Scale Insects, edition 2, a tool for the identification of potential pest scales at U.S.A. ports-of-entry (Hemiptera, Sternorrhyncha, Coccoidea)

Abstract We provide a general overview of features and technical specifications of an online, interactive tool for the identification of scale insects of concern to the U.S.A. ports-of-entry. Full lists of terminal taxa included in the keys (of which there are four), a list of features used in them, and a discussion of the structure of the tool are provided. We also briefly discuss the advantages of interactive keys for the identification of potential scale insect pests. The interactive key is freely accessible on http://idtools.org/id/scales/index.php


Introduction
Scale insects include notorious invasive species. They are small, often live in concealed portions of their hosts, and are transported in commodities that are common in international trade . Scales cause billions of dollars in damage to U.S. crops and in control costs annually (Kosztarab 1977(Kosztarab , 1990Kondo et al. 2008) and nearly all damaging scale pests are species that were inadvertently introduced Miller and Miller 2003).
Only a few specialists in the world can identify scale insects based on morphology. Of these, many are retired or approaching retirement, putting scale insect systematics in danger of being stifled by inadequate manpower (Kondo et al. 2008;Hardy 2013). Moreover, because scale insect pests may arrive from all over the world , regional expertise and scale insect diagnostic tools with a regional focus are inadequate for border patrol (Hardy 2013). Considering the common presence of scales in interceptions (Evans and Dooley 2013), we have produced a user-friendly, online tool that can be used at U.S.A. ports-of-entry to help with the identification of potentially invasive species.
Although many dichotomous keys have been developed for scales (Hardy 2013), they are regional in scope and do not have the flexibility of interactive keys. For instance, in traditional dichotomous keys, features appear in a fixed order, whereas in Lucid-based and other interactive keys, there are many paths to a correct identification. This is advantageous because some features are difficult to interpret, difficult to see on the specimen in hand, or are absent in a damaged specimen. Another benefit of interactive keys over dichotomous keys is that the user can express uncertainty by selecting more than one state per character, and view/modify the character states already selected (Walter and Winterton 2007;Cerretti et al. 2012;Dallwitz et al. 2013).
An earlier version of Scale Insects comprising 148 species was published on CDROM and uploaded to the USDA web site in 2007 (Miller et al. 2007). It was mentioned as useful by Hardy (2013) and has been part of training programs for state and federal identifier workshops, being often used by U.S. port personnel for making identifications (Greggory Evans, APHIS identifier, personal communication, 2014). The new edition includes 46 additional species, and updated information on the distribution of previously included species. Furthermore, the key to families was updated with new findings in the higher classification of scale insects (e.g., Hodgson 2012). The new tool also differs from the old one in how the glossary (pages containing pestspecific information) and the fact sheets are generated: in the old tool, fact sheets and the glossary are static html pages, whereas in the new tool they are dynamically generated and thus can be more easily updated.
As far as we know, one other Lucid key, with a regional scope, is available online to identify scales (Hardy and Kondo 2009). The general structure of Scale Insects, edition 2 emulates a number of Lucid-based tools for pest identification created and/or funded by the USDA Animal and Plant Health Inspection Service (APHIS), Identification Technology Program (ITP), and which can be found at http://idtools.org/id/. Our tool differs from all of them by being extensively linked to a comprehensive database on scale insects, ScaleNet (Ben-Dov et al. 2012). ScaleNet is unusual in that it is not only a specimen database per se, but a complex model of the scale insect literature (c. 25,000 references) (Hardy 2013). The links to ScaleNet give the user instant access to updated taxonomy, distribution and host records, and literature citations.
Despite the usefulness of Scale Insects, edition 2, it must be emphasized that it contains only a small subset of the world scale fauna. According to ScaleNet there are 7,792 species of scale insects worldwide. Therefore, a species that is not represented in the tool may key out as one that is (a false positive identification). For this reason, identifications should always be checked against information on the fact sheets and compared against authoritatively identified reference specimens for further assessment.
The following groups are represented in Scale Insects, edition 2 : Scale Families, encompassing the families most often encountered at the U.S.A. ports-of-entry; Soft Scales; Mealybugs and Mealybug-like families; and Other Scales, encompassing pests in various families with the exclusion of the Diaspididae. The tool makes extensive use of hypertext to link to images, glossaries, and other support material, and has four main components: Lucid-based interactive keys, fact sheets, glossary of morphological terms, and image gallery.

Format of the paper
This paper was written following the outlines for data papers provided by Penev et al. (2009Penev et al. ( , 2012 and the format was inspired on Cerretti et al. (2012) and Chrétiennot-Dinet et al. (2014).

Project description
There are four keys in Scale Insects, edition 2: Scale families, Mealybugs and Mealybug-like families, Soft Scales, and Other Scales (details for each key provided below). Terminal taxa in these keys are potential pests to U.S.A. agriculture, from all over the world, and were selected from a list of interceptions by APHIS-Plant Protection and Quarantine (PPQ) between 1995 and 2013. This list can be obtained from APHIS upon request, but has not been published. Below we describe the taxonomic coverage of each key, and provide lists of the features used to discriminate taxa in them. Features in the keys are from the adult female, are standard in scale insect taxonomy (Kondo et al. 2008) and, for the most part, can only be seen on a slide-mounted specimen.

General features
The key matrix is based on 66 morphological features of the adult female and includes 33 extant families as terminal taxa.

General features
The key matrix is based on 41 morphological features and includes 48 species in 21 genera of Coccidae.

Features used in the key
The morphological features to separate among soft scale pests are listed in Table 2. They were largely extracted from the comprehensive works of De Lotto (1965), Gill (1988), Gill et al. (1977), Gimpel et al. (1974), Hamon and Williams (1984), Hodgson (1994), and Williams and Watson (1990), from isolated descriptions, and analysis of specimens.

General features
The key matrix is based on 44 morphological features and covers 99 species in 27 genera and three families (96 species in 24 genera in Pseudococcidae, two species in one genus of Putoidae, and one species and one genus of Rhizoecidae).

Features used in the key
When the user first opens the Mealybugs and Mealybug-like Families key, he/she will be prompted to choose among the following families in the features window (in order to use this key, the user must know the family classification of the specimen): Pseudococcidae, Rhizoecidae or Putoidae. After the appropriate family has been selected, the features that differentiate among the species included in that family will open. While most features separate among species of Pseudococcidae, one character (presence/absence of tubular ducts in cerarii) is used to separate between the two species of Putoidae. This character was only coded for putoids. Since Rhizoecidae is represented by only one species (above), choosing this family will automatically result in species identification. Morphological features used to separate among species of Pseudococcidae are listed in Table 3 and were largely taken from the comprehensive works of Ferris (1950), McKenzie (1967), Miller and McKenzie (1973), Miller (1975), Cox (1987Cox ( , 1989, Williams and Watson (1988), Kosztarab (1996), Williams and Miller (2002), Miller and Giliomee (2011), Kaydan and Gullan (2012), Williams (2004), from isolated descriptions, and the analysis of specimens. Features to separate Putoidae species were taken from Williams and Granara de Willink (1992).

General features
The key matrix is based on 41 morphological features and covers 47 species (entities) in 26 genera and 11 families.

Technical specifications
Web location: http://idtools.org/id/scales/index.php Platform: a website Web Server: CentOS Programming language: PHP 5 and MySQL Application version: 2.0 Data base: MySQL Data: 2.0 Language: English License for use of the key: Attribution-Non-commercial Use of the primary data: available upon request.
The keys were directly built in Lucid builder (various versions up to 2007), then updated on Builder 3.5 (available at http://lucidcentral.org, Queensland, Australia) for the current version. A list of the Lucid3 key files (key data files and key program files) can be found in Penev et al. (2009).
Keys were deployed online using the Lucid Key Server (available at http://lucidcentral.org). The On-line Player is an alternative to the Lucid3 Application Player for interactive keys created using the Lucid3 Builder. Playing keys using the Lucid On-line Player does not require Java installation. This method may be advantageous to government workers because Java applications are often partially or totally blocked in government computers due to security concerns. Furthermore, some keys deployed using the Lucid3 Player may temporarily stop working or work defectively after a Java update, or may not work properly until an update has been implemented. However, it must be emphasized that the not all features of the Lucid3 Player are available in the On-line Player. For instance, the function to "prune redundant features" has been im-plemented in the former but not in the latter. Moreover, while the help icon (?) of the Lucid 3 Player is linked to a help file online, the help icon of the Lucid Online Player is linked to the Lucid webpage on (http://www.lucidcentral.org/en-us/software/lucid3. aspx); from there the user has to find the help file that has been written for the Lucid3 player. Despite those caveats, we believe that the On-line Player is a work in progress, and that in the long run the advantages of using it will outweigh the disadvantages of the Java player. A thorough discussion of the differences between the two programs for playing Lucid keys is warranted.
Information in fact sheets and glossary is managed using the Fact Sheet Manager (FSM), an interface for a database that stores all the data present in the dynamic pages of the website (fact sheets and glossary of terms). Both FSM and the database have been created and are maintained by the Identification Technology Program (ITP), USDA APHIS PPQ. When the user chooses an entity (by selecting a link of a particular species), a single fact sheet page that is coded with queries to the database populates the appropriate fact sheet content. This contrasts with the static HTML fact sheets in the previous version of Scale Insects (Miller et al. 2007), and allows for quick updates and corrections that go online almost immediately.

Tool details
In the portal Scale Insects, edition 2, the user has the option to find a fact sheet, use one of the keys, view the image gallery or consult the glossary of morphological terms (Fig. 1). Searching for a family in the "search fact sheets" box will list a fact sheet for that family and for the corresponding species in one of the keys, if present. For instance, a search for Aclerdidae will find three fact sheets, one for Aclerdidae in the Families key and two for species of Aclerda Signoret in the Other Scales key. Fact sheets in keys to species contain the following information: a link to a Catalog on ScaleNet, Common name, Field characters (diagnostic description), Validation characters (diagnostic description of the adult female), Comparison (with similar species), U.S.A. quarantine notes, and Important References (a link to ScaleNet). In fact sheets, most morphological terms are hyperlinked to the glossary, and two or three figures are present, which can be enlarged by clicking on them: a line drawing with structures labeled, a habitus picture and a picture of the slide mounted, whole body of a mature female. In the enlargement of the whole body, diagnostic features are marked with squares. When the user scrolls the mouse over a square, an enlargement of the corresponding feature will pop up in a window (Fig. 2).  The image gallery offers a quick way to perform identifications by matching the specimen in hand with a drawing or a photograph. Images can be filtered in two ways: by including or excluding images of the habitus, whole body picture or drawing; and by selecting a package (each package corresponds to a Lucid key and associated fact sheets and images).
From the Lucid On-line Player the Lucid key looks very similar to its Lucid3 Player counterpart (Fig. 3). The terminal taxa are represented as Entities on the right windows, whereas the features are represented as Features on the left windows. When a feature state is selected, the entities that do not have that feature will be moved into the "Entities Discarded" window (lower right), and the character state selected will be moved to the lower left window. All entities and feature states are richly illustrated with photographs and/or drawings. Clicking on the image thumbnail of a state brings up a larger image. Clicking on the taxon thumbnail brings up a larger window with thumbnails of a line drawing with important structures labeled, a whole body picture, and a habitus picture (when available). Clicking on the name of the taxon brings up a link to the fact sheet for that taxon.

Conclusions
Protecting the borders of large countries such as the United States from invasive scales often requires a very broad knowledge of the taxonomy the group, and consultation of books and papers that are scattered all over the place. Scale Insects, edition 2 may facilitate the job of target users, which include USDA APHIS pest survey specialists, identifiers at ports of entry, state and county identifiers, students, and scientists, in three ways. First, it condenses, in one online resource, a wide array of information on target species from various zoogeographical regions. Second, it maintains current . Features can be selected in any order or they can be selected by the program by clicking on the "best character" icon (indicated with black arrow). Entities with the character states selected remain in the "entities remaining" window (upper right window) whereas those that do not match the states selected are sent to the "entities discarded" (lower right window). taxonomic information through links to ScaleNet, a relational database that is updated regularly. Third, additional species of concern can be easily added to the Lucid keys and fact sheets. We believe that our tool will facilitate insect pest identifications and we hope that it will inspire taxonomists in other groups to build similar tools.