Introduction
Until quite recently, clones were assumed to be genetically identical and often described as such in the scientific literature, despite widespread knowledge since the turn of the 20th century of the existence of mutational processes (Endersby, Reference Endersby2008). The main reason for this was (i) it was scientifically convenient in many laboratory and field experiments to assume that a particular clone was genetically identical; and (ii) the means did not exist to prove or deny this belief (Lushai & Loxdale, Reference Lushai and Loxdale2002; Loxdale & Lushai, Reference Loxdale and Lushai2003; Loxdale, Reference Loxdale, Schön, Martens and van Dijk2009; Martens et al., Reference Martens, Loxdale, Schön, Schön, Martens and van Dijk2009). Now, with the widespread availability of high resolution molecular makers, it has become possible to test the genetic fidelity of clones, not only between clone mates within a single generation but also between vertically produced generations.
Genomes are doubtless subject to a plethora of mutational events, probably rapid, including point mutations, transposon-mediated ‘hot spots’, perhaps causing inversion polymorphisms (Cáceres et al., Reference Cáceres, Ranz, Barbadilla, Long and Ruiz1999) and slippage mediated mutational changes in highly repetitive sequences like micro- and minisatellites (Loxdale & Lushai, Reference Loxdale and Lushai2003). Meanwhile, in most studies of clonality in aphids, especially field studies, a relatively small number of predominantly neutral microsatellite markers (usually 4–12, rarely more) are employed, and clones so discriminated are labelled as ‘multilocus genotypes’ or MLGs. However, nothing is known concerning the rest of the genome in terms of its genomic status, mutational or not (Loxdale, Reference Loxdale2008a,Reference Loxdaleb, Reference Loxdale, Schön, Martens and van Dijk2009). With microsatellites, evidence suggests that whilst these markers clearly show mutational changes (usually by addition of repeat units, more rarely subtraction of these: Goldstein & Schlötterer, Reference Goldstein and Schlötterer1999), aphid clones seem to be relatively stable for such markers, at least over the short term (e.g. a growing season of about 14 asexual generations in temperate regions: Wilson et al., Reference Wilson, Sunnucks and Hales1999). When more loci are examined using other high-resolution DNA markers, e.g. AFLP (amplified fragment length polymorphic) markers, then considerable variability has been detected, even in the first generations of aphid lineages. Thus, Forneck et al. (Reference Forneck, Walker and Blaich2001) and Vorwerk & Forneck (Reference Vorwerk and Forneck2007), testing clonal lineages of the grapevine phylloxera, Daktulosphaira vitifoliae Fitch (Order Hemiptera: Superfamily Phylloxeroidea: Family Phylloxeridae), have found significant genetic variability within and between generations. For example in the latter laboratory study, the authors examined eight asexual lineages for 15 asexual generations. One hundred and forty one loci were tested of which up to 15 were polymorphic. Whilst mutations were found in every generation, even in early ones, sequencing of 37 selected polymorphic bands revealed such mutations to be mostly of non-coding origin. Five mutated loci were transmitted to the 15th generation. It is thought that such variations, especially in coding regions, but also sometimes in non-coding regions such as microsatellites (Li et al., 2002), are potentially selective and occasionally adaptive (Forneck et al., Reference Forneck, Walker and Blaich2001; Lushai et al., Reference Lushai, Loxdale and Allen2003).
In the present study, AFLPs have been used to examine genetic variability within and between clones of the grain aphid, Sitobion avenae (F.) (Order Hemiptera: Superfamily Aphidoidea: Family Aphididae), which feeds above ground, mainly on the ears of the growing plant, especially wild grasses and cultivated cereals (Family Poaceae). This aphid species is a common pest of cereals in many countries in the world, including Europe and North and South America, causing damage by direct feeding and by transmission of pathogenic plant viruses (Vickerman & Wratten, Reference Vickerman and Wratten1979; Blackman & Eastop, Reference Blackman and Eastop2000).
Our aim was to show that, like Phylloxera, species of aphid of the Superfamily Aphidoidea – which also have asexual propagation involving apomictic reproduction (Blackman, Reference Blackman, Blackman, Hewitt and Ashburner1980) – likewise show similar widespread mutational changes. For purposes of this study, we refer to a ‘clone’ broadly as the descendants of an asexual female founding aphid (=original lineage stem mother), derived in spring from egg hatch following autumnal sexual recombination (Blackman, Reference Blackman, Blackman, Hewitt and Ashburner1980; Hand, Reference Hand1989), or from a pre-existing asexual progenitor. As with many species of aphids, S. avenae undergoes seasonally-induced cyclic parthenogenesis, although unlike some aphids, it is monoecious on a single host all year round and shows a variety of life cycle types. These include holocyclic forms, anholocyclic forms, androcyclic forms and ‘intermediate’ forms (Simon et al., Reference Simon, Rispe and Sunnucks2002). It is known that in the field, the sexual forms – winged males and wingless sexual females (=oviparae) – can mate between some of the other lineages (i.e. males produced by holocyclic, androcylic or intermediate lineages can mate with oviparae from holocyclic or intermediate lineages), giving rise to the phenomenon of ‘sexual leakage’ (Llewellyn, Reference Llewellyn2000; Simon et al., Reference Simon, Rispe and Sunnucks2002).
Here, we propose two hypotheses: (1) that AFLP-traced mutations occur at random among lineages, which we refer to as ‘genetic noise’; and (2) that AFLP-traced mutations accumulate within specific lineages, such mutations being stable and transmitted in an asexual aphid line. Hypothesis (1) is an expected outcome and is the normal property of DNA, whilst (2) indicates observable intraclonal variation.
Materials and methods
Sitobion populations
Sitobion avenae single asexual founder lineages were kept on single wheat seedlings (cv. Avalon) in separate glass tubes according to Austin et al. (Reference Austin, Tatchell, Harrington and Bale1991) and reared under constant environmental conditions (long daylength conditions, i.e. 16 h L: 8 h D) at 20 °C. Generation time (i.e. time between live births by successive generations of adult female aphids in the temporal sequence within the strictly parthenogenetic lineages) ranged between 7–10 days. The two lineages selected for AFLP analysis were originally collected from the field in Hertfordshire, UK in 1992 (HF92a, later named SB) and 1995 (DAV95, later named SA), respectively, and maintained in long-term culture at Rothamsted Research, Harpenden, Hertfordshire, UK (fig. 1a, b). Each lineage was started from a single parthenogenetic aphid (=lineage stem mother), SA on 31st March 2003, SB on 8th April 2003. When first instar nymphs were produced, four or half of all individuals surviving were carefully placed, using a fine camel hair brush, as singletons on single leaf blades of wheat per individual Austin tubes for a further round of rearing, whilst the surplus with the asexual mother of that particular generation were stored in 95% ethanol until DNA extraction. This procedure was continued for subsequent generations until termination of the experiment at the fifth generation on 14th and 21st May 2003 for clones SA and SB, respectively. Strict clonal hygiene was maintained throughout the experiment, the white Formica-coated bench top being wiped with 95% alcohol between each new inoculation operation, i.e. placement of a new aphid on a leaf.

Fig. 1. Propagation scheme of asexual clonal lineages (a) SA and (b) SB over five propagation cycles. Black boxes indicate mutated individuals; numbers refer to mutations listed in table 2.
DNA extraction and AFLP-PCR
Total genomic DNA was extracted from single aphids (including adults and nymphal stages I-IV) using a column-based DNA extraction kit (QIAGEN, Hilden, Germany). RNAse digestion was included in the DNA extraction step.
Restriction of total genomic DNA was performed in a double digest reaction employing 3 U of EcoRI (Fermentas, Sankt Leon-Rot, Germany) and 2.4 U of TruI (Fermentas) and incubating samples for 90 min at 37 °C and 120 min at 65 °C. Restriction enzymes were inactivated by a final step of 15 min at 85 °C. Ligation of AFLP adapter sequences was performed directly after this by adding 5 pmol EcoRI-adapters, 50 pmol TruI-adapters, 1 U T4 DNA ligase (Fermentas), 100 mM ATP (Fermentas) and 1× T4 ligase buffer (Fermentas). A two-step procedure was followed for AFLP amplification as described by Vorwerk & Forneck (Reference Vorwerk and Forneck2006) using the primer combinations: E10/M8, E13/M8, E14/M8, E16/M8, E21/M3 and E21/M17 (table 1) revealed by a forgoing primer screening. AFLP-PCR reactions were performed in 10 μl volumes containing 10 pM of each primer, 2 mM dNTPs (Fermentas), 1 μl × 10 PCR Buffer, 2 mM MgCl2 and 0.25 U Taq Polymerase (Invitrogen, Karlsruhe, Germany) and approximately 5 ng of template DNA.
Table 1. AFLP primer sequences.

Samples were electrophoresed on 6% polyacrylamide gels on an ALF-sequencer (Amersham Biosciences, Freiburg, Germany) employing 5′ end cy5-labeled primers. Allele sizes were assigned using an external DNA size standard (50–500 base pairs (bp), Amersham Biosciences).
Technical quality and reproducibility of the samples were ensured as follows: DNA extractions were optimized and quality was tested prior to PCR-reaction. AFLP reactions were optimized and produced reproducible banding patterns with the primer combinations employed (data not shown). In order to avoid misinterpretation through gel electrophoreses, each gel was reproduced at least once and only reproducible bands were scored visually.
Data analysis
Reproducible markers were chosen between 70 and 400 bp and amplification products scored only in this size range. Datasets were revised for the analysis of lineages and sublineages by eliminating monomorphic and therefore non-informative markers, and these were treated separately. Arlequin 2.0 (Schneider et al., Reference Schneider, Roessli and Excoffier2000) was used for the analysis of the genetic variants observed and to estimate variance components in terms of AFLP-haplotypes within clonal lineages (intraclonal variation).
Results
General findings
Of the total of 1176 live aphids collected in alcohol by the end of the fifth generation throughout the generational experiment, 818 were of lineage SA, 358 lineage SB. Of the total accumulation of 818 SA aphids, 171 were stem mothers, 647 their offspring, giving a mother to offspring ratio of 1:3.78. Death amounted to 69/818 or 8.4%. Most of the extinctions occurred in generation 4 (33 mothers, 28 offspring) with four mothers dying in generation 3 and four in generation 4. Of the total accumulation of 358 SB aphids, 73 were stem mothers and 285 their offspring, giving a mother to offspring ratio of 1:3.90. Extinction amounted to 106/358 or 29.6% of the total. Again, most of the extinctions occurred in generation 4 (97 mothers, eight offspring), with only one mother dying in generation 3. From this data, it appears that lineage SA was considerably more fecund than lineage SB and generally suffered around a quarter the level of mortality as lineage SB when reared under the same conditions.
Genetic profiles
From the aforementioned samples collected and stored in alcohol, DNA was extracted from a combined total of 110 individuals from both S. avenae lineages (N = 54, SA; N = 56, SB). Of these, genetic variation was shown following analysis of 197 (SA) and 213 (SB) AFLP bands, of which ten (SA) and 13 (SB) were polymorphic, respectively, among all the descendants studied. Polymorphic bands were differentiated into addition and loss of markers (deletions). SA showed nine deletions and one addition, SB 12 deletions and one addition. The average gene diversity within the lineages was 0.024 ± 0.013 and 0.031 ± 0.016 for SA and SB, respectively.
Nature of mutations
Twenty-three AFLP loci were screened in the two asexual lineages analyzed, i.e. for bands which showed clear signals of mutational events – either random band changes in terms of electrophoretic mobilty or as new bands that appeared at random, or disappeared from previously documented banding profiles in the offspring of the two founder female aphids. Mutations were detected in the fourth and third propagation cycles of SA and SB, respectively. Six polymorphic loci were screened only once (table 2, column labelled N). We considered a mutation to be asexually transmitted in those cases when a mutation in the following generation occurred in at least one descendant. Asexually transmitted mutation events were detected in both lineages of which the markers 9 (E21M17-203) (SA) and 16 (E14M8-122) (SB), showed mutated loci in all descendants of a particular sublineage analysed (table 2, fig. 1a, b). Markers 8 (E21M3-291), 10 (E21M17-232) and 12 (E14M8-216) were detected in the majority, but not all, descendants of the sublineages in question. Other mutated loci could be found within some descendants of a particular sublineage and also sometimes in the offspring of other sublineages, such as markers 3 (E13M8-251), 4 (E13M8-257) in SA, or 11 (E10M8-213), 23 (E21M3-148) in SB, and may not be generally considered to be asexually transmitted at this point. Random mutations occurring once, were detected six times, i.e. 1 (E13M8-177), 13 (E10M8-222), 14 (E13M8-167), 15 (E13M3-222), E16M8-80 and 20 (E16M8-298), in three of 110 individuals analysed (SA21, SB9, SB15). It is noteworthy that these bands (among others) were found in highly polymorphic individuals observed in both lineages.
Discussion
In this study, we do not seek to challenge whether clones are ecological and evolutionary ‘dead-ends’, as propounded by Moran (Reference Moran1992) and Simon et al. (Reference Simon, Delmotte, Rispe and Crease2003) [but see Reusch & Boström, Reference Reusch and Boström2011 in the case of the marine angiosperm, eelgrass, Zostera marina L. (Alismatales: Zosteraceae) where the empirical evidence suggests that they are not] but rather we show empirically that aphid clonal lines do not have complete genetic fidelity; in effect such lineages comprise a spatio-temporal population with accumulating variance, as surmised from theoretical studies (e.g. Suomalainen et al., Reference Suomalainen, Saura, Lokki and Teeri1980; Kondrashov, Reference Kondrashov1988, Reference Kondrashov1993).
We show, in our study involving AFLP fingerprints from 110 individuals stemming from two asexual founder aphids, that spontaneous mutations occur – even in the early generations of the lineages studied. The marker technology here used involved selectively amplifying whole genome restriction fragments which were subsequently electrophoresed to give a complex band profile. The polymorphisms observed in our study were mostly due to the absence of existing bands rather than to the production of additional ones. Our work on molecular intraclonal genetic variation complements that of recent karyotypic studies of peach-potato aphid, Myzus persicae (Sulzer) clones which demonstrated widespread intra-clonal, intra-individual (i.e. inter-embryo) and even intra-embryo chromosomal variation, due to recurrent fragmentations of chromosomes X and autosomes 1 and 3 (Monti et al., Reference Monti, Mandrioli, Rivi and Manicardi2012a,Reference Monti, Lombardo, Loxdale, Manicardi and Mandriolib).
The fact that mutation rates seem very high (higher than ‘conventional rates’ of 10−9 to 10−6 per gene per generation) probably directly relates to the fact that most of the band mutations seen are in non-coding regions of the genome, and it is now well established emprically that such regions can have very high mutation rates, i.e. >10−2 per gene per generation, or one in 100 individuals. Transposonation rates may also be of this magnitude (see Loxdale & Lushai, Reference Loxdale and Lushai2003 and references therein), and there is some evidence that aphid lineages and DNA profiles (i.e. random amplified polymorphic DNA, RAPDs) are indeed influenced by transposons (Lushai et al., 1997; Loxdale, Reference Loxdale2008a,Reference Loxdaleb; see also Mittapalli et al., Reference Mittapalli, Rivera-Vega, Bhandary, Bautista, Mamidala, Michel, Shukle and Mian2011). The fact that bands actually disappear means that mutations (perhaps TE-mediated) either affect the primer binding sites per se during the AFLP amplification process or, more likely, insertion is involved such that the length between opposing primers is greater than the ability of the Taq polymerase to amplify the region concerned (i.e. usually no longer than 1 Kb). The mutational effect is also of course amplifed by the huge reproductive potential of aphids, one asexual female being able to produce up to 100 offspring in one generation (probably normally around 30 in S. avenae); and each offsring itself is capable of producing another large number, so that an exponential population rise occurs, more especially in the absence of biological control agents (parasitoid wasps, coccinelids, syrphid and lacewing larvae, etc.) and under favourable ambient conditons (Blackman, Reference Blackman1971; Dixon, Reference Dixon, Minks and Harrewijn1989; Harrington, Reference Harrington1994).
Use of AFLPs is a widely used approach for genetic screening of diploid organisms; however, it is prone to errors and must be interpreted carefully. Heterozygous mutations may not be detected and thus underestimated in the screenings. Furthermore, DNA quality affects the resolution of the method and like in any PCR technique, false positives may occur (Benjak et al., Reference Benjak, Konradi, Blaich and Forneck2006; see also Trybush et al., Reference Trybush, Hanley, Cho, Jahodova, Grimmer, Emelianov, Bayon and Karp2006). The fact that the majority of the mutations studied in this work arose as a consequence of the absence of bands needs stressing (i.e. since AFLPs are dominant markers, if only one allele of a genotype at a particular locus is affected (e.g. at the primer binding sites), a single band of diminished staining intensity might be expected to remain rather than complete absence of bands, as actually observed). Some individual aphids tested seemed to have accumulated such mutations (SA21, SB9, SB15). Other mutations which were not observed in all the descendants of a given asexual stem mother may be because they occurred in a part of the oocytes or following nymphal stages and thus may not be detected by the AFLP approach. At this point, we do not have an explanation for the simultaneous occurance of mutations (e.g. 23) in different sister lineages.
These findings thus take us beyond the realm of purely theoretical considerations and mean that caution must be exercised in assuming the genetic fidelity of aphid clonal cultures, more especially those kept for long periods, including on artificial diets (van Emden, Reference van Emden2009). Since it is possible that some of the mutations detected are in coding regions of the genome, as found in D. vitifoliae (Vorwerk & Forneck, Reference Vorwerk and Forneck2007), then such mutational changes may well have adaptive significance in aphids (Loxdale & Lushai, Reference Loxdale, Lushai, Emden and Harrington2007; Loxdale, Reference Loxdale, Loxdale, Claridge and Mallet2010), as well as in other organisms with periods of asexual reproduction (e.g. see de Witte & Stöcklin, Reference de Witte and Stöcklin2010 and Reusch & Boström, Reference Reusch and Boström2011 in relation to plant adaptation and clonal longevity, including somatic mutations). The fact that many of the asexual lines tested died out before the 5th generation suggests random, stochastic processes rather than loss of fitness of the two original lineages as such, more especially because some sublineages did persist to the end of the experiment. (Only advantageous or neutral mutations were maintained, whereas deleterious ones may have been selected out. Since most of these mutations were deletions, suggests that they were probably heterozygous in the original foundress.)
Lastly, the fact that the mutations seem to be at random than following through the lineages as specific haplotypes is contrary evidence to the notion of Muller's ratchet operating (Muller, Reference Muller1964), as also clearly demonstrated in our previously empirically-based AFLP study of asexual Phylloxera lineages (Vorwerk & Forneck, Reference Vorwerk and Forneck2007). Ultimately, in the light of these and similar recent findings, showing genetic variability within asexual lineages, there is a clear need to review and re-interpret the clone in all its many biological manifestations and misunderstandings (Hughes, Reference Hughes1989; Martens et al., Reference Martens, Loxdale, Schön, Schön, Martens and van Dijk2009; Schön et al., Reference Schön, Martens and van Dijk2009). It is known that aphids show high levels of phenotypic plasticity within asexual aphid lineages (e.g. Wool & Hales, Reference Wool and Hales1997; Dombrovsky et al., Reference Dombrovsky, Arthaud, Ledger, Tares and Robichon2009; Richardson et al., Reference Richardson, Lagos, Mitchell, Hartman and Voegtlin2011) but even here, probably there is a genetic basis to some of the phenotypic changes observed (Loxdale & Lushai, Reference Loxdale, Lushai, Emden and Harrington2007; Dombrovsky et al., Reference Dombrovsky, Arthaud, Ledger, Tares and Robichon2009; Loxdale, Reference Loxdale, Loxdale, Claridge and Mallet2010).
If we are not sure of its genetic fidelity, does the concept of the clone, which has now lasted over 100 years since its inception (Webber, Reference Webber1903), have any reality in a purely strict genetic sense? This may seem a minor problem in laboratory-based studies, but in the field, it marks a practical and potentially insurmountable one, i.e. the tracking of particular clones in the field using various forms of DNA fingerprinting (Loxdale, Reference Loxdale2008a). The alternative of entire or even partial sequencing seems preferable, but of course, this may be impracticable in large molecular ecological studies at the present time (The International Aphid Genomics Consortium, 2010), yet rapidly developing new technology may soon make it possible (e.g. Graham-Rowe, Reference Graham-Rowe2012). The instability of high-resolution, hypervariable banding profiles as currently observed, also throws up questions concerning their application in other molecular biological/ecological studies, e.g. human DNA fingerprinting approaches as applied to paternity studies where genetic stability of the markers is assumed between close generations and in forensic studies too. (The AFLP technique involves amplification of both coding and non-coding regions, some of which undoubtedly involve mini- and microsatellite regions, as used in DNA fingerprinting following the Jeffreys probe method.) If correct, aphids, because of their ease of asexual propagation and rearing in culture, may represent particularly valuable model systems (essentially representing ‘transgenerational individuals’ or TGIs) for investigating such topical and important biological issues, both from a fundamental as well as an applied point of view.
Acknowledgements
We thank Dr Richard Harrington of Rothamsted Research for kindly allowing us to use the two S. avenae clones kept in long-term culture there; Mrs Tracey Kruger for her expert help with aphid rearing; the three anonymous referees for their helpful and insightful comments; Prof. Rolf Blaich for supporting this study at the Department of Crop Sciences, University of Hohenheim, Germany; and Dr Ulrike Anhalt of the University of Natural Resources and Life Sciences Vienna, Department of Crop Sciences, Austria, for helpful discussions.