Hypogeococcus pungens, a mealybug native of southern South America, is devastating native cacti in Puerto Rico and threatening cactus diversity in the Caribbean, and potentially in Central and North America. The taxonomic status of H. pungens is controversial since it has been reported feeding not only on Cactaceae but also on other plant families throughout its distribution range. However, in Australia, where the species had been exported from Argentina to control weedy American cacti, it was never found on host plants other than Cactaceae. These conflicting pieces of evidence not only cast doubt on the species identity that invaded Puerto Rico, but also have a negative impact on the search for natural enemies to be used in biological control programs against this pest. Here we present reproductive incompatibility and phylogenetic evidences that give support to the hypothesis that H. pungens is a species complex in which divergence appears to be driven by the host plants. The nuclear EF1α and 18S and the mitochondrial COI genes were used as markers to evaluate the phylogenetic relationships among H. pungens populations collected in Argentina, Australia and Puerto Rico feeding on Cactaceae and/or Amaranthaceae. Additionally, we conducted reciprocal crosses between mealybugs from both hosts. Species delimitation analysis revealed two well-supported putative species within H. pungens, one including mealybugs feeding on Amaranthaceae (H. pungens sensu stricto), and a new undescribed species using Cactaceae as hosts. Additionally, we found asymmetric reproductive incompatibility between these putative species suggesting recent reproductive isolation. The Bayesian species delimitation also suggested that the Australian mealybug population may derive from another undescribed species. Overall, the patterns of genetic differentiation may be interpreted as the result of recent speciation events prompted by host plant shifts. Finally, the finding of a single haplotype in the Puerto Rico population suggests only one invasive event. We still need to identify the geographical origin of the pest in order to enable the use of biological control to reduce the threat to cacti diversity in the Caribbean.