Taxonomic and identification review of adventive Fiorinia Targioni Tozzetti (Hemiptera, Coccomorpha, Diaspididae) of the United States

Abstract This work provides general descriptions, illustrations, molecular diagnostic data, taxonomic keys, slide mounting recommendations, and Florida distribution records for Fiorinia Targioni Tozzetti species occurring in the USA. Species treated are F.externa Ferris, F.fioriniae (Targioni Tozzetti), F.japonica Kuwana, F.pinicola Maskell, F.phantasma Cockerell & Robinson, F.proboscidaria Green, and F.theae Green. New descriptions of second-instar males and females of all seven species in addition to first-instar nymphs and adult females of F.phantasma and F.proboscidaria are presented. Taxonomic keys to second-instar males and females are developed for the first time and previously available taxonomic keys to first-instar nymphs and adult females are improved. DNA sequences were used to further evaluate the monophyly of Fiorinia and provide additional diagnostic tools for Fiorinia species. Multigene phylogenetic analyses, COI barcoding methods, and examination of type material indicate that F.yongxingensis Liu, Cai & Feng, 2020, syn. nov. is a junior synonym of F.phantasma. A morphological survey of the genus demonstrates, for the first time, the utility of second-instar males for diagnostics. This study will help inform regulatory and pest management decisions by facilitating morphological and molecular identification of adventive Fiorinia species occurring in the USA.

Fiorinia phantasma, commonly known as phantasma scale, was described from the Philippine Islands in 1915. Subsequently, a major global expansion of F. phantasma occurred over the last decade through movement of nursery stock (Watson et al. 2015). Fiorinia phantasma is now documented from 19 countries (China (Hong Kong, Mainland China, Taiwan), France, French Polynesia, Grenada, Indonesia, Malaysia, Maldives, Nauru, Netherlands, New Caledonia, Papua New Guinea, Philippines, Reunion, Saint Barthelemy and Saint Martin, Singapore, Solomon Islands, Thailand, United States (American Samoa, Florida, Hawaii, Guam), and Vietnam). In some areas, F. phantasma may reach heavy infestations causing serious plant damage (Watson et al. 2015;García Morales et al. 2016). In one particularly impactful infestation of F. phantasma, approximately 6,000 palms were severely infested and declining at a resort in the Maldives (Watson et al. 2015). A polyphagous pest, F. phantasma has been reported on 25 families and 56 genera of hosts, including many nursery and ornamental plants, particularly palms, as well as several fruit crops (Watson et al. 2015;García Morales et al. 2016;Ahmed et al. 2021). For the nursery and greenhouse sector, palms account for sales of approximately $400 million annually in Florida and well over $1 billion annually in the USA (Khachatryan and Hodges 2012). Scale insects feed on all parts of their host plants, but F. phantasma is common on leaves, causing chlorosis, leaf drop, and ultimately plant death. This pest has the potential to cause economic harm in the USA to nurseries, landscape industries, and homeowners.
The first North American continental report of F. phantasma was in Florida and included more than twenty heavily infested Canary Island date palms (Phoenix canariensis Chabaud) along both sides of a road in Miami-Dade County (Ahmed 2018). The population was likely there for some time, considering the density of the scales and the presence of specimens on many trees. It is not surprising that the Florida infestation was not detected earlier because the scale is identical in field appearance to other Fiorinia species that occur in Florida (Ahmed 2018). Fiorinia species infestations start with the arrival of crawlers (first-instar nymphs), either by wind, or via infested plant material or garden tools because crawlers constitute the only mobile stage besides adult males, which do not feed. Crawlers settle on plant parts and molt into second-instar males and females within a few days.
The main pest management challenge is detection of new F. phantasma infestations. Fiorinia phantasma occurs in two Florida counties, Miami-Dade and Palm Beach, and is usually found on palms (FDACS-DPI Entomology Database 2021). Detection is complicated by the presence of F. fioriniae, which is commonly found on palms throughout most of Florida (FDACS-DPI Entomology Database 2021). Fiorinia japonica, another morphologically and behaviorally similar species, also infests palms, but is only found in California and several east coast states in the USA. Should F. japonica become established in Florida, it would be difficult to detect because the species looks identical in the field to the other Fiorinia species infesting palms. Heavy infestations of another Fiorinia species, F. proboscidaria, were recently recorded on citrus from residential areas in Florida. Regulatory efforts aimed at preventing its introduction to and establishment in commercial citrus growing areas in Florida are being implemented (Ahmed and Stocks 2020). To date, the only way to identify these species has been to mount adult females on a microscopic slide and examine them with a compound microscope. The regulatory and pest management situation surrounding Fiorinia species in the USA, and especially Florida, is dynamic and subject to identification challenges. Thus, it is important to develop identification tools for Fiorinia adult females and other commonly collected life stages using diagnostic molecular and morphological data. Without reliable and correct identification, one cannot properly make regulatory and control decisions.
The purpose of this study is to provide taxonomic keys for immatures of seven Fiorinia species occurring in the USA. We also provide line drawings and diagnoses of slide-mounted second-instar males and females, DNA sequence data for multiple loci for molecular diagnostics, and extensive records of the species' distributions in Florida. We newly describe and illustrate first-instar nymphs and adult females of two species, F. phantasma and F. proboscidaria. In addition, we provide updated taxonomic keys for first-instar nymphs (adapted from Howell 1977) and adult females (Watson et al. 2015).

Taxon sampling
Four species of Fiorinia (F. fioriniae, F. phantasma, F. proboscidaria, F. theae) were collected from Florida (Suppl. material 2: Table S1). Fiorinia externa samples were collected from Christmas trees imported from outside of Florida. First-and second-instar nymphs and adult females from infested plant materials were preserved in 100% ethanol for slide mounting and molecular analysis. Fiorinia japonica and F. pinicola specimens were borrowed from the United States National Museum of Natural History, scale insect collection, Beltsville, Maryland (USNM). Fiorinia pinicola specimens were provided to us by Natalia von Ellenrieder (California Department of Food and Agriculture) (Suppl. material 2: Table S1). The details for specimens examined for description and diagnosis is provided in the figure captions of each species. Due to regulatory issues surrounding F. yongxingensis Liu, Cai & Feng, 2020 from Hainan, China, its DNA sequences were obtained in China by one of us (DL). All samples were initially mounted in Hoyer's medium for visibility during illustration and were transferred to balsam medium for permanent preservation. This was done by placing the Hoyer's slide in a petri dish filled with water, just enough that the slide is slightly submerged, for a few hours depending on the age of the Hoyer's slide. Once the slide cover is detached and loosened, it can easily be removed. The specimen can then be removed from the slide without being damaged. Specimens were soaked in a watch glass filled with water overnight to rinse Hoyer's media out of the specimen. After soaking, specimens were placed on a new slide with a drop of balsam and covered with a new coverslip. Illustrations were made using a Leica DMRB compound microscope and a camera lucida. Morphological terminology follows that of Miller and Davidson (2005). Numerical values were taken from a minimum of five specimens, if available, from as many Florida localities as possible. All specimens were deposited in the Florida State Collection of Arthropods, Gainesville (FSCA) unless otherwise indicated. Other depositories included USNM (United States National Museum of Natural History, scale insect collection, Beltsville, Maryland), UMEC (University of Massachusetts Entomology Collection, Amherst, Massachusetts), and Entomology Museum, Northwest Agricultural and Forestry University, Shaanxi, China.
In addition to the freshly collected Fiorinia specimens described above, additional specimens and sequences were included in analyses of DNA sequences (Suppl. material 2: Table S1): fresh specimens of the outgroups Thysanofiorinia leei Williams and T. nephilii (Maskell) collected in Florida; ethanol-preserved specimens of Fiorinia sp. collected in Lambir Hills National Park, Malaysia, in 2013; cytochrome oxidase I (COI) sequences of Diaspididae from the BOLD database (Ratnasingham and Hebert 2007), along with one sequence of Pseudococcus sp. (BOLD record AMSMB002-15; BIN BOLD:ACZ2386) as an outgroup; and cytochrome oxidase I and II (COI-II), elongation factor 1a (EF1a), and large ribosomal subunit (28S) sequences of the genus Fiorinia reported in Normark et al. (2019), along with exemplars of other species of Fioriniina and one sequence of Unaspis yanonensis (Kuwana) as an outgroup.

Slide mounting of immature stages
Slide mounting is considered mandatory for morphological identification of armored scale insects because it is nearly impossible to identify taxonomic features without doing so. Moreover, for museum curation purposes, slide mounting is the best way to archive scale insects in a reference collection. There are studies available on methods for mounting hemipterans (Hodges and Evans 2005), but many are not specific to scale insects or armored scales (Wirth and Marston 1968). Previously published mounting methods for scale insects (McKenzie 1957;Wilkey 1990;Watson 2002) need to be reevaluated to meet the need for rapid identification as pest species are spreading swiftly through national and international trade. Recently published methods have focused on modifying slide mounting to enhance safety since the reagents can be corrosive, flammable, or carcinogenic, or can produce toxic fumes (Sirisena et al. 2013). Another recent study modified the watch glass with a sieve to process specimens in a shorter period (Barbecho and Lit 2015). Nevertheless, a reliable protocol for slide mounting of immature armored scale insects still needs to be established. Mounting methods are also biased towards adult scales, despite the importance of first and second-instar nymphs to armored scale biology. These immature life stages are commonly found in the field, but are taxonomically studied to a much lesser degree than adults. We evaluated several methods to enhance safety and reduce the time required to mount fresh and absolute ethanol-preserved specimens of first-and second-instar nymphs of Fiorinia species.

(i) Standard slide mounting method (6 steps)
Initially, a 67 mm beveled-edge watch glass (Prolab Scientific) and micro spatula were used. Fisher 10% potassium hydroxide (KOH) was used in step 1 for heating and maceration. Following this step, specimens were placed into a Humboldt mesh (Replacement Mesh Disk 5 cm dia. No. 325H-3807.325) container that was then placed inside the watch glass, eliminating the need for the micro spatula in the following steps until the final mount. Forceps (Bioquip Swiss style #4) were used to remove the mesh from the 4.8 cm watch glass (Item#742300, Carolina) while switching between steps. Glacial acetic acid (Fisher) was used in step 2 for removal of the remaining 10% KOH from step 1. Acid fuchsin stain (Bioquip) was used with a 3:7 dye to acetic acid ratio in step 3 to stain the specimens. For dehydration of the cuticle in the 4 th step, 75% and 95% EtOH were used. Clove oil (Spectrum Chemical) was used in step 5 to remove any remaining wax from the specimens. A disposable transfer pipette (13-711-9D, Fisher) which holds 3.2 ml, was used in steps 2-5. In the final step 6 filtered Canada balsam Fisher) was used as the medium and placed on a glass slide Fisher). A glass coverslip (12-545-80P, Fisher) was placed on the specimen in balsam to complete the mount. The 6 steps required for the standard slide mounting method are as follows: 1. Heating: specimens were set in a watch glass filled with 10% KOH and heated at 85 °C for 5-10 mins. After heating, gut contents were teased out using a microspatula to gently tap the dorsum.
2. KOH removal: specimens were moved to a watch glass of 95% glacial acetic acid for 10 mins to remove any remaining KOH.
3. Staining: Acid fuchsin stain was added and let sit for 5 mins. 4. Stain correcting: specimens were moved to a watch glass of 75% EtOH for 10 mins. Specimens were then placed in 95% EtOH for another 10 mins to dehydrate. 5. Wax removal: specimens were soaked in clove oil for 5 to 10 mins. This helps to remove any remaining wax or lipids and makes specimen bodies flexible to be easily spread on a slide. 6. Mounting: on a labeled slide, a drop of balsam was placed in a center and spread to avoid specimen drift. The specimen was then placed in the balsam dorsoventrally (i.e. ventral side up) and legs and antennae were positioned properly. A coverslip was placed on the balsam, and the slide was placed on a hot plate at 30 °C for 10 mins to remove any bubbles.
Due to the multiple steps in this method, which require each specimen to be moved from 5 different watch glasses before mounting, many first-instar nymphs were lost or damaged. Additionally, this method was time consuming. In an attempt to reduce the loss of first-instar nymphs, minimize damage, reduce the amount of chemical usage, and save time, we subsequently developed alternative methods -see below.

(ii) Modified slide mounting method A (1 step)
For fresh specimens (not preserved in ethanol).
1. Mounting: on a labeled slide, a drop of Hoyer's medium was placed in the center and spread to avoid specimen drift. Fresh specimens picked from plant material were placed in a Hoyer's medium dorsoventrally and legs and antennae were positioned properly. In this protocol, we omitted steps 1-5 of the standard method and mounted specimens directly into Hoyer's medium. This was effective in preventing loss of specimens and reducing the amount of chemical usage.
1. Heating: specimens were placed in a watch glass filled with 10% KOH and heated for 5 mins at 85 °C.
2. Rehydrating: specimens were placed in water and left to soak for 5-10 mins. We found that heating the specimens prior to submerging them in water aided in the rehydration process.
3. Cleaning: specimens were moved to a watch glass filled with Hoyer's medium. Because Hoyer's medium is a self-cleaning fluid (Anderson 1954), specimens were placed in the dish to accelerate the cleaning. 4. Mounting: on a labeled slide, a drop of Hoyer's medium was placed in the center and spread to avoid specimen drift. The specimen was placed in the Hoyer's medium dorsoventrally and legs and antennae were positioned properly.
(iv) Modified slide mounting method C -balsam method with mesh container (7 steps) For fresh specimens and ethanol-preserved specimens.
1. Heating and cleaning: specimens were placed in a watch glass filled with 10% KOH and heated at 85 °C for 5-10 mins. After heating, cavity contents were teased out using a micro-spatula. Once this step was completed, specimens were moved to a container modified using mesh placed in a watch glass (Fig. 1). The modified mesh container was made using a plastic 5 ml screw-top tube and fine wire mesh (Humboldt, Elgin, IL United States). The top was cut out of the screw-top tube and the mesh was put in its place, allowing liquid to move through the mesh while keeping the specimens inside.
2. Rehydration: specimens were placed in water and left to soak for 5 to 10 mins. 3. KOH removal: specimens were moved to a watch glass of 95% acetic EtOH (a few drops of glacial acetic acid with 95% ethanol) for 10 mins.
5. Stain correction and dehydration: specimens were moved to a watch glass of 75% EtOH for 10 mins. Specimens were then placed in 95% EtOH for another 10 mins to dehydrate the cuticles.
6. Wax removal: specimens were soaked in clove oil for 5 to 10 mins. 7. Mounting: on a labeled slide, a drop of balsam was placed in a center and spread to avoid specimen drift. The specimen was placed in the balsam and legs and antennae were positioned properly. A coverslip was placed on the balsam and the slide was placed on a hot plate at 30 °C to remove any bubbles.
Although the mesh is effective in keeping first-and second-instar nymphs in the container without damage, a few problems were noted. The mesh does not sit flat against the glass bottom of the watch glass, so the cleaning step cannot be done in the mesh. Cleaning must be done in a watch glass and then specimens must be moved back into the mesh for the remaining steps. Due to the smaller size of the mesh container, range of motion using microtools throughout this process can be limited. Similar to processing in a watch glass without mesh, specimens can get stuck on the upper sides of the modified dish. Visibility of first-instar nymphs can be hampered by the reflective coloration of the mesh.
There are several steps involved in traditional slide-mounting protocols (method i) that require each specimen to be moved to and from at least five different watch glasses before eventually being slide mounted. Many first-instar nymphs can be lost or damaged during these steps. We recommend using the mesh container during the slide-mounting protocol (method iv). Use of this container will decrease mounting time, reduce specimen loss, decrease the quantity of chemical reagents, and generate quality slides. All steps can easily be performed using the mesh container except for the cleaning step. Unfortunately, the cleaning step must be done in a watch glass and then the specimens should be moved back into the mesh container to finish the mounting process. Although this procedure is laborious, we recommend it when the aim is to make permanent mounts for deposit in archival collections. The other mounting procedure is to place first-instar specimens directly into Hoyer's mounting medium on a slide (method ii, iii). This protocol has fewer steps and less chance of specimen loss, and yields specimens with superior visibility. We recommend this protocol for rapid species diagnosis. Unfortunately, the mounts are only temporary unless slides are ringed to prevent deterioration.
DNA extractions, polymerase chain reaction (PCR), and sequencing DNA was extracted from individual Fiorinia and Thysanofiorinia specimens using the Qiagen Blood and Tissue Kit per the manufacturer's protocol. Extractions were nondestructive, and recovery of individual scale vouchers was attempted. DNA was quantified on a Nanodrop 2000 and PCRs had a target input of at least 5 ng of genomic DNA. PCRs were performed using the Kapa HiFi HotStart PCR Kit, in a total volume of 25 uL.

Data analysis
Cytochrome oxidase I barcode sequences (5'-COI) were initially aligned using an online version of MAFFT 7 (Katoh and Standley 2013) with the FFT-NS-2 strategy for relatively short, similar sequences. A few sequences with excessive ambiguities or large insertions were excluded from further analysis. The resulting barcode matrix included 1177 terminal taxa and was 649 bp in length.
Sequences were aligned using the default settings of MUSCLE (Edgar 2004) and Clustal W (Larkin et al. 2007) as implemented in MEGA X (Kumar et al. 2018). The lengths of the alignments were 645bp (5'-COI), 226 bp (3'-COI), 504 bp (COII), 708 bp (EF1α, introns omitted), and 425 bp (28S, regions of uncertain homology omitted). Alignments were concatenated as a single nexus file in Mesquite 3.51 (Maddison and Maddison 2018). PCR amplifications with 3'COI/COII primers failed to produce clear bands or clean sequence data on each attempt in this study. All of 3'COI/ COII sequences used in this study were from Normark et al. (2019).
Neighbor-joining and distance analyses of the 5'-COI matrix were conducted in MEGA X (Kumar et al. 2018). Neighbor-joining trees were constructed using the K2P model (Kimura 1980) with partial deletion of missing data and a site coverage cutoff of 95%. Node support was assessed using 10,000 bootstrap replicates. The resulting tree topology was adjusted in FigTree v1.4.3 (Rambaut 2012) to arrange nodes and collapse large clusters. Intra-and interspecific K2P distances among Fiorinia species were calculated with the same parameters as above using a separate alignment that only included Fiorinia barcodes.
Phylogenetic analyses were conducted using 3 sequence regions reported in Normark et al. (2019): portions of cytochrome oxidase I and II (using a 3' portion of COI nonoverlapping with the 5'-COI barcoding matrix: 3'-COI & COII), elongation factor 1a (EF1α), and the large ribosomal subunit (28S), as well as the 5'-COI region. The aim was to assess the monophyly of Fiorinia and the relationship of Fiorinia species to other species of Fioriniina.

Phylogenetic analyses
Maximum Likelihood analyses estimated a consensus bootstrap tree with a log-likelihood of -20,889.543 for the multigene tree (Fig. 2, Suppl. material 1: Fig. S2) and -3006.382 for the 28S tree (Suppl. material 1: Fig. S3). Parsimony ratchet analyses found five equally parsimonious trees with 4123 steps. A clade of grass-feeding Fioriniina (Unachionaspis MacGillivray + [Kuwanaspis MacGillivray + Nikkoaspis Kuwana]) was recovered, but the node was only weakly supported (Fig. 2). As in Normark et al. (2019), the Australasian Fioriniina (Pseudaulacaspis MacGillivray in part; Poliaspis Maskell; Anzaspis Henderson) were recovered as a clade by likelihood and parsimony methods, but with relatively higher support in some analyses (BS 80; SH-aLRT 92). These Australasian Fioriniina were sister to a clade of Fiorinia + Lineaspis MacGillivray + Pseudaulacaspis in part, with weak support except for SH-aLRT (92). The clade of Fiorinia + Lineaspis + Pseudaulacaspis was found by likelihood and parsimony, with some strong support (ML UF BS 95; SH-aLRT 93) (Fig. 2). Relationships within this clade were not entirely resolved, resulting in a polytomy. Fiorinia is monophyletic in our tree, with the exception of two isolates (Fiorinia sp., D4815B and D4815C) which were represented only by 28S data. These two isolates belong to an undescribed Fiorinia species from Malaysia. The remaining Fiorinia isolates formed a clade in likelihood analyses (SH-aLRT 99). Relationships among Fiorinia species were generally weakly supported. A terminal group of Fiorinia phantasma + F. yongxingensis was present in every analysis with strong support suggesting synonymy (Fig. 2).
The slide-mounted cuticle of D4815B and other specimens in the same lot have been re-examined by BBN and they clearly belong to a pupillarial species whose morphology is completely consistent with the genus Fiorinia. These results might imply that the lineage leading to D4815B and D4815C represents a second origin of the pupillarial habit in Fioriniina. These two isolates were placed within a section of a Fiorinia + Rolaspis + Pseudaulacaspis (in part) clade in the ML phylogenetic tree using only 28S data (Suppl. material 1: Fig. S3). They were placed with five species of Pseudaulacaspis (including P. biformis, P. cockerelli, P. momi, P. pentagona, and P. prunicola) with strong support (Suppl. material 1: Fig. S3). In addition to these five species of Pseudaulacaspis, three species of Rolaspis (including R. incisa, R. lounsburyi, and R. whitehilli), and one species of Pellucidaspis (P. epiphytidis) were also placed within this Fiorinia clade.

COI barcoding
This study produced 43 new sequences of the COI barcode region, 37 of which represent nine Fiorinia species (Fig. 3, Suppl. material 1: Fig. S4). The remaining 6 COI barcode sequences represent two species of Thysanofiorinia. These new barcode sequences range in length 562 bp-645 bp. In the neighbor-joining analyses of Diaspididae COI barcodes, Fiorinia species cluster near the species of Kuwanaspis, Unachionaspis, and Pseudaulacaspis (all members of Fioriniina), along with a sequence assigned to the genus Aulacaspis (subtribe Chionaspidina). (Fig. 3). Fiorinia species represented by multiple barcode sequences each formed well-supported clusters (100 BS) in the neighbor-joining tree, with one exception: F. theae. Fiorinia theae forms two well supported clusters whose relationship to each other is not resolved in this analysis (Fig. 3).
The alignment for calculating K2P distances among Fiorinia species included 37 terminal taxa and was 645 bp long. Based on the 95% site cutoff, calculations involved 560 total positions. Intraspecific K2P distances were low, except for specimens identified as F. theae (Table 1). Interspecific K2P distances between Fiorinia species ranged from 9.1% to 15.2% (Table 1). Sequences of F. phantasma from the population from Florida and Malaysia and sequences of F. yongxingensis were identical and were placed together in the tree with strong support (Fig. 3).

Second-instar females
With two definite pairs of lobes; third lobes and sometimes fourth lobes represented by series of points. Median lobes yoked, medial margins divergent or nearly parallel, longer than lateral margin, with series of notches. Second lobes bilobed, usually smaller than median lobes, sometimes wider, medial lobule largest, sometimes with small notches, lateral lobule sometimes with one or two small notches. Third lobes usually represented by raised sclerotized area with small series of notches, often divided into two lobules by seta marking segment VI. Fourth lobes sometimes represented by series of sclerotized points. Gland spine arrangement of two types: F. proboscidaria and F. theae with single gland spine on each side of each of segments II-VIII, gland spines on each side of segments II-IV larger than those on segments V-VIII, without gland spines on segment I; remaining species with single gland spine on each side of segments II-V, absent from segment VI, present on each side of segments VII and VIII, gland spines on each side of segments II-V larger than those on segments VI-VIII, with two or three smaller gland spines on each side of segment I; gland spines with barely perceptible sclerotization posterolaterad of each spiracle. Macroducts barrel shaped, marginal, with four or five on each side of pygidium from segments III or IV-VII. Microducts restricted to venter, three different patterns on abdomen; in F. proboscidaria and F. theae longitudinal lines on each side of abdomen from II-VI, each line composed of one or more ducts on each segment, mediolateral line on segments III or IV, V or VI, submarginal line on segments II-VI; in F. externa, F. fioriniae, F. japonica, and F. pinicola longitudinal lines on each side of abdomen from II-VI, each line composed of one or rarely two ducts on each side of each segment, mediolateral line on segments II-V or VI, submarginal line on segments II-VI; in F. phantasma longitudinal lines restricted to mediolateral areas of segments II-IV or V, other lines absent; microducts on head and thorax usually anterior of clypeus, laterad of labium, and posterior of each spiracle. Perispiracular pores associated with anterior spiracles only, with three loculi, one or two pores associated with each spiracle. Anal opening normally located in center of pygidium mesad of fourth marginal macroduct counting 0.00%-0.02% 11.8%-14.8% Fiorinia phantasma (n = 7) 0.00%-0.09% 9.1%-13.7% Fiorinia pinicola (n = 3) 0.00% 9.1%-15.2% Fiorinia proboscidaria (n = 6) 0.00%-0.02% 9.9%-14.2% Fiorinia theae (n = 5) 0.00%-8.00% 9.5%-14.8% Fiorinia sp. isolate D4778A (n = 1) N/A 9.5%-13.9% Fiorinia sp. isolate D4674F (n = 1) N/A 9.5%-12.7% Fiorinia sp. isolate D4682A (n = 1) N/A 9.1%-15.2% forward from posterior macroduct. Dorsal setae present near body margin on head and thorax, with one seta submarginally on each side of each abdominal segment; also present in mediolateral area on each side of body on any or all of abdominal segments I-VI; usually with one mediolateral seta on each side of head. Ventral setae in small numbers in marginal areas of head and thorax, with one seta usually present laterad of each spiracle; abdominal segments with one marginal and one submarginal seta on each side of each segment and with one mediolateral seta on each side of segments IV-VI. Antennae each normally with one long seta and two small sensillae. Cicatrices present or absent on each side of abdominal segment I. Two inconspicuous lobes present submarginally on head of F. proboscidaria and F. theae.

Notes
Characters most useful in distinguishing among species are: a) number of marginal macroducts; b) arrangement of gland spines; c) arrangement of microducts; d) presence or absence of cicatrices; e) presence or absence of lobes on head; f ) relative size of median lobes compared to medial lobule of second lobe; g) shape of median lobes. Second-instar females of Fiorinia species can be distinguished from most similar genera by having the following: median lobes yoked, usually divergent, medial margin longer than lateral, with one pair of setae between; dorsal macroducts confined to body margin, with four or five on each side of pygidium; with two pairs of definite lobes, second pair bilobular. However, we have been unable to distinguish between second-instar females of the Fiorinia species treated here and Pseudaulacaspis cockerelli (Comstock) and P. pentagona (Targioni Tozzetti). There are consistent differences in the distribution of the gland spines in most species of Fiorinia, but F. proboscidaria and F. theae are identical to P. cockerelli and P. pentagona. It is remarkable that the secondinstars are so similar, but the adult females are quite different.

Second-instar males
With two definite pairs of lobes; remaining body margin often with numerous projections, not organized into clear lobes. Median lobes spaced apart, without zygosis, usually with small medial lobule and large, conspicuous lateral lobule, medial lobule with one or two projections, lateral margin with several notches and projections. Second lobes usually associated with a dense cluster of marginal ducts, with series of projections, rarely bilobed, smaller than median lobes. Gland spines of three sizes: largest in clusters posterolaterad of each anterior spiracle, posterolaterad of posterior spiracle, and submarginal on abdominal segment I and sometimes II; medium-sized gland spines on body margin of anterior abdominal segments; small gland spines laterad of anterior spiracle on F. externa and F. theae. Macroducts barrel shaped, of two sizes: larger ducts grouped into communal ducts (= glanduliferous craters; Takagi 1999) that exit through single orifice with numerous fine filaments or series of short projections on margin; communal ducts either separate or associated with clusters of smaller macroducts; smaller macroducts ca. half as large as larger ducts, arranged singly or in clusters on prepygidial and pygidial margin. Microducts present on dorsum and venter, arranged in longitudinal lines, of two sizes: smaller size relatively slender, longer than wide, present on venter of most abdominal segments, on venter of head, in ventromedial areas of thorax, and on dorsum of posterior two or three segments; larger ducts ca. as long as wide present on venter in submarginal areas of thorax, on dorsum in submarginal areas of prothorax to anterior abdominal segments and submedially on anterior abdominal segments. Perispiracular pores associated with anterior spiracles only, with three loculi, 1-3 pores associated with each spiracle. Anal opening normally located in center of pygidium mesad of anterior edge of posterior cluster of macroducts. Dorsal setae present near body margin on head and thorax, setae associated with duct clusters long and conspicuous; also present in mediolateral area on each side of body on any or all of abdominal segments I-VI, usually with several mediolateral seta on each side of head. Ventral setae in small numbers in marginal areas of head and thorax, with one seta usually present laterad of each spiracle; abdominal segments with one marginal and one submarginal seta on each side of each segment and with one mediolateral seta on each side of segments IV-VII. Antennae each normally with one long seta and two small sensillae. Cicatrices absent.

Notes
Characters most useful in distinguishing among species are: a) arrangement and number of communal ducts b) organization of duct clusters c) arrangement of microducts; d) arrangement of gland spines. Second-instar males of Fiorinia are remarkably similar to the same instar of Pseudaulacaspis species by each having unusual lobes, duct clusters, and communal ducts (Takagi and Kawai 1967). Pseudaulacaspis species differ primarily by the presence of many ventral microducts on the head and barrel-shaped microducts in the medial and submedial areas of the abdominal venter, whereas Fiornia species possess no more than two ventral microducts on the head, and slender microducts on the submedial areas of the abdominal venter. In Normark et al. (2019), the subtribe Fioriniina comprises many genera and species with second-instar males that are similar in appearance to the species treated here. Howell (1977) gave a general description of the first-instar nymphs of the species that he examined. We will not repeat that here. Below, we present diagnoses of the two species that were not included in the Howell (1977), i.e., F. phantasma and F. proboscidaria.

First-instar nymphs
First-instar nymphs of Fiorinia species can be recognized by having the following combination of characters: antennae five segmented; apical segment annulate; large duct on each side of dorsum of head; submedial longitudinal line of microducts on each side of thorax; second lobes bilobulate. First-instar nymphs of Fiornia species are similar to some species of Pseudaulacaspis (P. cockerelli and P. pentagona) but differ by normally having a submedial longitudinal line of microducts on each side of thorax, whereas these ducts are absent from P. cockerelli and P. pentagona (Tippins and Howell 1983).

Fiorinia externa Ferris, 1942
Field characteristics. First-instar exuviae barely touching second-instar exuviae. Distinct indentation formed between attachment of first-and second-instar exuviae. Second-instar exuviae narrow, parallel sided, and elongate; longitudinal ridge absent or weakly developed. Second-instar exuviae reddish brown anteriorly and light brown to yellow posteriorly. Posterior end of adult female within second-instar exuviae rounded (Suppl. material 1: Fig. S1). First-instar nymph. Described in Howell (1977). Second-instar female. Median lobes slender, narrower than medial lobule of second lobe, not projecting beyond medial lobule of second lobes. With five pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually pointed. With four large gland spines on margin of each side of body from abdominal segments II-V; usually without small gland spine on each side of abdominal segment VI; with small gland spines on margin or submargin of abdominal segment I. With one microduct on each side of head. Longitudinal line of microducts present submarginally on venter of II-V, normally with one microduct on each side of each segment. Cicatrices absent.
Second-instar male. Three duct clusters on each side of body; posterior cluster composed of several small ducts and two communal ducts. Five longitudinal lines of microducts on venter of abdomen (one medial, two mediolateral, two submarginal). Cluster of small microducts with sclerotized orifice laterad of anterior spiracle. Fewer than five large-sized gland spines on each side of body between anterior and posterior spiracles. Antennae each with one enlarged seta.
Florida collection records. All records are on Christmas trees imported from states outside of Florida. This species is not established in Florida, and its common host, Abies fraseri, also does not occur naturally in Florida. It has been found on imported Christmas trees in the following localities in
First-instar nymph. Described in Howell (1977). Second-instar female. Median lobes broad, equal to or wider than medial lobule of second lobe, projecting ca. same or slightly less than medial lobule of second lobes. With five pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually rounded. With four large gland spines on margin of each side of body from abdominal segments II-V; usually without small gland spine on each side of abdominal segment VI; with small gland spines on margin or submargin of abdominal segment I. With three microducts on each side of head. Longitudinal line of microducts present submarginally on venter of abdominal segments II-VI, normally with one microduct on each side of each segment. Cicatrices absent.
Second-instar male. Submargin of abdominal segments II-VI with scattered small-sized macroducts, not in clusters; communal ducts absent. Medial longitudinal line of microducts absent. Cluster of small microducts with sclerotized orifice laterad of anterior spiracle absent. Fewer than five gland spines on each side of body between anterior and posterior spiracles. Antennae each with several enlarged setae.
Notes. The single specimen collected with identified adult females of this species is unusual and may not be the second-instar male of this species. U.S. populations of Fiorinia fioriniae have been reported to be parthenogenetic (Tippins 1970), so it is surprising to find a male, although many scale insect species with parthenogenetic populations also have sexual populations (Nur 1990). The specimen is unusual among second-instar males of Fiorinia in lacking tight clusters of marginal ducts. There exist a few other species of Fiorinia with males that similarly lack these ducts, for instance F. nachiensis Takahashi of Japan; thus it is plausible that this really is the male of F. fioriniae.
First-instar nymph. described in Howell (1977). Second-instar female. Median lobes broad, as wide as or wider than medial lobule of second lobe, projecting ca. same distance as or further than medial lobule of second lobes. With five pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually rounded. With four large gland spines on margin of each side of body from abdominal segments II-V; usually without small gland spine on each side of abdominal segment VI; with small gland spines on margin or submargin of abdominal segment I. With one microduct on each side of head. Longitudinal line of microducts present submarginally on venter of abdominal segment II or III-VI, normally with one microduct on each side of each segment. Cicatrices absent.
Second-instar male. One duct cluster on each side of body, composed of several small ducts and two communal ducts. Three longitudinal lines of microducts on venter of abdomen (one medial and two submarginal). Cluster of small microducts with sclerotized orifice laterad of anterior spiracle absent. Fewer than five gland spines on each side of body between anterior and posterior spiracles. Antennae each with one enlarged seta.
First instar. Similar to F. fioriniae and F. proboscidaria in having gland spines on abdominal segment VI at least half as long as gland spine on segment VII. Fiorinia fioriniae and F. proboscidaria differ by having (characters in parentheses are those of P. phantasma): pattern of derm surrounding large duct on head serpentine (globular); inner apex of large duct on head flat (rounded or mushroom like).
Second-instar female. Median lobes broad, as wide as or slightly narrower than medial lobule of second lobe, projecting ca. same amount or slightly less than medial lobule of second lobes. With five pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually pointed. With three large gland spines on margin of each side of body from abdominal segments II-IV, without gland spine on abdominal segment VI; with small gland spines on margin or submargin of abdominal segment I. With three microducts on each side of head. Longitudinal line of microducts absent submarginally on venter of abdomen. Cicatrices absent.
Second-instar male. One duct cluster on each side of body, composed of several small ducts and one communal duct. Five longitudinal lines of microducts on venter of abdomen (one medial, two mediolateral, and two submarginal), medial line sometimes incomplete. Cluster of small microducts with sclerotized orifice laterad of anterior spiracle absent. Fewer than five gland spines on each side of body between anterior and posterior spiracles. Antennae each with one enlarged seta.
Adult female. Body tapering at segment III to narrow pygidium. With three or four pairs of dorsal macroducts on each side of body, ducts similar in shape and size to microducts. Projection between antennae with many spicules. Antennae close together, with distinct projection.    Notes. We examined four paratype slides of F. coronata Williams & Watson from Guadalcanal, Solomon Islands deposited in the USNM collection at Beltsville, Maryland. Most of the specimens were punctured in the middle of the body between the posterior spiracles during the mounting process. However, we could still see that all had microducts between the posterior spiracles which were small and less numerous than specimens from elsewhere, but they definitely are there.
We also examined a paratype slide of F. phantasma in the same collection, but it is in such poor condition that only half of the pygidium is useful for diagnosis. It is impossible to even find the posterior spiracles, let alone microducts between them. In addition, the holotype of F. phantasma deposited in The Natural History Museum, London (NHMUK) was loaned to and examined by one of us (DL). It also was in poor condition; microducts close to anterior and posterior spiracles and in prepygidial abdominal segments were not visible. All of the examined specimens of F. phantasma that were in good condition had easily discernable microducts between the posterior spiracles.
First-instar nymph. Described in Howell (1977). Second-instar female. Median lobes broad, as wide as or wider than medial lobule of second lobe, projecting ca. same amount as medial lobule of second lobes. With five  pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually pointed. With four large gland spines on margin of each side of body from abdominal segments II-V; usually without small gland spine on each side of abdominal segment VI; with small gland spines on margin or submargin of abdominal segment I. With one microduct on each side of head. Longitudinal line of microducts present submarginally on venter of abdominal segments II-VI, normally with one microduct on each side of each segment. Cicatrices absent.
Notes. We have been unable to find characters that consistently separate second-instar females of F. japonica and F. pinicola. The swelling of the body adjacent to the abdominal macroducts is usually pointed in F. pinicola and is usually rounded in F. japonica, but we have too few specimens to understand the possible variation in this character.
Second-instar male. One duct cluster on each side of body, composed of several small ducts and two communal ducts. Three longitudinal lines of microducts on venter of abdomen (one medial and two submarginal). Cluster of small microducts with sclerotized orifice laterad of anterior spiracle absent. Five or more gland spines on each side of body between anterior and posterior spiracles. Antennae each with one enlarged seta.

Fiorinia proboscidaria Green, 1900
Field characteristics. First-instar exuviae overlapping second-instar exuviae. Without indentation formed between attachment of first-and second-instar exuviae. Secondinstar exuviae oval, convex marginally or parallel sided; light to medium dark brown; longitudinal ridge conspicuous and thick. Posterior end of adult female within secondinstar exuviae rounded. Heavily infested leaves with white secretion.
First-instar nymph. Similar to F. fioriniae and F. phantasma in having gland spines on abdominal segment VI at least half as long as gland spine on segment VII. Fiorinia fioriniae differs by having (characters in parentheses are those of P. proboscidaria): inner apex of large duct on head flat (rounded or mushroom like). Fiorinia phantasma differs by having (characters in parentheses are those of F. proboscidaria): pattern of derm surrounding large duct on head globular (serpentine); inner apex of large duct on head rounded or mushroom like (flat).
Adult female. Process between antennae without spicules, often clubbed. Head conical. Antennae close together. Macroducts usually 3-5 on each side of pygidium, thin, longer than wide, resembling microducts. Gland spines barely projecting from body margin. Gland tubercles nearly continuous along body margin from head to abdominal segment III. Microducts in medial areas of prepygidial segments dorsally and ventrally. Lateral margin of head with cluster of circular tubercles possibly representing eye.
Notes. There are a number of species with processes between the antennae; Wei et al. (2013) included 16 species in their key to the Fiorinia species from China. Only a few have an unusually elongate interantennal process and a conical head. Fiorinia proboscidaria resembles F. biakana Williams and Watson but differs by (characters in parentheses are those of F. biakana): space between median lobes less than width of median lobe (greater than width of lobe); macroducts ca. same width as gland spine ducts (wider than gland spine ducts); gland spines slightly protruding from derm surface (protruding at least half length of gland spine duct); gland tubercles continuous along body margin (grouped in clusters). This species also resembles F. turpiniae Takahashi but differs by (characters in parentheses are those of F. turpiniae): trilocular pores present near the anterior spiracle (absent); gland spines short, shorter than gland spine duct (long, longer than gland spine duct). Fiorinia proboscidaria differs from F. randiae Takahashi by (characters in parentheses are those of F. randiae): gland spines short, shorter than gland spine duct (long, longer than gland spine duct); median lobes nearly parallel (divergent). Florida specimens of F. proboscidaria are consistent with the description of Williams and Watson (1988) and Takagi (1970) except that both illustrated a lobe on each side of the head (Florida specimens lack these lobes), and that neither described the cluster of circular tubercles near the margin of the head or the ventromedial microducts anteriad of the pygidium. The illustration of Takagi has small lines at the end of the lobe on the side of the head which may be the same as the circular tubercles mentioned above, but they were not discussed in the description.
Second-instar female. Median lobes broad, as wide as or wider than medial lobule of second lobe, projecting ca. same amount or slightly less than medial lobule of second lobes. With four pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually rounded. With three large gland spines on margin of each side of body from abdominal segments II-IV; usually with small gland spine on each side of abdominal segments V and VI; without small gland spines on abdominal segment I. With three microducts on each side of head. Longitudinal line of microducts present submarginally on venter of abdominal segments II-VI, normally with 1-5 microducts on each side of each segment. Small lobular projections on anterior of head sometimes present. Cicatrix present on dorsal submargin of abdominal segment I.
First-instar nymph. Described in Howell (1977). Second-instar female. Median lobes broad, equal to or wider than width of medial lobule of second lobe, projecting ca. same amount or slightly less than medial lobule of second lobes. With four pairs of marginal macroducts. Swelling of body margin adjacent to macroduct usually rounded. With three large gland spines on margin of each side of body from abdominal segments II-IV; usually with small gland spine on each side of abdominal segments V and VI; without small gland spines on abdominal segment I. With three microducts on each side of head. Longitudinal line of microducts present submarginally on venter of abdominal segments II-VI, normally with 1-5 microducts on each side of each segment. Small lobular projections anteriorly on head sometimes present. Cicatrix present on dorsal submargin of abdominal segment I.
Notes. We have been unable to find characters that consistently separate secondinstar females of F. proboscidaria and F. theae.
Use of immature armored scales for identification is hampered by the fact that slide mounting protocols are tedious and laborious. Immature stages, especially firstinstar nymphs, are very small, ca. 0.1-0.2 mm in length, and can easily be lost during the mounting process. We reexamined previously published mounting protocols (McKenzie 1957;Wilkey 1990;Watson 2002) and addressed three issues: 1) avoiding specimen loss during mounting, 2) enhancing safety by reducing the amount of chemicals needed, since the reagents can be corrosive, flammable, carcinogenic, or produce toxic fumes, and 3) saving time if possible. Our comparative analysis of different slidemounting protocols and elaboration on their merits and drawbacks, especially for the incorporation of a mesh container during the slide-mounting protocols, enhance the potential for mounting immature armored scales.
One unexpected discovery during this project was that the morphology of secondinstar males was more reliable for species recognition than any other instar, including the adult female. For example, we were unable to distinguish between second-instar females of F. proboscidaria and F. theae, but their second-instar males were easily separated using the number of communal ducts. Second-instar males of F. fioriniae are remarkably different from the same instar of all other species of Fiorinia found in the USA even though other instars are quite similar to one another. Takagi (1975) discussed having difficulty separating F. nachiensis Takahashi and F. odaiensis Takagi based on adult females. At one point he treated them as synonyms, but based on major differences between the second-instar males he concluded that they were different species. Tippins (1970) published the first key and descriptions of the second-instar males of Fiorinia species and was surprised by the distinctive differences among species.
Recently, Liu et al. (2020) described a new species, F. yongxingensis from Hainan, China. It is similar to F. phantasma in the number and size of the marginal macroducts, the shape of the lobes, and the shape of the pygidium. The authors based their diagnosis in part on the detailed description and illustration of F. coronata (Williams & Watson, 1988), a junior synonym of F. phantasma (see Watson et al. 2015). Characters that appeared to be diagnostic for F. yongxingensis compared with F. coronata (= F. phantasma) are gland tubercles on the prothorax, microducts between the posterior spiracles, a gland spine on the prepygidium, and 0-3 pores near each anterior spiracle. Unfortunately, the type series of F. coronata did not contain the variation that we discovered in the Florida populations of F. phantasma. We have seen material with or without gland tubercles on the prothorax, a gland spine on the prepygidium, and 0-3 pores near each anterior spiracle. All specimens in the Florida populations have microducts between the posterior spiracles. Based on this information it appeared that the presence of these microducts was the key diagnostic character for F. yongxingensis. Because we needed to know the correct identity of the species introduced to Florida, several more steps were required. The next step was to examine the type series of F. phantasma and F. coronata. DL and JF borrowed the type specimens of F. phantasma from NHMUK and DRM examined another specimen from the type series deposited in USNM, but in each case the specimens were in such poor condition that it was impossible to see if microducts are present between the posterior spiracles. Type material of F. coronata also was studied; a type specimen deposited in the USNM has microducts between the posterior spiracles. A further step was to examine other relevant slides in the USNM. We studied slides from thirteen F. phantasma populations taken in quarantine from the Philippines, the type locality of F. phantasma, between 1965 and 1996, that are deposited in the USNM. We also examined slides taken in quarantine from Grenada, Hawaii, Thailand, Taiwan, and Vietnam. In all cases, microducts were present between the posterior spiracles, and there was overlapping variation in the other characters used to diagnose F. yongxingensis. We have yet to examine any adult female specimens of F. phantasma that lack these microducts and conclude that they are most likely a fixed character of the species.
The final step was to compare the results of multigene molecular analyses of the Florida population, the Chinese population, and two Malaysian populations (D1184 and D1185). The results clearly show that these populations are the same species. The morphological differences suggested as diagnostic of F. yongxingensis are within the range of variation that occurs in F. phantasma. Therefore, we here treat F. yongxingensis as a junior synonym of F. phantasma.
We obtained 37 5'-COI barcodes representing nine Fiorinia species in this study. Overall, low intraspecific genetic distances and high interspecific genetic distances ranging from 9.1% to 15.2% between Fiorinia species emphasize the reliability of 5'-COI barcodes in molecular diagnostics of armored scale species. Our rapid slidemounting protocol and the morphological keys to immatures and adults can provide time-and cost-effective diagnostics of Fiorinia species in the USA. However, for instances where specimens are damaged and cannot be mounted and where molecular diagnostics is the only option, barcodes will help to identify the species of Fiorinia. All of our DNA extractions are vouchered by permanently archived specimens in FSCA. This provides the opportunity for other researchers to validate the identifications of our specimens. We found an example of apparently misidentified specimens that were submitted to Genbank: the barcode of Aulacaspis rosarum Borchsenius (isolate wfsys017, accession number KP981086) was placed with 35 samples of Pseudaulacaspis cockerelli (Cooley) in our molecular analysis. A subtler discrepancy between DNA sequence and morphological identification, also seen in Normark et al. (2019), is the placement of F. vacciniae Kuwana (isolate D2453A, accession number KY219617) together with three samples of F. hymenanthis Takagi. Our study accentuates the importance of depositing morphological voucher specimen in an accessible collection.
Three populations of Fiorinia species (isolates D4674F, D4778A, and D4682A), collected from Lambir Hills National Park, Malaysia, September 26, 2013 from an undetermined host, identified as F. phantasma by BBN, were found to be genetically different from the F. phantasma populations from China, Florida and Malaysia. We reexamined the skins of the specimens used in our molecular analyses. The slides of isolates D4674F and D4778A are in poor condition and covered with fog, but we can see processes between the antennae and the shape of the pygidium, and they are consistent with the morphology of F. phantasma. The slide of isolate D4682A appears to have most characters of F. phantasma including the microducts between the posterior spiracles. This isolate is ca. 9% genetically distant from F. phantasma (based on COI) and is placed far from the subclade of F. phantasma (containing populations from China, Florida, and Malaysia) in the concatenated phylogenetic tree (Fig. 2). This may represent a cryptic species. More samples especially of second-instar males would help to confirm their identity.
Recently phylogenetic analyses in Normark et al. (2019) support the monophyly Fiorinia after the generic transfer of Ichthyaspis ficicola into the group. Our analysis agrees with the inference of Normark et al. (2019) with a few exceptions. Three Pseudaulacaspis MacGillivray species including P. cockerelli, P. pentagona (Targioni Tozzetti) and P. prunicola (Maskell) are placed in the same clade as Fiorinia in the case of the 5'-COI phylogenetic tree (Fig. 3, Suppl. material 1: Fig. S4). Likewise, in the case of the concatenated phylogenetic tree based on 28S, EF1-α, 5'-COI, 3'-COI, and COII, two samples of Fiorinia sp. (isolates D4815B, D4815C) fall out of the Fiorinia clade and placed with five Pseudaulacaspis species including P. biformis Takagi, P. cockerelli, P. momi (Kuwana), P. pentagona, and P. prunicola, with strong clade support (Fig. 2, Suppl. material 1: Fig. S2). Fiorinia was rendered polyphyletic by these two isolates (Fiorinia sp., D4815B and D4815C). They were collected from Malaysia in 2013 and determined as Fiorinia sp. by BBN. Given that the pupillarial habit has been gained and lost frequently in the history of Diaspididae (Normark et al. 2019), a second origin within Fioriniina would not be surprising. These two Fiorinia sp. isolates along with five Pseudaulacaspis species were placed together with strong support in a subclade within Fiorinia clade in our phylogenetic analysis based on 28S gene (Suppl. material 1: Fig. S3). Therefore, the placement of these two samples out of Fiorinia clade could be the result of an artifact of missing data or the methodology used in multigene tree and would require additional analysis for further confirmation. There are two samples of Lineaspis striata (Newstead) with P. simplex Takagi in the sister subclade that joins the subclade of Pseudaulacaspis/Fiorinia with strong clade support (Fig. 3, Suppl. material 1: Fig. S4). Overall, the main clade of the genus Fiorinia joins the Fiorinia/ Pseudaulacaspis/ Lineaspis clade with strong clade support (> 90%). Borchsenius (1966) separated Fiorinia from Pseudaulacaspis and placed them in different tribes due to their pupillarial habit. However, Takagi (1969) and Howell and Tippins (1973), based on the presence of communal ducts, suggested a relationship between Pseudaulacaspis and Fiorinia. Our phylogenetic analysis suggests that additional sampling of Fiorinia and Pseudaulacaspis from Asia will further clarify the monophyly of the genus Fiorinia.
Field habitus of adult females, especially the character of the overlap between the first-instar and second-instar exuviae, was used for the first time in this study. For example, in the case of F. externa, the first-instar exuviae are barely touching the second-instar exuviae and form a distinct indentation between the attachment of the first-and second-instar exuviae (Suppl. material 1: Fig. S1). In contrast to this, no indentation was observed in F. phantasma. In addition, we also compared the color and shape of the second-instar nymphs shed skins of Fiorinia species. Field habitus can assist growers and nursery workers in making preliminary identifications.
Fiorinia japonica was eradicated from California and has been rediscovered three times since its first report in 1910 (Watson 2009). The most recent reinfestation was observed in 2008 and was most likely eradicated in a subsequent year (Watson 2009). Our collaborator's attempt to collect fresh specimens of F. japonica in California for inclusion in this study was unsuccessful and its population has not been barcoded. It would be useful to trace its population in other states and to sequence its barcode. We also intended to include the population of F. phantasma from Hawaii, but efforts of our collaborators to collect it from Hawaii were unsuccessful. There have been at least two reinfestations of F. phantasma in Hawaii since its first report in 2004. The most recent heavy infestation was from palms reported in 2011 (Garcia 2011). Interestingly, in this most recent Hawaiian infestation, the second-instar nymph's shed cuticles had transverse brown stripes, whereas the Florida population lacks this character. It would be helpful to collect F. phantasma from Hawaii and to compare it with the Floridian F. phantasma population to determine if they are the same species. If the Hawaiian F. phantasma is the same as the Floridian species, that fact might imply that F. phantasma in Florida could follow the same pattern as it did in Hawaii and keep reappearing with heavier infestations in subsequent years. This study will facilitate regulatory and pest management decisions by enhancing morphological and molecular identification of seven adventive Fiorinia species occurring in the USA.

Conclusions
There are six main conclusions of our study. 1) The utilization of molecular barcodes is highly beneficial in diagnosing species of Fiorinia that occur in the USA. 2) The new keys in this study demonstrate that the USA species of Fiorinia can be identified using immature specimens. 3) Second-instar male morphology provided a reliable suite of characters for species-level identification. 4) Based on our comparative analysis of morphological characters and multigene molecular sequencing of specimens of F. phantasma and F. yongxinensis, it is clear that the latter is a junior synonym. 5) Of the different protocols tested for mounting immature specimens of Fiorinia, Hoyer's mounting medium was the best for discerning delicate morphological characters but it was not desirable for permanent slide preparations. Balsam was the best for permanent mounts but did not provide the morphological clarity of Hoyer's mounts. 6) The use of a mesh container in the process of mounting immatures is an effective method for preventing the loss of specimens. Overall, the use of the morphological and molecular data provides effective methods for early detection of new infestations and assists regulators in making control decisions.