Oecologia (2007) 153:323–329 DOI 10.1007/s00442-007-0730-2

PLANT ANI MAL INTE RA CTIONS

Genetic variation in the eVect of a facultative symbiont on host-plant use by pea aphids Julia Ferrari · Claire L. Scarborough · H. Charles J. Godfray

Received: 14 November 2006 / Accepted: 15 March 2007 / Published online: 6 April 2007 © Springer-Verlag 2007

Abstract Ecological specialisation on diVerent host plants occurs frequently among phytophagous insects and is normally assumed to have a genetic basis. However, insects often carry microbial symbionts, which may play a role in the evolution of specialisation. The bacterium Regiella insecticola is a facultative symbiont of pea aphids (Acyrthosiphon pisum) where it is found most frequently in aphid clones feeding on Trifolium giving rise to the hypothesis that it may improve aphid performance on this plant. A study in which R. insecticola was eliminated from a single naturally infected aphid clone supported the hypothesis, but a second involving two aphid clones did not Wnd the same eVect. We created a series of new pea aphid–R. insecticola associations by injecting diVerent strains of bacteria into Wve aphid clones uninfected by symbionts. For all aphid clones, the bacteria decreased the rate at which aphids accepted Vicia faba as a food plant and reduced performance on this plant. Their eVect on aphids given Trifolium pratense was more complex: R. insecticola negatively aVected acceptance by all aphid clones, had no eVect on the performance of four aphid clones, but increased performance

Communicated by Thomas HoVmeister. Julia Ferrari and Claire L. Scarborough contributed equally to the work. J. Ferrari · C. L. Scarborough · H. C. J. Godfray NERC Centre for Population Biology and Division of Biology, Imperial College London, Silwood Park Campus, Ascot, Berkshire SL5 7PY, UK Present Address: J. Ferrari (&) · H. C. J. Godfray Department of Zoology, University of Oxford, South Parks Road, Oxford, OX1 3PS, UK e-mail: [email protected]

of a Wfth, thus demonstrating genetic variation in the eVect of R. insecticola on pea aphid host use. We discuss how these results may explain the distribution and frequency of this symbiont across diVerent aphid populations. Keywords Acyrthosiphon pisum · Host-plant specialisation · Regiella insecticola · Secondary symbiont · Trifolium pratense

Introduction Herbivorous insect species often consist of host-adapted populations, which show both genetically determined preferences for and better performance on diVerent plant species (Mopper and Strauss 1998; Drès and Mallet 2002). The pea aphid, Acyrthosiphon pisum, is one of the best-studied model systems for the evolution of host-plant specialisation (Via 1991, 1999; Via et al. 2000; Hawthorne and Via 2001; Ferrari and Godfray 2003, 2006; Simon et al. 2003; Ferrari et al. 2006; Frantz et al. 2006). Host specialisation on Medicago sativa and Trifolium pratense has a genetic basis and loci involved in preference and performance have been mapped (Hawthorne and Via 2001). The host-associated pea aphid populations on these and other plant genera within the Fabaceae are genetically diVerentiated (Via 1999; Simon et al. 2003; Frantz et al. 2006). However, facultative bacterial endosymbionts can modify the aphids’ Wtness on certain host plants (Tsuchida et al. 2004). Here, we manipulate facultative symbionts to study their eVect on aphid performance. Nearly all aphids possess the obligate primary symbiont Buchnera aphidicola, a bacterium that resides in specialised cells and provides essential amino acids, which are absent in plant phloem, the nutritionally imbalanced diet of

123

324

aphids (Douglas 1998; Moran and Baumann 2000). Facultative or secondary symbionts can be found in the aphid haemolymph and in specialised cells similar to those that usually harbour B. aphidicola. The three best studied belong to the
-Proteobacteria and have recently received formal names: “Candidatus Regiella insecticola” (formerly known as U-type or PAUS), “Candidatus Hamiltonella defensa” (T-type, PABS) and “Candidatus Serratia symbiotica” (R-type, PASS) (Moran et al. 2005). These bacteria have been found in a wide range of aphid species (Haynes et al. 2003; Russell et al. 2003) though they have largely been characterised in the pea aphid A. pisum. They are vertically inherited during both the sexual and asexual phases of the aphid life cycle (Chen and Purcell 1997; Darby and Douglas 2003; Moran and Dunbar 2006), but have noncongruent phylogenies with aphids suggesting at least occasional horizontal transfer (Russell et al. 2003). There is both observational and experimental evidence linking facultative symbionts with host-plant use. A number of studies have shown that secondary symbionts are distributed non-randomly across host-associated populations (Tsuchida et al. 2002; Simon et al. 2003; Leonardo and Muiru 2003; Ferrari et al. 2004). The most striking pattern is that worldwide pea aphids on Trifolium tend to carry R. insecticola. This naturally suggested the hypothesis that R. insecticola carriage might improve performance on Trifolium. Tsuchida et al. (2004) tested whether the bacterium R. insecticola inXuenced performance on Trifolium by using antibiotics to eliminate R. insecticola from a naturally infected aphid clone collected on T. repens and by subsequently injecting R. insecticola into the “cured” line. Both the naturally and artiWcially infected lines had higher performance on T. repens compared to the cured line, while Wtness on vetch, Vicia sativa, was unaVected (Tsuchida et al. 2004). Leonardo (2004) conducted a similar experiment using two clones of pea aphid collected on T. repens which again naturally contained R. insecticola. In contrast to Tsuchida et al.’s (2004) results, removal of R. insecticola did not signiWcantly aVect performance on T. repens. She also showed that the removal of another facultative symbionts, Serratia symbiotica, from three aphid clones did not alter fecundity on M. sativa (Leonardo 2004). In addition to their eVects on host-plant use, facultative symbionts have a variety of other eVects on their hosts. S. symbiotica and H. defensa can increase the aphid’s tolerance to heat shock (Chen et al. 2000; Montllor et al. 2002; Russell and Moran 2006) and confer resistance to the parasitoid wasp Aphidius ervi (Oliver et al. 2003, 2005). In contrast, R. insecticola appears to decrease tolerance to heat shock (Russell and Moran 2006). This bacterium has, however, been shown to increase resistance to the lethal fungal pathogen Pandora neoaphidis (Scarborough et al. 2005) and to aVect the proportion of sexual morphs produced in

123

Oecologia (2007) 153:323–329

autumn and the fraction of winged forms produced throughout the year (Leonardo and Mondor 2006). Early work on aphid secondary symbionts tended to use only single genotypes of aphid and bacteria, but the diVerences between Tsuchida et al.’s (2004) and Leonardo’s (2004) results on how R. insecticola aVects performance on Trifolium suggests that is important to study multiple aphid genotypes. Similarly, Chen et al. (2000) found that S. symbiotica signiWcantly decreased pea aphid fecundity when feeding on M. hispida and Lathyrus odoratus in only one of the three aphid clones tested, while the eVect of R. insecticola on the timing of sexual morph production reported by Leonardo and Mondor (2006) occurred in only two out of three aphid clones. SigniWcant diVerences among aphid genotype are not universal; using the same set of aphid clones as those employed here we found little inXuence of aphid genotype on the resistance to fungi endowed by R. insecticola, while Oliver et al. (2005) found that one strain of H. defensa increased parasitoid resistance in a similar manner across Wve aphid clones. Only two studies have systematically searched for diVerences between bacterial genotypes. Oliver et al. (2005) showed that while all Wve genotypes of H. defensa increased aphid resistance to parasitoids, the fraction surviving varied from 19 to 100%, and Russell and Moran (2006) demonstrated that diVerent genotypes of S. symbiotica cause varying levels of tolerance to heat shock. We report here the eVect of introducing R. insecticola on the ability of Wve clones of pea aphids to utilise two diVerent host plants, T. pratense and V. faba. We found that performance on T. pratense depended on bacterial genotype and/or its interaction with aphid genotype, while the presence of R. insecticola consistently led to poorer performance on V. faba. We discuss how these Wndings may aVect the explanation of the association between secondary symbionts and diVerent host-plant-adapted aphid clones, and the frequency with which diVerent symbionts occur in pea aphid populations.

Materials and methods Aphid clones and bacterial strains Novel pea aphid–R. insecticola associations were created by injecting bacteria from naturally infected “donor” aphid clones into symbiont-free “recipient” aphid clones. The creation of these aphid lines has been described previously in our study of the eVect of R. insecticola on aphid resistance to fungal pathogens (Scarborough et al. 2005). BrieXy, Wve diVerent recipient aphid clones were used, three originally collected from V. sativa (clones 9813, 9820 and 9821) and two from Lotus pedunculatus (9815, 9817). These aphid

Oecologia (2007) 153:323–329

clones were chosen because they harboured no secondary symbionts and thus are not a random sample of pea aphid genotypes. Pea aphid clones collected from Vicia tend to perform relatively well on T. pratense and have similar plant preferences as aphid clones collected from T. pratense (Ferrari et al. 2006 and unpublished data), whereas aphid clones from L. pedunculatus are more variable in their performance on T. pratense (Ferrari and Godfray 2003). We conWrmed that these aphid clones were genetically distinct using four microsatellite markers. Each of these aphid clones was injected with haemolymph from a diVerent donor aphid clone; four of these originally came from T. pratense and one from V. sativa (Scarborough et al. 2005). We maintained a set of “recipient-control” lines that were genetically identical to the recipient aphid clones and which had been injected with distilled water. Initially, R. insecticola was identiWed by partially sequencing the mitochondrial 16S rDNA gene (Scarborough et al. 2005). The recipient aphid clones were maintained for at least eight generations and regularly screened for infection with R. insecticola using diagnostic polymerase chain reaction (PCR) (Sandström et al. 2001). At the beginning of the performance and preference experiment, the bacterial densities in the haemolymph of the artiWcially injected aphid clones were determined using a haemocytometer and were slightly, but not signiWcantly, higher than in the naturally infected donor aphid clones (Scarborough et al. 2005). The counts also conWrmed the diagnostic PCR result that the recipient-control lines harboured no secondary symbionts. The donor aphid clones carried no other symbionts apart from R. insecticola and B. aphidicola. This experimental design allows us to test the eVect of aphid genotype on host-plant use, but note that we cannot distinguish the eVect of bacterial strain from an interaction between aphid genotype and bacterial genotype. The aphids were cultured on broad bean, V. faba var. The Sutton, at 20°C, 70% relative humidity and a 16:8 h light:dark light cycle. V. faba is a plant species that most pea aphid clones perform well on (Müller 1962; Sandström and Pettersson 1994; Ferrari and Godfray 2006) and is commonly used as a rearing plant in the laboratory. For the experiments we used commercially available T. pratense seeds (Emorsgate Wild Seeds, UK) and V. faba var. The Sutton (Moles Seeds, UK). Acceptance of and performance on Trifolium and Vicia Acceptance and performance were assayed in no-choice tests using a protocol adapted from Ferrari and Godfray (2006). For each replicate, Wve young wingless aphids (12–13 days old) were transferred into a Petri dish that contained either three leaves of T. pratense or one leaf of V. faba with their stalks in a 0.5-ml Eppendorf tube

325

containing water. The total leaf area was approximately the same for both plant species. After 24 h, the number of oVspring produced was counted. We interpret this as a measure of test plant acceptance rather than of fecundity on that plant, as within a day it is unlikely that the aphid will have suVered substantially from feeding on an unsuitable host plant (Ferrari and Godfray 2006). To assess performance, ten of the oVspring produced in each dish were arbitrarily chosen and transferred to a new potted test plant. After 7 days the survivors were counted and one average-sized wingless survivor was chosen and transferred to another plant where it was left to reproduce for 8 days. Fecundity was estimated as the total number of oVspring produced over this period. In the analysis, we use the product of survival and fecundity as a composite measure of performance (Ferrari and Godfray 2006). The experiment was repeated 8 times for each aphid line (donor, recipient and recipient-control lines), with lines stratiWed across four temporal blocks. The donor, recipient and recipient-control lines were tested simultaneously. Preference for Trifolium or Vicia In order to test whether infection with R. insecticola aVects the preference of the aphid for T. pratense and V. faba, ten wingless adults (11–12 days old) were given the choice of two potted V. faba and two potted T. pratense plants of approximately the same size. The plants were arranged alternately in a 30 £ 30 £ 30-cm perspex cage. The aphids were released in the centre of the arena and after 3 days the number of adults on each plant species was counted. In such experiments pea aphids choose host plants quickly and their distribution across plants is constant after 72 h (Caillaud and Via 2000). This experiment was conducted in four temporal blocks, with two replicates for each aphid line in each block. Analysis The analyses for each measure of host use (acceptance, performance and preference) were performed in three steps. First, we compared the recipient and the recipient-control aphid clones to test the eVect of symbiont status on aphid traits. In this analysis, the unit of replication was the aphid line and the response variable was average host-plant acceptance or performance. We Wtted aphid clone as a discrete factor to account for aphid genetic variation and then tested the eVects of symbiont status, host plant and their interaction. Second, a similar analysis was performed to compare the artiWcially infected recipient aphid clones and the naturally infected donor aphid clones. Finally, we explored among-clone diVerences in the eVects of symbionts using the raw data from the replicate assessments of

123

326

Oecologia (2007) 153:323–329

aphid acceptance and performance. The analysis of hostplant preference was similar but used the proportion of aphids that was found on Vicia as the response variable (test plant is therefore not a factor in this analysis). All analyses controlled for temporal blocks.

Results Acceptance R. insecticola had a negative eVect on aphid acceptance which is deWned above as the number of progeny produced on Wrst encountering either of the two host plants (F1,12 = 17.03, P = 0.001; Fig. 1a). This eVect was evident on both host plants (no plant £ symbiont interaction, F1,12 = 1.20, P = 0.29), although overall acceptance was lower on T. pratense (F1,12 = 24.83, P < 0.001). The naturally infected donor aphid clones had overall higher levels of acceptance than the artiWcially infected recipient aphid clones on both plants (infection type, F1,16 = 56.16, P < 0.001; infection type £ test plan, F1,16 = 0.39, P = 0.54; Fig. 1b). There was signiWcant among-aphid clone variation in acceptance (F4,137 = 21.48, P < 0.001) and aphid clones also varied in their response to the test plants (clone £ test plant, F4,137 = 3.14, P = 0.02). The negative eVect of

(a) 70

R. insecticola was less severe for some aphid clones (clone £ symbiont, F4,137 = 3.79, P = 0.006). Performance Aphid performance was much higher on V. faba than on T. pratense (F1,12 = 63.55, P < 0.001; Fig. 1c). When comparing the naturally and artiWcially infected recipient aphid clones, there was no overall eVect of the symbiont on aphid performance (F1,12 = 0.06, P = 0.81), nor was there a signiWcant symbiont treatment by host plant interaction (F1,12 = 3.32, P = 0.09) as would be expected if R. insecticola consistently improved performance on T. pratense. Performance was higher in the naturally infected donor aphid clones compared to the artiWcially infected recipient aphid clones (F1,16 = 58.52, P < 0.001); this eVect was more pronounced on T. pratense than on V. faba (infection type £ test plant, F1,16 = 15.89, P = 0.001; Fig. 1d) There was signiWcant among-aphid clone variation in performance (F4,137 = 26.89, P < 0.001) and the aphid clones also varied in their ability to use both plants irrespective of symbiont status (clone £ plant, F4,137 = 10.07, P < 0.001). The eVect of R. insecticola on diVerent aphid clones was complex as indicated by signiWcant clone £ symbiont (F4,137 = 3.01, P = 0.02) and clone £ plant £ symbiont (F4,137 = 3.63, P = 0.008) interactions. R. insecticola

(b) 70

Recipient Clones

Performance

Acceptance

50 40 30 20

40 30 20 10

0

0

(d) 60

Recipient Clones

50 40 30 20 10

Donor Clones

50 40 30 20 10

0

0 9813

9815

9817

9820

9821

Aphid Clone

9824

9909

9911

9913

9915

Aphid Clone

Vicia, symbiont free

Trifolium, symbiont free

Vicia, R. insecticola

Trifolium, R. insecticola

Fig. 1a–d EVect of Regiella insecticola on use of Vicia faba and Trifolium pratense by the pea aphid. a Acceptance of diVerent host plants by symbiont-free and artiWcially infected aphid clones: R. insecticola negatively aVected acceptance of V. faba and T. pratense for all aphid clones, but the strength of this eVect varied among aphid clones; b acceptance by the naturally infected donor aphid clones; c performance

123

50

10

(c) 60

Donor Clones

60

Performance

Acceptance

60

of symbiont-free and artiWcially infected aphid clones: R. insecticola had a negative eVect on V. faba and a positive inXuence on performance on T. pratense for one aphid clone, but not the other four; d performance of the donor clones. Performance was measured as a composite of survival and fecundity. The bars show means per aphid line § SE

Oecologia (2007) 153:323–329

reduced the performance of all aphid clones on V. faba, while it had no eVect for four of the Wve clones on T. pratense. However, one aphid clone had signiWcantly increased performance on T. pratense when it harboured the symbiont. Preference All symbiont-free and artiWcially infected aphid lines had a clear preference for V. faba; in each line at least 85% of the aphids were found on V. faba. The symbiont did not alter this preference consistently (F1,4 = 0.08, P = 0.80). The naturally infected aphid clones chose T. pratense at signiWcantly higher frequencies (58–67%) than the artiWcially infected aphids (F1,8 = 164.87, P < 0.001). There was no signiWcant variation among the recipient aphid clones in preference (F4,67 = 2.19, P = 0.08). However, R. insecticola had a small, but signiWcant eVect on modifying this preference: the bacteria increased preference for V. faba in three aphid clones and decreased it in two aphid clones (clone £ symbiont, F4,67 = 2.75, P = 0.04). Preference for V. faba was signiWcantly correlated with relative performance across the Wve control and the Wve artiWcially infected aphid clones (r8 = 0.67, P = 0.04).

Discussion The endosymbiont R. insecticola inXuences the way that aphids choose and exploit two of their host plants, V. faba and T. pratense, but the eVect varied signiWcantly among the genotypes tested. In particular, despite the strong statistical association between the presence of R. insecticola and use of Trifolium (Tsuchida et al. 2002; Simon et al. 2003; Leonardo and Muiru 2003; Ferrari et al. 2004 and unpublished data), only one of the Wve aphid clones we tested showed enhanced performance on T. pratense, and even here the Wtness of the aphid on this host plant was low compared to that on V. faba. In comparison, Tsuchida et al. (2004) found worse performance by a single aphid clone on T. repens when R. insecticola was removed, while Leonardo (2004) did not Wnd any diVerences in two aphid clones. The eVect of R. insecticola on the pea aphid’s ability to use Trifolium is thus complex and strongly inXuenced by aphid genotype and by bacterial genotype or an interaction between bacterial and aphid genotypes. Injection of R. insecticola consistently decreased performance on V. faba and consistently reduced our measure of acceptance on both host plants. In many circumstances, R. insecticola carriage thus appears bad for the aphid. These negative eVects may be due to the lack of adaptation of the bacteria to the aphid in the relatively short time (eight

327

generations) since they were introduced, or to a more general and long-term cost of symbiosis. However, the fact that we found in this same set of lines that R. insecticola strongly and consistently improved aphid resistance to pathogenic fungi (Scarborough et al. 2005) indicates that the interaction is suYciently mature to provide conditional beneWts to the aphids. Donor aphid clones in general had higher Wtness measures in the experiments compared with recipient aphid clones. This diVerence may be due to a longer period of bacterial adaptation in these aphid clones, but is more likely due to clonal selection for high-Wtness combinations of bacteria and aphid genotypes. Preference for V. faba and T. pratense was only very slightly aVected by R. insecticola and the recipient aphid clones used had an overwhelming preference for V. faba. It is possible that a clearer eVect could have been observed with aphid clones that had a greater natural propensity to choose T. pratense and that had been reared on T. pratense prior to the experiment. The clear diVerence between the donor and recipient aphid clones in their preference for T. pratense shows, however, that the rearing plant V. faba did not obscure all diVerences in plant preference between aphid lines. Additionally, preference and relative performance of the recipient and recipient-control aphid clones on V. faba were positively correlated, indicating that Regiella might have similar eVects on performance and preference. In the United States, pea aphid populations are specialised on two important legume crops and forage, M. sativa and T. pratense (Via 1991, 1999). Similar host-adapted populations have been recorded from the Old World, from where pea aphids originated (Sandström 1996; Ferrari and Godfray 2003, 2006; Ferrari et al. 2006). Genetic analysis has identiWed quantitative trait loci for preference for, and performance on, T. pratense and M. sativa (Hawthorne and Via 2001). Current data suggest that maternal transmission of secondary symbionts occurs more frequently than paternal transmission (Moran and Dunbar 2006). In genetic analyses, any inXuence of bacterial symbionts on aphid performance would thus appear as a maternal eVect, but these have never been found to be strong (S. Via, personal communication). We established novel R. insecticola–pea aphid associations, using pea aphid clones that are poorly adapted to Trifolium. Our results show that R. insecticola does not or only marginally increases aphid performance on T. pratense, depending on the bacterial genotype or its interaction with the aphid genotype. This suggests that R. insecticola has not been a major factor in aphids switching to Trifolium from other legume genera. Of course, we have only tested a limited number of aphid clones, and it may be that symbionts do improve the performance of those aphid clones that are already at least partially adapted to Trifolium. The experiments presented here do not directly

123

328

address this possibility, but provide evidence that, at least in principle, the conXict between the two previous reports by Tsuchida et al. (2004) and Leonardo (2004) may be caused by genetic variation in the eVect of R. insecticola on host use. Another possibility that we have not explored here is that the eVect of R. insecticola on aphid Wtness varies depending on the Trifolium genotype. If symbiont presence does not have a major eVect on aphid performance, what might explain the observed nonrandom distribution of R. insecticola across host-associated pea aphid populations? R. insecticola may have other phenotypic eVects on the aphid that particularly beneWt aphid clones on Trifolium. As noted above, R. insecticola markedly improves aphid survival after infection by the parasitic fungus Pandora neoaphidis (Scarborough et al. 2005) and were this pathogen to be especially prevalent on Trifolium this might explain the association. Unfortunately, we know of no data on host-plant-speciWc pathogen or other natural enemy mortality. A related explanation is that harbouring R. insecticola carries a cost when feeding on V. faba, as shown here, and perhaps on other host plants, and that this cost is not outweighed by the beneWts of protection from the fungus. Either explanation would account for the lack of fungal resistance in pea aphid populations other than those found on Trifolium (Ferrari and Godfray 2003, 2006). A further issue is why secondary symbionts are able to invade pea aphid populations, yet do not reach Wxation like B. aphidicola. Recently, Oliver et al. (2006) found that aphid Wtness was strongly depressed when infected by multiple symbionts, possibly as a result of over-proliferation of one of the symbionts or antagonistic interactions between the bacteria. This provides an explanation of why multiple symbionts do not reach Wxation, but not why a single symbiont does not exclude the others. Possibly the beneWts of carrying diVerent symbionts vary in time and/or space, giving each the possibility of invading when rare. Symbiont spread may also be aVected by interactions with aphid genotype if only certain aphid clones beneWt from carrying the symbiont. This possibility is supported by the results reported here, and also by Oliver et al.’s (2005) and Russell and Moran’s (2006) studies. Local adaptation to diVerent host plants in plant-feeding insects has long been considered a possible route to ecological speciation (Bush 1969; Via 2001; Drès and Mallet 2002). A combination of strong selection for performance on diVerent host plants, plus a correlation between hostplant use and mating site or mate choice can lead to genetic diVerentiation which may then be strengthened by reinforcement. Bacterial endosymbionts could inXuence ecological speciation in aphids and other herbivorous insects by facilitating adaptation to diVerent host plants. We have shown here that although an endosymbiont can positively aVect host use, this depends on host genotype or its interaction

123

Oecologia (2007) 153:323–329

with bacterial genotype. We need now to understand these genotypic interactions in more detail and also to assess whether the endosymbionts’ eVect on host use is systematically linked to their eVect on further traits that may inXuence assortative mating and hence the speciation process (Leonardo and Mondor 2006). Only then will it be possible to assess whether or not facultative symbionts may be playing an important role in the ecological speciation process. Acknowledgements C. L. S. was supported by a NERC studentship. Our work complied with UK law.

References Bush GL (1969) Sympatric host race formation and speciation in frugivorous Xies of the genus Rhagoletis (Diptera: Tephritidae). Evolution 23:237–251 Caillaud MC, Via S (2000) Specialized feeding behavior inXuences both ecological specialization and assortative mating in sympatric host races of pea aphids. Am Nat 156:606–621 Chen DQ, Purcell AH (1997) Occurrence and transmission of facultative endosymbionts in aphids. Curr Microbiol 34:220–225 Chen DQ, Montllor CB, Purcell AH (2000) Fitness eVects of two facultative endosymbiotic bacteria on the pea aphid, Acyrthosiphon pisum, and the blue alfalfa aphid, A. kondoi. Entomol Exp Appl 95:315–323 Darby AC, Douglas AE (2003) Elucidation of the transmission patterns of an insect-borne bacterium. Appl Environ Microbiol 69:4403–4407 Douglas AE (1998) Nutritional interactions in insect-microbial symbioses: aphids and their symbiotic bacteria Buchnera. Annu Rev Entomol 43:17–37 Drès M, Mallet J (2002) Host races in plant-feeding insects and their importance in sympatric speciation. Philos Trans R Soc B Biol Sci 357:471–492 Ferrari J, Godfray HCJ (2003) Resistance to a fungal pathogen and host plant specialization in the pea aphid. Ecol Lett 6:111–118 Ferrari J, Godfray HCJ (2006) The maintenance of intraspeciWc biodiversity: the interplay of selection on resource use and on natural enemy resistance in the pea aphid. Ecol Res 21:9–16 Ferrari J, Darby AC, Daniell TJ, Godfray HCJ, Douglas AE (2004) Linking the bacterial community in pea aphids with host-plant use and natural enemy resistance. Ecol Entomol 29:60–65 Ferrari J, Godfray HCJ, Faulconbridge AS, Prior K, Via S (2006) Population diVerentiation and genetic variation in host choice among pea aphids from eight host genera. Evolution 60:1574–1584 Frantz A, Plantegenest M, Mieuzet L, Simon J-C (2006) Ecological specialization correlates with genotypic diVerentiation in sympatric host-populations of the pea aphid. J Evol Biol 19:392– 401 Haynes S, Darby AC, Daniell TJ, Webster G, van Veen FJF, Godfray HCJ, Prosser JI, Douglas AE (2003) Diversity of bacteria associated with natural aphid populations. Appl Environ Microbiol 69:7216–7223 Hawthorne DJ, Via S (2001) Genetic linkage of ecological specialization and reproductive isolation in pea aphids. Nature 412:904– 907 Leonardo TE (2004) Removal of a specialization-associated symbiont does not aVect aphid Wtness. Ecol Lett 7:461–468 Leonardo TE, Mondor EB (2006) Symbiont modiWes host life-history traits that aVect gene Xow. Proc R Soc B Biol Sci 273:1079–1084

Oecologia (2007) 153:323–329 Leonardo TE, Muiru GT (2003) Facultative symbionts are associated with host plant specialization in pea aphid populations. Proc R Soc B Biol Sci 270:S209–S212 Montllor CB, Maxmen A, Purcell AH (2002) Facultative bacterial endosymbionts beneWt pea aphids Acyrthosiphon pisum under heat stress. Ecol Entomol 27:189–195 Mopper S, Strauss SY (1998) Genetic structure and local adaptation in natural insect populations: eVects of ecology, life history, and behavior. Chapman and Hall, New York Moran NA, Baumann P (2000) Bacterial endosymbionts in animals. Curr Opin Microbiol 3:270–275 Moran NA, Dunbar HE (2006) Sexual acquisition of beneWcial symbionts in aphids. Proc Natl Acad Sci USA 103:12803–12806 Moran NA, Russell JA, Koga R, Fukatsu T (2005) Evolutionary relationships of three new species of Enterobacteriaceae living as symbionts of aphids and other insects. Appl Environ Microbiol 71:3302–3310 Müller FP (1962) Biotypen und Unterarten der “Erbsenlaus” Acyrthosiphon pisum (Harris). Z PXanzenkd PXanzensch—J Plant Dis Prot 69:129–136 Oliver KM, Russell JA, Moran NA, Hunter MS (2003) Facultative bacterial symbionts in aphids confer resistance to parasitic wasps. Proc Natl Acad Sci USA 100:1803–1807 Oliver KM, Moran NA, Hunter MS (2005) Variation in resistance to parasitism in aphids is due to symbionts not host genotype. Proc Natl Acad Sci USA 102:12795–12800 Oliver KM, Moran NA, Hunter MS (2006) Costs and beneWts of a superinfection of facultative symbionts in aphids. Proc R Soc B Biol Sci 273:1273–1280 Russell JA, Moran NA (2006) Costs and beneWts of symbiont infection in aphids: variation among symbionts and across temperatures. Proc R Soc B Biol Sci 273:603–610 Russell JA, Latorre A, Sabater-Muñoz B, Moya A, Moran NA (2003) Side-stepping secondary symbionts: widespread horizontal transfer across and beyond the Aphidoidea. Mol Ecol 12:1061–1075

329 Sandström J (1996) Temporal changes in host adaptation in the pea aphid, Acyrthosiphon pisum. Ecol Entomol 21:56–62 Sandström J, Pettersson J (1994) Amino-acid-composition of phloem sap and the relation to intraspeciWc variation in pea aphid (Acyrthosiphon pisum) performance. J Insect Physiol 40:947–955 Sandström JP, Russell JA, White JP, Moran NA (2001) Independent origins of horizontal transfer of bacterial symbionts of aphids. Mol Ecol 10:217–228 Scarborough CL, Ferrari J, Godfray HCJ (2005) Aphid protected from pathogen by endosymbiont. Science 310:1781 Simon J-C, Carré S, Boutin M, Prunier-Leterme N, Sabater-Muñoz B, Latorre A, Bournoville R (2003) Host-based divergence in populations of the pea aphid: insights from nuclear markers and the prevalence of facultative symbionts. Proc R Soc B Biol Sci 270:1703–1712 Tsuchida T, Koga R, Fukatsu T (2004) Host plant specialization governed by facultative symbiont. Science 303:1989–1989 Tsuchida T, Koga R, Shibao H, Matsumoto T, Fukatsu T (2002) Diversity and geographic distribution of secondary endosymbiotic bacteria in natural populations in the pea aphid, Acyrthosiphon pisum. Mol Ecol 11:2123–2135 Via S (1991) The genetic structure of host plant adaptation in a spatial patchwork—demographic variability among reciprocally transplanted pea aphid clones. Evolution 45:827–852 Via S (1999) Reproductive isolation between sympatric races of pea aphids. I. Gene Xow restriction and habitat choice. Evolution 53:1446–1457 Via S (2001) Sympatric speciation in animals: the ugly duckling grows up. Trends Ecol Evol 16:381–390 Via S, Bouck AC, Skillman S (2000) Reproductive isolation between divergent races of pea aphids on two hosts. II. Selection against migrants and hybrids in the parental environments. Evolution 54:1626–1637

123