A cladistic analysis of whiteflies, subfamily Aleyrodinae (Hemiptera: Sternorrhyncha: Aleyrodidae)

Relationships within the Aleyrodinae are investigated phylogenetically through a cladistic analysis of 94 (70 binary and 24 multistate) morphological characters derived from the pupal case. This is the first attempt to construct a phylogenetic classification at genus level on a world basis for the subfamily Aleyrodinae. More than 430 species, about one third of all described species, were examined. These belonged to 124 genera, of which 117 were aleyrodines (including two undescribed genera) and seven were out‐groups from the subfamily Aleurodicinae (Hemiptera: Aleyrodidae). Maximum parsimony analysis yielded more than 30 000 most‐parsimonious trees (length = 2730, RI = 0.672, CI = 0.137). Based upon the strict consensus tree, only 33 of the currently accepted genera, for which we had multiple representatives, were found to be monophyletic (44 genera of the 117 examined genera were monobasic and/or only one species was available for study). Monophyly of some economically important genera, e.g. Aleurolobus, Bemisia and Dialeurodes, as well as that of most previously proposed tribes, was not supported. No parallel diversification of whiteflies with host plants was detected, mostly because of within‐species host switching. Problems in determining generic level characters and applicability of tribal classification for this subfamily are discussed. One new genus, Pseudozaphanera Manzari gen. n., and seven new combinations, Aleuroclava lanceolata (Takahashi) comb. n., Crescentaleyrodes fumipennis (Hempel) comb. n., Pseudozaphanera niger (Maskell) comb. n., Pseudozaphanera papyrocarpae (Martin) comb. n., Pseudozaphanera rhachisreticulata (Martin) comb. n., Pseudozaphanera splendida (Martin) comb. n., Pseudozaphanera wariensis (Martin) comb. n., are proposed as well as three new generic synonyms; Taiwanaleyrodes Takahashi is placed in synonymy with Aleuroclava Singh syn. n., Hempelia Sampson & Drews with Aleurothrixus Quaintance & Baker syn. n. and Rositaleyrodes Meganathan & David with Aleurolobus Quaintance & Baker syn. n.


Introduction
The Aleyrodidae is composed of tiny insects (1-3 mm in body length), usually called whiteflies, though one of the most widespread pest species, Aleurocanthus woglumi Ashby, is called the citrus blackfly (Mound 1973). Mound and Halsey (1978) catalogued the whiteflies of the world and listed 1156 species, which has now increased to approximately 1450 (Martin et al. 2000). However, the real number is certainly much higher.
Many whiteflies are agricultural pests, especially in the tropics and subtropics (Calvert et al. 2001). Feeding activity damages their host plants and some species are known to transmit diseases, especially those caused by viruses (Gullan and Martin 2003). In spite of this, the family is poorly studied taxonomically, particularly in a phylogenetic context. Almost all publications dealing with whitefly taxonomy are restricted to alpha taxonomy. Major publications provide lists of species and descriptions of new species and/or the genera, mostly with restricted geographical compass (e.g. Singh 1931;Takahashi 1932;Corbett 1935;Russell 1948;Mound 1965;Cohic 1966;David and Subramaniam 1976;Bink-Moenen 1983;Jesudasan and David 1991;Martin 1999).
The taxonomy of the Aleyrodidae has long been problematic. Unusually amongst insects, whitefly generic classification is largely based on one of the immature stages, the so-called, pupal case (5puparium5fourth instar5last nymphal stage5last larval stage), rather than on the adults (Gill 1990). Quite often, third instar larvae have been used in species descriptions, having been mistaken for the so-called puparial stage (Martin 1999). Unfortunately, whitefly puparia are notorious for displaying variation induced by environmental and physical factors, such as temperature and humidity (Mohanty and Basu 1986), as well as by the type of leaf surface upon which they have developed (Russell 1948;Mound 1963;David and Ananthakrishnan 1976;Neal and Bentz 1999;Guershon and Gerling 2001). Their generic classification was largely laid out by Baker (1913, 1914) and is based mainly on pupal characters; the morphological traits of adults are currently poorly understood and so do not readily permit differentiation between genera or species (Frohlich et al. 1999).
Apart from a few genealogical diagrams proposed (e.g. Quaintance and Baker (1913) (Figure 1); Bondar 1923), the first attempt to construct a phylogenetic classification for whiteflies using morphological characters was made by Jensen (1999) for the species of Dialeurodes Cockerell. In Jensen's article, alternative generic names were not proposed despite the fact that his results showed the genus to be paraphyletic with respect to several others. Subsequently, Jensen (2001) proposed the use of three pre-existing generic names, Dialeurodes, Massilieurodes Goux and Singhiella Sampson, for many of the Dialeurodes of the world based on the placement of the type-species of these genera in the cladograms or based on the examination of original descriptions.
Phylogenetic analysis of 18S rDNA nucleotide sequences of hemipteran exemplars, shows the Sternorrhyncha appears to be monophyletic forming a sister group to all other hemipterans (the Euhemiptera, including Auchenorrhyncha and Heteroptera) (Campbell et al. 1994(Campbell et al. , 1995a(Campbell et al. , 1995bvon Dohlen and Moran 1995). However, relationships within the Sternorrhyncha (which includes psyllids, whiteflies, aphids and scales) are controversial with molecular and morphological data supporting different scenarios. Based upon the same molecular evidence, whiteflies form a sister group to aphids and scales, while psyllids form a sister group to all other Sternorrhyncha (Campbell et al. 1994(Campbell et al. , 1995a. However, some morphological evidence supports a sister-group relationship between whiteflies and psyllids (e.g. Quaintance and Baker 1913;Goodchild 1966;Hennig 1981;Shcherbakov 2000;Shcherbakov and Popov 2002), though Evans' (1963) morphologically based conclusions are more in agreement with the available molecular studies. Some molecular studies of their primary bacterial endosymbionts (P-endosymbionts) generally concur with the available morphological evidence, with an analysis of 16S rDNA symbiont sequences supporting a sister relationship between primary endosymbionts of psyllids and whiteflies von Dohlen 1998, 2001).
Efforts to determine the phylogenetic origin of whiteflies have been further impeded by gaps in the hemipteran fossil record, and especially a paucity of fossil whiteflies (Campbell et al. 1994). Owing to their size and fragility, adult aleyrodids would not be expected to be preserved as impression fossils often, and still less, their larvae (Evans 1963). The possibility that they arose in tropical latitudes may also account for their absence in fossil deposits now located in northern latitudes, where fossils of aphids are often found (Shcherbakov and Wegierek 1991). The oldest known fossil whiteflies, Juleyrodes gilli Shcherbakov and J. visnyai Shcherbakov, belong to an extinct subfamily, Bernaeinae, and are from the Late Jurassic or Early Cretaceous (more than 140 myr ago) (Shcherbakov 2000).
Monophyly of the Aleyrodoidea, and consequently of its single included family, Aleyrodidae, is well-supported by synapomorphies (Shcherbakov and Popov 2002). Five subfamilies have been established at various times within the extant Aleyrodidae, of which only two are now considered valid: the Aleurodicinae, found mainly in Central and South America, and the Aleyrodinae, which are more widespread (Mound and Halsey 1978;Gill 1990). The Udamoselinae was proposed on the basis of a single specimen of a South Figure 1. Genealogical diagram of the Aleyrodidae; after Quaintance and Baker (1913). polymorphisms. Maximum parsimony analysis (MP) was carried out using PAUP*, version 4.0b10 (Swofford 1998). Heuristic searches were carried out with 1000 random additions followed by branch swapping using tree-bisection-reconnection (TBR) holding a single tree (NCHUCK51, CHUCKSCORE51) (Morris et al. 1999;Sallum et al. 2000;Hall and Harvey 2001;Skevington and Yeates 2001;Vardal et al. 2002). The resulting tree(s) were used for a final round of branch swapping with maxtrees 30 000. The new strategy for estimating large phylogenies (changing the landscape) was employed to help find more parsimonious trees (Quicke et al. 2001). The approach was used with maximum and minimum retention index (RI), and maximum and minimum consistency index (CI).
The bootstrap support for individual branches was estimated using 100 pseudoreplicates each of 100 random additions (Felsenstein 1985), though this will be very conservative due Figure 2. Stylized whitefly puparium with major morphological features annotated (from Martin 1987). to the difficult nature of this data set (e.g. Gauthier et al. 2000). Parsimony Jackknifing with 36.79% deletion, i.e. resampling 63.21% of characters, proposed by Farris et al. (1996) was also carried out to find support for branches as well as finding Bremer support (Bremer 1994) for some clades. To carry out Bremer support using PAUP*, a constraint tree description for the taxa of the clade in question was created and a search then performed by enforcing the topological constraint, but with the programme to only search for trees that were NOT compatible with that constraint. As a result, if the node in question is strongly supported, the best trees found that do not contain that node, i.e. are not compatible with the constraint, will be at least one step longer than the most parsimonious tree found from the initial unconstraint analysis. This discrepancy in tree length is the decay index or Bremer support value.
Additionally, successive approximation weighting (SAW) (Farris 1969) was applied for unweighted analysis using each of the four above-mentioned indices to increase resolution. Analyses were performed constraining some traditional whitefly genera to be monophyletic, and the resulting trees compared with the initial unconstrained ones using the Templeton (Wilcoxon sined-ranks) test implemented in PAUP*. The same procedure was used to find most-parsimonious trees (MPTs) for constraint analysis, i.e. heuristic searches with 1000 random additions, and employing the new strategy. Characters were traced on to cladograms using MacClade 4.0 (Maddison and Maddison 2000) and were optimised using the accelerated transformation algorithm (AccTran). In all cases, the obtained trees were rooted using the out-group taxa.

Characters and character coding using in the analysis
Pupal case [1] Pupal case margin: (0) smooth ( Figure 3A); (1) undulate and/or indistinctly indentate ( Figure 3B); (2) distinctly toothed ( Figure 3C-L). Comments: in all cases, the word 'margin' is used to mean the real margin rather than the apparent margin, which may be submarginal in the case of marginal recurvature (see character 3). In other words, the real margin was used as a criterion to score the relevant characters. [2] If pupal case margin distinctly toothed: (0) crenate ( Figure 3C); (1) dentate ( Figure 3D); (2) lobulate ( Figure 3E); (3) truncate-lobulate ( Figure 3F, G); (4) serrate ( Figure 3H); (5) each tooth with apical notch, which gives the teeth the appearance of being twinned and sometimes being triangular ( Figure 3I-K); (6) closeset lanceolate processes, each arising from a blunt marginal tooth ( Figure 3L). [3] Recurvature of margin: (0) margin not recurved; (1) margin partially recurved ventrally; (2) margin completely recurved ventrally. Comments: when the ventral surface of pupal case is much smaller than the dorsal surface the margin is evenly reflexed, a part of the submargin is recurved or the whole submargin is recurved. In some species the margin may easily be recurved by the mounting process and can cause misinterpretation when the character is coded.
[16] Length of caudal setae: (0) distinctly shorter than vasiform orifice; (1) approximately as long as vasiform orifice; (2) distinctly longer than vasiform orifice. Comments: the length of caudal setae shows both interspecific and intraspecific variation in relation to the vasiform orifice, but appears to be fixed for the species of some genera.
[20] Setae of dorsal disc area (except those in characters 6, 8, 9, 10 and 12): (0) hairlike, short or long ( Figure 5A); (1) stout, apical acute and/or rounded, short or long ( Figure 5B); (2) somewhat swollen at base and pointed apically (lanceolate) ( Figure 5C); (3) capitate ( Figure 5D); (4) apical bended ( Figure 5E); (5) laciniate ( Figure 5F); (6) very stout, apical truncate-capitate ( Figure 5G); (7) almost Tshaped ( Figure 5H); (8) almost arrow-shaped ( Figure 5I). Comments: the length and position of setae as well as their number is variable within some species, e.g. Aleurolobus gruveli Cohic (Bink-Moenen 1983). Except for a few genera, for which the position and number of setae are characteristic, e.g. Corbettia Dozier, their number and position have not usually been used as diagnostic characters at genus level (unlike for the aleurodicine genera). This is most probably due to variability so that even for a given number of setae for a species, it seems they are submarginal when small, but are located more subdorsally when they are enlarged. Therefore, the number of setae and their positions (except being on submargin or dorsal disc area) were not coded as separate characters. In the case of lack of submarginal line (see character 28), it is sometimes very difficult to assign setae as being submarginal or subdorsal when they are located far from the margin. In such cases, other features such as submarginal ridges radiating from the margin, colour differences or different ornamentations depending on the examined taxa were used to help define the submarginal limit. It is worth mentioning that according to Bink-Moenen (1983), the division of the dorsal surface into submargin, subdorsum, submedian and median areas (the three latter areas all together is called dorsal disc area) is artificial, as can be seen from the variable position of some setae in many species. Even when the submargin is separated by a furrow, the same setae may occur as often on the submargin as on the subdorsum.
[21] Submarginal spines: (0) absent; (1) present. Comments: there is almost no distinct boundary between setae and spines in some cases in the literature and the same feature has been defined as a seta and/or a spine by different authors. In general, if there is no circle at the base of this character indicating a socket, it is considered as a spine. The length, position and number of spines are variable as mentioned above for the setae.
[26] Shape of transverse moulting suture: (0) like Figure 6A; (1) not bent near 90u in middle of each side ( Figure 6B); (2) curves anteriorly at the junction of longitudinal moulting suture and posteriorly at the middle of each side ( Figure 6C); (3) like Figure 6D; (4) each side curves anteriorly ( Figure 6E); (5) like Figure 6F; (6) like Figure 6G; (7) each side bends to anterior and meets in the midline ( Figure 6H); (8) each side curves anteriorly in the form of a pear together with a pair of membranous furrows ( Figure 6I); (9) each sides curves anteriorly in the form of an almost triangular together with a pair of membranous furrows ( Figure 6J). Comments: the shape of transverse moulting suture is quite variable and the 10 shapes chosen above were almost the main shapes and other intermediate states were assignable to one the main ones.
[27] Longitudinal moulting suture: (0) not reaching margin of pupal case; (1) reaching margin of pupal case. Comments: only in one species studied, Viennotaleyrodes curvisetosus Martin, the longitudinal moulting suture was absent. For this species the character was coded as a question mark similar to those few species that the distal extremity of the suture was not clearly discernible due to slide quality.
[28] Submarginal line or fold, which separates submargin from the dorsal disc area: (0) absent; (1) present. [29] If submarginal line or fold present: (0) parallel and longitudinal or almost bracketshaped, not meeting at both anterior and posterior; (1) not parallel and longitudinal, meeting at both anterior and posterior, concentric with margin or nearly so; (2) like state 1, but with breaks at eyespot regions; (3) like state 1, but interrupted at caudal area; (4) like state 1, but with breaks at eyespot regions and interrupted at caudal area. Comments: in the literature, the shape of submarginal line or fold has sometimes been cited but almost never with clear diagnostic states separated or compared between taxa. [30] Extra submarginal lines encircling the pupal case: (0) absent; (1) present.
Cockerelliella psidii (Corbett), the cephalothoracic fold joins to the distal extremities of the transverse and longitudinal moulting sutures and forms a ''trapdoor''. This feature is similar to character 26, state 7 but in the latter, each side of the transverse moulting suture curves anteriorly and is connected with the longitudinal moulting suture.

Vasiform orifice
[34] Vasiform orifice shape: (0) semicircular; (1) nearly circular ( Figure 7A); (2) nearly triangular ( Figure 7E Figure 7D). Comments: Maskell (1895) was the first to describe and name the vasiform orifice and its components the lingula, operculum and lingular setae. The shape of vasiform orifice is very variable and the same feature has been defined using different terms and/or short sentences by different authors. The shapes chosen in the characters 34-36 were the main shapes and other intermediate states were assignable to one of these states. The shape of vasiform orifice, at least in some cases, may be subjective but it has been used in many generic and species descriptions and so it has been included here and we have been as objective as possible in differentiating the states.
[39] Vasiform orifice posterior notch: (0) absent; (1) present. Comments: although the shape of notch also shows variation, this variation appears to be continuous rather than discrete. As far it is known from the literature, with the exception of Jensen's articles (1999,2001), the shape of vasiform orifice notch has never been used as a diagnostic character, at least at genus level. [40] Position of vasiform orifice: (0) inset from posterior margin of pupal case by less than its own length; (1) inset from posterior margin of pupal case by about its own length; (2) inset from posterior margin of pupal case by distinctly more than its own length (sometimes two or more times). [41] Operculum size: (0) occupying about 0.25 of orifice; (1) occupying about 0.5 of orifice; (2) occupying about 0.75 of orifice; (3) occupying about whole of orifice. Comments: there are two types of vasiform orifice: the orifice is nearly or completely filled by the operculum and the lingula is normally not visible in dorsal view, and in the other type, the operculum does not fill the posterior part of the orifice and the lingula is usually visible (Gill 1990). Here, these two main states were divided into four states so as to cover all those used by various authors in generic and/or species descriptions to date. [42] Operculum shape: (0) transversely elliptical ( Figure 8A); (1) nearly trapezoidal, rounded at angles ( Figure 8B); (2) nearly rectangular, wide about twice as long as length, posterior angles toothed and/or slightly indentate ( Figure 8C); (3) subcordate ( Figure 8D); (4) semicircular ( Figure 8E); (5) rectangular, convex-sided, hind margin straight and/or slightly curving anteriorly ( Figure 8F); (6) rectangular, convex-sided, hind margin with a wide projection ( Figure 8G); (7) cordate or slightly elongatecordate ( Figure 8H); (8) rectangular shaped pushed inward at posterior margin ( Figure 8I); (9) squarish, convex-sided, narrowed posteriorly ( Figure 8J). Comments: as with the vasiform orifice, the shape of the operculum is also variable and the same feature has been defined using different terms and/or short sentences used by different authors (see comments for character 34). This character has not been used as frequently as the shape of vasiform orifice in species or generic descriptions. [43] Opercular setae: (0) absent at posterior margin of the operculum; (1) present at posterior margin of the operculum (two setae). [44] Head of lingula: (0) completely obscured by opaque operculum, not visible; (1) completely obscured by transparent operculum, visible; (2) incompletely obscured by operculum, partly visible; (3) not obscured by operculum. [46] If head of lingula elongate: (0) lanceolate ( Figure 9H); (1) cylindrical, rounded posteriorly ( Figure 9I); (2) large and almost conical ( Figure 9J).
[48] Lingula: (0) not extending beyond posterior margin of vasiform orifice; (1) extending beyond posterior margin of vasiform orifice. Comments: sometimes the lingula may be protruded by the mounting process and can cause misinterpretation this character. It should be noted that in some species the length of lingula is longer than vasiform orifice but it folds inside the orifice. To avoid any error in determining its true length in such species, the normal state of lingula was chosen, i.e. the character was coded as having the first state (not extending beyond posterior margin of vasiform orifice). [49] Lingular setae: (0) absent or not apparent; (1) present. (C) nearly rectangular, wide about twice as long as length, posterior angles toothed and/or slightly indentate; (D) subcordate; (E) semicircular; (F) rectangular, convex-sided, hind margin slightly curves anteriorly and/or straight; (G) rectangular, convex-sided, hind margin with a wide projection; (H) cordate or slightly elongate-cordate; (I) rectangular shaped pushed inward at posterior margin; (J) squarish, convex-sided, narrowed posteriorly.
[50] Number of lingular setae: (0) two (if two more setae are present, these are minute and often visible only at 61000 magnification with a light microscope); (1) four. Comments: the lingula is usually spinulose and the number of conspicuous setae generally varies from without any seta to having four setae. In the Aleurodicinae, some of whose members were used as out-groups, the lingula almost always has four conspicuous setae. In the Aleyrodinae, only two apical setae have usually been mentioned in original descriptions. Russell (1947) was the first to report the presence of two more small lateral lingular setae (for some species of Trialeurodini that she described as a new tribe including three genera (see Appendix A)). In this study, examination of thousands specimens belonging to about one third of all described whitefly species at 61000 magnification indicates that with the exception of lack of lingular setae in some species, these two small lateral setae are also present in other genera and quite visible when the lingula is in a right position. It seems that they have been neglected, even by Russell in her other publications, because they are small and difficult to see at least at 6400. However, since these two small lingular setae somehow differ from those of the Aleurodicinae, the character was included here.
Comments: this character was only observed in one monobasic genus, Gagudjuia Martin. In the future the character may be seen in other species, so it was added to the character system to help for further studies, though in this study it is parsimonyuninformative. [55] Caudal furrow: (0) absent; (1) present. [56] Caudal furrow: (0) narrow, half-width or less than half-width of vasiform orifice, sometimes suture-like; (1) wide at base (near vasiform orifice), the rest part narrow; (2) wide, more than half-width of vasiform orifice; (3) narrow at base, the rest part wide. [57] Caudal ridges: (0) absent; (1) present.
Abdominal segments [58] Lengths of seventh abdominal segment: (0) length of abdominal segment VII not significantly (half-length or less than half-length of segment VI) reduced medially; (1) length of abdominal segment VII significantly reduced medially. Comments: the length of abdominal segment VII, comparing to abdominal segment VI, is quite variable but the reduction of its length medially has been used as a diagnostic character for some genera, e.g. Bemisia and Pealius Quaintance & Baker, without clearly defining that how much reduction is significant. To make it clear and practical to score the character and subsequently to justify its validity as being a diagnostic character, the above-mentioned criterion was used. [59] Submedian abdominal depressions: (0) absent; (1) present.
Papillae-tubercles-glands-pores [65] Submarginal papillae: (0) absent; (1) present ( Figure 11B). Comments: there is no clear distinction between papillae and tubercles in most cases in the literature. Their Figure 10. Different forms of the tracheal pore. (A) Not forming a distinct pore, margin slightly curved inwardly/ outwardly or thickened (arrowed); (B) somewhat invaginated and with slightly thickened cuticle; (C) crenulated to a lesser degree than the rest of margin (arrowed); (D)-(F) a comb of modified teeth, sometimes only two or three teeth; (G, H) C-shape pore without teeth; (I, J) C-shape with teeth; (K) deeply emarginate and covered by a smooth rounded lobe; (L) gland spines forming a comb.
presence, size and structure are very variable even within a species. Therefore, only the presence and absence of these characters on the submargin (this character and character 66) or on the dorsal disc area (characters 74 and 75) was scored. Compared with papillae, tubercles are more shapeless, usually irregularly distributed and have no associated pore. According to J. H. Martin (BMNH, personal communication), tubercles are often environmentally induced characters. [66] Submarginal tubercles (excluding the tubercle between caudal setae, when is present, and those at the bases of the larger setae): (0) absent; (1) present. Comments: see character 65. [67] Submarginal glands: (0) absent; (1) present ( Figure 11C, D). Comments: these structures were named by Russell (1958), thought they have also been called papillae, pores, or tubercles by other writers. [68] Shape of submarginal glands: (0) tips of glands smooth ( Figure 11C); (1) tips of glands dentate ( Figure 11D). Comments: the tips of the glands in the ten species described/redescribed by Russell (1958) were dentate. Later, Bink-Moenen (1983) described such glands for the monobasic genus, Arachnaleyrodes Bink-Moenen, but in that case their tips were smooth, and was used as one of the diagnostic character for the genus. Glands with smooth tip were only observed in one taxon, so this parsimony-uninformative character was automatically ignored by the analyses. [69] Submarginal simple pores: (0) absent; (1) present. [70] Submarginal geminate pore/porettes: (0) absent; (1) present ( Figure 11E).
Comments: these structures are also called combined pores and sometimes pores and porettes are widely spaced from each other. According to Russell (1948) and Bink-Moenen (1983), their distribution is of some value at the species level.
Comments: the same as for the character 70. [73] Large subdorsal simple pores: (0) absent; (1) present ( Figure 11F). Comments: these structures are usually a generic level diagnostic feature but at least in Dialeuropora decempuncta (Quaintance & Baker) these large simple pores are sometimes absent or much reduced in size. It is considered likely that this is a modification arising from environmental factors which remain obscure (Martin 1999). [74] Subdorsal papillae: (0) absent; (1) present. Comments: the same as for character 65. [75] Tubercles of dorsal disc area (except the tubercles mentioned in the character 60): (0) absent; (1) present. Comments: the same as for character 65. [76] Modified papillae of dorsal disc area, which are slightly elevated, nearly flat and platelike: (0) absent; (1) present ( Figure 11G). [77] Eyespots: (0) absent; (1) present ( Figure 11H). Comments: according to Bink-Moenen (1983), the presence of eyespots can be of taxonomic value at the species level, but presence and expression are often variable interspecifically. Both states were scored for those species that at least some individuals with them.
[84] Subdorsal plate with one or two conical pores connected with tracheal pore area: (0) absent; (1) present ( Figure 12G). [85] A row of scallop-shaped thickenings along the longitudinal subdorsal furrow: (0) absent; (1) present ( Figure 12H). [86] Prominent glandular areas with polygonal reticulate pattern, sometimes resembling compound eyes: (0) absent; (1) present ( Figure 12I). [87] Submarginal crescent-shaped pores: (0) absent; (1) present ( Figure 12J). [88] Submarginal wax-plat cluster: (0) absent; (1) present ( Figure 12K). Legs and antennae [89] Apical adhesion pad of legs: (0) absent; (1) present. [90] Tarsal claw: (0) absent; (1) present. [91] Basal mid-and hind leg seta: (0) absent; (1) present. Comments: this character has sometimes been used in species and/or generic descriptions, especially whenever the seta(e) have been long enough to be easily recognisable. In some cases, there is no clear distinction between setae and spines (character 92) and, apparently, the terms have been used synonymously by some authors. For example, for Acaudaleyrodes africana (Dozier), Mound (1965) stated ''A. africanus can be distinguished from the other species in Acaudaleyrodes by the absence of setae from the base of legs, ...''. Based on the criterion given in the comments for the character 21, what are absent from the base of legs in this species are spines not setae. Here, examination of thousands of specimens at 61000 magnification, clearly showed that at least one microseta is always present at the base of each mid-and hind leg. Sometimes one or rarely two or more setae can be seen in some species but because of slide quality it was almost impossible to count them precisely for each species for such a large-scale study. It seems that the existence of extra seta(e) at most can be important at species level, but it is sometimes variable even for the same species. This character was automatically ignored due to being always present, i.e. a constant character. [92] Basal mid-and hind leg spine: (0) absent; (1) present ( Figure 12L). [93] Antenna length: (0) limited to front legs or slightly more; (1) reaching middle of mid legs or more. Comments: sexual dimorphism occurs in some genera and species as a difference in the length of the antenna. In this case, the antennae of males are comparatively longer than the antennae of females. Such species were scored as polymorphic. [94] Position of antenna relative to the prothoracic legs: (0) mesal to the prothoracic legs; (1) lateral to the prothoracic legs.

Parsimony analysis
Maximum parsimony holding a single tree yielded one MPT (length52759, RI50.668, CI50.136). This shortest tree was hit only once during the 1000 random additions, showing that this is a rather difficult data set to analyse. Branch swapping with maxtrees set at 30 000 using this tree as the starting tree gave 30 000 equally MPTs with a length of 2743 (RI50.671, CI50.137). Application of the new strategy (changing the landscape) for these trees using maximum RI to weight characters found 30 000 (limited by computer memory) equally MPTs with a length of 2730 (RI50.672, CI50.137) (employing the same strategy using maximum CI, minimum CI or minimum RI to weight characters found no shorter trees). The strict consensus tree of these is shown in Figure 13. Although the strict consensus tree contains several monophyletic clades, the relationships among most of these are unresolved. Ignoring 44 of the 117 examined genera that are monobasic and/or only one species were available for study, the following 33 genera (the author's names of all genera are given in Appendix C) were found to be monophyletic: Acanthaleyrodes, Acaudaleyrodes, Africaleurodes, Aleurocerus, Aleurocybotus, Aleuroduplidens, Aleuroglandulus, Aleuroparadoxus, Aleuropteridis, Bellitudo, Combesaleyrodes, Corbettia, Dialeuropora, Fascaleyrodes, Filicaleyrodes, Fippataleyrodes, Indoaleyrodes, Leucopogonella, Metabemisia,   excluding S. vigintiseta Martin, and most of the species of the genus Dialeurolonga including the type-species as well as those species of Crenidorsum described by Russell (1945), including the type-species formed monophyletic clades. None of the mentioned excluded species are the type-species of their genera.

Successive approximation weighting
Applying SAW gave more resolution for unresolved nodes of the tree. More than 10% of the 30 000 MPTs obtained from the unweighted analysis (UW-MPTs) were selected so that to have at least 100 representatives from each 1000 MPTs. Then, the approach was used with four different indices, i.e. maximum RI, maximum CI, minimum RI, and minimum CI to weight characters, and four, eight, four, and eight cycles of reweighting were needed, respectively, before stability was reached. When all characters were set back to unit weight, the lengths of the trees were 2733, 2953, 2734 and 2927, respectively. Comparing these lengths with the length of UW-MPTs, i.e. 2730, indicates that the trees obtained from applying SAW using maximum and minimum RI (MPTs-SAW-MaxRI and MPTs-SAW-MinRI) are only three and four steps longer, respectively, and here the strict consensus tree of these trees (MPTs-SAW-MaxRI) is presented ( Figure 14) and discussed. It is worth mentioning that the topology of the strict consensus trees of MPTs-SAW-MaxRI and MPTs-SAW-MinRI was almost the same as each other, and both were more in agreement with the strict consensus tree of UW-MPTs (Table I). The strict consensus trees obtained from applying SAW using maximum and minimum CI (MPTs-SAW-MaxCI and MPTs-SAW-MinCI) showed two different topologies and were somewhat less in agreement with the strict consensus tree of UW-MPTs (Table I).
In the strict consensus tree of MPTs-SAW-MaxRI ( Figure 14), of the 33 genera recovered as monophyletic in the initial analyses, two, viz. Dialeuropora and Pogonaleyrodes, were no longer recovered, but the former excluding D. silvarum (Corbett) (not the typespecies) did form a monophyletic clade. However, Pectinaleyrodes was monophyletic in this tree, whereas it was not in the initial tree. The monophyly of other mentioned genera were the same.

Monophyly of large genera
The evidence against monophyly of two economically important genera, Bemisia and Trialeurodes, as well as several large genera, i.e. Aleuroplatus, Aleurotrachelus, Dialeurodes, and Tetraleurodes, which were not recovered to be monophletic in either unweighted or SAW analyses was investigated. Separate analyses were performed constraining each of these to be monophyletic, and the resulted trees compared statistically with the initial unconstrained ones using the non-parametric ranked-sign test of Templeton.
Constraining Bemisia to be monphyletic resulted in one MPT (length52772) and using the new tree searching strategy yielded more than 1000 MPTs (length52756). The Templeton test showed that there were significantly longer than the best unconstrained ones (P50.0196-0.0419) (Table II). Constraining Dialeurodes to be monophyletic yielded trees of length 2771 which were not different from unconstrained ones based on the Templeton test (P.0.07).
In all constraint analyses, only one MPT found after performing a heuristic search with 1000 random additions, which were significantly different from the UW-MPTs except for Bemisia and Trialeurodes. The final results changed for some genera after finding shortest Figure 14. Strict consensus tree of more than 10 000 MPTs-SAW-MaxRI. possible trees applying the new strategy (at least 500 trees were compared), which are shown in Table II. Generally, the monophyly of Aleurotrachelus and Bemisia was statistically rejected, while that of Aleuroplatus, Dialeurodes, Tetraleurodes, and Trialeurodes was not (Table II). These results will be discussed in detail later.
The finding that statistical significance can increase when difference in tree length decreases (as a result of finding most parsimonious trees) indicates that it is important to reduce the variance, i.e. noise, in the estimate of numbers of character steps, and the test is only meaningful if the shortest possible trees to be compared. If they are not the MPTs, there will be a degree of ''randomness'' in the character lengths and such random variation might occasionally be on the side of increasing significance, sometimes not.

Evaluating the current tribe classification
As already mentioned, 13 tribes have been proposed so far within the Aleyrodinae. In addition to the problems posed by unplaced genera, there is little agreement about the composition of some proposed tribes (Sampson 1943(Sampson , 1947Russell 1947;Takahashi 1954;David 1990). Before starting to evaluate the current tribe classification, it should be mentioned that some of the genera in Appendix A have been synonymized and these will be indicated in the discussion of each tribe. Furthermore, except the tribe proposed by Russell (1947), Trialeurodini, the other tribes were only defined in identification keys by the authors, who erected them (see above). Tribal level characters used by these authors for each tribe are given in Appendix B to avoid repeating them when each tribe is discussed.

Tribe Aleurocanthini
Three genera, viz. Aleurocanthus, Aleurotrachelus, and Pentaleyrodes, were placed in the Aleurocanthini by Takahashi (1954), and only Aleurocanthus by David (1990) for the whiteflies of Japan and India, respectively (Appendix A). As mentioned before, the latter excluding A. ceracroceus was found to be monophyletic and the monophyly of Aleurotrachelus even statistically rejected (Table II). However, the type-species of Aleurotrachelus, A. tracheifer (Quaintance), was never recovered in the same clade either with Aleurocanthus or Pentaleyrodes in the strict consensus tree of UW-MPTs ( Figure 13), whereas in the strict consensus tree of MPTs-SAW-MaxRI ( Figure 14) it was found in a different clade from the Aleurocanthus clade within a big clade. Sampson (1943) proposed this tribe on the basis of a single genus, Aleurochiton (Appendix A). Two of the four studied species of this genus, A. acerinus Haupt and A. aceris (Modeer) (the type-species), were recovered as forming a monophyletic group in this study (Figures 13 and 14). Although in the strict consensus trees of MPTs-SAW-MaxCI and MPTs-SAW-MinCI (trees not presented), A. pseudoplatani Visnya also came out with the former two species, A. forbesii (Ashmead) was not recovered with them in the same clade in any analysis. Interestingly, Sampson (1943) erected a new subgenus, Nealeurochiton Sampson, to accommodate A. forbesii, but this was subsequently synonymized with Aleurochiton by Mound and Halsey (1978).

Tribe Aleurilobini
This tribe was proposed by Takahashi (1954), for five genera based on Japanese species (Acanthobemisia, Aleurolobus, Apobemisia, Bemisia, Parabemisia). All studied species of Aleurolobus including the type-species, A. marlatti (Quaintance), were recovered in the same clade but not as a monophyletic group. The only examined species of Apobemisa, A. kuwanai (Takahashi) (the type-species), was not found with Aleurolobus in the same clade, and the type-species of Bemisia (not recovered as a monophyletic genus), B. tabaci, was never recovered either with Aleurolobus or Apobemisia. The type-species of Parabemisia was not studied but the five examined species did not form a clade. Acanthobemisia and the four species of Parabemisia were only recovered in the same clade in the strict consensus tree of MPTs-SAW-MaxRI ( Figure 14). David (1990) later transferred Bemisia to a separate tribe, Bemisini (see below), but Takahashi's Aleurolobini, even excluding Bemisia, was not supported in this study. David (1990) and Regu and David (1993) classified six genera from India in this tribe (Appendix A). Only the type-species of Aleuropapillatus and Orientaleyrodes were studied and these appeared in two distantly separated clades. Africaleurodes was found as a monophyletic genus, but did not form a clade with the other five genera. Two examined species of Asterochiton (but not the type-species) were recovered on two different branches. All species of Aleurolobus and Crescentaleyrodes were found in the same clade (Figures 13,14 and 19), and the possible monophyly of the latter will be discussed later.
Tribe Aleuroplatini David (1990) placed two genera in this tribe (Appendix A), of which Moundiella David was later synonymized with Viennotaleyrodes (David et al. 1994). Unlike the latter genus, Aleuroplatus was found to be non-monophyletic, though its monophyly was not rejected statistically using the Templeton test (Table II) (see also ''Discussion''). The type-species of Aleuroplatus, A. quercusaquaticae (Quaintance), and Viennotaleyrodes gathered in two different clades within a relatively big monophyletic clade in the strict consensus tree of MPTs-SAW-MaxRI ( Figure 14).

Tribe Aleyrodini
The Aleyrodini originally comprised 24 genera (Sampson 1943), and the author added another genus in 1947. Five more genera were added later (Drews and Sampson 1956; Sampson and Drews 1956) (Appendix A). Of these Aleuromigada Singh is considered as nomina nuda and the name Frauenfeldiella Gomez-Menor is not available in Aleyrodidae as it is preoccupied in the Cecidomyidae; the genus is now known as Aleurotuba (Mound and Halsay 1978;Tremblay and Iaccarino 1978). There were no specimens available for Hesperaleyrodes, Luederwaldtiana, Mexicaleyrodes, Nealeyrodes, and Neoaleurodes, and Laingiella was also omitted from the analysis (see ''Materials and methods''). Furthermore, Aleurocanthus, Aleurotrachelus, and Pentaleyrodes were transferred to Aleurochantini (see above), but David (1990) later reassigned Aleurotrachelus to Aleyrodini. Russell (1947) transferred Aleurotithius to Trialeurodini, and David (1990) relocated Acaudaleyrodes, Tetraleurodes, and Zaphanera in Neomaskellini, Tetraleurodini, and Zaphanerini, respectively (see below). Monophyly of Aleyrodini based on Sampson's classification even excluding Acaudaleyrodes, Aleurocanthus, Aleurotithius, Aleurotrachelus, Pentaleyrodes, Tetraleurodes, and Zaphanera was not supported in this study. Takahashi (1954) placed only five genera from Japan in the Aleyrodini (Appendix A), of which two genera are now considered as synonyms: Odontaleyrodes Takahashi was synonymized with Pealius by Martin (1999). This tribal classification, too, failed to be monophyletic. Except Aleyrodes, David (1999) transferred all genera placed by Takahashi in Aleyrodini, to Bemisini and placed two more genera in the latter tribe (see Bemisini below). David's (1999) classification of Aleyrodini with five genera from India, viz. Aleurocybotus, Aleuromarginatus, Aleurotrachelus, Aleurotulus and Aleyrodes, was not supported by this study.

Tribe Bemisini
Four of the six genera placed in this tribe by David (1990) (Appendix A) were briefly discussed above. The type-species of Pealius was not studied but the 10 examined species did not form a monophyletic clade. Setaleyrodes including the type-species, S. mirabilis Takahashi, also appeared non-monophyletic, but they were recovered together with some species of Pealius in the same clade (Figures 13 and 14). Indoaleyrodes (three species examined including the type-species, I. laos (Takahashi)) was recovered as monophyletic but in a separate clade. The type-species of Neopealius (the only examined species of this small genus) was on its own branch. As already mentioned, the monophyly of Bemisia was refuted statistically (Table II) (see also ''Discussion'') and its type-species, B. tabaci, was on its own branch.
Tribe Dialeurodini Sampson (1943Sampson ( , 1947 placed 32 genera in the tribe Dialeurodini and 13 more genera were later added by Sampson and Drews in 1956 (Appendix A). Of these, Aleuroclava and Aleurotuberculatus Takahashi are now treated under the former name (Martin 1999), as well as Corbettella Sompson, Neobemisia Visnya and Roucasia Goux, which were synonymized with Pealius, Asterobemisia (Mound and Halsey 1978) and Bemisia (Danzig 1964), respectively. Also, Nipaleyrodes and Stenaleyrodes have since been classified as members of the Aleurodicinae (Mound and Halsey 1978) (see also ''Discussion'' for Stenaleyrodes). For Aleuroporosus, Anomaleyrodes, Dialeurotrachelus, Neoaleurolobus, Metaleyrodes, Plataleyrodes, Pseudaleurolobus, Pseudaleyrodes, Trichoaleyrodes and Xenobemisia there were no specimens available for study. Although more than half of the 29 remaining genera were not found to be monophyletic, considering only their type-species, they never form a monophyletic clade, and the picture is not improved by excluding those genera that were later transferred to other tribes, viz. Acanthobemisia, Africaleurodes, Aleurolobus, Aleuroparadoxus, Aleuroplatus, Asterochiton, Bemisia, Parabemisia, Pealius, Setaleyrodes, and Trialeurodes (Russell 1947;Takahashi 1954;David 1990;Regu and David 1993). Takahashi (1954) placed only four genera from Japan in this tribe (Appendix A), of which Rhachisphora was recovered in a separate clade from the other three genera in all analyses (Figures 13 and 14) but not as a monophyletic genus. All examined species of Taiwanaleyrodes and Aleuroclava (5Aleurotuberculatus) (see above), and also a few species of Dialeurodes including the type-species, D. citri (Ashmead) were in the same clade together with some other genera in both strict consensus trees (Figures 13,14 and 17). Aleuroclava and Taiwanaleyrodes will be discussed in detail later. David (1990) and David and Sundararaj (1993) classified 19 genera from India in the Dialeurodini (Appendix A), of which three genera have now been synonymized (Martiniella Jesudasan & David and Aleurotuberculatus were synonymized with Aleuroclava by Martin (1999)). Also, the three subgenera of Dialeurodes, viz. Dialeuronomada, Gigaleurodes, and Rabdostigma of Quaintance & Baker, which were raised to the generic level by Indian authors (David and Sundararaj 1993;Sundararaj and David 1994), were here treated under Dialeurodes (the four examined species of the latter genus, viz. D. cerifera Quaintance & Baker, D. cinnamomi Takahashi D. ixorae Singh and D. minahassai Martin are assignable to Gigaleurodes, Gigaleurodes, Dialeuronomada and Rabdostigma, respectively, but none of them is the type-species). Kanakarajiella (only the type-species examined), Minutaleyrodes, and Singhius gathered in the same clade with Aleuroclava, Dialeurodes, and Taiwanaleyrodes ( Figure 17). The type-species of Asialeyrodes and Cockerelliella were not studied but one of the two examined species of the former and all examined species of the latter genus as well as six of the eight examined species of Singhiella including the type-species, S. bicolor (Singh), were also recovered in this clade, while Rusostigma was only found in this clade in the MPTs-SAW-MaxRI ( Figure 14). Furthermore, except D. cinnamomi, the other three species of Dialeurodes, all were found in this clade but only in the MPTs-SAW-MaxRI (D. ixorae was also found in the UW-MPTs), i.e. recovering the representatives of all three subgenera. The four remaining genera, Dialeurolonga, Dialeuropora, Fippataleyrodes, and Rhachisphora also appeared in four different clades.
Tribe Lipaleyrodini David (1990) included only Lipaleyrodes for this tribe (Appendix A) and the genus excluding L. atriplex was recovered to be monophyletic (including the type-species, L. phyllanthi Takahashi). L. atriplex had originally been described in Aleyrodes, and was subsequently transferred to Lipaleyrodes by Martin (1999), though it did not group even with the species of Aleyrodes in this study. Sampson (1943) based the monotypic Neomaskellini on Neomaskellia, and later Acaudaleyrodes was added by David (1990) (Appendix A). Both genera were recovered to be monophyletic but located in two distantly separated clades in all analyses (Figures 13 and 14).

Tribe Tetraleurodini
The only included genus, Tetraleurodes, was not found to be monophyletic in any analyses, though its monophyly was not rejected statistically (Table II) (see also ''Discussion'').

Tribe Trialeurodini
Russell (1947) erected this tribe and stated ''only the known genera Aleuroparadoxus and Aleurotithius of Quaintance and Baker and Trialeurodes Cockerell appear to be assignable to this tribe''. Subsequently, Russell (1967) added her new monobasic genus, Venezaleurodes.
In the strict consensus tree of UW-MPTs ( Figure 13), nine of the 11 examined species of Trialeurodes including the type-species, T. pergandei (Quaintance), as well as Aleurotithius (only the type-species studied), Venezaleurodes and several different genera formed a polytomy, and only two species of Trialeurodes were recovered in the same clade. Aleuroparadoxus was found to be monophyletic and formed a separate clade with several different genera. In the MPTs-SAW-MaxRI (Figure 14), the nine species of Trialeurodes, and Venezaleurodes were recovered in the same clade, and Aleurotithius formed a separate small clade with the two other species of Trialeurodes.

Tribe Zaphanerini
This tribe was proposed by David (1990) for Zaphanera, (Appendix A). Five species of this genus including the type-species, Z. cyanotis Corbett, were included in the present study. These formed two separate monophyletic clades which were supported by relatively high Jacknife values ( Figure 13). The African and oriental species (Z. capparis Bink-Moenen and Z. cyanotis, respectively) were recovered in one, and the three Australian species in another (Figures 13 and 14). Thus, the Australian species (Z. niger (Maskell), Z. papyrocarpae Martin, and Z. rhachisreticulata Martin) do not appear to be congeneric with the type-species and a new genus needs to be erected for them (see ''Nomenclatural changes'').

Whitefly-host plant relationships
Host plant information at family, order and higher group levels for all whitefly species included in this study is given in Appendix E. All botanical names follow the system of the Angiosperm Phylogeny Group (APG) (Bremer et al. 1998(Bremer et al. , 2003Soltis et al. 2000). Whitefly host plant information was obtained from Mound and Halsey (1978), Bink-Moenen (1983) and Martin (1999), as well as from more recent original descriptions, and from unpublished information available on slides examined.
Except for the only gymnosperm host record, Dioon spinulosum (Zamiaceae) for T. vaporariorum (Westwood), and also those few species recorded from Pteridophyta (see below), whiteflies are found mainly on angiosperms. Furthermore, few whitefly species are known to be monophagous, most being oligo-or polypahgous. The number of host plant orders attacked by the species studied here varies from one to 31 (probably the true number is higher) (Appendix E).
Twenty described, and three or four undescribed, species are reported from fern hosts (Pteridophyta) (Mound and Halsey 1978;Martin and Camus 2001), of which the fern host record for T. vaporariorum (which is highly polyphagous on Angiospermae) is uncertain, and the identification of one of these species as B. tabaci (also highly polyphagous on Angiospermae) is not definitive (Martin and Camus 2001). Here, 14 described species recorded from ferns (ignoring B. tabaci and T. vaporariorum) were scored. Of these, all described species of Aleuropteridis and Filicaleyrodes, were recovered as monophyletic, but they were distantly separated in trees from both analyses (Figures 13 and 14). The results were the same for the two fern-feeding species of Metabemisia (M. filicis Mound and M. palawana Martin), as well as the two fern-feeding species of Aleurotulus (A. nephrolepidis (Quaintance) (the type-species) and A. pteridophytae Martin). Neither the type-species of Metabemisia (not examined) nor the other species of Aleurotulus have been recorded from ferns. The latter genus appears not to be monophyletic and, apparently, its non-fernfeeding species are not congeneric with the type-species (it should be noted that one of its non-fern-feeding species, A. arundinacea Singh, is considered as species incertae sedis (Mound et al. 1994)). For Trialeurodes, three of the 11 examined species, T. bruneiensis Martin, T. dicksoniae Martin, and T. rex Martin, colonise ferns, but they were not recovered in the same clade. The only studied species of Mixaleyrodes were not recovered as a sister group to any of the fern-feeding species. In general, the position of these 14 species on the both cladograms did not support a clear Pteridophyta-whitefly relationship for the all fernfeeding species, although separate correlations were found for some species at genus level.
Almost the same results were obtained after mapping angiosperm hosts on both strict consensus trees. In general, some small monophyletic clades appear to be specialized on particular host plant orders, but in most cases, there was little of any indication of correlation between the species of a particular clade and host plant order.
Three genera for which a clear relationship with host plant order was apparent were Aleuromarginatus, Corbettia, and Viennotaleyrodes. The last two appeared to be monophyletic, but Aleuromarginatus was not found to be so, though most of its examined species formed a monophyletic clade with Corbettia ( Figure 15). Virtually all species of these three genera have been recorded from Fabaceae (the exception being C. graminis Mound on Poaceae, V. bosciae Bink-Moenen recorded from Brassicaceae, and C. millettiacola Dozier which has additionally been recorded from Apocynaceae) (there is no host record for A. serdangensis Takahashi but this species most probably does not belong to Aleuromarginatus; see ''Discussion'') (Appendix E). Furthermore, the monobasic genus Papillipes, which is also recorded from Fabaceae, formed a sister group to Viennotaleyrodes (Figures 13 and 14), but this clade did not form a sister group to Aleuromarginatus+Corbettia.

Discussion
The family Aleyrodidae is taxonomically difficult. The genera usually have no reliable suite of morphological characters to allow them to be recognised unambiguously, and this is obvious from many of the original descriptions. Many of the larger genera have some species included which differ in at least one of the characters given in the description of the genus (e.g. Aleurotrachelus asparagi (Lewis) and A. tarennae Bink-Moenen have a submarginal line or fold, whereas, based on the original description of the genus, the submarginal area is not separated from the dorsal disc (Quaintance and Baker 1914)) and, sometimes, there is really no reason for a species to have been placed in its current genus at all (e.g. Aleuromarginatus serdangensis; see below).
These and other problems created some difficulties in choosing possible genus level characters for the cladistic analyses. We attempted to code all puparial characters used by authors in generic descriptions, although in most cases there are no distinct boundaries between genus level characters and those of species level. For example, Bink-Moenen (1983) while describing three new genera, Arachnaleyrodes, Papillipes and Yleyrodes, and comparing them with their morphologically closely related genera, i.e. Orchamoplatus, Viennotaleyrodes and Africaleurodes respectively, used the absence and/or presence of first abdominal setae as one of the diagnostic characters to separate the new genera from their allied genera. This character, however, has also been used to differentiate two species of a particular genus, e.g. Dialeurolobus rhamni Bink-Moenen and D. pulcher Danzig (Bink-Moenen and Gerling 1990), and Crenidorsum millennium Martin and C. celebes Martin (Martin 1999). As stated by Martin (2003), there is nothing inherently wrong with the use of immature stages taxonomically, but the problem is that, with our current poor understanding of the true significance of many puparial characters, we have yet to maximise their value. The lack of such information has impeded our understanding of relationships between aleyrodids, rather than any lack of characters for use (immature stages have been used taxonomically in other insects such as Psylloidea (White and Hodkinson 1982) and Lepidoptera (Kitching 2002(Kitching , 2003Willmott 2003)). For example, as mentioned in comments for character 80 (glandular bases to marginal teeth), this structure has sometimes been described as ''a double row of teeth'' or rarely ''a translucent membranous area at the bases of the marginal teeth'', and according to Bink-Moenen (1983), these are probably wax-secreting pores or papillae. In the latter case, characters 65 (submarginal papillae) and 80, apart from their shape differences, can be considered the same in a broad sense. Table III shows some examples regarding the opinions of different authors about the relationships between some genera. In fact, with whitefly puparial taxonomy lacking objective criteria (it seems to be partly a matter of guess-work at genus level), there is likely to be quite a lot of apparent homoplasy (in addition to some true homoplasy) due to misinterpreting similar character states during the coding stage.
Owing to the high level of homoplasy in the data set, the maximum consistency index may be the best choice for reweighting characters for SAW, because it cannot weight characters as zero, so potentially useful characters would not be excluded from the analysis (Quicke et al. 2001). It should be noted that when stability was reached in SAW analyses and all characters then set back to unit weight, the lengths of the MPTs-SAW-MaxCI and MPTs-SAW-MinCI were much higher (223 and 197 steps, respectively) than that of UW-MPTs, while MPTs-SAW-MaxRI and MPTs-SAW-MinRI were only a few steps longer (see ''Results'').
Retention indices (ri) for each non-constant character were calculated to evaluate how much of the variation displayed by the characters could be attributed to true synapomorphy on trees, as well as calculating consistency indices (ci) to evaluate the amount of homoplasy of each character, and length. The results are shown in Table IV. In all UW-MPTs, the retention indices of characters 5, 14, and 86 were zero, i.e. contributing no synapomorphy to the cladograms. In contrast, retention index had the maximum value for the characters 22, 24, 30, 46 (only the best fit), 63, 64, 76, 78/79 (only the best fit), 82-84, 87/88 and 90, indicating the contribution of maximum synapomorphy to the cladograms (Table IV). Of  Excluding Stenaleyrodes from out-group taxa, puts it in the position of being a sister group to all ingroup taxa. Although Takahashi (1938) gave no information about the subfamily placement in his description of the monobasic genus Stenaleyrodes -as well as Mamet (1952) giving supplementary notes on the type-species, S. vinsoni Takahashi -but at the end of the description, the author stated ''related to Trialeurodes Laing, but differs in the elongate vasiform orifice, …. Resembles Bemisia Quaint. et Baker in the characters of vasiform orifice, …'' (Takahashi 1938). In other words, Takahashi was indicating similarities to two aleyrodine genera. As already mentioned, Sampson (1943Sampson ( , 1947 placed Stenaleyrodes in the Aleyrodinae, tribe Dialeurodini (Appendix A). Mound and Halsey (1978) listed the genus under Aleurodicinae and synonymized Dialeurodicus elongatus Dumbleton, which had already been transferred to Stenaleyrodes by Cohic (1968), with S. vinsoni. The former authors also stated ''the material in the BMNH from Tanzania listed here as Stenaleyrodes sp. indet. represents a distinct species, which as in the case of vinsoni, can only be distinguished from Dialeurodicus at present by the elongate shape of its pupal case''. No syntype materials of S. vinsoni were examined, but a paratype of D. elongatus and another specimen of S. vinsoni collected from the type locality (Reunion Island) and from the host of S. vinsoni (palm) as well as Tanzanian samples were studied. These samples lacked both the tarsal claw and apical adhesion pad. The illustration of hind leg in the original description of S. vinsoni (Takahashi 1938, Figure 5) is in agreement with this observation. It seems that Mound and Halsey (1978) did not notice the absence of the tarsal claw in Stenaleyrodes while comparing it with Dialeurodicus, in which the character is present, and only mentioned the elongation of pupal case to distinguish these genera from each other.
From a consideration of puparial characters, there is no reason for Stenaleyrodes to be in Aleurodicinae due to its lack of the tarsal claw and lack of compound wax pores and/or agglomerate pores, i.e. the two basic puparial characters of the Aleurodicinae, though the vasiform orifice and the general size are similar to those of aleurodicine members. However, the examination of recently collected material of an undescribed species of Stenaleyrodes in BMNH has revealed aleurodicine characters in adults. Therefore, it would seem expedient to retain Stenaleyrodes in Aleurodicinae.
All 13 remaining characters with maximum retention indices (ri51) (except characters 22, 24, and 46) are autapomorphic characters of certain genera, e.g. characters 63 (eighth abdominal bifid process) and 82 (chain-like design along some abdominal and thoracic sutures) support monophyly of Aleurocerus and Bellitudo respectively. Thus, these characters did not provide any phylogenetic information about relationships between genera.
According to Bink-Moenen (1983), character 3 (recurvature of margin) is mostly of value at the genus level, although in our analyses it had an intermediate ri and pretty low ci (Table IV). Furthermore, Mound (1961), who erected the genus Aleuropteridis, mentioned that ''this genus is near Tetralicia and Aleuropleurocelus in that the true margin is deflexed ventrally, i.e. the dorsal disc is larger than the ventral surface''. Although in these three genera the character is quite typical, it can also be seen in several species putatively belonging to other genera such as Bemisia hirta Bink-Moenen and Tetraleurodes caulicola Nakahara, which it is not a specific character of the latter genera. All examined species currently placed in Tetralicia, in spite of their sharing this character, were not found to be monophyletic. Aleuropteridis formed a sister group to Xenaleyrodes (also shows the character), and among the taxa gathered with Aleuropteridis in the same clade (Figure 16), two examined species of Leucopogonella lacked this character. In this clade, Tetraleurodes selachidentata Bink-Moenen was described as a species incertae sedis (Bink-Moenen 1983). Incidentally, some examined species of Corbettia and Viennotaleyrodes (both genera were recovered as monophyletic) lacked the character. Thus, character 3 seems to be far less reliable than past authors have suggested.
Submarginal line or fold (character 28) have been thought to be an important character at genus level and has been used by authors both in generic descriptions and identification keys (e.g. Mound 1965; Bink-Moenen 1983; Jesudasan and David 1991;Martin et al. 2000). It separates the submargin from the dorsal disc area and can appear in different forms (see character 29). In general, this character could not gather the genera possessed it in a separate clade, at least those that it appears in an identical shape. It should be noted that Aleurotuberculatus, in spite of having the character, was synonymized with Aleuroclava (Martin 1999). Although the 15 examined species (seven species had the character) of the latter large genus were not found to be monophyletic in this analysis (Figure 17), but  part of the tree, e.g. clades II and VI, it can be hypothesized that the character has independently evolved several times. The same hypothesis probably applies to characters 49, 55 and 57. Most taxa in the clade C ( Figure 18) lack lingular setae (or the setae are not apparent) (character 49), and those in the clade A lack both caudal furrow (character 55) and caudal ridges (character 57), but as for character 37, some other small clades in the other part of the tree lack these characters too, e.g. clades III and V for character 49, and clades I and IV for both characters 55 and 57. The applicability of any tribe level classification for the Aleyrodinae seems to be premature. Sampson (1943) was the first to classify aleyrodine genera to five tribes (Appendix A). Whereas, in 1956, in their identification key for the genera of the subfamily, Sampson and Drews (1956) placed almost all described genera by that time in two tribes, Aleyrodini and Dialeurodini, and the other three tribes was each comprised of a single genus. The authors also did not include the sixth tribe proposed by Russell (1947) in the key because of difficulty of finding characters to separate it from the others, and nothing was mentioned about the two tribes proposed by Takahashi (1954). It seems that erecting more tribes, as done by David (1990), with our current poor understanding the significance of puparial characters will not help. For example, none of the genera placed in the Alyrodini by three different authors (Appendix A) was recovered in the same clade as Aleyrodes in this study, and Indoaleyrodes, which was placed in the Bemisini by David (1990), formed a sister group to Aleyrodes ( Figure 15). Bearing in mind that defining a reliable suite of morphological characters at genus level is still problematic and many genera lack objective definitive, it will obviously be difficult to find some synapomorphic characters to define different tribes confidently.
Unfortunately, some strikingly obvious structural apomorphies have probably been independently derived in a number of quite closely related species. Such characters can easily confuse attempts at phylogenetic reconstruction as they are difficult to differentiate from genuine synapomorphies. Often such striking characters are considered to be of ''generic worth'' and in practice many species which show these kinds of characters are placed in separate monobasic or oligobasic genera (Gauld and Mound 1982). It seems that the ability to develop apomorphies of this type independently in closely related taxa can be found in Aleyrodidae. The taxa of the clade shown in Figure 19 may be good examples. In this clade, three genera, Aleuroparadoxus, Bellitudo, and Crescentaleyrodes, have obvious apomorphies, i.e. characters 76 (modified papillae of dorsal disc area, which are slightly elevated, nearly flat and plate-like), 82 (chain-like design along some abdominal and thoracic sutures) and 87 (submarginal crescent-shaped pores) respectively. Of these three genera, Crescentaleyrodes was not recovered as monophyletic and Aleurotrachelus fumipennis (Hempel) made it paraphyletic (Figure 19). David and Jesudasan (1987) proposed Crescentaleyrodes for Tetraleurodes semilunaris Corbett, which had already been transferred to Aleurolobus by Bink-Moenen (1983), as well as transferring two species of Aleurolobus, A. monodi Cohic and A. paulianae Cohic to Crescentaleyrodes. The main diagnostic character of Crescentaleyrodes, as the name implies, is the presence of crescent-shaped (semilunate) pores ( Figure 12J) along the submargin. It should be noted that the syntype materials of A. fumipennis were not examined and its original description (Hempel 1899) provides no illustration and is inadequate for the positive identification of the species. The examined specimens (from Brazil collected on grasses, the same locality and host plant as the syntypes) identified as A. fumipennis in BMNH had submarginal crescent-shape pores (character 87) and were in agreement with the redescription and illustration of A. fumipennis provided by Bondar (1923), i.e. belonging to Crescentaleyrodes (see also ''Nomenclatural changes''). Furthermore, there are three monobasic genera, Aleyrodiella, Harpaleyrodes and Rositaleyrodes, in this clade. The latter was proposed for Aleurolobus oplismeni Takahashi by Meganathan and David (1994), who erroneously proposed a new holotype and six paratypes from their samples collected for an extant species of another author. The authors stated ''this genus resembles Aleurolobus Quaintance & Baker in all the characters, but it differs from it, by having the submargin entirely demarcated from dorsum by a submarginal furrow and thoracic and caudal tracheal combs differentiated from margin by teeth and a pouch like structure''. The entire submarginal furrow can be seen in other species of Aleurolobus, e.g. A. hargreavesi Dozier, and also the differentiation of tracheal area from the margin is variable in this genus. It seems that there is no reason for proposing a separate genus (see also ''Nomenclatural changes''). Danzig (1966) erected Aleyrodiella and stated ''an extremely distinctive genus whose taxonomic position is uncertain''. The author then discussed its similarity to the three genera, Aleuroparadoxus, Aleurotithius, and Trialeurodes (Table III), as well as its differences from them. These three genera are those that Russell (1947) placed in the Trialeurodini (Appendix A) but only Aleuroparadoxus was recovered here in the same clade as Aleyrodiella. The latter two genera were found here to be more closely related to Aleurolobus than to Trialeurodes and/or Aleurotithius (Table III). The third monobasic genus in this clade, Harpaleyrodes, was erected by Bink-Moenen (1983) and, as the author stated, it resembles Aleurolobus. One of its differences from Aleurolobus is its strongly recurved submargin (Bink-Moenen 1983) (character 3), and as discussed above, this is variable in some genera. It is possible that Harpaleyrodes and Aleurolobus are congeneric, in spite of the alleged differences. It is also worth mentioning that according to Danzig (1964), Dialeurolobus (with two described species so far), which formed a sister group to all discussed taxa in this clade (Figure 19), is allied to Aleurolobus and differs from the latter in having no submarginal fold (character 28). The variability of this character has been discussed, above. Dialeurodes erythrinae Corbett has also much in common with the members of this clade ( Figure 19) rather than with Dialeurodes, and its placement in Dialeurodes is doubtful (species incertae sedis).
As is common in phylogenetic analyses, the sister groups of some clearly monophyletic genera are often paraphyletic assemblages, e.g. Aleurocerus rendering Aleurothrixus paraphyletic ( Figure 20). Seven of the 10 examined species, including the type-species, Aleurothrixus floccosus (Maslell), as well as Hempelia (see ''Nomenclatural changes'' for proposed synonymy of Hempelia with Aleurothrixus as well as the discussion of the other three species of Aleurothrixus) were recovered in the same clade as Aleurocerus. This clade was relatively highly supported in both bootstrap and parsimony Jackknifing analyses ( Figure 20). A close relationship between Aleurocerus and Aleurothrixus was also postulated by Russell (1986) and the presence of the eighth abdominal bifid process (character 63) ( Figure 11A) in Aleurocerus its only diagnostic character.
The same was found for Corbettia, recovered as monophyletic here, but rendering Aleuromarginatus paraphyletic ( Figure 15). Interestingly, Aleuromarginatus and Corbettia also colonize the same host plant family, Fabaceae (see ''Results''). Seven of the 10 examined species, including the type-species, A. tephrosiae Corbett, were found in the same clade as Corbettia. The two other species scored, A. marginiquus Martin and A. nemciae Martin, are apparently not congeneric with the type-species and this uncertainty was alluded to by the author (Martin 1999). Based on the results of this study, the placement of these species in Aleuromarginatus is uncertain (species incertae sedis). A. serdangensis, as already mentioned, seems to be a misplaced species differing in several characters (e.g. 4, 8, 9, 44 and 94) from the type and the other included species. A. serdangensis was recovered in a quite distantly separated clade that includes several diverse species putatively belonging to different genera (Figures 13 and 14).

Monophyly of large genera
Aleuroplatus, Aleurotrachelus, and Tetraleurodes are genera with many included species and worldwide distribution (Mound and Halsey 1978). Currently each includes many diverse species and lacks a suite of distinguishing characters. Therefore, there is no reason to assume that species currently classified in any of these, especially from different geographical regions, are related to one another. As expected, these genera were not found to be monophyletic, but only for Aleurotrachelus, was monophyly rejected statistically. The reason for this may be the lack of distinct generic characters. In other words, the species of these genera just have combinations of characters shared with various other genera. Regarding Aleurotrachelus, it should be noted that the specimens checked as A. fumipennis, as discussed above, had the typical character of Crescentaleyrodes and in the constraint analysis it was not excluded from Aleurotrachelus. Thus, it is quite likely that this species contributed largely to the statistical rejection of monophyly of Aleurotrachelus. To test this, another constraint analysis was performed for this genus excluding A. fumipennis. The length of the MPTs was 2773 compared with 2730 for the UW-MPTs. The result of the Templeton test (500 trees were compared) showed that the constrained trees were not consistently significantly longer with P values for the majority of the trees above 0.05 (P50.0390-0.0793; see also Table II). Such interpretation may also apply to Dialeurodes, which in spite of many of its some species having been transferred to Massilieurodes and Singhiella by Jensen (2001), it is still a large and heterogeneous genus. In contrast, although monophyly of Trialeurodes was not also rejected statistically, the same interpretation does not seem to apply. Trialeurodes is quite well characterised and most of its species are certainly of New World origin (Mound 1984). T. vaporariorum, which has long been known as a worldwide pest, almost certainly evolved in the south western part of North America (Russell 1948). In the strict consensus tree of MPTs-SAW-MaxRI (Figure 14), nine of the 11 examined species including the type-species, T. pergandei, as well as the closely related monobasic genus Venezaleurodes (probably should be synonymized with Trialeurodes) were recovered as a monophyletic group. These species were unresolved in the strict consensus tree of UW-MPTs ( Figure 13). The probabilities that this genus is monophyletic (Templeton test: P50.5590-0.7035) are much higher than for the other constrained genera (Table II). The factors mentioned above for Trialeurodes may also be applicable to Bemisia. According to Mound (1984), it seems probable that the Bemisia group primarily originated in the Palaearctic, though it was previously suggested that B. tabaci was Indian region in origin (Mound 1965). The main problem with this genus is that in spite of including the well-known pest, B. tabaci, it is not well defined. For this reason, it is expected that the genus includes at least several misplaced species which resulted its monophyly to be rejected statistically.

Host plant relationship
Relationships between whiteflies and their host plants are unclear. According to Mound and Halsey (1978), since the majority of aleyrodids are known only from the Angiosperms, it seems unlikely that this family evolved until after the radiation of the flowering plants. These authors compared the Cronquist's (1968) plant classification with the host plants of whiteflies and deduced that most whitefly species were recorded from plants belonging to four smaller, less highly evolved subclasses. The authors then stated: ''As the Asteridae and Rosidae are the most advanced groups of higher plants, it may be that the aleyrodids did not evolve at the same rate but retained a relationship with the more primitive groups of plants. Alternatively the relationship may be a reflection of the geographical distributions of the organisms, the Asteridae and Rosidae being most common in temperate regions, whereas the Aleyrodidae are mainly a tropical group.'' Based on the system of the Angiosperm Phylogeny Group (APG) (Bremer et al. 2003), 328 of the 439 whitefly species examined in this study are known from the two advanced groups, i.e. asterids (including euasterids I and euasterids II) and rosids (including eurosids I and eurosids II). In this system, euasterids II and euasterids I are two of the most recently derived groups. More than 100 of the examined whitefly species are known from these groups (21 from euasterids II, 78 form euasterids I, and 17 from both) (Appendix E). The positions of these species in the strict consensus tree of UW-MPTs as well in the MPTs-SAW-MaxRI show no pattern. Thus, parallel diversification with host plants and whiteflies is not apparent and it appears that widespread host switching has obscured any basic associations. Having said this, the evolution of aleyrodid host plant affiliations does not seem to be totally random as some groups have species feeding on related plants.

Nomenclatural changes
The results obtained from phylogenetic analyses presented above, indicate the great need for reclassification of the Aleyrodinae. It is quite possible that those species whose pupal cases differ considerably from each other, nevertheless have adults showing apparently unique apomorphies and therefore they should be placed in the same genus (Bink-Moenen 1992, discussing the genus Aleuroviggianus). On the other hand, adults can differ markedly despite of great similarities in their pupal cases, and so are placed in different genera (Russell 2000, discussing the genera Aleurocybotus and Vasdavidius). Since adults for the majority of described species are unknown, it is not desirable to change the current genus concepts defined based on pupal case character only, unless the true significance of the latter is known. This will be the next necessary step in improving the taxonomy of whiteflies.
Checking such a large number species, belonging to more than 100 genera, revealed the necessity for proposing some nomenclatural changes. These are only proposed for those taxa where the necessity of changing their current taxonomic positions has been supported by the pylogenetic analysis, in addition to the convictions obtained during the visual examination of specimens. Those genera mentioned here as New genus 1 and New genus 2 will be described somewhere else.
Aleuroclava Singh, 1931 Type-species: Aleuroclava complex Singh, 1931. Aleurotuberculatus Takahashi, 1932. Type-species: Aleurotuberculatus gordoniae Takahashi, 1932[synonymized by Martin 1999. Martiniella Jesudasan & David, 1990. Type-species: Aleurotuberculatus canangae Corbett, 1935[synonymized by Martin 1999. Taiwanaleyrodes Takahashi, 1932. Type-species: Taiwanaleyrodes meliosmae Takahashi, 1932. Syn. n. Comments: Takahashi (1932 erected Aleurotuberculatus mentioning that it was closely allied to Aleuroclava and may be a synonym of it. In the same article, he erected Taiwanaleyrodes and stated ''allied to Aleurotuberculatus Takah., but differs from it in lacking dorsal tubercles and thoracic tracheal clefts, but in possessing a distinct marginal rim on the venter, and also distinguishable from …'' (Takahashi 1932). Takahashi (1939) in the original description of Taiwanaleyrodes carpini Takahashi stated ''cephalothorax with a pair of large rounded tubercle-like markings on the pronotum, and also on the mesonotum, and 2 pairs of similar ones on the metanotum, …, thoracic tracheal folds very short, the clefts indistinct, very shallow, wide, without teeth''. In this article, the author transferred Aleurotuberculatus pyracanthae Takahashi to Taiwanaleyrodes (apparently Mound and Halsey (1978) did not noticed this transformation and listed the latter species under Aleurotuberculatus and/or did not accept it, but mentioned nothing), although in the original description of A. pyracanthae stated ''venter with no rim'' (Takahashi 1933). It is clear that the two species of Taiwanaleyrodes, which placed in this genus by the author of the genus, are not quite agree with the type-species, T. meliosmae Takahashi, in having the three diagnostic characters. Examination of many specimens of Aleuroclava, in addition to those included in this analysis, indicated that the dorsal tubercles and thoracic tracheal clefts definition is so variable and these characters should not be used for the separation of these genera. In the current study the type-species of Taiwanaleyrodes, as well as its one other included species were recovered within Aleuroclava (Figure 17). Taiwanaleyrodes is, therefore, here regarded as a junior synonym of Aleuroclava syn. n. Jesudasan and David (1990) proposed Minutaleyrodes for four species formerly included within Aleurotuberculatus. The genus was provisionally retained for its included species by Martin (1999), who synonymized Aleurotuberculatus and Martiniella with Aleuroclava. Aleuroclava similis (Takahashi) formed a sister group to Minutaleyrodes in this study ( Figure 17). Of the characters specified for Minutaleyrodes by Jesudasan and David (1990), demarcation of the submargin on the ventral surface, can be seen in some species of Taiwanaleyrodes (discussed above). The shape of transverse moulting suture is the same that of in A. similis, but the size of this species is not similar to Minutaleyrodes and submargin not ventrally demarcated in A. similis. Unlike Aleuroclava, the first abdominal setae in Minutaleyrodes are absent (not mentioned by the authors) but some species of the former genus, e.g. A. similis and A. nigeriae (Mound), also lack these setae. In the latter species, these setae are sometimes present (Bink-Moenen 1983). Therefore, Minutaleyrodes seems to be congeneric with Aleuroclava, but because of the three species of Dialeurodes recovered between these two genera (see Figure 17), it is not synonymized. These three species currently accommodated in Dialeurodes, are not quite typical for their current genus nor for Aleuroclava, and it appears they have much in common with Aleuroclava, especially D. mirabilis Takahashi, rather than Dialeurodes.
Comments: Specimens checked were in agreement with the original description and illustration, and indicate that the inclusion of this species in Aleuroclava is appropriate. As stated by Takahashi (1949), this species is not a typical form of Dialeurodes. A. lanceolata is closely related to A. filamentosa (Corbett) (Figure 17).
Comments: As discussed before, according to Meganathan and David (1994), Rositaleyrodes resembles Aleurolobus in all the characters, but it differs from it, by having the submargin entirely demarcated from dorsum by a submarginal furrow and thoracic and caudal tracheal combs differentiated from margin by teeth and a pouch like structure. The latter character is variable in Aleurolobus and the entire submarginal furrow can be seen in other species of this genus, e.g. A. hargreavesi. Rositaleyrodes is therefore here regarded as a junior synonym of Aleurolobus syn. n.
Comments: As already discussed, seven of the 10 examined species of Aleurothrixus including the type-species, A. floccosus (Maslell), as well as Hempelia were recovered in a monophyletic clade with Aleurocerus ( Figure 20). The syntype material of the type-species of monobasic genus Hempelia showed no clear differences from Aleurothrixus that appear to be of generic significance and here Hempelia is regarded as a junior synonym of Aleurothrixus syn. n.
Aleurothrixus antidesmae Takahashi and A. smilaceti Takahashi are closely related to each other and do not seem to be congeneric with the type-species. A different genus is needed for them; almost certainly a new one. The situation is the same for the American species, A. interrogationis (Bemis), but it differs from these two Asian species. It was recovered within some species of Aleuroplatus and its placement in Aleurothrixus is doubtful (species incertae sedis).
Comments: as already mentioned in detail, the redescription and illustrations provided by Bondar (1923) as well as the specimens examined, clearly indicate the appropriate inclusion of this species in Crescentaleyrodes. This species is very close to C. paulianae (Cohic) and maybe they are synonym, but this need to check the type material of the latter species. The specimens of C. paulianae studied for this study were from the same locality and host plant of type materials and showed slightly differences from C. fumipennis.
Description: Pupal case. Cuticle black. Margin toothed, thoracic and caudal tracheal pore areas not differentiated from margin. Submargin not separated from dorsal disc area by a line or fold. Longitudinal moulting suture reaching margin, transverse moulting suture curving anteriorly and reaching margin. Abdomen with exaggerated intersegmental sutures extending into outer subdorsum. Abdominal rhachis present or absent. Dorsum usually punctuated by geminate pore/porettes. Vasiform orifice cordate or subcordate, often elevated. Operculum almost filling vasiform orifice. Caudal furrow absent or faintly indicated. Eyespots absent or present.
Chaetotaxy: Cephalic and first abdominal setae absent, eight abdominal and caudal setae present. A row of submarginal setae present but their total number as well as their number on abdomen and cephalothorax variable (6-11, usually 7/8 pairs excluding the caudal setae), occasionally with a few pair subdorsal setae. All setae very small and often capitate.
Venter: Thoracic and caudal tracheal folds present. Antennae long, reaching middle of middle legs or more, male antennae much longer.
Comments: except the type-species, all above species included in this genus were originally been described in Zaphanera. The type-species has already been transferred to Aleurolobus, Tetraleurodes and Zaphanera by Baker (1914), Dumbleton (1956) and Martin (1999), respectively. Pseudozaphanera differs from Zaphanera in the number of submarginal setae as well as their shape, absence of cephalic setae, position of eight abdominal setae, shape of transverse moulting suture, and length of antennae. In Zaphanera, there are five pairs of setae (excluding the caudal setae), two pairs on the cephalothorax and three pairs on the abdomen, and these setae are not capitate. Antennae do not reach beyond front legs, and the eighth abdominal setae are located adjacent to postero-lateral margin of vasiform orifice. In three of the four described species (examined) retained in Zaphanera, including the type-species, Z. cyanotis, cephalic setae are present, and based on the redescription provided by David and Subramaniam (1976), they are not discernible in Z. publicus (Singh) (not examined). In Zaphanera abdominal segmentation is distinct only in the submedian area in contrast to Pseudozaphanera (see also tribe Zaphanerini). Felsenstein J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783-791. Takahashi R. 1939. Notes on the Aleyrodidae of Japan (Homoptera) VII. Kontyû 13:76-81. Takahashi R. 1949. Some Aleyrodidae from the Riouw Islands (Homoptera). Mushi 20:47-53. Takahashi R. 1954. Key to the tribes and genera of Aleyrodidae of Japan, with descriptions of three new genera and one new species (Homoptera). Insecta Matsumurana 18:47-53. Takahashi R. 1957. Some Aleyrodidae from Japan (Homoptera). Insecta Matsumurana 21: 12-21. Takahashi R. 1961. Some species of Aleyrodidae from Madagascar IV (Homoptera). Mémoires de l'Institut Scientifique de Madagascar (E) 12:323-339. Takahashi R. 1962. Two new genera and species of Aleyrodidae from Madagascar (Homoptera). Proceedings of the Royal Entomological Society of London B 31: 100-102. Takahashi R. 1963. Some species of Aleyrodidae from Japan (Homoptera). Kontyû 31: 49-57. Tremblay E, Iaccarino FM. 1978 Based on Tribe Included genera Sampson (1943) Aleurochitonini Aleurochiton Tullgren Aleyrodini Acaudaleyrodes Takahashi  Aleurochitonini Lingula elongate, much longer than wide. Dorsum completely covered with simple pores. Adults with radial 1 , radial sector, and cubital veins in fore wing. Aleyrodini Lingula elongate, much longer than wide. Dorsum with relatively few simple pores. Thoracic tracheal folds, combs, pores and anal fold absent. Adults with radial sector and cubital veins in fore wing and with tarsal paronychium. Dialeurodini Lingula elongate, much longer than wide. Dorsum with relatively few simple pores. Thoracic tracheal folds and combs or pores, or pores or combs only, and anal fold, or only anal fold, present. Adults with radial sector and cubital veins in fore wing and with tarsal paronychium. Neomaskellini Lingula extremely short, hardly longer than wide. Adults with radial sector vein only in fore wing. Siphoninini Lingula elongate, much longer than wide. Dorsum with elongate, siphon-like wax tubes and with relatively few simple pores. Adults with radial sector and cubital veins in fore wing and lacking tarsal paronychium. Russell 2 (1947) Trialeurodini Main points: a submarginal row of disc pores and porettes with which are associated variously conspicuous papillae. Submargin and subdorsum not separated by a furrow. Submarginal setae shorter than the width of a submarginal ridge or without distinguishable setae. Vasiform orifice ending abruptly, its sides without minute spines. Thoracic tracheal folds without angular markings and usually without spines. Antennae not reaching to posterior thoracic spiracles. Takahashi 2 (1954) Aleurocanthini Tracheal pores or clefts absent, tracheal combs sometimes developed. Vasiform orifice not notched at the hind end, rounded, not elongated, sometimes elevated; lingula concealed under the operculum. Caudal furrow absent. Eyespots present in some species. Seventh abdominal segment nearly as long as, or a little shorter than, the six. Aleurolobini Tracheal pores or clefts absent, tracheal combs sometimes developed. Vasiform orifice not notched at the hind end, subcordate, triangular or truncated at the hind end, elongated in some species; lingula exposed, knobbed and with a pair of long setae, sometimes large, knobbed part of lingula elongate, much longer than wide. Caudal furrow sometimes developed. Eyespots present in some species. Seventh abdominal segment much shortened at the median area in many genera. Aleyrodini Tracheal pores or clefts absent, tracheal combs sometimes developed. Vasiform orifice not notched at the hind end, subcordate, triangular or truncated at the hind end, elongated in some species; lingula exposed, knobbed and with a pair of long setae, sometimes large, knobbed part of lingula globular, not distinctly longer than wide. Caudal furrow sometimes developed. Eyespots present in some species. Seventh abdominal segment much shortened at the median area in many genera. Dialeurodini Tracheal pores or clefts usually present, if absent, vasiform orifice definitely notched at the hind end and caudal furrow distinctly defined, or lateral ridges (rhachis) developed on the abdomen and dorsum with many short spine-like setae, many of which are capitate. Lingula usually small, wanting long setae. Tracheal combs and eyespots lacking.

Based on Tribe Characters
David 2 (1990) Aleurocanthini Thoracic and caudal tracheal combs, clefts, or pores or furrows absent. Vasiform orifice elevated, rounded; lingula concealed by operculum. Margin with a single row of teeth or smooth. Submargin not separated from dorsal disc. Many prominent spines on the dorsum, often longer. Aleurolobini Thoracic and caudal tracheal combs, clefts, or pores or furrows present. Submargin separated from dorsal disc. Margin generally toothed. Seventh abdominal segment much shortened at the median area in many genera. Eighth abdominal segment often trilobed. Caudal furrow sometimes discernible. Vasiform orifice nearly triangular or subcordate; lingula knobbed and exposed. Aleuroplatini Tracheal pores, clefts, folds, furrows absent but thoracic and caudal tracheal combs indicated. Submargin generally not separated from dorsal disc, if separated, eighth abdominal segment not trilobed. Margin toothed. Pores on dorsum variously distributed. Dorsum with a prominent central ridge terminating cephalad in a more or less arrow-shaped figure, or elevated and fringed with rounded protrusions around. Lingula usually concealed. Aleyrodini Thoracic and caudal tracheal combs, clefts, or pores or furrows absent. Vasiform orifice not elevated; lingula generally exposed, knobbed, but in some concealed. Margin toothed or crenulate. Dorsum with rhachis and a pair of longitudinal fold in some genera. Seventh abdominal segment much shortened at the median area in many genera. Caudal furrow sometimes discernible. Bemisini Tracheal pores, clefts, folds, furrows present but combs absent. Submargin generally not separated from dorsal disc, if separated, eighth abdominal segment not trilobed. Rhachis present or absent. Dorsum generally with setae/ tubercles. Submargin with row of papillae. Margin with series of setae in some genera. Vasiform orifice subcordate or triangular or subrectangular, sometimes in a ribbed pyriform pit; lingula long or spatulate, setose and exposed. Dialeurodini Tracheal pores, clefts, folds, furrows present but combs absent. Submargin generally not separated from dorsal disc, if separated, eighth abdominal segment not trilobed. Rhachis present or absent. Dorsum generally with setae/tubercles. Submargin with row of papillae. Margin with series of setae in some genera. Vasiform orifice cordate or subcordate; lingula usually small, concealed and wanting setae. Lipaleyrodini Thoracic and caudal tracheal combs, clefts, or pores or furrows present. Submargin with wax plates in cluster arranged in a row. Neomaskellini Thoracic and caudal tracheal combs, clefts, or pores or furrows absent. Vasiform orifice elevated, elongately elliptic or transversely oval; operculum transversely rectangular and extremely short, exposing lingula. Margin with a single row of teeth or smooth. Submargin not separated from dorsal disc. Rhachis sometimes present. Submargin in some with a row of setae. Siphoninini Thoracic and caudal tracheal combs, clefts, or pores or furrows absent. Dorsum with elongate siphon-like wax tubes.