sister group of all other groups in Cephalotes, followed by theatratusgroup. In the recent molecular phylogenies (Price et al., 2014, 2016, see their Fig. S3) this relationship isinverse, and theatratusgroup is the sister-mostgroup in the topology ( Fig. 54). Theatratusand hamulus groups share the absenceof the soldier caste in some species of atratusand all species of hamulus. The species C. alfaroi, C. opacus, and C. serraticepshave known soldiers, while the species C. atratus, C. oculatus, and C. placidusdo not have soldiers, as far as we know. Of these, large colony series ofC. atratushave revealed that while this species certainly lacks morphologically differentiated soldiers and any pronounced allometric morphological scaling, the worker caste is highly variable insize within mature colonies (Corn, 1980). This species is broadly distributed, and frequently collected ( Fig. 18), and is the second most common species of the genus in Brazil, after C. pusillus. The other species without soldiers are not highly variable in size, and their distribution is more restricted (Amazon/Atlantic Forest ( Fig.18)).   Diagnosis:In workers and soldiers vertexal corners of head, in lateral view, with a pair of spines ( Fig. 2a). Pronotum always with a pair of long dorsolateral spines; a pair of short median spines can be present in some species, sometimes weakly developed. Postoccipital carinae with ventral expansions ( Fig. 17a-b).    Brazilian species of atratusgroup    Cephalotes atratus (Linnaeus, 1758) Obsolete combinations:Cryptocerus atratus, Formicaatrata = Cephalotes atratus crassispina Santschi, 1920= Cephalotes atratuserectus Kempf, 1951 = Cephalotes atratus nitidiventris Santschi, 1920= Cephalotes atratusquadridens (De Geer, 1773) Obsolete combinations: Cephalotesquadridens, Formicaquadridens = Cephalotes atratusrufiventris (Emery, 1894) Obsolete combination: Cryptocerus atratusrufiventris = Cephalotesdubitatus (Smith, 1853) Obsolete combination: Cryptocerus dubitatus = Cephalotes marginatus (Fabricius, 1804)  new synonymObsolete combination: Cryptocerus marginatus   Cephalotes oculatus (Spinola, 1851)Obsoletecombination: Cryptocerus oculatus= Cephalotesaethiops (Smith, 1853) Obsolete combination: Cryptocerus aethiops   Cephalotes opacus Santschi, 1920= Cephalotes abdominalis Santschi, 1929   Cephalotes placidus (Smith, 1860)Obsoletecombination: Cryptocerus placidus= Cephalotesfenestralis (Smith, 1876) Obsoletecombination: Cryptocerus fenestralis.   Cephalotes serraticeps (Smith, 1858)Obsoletecombination: Cryptocerus serraticeps  Note:The synapomorphiesso far accepted for C.marginatusinclude abundant suberect pilosity onthefirst gastral sternite, instead of sparse as in C. atratus, and median pronotal spines at least with ¼ of the size of the dorsolateral ones, instead of minute or absent as in C. atratus (De Andradeand Baroni Urbani, 1999).  Cephalotes marginatuswas described by Fabricius (1804), synonymized under C. atratusby Klug (1824), and then revived by De Andrade and Baroni Urbani (1999). In the latter study, the species C. decemspinosus Santschi, 1920was synonymizedunder C.marginatus. Kempf (1951) has already argued that the status of this species should be changed, as there were individuals in series of C. atratuswith the same characteristics of C. decemspinosus. After a careful examination of both species based on specimens from a wide range of distribution, we found great variation of the putative synapomorphic characters of C. marginatusamong samples of C. atratus. Cephalotes marginatusis known only for the Amazonian arch in South America, while C. atratusis widely distributed from Mexicoto north of Argentinaand isone of the most commonly sampled species of Cephalotes. Therefore, consideringmorphological and geographic evidence, we here propose the synonymy of C. marginatusunder C. atratus.   Key to the identification of Brazilian species of the atratusgroup of Cephalotesbased on workers and soldiers  ( Figs. 17a-g)     1 In lateral view, eyes positioned ventrally to the antennal scrobes ( Fig. 17a)...................................................................................................................2  1’ In lateral view, eyes positioned posteriorly the antennal scrobes ( Fig. 17b)..................................................................................................................3   2 Body shiny. In lateral view, postpetiolar dorsal spines shorter than the subpostpetiolar process ( Fig. 17d).............................................  C.atratus  2’ Body opaque. In lateral view, postpetiolar dorsal spines longer than the subpostpetiolar process ( Fig. 17e)…............................….  C. serraticeps   3 In dorso-oblique view, dorsal and lateral faces of mesonotum and propodeum meeting in a carina, not necessarily extending to the propodeal spines ( Fig. 17f)...............................................................  C. oculatus  3’ In dorso-oblique view, dorsal and lateral faces of mesonotum and propodeum continuous, without carina ( Fig. 17g)....................................4   4 Propodeal spines shorter than the declivous face of propodeum ( Fig. 17c)....................................................................................................  C.opacus  4’ Propodeal spines longer than the declivous face of propodeum ( Fig. 17d, e)............................................................................................  C. placidus