- Top of page
- Materials and methods
Collembola are polyphagous soil invertebrates that colonize virtually all soils. Owing to their large numbers they constitute an important component of decomposer food webs. Collembola are assumed to feed predominantly on fungi, but their diet also includes algae and detritus (Anderson & Healey 1972; Newell 1984; Ponge 2000). Polyphagous species generally appear to dominate in decomposer systems, which commonly has been ascribed to the lack of coevolution between consumer and resources (Scheu & Setälä 2002). This argument, however, applies only to primary decomposers, which predominantly live on dead organic matter (i.e. plant litter). Fungal feeding species such as Collembola are confronted with living resources and, therefore, one may expect coevolutionary processes similar to those between plants and herbivores to have taken place. Food choice experiments and analyses of gut contents in Collembola indicate that food specialists hardly exist, suggesting that coevolutionary processes in soil differ from those above the ground.
It has been documented that Collembola prefer certain fungal species as food substrate, but preferences appear to differ little between Collembola species, suggesting wide overlap in fungal food resources among Collembola taxa (Klironomos & Kendrick 1995; Maraun et al. 2003). Despite the prevalence of generalist feeders and the use of mixed diets, surprisingly little is known about why mixed diets dominate in Collembola (Hopkin 1997). Fungi arguably contain one of the most complex sets of secondary chemicals (Turner & Aldridge 1983), which should facilitate specialist feeders that are able to detoxify or tolerate fungal toxins. Prevalence of generalist feeding suggests that Collembola benefit from toxin dilution because of the combined ingestion of toxic and high-quality food, and/or from a more balanced food supply owing to the ingestion of food materials complementing each other (Bernays et al. 1994; Bernays & Minkenberg 1997; Hagele & Rowell-Rahier 1999). Indeed, S. Scheu & F. Simmerling (unpublished observations) show that Collembola generally appear to benefit from feeding on mixed diets consisting of different fungal species. The present study complements this work in including a wider range of diets, i.e. conidial fungi (wild type and melanin-deficient mutant of Aspergillus fumigatus), an ectomycorrhizal fungus (Laccaria laccata) and an algae (Chlorococcum infusorium).
Ectomycorrhizal fungi and algae have been assumed to be of high food quality (Wolters 1985; Schultz 1991; Hopkin 1997). However, there are only few studies that experimentally test these assumptions (Shaw 1985; Hiol, Dixon & Curl 1994); most studies on Collembola nutrition are based on food selection and not on animal growth and reproduction, i.e. fitness (Booth & Anderson 1979; Ohlsson & Verhoef 1988). In the present study we analysed the effects of single and mixed diets on animal reproduction. We were particularly interested in evaluating whether including fungal species of low food quality in mixed diets reduces Collembola fitness.
Polyphagous species need to be able to judge the proportion of dietary components that maximize fitness. A precondition for analysing whether Collembola are able to optimize fitness by mixing of food resources in certain proportions is that the proportions of the food materials ingested can be traced, i.e. the extent to which each of the resources contributes to animal nutrition is quantifiable. If two resources are offered, the proportion of carbon incorporated from each of the resources may be determined by using food materials with different stable carbon isotope signatures and then analysing the resulting stable isotope signatures of the Collembola (two source mixing model; Bakonyi, Dobolyi & Thuy 1995; Briones, Ineson & Sleep 1999). In the present study, fungi of different stable carbon isotope signatures were obtained by growing them on agar containing sucrose of either C4 or C3 plant origin.
A difficulty in using mixing models for evaluating incorporation of resources from mixed diets is that consumers may discriminate between different isotopes, i.e. their signature may not exactly correspond to the signature of the food resource. Fractionation is known to be low for 13C but significant in 15N (Petersen & Fry 1987; Michener & Schell 1994). To use stable isotope methodology in food web analyses, the extent of 13C and 15N fractionation in a wide range of taxa of different trophic groups must be evaluated experimentally (Scrimgeour et al. 1995; Gannes, O’Brien & Martinez del Rio 1997; Kelly 2000). Decomposer food webs are a particular problem as saprotrophic bacteria and fungi are one of the main resources for a great number of soil invertebrates. More detailed information on stable isotope fractionation in decomposer food chains is urgently needed. Stable isotope analyses are a very promising tool to understand better the structure and functioning of decomposer food webs (Scheu & Falca 2000; Ponsard & Arditi 2000; Oelbermann & Scheu 2002). In this study we also investigated stable isotope fractionation in decomposer food chain components including carbon resources, fungi/algae and Collembola and evaluated whether feeding on mixed vs single diets affects the fractionation of 15N in Collembola.
- Top of page
- Materials and methods
The dietary species used differed strongly in food quality as measured by Collembola growth and reproduction. Consistent with previous findings, and our expectations, the food quality of A. fumigatus wt was lower than that of A. fumigatus mdf (S. Scheu & F. Simmerling, unpublished observations). Higher food quality of A. fumigatus mdf may have been caused by melanin deficiency because of the lack of the pksP gene, which is known to control melanin synthesis. Furthermore, the content of nitrogen in A. fumigatus wt (1·77%) was lower than that in A. fumigatus mdf (2·13%).
The factors driving food quality of fungi for fungal feeding microarthropods are little understood. Toxic substances reduce and high nutrient content increases the food quality of fungi (Booth & Anderson 1979; Shaw 1985; Sadaka-Laulan et al. 1998). A particularly puzzling phenomenon is that a great variety of fungal feeding soil invertebrates appear to favour similar fungal taxa as food substrate, generally dark pigmented forms (Maraun et al. 2003). Moreover, it is enigmatic why Collembola may generally prefer saprophytic over VA mycorrhizal fungi (Klironomos, Bednarczuk & Neville 1999; Gange 2000) and ectomycorrhizal fungi over other fungal groups (Schultz 1991; Hiol et al. 1994). The lower food quality of A. fumigatus mdf compared with L. laccata in our study was unlikely to be due to toxins (see above). High food quality in L. laccata and C. infusorium was probably a consequence of high nitrogen content (2·45 and 3·39%, respectively).
From an evolutionary perspective, generalist feeding of virtually all Collembola species could be supported if the additional ingestion of a species of low food quality to that of high food quality would increase Collembola fitness. This view, however, appears counterintuitive and is hard to integrate into current theories on the advantages of generalist feeding. Generalist feeding is assumed to be advantageous because it dilutes toxins and/or complements food components (Freeland & Janzen 1974; Pulliam 1975; Rapport 1980; Bernays et al. 1994). Results of the mixed diet experiment of the present study support the suggestion that the addition of low-quality food to that of high quality further increases Collembola fitness. None of the single diets allowed maximum growth of Collembola, suggesting that no single diet was of optimum quality. This conclusion is supported by the fact that in each of the single diets, except A. fumigatus wt, Collembola reproduction stopped after 36–44 days, whereas in the mixed diet treatments it continued for at least 44 days. Presumably, in the single diets essential nutrients necessary for continuous reproduction were lacking.
The results indicate that the prevalence of generalist feeding in Collembola and other microbivorous soil invertebrates might be due to the fact that most species of fungi and algae in soil provide a non-optimum diet. Consequently, to maximize fitness, microbivorous invertebrates have to feed on a variety of diets. A puzzling question in this scenario is why there are no or few species of fungi and algae that allow maximum growth. A possible explanation is the lack of coevolutionary processes that allowed microbivorous soil invertebrates to maximize their fitness when feeding on single species of fungi or algae (cf. Scheu & Setälä 2002).
Microbivorous soil invertebrates live in close contact with their food resources; 1 g of soil may contain several kilometres of hyphae of very different fungal species (Kendrick 1992). It may therefore be almost impossible to specialize on a single fungal species, i.e. ingestion of a mixture of different species of fungi, algae and bacteria, and different detritus substrates may be unavoidable. Presumably, evolutionary processes in microbivorous soil invertebrates resemble those in ungulate herbivores and filter feeders rather than those in above-ground insect herbivores.
As indicated by 13C analyses, H. nitidus differentially fed on the dietary species offered in mixed diets and ingested the two species in a ratio that increased reproduction, i.e. fitness. In both juveniles and adults 10–20% of the animal carbon in the L. laccata with A. fumigatus treatments originated from the species of low food quality (A. fumigatus). Compared with the L. laccata treatments, the admixture of A. fumigatus in the C. infusorium treatments was higher and more variable. As indicated by the results from the single and mixed diet experiment, food quality of C. infusorium was somewhat lower than that of L. laccata. This suggests that ingestion of food materials is more balanced if H. nitidus is confronted with food substrates differing less in quality. Results from the single diet experiment indicated that A. fumigatus mdf was of higher food quality than A. fumigatus wt. Despite this difference, H. nitidus ingested A. fumigatus wt to the same extent as A. fumigatus mdf if offered in combination. This indicates that H. nitidus is unable to distinguish the two forms of A. fumigatus which is consistent with previous results (S. Scheu & F. Simmerling, unpublished observations).
Calculations of the contribution of each of the two dietary species offered as food substrate to H. nitidus nutrition was based on a two-source mixing model. Since we also determined 13C ratios in H. nitidus after feeding on single diets this calculation was independent of differential fractionation of 13C in H. nitidus feeding on different diets. However, for interpreting data on the variation in stable isotope ratios in animals collected in the field it is essential to know to what extent stable isotopes are fractionated in consumers. In the present study, 13C ratios in H. nitidus generally resembled those in the diet. However, compared with juveniles, 13C ratios in adults differed considerably more from those of the diet. In part, this might have been due to incomplete replacement of animal carbon in adult H. nitidus during the 7 weeks of incubation. However, fractionation in adults was also more pronounced in the A. fumigatus mdf treatment in which the animals became more depleted in δ13C during the experiment (Fig. 3). This indicates that fractionation in 13C in adults in fact exceeded that in juveniles. Similar differences in δ15N in juvenile and adult H. nitidus suggest that fractionation in 15N also varies with Collembola age.
Fractionation of 13C in juveniles of H. nitidus was in the range of one to two δ units, which is within the range observed in other animals (Petersen & Fry 1987; Michener & Schell 1994). However, compared with the average increase in 13C ratios per trophic level of 0·4 δ units calculated by Post (2002), fractionation in the present study was high and surprisingly, in certain treatments, H. nitidus was depleted in 13C compared with its diet.
Fractionation of 15N considerably exceeded that of 13C with a minimum in the A. fumigatus mdf diet of 0·81 and a maximum in the L. laccata diet of 6·95. 15N ratios have been documented to increase by 3·4 δ units per trophic level (Wada, Mizutani & Minagawa 1991; Post 2002). Our data are consistent with findings that natural variations in 15N increase per trophic level, but the data indicate that the increase may vary considerably depending on the food materials ingested. Variations in 15N ratios with the quality of the diet has been reported previously (Webb, Hedges & Simpson 1998; Adams & Sterner 2000; Rothe & Gleixner 2000).
For interpreting natural variations in 15N and 13C ratios in microbivorous soil invertebrates it is critical to know the variations in stable isotope signatures in dietary species, i.e. to what extent bacteria, fungi and algae fractionate stable isotopes. It has been documented that 15N and 13C signatures of saprotrophic fungi differ from those of mycorrhizal fungi (Hobbie, Weber & Trappe 2001). Furthermore, 15N signatures in mycorrhizal fungi have been documented to vary significantly between species (Kohzu et al. 1999) and with nitrogen nutrition (Lilleskov, Hobbie & Fahey 2002). However, only sporocarps have been investigated and little is known on stable isotope fractionation in fungal hyphae, the diet of fungal feeding soil invertebrates. Results of the present study indicate that, similar to fungal sporocarps, the fractionation of stable isotopes in fungal hyphae (including conidia) may differ markedly between species. Fractionation in 13C in A. fumigatus wt considerably exceeded the mean trophic fractionation of 0·4 δ units (Post 2002). Interestingly, the fractionation differed between A. fumigatus wt growing on C3 and C4 sucrose, a phenomenon that has been reported previously for sucrose by Henn & Chapela (2000). In contrast to 13C, fractionation in 15N in fungi was considerably lower than the mean trophic fractionation proposed by Wada et al. (1991) and Post (2002). Kohzu et al. (1999) also reported that fractionation of 15N in decomposer fungi (sporocarps of wood-decomposing basidiomycetes) is low. Overall, the results suggest that data on natural variations in stable isotope ratios in fungal-feeding microarthropods collected in the field have to be interpreted with caution. Signatures of 15N may vary considerably between fungal species, i.e. within a single trophic group. Differences in 15N signatures in fungal feeding taxa therefore may reflect feeding guilds rather than the trophic structure of the community as previously hypothesized (Scheu 2002). Presumably, variations in 15N signatures in fungal feeding microarthropods can be used to differentiate, for example, between species feeding on saprotrophic and those feeding on ectomycorrhizal fungi. The differentiation of feeding guilds by stable isotope methodology may therefore contribute substantially to the understanding of the functioning of decomposer systems.