The status and future of scale insect ( C occoidea) systematics

Two basic properties of scale insect biology make scale insect groups ideal systems for studies of evolutionary biology and ecology: (i) they are essentially sedentary and (ii) they are fixed at the nexus of multiple interspecific interactions with host plants, natural enemies, microbial mutualists and insect mutualists (Gullan & Kosztarab, 1997). Nevertheless, scale insects are seldom used as model systems in evolution (see Ross et al., 2012, 2013 for exceptions). A major impediment to the more extensive use of scale insects in this context is an incomplete systematic framework. In comparison to some other insect herbivore groups, e.g. butterflies (Papilionoidea) and fig wasps (Agaonidae), the species diversity and phylogenetic history of scale insects are poorly understood. Here, I assess the status of scale insect systematics, note recent accomplishments, point out current challenges and look to the future. Scale insect systematists were incredibly forward-thinking about organizing scale insect information and making it accessible via an online database, ScaleNet (Ben-Dov et al., 2012). ScaleNet was up and running for more than a decade before the inception of the National Science Foundation's Advancing Digitization of Biological Collections (ADBC) program. ScaleNet remains unusual among biodiversity databases in that it is not a specimen database per se, but a complex model of the scale insect literature (c. 25 000 references). Although certain properties of species, such as geographic distributions and interspecific associations, are better modelled with specimen data, ScaleNet remains one of the most complete synopses of the ecology and biology of any group of insects. In the past decade, major progress has been made in understanding phylogenetic relationships among scale insect lineages. Scale insect families are treated routinely as belonging to one of two informal groups (Gullan & Cook, 2007): the archaeococcoids (Fig. 1A–D) and neococcoids (Fig. 1E–I). Monophyly of the neococcoids has been demonstrated repeatedly with DNA sequence data (Cook et al., 2002; Gullan & Cook, 2007; Yokogawa & Yahara, 2009). Paraphyly of the archaeococcoids with respect to the neococcoids has been recovered from mitochondrial DNA data (Yokogawa & Yahara, 2009) and macropterous male data (Hodgson & Hardy, 2013). Among neococcoids, the first DNA sequence-based phylogeny estimates were produced for the four largest scale insect families: Diaspididae (Morse & Normark, 2006; Andersen et al., 2010b), Pseudococcidae (Downie & Gullan, 2004; Hardy et al., 2008), Eriococcidae (Cook & Gullan, 2004) and Coccidae (T. Kondo and L.G. Cook, personal communication). Amongst non-neococcoid taxa, i.e. 'archaeococcoids', the monophyly of several families has been demonstrated with a DNA sequence-based phylogeny estimate (Gullan & Cook, 2007), and the first intrafamily phylogenies were estimated for Ortheziidae (Vea & Grimaldi, 2012) and part of Monophlebidae (Unruh & Gullan, 2008a). Studies of scale insect species diversity for the most part have been couched in revisionary studies, and a few recent revisions have integrated different sources of data [e.g. Gullan et al. (2010) and Kaydan & Gullan (2012) for Ferrisia mealybugs] as advocated by proponents of integrative taxonomy (Schlick-Steiner et al., 2010). Examples of high-quality scale insect revisionary projects include: Williams (1989), Kozár & Miller (2000), Hodgson & Miller (2010), Kondo (2010), Hardy et al. (2011), Schneider & LaPolla (2011), Hodgson & Peronti (2012) and Hodgson (2012), and many papers by S. Takagi (summarized by Kawai et al., 2009). Recent worldwide faunal treatments of major groups include: Iceryini (Unruh & Gullan, 2008b), Ortheziidae (Kozár, 2004) and Rhizoecinae (now Rhizoecidae) (Kozár & Konczné Benedicty, 2007). In addition, excellent regional faunas have been produced: for example, the scale insect faunas of far eastern Russia (Danzig, 1980), the tropical South Pacific (Williams & Watson, 1988a,1988b, 1990) and California (Gill, 1988, 1993, 1997); the pseudococcid faunas of Australia (Williams, 1985), Central and South America (Williams & Granara de Willink, 1992) and southern Asia (Williams, 2004); and the eriococcid (Hoy, 1962), coccid (Hodgson & Henderson, 2000, 2004) and diaspidid (Henderson, 2011) faunas of New Zealand. Scale insects are small and cryptic plant parasites disproportionately common among documented invasive insects (Miller et al., 2005). Many species are established agricultural pests and inevitably many more will become established as pests in future (Miller et al., 2005; Ouvrard et al., 2013). Scale insect diagnostics is serious business. Scale insect systematists were early adopters of alternatives to dichotomous identification keys (which have limited use for nonexperts and with imperfect specimens), and have produced several diagnostic tools using new technologies. Chief among these is ScaleNet. One thing that sets ScaleNet apart from many other taxonomic databases is the ability for users to execute ecological queries that can facilitate specimen identification. For example, users can retrieve a list of all the coccid species known to occur in Argentina, or on host plants in the family Rhamnaceae. Geospatial queries are now possible in the Global Biodiversity Information Facility's (GBIF) data portal via filtered occurrence searches, whereas interspecific association queries are not. Another valuable diagnostic resource is the set of matrix-based interactive keys that are available online. For example, the USDA Systematic Entomology Laboratory developed matrix keys to identify the scale insect families, as well as the pseudococcid and coccid species that have been intercepted by US quarantine. Even with access to geospatial data, interspecific association data and user-friendly matrix-based keys, scale insect diagnostics can be challenging. In order to visualize diagnostic features, delicate specimens of the appropriate life stage (adult female) must be cleared in a weak base, stained, slide mounted and placed under high magnification, preferably on a phase-contrast compound microscope. Thus, scale insects are a group for which DNA barcoding is particularly germane. An African Centre for DNA Barcoding project is currently underway (led by Michelle van der Bank, Theresa Sethusa and Ian Millar) that aims to build a reference library of large nuclear ribosomal subunit (28S) sequences from scale insect species of economic importance to South Africa. This may serve as a model for similar projects in other parts of the world. The ScaleNet datastore has continued to evolve and expand, but the ScaleNet application has not. Currently, ScaleNet is built on an extinct and closed-source software platform (FoxPro) and is administered exclusively by retirees. Furthermore, the current ScaleNet data model is inadequately normalized, and as a result suffers from insufficient data validation. ScaleNet's architecture makes it difficult to extend, i.e. to build new functions into the application. ScaleNet, in its present form is unlikely to persist for the next ten years. After more than a decade of DNA sequence-based phylogenetic research, we still lack an inclusive and robust phylogenetic hypothesis for evolutionary relationships among scale insect families. The morphological study of Hodgson & Hardy (2013) constitutes a major advance in this area but, for many applications, is no substitute for a comprehensive DNA-sequence based estimate. Furthermore, we have few densely-sampled species-level scale insect phylogenies. Notable exceptions are the Ferrisia study of Gullan et al. (2010), and the ongoing studies of Diaspididae by B. B. Normark and collaborators (e.g. Andersen et al., 2010a; Gwiazdowski et al., 2011) and of the species diversity of Australian Myrtaceae-feeding felt scale insects by L. G. Cook and collaborators. More intensive species sampling of nonmorphological data is needed to resolve the demonstrated or probable nonmonophyly of some large genera, such as Pseudococcus Westwood and Dysmicoccus Ferris (Hardy et al., 2008), Eriococcus Targioni Tozzetti (Cook & Gullan, 2004) and Coccus Linnaeus (Lin et al., 2013). For studies of evolutionary biology and ecology, increasing the species density of scale insect phylogenies will be just as important as increasing their breadth. A major obstacle to estimating better scale insect phylogenies has been the high degree of sequence divergence among scale insect lineages, and the limited molecular toolkit for scale insect phylogenetics (Hardy, 2007). These challenges may be surmounted by new phylogenomic approaches, and may well be met with a commitment to traditional PCR + Sanger sequencing. The primary reason that we don't have more comprehensive scale insect phylogenies is that we have not tried to estimate them. The biggest challenges in scale insect alpha-level revisionary systematics are the same as those that face revisionary systematics of most insect groups: mass extinction due to habitat loss, invasive species, pollution and climate change (Thomas et al., 2004), and a system of incentives that discourages taxonomic research (Agnarsson & Kuntner, 2007). In particular, the scale insect faunas of parts of the southern hemisphere and palaeotropics are in need of study, and it is unclear who will do it. The number of scale insect systematists has been declining steadily as full-time professionals retire and are not replaced, despite efforts to train a new generation of scale insect taxonomists, such as the NSF Partnership for Enhancing Expertise in Taxonomy (PEET) project on Coccoidea. Although the US-funded PEET scheme was successful in training young taxonomists of many poorly known groups and in generating publications, it did not lead to employment in taxonomy for most trainees (Agnarsson & Kuntner, 2007). Most taxonomists with the skills to collect and identify scale insect material from under-studied areas are or soon will be retired, and those with the skills to incorporate that material into a modern systematic framework are un- or underemployed. Nearly 8000 species of scale insects have been described (Ben-Dov et al., 2012). As a matter of feasibility, most scale insect diagnostic tools have had a regional focus. Many regions lack any diagnostic resources beyond synoptic insect collections and disparate taxonomic publications. For example, there is no published key to the diaspidid genera known to occur in Australia. Many regions also have borders permeable to the incursion of new scale insect species (Miller et al., 2005) and, in that respect, regional expertise and regionalized diagnostics are inadequate. A more general problem is that the new diagnostic platforms—DNA barcode libraries and matrix-based keys—have certain drawbacks. Although DNA barcoding sidesteps many of the challenges of scale insect identification, DNA barcode reference libraries are never exhaustive, and assignment of identity involves arbitrary similarity thresholds that can lead to identification errors, especially among recently-diverged species (Meyer & Paulay, 2005). Moreover, DNA barcoding still relies on experts to associate taxonomic concepts with sequence data and in reference library construction and improvement. As noted above, this expertise is evaporating. Matrix-based keys are easier to use for nonexperts, and provide all users with greater flexibility than dichotomous keys (Walter & Winterton, 2007). On the other hand, they do a poor job of expressing the uncertainty associated with an identification (just how unlikely is the most likely identification?) and for a group such as scale insects, could better incorporate geospatial and ecological data. A big part of why scale insect workers have been able to assemble such an exhaustive catalogue of scale insect ecology (ScaleNet), is because those data have been relatively easy to come by. The ecological modelling of scale insect species is poised to become even better, with the Tri-Trophic Database project currently funded by the NSF ADBC program. That project seeks to database nearly all of the Hemiptera specimens present in U.S. insects collections along with specimens of their major host plants and natural enemies. However, the experience of molecular systematists has been that, relative to many other insect groups, scale insect DNA sequence data are difficult to acquire. As a result, scale insect phylogenetic data are far less complete than scale insect ecological data. Going forward, next generation sequencing approaches, such as transcriptomics and target-enriched genomic sampling, promise to obviate that difficulty. Just as important will be for scale insect systematists to agree on current priorities for the field and to cooperate in addressing them. With better phylogenies, scale insect classifications will be aligned with their history of divergent evolution. More importantly, better estimates of phylogeny will pave the way for evolutionary biologists and ecologists to begin addressing their research questions with scale insect data. In other words, scale insects systematics will do what it is supposed to: function as a basic representation of biodiversity that enables other types of general and applied science. With specimen-based (not literature-based) characterization of scale insect geographic distributions and interspecific associations, models of scale insect species ecologies will become more accurate and interoperable. It seems almost inevitable that research in scale insect systematics will form the basis for a greater understanding of ecology and evolution. Progress in scale insect systematics is in danger of being stifled by inadequate manpower. Nevertheless, fewer may be able to do as much (or at least enough), by continuing the tradition of consolidating, codifying and sharing scale insect knowledge through media such as DNA barcoding, remote microscopy, specimen databases, online diagnostic tools, and online repositories of taxonomic literature and images. Thanks to P. J. Gullan and D. J. Williams for helpful comments on a draft of the manuscript. Some of the ideas presented here came from discussion with B. B. Normark. I am grateful for the postdoctoral support I receive through NSF grant DEB1216309 awarded to G. S. Svenson at the Cleveland Museum of Natural History.