Abstract The honeydew composition and production of four aphid species feeding on Tanacetum vulgare, and mutualistic relationships with the ant Lasius niger ...
Oecologia (1999) 118:483±491
Ó Springer-Verlag 1999
Wolfgang VoÈlkl á Joseph Woodring á Melanie Fischer Matthias W. Lorenz á Klaus H. Homann
Ant-aphid mutualisms: the impact of honeydew production and honeydew sugar composition on ant preferences
Received: 2 March 1998 / Accepted: 16 November 1998
Abstract The honeydew composition and production of four aphid species feeding on Tanacetum vulgare, and mutualistic relationships with the ant Lasius niger were studied. In honeydew of Metopeurum fuscoviride and Brachycaudus cardui, xylose, glucose, fructose, sucrose, maltose, melezitose, and ranose were detected. The proportion of trisaccharides (melezitose, ranose) ranged between 20% and 35%. No trisaccharides were found in honeydew of Aphis fabae, and honeydew of Macrosiphoniella tanacetaria consisted of only xylose, glucose and sucrose. M. fuscoviride produced by far the largest amounts of honeydew per time unit (880 lg/ aphid per hour), followed by B. cardui (223 lg/aphid per hour), A. fabae (133 lg/aphid per hour) and M. tanacetaria (46 lg/aphid per hour). The qualitative and quantitative honeydew production of the aphid species corresponded well with the observed attendance by L. niger. L. niger workers preferred trisaccharides over disaccharides and monosaccharides when these sugars were oered in choice tests. The results are consistent with the ants' preference for M. fuscoviride, which produced the largest amount of honeydew including a considerable proportion of the trisaccharides melezitose and ranose. The preference of L. niger for B. cardui over A. fabae, both producing similar amounts of honeydew, may be explained by the presence of trisaccharides and the higher total sugar concentration in B. cardui honeydew. M. tanacetaria, which produced only low quantities of honeydew with no trisaccharides was not attended at all by L. niger.
W. VoÈlkl (&) á M. Fischer á M.W. Lorenz á K.H. Homann Department of Animal Ecology I, University of Bayreuth, POB. 10 12 51, D-95440 Bayreuth, Germany J. Woodring Department of Biological Sciences, Lousiana State University, Baton Rouge, LA 70803-1715, USA
Key words Aphididae á Honeydew á Mutualism Ant attendance á Sugar preferences
Introduction Many homopteran species produce a carbohydrate-rich excretion called honeydew, which is collected mainly by ants (Way 1963; Buckley 1987; HoÈlldobler and Wilson 1990), but also by other insects (MacVicker et al. 1990; Novak 1992; Stapel et al. 1997). In temperate regions, aphids are the most important honeydew producers, many of which have developed a mutualistic relationship with ants (Buckley 1987; HoÈlldobler and Wilson 1990). Aphids bene®t from this mutualism with ants by reduced predation and parasitism and by a reduced risk of fungal infection (e.g. Banks 1962; Way 1963; Samways 1983; VoÈlkl 1992, 1997; Fischer et al. 1997). Ant-attended homopteran species may therefore compete for the mutualistic services of ants (Addicott 1978; Cushman and Addicott 1991; Cushman and Whitham 1991). Success in this competition depends on various factors, among which are species-speci®c dierences between the involved homopteran and ant species (Addicott 1978; Bristow 1984), density-dependent eects (Itioka and Inoue 1996), host plant quality (Auclair 1963; Breton and Addicott 1992) and seasonal dierences in the honeydew demand of the ants (Sudd and Sudd 1983; Fischer et al. 1997). On the other hand, ants should respond more intensively to more pro®table resources (Davidson 1978; Krebs and Kacelnik 1991; Nonacs and Dill 1991; Roces and Nunez 1993; Bonser et al. 1998). In the ant-aphid mutualism, such signi®cant dierences in patch quality may result from dierences in aphid colony size (Addicott 1979), from dierences in colony distance to the nest (Scheurer 1971) and from dierences in honeydew quality (Sudd and Sudd 1983; Cherix 1987). In the present study, we sought plausible explanations for attendance hierarchies in the mutualistic
484
relationship between the ant species Lasius niger L. (Hymenoptera: Formicidae) and four aphid species (Homoptera: Aphididae), all living on tansy, Tanacetum vulgare L. (Asteraceae). Among these species, Aphis fabae Scop. and Brachycaudus cardui (L.) are polyphagous and host alternating. They hibernate on Evonymus europaeus L. and Prunus spp., respectively, and the emigrant generation appears on T. vulgare and various other secondary host plants between the end of May and early June. Metopeurum fuscoviride Stroyan and Macrosiphoniella tanacetaria (Kaltenb.) are monophagous monoecious species living exclusively on T. vulgare (BoÈrner 1952). All species are abundant in the study area, frequently occur simultaneously within the same plant population and regularly even on the same T. vulgare shoot (Fischer 1997). In ®eld studies and laboratory experiments, L. niger showed distinct preferences when collecting honeydew on T. vulgare (Fischer et al. 1997). Colonies of M. fuscoviride were most intensively visited, while colonies of A. fabae were least visited. B. cardui held an intermediate position, and the fourth aphid species, M. tanacetaria, was not attended at all. In the study of Fischer et al. (1997), environmental conditions, aphid colony size (Addicott 1979), host plant quality (Breton and Addicott 1992) and presence of secondary plant compounds (Molyneux et al. 1990; Givovich et al. 1992; van Helden et al. 1994) were controlled. It was therefore hypothesized that two aphid-speci®c factors might be responsible for the preferences of L. niger (Fischer et al. 1997): either honeydew quantity or honeydew composition (Bristow 1991; Cushman and Addicott 1991; Hendrix et al. 1992). Honeydew represents an aqueous mixture of various sugars which constitute more than 98% of the dry weight, amino acids and secondary plant compounds (Mittler 1958; Maurizio 1985; Douglas 1993; van Helden et al. 1994). Since L. niger meets its protein demand mainly by hunting and eating small insect prey, carbohydrates are very likely the most important nutritional factor in honeydew (Pontin 1963). The sugar composition of honeydew seems to depend both on the aphid species and host plant species (Nemec and Stary 1990; Hendrix et al. 1992). Most honeydews so far studied contained a mixture of monosaccharides (mainly fructose, glucose), disaccharids (sucrose, trehalose, maltose) and trisaccharides (melezitose, ranose, fructomaltose). Among these sugars, melezitose, which is particularly abundant in conifer aphid honeydew (Maurizio 1985), seems to be very important for ant-aphid relationships, since it is thought to be preferred by ants (Kiss 1981). In the present study, we examined whether the demonstrated preferences of L. niger for dierent aphid species on T. vulgare in the ®eld are correlated with aphid-speci®c dierences in honeydew production or in honeydew sugar composition. First, we analysed sugar composition and sugar concentration of honeydew using high-pressure liquid chromatography (HPLC). Second, we quanti®ed the honeydew production per unit time to investigate the amount of sugar available. Third, we
tested the sugar preferences of L. niger and tried to ®nd correlates between the ants' sugar preference, honeydew amount and composition, and the observed hierarchies in the ®eld.
Materials and methods Insect and plant material All aphid colonies were kept on potted tansy (T. vulgare). Only plants of the same clone were used, which were obtained in the previous year and which were planted into a de®ned soil (obtained from Bayreuth Botanical Garden) to standardize plant quality. The potted plants were kept in the ®eld at a sunny site, but transferred to a climate chamber (20°C, 16:8 L:D, 60% relative humidity, 3000 lux from the ceilings) 3 days before being used in the experiments. To obtain aphid stock cultures, one large colony (>100 individuals) of each of the four studied aphid species (M. tanacetaria, A. fabae, B. cardui, M. fuscoviride) was collected in the ®eld. Twenty-®ve to 30 individuals of a colony were transferred separately to potted T. vulgare plants in a climate chamber (for conditions, see above). For our experiments, aphid colonies from these stock cultures were transferred to the standardized plants. After 24 h, they were reduced to a standardized colony size (15 fourth-instar larvae) by removing surplus aphids. Single nests of L. niger (>1000 workers per nest; all nests contained queens) were established in small terraria (70 ´ 35 ´ 35 cm) and kept in a climate chamber under the same experimental conditions as the aphid species. The ants were provided with freshly killed crickets (Gryllus bimaculatus) as a protein source and with M. fuscoviride colonies on T. vulgare as carbohydrate source. All ant nests contained brood when used in the experiments. Analysis of honeydew sugar composition and sugar concentration For the analysis of aphid honeydew sugar composition and sugar concentration, we used aphids feeding on a de®ned developmental stage of T. vulgare (stem and leaves were completely developed, ¯owers had not yet opened). Honeydew was collected directly from the aphid's anus using a microcapillary (volume 0.5 ll). By collecting only freshly secreted honeydew, a comparatively precise estimation of the sugar concentration could be obtained, since very little water could evaporate from the honeydew droplet in such a short period of time (J. Woodring, unpublished data). Each sample consisted of honeydew obtained from 8±15 aphids from a single colony. Phloem sap exudation was obtained by cutting phloem elements extending near the surface of the the T. vulgare stem. For honeydew analysis by HPLC, an Alltech ExSil 100 NH2, 3 lm, 100 AÊ, 250 ´ 3 mm column was used. Sample elution was isocratic with 75% acetonitrile (ACN) and a ¯ow rate of 300 ll/ min. For the detection, a photodiodearray-detector (Jasco MD 910) at a wavelength spectrum between 195±400 nm was used. The software chose the optimal wavelength from the obtained spectra to create the respective chromatograms (maximum absorption mode). The system was optimized by the injection of sugar standards (arabinose, xylose, rhamnose, fructose, glucose, sucrose, maltose, trehalose, melezitose, ranose, stachyose, turanose; no fructomaltose standard was commercially available). The composition of aphid honeydew was based on comparison of retention time to that of standards. Coelution of sugars in the honeydew with known standards supported this assumption. There was no dierence in the retention time for maltose and trehalose. Therefore, these sugars could not be distinguished. However, the sugar maltose is more likely to be a common constituent of the studied honeydew than trehalose, since only maltose could be detected in honeydew of A. fabae from T. vulgare when another system was used (R. Neu-
485 fang and D. Jensen, personal communication). Some consistent peaks with very long retention times could not be identi®ed. Typically, 0.5 ll honeydew was collected in 50 ll of 80% ACN and a 10-ll subsample was injected directly onto the aminopropyl column. The chromatogram was calibrated to read out directly (in lg) the amount of each sugar in the 10-ll subsample, which is equal to lg sugar per 0.1 ll honeydew. Quanti®cation of honeydew production
[(average weight of ants with filled gaster leaving aphid species i ÿ
Standardized aphid colonies (n = 10; n = 15 aphids; see above) of the three ant-attended species were established on potted T. vulgare by using aphids from the stock culture. Workers of a L. niger colony (n = 10 colonies) received access to the small aphid colony through an arrangement of wooden sticks. Ants were deprived of all carbohydrate supply for 24 h to ensure that they would have a strong demand for sugars. We measured the honeydew collected by the ants by comparing the weight dierences of ant workers with empty and ®lled gaster (Herzig 1937). First, we counted for 15 min all ant workers which left an aphid colony, starting 1 h after the ants had access to the aphid colony. Each ant colony was tested once with one colony of the three aphid species within a time period of 8 days (i.e. there was at least a 2-day pause between single runs). The temporal sequence of presenting aphid colonies varied between nests. Four nests were tested in the sequence M. fuscoviride-B. cardui-A. fabae, three nests with A. fabae-B. cardui-M. fuscoviride and 3 nests with B. cardui-A. fabae-M. fuscoviride. Second, we collected for each aphid species at least 36 ant workers that had just left a standardized aphid colony (=ants with ®lled gaster, subsets 1±3), and additionally 50 ant workers that had just left the nest and entered a wooden stick in the direction of the aphid colony (=ants with empty gaster, subset 4). The ants were immediately anaesthetized with ethanol-soaked paper and weighed using a microbalance (Sartorius 2004 MP). Hind tibia length was used as a size measure. We pooled workers collected from all ten L. niger nests for subsets 1±4, since preliminary measurements had shown that there was no signi®cant size dierence between workers of the dierent nests if randomly collected individuals (25 workers of each colony that were caught at the nest entrance and on the edge of the terraria) were compared (one-way ANOVA: F = 1.14, df = 9, P = 0.338). This result is consistent with what is already known about size dierences in L. niger (Collingwood 1979). The four subsets of ant worker samples did not dier signi®cantly in size (one-way ANOVA: F = 1.66, df = 3, P = 0.178). Furthermore, there was a signi®cant correlation between size and weight in all samples (r = 0.517, n = 201, P < 0.001), and the slopes (ANCOVA: F = 1.07, df = 3, P = 0.363) and the intercept (ANCOVA: F = 0.67, df = 3, P = 0.678) for the four groups did
Table 1 Sugar combinations used in preference tests with Lasius niger. All tests were carried out with aqueous solutions of 10% (w/ v) (Fru fructose, Glu glucose, Suc sucrose, Mel melezitose, Raf ranose, Suc/Glu sucrose:glucose 5:1, Suc/Mel sucrose:melezitose 5:1, Cp-h cristalline honeydew from Cinara piceicola (host plant: spruce)
Laboratory experiments Fru Glu Suc Mel Field studies Glu Suc Mel
not dier signi®cantly. Therefore, we concluded that we used homogeneous ant subsets in our experiments. The average weight dierences between similar-sized ants with empty and ®lled gaster should represent the amount of honeydew collected by an individual ant. Since all excreted honeydew droplets were collected by L. niger workers and since no division of labor among L. niger workers was observed in the experiments, we estimated the honeydew production of an individual aphid per hour using the following formula:
Suc
Mel
Raf
Suc/Mel
Suc/Glu
Cp-h
x x
± ± x
± ± x x
± ± x x
± ± x x
± ± ± x
x
± x
± x ±
± x x
average weight of ants with empty gaster) average number of ants leaving a colony of aphid species i]/number of aphids of species i per colony: M. tanacetaria was not attended by ants. For this species, we placed an aluminium pan below the standardized colony for 15 min (n = 10 replicates). The honeydew weight (dry mass after drying the pan for 2 h at 60°C) was calculated as the weight dierence between the clean and the honeydew-covered pan using a microbalance (Sartorius 2004 MP). Wet weight was estimated by multiplying the dry weight by a factor of 12. This factor was estimated from the dierence between fresh weight and dry weight of individually collected M. tanacetaria honeydew droplets.
Sugar preference of L. niger Sugar preferences of L. niger were tested both in laboratory and ®eld studies. In the laboratory, we set up three kinds of experiments. First, we tested whether L. niger collected pure solutions of sugars that were identi®ed in the honeydew of aphids from T. vulgare. We oered, without any choice, a surplus amount (10 ml) of water only and of a 10% (w/v) solution of xylose, fructose, glucose, sucrose, trehalose, maltose, melezitose or ranose to a colony of L. niger which was deprived of any carbohydrate source for 6 h. Each sugar was tested with ®ve colonies. Second, we tested the preferences of L. niger for various singlesugar solutions and sugar mixtures in paired tests. A colony of L. niger was deprived of any carbohydrate source for 6 h, then a surplus amount of two dierent sugar solutions (10 ml each), was oered via two dierent bridges. After a habituation phase of 10 min, all ants which crossed the bridges with ®lled gaster in the direction of the nest were counted for 10 min. Each combination was replicated four times to derive a mean value for a colony. In total, ®ve L. niger colonies were used. The temporal sequence in which various combinations were oered varied between colonies and did not follow a speci®c pattern. Single sugars (fructose, glucose, sucrose, melezitose, ranose; 10%, w/v), or mixtures (sucrose:glucose 5:1; sucrose:melezitose 5:1; 10%, w/v), or solutions of cristalline honeydew (10%, w/v) from Cinara piceicola (a heavily attended aphid feeding on Picea abies; VoÈlkl and Novak 1997) were presented (Table 1). The C. piceicola honeydew used in the tests contained approximately 55% melezitose, 20% sucrose and 25% fructose (J. Woodring, unpublished data). Third, we tested four sugar solutions (10%, w/v) which were oered simultaneously: sucrose, melezitose, a sucrose:melezitose mixture (5:1) and a solution of cristalline C. piceicola honeydew. The same procedure as in the ®rst experiment was followed, i.e. four replicates, 10 min each, per L.niger colony, with a total of ®ve L. niger colonies. However, the laboratory experiments might have been biased by the previous experience of L. niger workers with M. fuscoviride colonies and a subsequent conditioning eect, and/or by the fact that no alternative sugar source was available. Therefore, we conducted an additional ®eld study with ants with dierent experiences. We used ®ve L. niger colonies which were nesting in earth®lled pots. These colonies had access to colonies of A. fabae on various host plants (T. vulgare, Chenopodium spp., Matricaria inodora) which constituted the only honeydew source in their environment. This honeydew source remained accessible throughout
486 the experiments. The honeydew sugar composition of A. fabae from these host plants does not dier qualitatively (M. Fischer, J. Woodring, H.W. Lorenz, W. VoÈlkl, K.H. Homann unpublished data). Therefore, no melezitose was available for the ants prior to the experiment. We followed a similar experimental design as in the laboratory experiments. A surplus amount of two dierent sugar solutions (10 ml each) in small Petri dishes was placed near the nest entrance of a L. niger colony. The ants had access to the food via a Y-shaped bridge to avoid any bias of placing one source close to an existing ant trail. After a habituation phase of 20 min, all ants which crossed the respective leg of the bridge in the direction of the nest were counted for 10 min. Each combination was replicated four times to obtain a mean value for a colony. In total, ®ve L. niger colonies were used. Single sugars (glucose, sucrose, melezitose, ranose; 10%, w/v), or a mixture of sucrose and melezitose (5:1; 10%, w/v) were presented (Table 1). This experiment was carried out in June 1997 during a period when adults and brood were present in the ant nest. All studies were conducted in the afternoon at ambient temperatures between 20°C and 25°C and during sunshine.
Results Honeydew sugar composition of aphids on T. vulgare The analysis of honeydew of aphids feeding on T. vulgare revealed the presence of xylose, fructose, glucose, sucrose, maltose, melezitose and ranose. In samples of B. cardui and M. fuscoviride, there were two unidenti®ed peaks, one of which might represent the trisaccharide fructomaltose (R. Neufang and D. Jensen, personal communication). Figure 1 shows a typical chromatogram of a B. cardui honeydew sample. The sugar composition of the honeydew varied considerably between the four aphid species (Table 2). Xylose, glucose and sucrose occurred in varying concentrations in all samples, independent of the aphid species, while the trisaccharides, melezitose and ranose, were restricted to honeydew from M. fuscoviride and B. cardui. No fructose, maltose or trehalose could be identi®ed in honeydew from M. tanacetaria.
Fig. 1 HPLC chromatogram of a honeydew sample from Brachycaudus cardui. For methods, see text
The total sugar concentration also varied between species (Table 2). Honeydew from M. fuscoviride contained a total of 66.4 lg sugar/ll honeydew on average (range: 23±123 lg/ll). The sugar concentration in B. cardui honeydew (124 lg sugar/ll honeydew) and M. tanacetaria honeydew (95.5 lg sugar/ll honeydew), was higher than that of M. fuscoviride. That of A. fabae was considerably lower (29.1 lg sugar/ll honeydew). The major constituent of M. tanacetaria honeydew was xylose (ca 85 lg/ll). Furthermore, M. tanacetaria was the only aphid species in which the honeydew contained no maltose/trehalose. A. fabae honeydew was characterized by a high proportion of sucrose (8.9 lg/ ll = 30%) and maltose (7.8 lg/ll = 27%). The honeydew of B. cardui contained a large proportion of sucrose (46.1 lg/ll = 37%) and melezitose (22 lg/ll = 17%). The fructose proportion was low in the honeydew of all species studied (Table 2). High amounts of trisaccharides were found in M. fuscoviride honeydew (22.1 lg/ll = 33%). M. fuscoviride honeydew also contained considerable amounts of xylose and sucrose (Table 2). Phloem sap exudation from growing T. vulgare plants revealed a total sugar concentration of 350 lg sugar/ll phloem sap, in which the plant sugar sucrose predominated (226 lg/ll). Additionally, xylose (114 lg/ll) and glucose (10 lg/ll) were present. Honeydew production of aphids on T. vulgare Ant workers leaving M. fuscoviride colonies were signi®cantly heavier than workers with empty gaster or workers leaving A. fabae or B. cardui colonies (Fig. 2). Workers leaving A. fabae or B. cardui colonies also weighed more than workers with empty crop but these dierences were not signi®cant. Furthermore, the mean number of ants leaving the colony per unit time was signi®cantly higher in M. fuscoviride colonies when compared to A. fabae or B. cardui colonies (Fig. 3). There were also signi®cant dierences between A. fabae and B. cardui. The estimated honeydew production per aphid per hour was more than sixfold higher in M. fuscoviride than in A. fabae and approximately fourfold higher than in B. cardui (Fig. 4). The lowest honeydew production was measured in M. tanacetaria, the species which is not attended by ants. B. cardui honeydew contained by far the greatest concentration of ant-relevant sugars (i.e. sugars which were collected by L. niger: fructose, glucose, maltose, sucrose, melezitose, ranose; see below) per microlitre honeydew, while M. fuscoviride produced the greatest amount of ant-relevant sugars per hour (Table 3). Furthermore, the total production of trisaccharides per hour was more than threefold higher in M. fuscoviride when compared to B. cardui (Table 3). A. fabae and M. tanacetaria produced a much lower amount of ant-relevant sugars per hour (Table 3).
487 Table 2 Honeydew sugar composition of four dierent aphid species feeding on Tanacetum vulgare and of T. vulgare phloem exudates. Sugar concentrations (average) are presented in lg su-
Metopeurum fuscoviride (n = 5) Brachycaudus cardui (n = 2) Aphis fabae (n = 2) Macrosiphoniella tanacetaria (n = 2) Tanecetum vulgare (n = 2)
lg/ll % lg/ll % lg/ll % lg/ll % lg/ll %
gar/ll fresh honeydew (Xyl xylose, Fru fructose, Glu glucose, Suc sucrose, Mal maltose, Mel melezitose, Raf ranose, n number of samples)
Xyl
Fru
Glu
Suc
Mal
Mel
Raf
16.7 25.2 17.6 14.1 8.7 29.9 85.0 89 114 32.6
1.7 2.6 1.1 0.9 0.2 0.7 ± ± ± ±
4.8 7.2 22.5 18.1 3.5 12.0 4.0 4.2 10 2.8
2.8 4.2 46.1 37.1 8.9 30.6 6.5 6.8 226 64.6
18.3 27.6 11.1 8.9 7.8 26.8 ± ± ± ±
6.0 9.0 22.0 17.7 ± ± ± ± ± ±
16.1 24.2 3.9 3.2 ± ± ± ± ± ±
Fig. 2 Average weight of Lasius niger workers (mean + SE) leaving colonies of three aphid species feeding on Tanacetum vulgare in comparison to L. niger workers with empty gaster (control). n = number of weighed L. niger workers. Workers of ten ant colonies were pooled. Means sharing the same letter do not dier signi®cantly at P < 0.05 (one-way ANOVA with subsequent Schee's test)
Sugar preferences of L. niger All colonies of L. niger readily collected fructose, glucose, sucrose, maltose, melezitose and ranose, while none of them responded to the presence of xylose, trehalose or water. In choice experiments with two sugars or sugar mixtures, L. niger visited food sources containing trisaccharides (melezitose, ranose) more frequently than sources without trisaccharides. Melezitose was visited more frequently than ranose (Fig. 5). Sucrose was preferred over glucose, while there was no signi®cant dierence between sucrose and fructose, or between sucrose and a 5:1 sucrose:glucose mixture. A 5:1 sucrose:melezitose mixture was preferred over plain sucrose, but visited less often than plain melezitose. The honeydew solution was preferred over sucrose, while there was no dierence between honeydew and melezitose. When L. niger workers had a choice between four dierent sugars or mixtures, signi®cantly more
Total sugar concentration 66.4 124.3 29.1 95.5 350.00
Fig. 3 Intensity of attendance by L. niger (mean + SE) in three aphid species feeding on T. vulgare (n = 10 replicates for each species). Means sharing the same letter do not dier signi®cantly at P < 0.05 (one-way ANOVA with subsequent Schee's test)
workers collected a honeydew solution or melezitose than a 5:1 sucrose:melezitose mixture, or sucrose alone (Fig. 6). Field experiments revealed the same pattern (Fig. 7), although the absolute number of ants per time unit visiting the sugar baits was considerably lower than in the laboratory experiments. L. niger workers collected melezitose and ranose more intensively than sucrose, and a 5:1 sucrose:melezitose mixture was preferred over sucrose.
Discussion The honeydew composition diered considerably between the four aphid species feeding on T. vulgare. The honeydew of M. fuscoviride and B. cardui was characterized by the presence of large amounts of melezitose and ranose, while neither of these trisaccharides could be detected in honeydew of A. fabae and M. tanacetaria (Table 2). M. tanacetaria honeydew diered from A. fabae honeydew in lacking maltose and fructose. The lack of fructose is not important, because the fructose
488 Table 3 Estimation of the quantitative sugar production of various aphid species on T. vulgare
M. fuscoviride B. cardui A. fabae M. tanacetaria T. vulgare phloem exudates
ll honeydew/ aphid per houra
lg ant-relevant sugars/ll honeydewb
lg ant-relevant sugars /aphid per hourc
lg trisaccharides/ aphid per hourc
0.77 0.19 0.11 0.04
49.7 106.7 20.4 10.5 226.0
38.3 20.3 2.3 0.4
17.0 4.9 ± ±
a
See Fig. 4; the speci®c weight of honeydew was 1.18 (measured with a microbalance Sartorius MC 210 P) See Table 2: ant-relevant sugars include fructose, glucose, sucrose, maltose, melezitose and ranose c Calculated by multiplying the amount of ant-relevant sugars/ll honeydew and of trisaccharides/ll honeydew (see Table 2) with the honeydew production per hour b
Fig. 4 Estimation of quantative honeydew production of the four aphid species (mean + SE) feeding on T. vulgare. Means sharing the same letter do not dier signi®cantly at P < 0.05 (Kruskal-Wallis ANOVA with subsequent Tukey-Kramer test). For details, see Materials and methods
Fig. 6 Sugar preferences of L. niger when four solutions (10% w/v) were oered simultaneously (mean + SE) (n = 5 replicates) [S/M mixture sucrose:melezitose 5:1, honeydew cristalline honeydew from C. piceicola (host plant: spruce)]. Means sharing the same letter do not dier signi®cantly at P < 0.05 (Friedman ANOVA)
Fig. 5 Sugar preferences of L. niger in paired tests (mean + SE) (n = 5 replicates per combination) in the laboratory. All tests were carried out with an aqueous solution of 10% (w/v) [F fructose, G glucose, S sucrose, M melezitose, R ranose, SG sucrose:glucose 5:1, SM sucrose:melezitose 5:1, Cp cristalline honeydew from Cinara piceicola (host plant: spruce)]. In pairs marked with an asterisk, means (comparing solid and hatched bars) diered signi®cantly at P < 0.05 (Wilcoxon test). For details, see Materials and methods
Fig. 7 Sugar preferences of L. niger in paired tests (mean + SE) (n = 5 replicates per combination) in the ®eld. All tests were carried out with an aqueous solution of 10% (w/v) (G glucose, S sucrose, M melezitose, R ranose, SM sucrose:melezitose 5:1). In pairs marked with an asterisk, means (comparing solid and hatched bars) diered signi®cantly at P < 0.05 (Wilcoxon test). For details, see Materials and methods
concentration was very low in the honeydew of all species studied. Only three sugars ± xylose, glucose and sucrose ± were detected in all species. Xylose, a pentose, has not been recorded in honeydew until now (Maurizio 1985; Hendrix et al. 1992), but was also present at a high
concentration in the phloem sap of T. vulgare (Table 2). Xylose is a common constituent of plant cell walls (Driouich et al. 1993; Richter 1996), but a high concentration in phloem sap is unusual (Ziegler 1975; Kandler and Hopf 1981). Since all aphid species were
489
feeding on clones of the same host plant species (whose phloem sap consisted of sucrose, xylose and glucose only), host-plant-speci®c eects can be excluded as an explanation for the observed dierences in honeydew composition (Nemec and Stary 1990; Hendrix et al. 1992). Instead, we assume that species-speci®c dierences in the aphids' ability to transform the ingested disaccharide sucrose ± either enzymatically (Fisher et al. 1984; Walters and Mullin 1988) or by the help of symbiotic bacteria (Teo and Woodring 1985; Baumann et al. 1995) ± are responsible for the observed dierences in honeydew composition. Apparently, A. fabae and M. tanacetaria were not able to synthesize oligosaccharides, which may play an important role in aphid osmoregulation (Fisher et al. 1984; Walters and Mullin 1988; Wilkinson et al. 1997). The four aphid species also diered signi®cantly in the amount of honeydew produced. The honeydew production of M. fuscoviride was similar to that observed for Tuberolachnus salignus on willow (Mittler 1958). Fourth-instar larvae of T. salignus, which are considerably larger than those of M. fuscoviride, produce approximately 1.0 ll honeydew per hour. The honeydew production of B. cardui and A. fabae was approximately one-fourth and one-sixth, respectively, of that of M. fuscoviride. However, the production of A. fabae and B. cardui was considerably higher than values reported for aphid species that are not attended by ants. Eucallipterus tiliae, Tuberculoides annulatus, Acyrthosiphon pisum, Myzus persicae and Hyalopterus pruni excreted between 0.010 and 0.040 ll/aphid per hour (Heimbach 1986; Novak 1989; Mittler and Meikle 1991; Wilkinson et al. 1997), and a similar amount of excreted honeydew was estimated for M. tanacetaria (Fig. 4). Dierences in honeydew composition (qualitative eects) and honeydew production (quantitative eects) seem to be primarily responsible for the hierarchies in ant attendance within competing aphid species. Ants should be expected to focus honeydew collection activities on aphid species which promise a high reward, either in terms of larger volumes of honeydew, or in terms of the presence of preferred sugars or amino acids (Cushman 1991; Cushman and Addicott 1991). For example, Cherix (1987) observed that Formica yessensis collected primarily honeydew with a high sugar and amino acid concentration. Cushman and Addicott (1991) reported that the presence of conifer aphids (Cinara spp.) on the spruce Picea engelmanni reduced the numbers of ant workers attending Aphis varians (which produces less honeydew) on the adjacent plant Epilobium angustifolium. L. niger showed distinct preferences when collecting honeydew on T. vulgare in the ®eld, visiting colonies of M. fuscoviride most intensively, colonies of A. fabae least intensively and never attending M. tanacetaria (Fischer et al. 1997). Furthermore, L. niger abandoned A. fabae colonies and switched to M. fuscoviride or B. cardui if one of the latter species was oered to ants attending
A. fabae (Fischer 1997). Our results show that these attendance hierarchies in the ®eld can be explained by a combination of qualitative and quantitative nutritional eects. M. fuscoviride, the most heavily attended aphid species, produced by far the greatest volume of honeydew (Fig. 4), and the greatest amount of total sugars and trisaccharides (Table 3) per unit time. The total sugar concentration per microlitre was also high in the honeydew of B. cardui. L. niger preferred trisaccharides in both the ®eld and laboratory tests regardless of previous experience with available carbohydrate sources (M. fuscoviride honeydew in the laboratory, A. fabae honeydew in the ®eld). Thus, both quantitative and qualitative nutritional factors probably explain the high attractiveness of M. fuscoviride. The observed tendency to collect more honeydew per visit of M. fuscoviride (Fig. 2) is consistent with the ®ndings of Bonser et al. (1998) who found, that L. niger workers remained longer in high-quality patches and carried larger loads back to the nest than when visiting low-quality patches. B. cardui, which held an intermediate position in the L. niger attendance hierarchy, produced a smaller honeydew volume but of similar composition to that of M. fuscoviride, with a considerable amount of trisaccharides and an even higher sugar concentration (Tables 2, 3). By contrast, the low attractiveness of A. fabae for L. niger in comparison to B. cardui could be explained by the dierences in honeydew composition. A. fabae produced less honeydew than B. cardui, although there was no statistical dierence, the sugar concentration was considerably lower (Tables 2, 3) and the honeydew contained no melezitose or ranose. Since L. niger is able to recognize melezitose at very low concentrations (Schmidt 1938) and preferred a 5:1 sucrose:melezitose mixture over sucrose, a qualitative eect might be responsible for the low ranking of A. fabae in the ant attendance hierarchy. Furthermore, the higher total sugar concentration in B. cardui honeydew might in¯uence the ant's preference (Schmidt 1938; Sudd and Sudd 1983). M. tanacetaria, which is not attended by ants at all, produced the least amount of honeydew per unit time (Fig. 4), the least amount of ant-relevant sugars and no trisaccharides. A similarly low honeydew production was reported for the aphid species E. tiliae, T. annulatus, H. pruni, A. pisum and M. persicae (see above). These species are also generally not attended by ants, although the honeydew of E. tiliae contains substantial proportions of melezitose (Bacon and Dickinson 1957; Nemec and Stary 1990), as does honeydew of the generally unattended Macrosiphum euphorbiae (Walters and Mullin 1988). All these aphid species are feeding in comparably small ``loose'' colonies (rather than in dense colonies like facultatively or obligatorily ant-attended species), and a foraging ant worker will thus receive only a relatively low quantitative reward. Therefore, it might not pay for ant workers to attend these species (Nonacs and Dill 1990, 1991; Krebs and Kacelnik 1991; Bonser et al. 1998),
490
even if the honeydew contains considerable proportions of the preferred trisaccharide melezitose. Acknowledgements We thank K. Fiedler, F. WaÈckers and W. Weisser for their discussions and for critical comments on earlier drafts of the manuscript. R. Neufang and D. Jensen (Dionex, Idstein) kindly provided informations on unpublished results. The German research council provided ®nancial support (Ho 631/18-1).
References Addicott JH (1978) Competition for mutualists: aphids and ants. Can J Zool 56:2093±2096 Addicott JH (1979) A multispecies aphid-ant association: density dependence and species-speci®c eects. Can J Zool 57:558±569 Auclair JL (1963) Aphid feeding and nutrition. Annu Rev Entomol 8:439±490 Bacon JSD, Dickinson B (1957) The origin of melezitose: a biochemical relationship between the lime tree (Tilia spp.) and an aphid (Eucallipterus tiliae). Biochem J 66:289±299 Banks CJ (1962) Eects of the ant, Lasius niger (L.), on insects preying on small populations of Aphis fabae Scop. on bean plants. Ann Appl Biol 50:669±679 Baumann P, Lai CY, Roubakhsh D, Moran NA, Clark MA (1995) Genetics, physiology and evolutionary relationships of the genus Buchnera ± intracellular symbionts of aphids. Annu Rev Microbiol 49:55±94 Bonser R, Wright PJ, Bament S, Chukwu UO (1998) Optimal patch use by foraging workers of Lasius fuliginosus, L. niger and Myrmica ruginodis. Ecol Entomol 23:15±21 BoÈrner C (1952) Aphidae Europae Centralis. Mitt ThuÈr Bot Ges 4:1±184 Breton LM, Addicott JF (1992) Does host-plant quality mediate aphid-ant mutualism? Oikos 63:253±259 Bristow C (1984) Dierential bene®ts from ant-attendance on two species of Homoptera on New York ironweed. J Anim Ecol 53:715±726 Bristow C (1991) Why are so few aphids ant-tended? In: Huxley CR, Butler DF (eds) Ant-plant interactions. Oxford University Press, Oxford. pp 104±199 Buckley R (1987) Interactions involving plants, Homoptera, and ants. Annu Rev Ecol Syst 18:111±135 Cherix D (1987) Relation between diet and polyethism in Formica colonies. Experientia Suppl 54:93±115 Collingwood C (1979) Fauna entomologica Scandinavica 8: the Formicidae (Hymenoptera) of Fennoscandia and Denmark. Scandinavian Science Press, Klampenborg Cushman JH (1991) Host plant mediation of insect mutualism: variable outcomes in herbivore-ant interactions. Oikos 61:138± 144 Cushman JH, Addicott JF (1991) Conditional interactions in antherbivore mutualisms. In: Huxley CR, Cutler DF (eds) Antplant interactions. Oxford University Press, Oxford, pp 92±103 Cushman JH, Whitham TG (1991) Competition mediating the outcome of a mutualism: protective services of ants as a limiting resource for membracids. Am Nat 138:851±865 Davidson DW (1978) Experimental tests of the optimal diet in two social insects. Behav Ecol Sociobiol 4:35±51 Douglas AE (1993) The nutritional quality of phloem sap utilized by natural aphid populations. Ecol Entomol 18:31±38 Driouich A, Faye L, Staehelin A (1993) The plant Golgi apparatus: a factory for complex polysaccharides and glycoproteins. Trends Biochem Sci 18:210±214 Fisher DB, Wright JP, Mittler TE (1984) Osmoregulation by the aphid Myzus persicae: a physiological role for honeydew oligosaccharides. J Insect Physiol 30:387±393 Fischer M (1997) Hierarchien im Mutualismus zwischen Lasius niger und verschiedenen honigtauproduzierenden Blattlausarten. Diploma thesis, University of Bayreuth
Fischer M, VoÈlkl W, Homann KH (1997) Hierarchien im Mutualismus zwischen Lasius niger und verschiedenen honigtauproduzierenden Blattlausarten. Mitt Dtsch Ges Allg Ang Entomol 11:721±724 Givovich A, Morse S, Cerda H, Niemeyer HM, Wratten SD, Edwards PJ (1992) Hydroxamic acid glucosides in honeydew of aphids feeding on wheat. J Chem Ecol 18:841±846 Heimbach U (1986) Freilanduntersuchungen zur Honigtauabgabe zweier Zierlausarten (Aphidina). J Appl Entomol 101:396±413 Helden M van, Tjallingii WF, Beek TA van (1994) Phloem sap collection from lettuce (Lactuca sativa L.): chemical comparison among collection methods. J Chem Ecol 20:3191± 3206 Hendrix DL, Wei Y, Leggett JE (1992) Homopteran honeydew sugar composition is determined by both the insect and plant species. Comp Biochem Physiol 101B:23±27 Herzig J (1937) Ameisen und BlattlaÈuse. Z Ang Entomol 24:367±435 HoÈlldobler B, Wilson EO (1990) The ants. Springer, Berlin Heidelberg New York Itioka T, Inoue T (1996) Density-dependent ant attendance and its eect on the parasitism of a honeydew-producing scale insect, Ceroplastes rubens. Oecologia 106:448±454 Kandler O, Hopf H (1981) Occurrence, metabolism and function of oligosaccharides. In: Peiss J (ed) The biochemistry of plants. Academic Press, New York, pp 221±279 Kiss A (1981) Melezitose, aphids and ants. Oikos 37:382 Krebs JR, Kacelnik A (1991) Decision-making. In: Krebs JR, Davies NB (eds) Behavioural ecology ± an evolutionary approach, 3rd edn. Blackwell, Oxford, pp 105±136 MacVicker JAK, Moore JS, Molyneux DH, Maroli M (1990) Honeydew sugars in wild-caught Italian phlebotomine sand¯ies (Diptera: Psychodidae) as detected by high performance liquid chromatography. Bull Entomol Res 80:339±344 Maurizio A (1985) Honigtau ± Honigtauhonig. In: Kloft WJ, Maurizio A, Kaeser W (eds) Waldtracht und Waldhonig in der Imkerei. Ehrenwirth, Munich, pp 268±295 Mittler TE (1958) Studies on the feeding and nutrition of Tuberolachnus salignus. II. The nitrogen and sugar composition of ingested phloem sap and excreted honeydew. J Exp Biol 35:74±84 Mittler TE, Meikle T (1991) Eects of dietary sucrose concentrations on aphid honeydew carbohydrate levels and rates of excretion. Entomol Exp Appl 59:1±7 Molyneux RJ, Campbell BC, Dreyer DL (1990) Honeydew analysis for detecting phloem transport of natural plant products: implications for host-plant resistance to sap-sucking insects. J Chem Ecol 16:1899±1909 Nemec V, Stary P (1990) Sugars in honeydew. Biologia (Bratislava) 45:259±264 Nonacs P, Dill LM (1990) Mortality risk vs. food quality trade-os in a common currency: ant path preferences. Ecology 71:1886± 1892 Nonacs P, Dill LM (1991) Mortality risk versus food quality tradeos in ants: patch use over time. Ecol Entomol 16:73±80 Novak H (1989) Untersuchungen uÈber Produktion und Konsum von Honigtau in ausgewaÈhlten Hecken Oberfrankens. Diploma thesis, University of Bayreuth Novak H (1992) Untersuchungen uÈber Produktion und Konsum von Honigtau in Hecken Oberfrankens. Anz P¯anzenzucht SchaÈdlingskde Umweltschutz 65:41±48 Pontin AJ (1963) Further considerations of competition and the ecology of the ants Lasius ¯avus (F.) and Lasius niger (L.). J Anim Ecol 63:565±574 Roces F, Nunez JA (1993) Information about food quality in¯uences load-size selection in recruited leaf-cutting ants. Anim Behav 45:135±143 Richter G (1996) Biochemie der P¯anzen. Thieme, Stuttgart Samways MJ (1983) Interrelationship between an entomogenous fungus and two ant-homopteran (Hymenoptera: Formicidae ± Hemiptera: Pseudococcidae and Aphididae) mutualisms on guava trees. Bull Entomol Res 73:321±331
491 Scheurer S (1971) Der Ein¯uû der Ameisen und der natuÈrlichen Feinde auf einige an Pinus sylvestris lebende Cinarinen in der DuÈbener Heide. Pol Pism Entomol 41:197±229 Schmidt A (1938) Geschmacksphysiologische Untersuchungen an Ameisen. Z Vergl Physiol 25:3551±3378 Stapel JO, Cortesero AM, Moraes CM de, Tumlinosn JH, Lewis WJ (1997) Extra¯oral nectar, honeydew, and sucrose eects on searching behaviour of Microplitis croceipes (Hymenoptera: Braconidae) in cotton. Environ Entomol 26:617±623 Sudd JH, Sudd ME (1983) Seasonal changes in the response of wood-ants (Formica lugubris) to sucrose baits. Ecol Entomol 10:89±97 Teo LH, Woodring J (1985) Digestive enzymes in the house cricket, Acheta domesticus with special reference to amylase. Comp Biochem Physiol 82A:871±877 VoÈlkl W (1992) Aphids or their parasitoids: who actually bene®ts from ant-attendance? J Anim Ecol 61:273±281
VoÈlkl W (1997) Interactions between ants and aphid parasitoids: patterns and consequences for resource utilization. Ecol Stud 130:225±240 VoÈlkl W, Novak H (1997) Foraging behaviour and resource utilization of the aphid parasitoid, Pauesia pini on spruce: in¯uence of host species and ant attendance. Eur J Entomol 94:211±220 Walters FS, Mullin CA (1988) Sucrose-dependent increase in the oligosaccharide production and associated glucosidase activities in the potato aphid Macrosiphum euphorbiae (Thomas). Arch Insect Biochem Physiol 9:35±46 Way MJ (1963) Mutualism between ants and honeydew-producing Homoptera. Annu Rev Entomol 8:307±344 Wilkinson TL, Ashford DA, Pritchard J, Douglas AE (1997) Honeydew sugars and osmoregulation in the pea aphid Acyrthosiphon pisum. J Exp Biol 200:2137±2143 Ziegler H (1975) Nature of transported substances. In: Zimmermann MH, Milburn JA (eds) Transport in plants. I. Phloem transport. Springer, Berlin Heidelberg New York, pp 59±94
