1. INTRODUCTION
Herbivorous insects are often involved in close interactions with their hosts since they are a food source and provide mating and oviposition sites (Schoonhoven, Van Loon, & Dicke, 2005). Such intimacy often entails the evolution of adaptations that allow insects to cope with specific features of their host plants. Hence, host plant shifts may affect the evolution of features associated with feeding location, oviposition and development on the host (Schoonhoven, Van Loon, & Dicke, 2005; Orsucci, et al., 2018). Adaptation to new hosts may cause, either as a direct consequence or as a byproduct, the evolution of sexual isolation, highlighting the role of host plant shifts in speciation (Funk et al., 2006; Nosil, 2012).
Phytophagous insects with presumptively wide host ranges, i.e. generalists, pose a concern from an applied view when dealing with insect pests. Many authors have suggested that polyphagous species are formed by locally adapted specialized populations or cryptic specialist species (e.g., Loxdale & Harvey 2016; Forbes et al., 2017). A species complex made up of cryptic specialists pose an additional complication in defining host range. The distinction between true polyphagous and cryptic specialist species is crucial for the design of biological control programs. In addition, cryptic species complexes offer the opportunity to investigate the role of host plants in the diversification of herbivorous insects since such complexes often include species of recent origin (Winter, Friedman, Astrin, Gottsberger, & Letsch, 2017; Malka et al., 2018; Bakovic et al., 2019).
Distinguishing between intra and interspecific genetic variation is particularly relevant in cases of suspected cryptic species, especially when members of a guild of cryptic species are involved in invasions of new areas. Invasive insects may cause global problems upon spread to new territories, not only to agriculture, but also to biological diversity since invasive species are among the main causes of biodiversity loss (Newbold et al., 2015). The proper identification of invasive species is a necessary task to design successful biological control strategies to prevent or reduce harmful effects of invaders.
With the advent of molecular data, an increasing number of studies showed that some putative polyphagous insects are, actually, species complexes embracing several deeply diverged species, each one specialized to different host plant use (Egan, Nosil, & Funk, 2008; Nosil et al., 2012; Powell, Forbes, Hood, & Feder, 2014; but see Vidal et al., 2019). So far most population surveys have been based on the evaluation of a single genetic marker, the so-called DNA‐barcoding gene encoding cytochrome oxidase subunit I (mtDNA) (Dinsdale, Cook, Riginos, Buckley, & De Barro, 2010; Stouthamer et al., 2017). However, it is well known that in many cases this marker provides information limited to the mitochondrial lineage, potentially resulting in misidentification of species boundaries due either to incomplete lineage sorting or introgressive hybridization (Eyer, Seltzer, Reiner-Brodetzki, & Hefetz, 2017; Després, 2019). Recently, population genetic studies benefited from availability of new methodologies based on high-throughput sequencing of genomic libraries containing a reduced-representation of nuclear genomes, known as genotype by sequencing, such as RADseq (Baird et al., 2008; Andrews, Good, Miller, Luikart, & Hohenlohe, 2016). These methodologies provide large multi-locus datasets that can be used to evaluate cryptic diversity in species complexes (e.g., in Elfekih et al., 2018), and to trace the origin of invading pests (Anderson, Tay, McGaughran, Gordon, & Walsh, 2016; Ryan et al., 2019).
The cactus mealybug pest invading Puerto Rico and the adjacent smaller islands, represents a threat for cactus diversity in the Caribbean, Central and North America. The pest was initially reported asHypogeococcus pungens Granara de Willink (Hemiptera: Pseudococcidae), under the common name Harrisia cactus mealybug (HCM), the successful biological control agent released against an invasive cactus in Australia and South Africa (McFadyen & Tomley 1978, 1981; Paterson, Hoffmann, Klein, Mathenge, Neser, & Zimmermann, 2011). Currently, H. pungens sensu lato is considered a species complex, and in its native range (South America) it was reported to feed on species of Amaranthaceae, Portulacaceae and Cactaceae (Ben-Dov, 1994; Claps & Haro, 2001; Zimmermann, Pérez Sandi Cuen, Mandujano, & Golubov, 2010). Recent studies based on molecular data and assessing reproductive compatibility (Poveda-Martínez et al., 2019) and performance on different hosts plants (Aguirre et al., 2016) suggests that populations collected in Argentina, initially identified asH. pungens sensu lato , were, actually, a species complex comprising at least two species; one mealybug species feeding on Amaranthaceae (H. pungens sensu stricto ), and the other an undescribed species feeding on cactus. Interestingly, the Puerto Rican cactus pest appears closer to H. pungens sensu stricto in phylogenetic trees suggesting that the pest shares a recent ancestor with the latter rather than with the sympatric, cactus feeding new species from Argentina. However, none of the mitochondrial haplotypes found in Argentina matched the single haplotype detected in the Puerto Rican mealybugs feeding on Cactaceae, suggesting that the source population of the pest was not Argentina (Poveda-Martínez et al., 2019).
In the present study, we extended the sampling effort to a large geographic area comprising both the native (Argentina, Paraguay and Brazil) and non-native (Puerto Rico and southern United States) ranges by collecting mealybugs on host plants recognized as part of the diet ofH. pungens sensu lato . We used a combination of genome-wide SNPs and mtDNA variation to investigate: i) genetic diversity within H. pungens species complex, ii) host plant ranges of each one of the putative members of the complex; iii) whether host plant shifts drove the diversification in the species complex, and iv) the source population of the Puerto Rican cactus pest. Based on these results, we will be able to search for and select biological control strategies using natural enemies, either those that co-evolved with the pest (classical biological control) or those that did not co-evolve but attack closely related species to the pest (new association biological control).