Lucerne or alfalfa Medicago sativa is the most important temperate forage legume worldwide. Only one or two varieties of lucerne were grown in the U.S.A. and Australia (the two leading exporters of lucerne) before the late 1950s and late 1970s, respectively. These dates coincided with the arrival of aphid species, which devastated lucerne stands and prompted the development of aphid resistant cultivars.
Lucerne-feeding aphids, including bluegreen aphids Acyrthosiphon kondoi, pea aphids Acyrthosiphon pisum, spotted alfalfa aphids Therioaphis trifolii maculata and cowpea aphids Aphis craccivora, however, still present significant risks for the lucerne industry worldwide and account for 25% of global production losses. Moreover, increased production costs, negative environmental effects and emerging aphid resistance to insecticide applications have led to a narrowing of management options against lucerne aphids.
Understanding lucerne aphid biology and trophic ecology will be needed to develop future management practices, including biological control. We review and synthesize research on the four lucerne aphid species, focussing on cultivar resistance and their interactions with other organisms, including predators, parasitoids, entomopathogens and bacterial symbionts. The effects of global climate change are considered, with a particular emphasis on the potential for compromised aphid resistance in lucerne cultivars under future climates.
We conclude by identifying future research questions and perspectives for the sustainable management of lucerne aphids. These include the characterization of plant secondary metabolites associated with natural enemy recruitment, an understanding of the role of endosymbionts in cultivar resistance and a better comprehension of multi-trophic interactions of lucerne aphids, both with other herbivores and higher trophic groups.
Lucerne or alfalfa Medicago sativa L., the principal economic species of the 87 species of Medicago and the world's most important and widely grown temperate forage crop, is a perennial legume adapted to a wide range of climatic conditions (Yang et al., 2008; Small, 2011). Originating in south-western Asia, lucerne was first cultivated in Iran and is now distributed worldwide as a result of its popularity as an agricultural species (Sullivan, 1992). Its high nutritional quality (e.g. high protein content) makes it desirable as a hay, silage and pasture crop for sheep and cattle, with new uses, including sprouts for salad and nutritional supplements for human diets (Bouton, 2012). Lucerne possesses many ecologically important characteristics, including the ability to fix atmospheric nitrogen via its association with rhizobial bacteria that form nodules on the roots. This has the potential to mitigate nitrogen-limitation induced by elevated atmospheric CO2 (Soussana & Lüscher, 2007) and reduces the requirement for nitrogenous fertilizers for lucerne and for subsequent crops (McDonald et al., 2003). The deep root system of lucerne gives it the ability to reduce the likelihood of soil salinity by absorbing water that would otherwise raise the watertable (Slavich et al., 2002; McDonald et al., 2003). Dryland salinity, resulting from the continual use for cropping of areas previously covered by trees, is an increasing problem worldwide with, for example, a large proportion of south-western Australia (4.4 million ha) at high risk from shallow watertables (National Land and Water Resources Audit, 2001). The high water-use of lucerne and its ability to help control this problem has therefore attracted much attention (Crawford & Macfarlane, 1995; Ridley et al., 1998; Humphries & Auricht, 2001). The lucerne industry generates US$7 billion annually and covers 32 million ha, one-third of which is in the U.S.A. (Michaud et al., 1988; Şakiroğlu & Brummer, 2007). In Australia, the lucerne seed industry alone in 2007 was valued at A$95 million, with an export value of A$28 million (Carter & Heywood, 2008). These benefits and the high economic value of the lucerne industry in Australasia (Bullen, 2002), North America (Small, 2011) and Europe (Veronesi et al., 2006) necessitate management strategies to maximize crop security.
Over 100 insect species, including weevils (Hypera spp. and Sitona spp.), grasshoppers (Melanophus spp.), blister beetles (Epicauta spp.) and caterpillars (e.g. Colias eurythene) are significant pests of lucerne (Frame et al., 1998; Manglitz & Sorensen, 1998). Amongst the most globally-significant pests of lucerne are aphids (Hemiptera: Aphididae), which, in some regions, feed continuously and exclusively on lucerne throughout the year (Klingler et al., 2007). As invasive species, aphids can devastate lucerne and have been traditionally controlled by insecticides, although increased production costs, negative environmental effects and selection for resistance associated with insecticide applications has led to a narrowing of management options for lucerne producers (Oerke & Dehne, 2004). A greater understanding of the biology of lucerne aphids is therefore required to develop alternative control strategies, such as biological control. The present review synthesizes existing knowledge about the four aphid pests of lucerne [spotted alfalfa aphid (SAA) Therioaphis trifolii maculata (Buckton), bluegreen aphid (BGA) Acyrthosiphon kondoi (Shinji), pea aphid (PA) Acyrthosiphon pisum (Harris) and cowpea aphid (CA) Aphis craccivora (Koch)] (Fig. 1), including their interactions with lucerne and other organisms, the effects of climate change and existing management options. In particular, we focus on cultivar resistance and identify future prospects for sustainable control of lucerne aphids to maximize lucerne crop security.
Aphids as lucerne pests
Aphids, with their piercing and sucking mouthparts, are specialized to feed from a single cell type, the phloem sieve element, and can feed from the same cell for hours on lucerne (Edwards et al., 2008). Similar to many other species, lucerne aphids can move their stylets between cells to evade plant defences and have recently been shown to manipulate the host plant by secreting saliva into the phloem sieve elements (Will et al., 2007; Gao et al., 2008). For example, Jiang and Miles (1993) suggested that SAA may induce redox changes in lucerne as a result of oxidases introduced in the aphid's saliva. Aphid populations, including lucerne aphids, expand rapidly under optimal conditions, aided by asexual reproduction (viviparous parthenogenesis) and the telescoping of generations (i.e. females give birth to live young in which embryos of the next generation are already developing) (Moran, 1992). When a host plant can no longer support the growing population, aphids produce winged (alate) morphs that disperse with the wind (Lawrence, 2009).
In temperate climates, lucerne aphids reproduce by viviparous parthenogenesis in spring and throughout summer. In late autumn, sexual females and males are produced, which mate and produce eggs that overwinter and hatch in spring. In warm temperate climates, including some regions of Australia, the sexual phase is not necessary for producing overwintering eggs and so aphids reproduce asexually throughout the year (Edwards et al., 2008). The harmful effects of aphids derive from the transmission of plant viral diseases [e.g. alfalfa mosaic virus (AMV)], the injection of toxins, secondary fungal growth (sooty mould) on honeydew (i.e. sap-feeding-induced sugary liquid secreted by aphids) and the direct ingestion of plant nutrients. They also have the ability to overcome host plant resistance and develop resistance against insecticides through the evolution of new aphid biotypes or host races (i.e. populations of a species that have been partially reproductively isolated from other conspecific populations as a direct consequence of adaptation to a specific host) (Garran & Gibbs, 1982; Diehl & Bush, 1984; Gao et al., 2007b; Lawrence, 2009).
Lucerne-feeding aphids are the most devastating insect pests of lucerne, causing an estimated production loss of 25% worldwide (He & Zhang, 2006). They present a significant problem for the lucerne industry by inhibiting seedling establishment, plant growth and persistence, and by reducing the quality and quantity of lucerne produced (Irwin et al., 2001). For example, during the late 1970s, the arrival of two exotic aphids, SAA and BGA, followed shortly after by PA in 1980 caused significant Australian lucerne declines (Berg & Boyd, 1984). Similarly, during the late 1950s, SAA invaded and devastated lucerne crops in western U.S.A., leaving only a small proportion of resistant varieties standing, which were subsequently selected for breeding (Hughes & Hughes, 1987).
Therioaphis trifolii maculata, originally with a Palaearctic distribution (Europe, the Mediterranean, North Africa, the Middle East, India and Pakistan), spread to south-western U.S.A. (approximately 1953), Australia (approximately 1977), South Africa (approximately 1980), Japan (approximately 1980) and New Zealand (approximately 1982) (Blackman & Eastop, 2000). It is a yellowish–green aphid, approximately 1.5 mm long, with four to six rows of dark spots running along the length of its back. Spotted alfalfa aphids are adapted to warm, dry conditions, with peak activity during late spring, summer and autumn. They are prominent pests of lucerne and are most often found on the underside of lower leaves, which turn yellow and die as a result of both the removal of phloem sap (i.e. general degeneration) and the toxic nature of the aphids' saliva (Berg & Boyd, 1984). They also cause veinal chlorosis in the growing tips and secrete large amounts of honeydew, often leading to the growth of black sooty mould (Madhusudhan & Miles, 1998). A combination of these effects greatly reduces the palatability and productivity of lucerne.
Acyrthosiphon kondoi, originating in Asia (known from Japan, Korea, India, Pakistan, Afghanistan and Iran) was accidentally introduced to California and New Zealand in 1974–1975 and subsequently spread to Australia, U.S.A., Chile and South Africa (Blackman & Eastop, 2000). BGA adults are approximately 2.5 mm long and, similar to PA, have two long cornicles protruding from their abdomens. They are blue–green in colour and feed primarily on the growing tips of lucerne stems, injecting salivary toxins as they feed (Knowles, 1998). High population densities can severely stunt the plant, causing leaves to yellow and die.
Acyrthosiphon pisum is a large (approximately 3 mm long as adults), green or pink aphid with long appendages, usually abundant from spring to early summer and autumn. PA is adapted to warmer temperatures than BGA but cooler temperatures than SAA (Rohitha & Penman, 1983). Originally a Palaearctic species, PA now occurs in most parts of the world and is a well known pest of lucerne in Europe, America and Oceania (Julier et al., 2004). Strong genetic diversity can occur between populations with different host ranges and preferences (van Emden, 2007). For example, populations colonizing peas, lucerne and red clover in France were found to be genetically divergent from one another (Simon et al., 2003). PA, apart from being slightly larger, resembles BGA in appearance. It is often difficult to differentiate them without magnifying their antennae, which possess several dark, narrow bands in PA but are uniformly brown in BGA. PA also infests the growing tips of lucerne stems but tends to be less specific about where they feed on the plant than BGA (Knowles, 1998). They cause general degeneration in lucerne but no local symptoms, and so are not considered ‘toxic’ like SAA and BGA. Infested plants wilt, the top leaves become paler, and lower leaves turn yellow and drop off.
Aphis craccivora is a small (approximately 2 mm long as adults), dark brown aphid with a wide range of host plants in different families (approximately 50 crops across 19 families). Similar to PA, CA has an almost worldwide distribution but probably originates from warmer areas of southern Europe (van Emden, 2007). It is a comparatively minor pest of lucerne, with most activity occurring during early spring and late autumn (Berg & Boyd, 1984). It is most commonly associated with grain legumes in Australia but has the potential to become more destructive to pasture legumes (e.g. lucerne) with climate change (Edwards, 2001).
Because of the costly nature and negative environmental impacts of insecticide applications, the use of cultivar resistance (i.e. bottom-up control) against aphids presents a more economical and practical option for increasing lucerne yield, quality and stand persistence (Irwin et al., 2001; Edwards & Singh, 2006; He & Zhang, 2006; Gao et al., 2007b). Aphid-resistant varieties of lucerne were selected by plant breeders soon after the late 1950s when SAA devastated lucerne stands in the U.S.A., which at that time, were dominated by the cultivars ‘Vernal’ and ‘CUF101’ (Hughes & Hughes, 1987; Bouton, 2012). Research was undertaken to investigate the nature of aphid resistance. Painter (1968) defined three mechanisms of plant resistance to insect attack: (i) antibiosis (Greek for ‘resistance to life functions’), which is the ability of a plant to reduce insect survival, extend insect development time, or reduce insect growth or fertility; (ii) tolerance, which is the ability of a plant to show a reduced damage response (e.g. yields maintained despite aphid attack); and (iii) nonpreference or antixenosis (Greek for ‘resistance to outsiders’), which refers to plant properties that reduce the likelihood of colonization (Kogan & Ortman, 1978; van Emden, 2007). Reports of a breakdown in plant resistance to new aphid biotypes were soon prevalent as a result of various factors, including time of year (Barnes, 1963), plant age (Howe & Pesho, 1960) and plant nutritional differences (Kindler & Staples, 1970), requiring further efforts aiming to evaluate and investigate the mechanisms for resistance.
Before the arrival of SAA and BGA in 1977, the Australian lucerne industry was reliant on one cultivar of lucerne, ‘Hunter River’, a locally-adapted type derived from the French variety ‘Provence’, which was highly susceptible to aphid attack (Lodge & Frecker, 1989). In response to the devastation of ‘Hunter River’, seeds of well-developed aphid-resistant cultivars were imported from the U.S.A. (Holtkamp & Clift, 1993). Australian breeding programmes have used ‘CUF101’ extensively as a progenitor of SAA and BGA resistance (Irwin et al., 2001). ‘Trifecta’, released in 1983, was the first cultivar to achieve at least moderate (>20% of plants) levels of aphid resistance (Oram, 1990), which led to significant yield increases over ‘Hunter River’ and ‘CUF101’ (Clements et al., 1984). The problem with cultivar introductions overseas, however, is that aphid responses can vary outside their country of origin (He & Zhang, 2006) and so further studies are required to assess local levels of lucerne resistance, environmental adaptation and mechanisms for resistance to aphids. After more than 30 years of breeding and variety introductions, over 50 lucerne varieties are now available for Australian growers (Lattimore, 2012).
To standardize the assessment of resistance to aphids, a rating scheme was developed (McDonald et al., 1988) using five classes ranging from susceptible to highly resistant. Multiple studies, both laboratory- and field-based, have focussed on the responses of different lucerne cultivars to aphid attack, and vice versa (Turner et al., 1981; Lloyd et al., 1983; Brownlee & Pitt, 1984; Holtkamp & Clift, 1993; Humphries & Auricht, 2001; Gao et al., 2007b; Goławska et al., 2008). Glasshouse studies, however, may not accurately reflect the field performance of resistant lucerne cultivars (Hamilton et al., 1978; Ridland & Berg, 1981). Lodge and Greenup (1983) investigated whether the field performance of a number of cultivars in New South Wales reflected the range of SAA resistance values observed by glasshouse screening tests. In general, these tests were poor indicators of resistance in the field. Field tests showed that total aphid numbers per seedling and dry matter yield were poorly correlated with observations in glasshouse studies, whereas total numbers of SAA and numbers of large nymphs per stem generally reflected resistance levels found in glasshouse studies. Resistance differences between lucerne growth stages have also been questioned. Bishop et al. (1982) found an increase in resistance to BGA with increasing plant size. Lloyd et al. (1983), however, observed no change in the order of resistance to SAA between growth stages of mature plants and seedlings, although mature plants were more tolerant of SAA than seedlings.
Various methods (both qualitative and quantitative) are used to measure resistance to aphids. In an antibiosis test, biological traits (e.g. aphid mortality, larval development, reproduction) are scored. Girousse and Bournoville (1994), for example, evaluated lucerne cultivar resistance to PA by estimating the net reproductive rate over 2 weeks. Similarly, Sandmeyer et al. (1971) used longevity, number of nymphs, development and reproductive period of PA and SAA to categorize resistance of six lucerne cultivars. Antibiosis studies, although reliable, are resource-intensive, especially when large numbers of cultivars are tested. Additionally, they do not take into account differences in plant tolerance of aphids (Julier et al., 2004). Tolerance has been addressed elsewhere (Berberet et al., 1991; Girousse et al., 1998; Bournoville et al., 2001), focussing on lucerne development (e.g. seedling biomass) under aphid infestation. He and Zhang (2006) evaluated the resistance of nine lucerne cultivars to three aphid species (SAA, CA and PA) in China, focussing primarily on ‘aphid damage index’ and ‘percentage of resistant plants’ as parameters to evaluate cultivar resistance.
Various studies have focussed on secondary chemistry of lucerne as mechanisms for resistance. For example, lucerne contains saponins, naturally occurring glycosides that are induced by herbivore attack, which, similar to constitutive (constant and active) defences, can vary with cultivar (Njidda & Ikhimoya, 2011; Goławska et al., 2012). High saponin concentrations in some lucerne cultivars was correlated with resistance against PA but not SAA (Pedersen et al., 1976) and reduced reproduction and survival was seen in PA when fed on a high-saponin line of lucerne (Goławska et al., 2006). Goławska et al. (2008) compared the feeding behaviour of PA on two European cultivars and suggested that the lucerne cultivar ‘Radius’, containing a higher concentration of saponins and a lower concentration of flavonoids than cultivar ‘Sapko’, had a higher resistance to PA (Fig. 2). In particular, reproductive period, fecundity and survival were higher on ‘Sapko’ (Fig. 2A–C) and the number of potential drops (cell punctures by aphid stylets) was higher on ‘Radius’ (Fig. 2D). Goławska et al. (2012) extended this research to include four cultivars, similarly showing that PA numbers were inversely related to saponin content. They also identified specific saponins (e.g. the low saponin line did not contain zanhic acid or medicagenic acid), which are important to consider because the biological activity of saponins and other plant secondary metabolites (PSMs) is often dependent on specific components (i.e. qualitative differences) as opposed to total concentrations (Bagheri et al., 2001; Francis et al., 2002). The presence of epicuticular lipids on lucerne leaves has also been suggested as a mechanism for resistance to SAA, which were found to feed on older leaves because they contain lower lipid concentrations (Bergman et al., 1991). An alternative mechanism suggests that higher levels of an aphid alarm pheromone [(E)-β-farnesene] observed in some lucerne cultivars increases BGA resistance by acting as a chemical cue for aphid nonpreference (i.e. repellent) (Mostafavi et al., 1996). This provides scope for the incorporation of increased levels of (E)-β-farnesene into new lucerne cultivars, a method already incorporated into wheat to repel cereal aphids (e.g. Sitobion avenae) (Rothamsted Research, 2013). Soluble antioxidants, such as ascorbate and glutathione, have been identified as effective enhancers of the plant's defensive system, with a reduction in reproductive rate seen in BGA and SAA that were fed on lucerne cultivars artificially infiltrated with ascorbate and glutathione (Miles & Oertli, 1993). Febvay et al. (1988) noted a few amino acids that may be important for lucerne resistance to PA with, for example, low concentrations of methionine and lysine found in the resistant cultivar ‘Lahontan’ compared with the susceptible cultivar ‘Resistador’ lacking no single amino acid. Girousse and Bournoville (1994), however, found no evidence that sugar or amino acid deficiencies in the phloem from two lucerne cultivars were responsible for resistance to PA, although variations in a few amino acids (alanine, leucine, isoleucine, arginine and ornithine) were observed, which may be linked to resistance. They identified a reduced flow of phloem in the resistant cultivar, suggesting ingestion difficulty as a potential mechanism for resistance (Klingler et al., 2005).
Improvement programmes have focussed on the development and maintenance of aphid resistance as newly-adapted aphid biotypes emerge (Humphries & Auricht, 2001; NAAIC, 2004). Recently, for example, a new BGA biotype was found in south-eastern Australia that causes substantial damage to previously resistant lucerne cultivars (Humphries et al., 2012). Similarly, in 1991, previously resistant lucerne varieties in California were devastated by a new BGA biotype (Zarrabi et al., 1995). Cultivar resistance may reduce aphid growth (antibiosis) or preference (antixenosis), or ameliorate the negative effect of aphid infestation on plant growth (tolerance), although the effectiveness of most plant resistance mechanisms are limited to a single aphid species, or even to single biotypes of a species (Clements et al., 1984; Gao et al., 2007b). For example, Salisbury et al. (1985) found a strong correlation between BGA and PA resistance in Australian-bred lucerne lines but no correlation in lines from the U.S.A. Trials incorporating multiple pest-resistance are ongoing, with, for example, new cultivars outyielding ‘CUF101’ by up to 58% in California (Putnam et al., 2011). Similarly, in eastern Australia, cultivars have out-yielded ‘Hunter River’ by over 300%, suggesting that trials incorporating multiple pest resistance have dramatically increased lucerne productivity (Irwin et al., 2001), although other growth factors incorporated in breeding programmes must be considered. Continual field testing of aphid-resistant varieties, involving phenotypic selection for resistance and control of AMV, is required to maintain aphid resistance. Regular collections of aphids and laboratory aphid culture replacements are also important to ensure that virulent strains are being used (Humphries & Auricht, 2001).
Alfalfa mosaic virus, a common seed-borne virus that causes significant lucerne crop losses, is the only lucerne-infecting virus known to be transmitted by aphids (Ohki et al., 1986; Miczynski & Hiruki, 1987; Hajimorad & Francki, 1988). All four lucerne-feeding aphids (BGA, PA, CA and SAA) have been shown to transmit AMV in a nonpersistent manner (i.e. short-term transmission) (Jaspars & Bos, 1980; Jones, 2004). Lucerne is a perennial crop so acquisition of AMV as a result of aphid feeding has significant consequences for crop health and economic value over a number of cropping seasons (as opposed to annual crops). AMV incidence levels and rates of transmission through lucerne seed vary depending on virus strain, environmental conditions, plant age, aphid preference and cultivar type (Bailiss & Offei, 1990). Feeding-preference tests by Garran and Gibbs (1982), for example, revealed a preference by SAA for AMV-infected lucerne cultivar ‘Siriver’ over healthy ‘Siriver’ but no preference was observed between AMV-infected and healthy ‘Hunter River’ lucerne. Additionally, younger lucerne crops in Australia were found to have a higher incidence of AMV than mature crops, although contrasting results were seen in Great Britain in the 1950s where mean AMV incidences of 5%, 7% and 17% were observed in crops that were 2, 3 and 4 years old, respectively (Gibbs, 1962). This contradicts the idea that seed-borne viruses are often present in younger crops (Tomlinson, 1962), although the difference may be related to the lack of AMV found in some of the commercial seeds of the cultivar ‘Du Puits’ in Great Britain in 1960, therefore resulting in less seed–seed transmission compared with Australia. AMV was reported in Australia in the 1960s, most likely introduced through importation of contaminated seed stock, although it was not widespread until the 1980s when lucerne-feeding aphids arrived. Similarly, the introductions of these aphids to New Zealand coincided with an increase in AMV in lucerne crops (Forster et al., 1985). Controlling aphids should help to reduce the incidence of AMV in lucerne.
Several trophic interactions have been reported for lucerne-feeding aphids, including interactions with bacterial endosymbionts, predators, pathogens and parasitoids (Hughes & Bryce, 1984; Leathwick & Winterbourn, 1984; Rakhshani et al., 2006; Simon et al., 2011). The first of these interactions, endosymbiosis, can represent a mutualistic interaction, whereas the latter three are antagonistic interactions, which are often considered in the context of biological control (Fig. 3).
Most aphids examined to date harbour the primary (or obligate) bacterial symbiont Buchnera aphidicola, which compensates for the deficiency of nutrients in their phloem diet by synthesizing essential amino acids such as arginine and tryptophan, as well as other essential nutrients (Prosser & Douglas, 1991; Liadouze et al., 1995; Shigenobu et al., 2000; Wilkinson et al., 2007). Buchnera aphidicola lives in specialized host cells called bacteriocytes (or mycetocytes) in the aphid haemocoel and is transmitted vertically (maternally). Some aphids also carry one or more secondary (or facultative) symbionts that are mostly vertically, and on an ecological timescale rarely horizontally, transmitted and exert diverse effects on hosts [interaction (i) Fig. 3]. These effects have mostly been studied in PA. The best-characterized secondary symbionts of PA include ‘Candidatus Regiella insecticola’ (pea aphid U-type symbiont), ‘Candidatus Hamiltonella defensa’ (pea aphid Bermisia-type symbiont or T-type symbiont), ‘Candidatus Serratia symbiotica’ (pea aphid secondary symbiont or R-type symbiont) and pea aphid Rickettsia symbiont (Rickettsia) (Moran et al., 2005). As the majority of lucerne-feeding PA clones carry H. defensa, compared with R. insecticola in clover feeders (Darby et al., 2003; Ferrari et al., 2004; McLean et al., 2010) it has been suggested that secondary symbionts can promote host plant specialization (or plant choice) by lucerne aphids [interaction (ii) Fig. 3], however, this is not as clear as originally interpreted. In a first set of experiments, Chen et al. (2000) evaluated the effects of S. symbiotica and Rickettsia on PA and BGA (Fig. 4). Rickettsia had no significant effect on fecundity of PA feeding on M. sativa but significantly reduced the fecundity of PA feeding on bur clover M. hispida. Serratia symbiotica reduced fecundity and longevity of BGA on both lucerne (Fig. 4B, C) and clover. Similarly, Leonardo and Muiru (2003) showed that pea aphids with R. insecticola had twice as many offspring as other aphids on clover, although populations were declining on lucerne, suggesting that aphid specialization on host plants is influenced by symbiont identity. Although many original studies identified an association between host plant and endosymbiont presence (e.g. Tsuchida et al., 2004), it is difficult to show that endosymbionts unequivocally influence plant utilization; the relationship may be incidental (e.g. different bacteria may simply have colonized after genetically different aphid lineages specialized on a particular host plant) or it might reflect differences in natural enemy pressures experienced by aphids on different plant species. Using antibiotic treatment to remove R. insecticola from PA, Leonardo (2004) found that R. insecticola was not responsible for causing host plant specialization. Similarly, McLean et al. (2010) found no evidence that secondary symbionts influence plant host specialization, although effects varied among aphid and host plant genotypes.
Furthermore, endosymbionts have been shown to influence a wide range of PA characteristics, including body colour (Tsuchida et al., 2010) and resistance to heat stress (Montllor et al., 2002; Russell & Moran, 2005; Dunbar et al., 2007). Experiments have also shown that secondary symbionts can infer a physiological defence mechanism in PA against parasitoids (Oliver et al., 2009, 2010) and entomopathogenic fungi (Scarborough et al., 2005) by restricting parasitoid egg development and lowering the rate of fungal transmission, respectively. Oliver et al. (2005) found one strain of H. defensa derived from CA that conferred resistance to the parasitic wasp Aphidius ervi in PA, suggesting that H. defensa may provide a protective role in other aphid species and that resistance may be transferred interspecifically. It is important to consider the aphid–symbiont association holistically as any changes in host plant nutritional composition and aphid activity (e.g. cultivar preference and reproduction) will inevitably alter the identity or frequency of symbionts present in aphid populations. Effects of symbionts on different aphid characteristics are also likely to influence other multi-trophic interactions with antagonists (Tsuchida et al., 2010). They thus play a fundamental role in aphid biology, beyond providing essential nutrients required for aphid survival, and offer potential targets for novel biological control strategies (Goggin, 2007).
Natural enemies of aphids include predators, pathogens and parasitoids, all of which have the potential to reduce aphid populations (Ekbom, 1994). Only a few studies have considered the importance of arthropod predators in controlling lucerne aphid numbers. Milne and Bishop (1987) identified brown lacewings (e.g. Micromus sp.), ladybirds (e.g. Coccinella repanda), predatory mites and syrphid larvae as important predator groups of BGA and SAA in lucerne. These predators significantly reduced aphid abundance in New South Wales, Australia, and were identified as potential pest management tools [interaction (iii) and (iv) Fig. 3]. By contrast, Ekbom (1994) found no evidence that predators have an important impact on numbers of PA in lucerne. Assessing the impact of predators on aphid populations requires careful analysis of temporal and spatial synchrony, prey choice and environmental condition (e.g. temperature) preferences (Cameron et al., 1980; Leathwick & Winterbourn, 1984; Ekbom, 1994).
Pathogens also provide an important means of biological control. In California, after the outbreak of SAA in 1954, two pathogenic fungi, Entomophthora virufenta and Entomophthora exitialis, were widely distributed, although the occurrence of natural epizootics confounded their effectiveness (Hall & Dietrick, 1955; Hall & Dunn, 1957). After the outbreak in Australia, Milner and Soper (1981) screened five species of fungi for activity against SAA. Entomophthora sphaerosperma, an important pathogen of SAA in Israel, was highly active and presented a promising microbial control agent, which was confirmed by releases in Australia. Another entomopathogenic fungus, Zoophthora radicans, known to cause extensive epizootics in SAA populations in Israel (Kenneth & Olmert, 1975), was imported and released into Australian lucerne fields during 1979 [interaction (v) Fig. 3]. These fungi can be mass-produced and easily applied, are able to persist between seasons and have great potential as microbial control agents, although a high relative humidity (approximately 90%) is required for effective transmission (Shah & Pell, 2003). Significant SAA declines were partly a result of Z. radicans, as well as the introduction of the parasitic wasp Trioxys complanatus (Butt et al., 2001).
Various parasitoids (Hymenoptera: Braconidae) attack lucerne aphids, therefore providing a mechanism for protection of lucerne crops [interaction (vi) and (vii) Fig. 3]. In Iran, for example, specific parasitoids of SAA include T. complanatus and Praon exsoletum. Common parasitoids of BGA and PA include Aphidius ervi, Aphidius smithi and Aphidius eadyi and CA is mainly attacked by Lysiphlebus fabarum (Rakhshani et al., 2006). In California during the period of severe infestation of SAA around 1955, three hymenopteran parasitoid species were successfully introduced and contributed significantly to the biological control of SAA (van den Bosch et al., 1964). One parasitoid in particular, T. complanatus, was generally dominant over the others. The three parasitoids were also released in Arizona in around 1956 and T. complanatus became an important control agent (Barnes, 1960). Delays in getting the parasitoid established, however, meant that other factors, including the use of insecticides against lucerne weevils and the incorporation of resistant lucerne varieties, may have influenced the aphid decline (Hughes et al., 1987). In 1958, the parasitic wasp, A. smithi, identified as an important factor in the natural control of PA in India, was imported and released in California. Aphidius smithi became established and exerted considerable control of PA (Hagen & Schlinger, 1960).
Many parasitoids were introduced into Australia in the 1980s after the outbreak of SAA, BGA and PA. Long periods of time are required to confirm the effectiveness of biological control mechanisms and perhaps the first comprehensively evaluated case of biological control of an aphid species was the control of SAA in Australia by T. complanatus. In New South Wales alone, this biological control has saved approximately A$2 million per year by avoiding the need to apply insecticides to lucerne fields and reducing the urgency for replacing susceptible lucerne with aphid-resistant cultivars (Hughes et al., 1987). Trioxys complanatus has also been shown to exhibit a preference for aphids feeding on lucerne rather than clover, despite the presence of aphids on both legumes (Milne, 1997). During 1978–1981, the Commonwealth Scientific and Industrial Research Organisation (CSIRO) imported the parasitoid A. ervi from California and Europe in an effort to control BGA in New South Wales. Surveys confirmed the successful dispersal and establishment of A. ervi despite some problems caused by drought in some areas (Milne, 1999). The impact of A. ervi on aphid abundance, however, was determined by the initial ratio of parasitoids to aphids (Wellings, 1986). The three Aphidius parasitoids of BGA and PA were also introduced into New Zealand around the same time. Aphidius smithi failed to establish but A. eadyi established quickly and spread rapidly throughout the North and South Islands as a result of its fast development and high fecundity. It took several attempts to establish A. ervi, which eventually replaced A. eadyi as the predominant parasitoid in the North Island (Cameron & Walker, 1989).
Successful biological control of aphids by parasitoids depends on the ability of the parasitoid to reduce aphid reproductive potential and the age structure of aphid populations can determine this success. Parasitoids that attack early-instar aphids are likely to be more effective at reducing aphid populations than those that attack late-instar aphids (Sequeira & Mackauer, 1988). He et al. (2003, 2005) found that parasitism of early-instar PA by A. eadyi prevented any reproduction and, when late instars and adults were parasitized, progeny were still produced but reproductive potential was lower than unparasitized aphids (Fig. 5). Specifically, parasitized fourth-instar and adult aphids had a shorter reproductive period (Fig. 5A) and lower fecundity (Fig. 5B) than unparasitized adults. Growing aphid populations often contain large numbers of early-instar individuals, which are present early in the season (Schowalter, 2006). Parasitoid attack early in the season should therefore significantly reduce aphid populations and prevent build-up later in the season. This would particularly apply in cooler conditions when aphids develop slowly, making populations vulnerable to attack by parasitoids. If similar effects on aphid populations are seen between species, this model may help to predict population dynamics regardless of species (Sequeira & Mackauer, 1988). The impact of this biological control mechanism, however, may also be dependent on the population growth of the parasitoid itself, which can be influenced by, for example, the presence of alternative aphid species (that are not feeding on lucerne) as hosts, and so variation in parasitoid population dynamics would make long term predictions difficult (He et al., 2005). Long-term studies are therefore required to assess changes in trophic synchrony over time.
Interactions between natural enemies
Competition between natural enemies (e.g. predators killing other predators or parasitoids) may also make predictions difficult. Snyder and Ives (2003) investigated how different natural enemies, including parasitoids and predators, and their interactions, can affect the successful control of PA on lucerne. Parasitoids caused a delayed reduction in PA populations relating to the generation time of the parasitoid, whereas generalist predators, including coccinellid and carabid beetles, web-building spiders and Nabis and Orius bugs, caused an immediate reduction in population growth. In an earlier study, carabid beetles had little influence on the population of PA as a result of avoidance tactics by the aphids but they fed on parasitized aphid mummies, therefore disrupting the biological control of aphids by parasitoids [interaction (viii) Fig. 3] (Snyder & Ives, 2001). A further level of complexity is added by the potential for endosymbionts to increase the resistance of aphids to parasitoids. The parasitoid A. ervi, for example, has been shown to increase the deposition of eggs in PA infected with the symbiont H. defensa to overcome symbiont-based defence (Oliver et al., 2012). Additionally, changing abiotic environmental conditions can alter interactions with, for example, A. ervi proving less effective at reducing aphid populations in drier, inland areas and most effective in high rainfall areas of eastern Australia (Milne, 1999). Similarly, the difficulty in establishing A. eadyi and A. ervi in New Zealand indicates that they were suited to different conditions (Cameron & Walker, 1989) and the parasitoid T. complanatus was less effective at high Australian temperatures (Hughes et al., 1987).
Future climate change
Few studies to date have considered the effect of climate change on aphids feeding on lucerne. Amino acid concentrations in the phloem, which are important for aphid performance, can be altered by a variety of abiotic environmental factors, including temperature, atmospheric carbon dioxide and water availability (Docherty et al., 2003; Douglas, 2003; Aslam et al., 2012). In general, elevated temperatures (eT), anticipated to occur with climate change, are expected to increase nutrient uptake efficiency resulting in larger plants that contain lower concentrations of some PSMs (Rufty et al., 1981; Bale et al., 2002; Richardson et al., 2002). Drought is predicted to reduce nutrient uptake efficiency and eT may increase the impacts of drought by creating a higher demand for water (Gregory, 2006; Staley & Johnson, 2008), whereas plants under elevated carbon dioxide concentrations (eCO2) may perform better under drought because stomata close to minimize water loss and water use efficiency increases. Additionally, the nutrient composition of the phloem is likely to be particularly affected as a result of increasing demands for phloem solutes under drought (Pritchard et al., 2007). In many plants, eCO2 is expected to increase plant productivity by increasing photosynthetic rates. This, however, generally leads to a reduction in tissue quality and, specifically, an increase in the C : N ratio. This can be exacerbated when increased plant growth leads to nitrogen-limitation (Bezemer & Jones, 1998; Veteli et al., 2002; Ainsworth & Long, 2004; Johnson & McNicol, 2010). Legumes such as lucerne, however, may partially overcome eCO2-induced nitrogen-limitation by increasing biological nitrogen fixation, which may alter the dynamics of aphids feeding on the plant (Johnson & McNicol, 2010). Despite a wealth of studies on the effects of eCO2 on insect herbivores, however, the response of aphids and, in turn, the response of higher trophic levels, has been difficult to generalize (Pritchard et al., 2007; Sun & Ge, 2011).
Aphids have the potential to respond rapidly to climate change as a result of their short generation times and high reproductive capacity (Harrington et al., 1995), although changes in climate will affect aphid species that possess idiosyncratic characteristics differently. PA, for example, is prevalent in the summer, whereas BGA is adapted to cooler temperatures and predominates in early spring and autumn when there are fewer natural enemies (Rohitha & Penman, 1983). Future temperature changes may increase the temporal overlap of these aphids and warmer winters may lead to the disappearance of sexual overwintering stages, causing greater devastation over extended periods, although many other factors will play a role. Chen et al. (2000) observed no effect of the endosymbionts S. symbiotica and Rickettsia on the fitness of PA at 20 °C but both increased the fitness of one parent clone reared for three generations at 25 °C on lucerne. These fitness effects, however, were not observed for two other clones at 25 °C. Montllor et al. (2002) also identified a beneficial effect of S. symbiotica and Rickettsia on PA under heat stress, although other studies have suggested that obligate symbionts (e.g. B. aphidicola) limit the ability of aphids to cope with rising temperatures (Harmon et al., 2009; Wernegreen, 2012).
One study (S. N. Johnson, J. M. W. Ryalls and A. J. Karley, unpublished data) has considered the effect of eCO2 on pea aphids feeding on five different lucerne cultivars of varying resistance. Compared with ambient CO2, they saw an increase in colonization success (from 22% to 78% of plants) and reproduction rates (from 1.1 to 4.3 nymphs per week) under eCO2 on the resistant cultivar ‘Sequel’. The moderately resistant cultivar ‘Genesis’, however, became more resistant at eCO2, as shown by a decrease in colonization success (from 78% to 44% of plants) and reproduction rates (from 4.9 to 1.7 nymphs per week). Additionally, under eCO2, concentrations of the essential amino acid lysine increased in ‘Sequel’ but decreased in ‘Genesis’. This demonstrates the importance of considering different host varieties when studying the effects of climate change. Additionally, future CO2 concentrations are likely to be accompanied by elevated temperatures and, although a number of studies have observed the effect of eCO2 on aphid populations, the combination of eT and eCO2 has received little focus (Newman, 2003; Himanen et al., 2008; Auad et al., 2012). To predict the effects of eCO2 and eT on lucerne resistance to aphids, it is necessary to understand how these abiotic factors influence lucerne nutrition and defence, including what nutritional and defensive properties affect resistance, of which little is known at present.
Conclusions and future perspectives
Although the impact of aphids on lucerne productivity may be influenced by a wide range of factors under eCO2 and eT, the effectiveness of natural enemies in controlling lucerne aphids will be a major influence (Hughes et al., 1987; He et al., 2005; Thomson et al., 2010). The interaction of abiotic and biotic variables will determine future outcomes. Any efforts to conserve or facilitate natural enemies for biological control may be negatively influenced by interactions between abiotic and biotic variables and inhibited by trophic asynchrony (Dyer et al., 2013). Although PA is best-studied at present, it is important to consider the other lucerne-feeding aphids to determine interspecific differences and provide effective management strategies. Few studies of CA, for example, have been reported, although this aphid may become more problematic in future as conditions change. Many effective resistant lucerne cultivars are already established (Lattimore, 2012) but future resistance studies should include CA to allow for effective control at the risk of an outbreak.
Longer-term studies are required to determine bacterial, predator, pathogen and parasitoid roles, as well as the effects of climate change, plant nutrition and PSM quality and quantity. Most studies focus on total concentrations of PSMs, with few considering specific components that may explain variations in plant–herbivore–symbiont–natural enemy (multi-trophic) dynamics. Factors to consider may include the effect of light availability and plant age on PSMs, such as saponins (Francis et al., 2002). PSMs, such as indole acetaldehyde in lucerne (Hatano et al., 2008), and aphid pheromones, such as (E)-β-farnesene (Verheggen et al., 2012), can act as cues for natural enemies of aphids, which may differ according to cultivar type and in response to climate change. Further studies may also focus on jasmonic acid signalling, a phytohormonal response that has been implicated in R gene resistance (i.e. genes in the plant genome that confer resistance) to BGA in another Medicago species, Medicago truncatula (Gao et al., 2007a). Incorporating resistance genes into commercial lucerne lines may provide further scope for effectively managing lucerne crops and preventing aphid infestation (Goggin, 2007; Bouton, 2012).
Secondary endosymbionts of aphids influence a wide range of lucerne aphid characteristics. Studies should also be undertaken to determine whether symbionts play a role in cultivar specialization within the same host plant species, which may alter resistance test outcomes. It is, however, important to consider factors that may influence associations between aphid endosymbionts and host plant use, including the risk of attack by natural enemies and symbiont colonization history (Frago et al., 2012). Biological control of lucerne aphids by parasitoids, pathogens or predators presents an effective management strategy, although intraguild or predator–parasitoid predation must be considered at the same time. For example, Snyder and Ives (2003) found biological control to be most effective when both types of natural enemy were present but they suggest that long-term experiments would lead to non-additive effects on aphid control because the predators would displace parasitoids. It is also important to consider the effects of climate change on host–enemy synchronization (Hance et al., 2007) and the effectiveness and establishment of biological control agents, which are often suited to different environmental conditions (Milne, 1999).
Aphids also have to contend with indirect influences from spatially-separated root herbivores present below-ground. Indeed, in a recent meta-analysis, root herbivores had the biggest effect on aphids compared with other above-ground herbivores (Johnson et al., 2012). Very few studies concerned with crop security have considered above-ground–below-ground interactions (van der Putten et al., 2009) and no studies to date have considered the effect of below-ground organisms on above-ground aphids feeding on lucerne. Lucerne aphids might be particularly susceptible to effects of below-ground herbivores. In particular, the lucerne weevil Sitona discoideus and white fringed weevil Naupactus leucoloma have the capacity to reduce nitrogen fixation in lucerne by damaging root nodules that house nitrogen fixing bacteria (Keane & Barlow, 2002; Barnes & De Barro, 2009). Given the effects of such root herbivory on nitrogen fixation, it is likely that the amino acid composition of the foliage and phloem could also vary, which in turn could affect both aphids and their primary and secondary endosymbionts (Douglas, 1993, 1998). Replicating trophic complexity in an experimental setting is challenging but necessary to provide a more holistic insight into the mechanisms underpinning multi-trophic interactions in natural terrestrial systems (Bardgett & Wardle, 2010). In general, sap-feeding above-ground herbivores, such as aphids, are assumed to benefit most from plant attack by below-ground herbivores. As phloem-feeders they may circumvent any induced defence compounds, which usually occur in low concentrations in phloem sap, at the same time as reaping the benefits of stress-induced increases in above-ground nutrition as a result of root damage below-ground (Bezemer et al., 2005; Johnson et al., 2008), although many factors play a role. The opposite might be true in lucerne because the root nodules (and therefore sites of nitrogen acquisition) are often the principal site of attack by root herbivores.
There is not one lucerne variety that will persist everywhere, which is why it is important for new cultivars to be developed and targeted for different geographical ranges (Bouton, 2012). In southern Queensland, for example, aphid resistant varieties are desirable because insecticidal control is less economical on rain-grown pastures (Franzmann et al., 1979). Similarly, new cultivars are required to counteract new aphid biotypes that are able to break resistance. The capacity of lucerne aphids to develop resistance-breaking biotypes has already been demonstrated by BGA in the U.S.A. (Zarrabi et al., 1995) and Australia (Humphries et al., 2012). Lucerne breeders, however, need to focus on what producers are willing to pay. Communication failures between researchers, breeders and producers are often the reason why only a small percentage of resistance mechanisms have been incorporated into commercial lines (Edwards & Singh, 2006). Focus should be placed on incorporating multi-trophic interactions into long-term field studies, as opposed to short-term glasshouse studies, using mixed or various individual cultivars. Gutierrez and Ponti (2013) identified host plant resistance as the most important mortality factor of SAA in California. They also predicted that, alone, each mortality factor failed to control SAA, whereas successful control was achieved by a combination of factors. An integrated pest management approach combining top-down (biological control) and bottom-up control (cultivar resistance) should provide a more accurate representation of potential management practices aimed at reducing lucerne aphid populations and avoiding outbreaks.
This work was undertaken as part of a research project funded by the Hawkesbury Institute for the Environment. We thank Cavan Convery of the James Hutton Institute (U.K.) for providing the trophic schematic diagram. We also thank the Department of Agriculture, Fisheries and Forestry for providing high-resolution aphid images.