Phloem-sap feeding by animals: problems and solutions

Abstract

The incidence of phloem sap feeding by animals appears paradoxical. Although phloem sap is nutrient-rich compared with many other plant products and generally lacking in toxins and feeding deterrents, it is consumed as the dominant or sole diet by a very restricted range of animals, exclusively insects of the order Hemiptera. These insects display two sets of adaptations. First, linked to the high ratio of non-essential:essential amino acids in phloem sap, these insects contain symbiotic micro-organisms which provide them with essential amino acids. For example, bacteria of the genus Buchnera contribute up to 90% of the essential amino acids required by the pea aphid Acyrthosiphon pisum feeding on Vicia faba. Second, the insect tolerance of the very high sugar content and osmotic pressure of phloem sap is promoted by their possession in the gut of sucrase-transglucosidase activity, which transforms excess ingested sugar into long-chain oligosaccharides voided via honeydew. Various other animals consume phloem sap by proxy, through feeding on the honeydew of phloem-feeding hemipterans. Honeydew is physiologically less extreme than phloem sap, with a higher essential:non-essential amino acid ratio and lower osmotic pressure. Even so, ant species strongly dependent on honeydew as food may benefit from nutrients derived from their symbiotic bacteria Blochmannia.

Essential amino acids, Hemiptera, honeydew, phloem sap, sucrose

Introduction

Sugar-rich diets taste good to most animals. Plants provide three types of sugar-rich foods for animals: nectar, fleshy fruits (berries, drupes, etc.), and phloem sap. The relationship between these plant products and animals is exploitative. For nectar and fruits, the exploitation is reciprocal and for phloem sap, the exploitation is unidirectional (animals exploit plants). Both nectar and fruits have evolved in response to the ‘sweet tooth’ of animals, are accessible to animals, and are of a composition that maximizes animal foraging: as a reward for animal-mediated pollination (floral nectar), animal protection from herbivores (extrafloral nectar), and animal dispersal of seeds (fruits) (Herrera and Pellmyr, 2002). Plant phloem sap has as its principal function the long-distance transport of nutrients, especially photosynthate, around the plant (Fisher, 2000), and the diversion of these nutrients to animals is not in the plant's selective interests. In other words, the relationships of plants with nectar and fruit feeders are generally mutualistic and those with phloem-feeders are generally antagonistic.

Current understanding of phloem sap utilization by animals is shaped by two facts. First, just one group of animals, insects of the order Hemiptera, includes members that use phloem sap as their dominant or sole food source. This lifestyle has evolved multiple times among the Hemiptera, and is displayed by most Sternorrhyncha, including whiteflies, aphids, mealybugs, and psyllids, many Auchenorrhyncha, for example, most planthoppers and many leafhoppers, and most plant-feeding Heteroptera, including the lygaeids, pentatomids, and coreids (Dolling, 1991). A few other animal groups (including some thrips and lepidopterans among the insects, sapsuckers and hummingbirds among the birds, and some primates) consume phloem sap occasionally, but none require it as a component of their diet (Dailey et al., 1993; Passamani and Rylands, 2000). Second, all phloem-feeding hemipterans possess symbiotic micro-organisms that are vertically-transmitted, i.e. passed from mother to offspring. In the sternorrhynchans and auchenorrhynchans, the micro-organisms are intracellular, restricted to specific cells; and in the heteropterans they are localized in specialized diverticula of the gut (Buchner, 1965; Douglas, 1989). Several groups have reverted secondarily from feeding on phloem sap to whole plant cells. These insects (e.g. typhlocybine leafhoppers, phylloxerid aphids) lack symbiotic micro-organisms (Buchner, 1965), suggesting that the micro-organisms are advantageous to the insect only in the context of phloem sap feeding.

These considerations raise two questions: in what ways are hemipteran insects uniquely predisposed to use phloem sap as food; and why is symbiosis with micro-organisms linked to this habit? In this article, as one contributory factor, it is suggested that phloem sap poses nutritional barriers to utilization by animals that only the hemipterans have overcome, partly through the nutritional contribution from their symbiotic micro-organisms. Other hemipteran traits important to phloem feeding include the anatomy and function of the insects mouthparts and gut, and these are addressed by Goodchild (1966) and Douglas (2003).

Although phloem feeding is restricted to hemipterans, other animals use phloem sap by proxy. These are the so-called ‘cryptic herbivores’ (Hunt, 2003) that feed on the honeydew of hemipterans. Honeydew is phloem sap modified in composition by passage through the hemipteran gut and released via the anus. In the latter part of this article, the phloem sap and honeydew are compared as food.

Phloem sap as a food source

In some respects, phloem sap is an excellent diet for animals. For most plants, it approximates to a ‘predigested’ food with high concentrations of sugars providing an abundant source of carbon and energy, and nitrogen predominantly in the form of free amino acids. It is also generally free of toxins and feeding deterrents, a consequence of its being a highly specialized cytoplasm (plant secondary compounds tend to be localized in the apoplast and cell vacuole, and not the cytoplasmic compartment). There are exceptions to these generalities, including the high protein content in the phloem sap of cucurbits, at 10–40 μg μl⁻¹ compared with 0.1–2 μg μl⁻¹ in most other plants (Thompson and Schulz, 1999), and the phloem-mobile secondary compounds in various plants, for example, glucosinolates in crucifers and other Capparales (Merritt, 1996; Brudenell et al., 1999), cardenolides in the Asclepiadaceae (Botha et al., 1977), and pyrrolizidine alkaloids in various groups (Hartmann, 1999). These exceptions notwithstanding, phloem sap remains a poorly-defended, nutrient-rich food source for those animals that can access it.

Despite this, phloem sap poses two major nutritional problems for animals. These problems can be described as the ‘nitrogen barrier’ and ‘sugar barrier’ that animals must overcome to use phloem sap. The nature of these barriers and the response of phloem-feeding insects to them will be considered.

The nitrogen barrier to phloem sap utilization

The growth and fecundity of phytophagous insects are generally limited by nitrogen, in two ways: the quantity of nitrogen, i.e. the total amount of nitrogen available; and the quality of nitrogen, or its composition. The issue of quality arises because animals are metabolically impoverished, lacking the ability to synthesize nine of the 20 amino acids that make up protein. (These nine ‘essential’ amino acids are listed on the x-axis of Fig. 1A). If the concentration of just one of these essential amino acids is in short supply, protein synthesis and growth of an animal are constrained.

Fig. 1.

The nitrogen barrier to phloem sap utilization: amino acid relations of the pea aphid Acyrthosiphon pisum line LL01 feeding on Vicia faba. (A) Amino acid content of V. faba phloem sap and aphid protein (excluding the non-essential amino acids cysteine and proline, which cannot be quantified by the method adopted) asx: aspartic acid and asparagine; glx: glutamic acid and glutamine. (B) Amino acids derived from Buchnera symbionts during the 2 d of the final larval stadium, as calculated from the difference between amino acids required for protein growth and acquired by feeding on plant phloem sap. [Unpublished data of LB Minto, E Jones and AE Douglas and data from Wilkinson and Douglas (2003) which has been reproduced here with kind permission of Blackwell Publishing.]

The nitrogen barrier to phloem sap utilization is its low nitrogen quality. Broadly speaking, the ratio of essential amino acids:non-essential amino acids in plant phloem sap is 1:4–1:20, considerably lower than the ratio of 1:1 in animal protein. Data for the phloem sap of the broad bean Vicia faba and the legume-feeding pea aphid Acyrthosiphon pisum illustrate this mismatch (Fig. 1A). The amino acids in the phloem sap of V. faba are dominated by asparagine, which accounts for 72% of the total amino acids in this dataset. All the essential amino acids are detectable in the phloem sap samples, but their combined concentration represents just 8.2% of the total and the concentration of all but one essential amino acid (histidine) is proportionately lower in phloem sap than in aphid protein.

The essential amino acid content of phloem sap is insufficient to support the observed growth rate of the aphids. This can be illustrated for final instar larvae of A. pisum that ingest, on average, 1.92 μl phloem sap over 2 d (from days 6 to 8 after birth), during which time they increase in weight from 0.7 mg to 1.32 mg, equivalent to protein growth of 31 μg (AE Douglas, unpublished data). Even if the unlikely assumption is made that the aphids assimilate the ingested essential amino acids and convert them to protein with 100% efficiency, the phloem-derived amino acids are inadequate to support the observed protein growth for all essential amino acids other than histidine. The shortfall varies from 0.06 nmol tryptophan to 21.6 nmol leucine per aphid (Fig. 1B), and represents 5% and 60–86% of the total aphid requirement for tryptophan and the other essential amino acids, respectively. The shortfall in the dietary supply of amino acids is met by an endogenous source. As described below, there is overwhelming evidence that the endogenous source is symbiotic bacteria, Buchnera sp., which synthesize and provide these nutrients to the aphid. In other words, Buchnera cells enable aphids to overcome the nitrogen barrier to phloem sap utilization.

The biology of Buchnera has been reviewed recently in Douglas (2003). The key information required in the present context is that Buchnera sp. is a coccoid γ-proteobacterium that dominates the microbiota of aphids, accounting for >90–99% of all microbial cells in the aphid tissues. Buchnera is obligately intracellular, restricted to the cytoplasm of specialized insect cells, known as bacteriocytes, in the aphid haemocoel (body cavity) and transferred vertically to eggs (or early embryos for viviparous morphs) in the female reproductive organs.

The evidence that Buchnera provide aphids with essential amino acids is 3-fold: nutritional, physiological, and genomic. The nutritional and physiological approaches have depended on the development of two sets of techniques: to eliminate the Buchnera from the aphids using antibiotics, generating aphids known as aposymbiotic aphids (Wilkinson, 1998); and to rear the aphids on chemically-defined diets that can be manipulated (Dadd, 1985). The key results of nutritional research over many years are that aphids with their normal complement of Buchnera can be reared on diets from which individual essential amino acids are eliminated, but aposymbiotic aphids have an absolute requirement for all the essential amino acids (reviewed in Douglas, 1998). The complementary physiological experiments demonstrate that aphids with Buchnera, but not aposymbiotic aphids, can synthesize essential amino acids from dietary precursors such as sucrose and aspartate (Douglas, 1988; Febvay et al., 1995, 1999; Wilkinson et al., 2001; Birkle et al., 2002). Together, these experiments indicate that bacteria are responsible for the capacity of aphids to utilize phloem sap poor in essential amino acids by providing the aphid tissues with these nutrients.

The genomic evidence supporting the central role of Buchnera in providing essential amino acids to aphids comes from annotation of the complete genome sequences of Buchnera, now available for isolates from A. pisum (Shigenobu et al., 2000), Schizaphis graminum (Tamas et al., 2002), and Baizongia pistacea (van Ham et al., 2003). The Buchnera in all these aphid species have a small genome (0.62–0.64 Mb) with few genes (553–630, including pseudogenes). Most unusually for bacteria, nearly all the genes have orthologues in other bacteria, including Escherichia coli. In other words, the Buchnera genome approximates to a subset of the E. coli genome. Exceptionally, Buchnera have retained genes coding for most enzymes in the biosynthetic pathways for essential amino acids, even though they have lost many other metabolic capabilities, including the capacity to synthesize most non-essential amino acids (Table 1).

Table 1.

Amino acid biosynthetic capability of Buchnera, the symbiotic bacterium in aphids: + genes present; − genes absent: data collated fromShigenobu et al. (2000), Tamas et al. (2002), andvan Ham et al. (2003)

Essential amino acids	Genes for biosynthetic pathway	Non-essential amino acids	Genes for biosynthetic pathway
Histidine	+	Alanine	−
Isoleucine	+	Arginine	+b
Leucine	+	Asparagine	−
Lysine	+	Aspartic acid	−
Methionine	+a	Cysteine	±c
Phenylalanine	+	Glutamic acid	−
Threonine	+	Glutamine	−
Tryptophan	+	Glycine	+
Valine	+	Proline	−
		Serine	−
		Tyrosine	−

a

The biosynthetic pathway is truncated in Buchnera relative to E. coli. Buchnera has metH, suggesting that it can synthesize methionine from homocysteine, but lacks metABC required for the conversion of homoserine to homocysteine.

b

Animals including most insects can synthesize arginine from ornithine. Arginine is a non-essential amino acid by the definition used here, i.e. it is synthesized de novo by the animals, but it is sometimes treated as an essential amino acid because it is synthesized at insufficient rates to support optimal growth rates of young rats. Present in Buchnera from Acyrthosiphon pisum and Schizaphis graminum, absent from Buchnera from Baizongia pistacea.

c

Present in Buchnera from Acyrthosiphon pisum and Baizongia pistacea, absent from Buchnera from Schizaphis graminum.

These studies suggest strongly that aphids overcome the nitrogen barrier to phloem sap utilization by their acquisition of essential amino acid-overproducing bacterial symbionts. As considered in the Introduction, all phloem-feeding hemipterans bear symbiotic micro-organisms, a minority of which have been studied phylogenetically. For example, psyllids bear γ-proteobacteria, known as Carsonella sp. (Thao et al., 2000); mealybugs have Tremblaya sp. in the β-proteobacteria (Baumann et al., 2002); and some planthoppers and a single tribe of aphids, the Cerataphidini, bear ascomycete fungi of the family Clavicipitaceae (Suh et al., 2001), misleadingly called ‘yeast-like’ symbionts in the early literature. By analogy with the aphid–Buchnera relationship, these micro-organisms can plausibly be argued to provide their insect hosts with essential amino acids, but direct evidence is lacking. If future research demonstrates uniformity in the nutritional role of the various micro-organisms, it suggests that the key symbiotic trait of essential amino acid provisioning has evolved independently in many different microbial lineages and is not an evolutionarily ‘difficult’ transition.

The sugar barrier to phloem sap utilization

The dominant compounds in phloem sap are sugars derived from photosynthetic carbon fixation. In many plants, most of the sugar is in the form of sucrose, a chemically-stable disaccharide of low viscosity. The phloem-mobile sugars in some plants, including many labiates, include oligosaccharides of the raffinose series, especially raffinose and stachyose (with one or two galactose units, respectively, transferred to the glucose moiety of sucrose); and some plants have appreciable levels of sugar alcohols, for example, mannitol in the Apiaceae, sorbitol in the Rosaceae (Ziegler, 1975).

The basis of the sugar barrier to phloem sap feeding is its very high concentration in phloem sap, up to and often exceeding 1 M sugar, and a resultant osmotic pressure 2–5 times greater than the osmotic pressure of the insect's body fluids. The insects ingest the phloem sap at a high rate, partly because phloem sap has high hydrostatic pressure and partly to ensure a sufficient supply of other phloem nutrients at low concentrations. The predicted consequence of the continuous flow of fluid at high osmotic pressure into the gut is the transfer of water from the body fluids to the gut contents and osmotic collapse of the insect. In other words, phloem-feeding insects are expected to shrivel as they feed.

The key evidence that aphids overcome the sugar barrier by osmoregulation is that the osmotic pressure of the honeydew, the egesta voided from the gut, is comparable to that of the body fluids and is lower than the ingested food (Fig. 2A). Equivalent data for other hemipterans are lacking. Analysis of honeydew sugar composition provides a clue as to how the osmotic barrier is overcome. When aphids are reared on chemically-defined diets with sucrose as the sole sugar, the dominant honeydew sugars are the monosaccharides, glucose and fructose, at low dietary sucrose concentrations (0.2–0.3 M), but oligosaccharides comprising mostly glucose moieties at high concentrations (0.5–1.0 M) (Fig. 2B); the sucrose-derived fructose is assimilated with high efficiency and used principally in respiration (Ashford et al., 2000). It is proposed that the transformation of ingested sucrose to oligosaccharides would tend to reduce the osmotic pressure of the gut contents because the osmotic pressure exerted by solutes is determined by their molality and not their weight.

Fig. 2.

The sugar barrier to phloem sap utilization: osmotic relations of the pea aphid Acyrthosiphon pisum. (A) The osmotic pressure of aphid honeydew is similar to that of aphid haemolymph (body fluids) and lower than that of the phloem sap of Vicia faba plants on which they were feeding. (B) The oligosaccharide content of aphid honeydew is strongly dependent on the sucrose concentration ingested from the chemically-defined diet. (Data from Wilkinson et al., 1997).

These data suggest that the sugar relations of phloem-feeding insects are intimately linked with osmoregulation. At an enzymological level, the fate of ingested sugars is best-understood in aphids. For example, pea aphids have very high sucrase activity localized in the gut distal to the stomach (Ashford et al., 2000; Cristofoletti et al., 2003). The sucrase is an α-glucosidase (i.e. it is specific for the α-glucosyl residue of sucrose) and not a β-fructosidase (specific for the β-fructosyl residue) (Ashford et al., 2000). It has been proposed that the sucrase enzyme may also mediate the synthesis of oligosaccharides by transglucosidation, i.e. by inserting glucose and not water at the glucosidic bond (Walters and Mullin, 1988; Ashford et al., 2000). Transglucosidation activity has also been demonstrated in the whitefly Bemisia, generating an array of honeydew sugars (Byrne et al., 2003), but its incidence in other phloem-feeding insects is unknown.

An important implication of these results is that the microbiota play no part in the capacity of insects to utilize sugar-rich phloem sap. Although some early literature has suggested a role of micro-organisms in osmoregulation and in the sugar relations of phloem-feeding insects (reviewed in Douglas, 1989), the balance of current evidence is that micro-organisms have no direct role. In particular, aphids treated with antibiotics to eliminate Buchnera maintain their capacity to osmoregulate their body fluids to a constant value over a wide range of dietary sucrose (Wilkinson et al., 1997).

Phloem sap as a variable resource

Phloem sap varies in composition over multiple timescales, from the diurnal cycle to the season, with the developmental age of the plant, and with abiotic factors such as temperature and water availability (Douglas, 1993; Geiger and Servaites, 1994; Kehr et al., 1998; Ponder et al., 2000; Corbesier et al., 2001; Karley et al., 2002). The capacity of phloem-feeding insects to respond to long-term changes, for example, associated with plant development or season, has been amply demonstrated for aphids. This has been achieved by the use of chemically-defined diets with a range of different compositions that reflect different phloem sap compositions (Karley et al., 2002).

By contrast, there is very little information on the response of insects to diurnal variation in phloem sap composition, even though this variation can be considerable. Phloem sugar levels may differ between day and night by 10–20% to 2-fold, varying with irradiance, temperature, and other factors (Geiger and Servaites, 1994), and diurnal variation in amino acid composition has also been reported (Winter et al., 1992; Corbesier et al., 2001). This short-term variation cannot be mimicked readily using chemically defined diets, which are of fixed composition. Much of the insect response may be behavioural, through variation in the rate of food uptake according to the nutrient content and osmotic pressure. Where compositional changes are very rapid or large, the insect has the option to withdraw the stylets and probe for a different sieve element. [Phloem nutrient composition is far from uniform across sieve elements, even in plants reared under tightly controlled conditions, as illustrated by the 3–5-fold variation in amino acid concentration among replicate samples of exudates from severed stylets (Girousse et al., 1996; Telang et al., 1999).] Rapid post-ingestive responses may also be involved, including changes in gut sucrase activity and the function of transporters in sugar and amino acid assimilation, but the underlying biochemical and molecular processes remain to be established.

There is a real possibility that the capacity of an insect to utilize a plant is influenced by the scale of diurnal variation in phloem sap composition, as well as by the composition quantified at any one point in the diurnal cycle. In this context, datasets on phloem sap composition obtained during the light period, and usually at a fixed time, may not reflect accurately the total nutritional inputs to phloem-feeding insects.

Phloem feeding by proxy

After digestion and assimilation of ingested phloem sap in the hemipteran gut, the residue is voided via the anus as honeydew. Honeydew is often produced in copious amounts. For example, first instar larvae of the willow aphid Tuberolachnus salignus produce more honeydew per hour than their body weight (Mittler, 1958).

Honeydew is used as a food by various animals which can be considered as secondary or proxy phloem feeders that exploit the capacity of hemipterans to access plant sieve elements. Many insects, including flies, wasps, bees, beetles, butterflies, and moths, as well as nectarivorous birds and flying foxes, consume honeydew that has fallen onto plants or other surfaces; and some animals take honeydew droplets directly from the anus of the hemipterans. This behaviour, called tending, is widely displayed among ants, especially among the dolichoderines and formicines, and also by polybiine wasps, silvanid beetles and, remarkably, some Madagascan geckos that specifically tend planthoppers (Folling et al., 2001). Most research has concerned ant-tending relationships. They are mutualistic: the tending ants gain food and the tended phloem-feeding hemipteran is protected from natural enemies by the ants. It has recently been argued that the evolution of tending for hemipteran honeydew contributed to the ants ‘breaking out’ from their lifestyle as ground predators to colonize arboreal habitats and exploit phloem sap by proxy some 40–50 million years ago, triggering their subsequent dramatic diversification (Wilson and Holldobler, 2005).

A vivid demonstration of the quantitative significance to ants of honeydew feeding comes from stable isotope analysis, specifically of ¹⁵N:¹⁴N (δ¹⁵N). The ¹⁴N isotope is lost preferentially in catabolic reactions and therefore herbivores are predicted to be enriched in ¹⁵N relative to plants, their predators enriched relative to herbivores, and so on through the trophic levels (Griffiths, 1998). In a detailed analysis of ants in tropical rainforests of Peru and Brunei, Davidson et al. (2003) established that δ¹⁵N of ant species tending hemipterans is similar to that of phloem sap-feeding hemipterans and chewing herbivores, but lower than that of predatory ants (Fig. 3). In ecological terms, tending ants are herbivores, gaining access to phloem sap through mutualism with hemipterans.

Fig. 3.

Mean δ¹⁵N values () of ant species in a Borneo rainforest, classified according to predominant feeding mode, and calibrated against δ¹⁵N of plants (A), hemipterans (B), chewing herbivorous insects (C), and arthropod predators (D). [Redrawn with permission from Fig. 2 of Davidson DW, Cook SC, Snelling RR, Chua TH. 2003. Explaining the abundance of ants in lowland tropical rainforest canopies. Science300, 969–972. Copyright 2003 AAAS omitting species with mixed and uncertain foraging strategies.]

The conclusion that hemipteran honeydew is a quantitatively important component of the diet of many ants raises the issue of the nutritional suitability of honeydew as a food for animals. Honeydew is nutritionally distinct from phloem sap because of the enzymatic and assimilatory capabilities of the hemipteran gut. In particular, the sugars are modified by hydrolysis and transglucosidation, and the amino acid profile is altered by differential assimilation.

With respect to the sugar barrier to phloem feeding, the osmotic challenge posed by high phloem sugar is negated by the osmoregulatory capabilities of phloem feeders, making honeydew an osmotically-neutral foodstuff of complex and variable sugar composition. (The osmotic pressure of honeydew is, however, expected to rise with increasing time after deposition through the evaporative loss of water, posing osmotic problems for animals feeding on honeydew on plant surfaces etc). Furthermore, the capacity of ants and other honeydew feeders to digest the complex honeydew sugars, and its enzymological basis, remain to be investigated in detail.

The significance of the nitrogen barrier to honeydew feeding depends critically on the amino acid assimilation patterns of hemipterans. In the aphid species studied in this laboratory, the amino acid composition of honeydew is generally more balanced than in phloem sap because aphids preferentially assimilate non-essential amino acids (Adams, 1997). For example, when the black bean aphid Aphis fabae, which is facultatively tended by ants, feeds from Vicia faba, the essential amino acids in its honeydew account for 30% of the total amino acid content, about 2-fold higher than their percentage-contribution to phloem sap, at 13% (Fig. 4). In this dataset, all of the essential amino acids and just two of the non-essential amino acids (serine and aspartic acid) are proportionately enriched in honeydew, relative to phloem sap. However, it would be inappropriate to generalize from these data because there could be considerable variation among plant–hemipteran relationships. Also, ant-tending could affect amino acid assimilation patterns, and so honeydew composition.

Fig. 4.

Amino acid content of phloem sap and aphid honeydew for the black bean aphid Aphis fabae feeding on Vicia faba, expressed as percentage of the total amino acids. Asparagine (63% and 47% of phloem sap and honeydew amino acids, respectively) is not included in the figure. [Data of Adams (1997).]

Is the nitrogen nutrition of honeydew-feeding ants promoted by microbial symbionts? This possibility is suggested by the presence in some ants of an intracellular bacterium allied to Buchnera and known as Blochmannia sp. The genome of Blochmannia sp. has been sequenced and includes the full gene complement for synthesis of all nine essential amino acids (Gil et al., 2003). In principle, therefore, Blochmannia could provide ants with these nutrients. Consistent with this interpretation, ant species with Blochmannia have been cited to ‘show a preference for honeydew and other sweet secretions’ (Zientz et al., 2004), although a detailed analysis of the incidence of Blochmannia in ants with different nutritional ecologies remains to be conducted. If ¹⁵N:¹⁴N stable isotope analysis (as used in Fig. 3) were used as an index of nutritional ecology, interpretation would require great care because the amino acid biosynthetic function of the microbial symbionts would tend to deplete the ¹⁵N content of the ants, so underestimating ant dependence on honeydew or other plant products.

Mutualisms are a recurring theme in animal utilization of sugar-rich plant products. Phloem sap feeding differs from nectar and fruit feeding in that it is generally an antagonistic animal–plant interaction (see Introduction), but it does involve other mutualisms both between animals (the phloem feeders and their tenders) and between animals and micro-organisms. Furthermore, on some plants, ants both tend phloem-feeding hemipterans and protect the plant from herbivory (Heil and McKey, 2003), thereby transforming an antagonistic relationship between plant and phloem feeders into a crucial element to a three-way mutualism. As an example, the Crematogaster ants that form nests in hollow stems of Macaranga species that lack extrafloral nectaries both tend scale insects and protect the plant from natural enemies (Heckroth et al., 1999).

Future research directions

Recent research summarized in this article has identified two key elements underpinning phloem sap utilization by animals: possession of symbiotic micro-organisms that provide essential amino acids; and carbohydrases with transglucosidase function that reduce the osmotic pressure of ingested phloem sap. However, these explanations are partial in several ways. In particular, the molecular basis of these capabilities is largely unknown. The transformations of dietary sucrose in the insect gut are not understood at the levels of either enzymological or gene function and, to my knowledge, the only published study of insect utilization of phloem sugars other than sucrose was conducted 50 years ago (Duspiva, 1955). Similarly, the molecular basis of the nutrient and signal exchange underpinning the production and release of essential amino acids by Buchnera in aphids and by the symbiotic micro-organisms in other phloem-feeding hemipterans is still unknown. The other key limitation to our understanding is the narrow perspective of current research, with near-exclusive focus on aphid utilization of the most abundant phloem constituents, sugars and amino acids. This is unsurprising given the amenability of aphids to experimental manipulation and analysis of minor constituents of phloem sap. However, other potentially important phloem constituents are lipids, the phloem mobility of which is poorly known, and minerals. Recent developments in molecular and analytical approaches to study the molecular physiology of hemipterans (Douglas, 2003) and to dissect plant sieve element function and phloem sap composition (see the other papers in this Focus section) offer exciting new opportunities to study the traits of hemipteran insects which promote the phloem-feeding habit. Only then will it be possible to resolve the fundamental question in this field: why has utilization of phloem sap as dominant or sole diet uniquely evolved in hemipteran insects?

I thank Professor Doyle McKey, Professor Diane W Davidson, and Dr Steven C Cook for helpful comments on a draft of this manuscript. Previously unpublished data presented in this article was obtained in research funded by BBSRC grant 87/S16725.