Screening germplasm for aphid resistance led to the discovery of resistant accessions in several crop species against various aphid species (Table 1). However, sources of aphid resistance are limited and usually scarce. More often, aphid resistance was identified in unimproved landraces, in wild accessions or even in related species, thus requiring a long breeding process to introduce the resistance into cultivated varieties. In soybean, eleven aphid-resistant accessions were identified after screening more than 3500 soybean germplasm accessions [3,4]. In lettuce, the screening of 1200 accessions of Lactuca sativa and related species allowed the identification of two L. serriola and L. virosa aphid-resistant accessions; the genetic relationship with the previously known resistance conferred by Nr is yet unknown [5]. In wheat, over 40 000 accessions have been evaluated for seedling reaction to the Russian wheat aphid Diuraphis noxia and 300 have shown resistant or moderately resistant reactions. Very few of them, mostly identified in relatives of wheat, are currently being used in breeding programs and incorporated into the elite germplasm [6]. More recently, a screening effort was performed to find new resistance sources to the now predominant, virulent biotype 2 of the Russian wheat aphid; about 8% of accessions were resistant and belonged to distinct phylogenetic subgroups, suggesting that new genes or alleles may be identified [7,8].

The relative high number of resistant accessions discovered in certain species should not mask the fact that aphid resistance usually relies on a small number of genes with limited numbers of resistance alleles. In most cases, the genetic studies have still to be made to determine if the selected accessions are sources of novel resistance genes. In melon, about 50 accessions out the 500 tested were found resistant to the melon-cotton aphid Aphis gossypii. Among them, a large majority carries the same resistance Vat allele, whatever their geographical origins and only very few accessions carry a distinct allele [9]. Thus, aphid resistance in crop species is a valuable resource and our first rule should be to manage this natural biodiversity in a sustainable perspective.

Aphid resistance is manifested by different phenotypes depending on host plant and aphid genotypes. Entomologists have distinguished two mechanisms of resistance that affect insects: antixenosis (also known as non-preference), which affects the behaviour of insects and deters primary infestation of the crop, and antibiosis, which affects their biotic potential, e.g. growth, development, and reproduction [10]. Antixenosis was described in melon toward the melon-cotton aphid A. gossypii [11] and in barrel medic (Medicago truncatula) towards the bluegreen aphid, Acyrthosiphon kondoi [12]. Antibiosis affects aphids primarily by reducing fecundity or increasing mortality. For example, Vat-mediated resistance in melon reduces fecundity by 80 to 90% within three days [13], and the Mi-1-mediated resistance in tomato causes 100% mortality of the potato aphid Macrosiphum euphorbiae within 10 days [14]. However, in several cases, the reduction of the aphid biotic potential (antibiosis) results from a modification of the aphid feeding behaviour (antixenosis). Thus, in melon, barrel medic and tomato, the resistance is mainly due to a drastic reduction of the phloem sap ingestion by the aphids, A. gossypii, A. kondoi, and M. euphorbiae, respectively [15,16,12].

Aphid resistance in certain plant genotypes is associated with localized cell death at the aphid-feeding site, analogous to a hypersensitive response and associated with production of reactive oxygen species (ROS) like hydrogen peroxide (H₂O₂). Necrotic lesions were reported in some wheat resistant varieties after D. noxia feeding. Interestingly, the hypersensitive-like response was observed after infestation with one biotype but not another, while the resistance was effective towards both biotypes [17]. The RAP1 gene confers resistance to the pea aphid Acyrthosiphon pisum in barrel medic, independent of the hypersensitive reaction [18]. These data clearly indicate that the hypersensitive response is not required for aphid resistance. Furthermore, in several cases, no signs of localized cell death were reported, such as in A. kondoi resistant barrel medic plants (while susceptible plants develop necrotic patches and reddening on aphid feeding sites) [12] or in Mi-1 M. euphorbiae-resistant tomato plants [19]. In melon, although aphid resistant plants exhibit no visible necrotic symptoms after A. gossypii infestation, a microscopic hypersensitive response was reported [20].

Aphid resistance in some crops is enhanced by prior aphid infestation. This is the case in peach, where the resistance to the green peach aphid, Myzus persicae, conferred by the dominant gene Rm2 is enhanced by prior infestation [21,22], and in barrel medic against A. kondoi [12].

Inheritance of aphid resistance may be monogenic or polygenic across plant and aphid species. Single, dominant R genes control aphid resistance in cereals, forages, fruits and vegetables (Table 1). In wheat, twelve genes are involved in resistance to the Russian wheat aphid D. noxia. Eleven of the genes are dominant, each conferring resistance in a different resistance source; most of them are located on group 1 and 7 chromosomes of Triticeae, and may be allelic or closely linked to the same cluster [23,24]. In wheat, barley and rye, resistance to the greenbug, Shizaphis graminum, is conferred by single dominant genes [25]. Dominant genes provide resistance to different biotypes of the rosy leaf-curling aphid, Dysaphis devecta (Sd1, Sd2 and Sd3), and of the woolly apple aphid, Eriosoma lanigerum (Er1, Er2 and Er3), in apple and related species (Malus domestica and Malus spp.) [26]. A dominant gene (Dp-1) provides resistance to Dysaphis pyri in pear (Pyrus spp.) [27]. In soybean, three dominant genes, Rag1, Rag2 and Rag3, for resistance to the soybean aphid, Aphis glycines, were mapped to independent soybean linkage groups [28,29].

Several aphid resistances were reported with a recessive inheritance but researches on these resistances are scarce. Recessive gene dn3 confers resistance to the Russian wheat aphid, D. noxia, in Triticum tauschii [30] and gb1 from Triticum durum confers resistance to the greenbug, S. graminum (cited in [6]). Monogenic recessive resistances to the cowpea aphid, Aphis craccivora, and to the corn aphid, Rhopalosiphum maidis, were also reported in peanut and in maize, respectively [31,32]. Resistance to the soybean aphid, A. glycines, is conferred by two recessive genes with duplicate, dominant epistasis (ratio 15S:1R in F₂ populations) in soybean accessions PI 567541B and PI 567598B [33].

In several cases, aphid resistance is quantitative and polygenic. Genome locations of the genetic factors or QTL (Quantitative Trait loci) involved in aphid resistance have thus far been reported in very few instances. Aphid resistance was measured as intensity of infestation in field conditions or after controlled infestation in most cases. QTL analysis performed in sorghum for resistance to the greenbug, S. graminum, using different resistance sources and different aphid biotypes revealed three to nine genomic regions involved in the resistance that are likely distinct in the different resistance sources [34,35]. Two QTL in barley with large effects and a third QTL with minor effect confer resistance to the Russian wheat aphid, D. noxia; these three QTL are not located in syntenic regions with the monogenic dominant Dn genes of wheat [36]. Two QTL in cowpea control soybean aphid abundance; they have additive and epistatic effects [37]. A first study of quantitative abundance of aphids in field conditions in a segregating population of apple localized a putative QTL (detected only once) of resistance to the rosy apple aphid, Dysaphis plantaginea, and a QTL of resistance to the green apple aphid, Aphis pomi, which both explained from 8 to 20% of the variation, depending on sites and years [38]. Quantitative resistance was measured in two instances using parameters that affect behavioural and life history traits of aphids. In peach, two QTL for resistance to the green peach aphid, M. persicae, were identified, that together affect the infestation rate, the leaf curling of infested plants and, interestingly, the aphid feeding behaviour as determined by the electronic penetration graph technique [39]. In melon, four additive QTL and two couples of epistatic QTL affect the melon-cotton aphid A. gossypii. Among them, a major QTL affects both the behaviour and the biotic potential of A. gossypii; it co-localizes with and likely corresponds to the Vat cloned gene [40].

Aphid resistance genes are often located within clusters of resistance genes in the same chromosomal region, as is the case for many pathogen resistance genes. These ‘hot spots’ of resistance genes combine genes that confer resistance to aphids and other insects and pathogens. The barrel medic, M. truncatula, AKR, TTR and RAP1 genes, each of which confers the resistance to a distinct aphid species (the bluegreen aphid, A. kondoi, to the spotted alfalfa aphid, Therioaphis trifolii f. maculata, to the pea aphid A. pisum) are located within about 40 cM [41,18]. The apple genes, Er1 and Er3, for resistance to the woolly apple aphid, E. lanigerum, map to the same genomic region with a major gene for powdery resistance and with resistance gene analogues [42]. The Mi-1 gene was located on the short arm of tomato chromosome 6, which carries an impressive collection of resistance genes effective against fungi, oomycetes and nematodes [43]. These clusters of resistance genes targeting taxonomically distinct pests and pathogens suggest that genes with a similar nature confer resistance to different organisms; duplication, recombination and multiple rearrangements events during evolution may have contributed to the development of new resistance specificities.

More than 40 genes conferring resistance to diverse pathogens, such as bacteria, fungi, nematodes and viruses have been cloned during the last 20 years. In contrast, very little was known until recently about molecular mechanisms underlying aphid resistance. Two aphid resistance genes have been isolated and results to date suggest that plant aphid resistance is mediated by the specific recognition of aphid-effector proteins that triggers signalling cascades that rapidly activate plant defences against aphids in a similar scheme that was widely described for most plant-pathogen interactions.

The Mi-1 gene, which confers resistance to three species of the root knot nematodes Meloidogyne, was isolated in tomato. The same gene/allele was shown to confer resistance to a biotype of the potato aphid M. euphorbiae as well as other insects, whiteflies and the psyllids [44,45,46,47,48,49]. The melon Vat gene confers resistance to the melon-cotton aphid A. gossypii. It has also the unique feature of conferring resistance to non-persistent viruses when vectored by A. gossypii [50]. We isolated the Vat gene by a map-based cloning strategy [51,52]. Both isolated aphid resistance genes, Vat and Mi-1, are members of the nucleotide-binding-site and leucine-rich repeat region (NBS-LRR) family of resistance genes, to which belong the majority of the genes, isolated to date, conferring resistance to bacteria, viruses, fungi and nematodes [53]. Genes Vat and Mi-1 share structural similarities. Their encoded proteins belong to the coiled-coil (CC)-NBS-LRR, subfamily resistance proteins, which possess a coiled coil domain in N-terminal extremity of the NBS region. They differ, however, in several specific features. The gene Mi-1 possesses a long N-terminal leucin-rich extension of about 200 amino acids, which was shown to play a role in regulating signal transduction and cell death [54]. In contrast, the N-terminal region of the Vat gene is short. The C-terminal region of the Vat gene comprises four near-perfect repeats of 65 amino acids flanked by highly imperfect copies of a LRR motif, absent in Mi-1. Both genes are constitutively expressed at low levels and encode proteins predicted to be located in the cytoplasm.

The identification of signalling cascades activated by aphid resistance genes is still at its infancy, but the data available to date indicate they partially overlap with those activated by pathogens [55]. Moreover, there is evidence that different aphid resistance genes activate distinct signalling pathways. Mi-1-mediated resistance is dependent upon the salicylic acid (SA) signalling pathway, while the jasmonic acid-responsive gene pathway is predominantly or exclusively induced in bluegreen aphid-resistant M. truncatula [56,57].

Several other aphid resistance genes may encode NBS-LRR proteins. Resistance to the bluegreen aphid A. kondoi in M. truncatula were mapped in a cluster of CC-NBS-LRR sequences and may be encoded by one of them [12]. Resistance to the soybean aphid A. glycines in soybean was mapped in a 115 kb region including two candidate TIR-NBS-LRR genes [58]. The lettuce Ra gene, which confers resistance to the lettuce root aphid, Pemphigus bursarius L., was shown to belong to the large RGC2 NBS-LRR gene cluster in lettuce, which comprises several downy mildew (Bremia lactucae) resistance genes [59]. Several NBS-LRR analog sequences have been mapped in the vicinity of loci associated with resistance to the rosy leaf curling aphid in apple [26] and resistance to aphids in barley and wheat [60]. As knowledge of crop plant genomes develops, the cloning and sequencing of new aphid resistance genes from different crop plants will provide valuable information on the identity of resistance genes and the specific features of the molecular basis of the interaction between plants and aphids.

The molecular bases of recessive resistances to aphids have yet to be elucidated. Recessive resistances to viruses and fungi showed that resistance alleles correspond to mutations or deletions in a gene required for the infection cycle of the pathogen or in a resistance repressor [61,62,63,64]. Recessive resistance to aphids may result from the interruption of mechanisms used by the aphid for efficient feeding. The isolation of such recessive genes will be of great interest and is an exiting new avenue of research.

Several crops are hosts of different aphid species e.g., apple, wheat, raspberry, lettuce, barrel medic, whereas plant resistance to aphids is specific to an aphid species or even to a biotype within the species. A biotype is a clone able to survive, reproduce on, and/or cause injury to a cultivated plant that is resistant to other clones of the same species. Thus, most aphid resistances were shown biotype-specific, such as resistance to the greenbug, S. graminum, (Gb genes) to the European raspberry aphid, Amphorophora idaei, (Ag genes), to the soybean aphid, A. glycines, (Rag genes), and to the woolly apple aphid (Er genes) [6,42,65,66]. Some accessions are resistant to several biotypes, such as the barley ‘Post’ cultivar, resistant to 7 of the 11 known biotypes of greenbug [6]. The gene Dn7 from rye and the two genes present in the wheat introduction CItr2401 confer resistance to all the biotypes of the Russian wheat aphid known to date [67]. Some genes such as Rag2 in soybean are known to confer resistance to different biotypes [28]. In melon, an allele of the Vat gene confers resistance to only a biotype of A. gossypii, while another allele confers resistance to two biotypes which are genetically distant [9].

Evidence of biotypes in some aphid species suggests potential for breakdown of major genes of resistance to aphids. Breakdown of resistance conferred by the Nr gene in lettuce to the lettuce aphid, P. bursarius, occurred in Europe ca. 10 years after the initial release [68] and subsequent wide deployment of this resistance. The resistance gene, Ag1, to the larger raspberry aphid, Amphorophora agathonica, was extensively used in raspberry from the 1930s to the 1980s, when a resistance-breaking biotype appeared [69]. Nevertheless, since the 1990s, the Vat resistance allele has been used extensively, incorporated into over 30 cultivars and representing over 40% of the area of melon cultivation in France (ca. 15 000 ha in total). Even if a few adapted biotypes were reported [70], we did not observed any extension of resistance-breaking biotypes.

Occurrence of sexual reproduction in pathogen species significantly increases the risk of resistance-breaking biotypes or races [71]. The relevant data for aphids are far less numerous than for pathogens. The melon-cotton aphid, A. gossypii, is anholocyclic (only asexual reproduction occurs) in southern France where the resistance has been deployed without breaking down for 20 years. The Russian wheat aphid, D. noxia, is holocyclic (sexual and asexual reproduction occur) in its native range of Central Asia, but is anholocyclic in the Americas and South Africa; since 2003, new biotypes were found in the United States capable to overcome the resistance gene Dn4, which has been used since 1994 [72]. The greenbug, S. graminum, is considered holocyclic only beyond the 35th parallel; resistance-breaking biotypes have been discovered within 2 to 5 years of the release of every greenbug-resistant sorghum cultivar [73]. Virulent biotypes of greenbug were observed prior to deployment of some resistances [74], which suggests existence of genetic diversity in greenbug populations on non-cultivated grasses including future biotypes able to overcome newly deployed resistances.

Three strategies for deployment of resistance genes have been suggested. First, continual identification and introduction of new resistance genes in order to stay ahead of aphid populations that adapt to host resistance genes [6]. The number of genes and alleles of resistance are limited, as reported above, and it is feared that adapted biotypes spread into or develop de novo in areas, where biotype-specific resistance genes are deployed. This is a short-sighted strategy, especially for aphid species predicted to adapt quickly (holocyclic species). Second, combine (pyramid) as many resistance genes as possible in order to reduce the probability of new resistance-breaking biotypes. This pyramiding strategy requires the use of resistance genes, which are not already overcome individually. Three, identify and introduce resistance controlled by multiple, quantitative loci or by recessive loci demonstrated to be more durable. The combination of a major, recessive gene with a resistance QTL increased the durability of a major gene for virus resistance in pepper [75]. These different strategies have yet to be supported by experimental studies on aphids.

Thus it is critical that entomologists and plant breeders collaborate in development of aphid-resistant germplasm and cultivars. They need to better know the mode of reproduction and the genetic structure of the aphid populations that affects breeding crops. They have to investigate whether aphid biotypes, able to defeat resistance genes, pre-exist before the deployment of new resistant varieties. Such efforts will more likely challenge putative sources of aphid-resistance with different aphid populations, perhaps by testing in different environments (locations), and reveal occurrence or emergence of resistance-breaking biotypes. When controlled infestations are done, aphid colonies, collected in different production areas and genetically well characterized, have to be monitored for resistance bioassays.

The development of genomic resources and the first sequencing of an aphid genome, that of the pea aphid A. pisum, have begun to provide new insights into the structure and function of aphid genes [76,77]. In order to gain understanding of the basis of resistance durability, one of the most important developments likely to occur in the coming years is the elucidation of mechanisms involved in the break-down of plant resistance by aphids. The molecular basis of aphid avirulence is presently unknown, but will be revealed using the new developed genomic tools, and an aphid elicitor has yet to be identified. Aphid salivary secretions have a central role in aphid feeding activities and in the interaction between aphids and plants [78]. Thus it is highly likely that aphid saliva contains effector proteins that interact with plant proteins of resistant plants. Aphids secrete two kinds of saliva: gelling, which solidifies to form the stylet sheath, and non-gelling or watery. Aphids continuously inject watery saliva during their feeding activity. They penetrate and salivate into epidermal and mesophyll tissues during their probing activity, into parenchyma cells during their intercellular pathway before they reach the phloem, and they salivate into sieve tubes before and during sustained ingestion of phloem sap. Elicitors present in saliva may recognize plant cellular factors and trigger plant defence responses. Watery saliva contains enzymes, such as phenoloxidases, peroxidases, and pectinases, and other factors, which promote colonization of the host [79] and prevent the occlusion of sieve elements during aphid feeding [80]. Proteomic approaches initiated in several labs have identified some major proteins in the saliva of the pea aphid and green peach aphid [81,82]. The pattern of aphid salivary proteins strikingly differs among the aphid species and several minor and highly variable proteins still need to be identified [83].

Host plant resistance offers a very effective and promising way to control aphids in cultivated crops, and effective sources of resistance are available for a number of crops. Resistant cultivars must be developed and deployed with an awareness of the potential for emergence of resistance-breaking biotypes. In order to increase our understanding of aphid resistance, new resistance genes should be isolated, exploiting the genomic resources, which are available (or will be in a near future) for an increasing number of crop species. A particular effort should be dedicated to elucidate the molecular basis of recessive resistance genes, none of which have been isolated to date; this will likely provide new research avenues on aphid and plant interactions. Molecular characterization of aphid avirulence factors may enable evaluation and prediction of the potential durability of specific host plant resistance genes.