Aphididae are divided into several subfamilies, tribes and subtribes (we use here the most recent comprehensive classification of [6]). Among subfamilies and tribes, few are strictly associated with gymnosperms, some use herbaceous monocotyledons and most use various angiosperm families. Few molecular phylogenies have investigated the origin of these tribes and their relationships; we have attempted here to summarize these studies and what they tell us about the evolution of host plant association (Fig. 1).

Phylogenetic reconstruction places the coniferous feeding family of Adelgidae as a sister group of Aphididae [16]. Several lineages of Aphididae also use coniferous hosts, and these may represent ancient associations [17]; one such lineage is the subfamily Neophyllaphidinae, which has a Gondwanan distribution and is associated with two gymnosperm genera, Podocarpus and Araucaria. It has been suggested that early Aphididae ancestors lived on coniferous hosts, a hypothesis consistent with fossil records [18,19]. For example, Neophyllaphidinae look very similar to the Cretaceous fossil Aniferella, which by inference were probably associated with Podocarpaceae or Araucariaceae [20]. These two gymnosperm families were likely ancestral aphid hosts in the Early Cretaceous before angiosperms arose, along with several other abundant gymnosperm groups [21]. Fossils have allowed the calibration of molecular phylogenetic reconstructions: the age of the common ancestor of extant subfamilies in the Aphididae has been estimated at 84–99 mya [21], which corresponds to the Late Cretaceous epoch and roughly the diversification of woody angiosperms. The lack of resolution of molecular phylogenetic relationships among Aphididae subfamilies and tribes and the short branches found in the deep nodes of the phylogenetic tree suggest a rapid radiation of aphids at this period [21,22]. A more recent study found tentative support for three main lineages, with Lachninae as sister group to all other aphids; as one of the few subfamilies with conifer feeders, the position of Lachninae also supports the hypothesis of an ancestral gymnosperm feeding in aphids [23] with the caveat that conifer feeding in lachnines is thought to be a derived state by some systematists [24,25]. That most of present-day aphid diversity likely arose after one or more shifts from gymnosperms to angiosperms strongly suggests that the radiation of aphids has been driven by the diversification of flowering plants [26]. Fewer than 30 extant species remain in two subfamilies thought to be primitively conifer feeding (Neophyllaphidinae and Mindarinae), compared to more than 4000 angiosperm-feeding species. These observations suggest that either aphids could have speciated in parallel with particular groups of angiosperms (i.e., have partly cospeciated) [27] or that their diversification has rapidly followed the radiation of angiosperm families, favoured by the diversity of new ecological niches offered by these plants.

Compared to other phytophagous insect groups of similar age, aphid biodiversity is rather low. Furthermore, aphids are relatively rare in the tropics, where plant diversity is the highest, and underrepresented in the southern hemisphere. The plant families cited by Eastop [24] and Dixon [9] as having no specific aphids are mostly diversified in tropical or hot regions and/or in the southern hemisphere. Thus, absence of diversification on certain plant families may have little to do with aphid feeding preferences. Several hypotheses might explain the paucity of aphids in the tropics and southern hemisphere. One of them suggests that aphids have been less successful in tropical regions due to the difficulty of finding their specific host plants in the highly diversified tropical ecosystems; aphids seem to locate their hosts somewhat randomly and may depend on high plant apparency to do so [9,28]. Another hypothesis suggests that cyclical parthenogenesis, the ancestral life cycle of all sexually reproducing species, is maladapted to the tropics [29]. Low diversity in the southern hemisphere compared to the northern one has also been explained in a historical biogeographic context, predicated on the idea that aphids originated in the northern hemisphere and underwent extensive species radiations there during the Tertiary Period; subsequently the tropics acted as a barrier to southward migrations [30]. Aphid geographical distribution and plant association is most certainly the result of both ecological and historical constraints. Comparative biogeographical studies of aphids and their host plants integrating advances in molecular phylogenetics are still quite rare [31,32], and more are needed to elucidate several of the major transitions in aphid diversification.

Host alternation, a characteristic trait of certain aphid life cycles, may have played an important role in favouring the colonization of new plants and speciation in aphids. About 10% of aphid species alternates seasonally, and mostly obligately, between two unrelated plant species [7,33,34]; these aphids are said to be host alternating or heteroecious, in contrast to the vast majority of species that complete their entire life cycle on the same host (known as non-host-alternating or monoecious species). The evolution of host alternation has occurred in several aphid families, in which the cycle follows a roughly similar pattern of host use: after egg hatching followed by parthenogenetic development of colonies on their primary/winter host (most generally a woody host), aphids alternate to secondary/summer hosts (most generally herbaceous plants) on which they reproduce parthenogenetically; later generations return to their primary host for sexual production of eggs and usually overwintering. The two host plants of host alternating species are almost never closely related, and some consist of rather surprising combinations. For instance, the Nearctic species Melaphis rhois (Fordini) and the related eastern-Asian species of Nurudea, Kaburagia and Schlechtendalia (Fordini) alternate between Rhus (Anacardiaceae) and mosses [35]. The exact number of evolutionary gains and/or losses of this life cycle throughout aphid history is not known, but almost certainly both gains and losses have occurred multiple times, and some proportion of extant aphid-host associations probably have been shaped by such transitions [35]. The evolution of host alternation has two intricately linked consequences:

• (i) the colonization of new ecological niches via a new life cycle;
• (ii) the broadening of host plant range.

Host-alternating species often have a much larger range of secondary hosts than primary hosts (see section 2.2 for possible causes). Hence, host alternation seems to allow for an escape from host plant specialization. This transition in aphid diet might be followed by a shift of all or part of the population to permanent existence on some of the new secondary hosts, and eventually leading to speciation. These sequences of evolutionary changes could have fuelled the diversification of certain aphid lineages [34].

Such a scenario entailing acquisition of secondary hosts and broadening of host range, followed by specialization and speciation on newly acquired hosts, might be exemplified by the diversification of Aphidinae. With about 57% of all described aphid species [24], Aphidinae constitutes the most species-rich subfamily. It encompasses many genera that live exclusively or partially on Rosaceae and are also associated with modern grasses. It also includes many highly polyphagous species among which most of the agricultural aphid pests are found. The scarcity of Eocene and Oligocene fossils suggests that this lineage probably emerged and diversified in the latter half of the Tertiary (Miocene and later), when plant communities dominated by grasses became abundant [19,26]. Molecular phylogenetic reconstruction implies several evolutionary transitions in life cycles in this subfamily [36]. Further, phylogenies suggest that several shifts to and from Rosaceae have occurred. One plausible evolutionary scenario is that ancestors of Aphidinae fed on a woody host, possibly an early Rosaceae species [27]. This lineage then radiated following the acquisition of host alternation to herbaceous hosts; in some cases descendent lineages lost their primary rosaceous hosts to become monoecious on more recently acquired secondary hosts (also called secondary monoecy) [8,19,36,37]. Several genera in Aphidinae exhibit all the life cycle states in this hypothetical process (primary monoecy, host alternation, and secondary monoecy), including Dysaphis, Cryptomyzus, Myzus, and Brachycaudus. In contrast, no other aphid lineages with host alternation (Eriosomatinae, Hormaphidinae, Anoeciinae) exhibit complete secondary monoecy including a sexual generation. Given the disparity in species diversity between Aphidinae and other lineages, it seems possible that acquisition followed by loss of heteroecy may have accelerated the diversification process in this lineage [34].

A true comparative analysis of the diversification rate is difficult to conduct based on the current phylogenetic reconstructions, but it would be interesting to test whether increased diversification rates are correlated with acquisition and loss of host alternation. Such comparative tests could be conducted at several taxonomic levels: among subfamilies, and among genera within Aphidinae. These analyses could yield insights into the historical role of evolutionary transitions in life cycle and host plant range in aphid speciation.

As mentioned above, aphids are specialized at several taxonomic levels. More than half of all aphid species feed on a single plant species, and many aphid genera are restricted to a single plant genus or several closely related plant genera [24]. Among the 445 non-host-alternating European Aphidinae, 407 (91%) feed on host plant species confined to a single genus [24]. Such host specificity extends across all aphid lineages: for example, all primary hosts of Pemphigus (Eriosomatinae) belong to Populus [8]; all species of Chaitophorus (Chaitophorinae) are confined to the genera Salix and Populus (both in Salicaceae). Some tribes are restricted to one or more genera within a single plant family: all primary hosts of Eriosomatini belong to two genera of Ulmaceae; Hormaphidini use primary hosts in Hamamelis (Hamamelidaceae); Siphini (Chaitophorinae) live on numerous genera of Poaceae, etc. Some aphid genera are confined to one or two plant families, e.g., Uroleucon feed only on Asteraceae and Campanulaceae [8,38]. Of course, such a pattern could result from the fact that many aphid taxa were circumscribed based on host association. However, these affinities of closely related aphids for particular host plant groups are confirmed by phylogenetic studies, and in some cases, aphid phylogenetic relationships are better explained by their host plant similarity than by morphological criteria. The molecular phylogenetic reconstruction of the Aphidinae clusters some aphid genera that are morphologically divergent but use related plants [36]: Myzus and Brachycaudus were thought unrelated based on their morphology [39], but nevertheless were found closely related in molecular studies, and both contain many species using Prunoideae as a host plant (as sole hosts or primary hosts). Similarly, pterocommatines (Pterocomma, Plocamaphis, etc.), feeding on Populus and Salix (Salicaceae), are so morphologically different from any other aphids as to be placed in a separate subfamily (Pterocommatinae) by a recent classification [6]; however, molecular phylogenetics placed pterocommatines firmly within Aphidinae-Macrosiphini as a sister group to Cavariella, the only other lineage of Aphidinae that has also diversified on Salicaceae [36].

The degree of host plant specialization in phytophagous insects is difficult to define and often subjective. Depending on the breadth of their host range, aphids have been classified as monophagous, oligophagous, polyphagous and even pantophagous [24]. An aphid species that uses several plant species from the same genus probably exhibits a higher level of host specialization than a species that uses only two hosts, which are drawn from distantly related plant families. The former suggests stronger phylogenetic constraints in the colonization of new host plants and is more indicative of specialization from an evolutionary point of view. Blackman and Eastop [7,29,33] compiled an impressive body of data on host-plant association for all aphids. However, the level of host specialization in aphids has still not been extensively compared between subfamilies and genera. To date, only a few generalities have been drawn, such as all aphid species feeding on conifers are monophagous or oligophagous; aphids associated with monocotyledons are less specific than aphids associated with dicotyledons [18,40]; galling aphids (e.g., Hormaphidinae) are often very specialized [41]. Furthermore, levels of specialization in aphids have rarely been compared with other groups of phytophagous insects. Compared to other hemipterans, such as psyllids [24], aphids might not be as strictly monophagous, but they exhibit greater phylogenetic structure in their host plant ranges.

Several types of selective pressures could favour a specialized diet in aphids. Here we briefly review behavioural and life-history traits of aphids that are considered as important in shaping host-plant association ([9,42] for more comprehensive discussions). Plant chemical defences (secondary compounds), which are usually involved in plant-insect coevolution [43–45], have rarely been implicated in shaping long-term patterns of aphid-plant associations. As phloem feeders, aphids probably circumvent many of those defences ([9] pp. 32–38 regarding the occurrence of chemical defences in phloem). One hypothesis to explain their specialist diet is simply that particular aphid species just perform better on particular plants that are more suitable as a resource for them; specificity would thus be the result of metabolic adaptation (see section 3) [42]. Mechanisms (behavioural, chemical) of host-plant location and recognition in aphids may also play an important role in either favouring or discouraging specialization [9]. However, the fact that some aphids host alternate between very different plant species argues against host specificity being solely a consequence of adaptation for finding and using the resources offered by a specific plant. An alternative hypothesis has been proposed by Ward [46], who suggested that host specificity was favoured because host plants serve not only as a source of nutrients but also as a mating site (termed the “rendezvous” hypothesis [9]); hence, host-range limitation is probably strongly linked to mating success. This hypothesis is clearly in agreement with the observation that heteroecious species exhibit a wider diversity of secondary hosts and a narrower range of primary hosts, where the sexual phase of the cycle is produced and mating occurs. Of the 141 heteroecious European Aphidinae, only nine have their secondary host plants confined to one genus [24]. Another strongly debated hypothesis is that of “fundatrix specialization” [37,47,48]. Under this hypothesis, host specialization in aphids is due largely to the phylogenetic constraints on the morph that hatches from overwintering eggs in spring, the fundatrix, and perhaps also the sexual morphs. These morphs would be strongly adapted to the primary, often ancestral hosts, and less able to switch hosts than would other morphs. This hypothesis is also consistent with the general pattern of a narrower primary host-plant range than secondary host plant range in heteroecious species.

There are no studies on the evolution of host association at the level of Aphididae as a whole; however, several case studies have focused on the historical association of certain aphid tribes and genera and their host plants in order to investigate how specialization, cospeciation and host shifts have influenced diversification. These studies are summarized in Table 1 and reviewed below.

Through a phylogenetic reconstruction of North American Cinara, Favret and Voetglin [49] proposed that speciation was accompanied by host shifts. Closely related species in the Cinara phylogenetic tree are often found associated with different Pinus species, while they maintain similar feeding sites (they are either branch or shoot feeders). This suggests that these aphids are better at switching hosts than at colonizing different ecological niches on the same host plant. Although the Pinus species serving as hosts of closely related Cinara are sympatric, no information exists on whether these host shifts and specialization occurred in sympatry or whether they were facilitated by geographical separation of populations.

Several phylogenetic and ecological studies of galling aphids in Fordini have been published [50–52]. Fordines are often host specific and repeated host shifts between related hosts have been inferred [51]. However, several occurrences of aphid sister species sharing the same host plant species suggest that host shifts followed by host specialization and speciation are not the sole mechanism promoting diversification in this group. Study of two sympatric Pemphigus species sharing the same host (P. populi and P. obesinymphae on Populus deltoides) suggests that allochronic speciation (temporal isolation) has occurred. Hence, variation in plant phenology could be an important factor in generating aphid diversity [53].

Containing more than 197 valid species, Uroleucon constitutes one of the many genera that probably arose during the diversification of aphids on herbaceous plants in the Tertiary. Most Uroleucon species are generally restricted to a single host plant species or to several closely related species within Asteraceae or Campanulaceae. In some cases, species complexes are associated with groups of closely related plants; thus, Uroleucon is a seemingly good candidate for an example of aphid-host parallel diversification. However, a molecular phylogenetic reconstruction of the genus, based on only 17 species, clearly does not support a cospeciation hypothesis [38]: Uroleucon species have shifted repeatedly between Asteraceae tribes. Furthermore, the phylogeny implies that Uroleucon ancestors were not associated with a common ancestor of Campulanaceae and Asteraceae, as would be expected under a cospeciation scenario; rather, the association with Campulanaceae is a derived condition. Estimation of the age of the radiation suggests that Uroleucon species have diversified after Asteraceae became abundant in the fossil record, and speciation linked to host specialization has more likely originated through the acquisition of existing host taxa. Again, it is difficult to conclude whether host specialization followed by speciation could have occurred in sympatry based on molecular phylogenies alone. The authors of this study mentioned that most host plants of Uroleucon overlap over large parts of their geographic range; therefore, because mating does occur on the specific host plant, reproductive isolation could be favoured even in sympatry. Rapid diversification (as indicated by the lack of resolution in the deep nodes of the phylogeny) is also in favour of a sympatric mode of speciation [38].

A thorough study of speciation mode in Cryptomyzus aphids was conducted using phylogenetic studies and experimental work [54,55]. There are 17 valid species in this genus, with some species alternating between Ribes and several herbaceous species of the Lamiaceae family, and other species being monoecious on Lamiaceae. Two possible pathways of speciation were proposed in Cryptomyzus: host shift of a monoecious species onto a new herbaceous host followed by specialization and reproductive isolation (similar to the scenario inferred for Uroleucon [38]), and speciation via loss of the heteroecious life cycle and shift of all primary-host generations onto some of the secondary hosts of the original life cycle [54,55].

The genus Brachycaudus presents a pattern of life cycle evolution resembling the situation in Cryptomyzus. It includes heteroecious species (often alternating from a Prunus tree to herbaceous hosts from very diverse angiosperm families) and monoecious species that are often found on the secondary hosts of closely related heteroecious species [56]. Brachycaudus species also vary greatly in their host plant repertoire, ranging from strictly monophagous (most monoecious species) to highly polyphagous on secondary hosts when heteroecious; the latter condition might favour the capture of new hosts. This genus potentially illustrates speciation via acquisition of new, sometimes distantly related, hosts during the “summer” phase of the heteroecious life cycle, and through changes in life cycle (Jousselin et al. in prep).

Overall, the few studies of the history of host association within aphid genera all demonstrate that speciation is often associated with a host shift. However, these macroevolutionary patterns do not firmly demonstrate the role of plant use on reproductive isolation in aphids. At the most, they provide hints that host plant specialization, together with life cycle evolution, has probably played central roles in aphid diversification. Assessing the role of host plant association as a promoter of speciation necessitates intraspecific studies within natural populations, which are reviewed below.

Host plants exert considerable selective pressures on their associated aphids, constituting their only food resource and habitat (including mating and oviposition sites). Because the vast majority of aphid species are host-specific (see section 2), it is tempting to infer their origin and divergence as a direct consequence of adaptation to different host plants, a hypothesis that was proposed long ago for phytophagous insects [57]. However, distinct host specificities may just result from local adaptation and competition between species that have previously diverged in geographical isolation (allopatry) [58]. Under such processes, host specialization would merely allow the coexistence of host-adapted insects as a consequence of reproductive isolation, rather than causing it. This hypothesis does not appear to be considered in the scientific literature on aphid ecology [9].

By contrast, speciation promoted by divergent selection in contrasting environments, a process termed “ecological speciation” [59], is actively studied in a variety of organisms, including fishes, birds, fungi and plants [59,60]. Premating isolation (reduction of hybridization rate) may arise from mating occurring within the preferred habitat. In addition, the contrasting parental environments may not be suitable for hybrids, thereby causing post-zygotic isolation (hybrid unfitness). While the contribution of divergent natural selection to particular components of reproductive isolation has been quantified only in a handful of biological models [61–63], phytophagous insects, including aphids, offer good systems for studying these ecological reproductive barriers [64–66]. Ecological speciation can theoretically occur under any geographical context [67], but its study requires populations in partial reproductive isolation, in which ecological and non-ecological reproductive barriers can be distinguished (this would not be possible if no hybrid is produced). In phytophagous insects, these populations would correspond to host races [64].

Few insect species have been shown to encompass host-specialized populations undergoing appreciable hybridization (typically at a rate greater than 1%) while maintaining significant genetic differentiation [64,68], such as the well-known races of the apple maggot fly, Rhagoletis pomonella [69–71]. Within this scarcity of host race systems, there is still a non-negligible proportion of aphids, comprising attested or serious candidate models, most belonging to the most diverse subfamily, Aphidinae. The pea aphid, Acyrthosiphon pisum that essentially comprises monoecious populations (reviewed in reference [72]) encompasses at least eight conspecific host races specialized on different legume species [73], embracing the much-studied “clover” (Trifolium sp) and “alfalfa” (Medicago sativa) races in Europe [74,75] and North America [76,77]. In Hyalopterus pruni, two populations using either plum (Prunus domestica) or almond (P. dulcis) as primary hosts are highly genetically distinct in Europe and North America (a zone of introduction), but they hybridize on a shared host, apricot (P. armeniaca), at a significant rate (approximately 4%) [78]. The polyphagous aphid Myzus persicae comprises a specialized form on tobacco (Nicotiana tabacum) as a summer host [79] that is formally designated as the subspecies M. persicae nicotianae [7]. In Greece, sexual populations in tobacco-growing areas are genetically differentiated from those in other regions [80]. Distance alone cannot explain this differentiation, as the tobacco-feeding population is genetically uniform in most of the country. Reproductive isolation is likely ongoing, as suggested by behavioral studies of mating between tobacco lineages and those feeding on other crops, which occurs on the shared primary host, peach (Prunus persica) [80]. Given the moderate genetic differentiation between host-associated populations (populations analyzed at many microsatellites do not form genetic clusters as distinct as in the host race models presented above), biotypes of M. persicae probably have not reached species status, and may constitute host races in areas of sexuality and sympatry.

In several other aphid species, host races are suspected but not yet confirmed due to unknown hybridization rates, or even to a lack of any population genetic analysis. Here, the distinction of “biotypes”, often named as subspecies, denotes any degree of population divergence below or at the species level [64] and the presence of minute morphological variations (body color or the presence of wings in specific morphs), sometimes confirmed on the basis of performance and preference assays on different host species. Müller [11,81] reviewed complexes of biotypes belonging to monoecious species, such as Aulacorthum solani, Acyrthosiphon malvae and Macrosiphum euphorbiae. Each complex presents various subspecies of unclear status, with various degrees of sympatry, host specificity and ability to hybridize. The Aphis fabae complex encompasses at least three subspecies alternating between spindle (Euonymus europaeus) and various herbaceous plants, and a fourth monoecious subspecies that remains on the primary host (spindle) all year round (case reviewed in reference [82]). These subspecies show distinct summer host preference and some degree of behavioral isolation [83,84], but most of them can be hybridized in the laboratory [82]. A comparable system is found in Pemphigus bursarius (Pemphigideae), whose specialized populations on distinct Asteraceae are highly genetically differentiated despite reproducing on a shared primary host, poplar [85]. The case of Cryptomyzus galeopsidis is opposite, in that two proposed host races (within a larger complex) colonize different primary hosts, redcurrant (Ribes rubrum) and blackcurrant (Ribes nigrum), but share their summer host (case reviewed in reference [64]). The absence of mating preference in the laboratory and the identification of possible hybrid clones [86] suggest host race status, but hybridization in the field is limited by preference to different primary hosts [87].

In addition to typical aphid populations that alternate asexual and sexual generations on a regular basis (see Simon et al., this issue), we may mention a number of host-associated biotypes comprising asexual lineages, and sometimes single clones. These occur in Schizaphis graminum on cereals [88], Therioaphis trifolii on clover and alfalfa in Australia [89] and Aphis gossypii, which presents specialized clones on unrelated host families (cotton, potato and Cucurbitaceae, amongst others) [14]. In the following, we do not consider asexual biotypes, as host plants cannot promote their reproductive isolation.

Hence, intraspecific differentiation onto different host plants is not unusual in aphids, at least in the Aphidinae. Moreover, many host-associated biotypes that do not differ morphologically might have gone unnoticed.

3.2.1 Sources of divergent selection and post-zygotic isolation

3.2.2 Host plants as a source of premating isolation

The mechanisms by which host plants may induce reproductive isolation in aphids (Table 2) constitute various aspects of host specialization. However, widespread host specialization in aphids may actually have evolved as a consequence of divergence and does not constitute evidence for a direct role of host plants in speciation. In only few models have some specific predictions of ecological speciation been verified (Table 2, bottom). An experiment by Shaposhnikov on Dysaphis [133–135] (Table 2) might still constitute the most remarkable demonstration of host-induced speciation. In Dysaphis anthrisci (feeding on Anthriscus), Shaposhnikov successfully transferred one clone to Chaerophyllum, a normally unsuitable plant that hosts the related species D. chaerophyllina. In about 20 parthenogenetic generations, the D. anthrisci clone apparently adapted to its new host, gained some morphological resemblance and reproductive compatibility with D. chaerophyllina, while almost losing all ability to produce viable eggs with its parent species! This unique case of experimentally repeated, or reversed, ecological speciation (depending on the original direction of the host shift in nature) should however be taken with great caution, as the possibility of contamination of the reared D. anthrisci clone by some D. chaerophyllina line has not been excluded.

The geographical context of divergence may also indirectly suggest an active role of host adaptation, if prolonged allopatry is excluded. Sympatric speciation in animals is a hotly debated topic [125,136] that has long been proposed in phytophagous insects, especially after the work on Rhagoletis pomonella host races [70]. This mode of ecological speciation assumes that host-associated populations remain within each other's dispersal range and that gene flow, starting at 50% (panmixia), is maintained until speciation is over. In practice, allopatric divergence cannot be excluded in aphids, as known species complexes have wide continental ranges, and because geographical distributions are not thoroughly known. For instance, certain European countries do not seem to harbor some subspecies of Aphis fabae [11], suggesting some possible geographic separation. Biotypes of Hyalopterus pruni occuring in sympatry across their widely studied range are weakly geographically differentiated [15], but their origin of several million years ago leaves ample time for earlier allopatry. On the other hand, the Myzus persicae nicotianae subspecies may present a rare case of divergence onto a recently introduced crop, tobacco (cultivated in Eurasia since the sixteenth century), similar to the emergence of the Rhagoletis pomonella race after apple was introduced in North America. Given the extreme polyphagy and wide distribution of Myzus persicae s.s. [7], strict allopatric divergence of the subspecies nicotianae appears improbable. Similarly, the very recent divergence of pea aphid races and species may have been promoted by the increasing availability of some of their legume hosts through agriculture and climate warming in the Holocene [137]. It is not clear how such a context would favor repeated allopatric divergences on plants that are now broadly sympatric, given that the radiation occurred after the last glacial episode. Furthermore, the eleven pea aphid biotypes show variable rates of hybridization and genetic differentiation that may illustrate the gradual evolution of host races into good species [73].