Experimental hybridisation between Aphis grossulariae and Aphis triglochinis ( Sternorrhyncha : Aphididae )

Aphis triglochinis and A. grossulariae clones from southern Poland produced fertile hybrid eggs under experimental con­ ditions. Established hybrid clones expressed normal parthenogenetic reproduction but bisexual generations were obtained only in three hybrid clones out of twenty six. Fertile Fi hybrid eggs were obtained in one hybrid clone. Morphological and host-specificity features of A. grossulariae dominated in the majority of hybrid clones. The present results do not exclude the possibility of natural hybridisation of studied aphid species. Natural hybrids may be difficult to detect because of their “pure” morphological and hostspecificity features.


INTRODUCTION
The possibility of hybridising and producing viable and fertile progeny is an important feature of biparental spe cies, being emphasized by the reproductive species con cepts, including the biological species concept (e.g.Mayr, 1982;Dobzhansky, 1970;Paterson, 1993).Aphids are mostly biparental species, and reproductive isolation is important feature of aphid species (Shaposhnikov, 1987;Blackman, 1995;Rakauskas, 1998a).Hybridisation studies might supply important information on the taxo nomic status of the forms involved in a complex (Müller, 1985;Shaposhnikov, 1987;Guldemond, 1990a;etc.),although aphid crossing experiments are rather compli cated (Hales et al., 1997).
Species of the genus Aphis L. inhabiting currants in Europe [Aphis grossulariae Kaltenbach, 1843, A. tri glochinis Theobald, 1926, and A. schneideri (Börner, 1940)] present certain taxonomic problems (Hille Ris Lambers & Dicker, 1965;Stroyan, 1984;Rakauskas, 1998b).Detailed biosystematic studies of this species complex have therefore been undertaken.It appeared that the three species are rather distinct in their life-cycles and host specificity, although they share the same winter hosts.A. schneideri is monoecious holocyclic on Ribes spp., A. grossulariae is holocyclic facultatively heteroecious between Ribes spp.and Onagraceae herbs, and A. triglochinis is holocyclic obligatorily heteroecious between Ribes spp.and Brassicaceae, Boraginaceae and Asteraceae herbs (Rakauskas, 1993).Sibling species may occur on the summer hosts of A. triglochinis (Rakauskas, 1998b).Many of the morphological characters exploited in the keys to discriminate between the three species appeared to be unreliable.Nevertheless, morphometric analysis of numerous specimens of all morphs revealed morphological features, ensuring separation of all morphs of A. grossulariae, A. triglochinis and A. schneideri (Rakauskas, 1998c).All three species have the same chromosome number (2n = 8), and preliminary karyotype analysis suggests that A. triglochinis is more closely related to A. schneideri than to A. grossulariae, but this is inconsistent with morphological and host specificity data (Turěinaviěiené et al., 1997).DNA-s of the three species appeared to be different when analysed by means of the randomly amplified polymorphic DNA polymerase chain reaction technique.Four out of 13 primers applied pro duced bands that were polymorphic among the three spe cies.Based on the numbers of bands shared in common, A. grossulariae seems to be more closely related to A. schneideri when compared with A. triglochinis (Turcinaviciené et al., 1999).Thus, morphological, life-cycle, host specificity and DNA analysis data suggest that A. grossulariae, A. triglochinis and A. schneideri are good, well-defined species.Nevertheless, with regards to the similarity of karyotype and controversial references on the bionomics of the three species (see Rakauskas, 1998b, c), experimental interspecific hybridisation studies have been undertaken.Data on A. grossulariae x A. schneideri and A. schneideri x A. triglochinis crossing results are alreadypublished (Rakauskas, 1999a, b).
The aim of this work was to study the possibility of hybridisation between A. grossulariae and A. triglochinis under experimental conditions.

MATERIAL AND METHODS
Five clones of A. grossulariae and three clones of A. tri glochinis originating from southern Poland were used for inter specific crossing experiments in Katowice (southern Poland) in 1987, each clone starting from a single fundatrix or fundatrigenia (Table 1).Hybridisation experiments were the continua tion of morphological, life-cycle and host specificity studies of the two species (Rakauskas, 1993).This ensured the precise documentation of the morphology and bionomics of parental clones and provided data for obtaining key morphological char acters and canonical discrimination functions to distinguish between various morphs of A. grossulariae and A. triglochinis (see below).Sixteen A. grossulariae 2 x A. triglochinis k and ten reciprocal crossings were tried.Ten oviparae and one male of the alternative species were isolated in muslin branch-tip cages on currant bushes for each cross.The construction of cages ensured the isolation of 10 cm of the terminal part of the currant shoot (see Rakauskas & Rupais, 1983).Groups of 5 to 10 gynoparae of the same clone were used to receive newly born oviparae larvae.Gynoparae were obtained from cages of clones of each respective species and their morph was con firmed under a stereoscopic microscope (16*) before releasing them into branch-tip cages on currants.After depositing progeny gynoparae were removed and fixed in alcohol for sub sequent morphological analysis, as were males.This ensured that the oviparae were virgin.It is noteworthy that A. grossu lariae and A. triglochinis have no sexuparae: those remigrating from summer host plants are gynoparae producing only larvae of future oviparae.Currant shoots were isolated early in autumn, at the moment when the first gynoparae appeared, and males were not yet present.This (and careful examination of the shoot by means of 2.5* magnifying glass) eliminated the possibility of any wild eggs inside the cage.Eggs of these species have never been found in Katowice at the beginning of September.Hybrid eggs were obtained from thirteen crosses (Table 2).Intraspecific interclonal and intraclonal crosses were also per formed.Eggs were subsequently transferred (together with frag ments of shoots on which they were deposited) to Vilnius (Lithuania) and maintained in field conditions throughout the winter.This was performed by attaching the fragments of the shoots containing hybrid eggs to the appropriate tip shoots of the field grown black currant bushes (mid-ripening variety "Derliai") inside muslin branch-tip cages.Currant shoots used were carefully checked using 2.5* magnifying glass, to confirm  (Rakauskas, 1993).The list of clones is pre sented in Tables 3 and 4: clone number indicates also the origin of the hybrid clone: gt1 means the first hybrid clone from the crossing scheme A. grossulariae l * A. triglochinis k , tg1 -the first hybrid clone from the reciprocal crossing.
The fundatrix and twenty specimens (when available) of the main morphs (alatae and apterae from winter and summer hosts) of each hybrid clone were mounted in Faure-Berlese fluid on microscope slides for morphological analysis.Two methods were used for the morphological identification of hybrid clones.First, the identification was attempted using common key char acters (Rakauskas, 1998c).Numbers of additional hairs on the ultimate rostral segment (for fundatrices, apterous viviparous females from currants and summer host plants, alate viviparous females from currants, gynoparae, oviparae and males) and numbers of secondary rhinaria on the third antennal segment (for alate viviparous females from summer hosts) were used as key characters.Second, canonical variates analysis, a method that has proved very useful in distinguishing closely related aphid species (e.g., Blackman, 1992;etc.),was applied.Mor phometric data of pure A. schneideri and A. grossulariae clones (see above) were used for calculating the canonical discrimina tion functions (CDF) for each morph.Variables used in the CDF were selected on the basis of their discriminatory power: those having the smallest partial Wilks' Lambda were taken when cal culating the CDF for every morph (for details see StatSoft, 1995, Chapter 2).List of variables used for calculating the CDF for every morph is presented in Table 5. Wider information on the aphid material used has been already published (Rakauskas, 1998c).The obtained CDF values were subsequently counted for every hybrid specimen of every morph, and standard box and whisker plot procedure was applied for morphological determination of various morphs of every hybrid clone.Exam ples illustrating the morphological identification procedure of the alate currant-inhabiting viviparous females are presented in Fig. 1 (using the key characters) and Fig. 2 (exploiting the CDF).Every morph of each hybrid clone was treated as having the morphology of a particular species if the range of the studied character or CDF values in that morph was covered by the range of the same character of that particular species.Thus, alate viviparous females (currant morph) of the hybrid clones gt6-11, g tl5-18 were determined as having the morphology of A.grossulariae both by means of key character (Fig. 1) and CDF values (Fig. 2).The hybrid clone morph was treated as morpho logically tending towards a particular species if the 25-75% box area of the studied character or CDF value of that morph was overlapped by the range of that particular species.For example, alate viviparous females (currant morphs) of hybrid clones gt3-5, gt12-14 were determined by means of key character as tending morphologically towards A. grossulariae (Fig. 1).The hybrid morph was treated as morphologically intermediate if the 25-75% box area of the studied character or CDF value for that morph was between the ranges ofboth species.Graphical data for other morphs of all hybrid clones (similar to those presented in Figs 1-2 for alate viviparous females) are available from the author on request.Morphological identifica tion of fundatrices was different, because only one fundatrix of every clone was available.Scatterplot analysis procedure was performed in this case, an example being presented in Fig. 3.In total, five morphs of the majority of hybrid clones were evalu ated morphologically by means of key characters and CDF.Thus, ten evaluations (key characters and CDF for each morph) of every hybrid clone were obtained.Every hybrid clone was afterwards summarized as having certain overall morphological features on the basis of these ten evaluations.For example, hybrid clone gt1 (Table 3) had 2 evaluations (when using CDF) as tending morphologically to A. grossulariae, and 3 evaluations (when applying key characters) as being morphologically iden tical with A. grossulariae.In overall determination, this hybrid clone was treated as tending morphologically to A. grossulariae, since CDF performs identification on the basis of more charac ters.Following the same procedure, hybrid clone gt2 was deter mined as morphologically identical with A. grossulariae, clones gt3-4 as tending morphologically to A. grossulariae, and so on (Tables 3-4).Discussion of morphological characters elsewhere in this paper concerns this overall morphological determination of the clone, unless otherwise stated.
All calculations were done using the STATSOFT statistical package STATISTICA for WINDOWS 5.1 (StatSoft, 1995).
Host specificity and life cycle analysis of every hybrid clone were performed in the same way as described earlier (Rakauskas, 1993).Potted Epilobium adenocaulon Hausskn., Chamaenerion angustifolium (L.) Scop.(Onagraceae, summer hosts of A. grossulariae), Cardamine amara L. (Brassicaceae) and Myosotis palustris L. (Boraginaceae; both summer hosts of A. triglochinis) plants were tested as potential summer hosts for every clone.Transfers of alate females were repeated (if first transfers were unsuccessful) at weekly intervals until this morph was no longer available.Groups of five migrants were used for each transfer test.This was one of the reasons for insufficient numbers of certain morphs used for the morphological analysis in some clones (e.g.lack of alate viviparae from currants in clone gt16).When only a few winged viviparae were obtained, they all were used for transfer experiments.Hybrid clones that produced sexuales were crossed both intra-and inter clonally (Table 6).Five oviparae and one male were used in each F1 crossing variant.Backcrossing with pure A. triglochinis and A. grossulariae clones was not performed because of the lack of sexuales in pure clones of these species in autumn 1988 in Vil nius.   3 -4 ).O n the other hand, the m orphological an d hostspecificity features o f the h ybrid clones are puzzling.Inform ation on secondary h ost specificity and the m o r phological features o f hybrid clones can be sum m arized as follow s (overall m orphological determ ination above the line, h ost specificity below the line; -gross., ->trigl.overallm orphology or host specificity tending to the respective species; m onoec.? -hybrid clone probably m onoecious on currants, rejected all proposed secondary hosts; ?-lack o f inform ation).
In A. triglochinis k x A.grossulariae 2 crosses: 3 gross.: 2 gross.: ^ gross.: ^ trigl.3 gross.: 2 -*gross.: 1 ?: 1 ^trigl.A confusing phenom enon appeared in hybrid clones gt3, gt5, gt10 and gt14: certain alate viviparous fem ales (currant m orphs) h ad som e o f the antennal hairs finely acute an d relatively long (Fig. 4d).T his is one o f the key characters o f the th ird currant-inhabiting species, Aphis schneideri (Fig. 4b).The existence o f A. grossulariae specim ens having A. schneideri-like antennal hairs has been already docum ented for pure clones o f A. grossu lariae (R akauskas, 1998c).6 ).H ybrid gt5 and tg7 oviparae w ere m orphologically A. grossulariae, w hereas the only analysed hybrid gt17 ovipara w as m orphologically interm ediate (Fig. 7).It is rem arkable th at num bers o f scent plaques on the hind tibiae o f hybrid oviparae w ere m arkedly red u ced w hen com pared w ith b o th parental species: in tw elve analyzed hybrid oviparae, the num bers o f scent plagues on the hind tib ia w ere from 3 to 78 (m ean value 33.36).The resp ec tive figures for A. grossulariae are 4 3 -1 1 6 (85.00) and for A. triglochinis 56-83 (72.20).It is noteworthy that the absence of scent plagues on the hind tibiae of oviparae is characteristic of A. schneideri (Rakauskas, 1998c).The malformation of scent plagues in hybrid oviparae might explain the low success of Fi crossing (Table 6).Intra clonal crosses of hybrid clones resulted in the appearance of winter eggs only in hybrid clones gt5 and tg7, but nothing hatched from these eggs the following spring.Fi crosses with the A. schneideri-like hybrid clone sti (A.schneideri x A. triglochinis crosses, Rakauskas, 1999b) resulted in eggs, except for the case when hybrid clone gti 7 oviparae were used.Successful overwintering and hatching of these eggs occurred only in the st1 x tg7 cross.Unfortunately, all five Fi cross fundatrices were killed by an Anthocoris predatory bug (as the cage muslin was damaged) previous to their maturation.The Fi cross results suggest the possibility that hybrid A. grossulariae x A. triglochinis clones possessing normal potential for bisexual reproduction may be found in the field.

DISCUSSION AND CONCLUSIONS
A. triglochinis and A. grossulariae clones from southern Poland are able to produce fertile hybrid eggs under experimental conditions.Established hybrid clones expressed normal parthenogenetic reproduction, and the bisexual generation appeared in three hybrid clones.Nev ertheless, Fi crosses inside the hybrid clones were not successful.Pure clones of A. triglochinis and A. grossu lariae also demonstrated reduced possibilities for bisexual reproduction in 1988 in Vilnius (for details, see Rakauskas, 1999a).Therefore, it remains unclear whether hybrid A. grossulariae x A. triglochinis clones had reduced potential for bisexual reproduction and hybrid breakdown is a possible postzygotic isolating mechanism between these species.The fact that the crossing of the hybrid A. triglochinis x A. grossulariae clone tg7 male with the hybrid A. schneideri x A. triglochinis clone sti oviparae resulted in fertile egg production supports the hypothesis of natural hybridisation between the studied species.Nevertheless, the present results are rather pre liminary.Successful experimental hybridisation does not necessarily mean the possibility of natural hybridisation.Experimental interspecific crossings have been success fully performed in aphid genera Dysaphis (Shaposhnikov, 1987, etc.), Myzus (Müller, 1969), Cryptomyzus (Guldemond, 1990a, b, etc.), Ovatus (Müller & Hubert-Dahl,   , 26.vi.1984, cultivated red currant;b -Vilnius, 10.v.1983, cultivated red currant;c -Vilnius, 26.v.1988, cultivated black currant;d -Vilnius, 20.vi.1988, cul tivated black currant.Scale: 0.05 mm. 1979), and the Aphis fabae Scopoli complex (Müller, 1982;Thieme, 1988).Natural prezygotic isolating mecha nisms might be rather sophisticated and fragile (Guldemond et al., 1994;Guldemond & Dixon, 1994;Thieme & Dixon, 1996), and be easily circumvented by the experi mental procedure.So, information on successful experi mental hybridisation reinforces the need for the study of natural isolating mechanisms between the species involved, such as sex pheromone specificity, the circadian rhythms of sex pheromone release and male activity, and other aspects of possible species-specific mate recogni tion systems.Crossing experiments need to be repeated with clones from other parts of the species distribution area, because the possibilities to produce hybrids may differ in different populations (Hewitt, 1990).DNA analysis of parental and hybrid clones (e.g.microsatellites and mitochondrial DNA techniques, see Hales et al., 1997;Sunnucks et. al., 1997) would help to confirm the identity of the crosses and indicate the degree of intro gression between natural populations of this species group.The question of natural hybridisation between A. triglochinis and A. grossulariae is important not only in a taxonomic context, but also in terms of the practical needs of currant pest management.The appearance of hybrid specimens having certain morphological characters of A. schneideri (Fig. 4d), the third European species of the currant-inhabiting complex of the genus Aphis, is of spe cial interest.Because of hybridisation, clones with the morphological features of A. grossulariae might appear, but being monoecious (as hybrid clone gt20) or faculta tively monoecious (as hybrid clone gt5) on currants.This might explain previous data on monoecy in A. grossu lariae (Gusynina, 1963;Savzdarg & Ponomareva, 1978).We have recently found A. triglochinis clones in Finland that seem to be facultatively heteroecious (Rakauskas, Turcinaviciene, unpubl.), a characteristic of A. grossulariae.
The present data raise certain questions concerning the genetic control of morphological characters.Despite the  14 characters used for calculating the CDF for the dif ferent morphs, all hybrid clones were identical with (23 clones of 26) or clearly tended to (3 clones) one of the species.The absence of morphologically intermediate clones is consistent with a hypothesis of the monogenic control of morphological characters, with A. grossulariae characters being dominant.That does not conform to the common understanding of the genetic control of develop mental processes (Blackman, 1999, pers. comm.), although it is not absolutely impossible, e.g., the "poly phene" mutation in Drosophila affects various morpho logical characters, such as eyes, thorax, tarsi and wings, but it is known to be a single gene mutation (Severtsov, 1987: 24).Single gene pleiotropic effects on general mor phology has been also reported in plants (Bohmert et al., 1998).The phenomenon, whereby all hybrid clones are identical with maternal species, can be explained as due to action of maternal genes (Lewin, 1996(Lewin, : 1141(Lewin, -1179)), or fertilization errors, such as gynogenesis (Ham & Veomett, 1980: 605), also due to hybridogenesis (Cherfas, 1981).Nevertheless, most of our hybrid clones that were obtained when using A. triglochinis oviparae (6 out of 7) were also identical with A. grossulariae.
The majority of hybrid clones, 22 of 26 tested, had the host specificity of A. grossulariae: they developed on Epilobium adenocaulon and Chamaenerion angustifolium, and rejected Cardamine amara and Myosotis pal ustris.The scarcity of hybrid clones having intermediate host preferences between A. grossulariae and A .triglochinis suggests monogenic control of this character.Monogenic control of host specificity has already been reported for the raspberry aphid Amphorophora rubi (Kaltenbach) (Briggs, 1965).Guldemond (1990a) also suggested that host plant specificity in the aphid genus Cryptomyzus might be controlled by only a few genes.
It may be that natural crosses between A. grossulariae and A. triglochinis (if they exist in the field) are hardly detectable, because of their similarity to one of the parental species.The dominance of morphological and host specificity features of A. grossulariae is in accor dance with the information on the rarity of A. triglochinis: it is rather uncommon on currants, at least in Europe (Hille Ris Lambers & Dicker, 1965;Cichocka, 1980; , 1981). A. triglochinis m ay be rare, if its phenotype is ju st a recessive hom ozygote o f the gene, w hose dom inant allele causes A. grossulariae features.M orphology o f A. schneideri, A. grossulariae and A. tri glochinis suggests that the latter species should be m ore phylogenetically distant from the first tw o (Stroyan, 1984;R akauskas, 1998c).R em audiere (1993) even places them in separate subgenera.A ccording to this, A. grossu lariae x A. schneideri crosses should be m ore successful th an A. grossulariae x A. triglochinis, b u t it seem s n o t to be true (R akauskas, 1999a).O n the other hand, the m ore distantly related are the parental species, the higher is p robability th at hybrids w ill look like one or the other o f them .T he phenom enon has been explained in other sys tem s b y preferential gene expression o f the m aternal allele (W u et al., 1997), m aternal gynogenesis (M akeeva, 1989), or elim ination o f paternal chrom osom es during distant hybridogenesis (Cherfas, 1981).The explanation in the present case is n o t y et know n.

Fig. 1 .
Fig. 1.Box and whisker plot of the key morphological character for the alate viviparous females of A. grossulariae and A. triglochinis and hybrid clones grossulariae x triglochinis (currant morphs).

Fig. 2 .
Fig. 2. Box and whisker plot of canonical discrimination function values for the alate viviparous females of A. grossulariae and A. triglochinis and hybrid clones grossulariae x triglochinis (currant morphs).
A n interesting resu lt is th at none o f the 26 hybrid clones p o ssessed interm ediate m orphology.22 clones w ere m orphologically identical w ith A. grossulariae, 3 clones ten d ed m orphologically to this species, 1 clone w as identical w ith A. triglochinis.A. grossulariae m o r phology w as dom inant w hether A. grossulariae or A. tri glochinis oviparous fem ales w ere used for the crossing experim ents.

N
one o f the h y b rid clones had the pure h o st specificity o f A. triglochinis.E ven hybrid clone tg l, being m o rp h o logically indistinguishable from A. triglochinis, d ev el oped poorly on A. triglochinis sum m er hosts.A late viviparous fem ales o f this clone accepted Cardamine amara an d Myosotis palustris as sum m er hosts, fed and deposited progeny, although no pro g en y reached ad u lt hood on these plants.A late viviparous fem ales o f this clone rejected Epilobium adenocaulon and Chamaenerion angustifolium as sum m er hosts.F ourteen hybrid clones norm ally accepted Epilobium adenocaulon and Chamaenerion angustifolium as sum m er hosts, eight hybrid clones propagated on these plants m ore poorly.O ne hybrid clone (gt20, having A. grossulariae m o r phology) did n o t accept an y o f the tested herbaceous plants as sum m er hosts, an d also failed to finish its entire life-cycle on currants.In hybrid clone gt5 (m orphologi cally A. grossulariae), som e gynoparae and oviparae appeared on currants, b u t no sexuales w ere p roduced on sum m er hosts.This is in accordance w ith the previous data on facultative h eteroecy an d m onoecy in A. grossu lariae (G usynina, 1963; Savzdarg & Ponom areva, 1978).T hree hybrid clones succeeded in producing a bisexual generation.In A. grossulariae 2 x A. triglochinis k crossings, h y b rid clone gt5 p roduced gynoparae an d sub sequently oviparae on currants.H ybrid clone gt17 gyno parae and m ales w ere p roduced on Epilobium adeno caulon, an d oviparae (after transfer o f gynoparae to w inter host) on currants.In reciprocal crosses, hybrid clone tg7 p roduced p len ty o f gynoparae an d m ales on Epilobium adenocaulon, an d subsequently oviparae on currants.M orphologically, gynoparae an d m ales o f all three hybrid clones w ere sim ilar to A. grossulariae (Figs 5 -

Fig. 3 .
Fig. 3. Scatterplot of the individual main key character values plotted against the body length of hybrid fundatrices (A.triglochinis x A. grossulariae crossings) showing the distribution of the same values in A. grossulariae and A. triglochinis fundatrices.

Fig. 5 .
Fig. 5. Scatterplot of the individual canonical discrimination function values plotted against the body length of A. grossulariae, A. triglochinis and hybrid gynoparae.

Fig. 6 .
Fig. 6.Scatterplot of the individual canonical discrimination function values plotted against the body length of A. grossulariae, A. triglochinis and hybrid males.

Fig. 7 .
Fig. 7. Scatterplot of the individual canonical discrimination function values plotted against the body length of A. grossulariae, A. triglochinis and hybrid oviparae.

Table 3 .
Morphological and biological features of the experimental hybrid clones (A. grossulariae 2 x A. triglochinis k ) showing the summer host specificity (+ , normal propagation on respective hosts; ± , poor propagation; -, no propagation), morphological pecularities of dif ferent morphs of each clone (fx -fundatrix; apt, al -apterae and alatae from currants; aptll, alii -apterae and alatae from summer hosts; gyngynoparae; male -males; ovip -oviparae; t, ^ t -morphology as in A. triglochinis or tending to it; g, ^ g -morphology as in A. grossulariae or tending to it; i -intermediate morphological features; n -morph not obtained; 0 -morph obtained, but not measured) when performing iden tification by common key characters (key) or by means of canonical dis crimination function (CDF, see in material and methods), and the overall morphology of the clone (summary).Figures in morph column -No. of analysed specimens of respective morph.

Table 4 .
Morphological and biological features of the experimental hybrid clones (A. triglochinis 2 x A. grossulariae k ).Abbreviations as in Table3.

Table 5 .
Morphological characters exploited for calculating canonical discrimination functions used for the discrimination of the respective morphs of A. triglochinis and A. grossulariae.Abbreviations as in Table4.

Table 6 .
Hybrid A.triglochinis x A. grossulariae Fi crossing scheme, with information on the amount of live (black shining) eggs obtained and hatching success.