Distribution of bisexual and unisexual species in the aphid genus Colopha Monell (Aphididae: Eriosomatinae), with the description of a new species in Japan

Aphids of the genus Colopha are represented by three bisexual and three unisexual species. The bisexual species are associated with two Ulmus species that are most closely related, being distributed disjunctively in Europe and eastern North America on the host plants. A new unisexual species of Colopha, collected from Setaria chondrachne (Poaceae) in Japan, is described under the name Colopha setaricola sp. nov. The distribution of the three unisexual species is discussed in relation to aphid‐plant associations and historical changes in the distribution of the host plants. It is suggested that these unisexual species have persisted on the secondary hosts through parthenogenetic reproduction since the extinction of the primary host plants, Ulmus species section Blepharocarpus. Therefore, the unisexual species in Colopha exemplify Mordvilko's hypothesis that unisexual species on the secondary host plants have remained as relics in the region where the primary host was once distributed but then became extinct.


Introduction
Aphids of the Eriosomatine and Hormaphidine are typically gall formers on broad-leaved deciduous trees and highly specific to their host plants (Blackman and Eastop 2000). Biogeographic evidence suggests that their host associations have a long history tracing back to the early Tertiary (Akimoto 1985;Moran 1989). Throughout the Tertiary and Quaternary, hardwood taxa drastically shifted their distributional ranges depending on global climatic changes, which have resulted in the disjunctive distributions of some taxa, for example in East Asia and eastern North America (Pielou 1979;Tiffney 1985a;Wen 1999). Hardwood trees adapted to mild climates are known to have survived glacial epochs in some fragmented refuges and expanded their distribution to the north in interglacial epochs, with the retreat and expansion of their distribution having been repeated during the
Eyes with three ommatidia and, in the space surrounded by them, with about 15 small sclerotized tubercles. Antennae (Figure 3) five-segmented, 0.145-0.170 (0.157) mm long, 0.20-0.24 (0.22) times as long as body, 1.02-1.21 (1.13) times the length of hind femorotrochanter. Antennal segments I-III smooth, segments IV and V spinulose. Antennal setae sparse; setal arrangement on each segment as follows: I 3 (one of which is feeble), II 2, III 0, IV 4, and V 2 + 5. Primary rhinarium surrounded by cilia in a single row, with a circular opening, from which a semi-transparent tonguelet projects. Tonguelet on segment IV small and furcated, that on segment V extending apically like a horn. Some accessory rhinaria present close to primary one, surrounded by cilia in a single row, without a tonguelet. Rostrum reaching abdominal segment IV or V. Ultimate segment (Figure 4) spinulose, with two pairs of accessory setae, 0.075-0.88 (0.82) mm long, 0.55-0.63 (0.59) times the length of hind femorotrochanter.
Wax gland plates ( Figure 6) circular or elliptical, consisting of one or two central fields and circumferential cells. Central field microscopically bright, with a broad blackish rim, usually larger than circumferential cells. Circumferential cells with minute blackish points scattered over, surrounding the central field(s) in a single layer. Head (Figures 1, 2) with five pairs of wax gland plates; on the dorsum one pair medio-posteriorly and one pair latero-posteriorly; on the frons one pair; on the ventrum one pair medio-anteriorly and one pair latero-facially. Prothorax with one dorso-spinal pair of wax gland plate and one ventrolateral pair. Meso-and metathorax and abdominal segments I-VI each with three pairs at dorso-spinal, dorso-pleural and ventro-lateral positions. Abdominal segment VII with one dorso-lateral and one ventro-lateral pairs. Lateral pair rather located on the dorsal side on anterior segments, especially on thorax.
Body setae all simple and short. Head (Figures 1, 2) with eight pairs of setae; on the dorsum one pair outside medio-posterior wax gland plates, one pair outside latero-posterior wax gland plates, one pair laterally just above antennal bases, and one pair medioanteriorly; on the frons one pair; on the ventrum three pairs facially. Pronotum with two spinal pairs of setae and two lateral pairs. Meso-and metanotum each with one spinal, one pleural and two lateral pairs. Abdominal tergites I-VI each with three pairs at spinal, pleural and lateral positions. Abdominal tergite VII with one spinal and one lateral pair. Abdominal tergite VIII with one pleural pair. Cauda with two setae. Abdominal sternite VIII with four setae.
Abdominal segments I-VI each with one pair of sclerotized bands, on which spinal and pleural setae, and spinal and pleural wax gland plates are situated. Abdominal segments VII and VIII sclerotized dorsally.
Eyes with about 10 ommatidia, without small sclerotized tubercles in the space surrounded by them. Arrangement of cephalic setae and wax gland plates (Figure 7) as in the first instar exule, but some wax gland plates rudimentary or absent. Rostrum short, reaching a little beyond fore coxae. Antennae (Figure 8) six-segmented with the division between segments III and IV sometimes incomplete, or rarely five-segmented with segment III completely fused with IV, 0.198-0.251 (0.223) mm long, 0.12-0.16 (0.13) times as long as body, 0.89-1.15 (0.98) times the length of hind femorotrochanter. Primary rhinarium not transversely extended, surrounded by cilia in a single row, with a circular opening, from which a semi-transparent tonguelet projects. Some accessory rhinaria present on segment VI. Antennal setae short and scarce; their arrangement as follows, I 2-3 (one of which is feeble), II 2, III 0-1, IV 0, V 4, and VI 2+425. Ultimate segment ( Figure  9) slightly spinulose, 0.070-0.093 (0.081) mm long, 0.32-0.38 (0.35) times as long as hind femorotrochanter, with two pairs of accessory setae.
Wax gland plates ( Figure 12) as in the first instar exule, but central fields weakly rimmed as in circumferential cells with rough blackish points scattered over and sometimes not distinguishable from circumferential cells. Arrangement of wax gland plates as in the first instar exule. Body setae short, as long as the diameter of wax gland plates, arranged in a transverse row. Cornicles absent. Genital plate ( Figure 13) consisting of two lobes, altogether with 10-17 setae, two of which are much stouter and longer and situated at the centre of each lobe. Anal plate ( Figure 13) with slightly lifting three lobes, with 11-18 setae and 5-9 feeble setae, the latter of which are scattered along anterior margin of a depression surrounded by three lobes. Cauda with two setae.

Sexupara
Materials from the two localities included males and bisexual females in the abdomen, suggesting that they are sexuparae, not alate exules.
Antennae ( Figure 14) six-segmented. Antennal segment III the longest, as long as segments IV, V and VI combined, 0.23-0.28 (0.25) mm long. Segment IV a little longer than segment V, but shorter than segments V and VI combined, 0.085-0.11 (0.097) mm long. Segment V a little longer than segment VI, 0.078-0.090 (0.83) mm long. Segment VI slightly spinulose, 0.065-0.083 (0.74) mm long. Antennal setae short, as long as basal width of segment III, arrangement of which as follows; II 2, III 1-6, IV 0-1, V 4, and VI 2+425. Secondary rhinaria microscopically represented as bright lines, present on segments III-VI. Those on segments III-V covering usually about half the circumference from the ventral side; some rhinaria much shorter. Secondary rhinarium on segment VI short and rudimentary. Primary rhinarium on segment V not ciliated, without an opening, always connected with the apical secondary rhinarium. Primary rhinarium on segment VI slightly ciliated with a semi-transparent tonguelet, often connected with the apical secondary rhinarium, which is more irregular in shape and sometimes thicker than other secondary ones. Some accessory rhinaria present dorsally on segment VI. Antennal segments III, IV, V, and VI with 12-18, 3-6, 3-5, and 1-3 secondary rhinaria, respectively. Head ( Figure 15) dorsally with five pairs of minute setae; one pair medio-posteriorly, one pair latero-posteriorly, one pair medio-anteriorly, one pair latero-posteriorly and one pair frontally. Rostrum short, a little exceeding fore coxae. Ultimate segment (Figure 16) slender, spinulose, with three to five accessory setae, 0.088-0.11 (0.094) mm long, 0.83-1.02 (0.90) times as long as the second segment of hind tarsus.
Cornicles absent. Cauda with two to six setae. Genital plate ( Figure 17) depressed in the middle, forming two lobes, altogether with 20-26 setae mainly scattered on lobes. Anal plate ( Figure 17) with 17-25 setae. Fore wings with once-branched media. Hind wings with two oblique veins.

Host plant
Setaria chondrachne (Poaceae). This species forms colonies on the aerial parts of the secondary host, and the colonies are densely covered with a large amount of wax ( Figure 18). Primary host plant unknown.

Remarks
Adult exules of C. setaricola can be distinguished from those of other Colopha species in the antennae, eyes and wax gland plates. C. setaricola has usually six-segmented antennae, while other species have four-or five-segmented antennae. C. setaricola has eyes consisting of about 10 ommatidia, while other species have only three ommatidia. In C. setaricola the wax gland plates have a central field that is weakly rimmed as in circumferential cells, while in other Colopha species wax gland plates have a conspicuously rimmed central field.
First instar exules of C. setaricola can be distinguished from those of C. kansugei, the only species described for this stage, by the abdomen, tarsi and ultimate rostral segment; C. setaricola has sclerotized bands on abdominal tergites I-VII, while C. kansugei has abdomen not sclerotized; the ultimate rostral segment of C. setaricola is densely spinulose, while that of C. kansugei is very slightly spinulose; C. setaricola has symmetrical claws and symmetrical dorso-apical setae on middle and hind tarsi, while C. kansugei has asymmetrical claws and asymmetrical dorso-apical setae on middle and hind tarsi.

Life cycle
It is likely that this species reproduces parthenogenetically on the secondary host plants, although sexuparae are produced in autumn. No alate exules have been found. Despite our extensive survey and a long history of galling aphid studies in Japan (Matsumura 1917;Monzen 1929;Shinji 1941Shinji , 1944Yukawa and Masuda 1996), no galls of Colopha species have been found on Ulmus spp. or Zelkova serrata.

Distribution and host plants of other Colopha species
Approximate distributional ranges of six Colopha species are illustrated in Figure 19, and the primary and secondary host species for six Colopha species are tabulated in Table I.
C. kansugei is know as an anholocyclic (unisexual) species, having a wide range stretching from Nepal through China to south-western Japan (Akimoto 1985) and south to northern Thailand (Sano and Akimoto unpublished). This species is associated with Carex lenta in Japan (Moritsu 1983;Matsumoto 1999Matsumoto , 2000 and Carex brunnea in Taiwan (Takahashi 1937). A record from Carex morrowii by Uye (1924) is doubtful, because at the time of the record several species of Carex were confused under the name of Carex morrowii; C. kansugei has not been found from Carex morrowii (Matsumoto 1999(Matsumoto , 2000. C. graminis and C. ulmicola are holocyclic (bisexual) and widely distributed in North America (Smith and Parron 1978;Blackman and Eastop 1994). Both species are collected mainly from U. americana, and occasionally from U. rubra (5fulva) as the primary host, and from the roots of some grass species as the secondary host (Patch 1910;Hottes and Frison 1931;Gillette and Palmer 1934;Knowlton 1983;Smith 1985).

Mordvilko's hypothesis
Most host-alternating species of eriosomatine and hormaphidine aphids are able to persist on the secondary host parthenogenetically all year round. Obligatorily or facultatively unisexual strains may often coexist with bisexual strains on the secondary host (Bodenheimer and Swirski 1957;Moran 1991). Obligatorily unisexual populations are sometimes distributed beyond the ranges of the primary host plant, as shown in C. compressa. A few species consist of unisexual populations alone, as in C. setaricola. Mordvilko (1935) hypothesized that unisexual strains would be distributed outside the range of the primary host if bisexual strains became extinct due to local extinction of the primary host in cold and arid climates but unisexual strains survived on the more coldtolerant secondary host. Mordvilko (1935) illustrated this theory with Tetraneura rubra (5T. caerulescens) that is found on the secondary host in Egypt where the primary host (Ulmus spp.) does not occur at present. If Mordvilko's hypothesis is true, unisexual populations may have reproduced parthenogenetically over a long period of time since the extinction of the primary host. Based on molecular phylogeny, for example, von Dohlen et al. (2002) inferred that a unisexual North American Hamamelistes species has persisted on the secondary host for 2-4 million years.
Several eriosomatine and hormaphidine species can migrate over a long distance through dispersal of alate exules that fly from secondary host to secondary host. Some authors (e.g. Bodenheimer and Swirski 1957;Hille Ris Lambers 1970) criticized Mordvilko's hypothesis on the grounds that unisexual strains can expand their range by alate migration beyond the range of the primary host where bisexual and unisexual strains coexist. If this criticism is true, the origins of unisexual populations should be more recent events than Mordvilko's hypothesis predicts. Whether Mordvilko's hypothesis is applicable to the origin of two East Asian unisexual species, C. setaricola and C. kansugei, is evaluated by invoking fossil records of the host plants.

Disjunctive distribution and Land Bridge
The bisexual species C. compressa, C. graminis and C. ulmicola are mostly recorded from Ulmus laevis or U. americana, with occasional records from other Ulmus species (Table I). A recent study based on molecular phylogeny indicates that U. laevis and U. americana are sister species, comprising the section Blepharocarpus (Wiegrefe et al. 1994), distributed disjunctively in Europe (U. laevis) and eastern North America (U. americana) (Tutin 1964;Bate-Smith and Richens 1973;Sherman-Broyles et al. 1997). Disjunctions between East Asia and eastern or western North America are more common patterns of intercontinental distribution in temperate biota, so the distribution pattern of this section requires explanation. Intercontinental disjunctions have usually been explained by ancestral, continuous distribution across the Bering Land Bridge or the Atlantic Land Bridge and subsequent separation and isolation into two continents. The Bering Land Bridge that connected East Asia and North America was inhabited by deciduous broad-leaved trees from the Eocene to Middle Miocene, whereas the Atlantic Land Bridge that connected Europe and eastern North America was available for dispersion of temperate biota from the Paleocene to the Early Eocene (reviewed by Pielou 1979;Wen 1999;Sanmartín et al. 2001).
The fossils of Ulmus fruits that provide diagnostic characters for the genus are traceable back to the sediment of the Eocene in western North America and the Oligocene in East Asia (Manchester 1989). Fossil leaves of Ulmus are recorded from the sediment of the Eocene in Japan (Endo 1968;Tanai 1972) although fossil records without fruits are less reliable for identification. In Europe, although a number of fossil records of Ulmus fruits and leaves have been documented, none of them is found before the middle Oligocene (Manchester 1989). Dispersion of Ulmus through the Atlantic Land Bridge would predict much older records of Ulmus fossils in Europe. Temperate broad-leaved deciduous forests had expanded from East Asia to western North America through the Beringia by the Middle Miocene, with the same species found both from Asia and from Alaska, such as Ulmus newberryi and Zelkova oregoniana (Wolfe and Leopold 1967). These lines of evidence are consistent with the hypothesis of the dispersion of Blepharocarpus across the Bering Land Bridge rather than across the Atlantic Land Bridge. If this hypothesis is true, the section Blepharocarpus may have had a continuous and extensive distribution from Eurasia to North America.

Origin of Colopha and unisexuality
Gall-forming aphids are highly host specific (Eastop 1973;Wool 1984). The three bisexual species form closed, cock's comb-shaped galls Eastop 1994, 2000) on Ulmus species of Blepharocarpus. This gall shape and the morphology of fore tarsi in exules are peculiar to Colopha among eriosomatine aphids, suggesting that Colopha is monophyletic. If the speciation and diversification of Colopha species have proceeded in association with this section of Ulmus, the origin of Colopha can trace back prior to the Middle Miocene when the section Blepharocarpus is considered to have expanded in Asia and North America through the Bering Land Bridge. The ancestral stock of Colopha may have had an extensive distribution on this section of Ulmus, resulting in the present disjunctive distribution.
Two unisexual species, C. setaricola and C. kansugei, are distributed in East Asia where no Ulmus species of the section Blepharocarpus are present. However, fossils of this section were found from the Tertiary flora of Japan and Kazakhstan (Grudzinskaya 1967). Furthermore, Tanai (1961) pointed out that the fossil U. appendiculata from the Neogene sediment in Japan closely resembles the extant U. americana. Thus, the ancestors of C. setaricola and C. kansugei may have formed galls on the ancestors of Blepharocarpus, which have probably become extinct from East Asia. The ranges of C. setaricola and C. kansugei, stretching from Nepal to south-western Japan (Figure 19), are included in the area that functioned as a refuge for temperate biota during the glacial epochs (Haffer 1982). Unisexual populations of C. setaricola and C. kansugei may have survived on the secondary host as relics in East Asia since the extinction of the primary host.
C. hispanica is distributed in northern Spain where no U. laevis occurs naturally (Tutin 1964). The Iberian Peninsula is known to have been one of the refuges for temperate biota during the glacial epochs (Haffer 1982;Taberlet et al. 1998;Hewitt 1999). Therefore, it is also possible that C. hispanica has survived in the refuge after the local extinction of the primary host.
There is no sound evidence that a unisexual species was derived from another unisexual species, but it is conceivable that unisexual populations were genetically differentiated among local regions during the expansion of the range to the extent that the morphology of a local unisexual population is recognized as a distinct species. If this were true, the most recent ancestors of C. setaricola and C. kansugei may not have been gall-formers. But this would require the production of alate exules, by which unisexual populations of, for example, C. compressa may have expanded the distribution to East Asia. However, alate exules have not been reported for C. setaricola nor for most populations of C. kansugei except for the Nepalese population (Akimoto 1985). Therefore, it is most likely that C. setaricola and C. kansugei originated in East Asia.
The distributional ranges of the three unisexual species of Colopha are outside the present range of the section Blepharocarpus. However, the disjunctive distribution of Blepharocarpus suggests that the elements have become extinct in several localities including the present ranges of the unisexual species. Therefore, it is concluded that the origins and present distributional pattern of the unisexual species are consistent with Mordvilko's hypothesis. Many fossil species of Ulmus that are now extinct are reported from the Neogene flora of Japan (Tanai 1961). If the section Blepharocarpus has become extinct from East Asia since the Neogene, C. setaricola and C. kansugei would have a history of parthenogenesis of more than 1.6 million years.