BackgroundMaternal care (egg-nymph guarding behavior) has been recorded in some genera of Acanthosomatidae. However, the origin of the maternal care in the family has remained unclear due to the lack of phylogenetic hypotheses. Another reproductive mode is found in non-caring species whose females smear their eggs before leaving them. They possess pairs of complex organs on the abdominal venter called Pendergrast’s organ (PO) and spread the secretion of this organ onto each egg with their hind legs, which is supposed to provide a protective function against enemies. Some authors claim that the absence of PO may be associated with the presence of maternal care. No study, however, has tested this hypothesis of a correlated evolution between the two traits.ResultsWe reconstructed the molecular phylogeny of the subfamily Acanthosomatinae using five genetic markers sequenced from 44 species and one subspecies with and without maternal care. Eight additional species from the other two acanthosomatid subfamilies were included as outgroups. Our results indicated that maternal care has evolved independently at least three times within Acanthosomatinae and once in the outgroup species. Statistical tests for correlated evolution showed that the presence of maternal care is significantly correlated with the secondary loss or reduction of PO. Ancestral state reconstruction for the node of Acanthosoma denticaudum (a non-caring species in which egg smearing with developed POs occurs) and A. firmatum (a caring species with reduced POs) suggested egg smearing was still present in their most recent common ancestor and that maternal care in A. firmatum has evolved relatively recently.ConclusionsWe showed that maternal care is an apomorphic trait that has arisen multiple times from the presence of PO within the subfamily Acanthosomatinae. The acquisition of maternal care is correlated with the reduction or loss of PO, which suggests an evolutionary trade-off between the two traits resulting from physiological costs. This prediction also implies that presence of maternal care can be highly expected for those groups lacking behavioral data, which invariably also lack the organ. No secondary loss of maternal care was detected in the present tree. We suggest that the loss of maternal care may be suppressed due to the vulnerability of the PO-free condition, which thus maintains maternal care.Electronic supplementary materialThe online version of this article (doi:10.1186/s12862-015-0537-4) contains supplementary material, which is available to authorized users.