THAUMASTELLIDAE Seidenstücker Historical

Grazia, Jocelia, Schuh, Randall T. & Wheeler, Ward C., 2008, Phylogenetic relationships of family groups in Pentatomoidea based on morphology and DNA sequences (Insecta: Heteroptera), Cladistics 24, pp. 932-976 : 966-967

THAUMASTELLIDAE Seidenstücker Historical : The superfamilial position of this family group has been controversial since the time of its description. Having been originally placed in the Lyg-

aeoidea, the Thaumastellidae was transferred to the Pentatomoidea at family rank by Štys (1964a) and later reduced to a subfamily of the Cydnidae by Dolling (1981) and Jacobs (1989). Jacobs (1989) mentioned the presence of a microchromosome (m-chromosome) in the Thaumastellidae , which is also present in many Coreoidea, Lygaeoidea, and Pyrrhocoroidea ( Largidae ). Henry

(1997) suggested that the m-chromosome is a synapomorphy defining this broad group within the Pentatomomorpha , albeit lost in a few taxa ( Berytidae , Lygaeinae , and Piesmatidae ), and that the Thaumastellidae , due to

the presence of an m-chromosome, may therefore not belong in the Pentatomoidea . Nonetheless, many apomorphies of the Pentatomoidea are found in the Thaumastellidae , including the presence of the foretibial apparatus, expanded mandibular plates, the base of the head not forming a ‘‘neck’’, post-ocular tubercles absent, the structure of the female genitalia, and the paired lateral trichobothria.

Analytical result: Because of the controversy over placement of the Thaumastellidae , and their rarity in collections, we felt fortunate to be able to secure sequences of two species of Thaumastella Horváth, offering new evidence to test the relationships of the taxon. In our morphological analyses ( Figs 42–44 ) the Thaumastellidae is placed within the Cydnidae , in conformity with the theory proposed by Dolling (1981), an unsurprising result. The strict consensus of molecular data with a 1: 1 cost ratio places the Thaumastellidae as the sister group of the Pentatomidae ( Fig. 45 ); changing the cost ratio to 2: 2 places Thaumastella in a clade that includes the Urostylididae , Dinidoridae , Tessaratomidae , Canopus , and Phloea ( Fig. 46 ). These results are more or less duplicated in the 52-taxon combined analyses ( Figs 51 and 52 ), although there Thaumastella becomes the sister group of the Acanthosomatidae + Lestoniidae under a 1: 1 cost ratio ( Fig. 51 ), rather than of the Pentatomidae . The 92-taxon combined analyses always place the Thaumastellidae in a clade containing most of the other ‘‘cydnoid’’ taxa ( Figs 53–55 ). Although it seems abundantly clear that Thaumastella belongs within the Pentatomoidea , as opposed to the Lygaeoidea, because it never falls outside the ingroup in any of our analyses, we continue to maintain it at the family level, in the absence of a more clear-cut signal regarding the details of its phylogenetic placement within the group, and more particularly a more rigorous test of the monophyly of the Cydnidae sensu Dolling. Morphological features treated as diagnostic for the Thaumastellidae in our analyses are the short scutellum that at most barely surpasses the posterior margin of the metathorax (160) and the claval commissure reduced, being no more than half the length of the scutellum (171).

Figures

Fig. 42. Strict consensus of 96 most parsimonious trees for full-taxon morphological data set, with unsupported nodes supressed. Length = 207; consistency index = 42; retention index = 86. (d) Non-homoplasious; (s) homoplasious.

Fig. 43. Strict consensus of three trees derived from successive weighting of the results shown in Fig. 42. (d) Non-homoplasious; (s) homoplasious.

Fig. 44. Strict consensus of 12 trees derived from implied weighting analysis of morphological data using PIWE. (d) Non-homoplasious; (s) homoplasious.

Figs 45–48. 45. One of two trees derived from analysis of combined molecular data with 1: 1 indel ⁄transition–transversion cost ratio. 46. Single tree derived from analysis of combined molecular data with 2: 2 indel ⁄ transition–transversion cost ratio. 47. Single tree derived from analysis of ~500 bp of 16S rRNA data using 1: 1 indel ⁄ transition–transversion cost ratio. 48. Single tree derived from analysis of ~1800 bp of 18S rRNA using 1: 1 indel ⁄ transition–transversion cost ratio.

Figs 49–52. 49. Single tree derived from analysis of ~470 bp of 28S rRNA using 1: 1 indel ⁄ transition–transversion cost ratio. 50. Single tree derived from analysis of ~1100 bp of COI mtDNA using 1: 1 indel ⁄transition–transversion cost ratio. 51. Total evidence analysis with POY of 52- taxon data set using 1: 1 indel ⁄ transition–transversion cost ratio. 52. Total evidence analysis with POY of 52-taxon data set using 2: 2 indel ⁄ transition–transversion cost ratio.

Fig. 53. One of six trees from total evidence analysis with POY of 92-taxon data set using 1: 1 indel ⁄ transition–transversion cost ratio. (d) Non-homoplasious; (s) homoplasious.

Fig. 54. One of four trees from total evidence analysis with POY of 92-taxon data set using 1: 2 indel ⁄ transition–transversion cost ratio. (d) Non-homoplasious; (s) homoplasious.

Fig. 55. One of three trees from total evidence analysis with POY of 92-taxon data set using 2: 2 indel ⁄transition–transversion cost ratio, which had the lowest MRI value. Bremer support values are shown.