Introduction

Mutualistic seed dispersal by ants, myrmecochory, can offer a wide range of benefits to plants (Beattie, 1985). Ants can increase dispersal distances, reduce intraspecific competition, and take the seeds to sites with improved conditions for germination and survival (Andersen, 1988; Bond et al., 1991; Espadaler and Gomez, 1996, 1997). Within the ant nest, seeds may be protected from fire and predators (Handel and Beattie, 1990; Bond et al., 1991; Christian and Stanton, 2004).

However, not all ants that carry seeds are plant mutualists. Granivorous ants take a wide range of seeds and consume the majority, although they may have a minor dispersal role by dropping some seeds (Hulme, 2002; Retana et al., 2004). Granivores tend to have a specialized seed diet (Hölldobler and Wilson, 1990), while mutualistic ants are generally scavengers with a predominantly invertebrate diet (e.g., Myrmica, Formica, Lasius) and collect only seeds with an elaiosome, which they remove and leave the seed unharmed (Hölldobler and Wilson, 1990; Gomez and Espadaler, 1998).

Thus, elaiosomes are a specific adaptation that facilitates non-predatory seed dispersal. A wide variety of tropical and temperate plants are myrmecochores (i.e., have an elaiosome)—currently 2100–3000 species in 60–80 families (Brew et al., 1989; Hughes and Westoby, 1992; Gomez and Espadaler, 1998)—and four of the seven ant families distribute seeds (Beattie, 1991). Elaiosomes are usually white or yellow appendages on the seed coat (Marshall et al., 1979) and are nutrient-rich, containing lipids, proteins, starch, sugars, vitamins (Kusmenoglu et al., 1989; Beattie, 1991; Lanza et al., 1992; Ohkawara et al., 1997), and essential nutrients that cannot be synthesized by ants, such as linoleic acid and sterols (Gammans et al., 2005).

Many studies have concluded that the elaiosome alone attracts mutualistic ants and induces them to carry the seed. Seeds with the elaiosome removed are often ignored in comparison with intact seeds or separated elaiosomes (Kjellsson, 1985; Skidmore and Heithaus, 1988; Brew et al., 1989). However, the mechanism by which the elaiosome affects ant behavior is not well understood. One hypothesis is that ants are responding simply to a food stimulus (Brew et al., 1989), but the chemical behavior-releaser hypothesis proposes that one or more chemical compounds from the elaiosome induces seed-carrying behavior (Sheridan et al., 1996). Several studies have suggested that this chemical inducer may be the fatty acid oleic acid or the diglyceride 1,2-diolein (Wilson et al., 1958; Marshall et al., 1979; Gordon, 1983; Skidmore and Heithaus, 1988; Brew et al., 1989; Kusmenoglu et al., 1989; Lanza et al., 1992; Table 1). For example, Marshall et al. (1979) investigated the interaction between the ant Aphaenogaster rudis and the plant Viola odorata. Ants were found to react most strongly to lipid fractions in bioassays, and further fractionation of lipids into fatty acids, sterols, and diglycerides showed that the last induced the greatest reaction from Aphaenogaster. An alternative hypothesis to chemical cues is that the brightly colored elaiosome acts as a visual attractant (Davidson and Morton, 1981), but this has not been tested.

Table 1 Studies that have suggested elaiosome chemicals may attract ants
Full size table

In general, studies of elaiosome attraction of ants have not differentiated between granivorous and mutualistic ants. In fact, certain ant species tested, such as Pogonomyrmex, Pheidole, Rhytidoponera, and Messor (see Rice and Westoby, 1986; Gomez and Espadaler, 1998), are usually seed eaters. If elaiosomes attract all ants equally, there is clearly a trade-off between the benefit of attracting mutualists and the disadvantage of attracting predators. There should be a selective advantage in being able to attract only mutualists. However, there has been no comparison of the attractiveness of elaiosome chemicals to mutualistic vs. granivorous ants.

Myrmica ruginodis is a mutualistic disperser of a wide range of myrmecochores (Kjellsson, 1985; Mark and Olesen, 1996; Heinken, 2004), including Ulex species on European lowland heaths (Stokes et al., 2003; Gammans et al., 2005). Tetramorium caespitum is a granivore that eats seeds of many plants including Ulex spp. (Brian, 1977; Skinner, 1987). In this article, we used laboratory colonies of M. ruginodis and T. caespitum to test the following hypotheses. (1) Chemicals on the surface of the elaiosome attract the mutualist M. ruginodis to the seed, but the granivore T. caespitum does not respond to these chemical cues. (2) M. ruginodis responds most strongly to the diglyceride fraction. (3) The yellow elaiosome on Ulex seeds is a visual cue for M. ruginodis. Our expectation was that patterns of response of either M. ruginodis or T. caespitum to elaiosome chemical extracts should be similar for both Ulex species.

Methods and Materials

Study Species

In southern England, Ulex minor produces seeds from the end of May for approximately 2 wk and U. europaeus from the middle of June until the end of July. The seeds are ejected explosively from pods that contain one to two (U. minor) or three to six (U. europaeus) seeds. Ants collect the seeds where they fall. U. minor seed length without the elaiosome is 2.0 ± 0.1 mm (mean ± S.E.). The elaiosome length is 1.2 ± 0.1 mm. U. europaeus seeds are slightly longer, 2.2 ± 0.1 mm, and the elaiosome is 1.2 ± 0.1 mm.

Myrmica ruginodis is a red ant 3–6 mm in length that forms polygynous (multiple queen) colonies with between 18 and 6560 workers (Wardlaw and Elmes, 1996). Myrmica species are generally scavengers with ∼80–90% of their diet made up of homopterans, hemipterans, dipterans, and arachnids, but they are also known to collect plant matter (Brian, 1977). In heathland habitats, nests of M. ruginodis are found under small heather (Calluna, Erica) or gorse (Ulex) bushes, and they forage in close proximity to Ulex species (Gammans et al., 2005). In contrast, T. caespitum eats seeds of grasses, Calluna vulgaris, Erica cinerea, and Ulex spp. on English heaths, although it will occasionally eat insect prey (Brian, 1977). It is a small black ant, typically 2–3 mm in length, which forms colonies of up to a few thousand workers (Skinner, 1987) in open habitats on southern English heaths (Brian, 1964).

Collection of Study Species

In 2004, ripe seed pods of U. minor and U. europaeus were collected from three heathland sites in the county of Dorset, southern England. Seed pods were stored in paper bags at 4°C. Seeds were extracted when needed. During March 2004, seven M. ruginodis colonies were collected from four Dorset heathlands. These colonies were split to form 40 laboratory nests comprising 50 workers and ∼20 larvae. Larvae were added to all nests to encourage the workers to forage for food. Each nest was placed into a nest box of 22 × 22 × 10 cm, which was attached to another smaller “foraging box” of 7 × 7 × 4 cm by a 20-cm-long plastic tube. Six T. caespitum colonies were collected from four heathlands in Dorset. Thirty laboratory nests (50 workers with 20 larvae) were set up in the same way, except that the boxes were slightly smaller at 12 × 8 × 2 cm and 7 × 4 × 2 cm, due to the smaller size of this species.

All colonies were fed each week with a standard diet of white sugar and Drosophila melanogaster larvae placed into the foraging box, and were watered regularly (Brian and Abbott, 1977; Elmes et al., 2004; Gammans et al., 2005). All colonies survived the experiment and there was a low mortality of ants, similar to other studies (Wardlaw et al., 1998; Gammans et al., 2005).

Standard Procedure for Laboratory Bioassays

All colonies were allowed to habituate to the laboratory nests for 1 wk before the first bioassays. Nests were selected at random, and each nest was used only once in each set of bioassays. When no bioassays were conducted, ants were allowed to roam freely between the nest and foraging boxes. Before each bioassay, any remaining food was removed from the foraging box, which was then cleaned with water to remove any food residue and any possible pheromone trails. Any ants within the foraging box were removed except for a randomly chosen “focus” ant, which was allowed to settle, and the foraging tube was then closed. If the focus ant left the foraging box during the bioassay, the foraging tube was opened and remained so until a new ant entered under its own volition. If no ant was present in the foraging box at the start of the experiment, the tube was left open until an ant entered. This procedure minimized disturbance and ensured that the workers observed were active foragers.

During bioassays, seeds, elaiosomes, or filter paper with chemical extracts were always handled with sterile forceps to prevent contamination. The lid of the foraging box was kept open during the bioassay to prevent build up of any volatile compounds.

Bioassays were conducted during times of day when both ant species normally forage; for M. ruginodis 06:00–12:00 and for T. caespitum 12:00–18:00 hr. Experiments were conducted with U. minor from May 26 to July 9, 2004, and with U. europaeus from July 12 to August 2, 2004. Although this approach meant that bioassays were conducted on the two Ulex species at different times of the year, it was preferred to ensure that seeds were as fresh as possible for the bioassays. Data analysis reflected this lack of randomization by analyzing results for the two Ulex species separately.

Ant Responses to Seeds, Elaiosomes and Diaspores

Bioassays were carried out to assess ant responses to diaspores, separate elaiosomes, and seeds from which the elaiosome had been removed. Three of each “treatment” (diaspore, elaiosome, or seed) were placed into a foraging box, and the number removed to the nest box was counted after 24 hr. Each treatment was replicated in 20 nests for each ant species.

Ant Behavior Toward Diaspore Surface Compounds

Further bioassays were done using extracts of elaiosome chemicals on filter papers. The surface chemicals of 90 U. minor or U. europaeus diaspores (seed plus elaiosome) were extracted by immersion in diethyl ether for 8 hr. When required, the extract was added to filter paper squares of 2–3 mm, which were then left overnight at room temperature to allow the ether to evaporate. Filter paper was used as the substrate to test reaction to chemical extracts on an object that did not resemble a diaspore. The amount of extract added was equivalent to one diaspore. Control paper squares were immersed in diethyl ether, and then left overnight to allow evaporation. Three control or extract filter paper squares were placed into the foraging boxes of randomly selected nests with only one treatment used per nest. Fifteen colonies were given the extract and 15 the control. Ant behavior was recorded for a period of 20 min. Four distinct ant behaviors were recorded as follows: (1) ignore: the ant paid no attention to the filter paper; (2) antennate: the ant touched the filter paper with its antennae; (3) bite: the ant bit the filter paper with its mandibles; (4) carry: the ant carried the filter paper in its mandibles. We also noted the time between the start of the bioassay and the first antennation of the filter paper. In preliminary experiments, it was noticed that the focus ant would repeat behaviors, especially when showing interest in the filter paper, but might then ignore it for some time, before returning and showing renewed interest. Repetition of behaviors also occurred when a focus ant was replaced by another. Therefore, the total time spent (to the nearest second) and the number of repetitions of all four behaviors was recorded.

Lipid Fractions from Elaiosome Extracts

For each Ulex species, 70 elaiosomes were removed from the diaspore and slightly macerated and submerged for 24 hr in 3 ml of chloroform to separate the different lipid fractions of the elaiosomes. The lipid fractions of these extracts were separated with solid phase extraction (1 ml SPE NH2; Qmx) using the standard methodology described by Kalunzy et al. (1985). Solvents were used in the following sequence: 15% ethyl acetate in hexane to separate diglycerides; hexane for cholesterol esters; 2:1 chloroform/methanol for monoglycerides; 1% diethyl ether in 10% methanol in hexane for triglycerides; 5% ethyl acetate in hexane for cholesterol; a second elution with 5% ethyl acetate in hexane for residual cholesterol; 2% acetic acid in diethyl ether fatty acids; and methanol for phospholipids. The eluent from each wash was stored at −20°C until use in bioassays. The diglyceride fraction (to which M. ruginodis reacted) for U. minor was sent to the MRS Lipid Analysis Unit in Dundee for quantitative analysis.

For all bioassays, a volume of the lipid fraction equivalent to one elaiosome was added to a filter paper square, and the respective solvent used in the fractionation was added to filter paper as a control. Both sets of papers were left overnight at room temperature to allow the solvent to evaporate. As before, three filter papers of either treatment or control (only one treatment used per nest) were then placed into the foraging box of a randomly selected nest. Thirty nests were used for each ant species; half were given the extract and half the control. All replicates and controls for each particular fraction were run consecutively. The different fractions were used in random order. The behavior of the focus ant was recorded for 20 min using the protocol described above.

Visual Cues

An experiment was designed to investigate whether visual cues are used by M. ruginodis when finding Ulex diaspores. U. minor diaspores (N = 270) were submerged in diethyl ether for 8 hr to remove any surface chemicals. In sets of 30 diaspores, the elaiosomes were painted different colors with nontoxic acrylic paint: brown, yellow, green, red, and blue; one set was left unpainted. A further 30 diaspores were also used without the surface chemicals removed. Three diaspores per color were placed into 10 randomly chosen nests (some nests were repeated), and ant behavior was recorded for 20 min using the protocol described above. Diaspores were left in the foraging arena, and their final location was recorded after 24 hr.

Analysis

All results were analyzed with the Minitab statistical package by using analysis of variance ANOVA where the data were normally distributed. Some data sets had many zero values and remained nonnormally distributed after transformation. These data sets were analyzed by Kruskal–Wallis tests. Treatments and, where relevant, ant species were included as factors in the analyses, with a fully randomized design. Because the Ulex species were subjected to bioassays on different dates, they were analyzed separately.

Results

Ant Responses to Diaspores, Seeds and Elaiosomes

The ant M. ruginodis moved more elaiosomes (58%; i.e., proportion of the three elaiosomes removed across 20 replicates) than diaspores (18%), and moved seeds without diaspores the least (6%), while the respective percentages for T. caespitum were 30%, 8%, and 5%. Therefore, while the strong preference of both species for elaiosomes led to a treatment effect (F 2,108 = 29, P < 0.001), the preference of M. ruginodis for diaspores over seeds compared with the relative lack of distinction between the diaspore and seed-only treatments by T. caespitum led to an ant species × treatment interaction (F 2,108 = 3.2, P < 0.05).

Surface Compounds

The major response of ants to filter papers involved antennation. The biting and movement of papers was rare, so the data were not analyzed. ANOVA indicated that M. ruginodis antennated the filter papers containing the surface extract for longer and for a greater number of times than the control for both Ulex species, although time till first antennation was significantly different between treatments only for U. europaeus (Table 2). T. caespitum showed no difference in any behavior in response to the extracts of either Ulex species (Table 2).

Table 2 Behavioral responses (Mean ± S.E.) of M. ruginodis and T. caespitum ants to elaiosome surface chemical extracts of two Ulex species
Full size table

Lipid Fractions from Elaiosome Extracts

Sixteen analyses (i.e., eight fractions × two Ulex species) were performed on the data from the lipid fraction experiments for each behavior of each ant species. A 5% significance threshold means 16/20 = 0.8 tests on average would be significant by chance. However, if the behavioral responses toward a particular fraction are consistent, and biologically meaningful, one would expect an ant species to show responses to a fraction that is similar for the two Ulex species. Tests on the two Ulex species were independent, and the probability of significant response in terms of a particular behavior to the same fraction of both Ulex species is 8/20 (probability of one test in eight being significant at 5%) × 1/20 (probability of the same fraction in the second Ulex species having a significant effect at 5%) = 2%.

The comparison of M. ruginodis behaviors toward each lipid fraction and its solvent revealed that the only fraction that caused significantly greater reactions than its solvent was the diglyceride fraction for both Ulex species. The number (U. minor extract mean = 4.5, control mean = 0.9; U. europaeus extract = 8.0, control = 2.0) and duration (U. minor extract = 7 sec, control = 0.9 sec; U. europaeus extract = 16.2 sec, control = 3.7 sec) of antennations were greater for both Ulex species, and the time till first antennation was faster for U. minor (extract = 127 sec, control = 446 sec; Tables 3 and 4). For U. europaeus, the hexane solvent was also antennated for significantly longer than the cholesterol ester fraction (Table 4).

Table 3 M. ruginodis and T. caespitum behaviors toward the lipid fractions of U. minor elaiosomes
Full size table
Table 4 M. ruginodis and T. caespitum behaviors toward the lipid fractions of U. europaeus elaiosomes
Full size table

Unlike M. ruginodis, T. caespitum did not show a changed behavior toward the diglyceride fraction (Tables 3 and 4) and there were no significant difference among any fractions and their solvents for U. europaeus (Table 4). Comparing the U. minor fractions with controls, the time until first antennation was significantly faster for the triglyceride fraction (extract mean = 205 sec, control mean = 734 sec), the cholesterol residual was antennated for a greater duration (extract = 5 sec, control = 0.4 sec) and frequency (extract = 2.7, control = 0.4), and the cholesterol fraction was antennated more times than its solvent (extract = 3, control = 0.4; Table 3).

According to the argument above, the responses of M. ruginodis to the diglyceride were greater than expected by chance. However, the T. caespitum responses were unconvincing, being inconsistent across the two Ulex species and, on average, only slightly more frequent than expected by chance.

The main constituents of the diglyceride fraction were hexadecadienoic acid, stearic acid, and arachidic acid. Oleic acid was present but in small amounts (Table 5).

Table 5 The fatty acid composition of the diglyceride fraction of U. minor elaiosomes
Full size table

Visual Cues

ANOVA and Kruskal–Wallis tests showed significant differences in all behaviors of M. ruginodis among the five artificial elaiosome colors, the diaspore with surface chemicals removed, and the unaltered diaspore (Table 6). Comparison of means suggested that the unaltered diaspore was more attractive than all the other treatments. When the unaltered diaspore treatment was removed from the analyses, there were no differences among the remaining treatments. The mean values for behaviors to all these other treatments were low, suggesting they were generally unattractive.

Table 6 Behavioral reactions (Mean ± S.E.) of M. ruginodis to U. minor diaspores with natural and artificial coloration
Full size table

Discussion

The data suggest that elaiosomes attract mutualistic ant species through chemical cues. M. ruginodis responded both to the general diaspore surface chemicals extracted with diethyl ether, and more particularly to the diglyceride fraction of elaiosome chemicals. Both mutualistic and granivorous species carried the diaspore, and both carried elaiosomes more frequently than the diaspore, probably because the elaiosome was easier to carry. The seed with the elaiosome removed produced less of a response, especially from M. ruginodis, which suggests the importance of the elaiosome in attracting this ant compared with T. caespitum. The lack of reaction of T. caespitum to the diethyl ether extracts suggests that the granivore did not respond to chemical cues.

The testing of T. caespitum responses to the chemical fractions showed four significant effects. These may be chance events because they were inconsistent across the two Ulex species, only slightly more frequent than expected by chance, and they conflict with the lack of response to the complete extract. In contrast, the M. ruginodis responses toward the diglyceride fraction were consistent for both U. minor and U. europaeus and, therefore, probably biologically meaningful.

Our results agree with those of Marshall et al. (1979), Brew et al. (1989), and Kusmenoglu et al. (1989) who found that ants responded to the diglyceride fraction of elaiosomes. However, this response seems to be more specific than has been assumed previously, and our results support the hypothesis that selection has acted on the evolution of cues that attract mutualists but not predators. The granivore detects seeds, possibly by simple mechanical cues, whereas the mutualist has more sensitive and accurate chemically-based recognition. M. ruginodis dependence on this chemical cue for finding Ulex diaspores was further demonstrated by the experiment on visual cues. It seems that visual cues are not used, as no color produced a greater reaction than another. In contrast, the removal of surface chemicals in this experiment decreased the responses of ants to the diaspore.

It is unclear whether the recognition of chemicals is gustatory, by antennation on or near the elaiosome, or by olfaction of volatiles. Slingsby and Bond (1981) suggested that ants can detect elaiosome chemicals over some distance, but Sheridan et al. (1996) produced evidence that elaiosome chemicals are detected by gustation. We often observed M. ruginodis antennating the air within 1–2 cm of elaiosome extracts or diaspores, suggesting olfaction. Diglycerides are not volatile, however, (Sheridan et al., 1996), and an olfactorial response is likely to involve a more volatile and as yet undetected compound(s). However, if it exists, there should be a trade-off between the volatility and the longevity of such a cue. Such trade-offs have been encountered with the parasitoid Ichneumon eumerus that attacks caterpillars of the butterfly Maculinea rebeli inside host ant colonies. A mimic of the ants alarm pheromone is used. However, mimic compounds have a greater molecular weight and are less volatile than the template, providing the intruders with more time to attack the caterpillar (Thomas et al., 2002).

M. ruginodis rarely bit or picked up filter papers, in contrast to behavior toward the diaspore. This result contrasts with Brew et al. (1989), who found that piths treated only with oleic acid, 1,2-diolein, and triolein were removed by ants at the same rate as elaiosomes. This suggests that in the Ulex–M. ruginodis system diglycerides act only in the initial stages of attraction, and that the stimulus to remove the diaspore comes subsequently. This may be a simple food cue released when the elaiosome is damaged by biting.

In this myrmecochorous system of Ulex seeds and M. ruginodis ants, there appears to be a highly evolved relationship between the species. Ulex elaiosomes release chemicals that attract M. ruginodis and contain essential nutrients, linoleic acid, and sterols, which increase nest productivity (Gammans et al., 2005). It may be that the specific chemical attractant arose because M. ruginodis evolved a refined ability to detect such chemicals, or Ulex elaiosome chemicals have evolved and become specifically attractive to a mutualistic ant, or both processes have occurred through coevolution.

It is well established that many plant species use chemicals to manipulate insect behavior for their benefit, e.g., through repelling herbivores (De Moraes et al., 2001) or attracting ants that defend the plant (Raine et al., 2004). The study of chemical attractants is important for understanding biotic interactions that affect populations and communities, and is an area in which further research is needed (Weissburg et al., 2002; Muller and Riederer, 2005).