Molecular and morphological differentiation between Aphis gossypii Glover (Hemiptera, Aphididae) and related species, with particular reference to the North American Midwest

Abstract The cotton aphid, Aphis gossypii, is one of the most biologically diverse species of aphids; a polyphagous species in a family where most are host specialists. It is economically important and belongs to a group of closely related species that has challenged aphid taxonomy. The research presented here seeks to clarify the taxonomic relationships and status of species within the Aphid gossypii group in the North American Midwest. Sequences of the mitochondrial cytochrome oxidase 1 (COI), nuclear elongation factor 1-α (EF1-α), and nuclear sodium channel para-type (SCP) genes were used to differentiate between Aphid gossypii and related species. Aphis monardae, previously synonymised with Aphid gossypii, is re-established as a valid species. Phylogenetic analyses support the close relationship of members of the Aphid gossypii group native to North America (Aphid forbesi, Aphid monardae, Aphid oestlundi, Aphid rubifolii, and Aphid rubicola), Europe (Aphid nasturtii, Aphid urticata and Aphid sedi), and Asia (Aphid agrimoniae, Aphid clerodendri, Aphid glycines, Aphid gossypii, Aphid hypericiphaga, Aphid ichigicola, Aphid ichigo, Aphid sanguisorbicola, Aphid sumire and Aphid taraxicicola). The North American species most closely related to Aphid gossypii are Aphid monardae and Aphid oestlundi. The cosmopolitan Aphid gossypii and Aphid sedi identified in the USA are genetically very similar using COI and EF1-α sequences, but the SCP gene shows greater genetic distance between them. We present a discussion of the biological and morphological differentiation of these species.


Introduction
Host plant association is often one of the main characters used to distinguish between closely related aphid species. However, host association can also be one of the main sources of misidentification of host-alternating aphids. These aphids migrate between taxonomically distant hosts, usually between woody and herbaceous plants. Taxonomic problems have been created when aphid morphs from primary (woody) host plants have been treated as separate species from those found living on secondary (herbaceous) or summer host plants. Host alternation provides an opportunity for aphids to acquire new hosts and may be a key to the rapid diversification of some groups of aphids (Eastop 1971, Dixon 1973, von Dohlen and Moran 2000, thereby leading to hard-to-distinguish species complexes. The evolution of Aphis, the largest aphid genus by a margin, is associated with the rapid diversification of herbaceous angiosperms (Heie 1996).
The Aphis gossypii group contains economically important and taxonomically problematic species, with A. gossypii Glover itself being the most biologically diverse and hence taxonomically challenging (Blackman and Eastop 2007). It has many different primary and secondary host plants and exhibits both holocyclic and anholocyclic life cycles (Kring 1955, Blackman and Eastop 2006, Margaritopoulus et al. 2006. Its taxonomic complexity is attested to by its 42 available synonyms, including the native North American species, Aphis monardae Oestlund (Favret 2014). Eastop and Hille Ris Lambers (1976) established this synonymy without comment. Lagos (2007) found that A. monardae is distinct morphologically from A. gossypii and treated it as a valid species. No type specimen of A. monardae could be found at the time, and no molecular or biological evidence was available to support this decision. The research presented here contains both molecular and biological evidence as well as an examination of material collected by Oestlund from Monarda spp.
In Europe, there are approximately 20 aphid species morphologically similar to A. gossypii (Stroyan 1984, Heie 1986. Several studies using mitochondrial, nuclear, and intron length polymorphism in the sodium channel para-type (SCP) genes have achieved some resolution discriminating A. gossypii and other Aphis species (Coeur d'acier et al. 2007, Foottit et al. 2008, Coccuzza et al. 2009, Carletto et al. 2009b, Kim et al. 2010a, Komazaki et al. 2010, Favret and Miller 2011. In North America, morphological studies show that species of the A. gossypii group can be misidentified easily (Voegtlin et al. 2004, Lagos 2007. The discrimination of species closely related to A. gossypii is of particular importance due to the recent introduction into North America of the soybean aphid, A. glycines Matsumura (Voegtlin et al. 2004). This species is obligately holocyclic and heteroecious, feeding on soybean, Glycine max (L.) Merr., as secondary host, and on Rhamnus spp. as primary host. Aphis gossypii has also been reported to colonize soybean in North America (Blackman and Eastop 2007), and while its colonization on soybeans is uncommon in the north central United States, some soybean-collected insect samples from Alabama, Georgia, Kansas, Louisiana, and Mississippi contained only A. gossypii or a mixture of both species (personal observation, Illinois Natural History Survey (INHS) insect collection records). These collections suggest that A. gossypii may be more common on soybeans in southern regions. There are no records of the exotic A. nasturtii Kaltenbach feeding on soybeans, and attempts to culture A. nasturtii on soybeans were not successful (David Ragsdale, personal communication); however, this species shares the primary host, Rhamnus spp., with both A. glycines and A. gossypii.
We here elucidate the phylogenetic relationship of species morphologically close to A. gossypii and the taxonomic status of A. monardae in the North American Midwest.

Materials and methods
Aphid collections: Aphids were collected from their primary and/or secondary host plants from different sites in China, France, Italy, Japan, Spain and the USA, with the majority of the material originating from the Midwest of the USA. When possible, aphids were collected alive and reared on the host plant for the maturation of late instar nymphs. Adults were preserved in 95% ethanol and stored at -20°C until DNA extraction and microscope slide preparation. Collection data with INHS Insect Collection specimen voucher numbers are presented in Suppl. material 1.
Morphology: Archival microscope slides were prepared using the technique described by Pike et al. (1991). Individuals were selected from the same colonies as those selected for DNA extraction. Photographs of mounted specimens and measurements were taken using a Leica DM 2000 digital camera and SPOT Software 4.6 (Diagnostic Instruments, Inc). Analyses of variance of diagnostic characters, such as the distance from the base of the third antennal segment to the first secondary sensorium, the ratio of the lengths of the processus terminalis and the base of the sixth antennal segment, and the ratio of the lengths of the siphunculus and the cauda, were tested using JMP, Version 7 (SAS Institute Inc., Cary, NC, 1989-2007. Species identification of slidemounted material was done by the first author, using published keys (Oestlund 1887, Gillette 1927, Hottes and Frison 1931, Palmer 1952, Kring 1955, Cook 1984, Voegtlin et al. 2004) and authoritatively identified specimens in the insect collections of the INHS and the University of Minnesota. Identifications of slide-mounted specimens were referenced to the aphid colony-mates used in the molecular analyses.
Phylogenetic analysis: Modeltest 3.7 (Posada and Crandall 1998) was used to select the best-fit nucleotide substitution model. Single sets of gene sequences were analyzed using MrBayes 3.1.2 (Huelsenbeck and Ronquist 2003) to execute Bayesian analyses. For single analysis, four chains were run. The number of generations was 5,000,000 with a burn-in of 250 trees and frequency sampling of 100 generations with rates equal to variable gamma as a model of substitution of nucleotides. Rhopalosiphum maidis (Fitch) (Aphidinae: Aphidini), and Hyadaphis tataricae Aizenberg and Uroleucon helianthicola (Olive) (Aphidinae: Macrosiphini) were selected as outgroups.
Aphid biology: Two growth chambers were used to examine various aspects of the biology of A. monardae, A. gossypii, and A. sedi Kaltenbach, in order to discern differences in their life cycle. Experimental plants were grown in a greenhouse in 12.7 cm diameter pots and isolated in 13.5 by 13.5 by 22.5 inches cages. Chamber A was set at 12°C and short photoperiod (8L:16D), conditions that will trigger the development of sexual morphs. Colonies of A. monardae on Monarda fistulosa L., A. sedi on Hylotelephium telephium (L.) H.Ohba, and A. gossypii on Cucurbita pepo L. and Rhamnus cathartica L. were exposed to these conditions for extended lengths of time. Samples of A. monardae and A. sedi were collected on a weekly basis from the host plants listed above and examined for the presence of sexual morphs. In the cages of A. gossypii weekly samples were taken from R. cathartica.
The B chamber was set at 24°C with constant illumination (24 hours) to keep colonies and test host plant specificity of the three species mentioned above. The following experiments were done in chamber B: a Monarda fistulosa plant infested with A. monardae was placed into a cage with an aphid-free C. pepo plant and left for a several weeks. Biweekly examination of the C. pepo plants was made to determine if A. monardae had colonized them. A Cucurbita pepo plant infested with A. gossypii was placed into a cage with aphid-free M. fistulosa and H. telephium and left for several weeks. Biweekly examination of M. fistulosa and H. telephium was made to see if A. gossypii had colonized them. A Hylotelephium telephium plant infested with A. sedi was placed into a cage with aphid-free C. pepo and left for several weeks. Biweekly examination of C. pepo was made to see if A. sedi had transferred to them. An entire tree of R. cathartica infested with A. gossypii was isolated in a 2 by 2 by 2-m walk-in cage in May of 2011 on the grounds of the South Farms of the University of Illinois (Suppl. material 1). The temperature ranged between 10 and 22 °C, http://www.isws.illinois.edu/atmos/ statecli/cuweather/. Aphid-free C. pepo, H. telephium and Glycine max were placed into the cage to document the potential infestation of these secondary hosts under natural environmental conditions.

Phylogenetic analysis
A total of 160 COI sequences from 28 species, 133 EF1-α sequences from 36 species, and 13 SCP sequences from 6 species were used in this study. After alignment and excluding the primer sites, 1,290, 1,078 and 703 bp for COI, EF1-α (including gaps and introns) and SCP were used in the analysis, respectively. COI sequence divergence between species of the A. gossypii species group ranged from 0.08% (between A. gossypii and A. sedi) to 3.04% (between A. gossypii and A. monardae). The sequence divergence of A. glycines and A. nasturtii (sharing a winter host plant with A. gossypii), as compared with the species of the gossypii group, ranged from 5.25% (between A. gossypii and A. glycines) to 6.97% (between A. nasturtii and A. sedi) ( Table 1). The sequence divergences of EF1-α and SCP are presented in Table 2. Generally, COI sequences were more conserved than EF1-α, which in turn were more conserved than SCP.
The cladograms using COI ( Figure 1) and EF1-α ( Figure 2) showed a high level of agreement. The COI analysis supported the monophyly of a group of species (Clade A) related to A. gossypii, including A. glycines, A. nasturtii, and a still more closely related group that are regarded as members of the A. gossypii complex (Clade D). Within Clade A are several supported groups: Clade H of A. rubifolii (Thomas) and A. rubicola Oestlund (PP:1.00); Clade I of A. nasturtii and A. urticata Gmelin (PP:1.00); Clade B of A. glycines with the A. gossypii complex (PP:0.99). The A. gossypii complex is itself well supported (Clade D, PP:0.99), and includese two groups: Clade E of A gossypii and A. sedi (PP:0.99) and Clade F of A. oestlundi Gillette, A. monardae, and several possible new species.
The dendrogram inferred by MrBayes using EF1-α ( Figure 2) is congruent with that of COI for some taxa mentioned above (clade A, PP:0.99), although lack of resolution prevented recovery of a monophyletic Midwest A. gossypii complex. The close relationship of A. nasturtii and A. urticata is robustly supported (Clade G, PP:1.00) as is clade B (PP:0.99). Clade A in the COI analysis is polyphyletic in the EF1-α analysis and includes Asian species A. ichigicola Shinji and A. ichigo Shinji (Clade F, PP:0.97); A. glycines and A. sanguisorbicola Takahashi (Clade E, PP:0.98). Clade C is poorly supported and presents polytomies of species closely related to A. gossypii: Table 1. Range of Kimura 2 Parameter pair-wise inter-and intraspecific sequence divergence (%) for COI sequences.

Biological evidence
After four weeks under conditions of reduced temperature and photoperiod, colonies of A. monardae reared on M. fistulosa produced oviparae and apterous males ( Figure  4A-B). Also, sexual morphs were collected in the field (Middlefork Savanna, Lake County) at the beginning of October (Suppl. material 1). Aphis sedi on H. telephium produced oviparae ( Figure 4C) but no males were found. Voucher slides of both species are deposited in the INHS insect collection with the following catalog numbers: A. monardae, 512858-512865; A. sedi, 511202-511208 and 511559-511573. In chamber A, no sexual morphs of A. gossypii were found after two months exposure to the low temperatures and reduced photoperiod and weekly collections on R. cathartica.
The outdoor experiments located at the South farms of the University of Illinois were evaluated after 25 days. Alate viviparae of A. gossypii were seen on H. telephium and G. max but they did not produce offspring, however, alates that moved to C. pepo did produce apterous and alate viviparae. Voucher slides are deposited in the INHS insect collection numbers: 512851-512857. The colonies of A. gossypii reared on C. pepo were set in a growth chamber B where they grew rapidly. Potted M. fistulosa were placed in this chamber and were colonized by A. gossypii. Clean plants of C. pepo that were later exposed in the same chamber to a colony of A. monardae were not colonized. A colony of A. sedi begun with fundatrices from H. telephium was exposed to C. pepo in growth chamber B for several weeks, but the aphids did not transfer to and establish on this plant.

Comparison of Aphis monardae and Aphis gossypii
In both the COI and EF1-α analyses, A. monardae was readily distinguished from A. gossypii (Figure 1 Clade G, Figure 2 Clade D). Aphis monardae and A. gossypii are also differentiable morphologically: 1) the siphunculi of apterous morph are darker in A. gossypii than in A. monardae, and 2) and secondary sensoria on antennal segment IV are always absent in alate viviparae of A. gossypii, but present in A. monardae (Suppl. material 2). A third, novel morphological character, the distance from the base of antennal segment III to the first secondary sensorium (DBIII) in alate viviparae also separates these species consistently. In A. gossypii, the secondary sensoria are uniformly distributed along the segment ( Figure 6B) but not in A. monardae ( Figure 6C). The means of the distance from the base of antennal segment III to the basal margin of the first secondary sensorium of A. gossypii and A. monardae are 0.06 and 0.08 mm, respectively ( Figure 5A, F ratio=152.3, df=1, P<0.0001). Evidence in support of the reproductive isolation of this species is the presence of oviparae ( Figure 4A) and apterous males of A. monardae ( Figure 4B) on M. fistulosa (INHS insect collection numbers: 511335-511344 and 512858-512865, respectively), as well as a COI sequence divergence of 2.7-3.04% between A. gossypii and A. monardae (Table 1).
Apterous viviparae (n= 40). Color in life ( Figure 8B): Head, thorax and abdomen vary from light yellow to light green. Color of cleared specimens ( Figure 4I): Head: dusky. All antennal segments pale, except the sixth throughout, which is dusky. Secondary sensoria absent. URS does not reach the hind coxae. Thorax: Coxae, trochanters and all femora dusky. All hind tibiae dusky and dark distally. Abdomen: Cauda first and second dark, the rest dusky. Secondary sensoria present on and III and IV. Arrangement of secondary sensoria in a single row on the distal half ( Figure 6C). Thorax: All femora dusky except in the base. Hind coxa dark. Hind trochanters paler than coxa. Hind tibiae dark distally. Abdomen: Cauda pale or slightly dusky. The cauda parallel-sided with constriction near the base. Siphunculi dark throughout,imbricated with flange. Pre-siphuncular sclerites absent. Post-siphuncular sclerites dusky. Marginal sclerites pale. Marginal tubercles only present on abdominal segments I and VII. Dorsal abdomen with small transverse sclerites on VI, VII and VIII. Abdominal tergite VIII with 2 setae. Subgenital plate complete, slightly dusky, with 2-7 setae on anterior part. Cuticle without reticulation.
Alate male (n=17). Color in life ( Figure 8D): Head: brownish. Antennae: blackish. Thorax: greenish. Legs light brown and tibiae distally dark as well as tarsi. Abdomen: Cauda dark green. Siphunculi lighter than dark green abdomen. Color on slide and morphological characters ( Figure 4B): Head: dark. Antennae dark with secondary sensoria scattered on segments III, IV, and V. Abdomen: Cauda pale or dusky, parallel-sided with blunt tip and bearing 3-6 setae. Marginal tubercles present on abdominal segments I and VII. Dorsum of abdomen without large transverse sclerites. Male genitalia with 2 short claspers anteriorly and aedeagus centrally.

Comparison of Aphis sedi and Aphis gossypii
The distinction of A. sedi from A. gossypii is supported by phenotypic characters of specimens in collections included in Tables S1 and S2. In addition, morphological characters such as the ratio of the lengths of the processus terminalis and the base of the sixth antennal segment (Suppl. material 2, Figure 5B: F ratio=498.1, df=1, P<0.001) and the ratio of the lengths of the siphunculus and the cauda (Suppl. material 2, Figure  5C: F ratio=168.5, df=1, P<0.001) of apterous viviparae can be useful to discriminate these species. Interestingly, only oviparae of A. sedi reared on Hylotelephium telephium were collected under laboratory conditions (Figures 4C and 8G). In contrast with the morphological differences, the interspecific genetic divergences using COI and EF1-α sequences of A. gossypii and A. sedi are less than 1% (Tables 1 and 2). SCP showed greater genetic divergence between these two species, namely 0.84-1.84% (Table 2).

Comparison of Aphis gossypii with Aphis forbesi, Aphis glycines and Aphis nasturtii
These species are sometimes misidentified because they share some morphological characters on either apterous or alate morphs. Moreover, the pair-wise sequence divergences using COI sequences between A. gossypii and A. forbesi Weed, A. glycines and A. nasturtii are up to 5% (Table 1). Here we present some characters that can be useful for their discrimination. Apterous viviparae of A. gossypii can be differentiated from those of A. forbesi by the width of the marginal tubercles on abdominal segments I and VII (maximally 0.011 in A. gossypii and minimally 0.025 in A. forbesi; range for A. gossypii is given in Suppl. material 2), number of antennal segments and color pattern of siphunculi. Apterae of A. gossypii are differentiated from those of A. glycines by the shape of the cauda (Figures D-F, H) and the number of caudal setae, and from those of A. nasturtii by the absence of marginal tubercles on abdominal segments II and VI ( Figure  4J). Alate viviparae of Aphis gossypii can be differentiated from those of A. forbesi by the number of secondary sensoria on III and the DBIII (Suppl. material 2), from those of A. glycines by the color of the hind coxae and marginal sclerites ( Figure 7A) and from those of A. nasturtii by the number of secondary sensoria on antennal segments III, IV and V, absence of marginal tubercles on abdominal segments II and VI, and shape of cauda ( Figure 7F). More figures and morphological characters have been uploaded in Lagos et al. 2014a.

Dichotomous keys to apterous and alate viviparous females of the Aphis gossypii complex in the Midwest
Many dichotomous keys to subsets of Aphis have been written (Hottes and Frison 1931, Palmer 1952, Rojanavongse and Robinson 1977, Cook 1984, Stroyan 1984, Heie 1986, Brown 1989, García Prieto et al. 2005, Blackman and Eastop 2006 when morphological characters were not useful to discriminate between species, host plant associations have been used. Unfortunately, in the Midwest A. gossypii has been found on most of the host plants of other Aphis species included in this complex (A. gossypii, A. monardae stat. nov., A. oestlundi and A. sedi). The alternative key that we present below is based on specimens from collections made in the Midwest, and molecular data for specimens from these collections (Tables 1 and 2) supports our morphologically based identifications. Morphological data for these species is shown in Suppl. material 2. For some comparative morphometric data of European specimens of A. gossypii and A. sedi see Stroyan (1984), Heie (1986), Brown (1989) and García Prieto et al. (2005).
The key is specific to Midwest collected specimens and may not be reliable in other geographic regions. It also demonstrates the difficulty of separating these closely related species using only morphological characters.

Discussion
The analysis of different species included in this study largely corroborates the results obtained by Coeur d'acier et al. (2007), Kim and Lee (2008), Kim et al. (2010a), Kim et al. (2010b), Kim et al. (2011) andLagos et al. (2014b). The gossypii complex in the North American Midwest contains the following native species, A. oestlundi and A. monardae, and the invasive species A. gossypii and A. sedi. Collection host records for A. gossypii show that it has been collected on Oenothera and Monarda, the host plants of the native Aphis species listed above (Blackman and Eastop 2006). Collection records for the native species suggest a very limited host range, in contrast with the highly polyphagous A. gossypii.
Our results indicate that these species can be differentiated by morphological characters as well as host association. Data from this study confirms the finding of Lagos (2007) that A. monardae is a valid species and not a synonym of A. gossypii (Eastop and Hille Ris Lambers 1976 (Oestlund 1887). The comparison of apterae, alatae and oviparae of Oestlund's collections match the morphological characters of those collected recently (Suppl. material 1). Some slides made by Oestlund were remounted in 1968 so it was possible to better see the characters. For a neotype we chose a more recently collected specimen taken in Minnesota as it more clearly shows color pattern and other characters used in the redescription. The discrimination of A. gossypii and A. sedi is clear when the aphids are alive (Figure 8). The identification problem arises when we examine samples that have lost their color by being stored in ethanol. Molecular data also are helpful. The pair-wise sequences divergences between these species using SCP are higher than for COI and EF1-α sequences (Tables 1 and 2). This marker also successfully differentiated the cryptic species A. gossypii and A. frangulae (Carletto et al. 2009a). Results obtained in this study corroborate the biological and morphological findings of Kring (1955), who found that A. sedi is holocyclic monoecious on Hylotelephium. In this study, only apterous oviparae were collected under laboratory conditions conducive to the production of sexuales ( Figure 4C). Kring's morphological observations showed that the ratio of the processus terminalis to the base of the last antennal segment (PT/B), and the ratio of the length of the siphunculus to the length of the cauda (SIPH/CA), are both greater in A. gossypii than A. sedi for all morphs (Suppl. material 2). Although the above characters are useful to differentiate these species, their identification (especially the alate viviparae) is still problematic because of their similar morphology and because these ratios overlap ( Figures 5B-C).
The inclusion of A. glycines, A. gossypii and A. nasturtii in strongly supported clades (Clade A, Figures 1 and 2) is consistent with the findings of Foottit et al. (2008) but in disagreement with those of Kim et al. (2010a). Interestingly, these three invasive species share a winter host plant, Rhamnus spp., but this is not the only known overwintering host for A. gossypii. This indiscriminate behavior, in addition to multiple species sharing winter hosts, raises the possibility of interspecies hybridization (Müller 1986, Rakauskas 2003. Hybridization may or may not be successful but should be detectable in studies of gene flow and phenotypic characterization of putative hybrids. The species regarded here as members of the A. gossypii complex, A. gossypii, A. sedi, A. oestlundi and A. monardae (Clade D), exhibit interesting biological, morphological and molecular patterns. Aphis gossypii has been shown to colonize numerous secondary host plants including those of closely related taxa (Stroyan 1984, Heie 1986, Blackman and Eastop 2006. Moreover, it is one of the few Aphis species with multiple primary host plants (Blackman and Eastop 2006). By contrast, the native taxa related to it and found in the Midwest have or are presumed to have monoecious holocyclic life cycles (see Suppl. material 1 for host plant information). Aphis oestlundi, A. monardae and A. sedi have wingless males, a characteristic that would contribute to the genetic isolation of these species. These sibling species possess morphological characters useful for diagnostic purposes (Suppl. material 2) and the values that support interspecific sequences divergences (Table 2) are similar to those found by Foottit et al. (2008) and Favret and Miller (2011). The identification of species related to A. gossypii is made more difficult because they feed on host plants that can also serve as host to A. gossypii. Interestingly, however, their colors in life differ and can be useful for identification. For example, A. gossypii is dark green or light brownish and its siphunculi are dark throughout ( Figures  8A, E), although this can vary in summer dwarf specimens. Aphis gossypii is mostly darker than A. monardae, which is light yellow or green ( Figures 8B, C), and A. oestlundi is light yellow ( Figure 8F). The color of A. sedi is dark green ( Figure 8G), like A. gossypii, although it has more white wax on its body (García Prieto et al. 2005).
The COI sequence divergence values obtained in this study are similar to those obtained in other studies (Cocuzza et al. 2009, Cognato 2006, Coeur d'acier et al. 2007, Foottit et al. 2008, Favret and Miller 2011, Wang and Qiao 2009. Moreover, the low pair-wise sequence divergences found between some species such as A. gossypii and A. sedi (Table 1) are consistent with those obtained by other workers such as Piffaretti et al. (2012). While COI data have been found useful to discern the phylogenetic relationships of many taxa, the use of COI sequence divergences to set cut-off points that can differentiate Aphis species should be used with caution, since it may lead to the misidentification of new species, a conclusion drawn by other studies for several orders of insects (Blaxter 2004, Hebert et al. 2004, Nadler 2002, Will and Rubinoff 2004, Smith et al. 2008).
Our work suggests the possible existence of three undescribed Midwestern species (Aphis spp. 1, 2, and 3) within the gossypii complex. Further studies need to be done to validate their status. It is likely that additional new species will be found within this group as material is gathered from a larger geographical area and combined molecular, morphological and biological data are used to analyze the new taxa. The use of multiple primary hosts is unusual for any species, thus lineages within the gossypii complex that select and limit themselves to specific hosts may be driving the speciation process within this group (Peccoud et al. 2010, Kim et al. 2011 grateful to Armelle Coeur d'acier, Thelma Heidel, Wayne Ohnesorg, Benjamin Puttler and Andrew Williams for supplying with specimens used in this study. Many thanks to the associate curator of the University of Minnesota Insect Collection, Robin Thomson, who kindly sent Oestlund's slides and provided quick correspondence. We gratefully acknowledge the comments and examination of specimens by Drs. Juan Nieto Nafría, Shun'ichiro Sugimoto and Giuseppe E. Cocuzza. The manuscript benefited greatly from the comments provided by Susan Halbert, two anonymous reviewers, and the subject editor, Roger Blackman.