Introduction

To the extent that certain herbivores can alter the physiology of their host plant, and hence its nutritional value, there exists the potential for complex interspecific interactions among herbivores that are mediated by the host plant. As aphids feed from plant phloem elements, they inject saliva that often brings about changes in plant physiology (Prado & Tjallingii, 1994), usually for their own nutritional benefit (Petersen & Sandström, 2001). When more than one aphid species feed on the same plant the net changes in plant physiology will be some function of their combined effects. For aphid species that form closely spaced colonies, individuals may benefit from faster development when feeding in groups (Qureshi & Michaud, 2005). However, alterations in plant physiology induced by aphid feeding could potentially have positive, neutral, or negative effects on heterospecific aphids attempting to colonize the same plant. Amensalism (one species is harmed or inhibited and the other is unaffected), commensalism (one species derives some benefit while the other is unaffected), mutualism (an association in which both species benefit), and antagonism (both species are negatively affected via an indirect mechanism) are all plausible outcomes of interspecific interactions among aphid species that co-infest a host plant. Thus, host plant-mediated interactions among aphid species, rather than any direct form of competition, may be key factors influencing patterns of aphid species co-occurrence or niche partitioning in nature.

Three economically important aphids occur sympatrically in fields of wheat, Triticum aestivum L., throughout many regions of the USA where cereals are grown. These are the greenbug, Schizaphis graminum, the Russian wheat aphid, Diuraphis noxia, and the bird cherry-oat aphid, Rhopalosiphum padi (Homoptera: Aphididae) (Turney & Hoelscher, 1986; Schotzko & Bosque-Perez, 2000; Bosque-Pérez et al., 2002). It has been shown that S. graminum and D. noxia are able to alter the amino acid profile of phloem contents in susceptible wheat (cultivar Arapahoe) for their own benefit (Telang et al., 1999; Sandström et al., 2000; Sandström & Moran, 2001), although they likely do so in very different ways. Notably, Telang et al. (1999) did not find any such effect for D. noxia on resistant wheat, cultivar Halt. Havlickova (1986) reported that feeding by R. padi on young wheat plants cultivars Mironovskaya 808 and Slavia resulted in increased concentrations of free amino acids, sucrose, glucose, and some phenolic compounds in above-ground plant parts, but caused a reduction in the concentration of free amino acids and other compounds in the roots. In contrast, Gianoli & Niemeyer (1997, 1998) have shown that R. padi infestation triggers the induction of defensive chemicals in the first leaf of wheat seedlings and that this response can vary with the tissue infested. Thus, some aspects of antibiosis-based plant resistance to aphids may involve mechanisms that interfere with the processes whereby aphids attempt to alter host plant physiology for their own benefit. Here we refer to changes in host plant chemistry brought about by aphid feeding as ‘plant induction’, whether the consequences are positive or negative for the aphids. If plant inductive processes are species-specific among aphids, we also might expect plant antibiosis mechanisms to be quite specific to aphid species. Notably, wheat varieties resistant to greenbug (e.g. cultivar TAM 110) are not resistant to Russian wheat aphid, nor vice versa (e.g. cultivars Halt, Stanton).

A variety of interactions have been reported for aphid species feeding on plants previously infested by conspecifics or different aphid species. For example, Messina et al. (2002) found that previous infestation of wheat by R. padi reduced the subsequent growth rate of a conspecific colony by 50%, but had no effect on a population of D. noxia. Similarly, biotype E of S. graminum had significantly improved fecundity on wheat cultivar Newton previously conditioned by D. noxia, but not on wheat conditioned by conspecifics (Formusoh et al., 1992), a finding indicative of commensalism. Qureshi & Michaud (2005) observed that developmental time was decreased for nymphs of S. graminum, and increased for nymphs of R. padi, relative to that of their respective mothers that completed development on the same plants of wheat cultivar TAM 107, whereas there was no such effect for D. noxia. Gianoli (2000) found that the reproductive rate of English grain aphid, Sitobion avenae, on tillering wheat plants was negatively affected by a previous infestation of R. padi, a result suggesting amensalism.

Interactions among simultaneously occurring cereal aphids on the same plants have rarely been studied, but studies of other phloem-feeding Homoptera have revealed interactions ranging from synergism to antagonism. Alla et al. (2001) examined interactions between R. padi and the wheat leafhopper, Psammotettix alienus. They found that infestation by R. padi resulted in delayed development and increased mortality of P. alienus on the same wheat plants. Furthermore, in the presence of R. padi, P. alienus left their preferred feeding sites on the lower part of the plant and moved to upper plant parts. In contrast, Kidd et al. (1985) found a beneficial interspecific association between the grey pine aphid, Schizolachnus pineti and the spotted pine aphid, Eulachnus agilis on Scots pine, Pinus sylvestris. By feeding on the same shoots and needles as S. pineti, E. agilis was found to benefit from commensalism in terms of increased survival and faster growth, presumably by exploiting the plant induction brought about by S. pineti.

Schizaphis graminum, D. noxia, and R. padi have all been observed to simultaneously infest wheat plants (Bosque-Pérez et al. 2002, JAQ unpublished). In the course of rearing these species in the laboratory, we often find various combinations of these species developing together on the same plants, to the point where prevention of cross-contamination among colonies is a constant challenge. However, contamination of D. noxia colonies by S. graminum and R. padi seem to occur most frequently, as though colonies of D. noxia were preferentially invaded by the latter species. We have previously shown that these three species prefer different feeding sites on the plant and vary in their tendency to move among plant parts in the course of development (Qureshi & Michaud, 2005). In the present study, we examined interactions among S. graminum, D. noxia, and R. padi by comparing their acceptance of TAM 107 wheat, frequency of intra-plant movement, developmental time, and reproductive performance as they developed and reproduced on the same plants in heterospecific and conspecific pairs.

Materials and Methods

Results

Mean daily temperature was 21.23 ± 0.8° C during the first experiment. Survival of first instars to first reproduction averaged 90–100% across treatments and was not significantly different for any species when conspecific pairs were compared to heterospecific pairs (P > 0.05). The GLIMMIX MACRO model used to test survival indicated a good fit for the analyzed data sets because deviance values were close to χ² critical values and extra dispersion scale values were above 0.9 and close to 1.0.

In the first experiment, when seedling wheat plants were co-infested with first instar nymphs of all species combinations, D. noxia nymphs took longer to mature in the presence of S. graminum than in the presence of conspecifics (F = 4.41; df = 2, 55; P = 0.043), whereas their development rate in the presence of R. padi was intermediate and not significantly different from the other treatments (Fig. 1). The developmental times of S. graminum and R. padi nymphs did not differ from those of conspecific pairs when they developed in any heterospecific combination (F = 2.84; df = 2, 56; P = 0.067) (F = 1.01; df = 2, 54; P = 0.372). Diuraphis noxia maturing in conspecific pairs had higher fecundity than those maturing in pairs with either S. graminum or R. padi (F = 5.78; df = 2, 55; P = 0.005), whereas the fecundity of S. graminum and R. padi maturing in heterospecific pairs did not differ from those maturing in conspecific pairs (F = 0.02; df = 2, 56; P = 0.977) (F = 1.31; df = 2, 54; P = 0.278), (Fig 2).

We also used data from the first experiment to compare the developmental time and fecundity of conspecific aphid pairs that developed on the same plant tissue for at least five consecutive days with those that developed feeding on different plant tissues. Nymphs of D. noxia feeding consistently on the same plant tissue developed faster than those that developed on different tissues, whereas S. graminum pairs maturing on the same tissue had higher fecundity than those that fed and developed on different tissues (Table 1). All other comparisons of developmental time and fecundity were not significantly different between aphids that fed together versus separately.

The survival of progeny produced by S. graminum foundresses maturing in conspecific pairs was not different from that of foundresses that matured in the presence of D. noxia, but was lower for progeny whose mothers had matured with R. padi (F = 5.03; df = 2, 56; P = 0.009), (Fig. 3). Progeny survival for D. noxia and R. padi maturing in conspecific pairs was not different from that observed in any heterospecific combination (F = 0.25; df = 2, 55; P = 0.780 and F = 0.25; df = 2, 54; P = 0.782, respectively).

There was no difference among treatments in the frequency of position change on the plant for developing S. graminum nymphs (F = 1.59; df = 2, 57; P = 0.213), (Fig. 4). Diuraphis noxia changed position on the plant more often when developing in the presence of S. graminum than in conspecific pairs, but demonstrated an intermediate value in the presence of R. padi (F = 3.27; df = 2, 56; P = 0.045), (Fig. 4). Similarly, R. padi changed position more often in the presence of S. graminum than in the presence of D. noxia or a conspecific aphid (F = 3.43; df = 2, 58; P = 0.039), (Fig. 4).

Developing S. graminum nymphs spent 80–85% of their time on the primary leaf whether they were developing with a conspecific (F = 52.68; df = 5, 114; P < 0.0001) with D. noxia (F = 72.95; df = 5, 114; P < 0.0001) or with R. padi (F = 39.20; df = 5, 114; P < 0.0001). The order of plant tissue preference for S. graminum was primary leaf > stem > secondary leaf = tertiary leaf = flag leaf. Nymphs of D. noxia were present on the tertiary leaf for 57–70% of observations, significantly more than on any other plant tissue whether developing as conspecific pairs (F = 15.92; df = 5, 114; P < 0.0001) with S. graminum (F = 19.00; df = 5, 114; P < 0.0001) or with R. padi (F = 26.42; df = 5, 108; P < 0.0001). The order of tissue preference for D. noxia in conspecific pairs was tertiary leaf > primary leaf = secondary leaf > stem = flag leaf, whereas with S. graminum it was tertiary leaf > primary leaf > secondary leaf = stem = flag leaf and with R. padi it was tertiary leaf > primary leaf = secondary leaf with primary leaf > stem = flag leaf and secondary leaf = stem = flag leaf. Rhopalosiphum padi nymphs spent 66–88% of their time on the stem, significantly more than on any other plant tissue whether developing in conspecific pairs (F = 276.80; df = 5, 114; P < 0.0001) with S. graminum (F = 24.68; df = 5, 114; P < 0.0001) or with D. noxia (F = 59.41; df = 5, 108; P < 0.0001). The order of tissue preference for R. padi was stem > primary leaf > secondary leaf = tertiary leaf = flag leaf. The two tissues on which a species spent more time in all three species-pair combinations (primary leaf and stem for S. graminum, tertiary and primary leaf for D. noxia, and stem and primary leaf for R. padi) did not differ among species combinations in any case (P > 0.05).

The mean daily temperature was 22.32 ± 0.34° C during the course of the second experiment. Neither the fecundity of aphids transferred to experimental plants as fourth instars, nor the survival of their progeny, were significantly different among treatments for any aphid species, when aphids in conspecific and heterospecific pairs were compared (Table 2).

Discussion

Changes in plant physiology due to feeding by these aphid species have already been described in wheat (Havlickova, 1986; Prado & Tjallingii, 1994; Telang et al., 1999; Sandström et al., 2000; Sandström & Moran, 2001). All three aphid species appeared to initiate distinct, species-specific, plant inductive processes that were systemic to varying degrees. Consequently, interactions among these aphid species appeared to be mediated by the host plant, although direct interactions among the aphids themselves cannot be ruled out. It is unlikely that the observed effects could have resulted from exploitation competition as the pairs of aphids employed in these experiments would not be able to extract more than a small fraction of available nutrients from the phloem, nor effect any appreciable resource depletion in the plant. Although certain aphid species exhibit direct aggression towards other insects, usually predators, this is known only in social aphids with sterile soldier castes (Rhoden & Foster, 2002) and has not been reported for these cereal aphids to our knowledge. In these experiments, using only two aphids per plant, we were able to document amensalism by both S. graminum and R. padi on D. noxia, and by R. padi on S. graminum, with all other interactions essentially neutral (Price 1997).

Of the three species, S. graminum can be considered the most aggressive in exploiting its host plant; it has the highest reproductive rate and causes rapid deterioration of plant quality via chlorosis (Gellner et al., 1990; Qureshi & Michaud, 2005). The higher fecundity of S. graminum that developed in conspecific pairs on the same plant tissue for at least five consecutive days compared to those that fed on different tissues (Table 1), suggests a benefit of conspecific group-feeding on reproductive rate that is localized within infested tissues. Schizaphis graminum is also known to significantly reduce translocation from the immediate vicinity of its feeding site without altering the integrity of phloem elements (Burd, 2002). Although both D. noxia and S. graminum feeding causes leaf chlorosis in wheat, the symptoms are slower to develop with D. noxia than with S. graminum and plants with uncontrolled colonies of D. noxia can survive considerably longer. In contrast, R. padi feeding symptoms appear much later than those of D. noxia or S. graminum, and uncontrolled colonies do not result in plant death, although plant productivity may be affected (Kieckhefer & Gellner, 1992). Thus S. graminum negatively impacted both development (Fig. 1) and reproduction (Fig. 2) of D. noxia, whereas R. padi negatively influenced only D. noxia reproduction and the nymphal survival of S. graminum. These results are consistent with plant induction by R. padi proceeding more slowly compared to S. graminum with more delayed effects on co-infesting aphid species. The fact that D. noxia developmental time was not significantly extended in the presence of R. padi confirms that a negative influence of S. graminum, rather than a lack of conspecific pair feeding benefits, was responsible for the delayed development. Furthermore, the higher rate of intra-plant movement by both D. noxia and R. padi in the presence of S. graminum (Fig. 4) is indicative of a response by the former species to plant induction by the latter.

Diuraphis noxia causes specific changes in host plant architecture (leaf rolling) that result in the creation of a ‘cryptic niche’ (Burd et al., 1998), and the species has likely adopted sedentary feeding habits to capitalize on the benefits of feeding within this protected microhabitat (Telang et al., 1999; Qureshi & Michaud, 2005). The faster development of conspecific pairs of D. noxia nymphs that remained feeding on the same leaf for at least five consecutive days compared to those that fed on different leaves (Table 1) is evidence of a benefit of group-feeding that also appears localized within the plant tissue, although D. noxia appears to have little effect on vein loading or phloem translocation at its feeding site (Burd, 2002). Consequently, it is conceivable that part of the cost paid by D. noxia developing in tandem with S. graminum was attributable to its increased frequency of intra-plant movement, either through a reduction in total feeding time, or because its own plant inductive processes were not localized on one part of the plant. This cost was apparently not paid by R. padi, possibly because this species is adapted to frequent position changes on the plant, even in the absence of heterospecific aphids (Qureshi & Michaud, 2005). Similarly, the higher fecundity of S. graminum that developed in conspecific pairs on the same tissue for at least five consecutive days compared to those that fed on different tissues (Table 1) indicates a benefit of group-feeding on reproduction in this species.

The presence of R. padi not only reduced D. noxia fecundity, it also reduced the survival of second generation S. graminum nymphs, both indications that feeding by the former species induces defensive host plant responses with amensal consequences for co-infesting aphid species. Amensal interactions have been previously demonstrated between R. padi and S. avenae (Gianoli, 2000) and R. padi and the wheat leafhopper, P. alienus (Alla et al., 2001). These effects may result from R. padi feeding triggering the localized release of defensive compounds in the wheat plant (Gianoli & Niemeyer, 1997; 1998) that it seeks to avoid by frequent position change on the plant. Developing S. graminum nymphs, being relatively sedentary feeders, may have suffered from plant responses to R. padi feeding by not responding to their induction with position change on the plant. Interestingly, R. padi was the only aphid species to avoid any measurable negative impact in both heterospecific combinations and even displayed a trend toward higher fecundity after maturing in the presence of D. noxia (Fig. 2), although the effect was not significant because of high within-group variances in reproduction. Thus R. padi, perhaps the most generalist feeder of the three species and the only host-alternating aphid (Leather & Lehti, 1982), appeared to be the least negatively impacted in heterospecific interactions.

Although inter-specific interactions mediated by the host plant were clearly evident in this study, many of the host plant inductive processes seemed localized to some degree within plant parts. These aphid species differ significantly with respect to their preferred feeding locations on the plant (S. graminum: primary leaf, D. noxia: tertiary leaf, R. padi: stem) and these preferences were not altered by the presence of heterospecific aphids. Disparate feeding locations may reflect some degree of niche partitioning among these aphid species that share a range of host plants. The fact that there were no significant differences in fecundity or progeny survival among aphids transferred into conspecific or heterospecific pairs as fourth instars (Table 2) suggests that a period of more than several days of co-infestation by pre-reproductive aphids is required before any heterospecific impact can be realized. Thus interspecific interactions among these aphid species through altered plant physiology might not be an increasing, monotonic function of density as envisioned for classical interspecific competition (Faeth, 1992).