The contribution of fungus to the diets of three mycophagous marsupials in Eucalyptus forests, revealed by stable isotope analysis

Authors

  • A. P. Mcilwee,

    1. Cooperative Research Centre for Tropical Rainforest Ecology and Management, Department of Zoology and Tropical Ecology, James Cook University of North Queensland, Townsville Qld 4811, Australia
    Search for more papers by this author
  • C. N. Johnson

    1. Cooperative Research Centre for Tropical Rainforest Ecology and Management, Department of Zoology and Tropical Ecology, James Cook University of North Queensland, Townsville Qld 4811, Australia
    Search for more papers by this author

Abstract

1. Many field studies have shown that small herbivorous mammals include fungus (usually hypogeous sporocarps of ectomycorrhizal fungi) in their diets. However, the dietary importance of fungus relative to other foods is generally unclear because of limitations on the power of conventional techniques of diet analysis. Stable isotope analysis in conjunction with faecal analysis was used in an attempt to overcome these limitations.

2. Two foregut-fermenting marsupials (the Northern Bettong Bettongia tropica and Rufous Bettong Aepyprymnus rufescens) and a hindgut fermenter (the Northern Brown Bandicoot Isoodon macrourus) were studied. The Northern Bettong and Northern Brown Bandicoot are of similar body size (around 1 kg); the Rufous Bettong is significantly larger at 3 kg. Faecal analysis showed that the two bettongs ate a variety of grasses, lilies and fungi; the bandicoot ate these foods and also invertebrates.

3. Ratios of 15N/14N and 13C/12C differed in major food types collected in the field (fungus, grass, lily and invertebrates). Grass was clearly separated from the other food types by its low 13C/12C ratio, while fungus was separated from the other types by its high 15N/14N ratio. Invertebrates and lilies differed slightly in 13C/12C ratios.

4. Isotope ratios in body tissue (sampled in hair) of the three mammals were also discrete, showing that the species differed in the predominant sources of their C and N. Estimates of the proportion of C assimilated in body tissue that was derived from grass were 80% for the Rufous Bettong, 40% for the Northern Bettong and 45% for the Northern Brown Bandicoot. Analysis of 15N/14N ratios suggested that the Northern Bettong derived almost all its N from fungus, the Northern Brown Bandicoot derived practically no N from fungus, and the Rufous Bettong was intermediate.

5. The results confirm that for the Northern Bettong, fungus is a predominant source of N and C assimilated into body tissue. Differences between the use of fungi by the Northern Bettong and the Northern Brown Bandicoot strengthen conclusions from other studies that foregut fermentation confers on small mammals a greater ability to utilize fungus than does hindgut fermentation. It is hypothesized that the limited use of fungus by the Rufous Bettong is due to the patchy distribution of hypogeous sporocarps, which would result in a high energy cost of foraging for this larger-bodied species with a higher absolute food requirement.

Introduction

Many species of small mammals include fungus in their diets. This is especially true in coniferous forests of North America and eucalypt forests of Australia, where most ground-dwelling herbivorous and omnivorous mammals with a body mass of less than 2 kg are at least partially mycophagous (Maser, Trappe & Nussbaum 1978; Claridge & May 1994; Johnson 1996). The fungi eaten by these mammals are typically hypogeous ectomycorrhizal species that produce large truffle-like sporocarps. Such fungi are common and diverse in conifer and eucalypt forests and their sporocarps may be very abundant, at least seasonally (Luoma, Frenkel & Trappe 1991; Johnson 1994a).

The widespread occurrence of mycophagy suggests that hypogeous fungi are an important source of energy and nutrients for small forest-dwelling mammals, and that production of hypogeous sporocarps may be a limiting factor for some species. However, this conclusion is tenuous, for two reasons. First, mycophagous mammals tend to have mixed diets which include a variety of plant foods (leaf, fruit, seeds, etc.), and often some invertebrates, as well as fungus. The relative importance of these food types has been judged by faecal analysis or stomach-content analysis. Faecal analysis is subject to large biases produced by differential digestibilities of different food types. Many of the non-fungal foods eaten by mycophagous mammals are readily digestible and high in nitrogen, but chemical analysis of sporocarps of hypogeous fungi suggest that they contain many complex carbohydrates of low digestibility and that much of their nitrogen is bound up in forms not readily available to mammals (Cork & Kenagy 1989a; Claridge & Cork 1994). Faecal analysis, which characterizes the indigestible fraction of the ingested diet, is likely to overestimate the dietary contribution of foods of low digestibility and to underestimate consumption of high-quality foods. Stomach-content analysis will also overestimate the dietary contribution of bulky, low-quality foods.

Second, it appears that many small mammals that include hypogeous fungus in their diets do not have digestive adaptations that allow them to utilize it efficiently. Cork & Kenagy (1989a) showed in feeding trials that hypogeous fungi were a poor-quality food source for a mycophagous ground squirrel. Sciurids have a limited capacity for fibre digestion, which takes place in the hindgut only, and ingesta is passed rapidly through the gut (Cork & Kenagy 1989b; Hume 1994). Subsequently, Claridge & Cork (1994) showed that a mycophagous rat-kangaroo achieved high digestibilities of carbohydrate and nitrogen on a pure diet of hypogeous fungi. They suggested that mammals, such as rat-kangaroos, that have digestive systems specialized for foregut fermentation (Hume 1982; Cork 1994) are better able to utilize hypogeous fungi than are species that have hindgut fermentation.

This study investigated the utilization of hypogeous fungi in wild populations of three species of mycophagous marsupials in Eucalyptus forests in northern Australia. The major objective was to quantify the extent to which the carbon and nitrogen assimilated into the body tissues of each species was derived from fungus. This was achieved using stable isotope analysis: in this technique, stable isotope ratios in different foods are measured and compared with ratios of the same isotopes in body tissues (Rundel, Ehleringer & Nagy 1988). Because stable isotopes in foods are transferred to the body tissues of animals that eat them, differences in stable isotope ratios between food types provide markers that show the extent to which nutrients from different food types are assimilated into body tissue of the consumer. The stable isotope analysis was preceded by conventional faecal analysis to identify the range of food types used by each species.

The utilization of fungus was compared in three marsupial species, two rat-kangaroos (the Northern Bettong Bettongia tropica and the Rufous Bettong Aepyprymnus rufescens) and the Northern Brown Bandicoot (Isoodon macrourus). The two bettongs are specialized foregut-fermenting herbivores, while the bandicoot possesses a relatively simple digestive tract with hindgut fermentation (Hume 1982) more typical of most small herbivorous/omnivorous mammals. Inclusion of the Northern Bettong and Northern Brown Bandicoot allowed the comparison of utilization of fungus by a foregut-fermenting and a hindgut-fermenting species, with both species of similar body size (1–2 kg) and occupying the same habitat. Further, the inclusion of the Rufous Bettong, which has about twice the body mass of the Northern Bettong, allowed the effect of body size on the utilization of fungi in specialized foregut-fermenting herbivores to be considered.

Methods

STUDY SPECIES

The Rufous Bettong is widespread in dry forests and woodlands in eastern Australia. Adults typically weigh around 3 kg, and previous diet studies have shown that they eat roots, grasses and forbs as well as fungi (Schlager 1981; Seebeck, Bennett & Scotts 1989). The Northern Bettong (1–1·5 kg) has a restricted distribution in eucalypt forests and woodlands on the fringes of rainforest in northeastern Queensland. Field observations suggest that Northern Bettongs are primarily mycophagous, but also include a range of roots, tubers and grasses in their diet (J.W. Winter, personal communication). The Northern Brown Bandicoot (1–2 kg) is common in eucalypt forests in eastern and northern Australia. Bandicoots typically eat invertebrates and plant food as well as fungi (Claridge et al. 1991).

STUDY SITES

The study was conducted at two sites (designated A and B) on the western side of the Lamb Range near Mareeba in northeastern Queensland. Both were Eucalyptus-dominated woodlands (mainly E. crebra, Corymbia intermedia, C. citriodora and E. acmenoides) with a grassy understorey on granite-derived soils of low fertility. The two sites were 12 km apart, and had similar floristics and vegetation structure, but site A received ≈ 300–400 mm of rainfall per year more than site B. The Northern Bettong occurred at both sites, the Rufous Bettong occurred only at site B, and the Northern Brown Bandicoot was common enough for study only at site A. Mean annual rainfall for the general area over the study period was 1275 mm, 60% of which fell in the three months from January to March. Summers were warm and humid, winters were dry and mild. Field trips were conducted in September/October 1993 (late dry season), December 1993 (early wet season), January/February 1994 (mid wet season) and June/July 1994 (early dry season).

FAECAL ANALYSIS

Animals were captured at night in baited wire-mesh cage traps, and samples of faecal pellets were collected from beneath traps. Faecal samples were collected from each animal on the first capture only for each field trip, to avoid contamination of faeces with bait. For each animal species and season, faecal samples from the first eight individuals caught were retained for diet analysis. Samples were prepared for analysis as follows. A single pellet of standard size was taken from each sample, lightly ground in a mortar and left in Hertwigs solution for 6 days. It was then strained through a sieve (0·2 mm) and the retained material washed with 5% ammonia. A subsample of the remaining coarse fragments was stained in Lugols iodine solution, rinsed with 5% ammonia, then stained with 1% gentian violet. The material was again rinsed and mounted on microscope slides in glycerol jelly.

Fragments in faeces were assigned to the following types: fungus, grass, lily or forb, root or tuber, fruit or seed and invertebrate. Reference material for each type of plant food and fungus was prepared by first cutting samples into 5-mm lengths and boiling these lightly in Jeffries Solution (10% chromic acid and 10% nitric acid). Samples of epidermis could then be separated from underlying plant tissue, and were mounted in glycerol jelly on microscope slides after staining in gentian violet. Thin sections of other food types were mounted in the same manner.

Slides were viewed at 100× magnification, and all fragments in each food type were scored and recorded in 30 fields of view, randomly located along parallel transects across the slide. All fragments in each field of view were given scores of: (1) item covering less than 25% of view; (2) item covering 25–50% of view; (3) item covering 50–75% of view; or (4) item covering greater than 75% of view. For each slide, food types were represented as a percentage by calculating the total score for that type divided by the total for all types multiplied by 100. A small proportion of material (x = 7·56 ± 0·53%, n = 128) was indistinct, and could not be ascribed to any food category. Similarly, a small proportion of plant material (x = 7·65 ± 0·55%, n = 128) could not be ascribed to any plant group. This material had no organized cellular structure and was perhaps matrix material between cells or tissue layers.

Statistical comparisons among the proportion of different food groups eaten by the three species were made using analysis of variance, performed on arcsine-transformed values to normalize the percentage contributions of different food types (Zar 1974). Fruits and seeds formed only a minor fraction of faecal material with a combined mean of 0·88 ± 0·27% (n = 128) for all three species of mammal, and were excluded from statistical analysis.

STABLE ISOTOPE ANALYSIS

Stable isotope ratios of carbon and nitrogen were analysed in various food items known (from faecal analysis) to be consumed by the three study species. Samples were collected from both sites A and B. At each location, three to six individuals for each species were collected and combined to form a pooled sample for that species. Samples analysed from each site included five species of grass, four species of lily and forb and five species of hypogeous fungi. In addition, two samples of 10 ground-dwelling insects (cockroaches, crickets, cicadas and grasshoppers) and one sample of 10 subterranean invertebrates (worms and beetle larvae) were also analysed. These samples were immersed in liquid nitrogen and pulverized with a mortar and pestle. To remove possible liquid nitrogen residues, ground samples were oven dried at 60 °C for 2 days. All samples were oxidized and the resultant CO2 and N2 analysed with an isotope ratio mass spectrometer (Europa Tracermass, Oxford, UK). Isotope ratios are expressed in (notation as parts per thousand (‰) according to the following:

δX = [(Rsample/Rstandard) – 1]× 1000

where X is the δ13C or δ15N and R is the corresponding ratio 13C/12C or 15N/14N. Carbon and nitrogen isotope ratios are expressed relative to official Pee Dee Belemnite and Air standards.

Hair samples from six individuals were collected from each of the three mammal species late in October 1993 and in February 1994; and analysed using the same technique as for food samples.

Results

FAECAL ANALYSIS

Figure 1 compares the composition of faecal pellets of the Northern Bettong and Northern Brown Bandicoot at site A, and of the Northern Bettong and Rufous Bettong at site B. For the Northern Bettong, at both sites, fungus comprised most of the faecal matter. There was no significant seasonal variation in the contribution of fungus to faecal matter at site A (F3,24 = 1·43, P = 0·26), but at the (drier) site B, representation of fungus was highest in the wet season (F3,24 = 10·38, P < 0·0001). The fall in representation of fungus in the dry season was compensated by a significant increase in the representation of grass (F3,24 = 21·48, P < 0·0001).

Figure 1.

. Seasonal diets of the Northern Bettong, Bettongia tropica, the Rufous Bettong, Aepyprymnus rufescens, and the Northern Brown Bandicoot, Isoodon macrourus, as revealed by faecal analysis. Values are percentages (± SE) of fragments identified in faecal pellets, n = 8 for each point. Food categories are •, fungus, ▪, grass, ▴, lilies and forbs, ▾, roots and tubers and ◆, invertebrates.

Compared with the Northern Bettong, faeces of the Northern Brown Bandicoot at site A contained more grass (F1,48 = 17·69, P < 0·0001) and invertebrates (F1,48 = 165·98, P < 0·0001), and less fungus (F1,48 = 246·22, P < 0·0001) and roots and tubers (F1,48 = 4·88, P < 0·05). No significant seasonal variations were found for the contribution of invertebrates, grasses or fungi to bandicoot faeces. Faecal pellets of the Rufous Bettong differed from those of the Northern Bettong at site B by containing higher proportions of grass (F1,48 = 59·71, P < 0·0001) and lower proportions of fungus (F1,48 = 164·32, P < 0·0001). The representation of fungus in faeces of the Rufous Bettong was lower in the dry than in the wet season (F3,48 = 19·46, P < 0·0001), while the representation of grass and roots and tubers were both higher in the dry season (F3,48 = 13·17, P < 0·0001 and F3,48 = 18·82, P < 0·0001, respectively).

STABLE ISOTOPE ANALYSIS

The δ13C and δ15N values, with 95% confidence limits, for the major types of food eaten by bettongs and bandicoots are shown in Fig. 2. All food types had distinct isotope signatures. Grasses (including roots as well as stem and leaf) had the highest δ13C values and lilies and forbs had the lowest. Hypogeous fungi and insects had δ13C values significantly greater than lilies and forbs (F2,29 = 9·87, P < 0·01). Similarity in N isotope ratios was greater than for C; however, fungi had consistently higher δ15N values than grasses, forbs and lilies, and insects (F3,39 = 3·13, P < 0·05).

Figure 2.

. Mean stable isotope ratios of carbon and nitrogen (± 95% confidence limits) for hair samples, n = 6 for each point, collected from Northern Bettongs, Bettongia tropica–▪, site A, ▴, site B; Northern Brown Bandicoots, Isoodon macrourus–◆, site A; and Rufous Bettongs, Aepyprymnus rufescens–▾, site B. Samples were collected late in the dry season (October 1993) (open symbols) and mid wet season (February 1994) (closed symbols). Stable isotope ratios (95% confidence limits) are also given for known food items; hypogeous fungi, n = 10; grasses, n = 10; lilies and forbs, n = 8; and insects, n = 3 (pooled samples).

The combined stable isotope ratios of C and N in hair samples of the three species of marsupials were also discretely different (Fig. 2). The Northern Bettong was separated from the other two species by possessing the highest mean δ15N values, and lowest mean δ13C values. The Rufous Bettong differed from the Northern Bettong in having low to intermediate δ15N values and high δ13C values, and the Northern Brown Bandicoot was separated from the other two species by its low δ15N and intermediate δ13C values.

The position occupied by each mammal species on the plot of δ15N against δ13C, relative to the positions of the major food types, can be used to estimate the contribution of each food type to the N and C supply for each species. In making these estimates, it is necessary to allow for the fact that δ15N values in body tissue of mammals are typically enriched by 3–5‰ relative to values in food (Steele & Daniel 1978; DeNiro & Epstein 1981; Peterson & Fry 1987), and δ13C values are typically enriched by about 1‰ in mammalian hair (Tieszen et al. 1983). Taking these enrichment values into account, the δ15N value of the diet of the Northern Bettong must have been ≈ 8‰, very close to the value of 8·8‰ for hypogeous fungus, and much higher than the values of 3·2–3·9‰ for the other food types studied. This suggests that the Northern Bettong derived almost all its N from fungus. δ15N values of the Northern Brown Bandicoot suggest that it derived almost no N from fungus, and are consistent with invertebrates as the predominant source of N. The low δ15N value for the Rufous Bettong sample from the wet season also suggests that very little N for that sample came from fungus, but the higher value in the dry season suggests a contribution by fungus of approximately half its N supply.

In the case of the C supply for the two species of bettongs, invertebrates can be excluded as a significant source on account of their very low occurrence in faeces. This leaves grass, fungus and lilies and forbs as the three possible sources of carbon. Because these three sources could be mixed in various proportions to produce the C isotope ratio found for each mammal species, a simple and accurate measurement of the contribution from each source is not possible. However, it is clear that grass contributed significantly to the C budget for both bettongs because the C isotope ratio for grass was so extreme and the value for each bettong was clearly displaced towards it. The degree of this displacement provides a clear indication of the relative importance of grass to each species, and given some assumptions about the relative contribution of fungus vs lilies and forbs, this importance can be quantified.

The diet of the Rufous Bettong must have had a δ13C value of approximately – 15‰; this is close to the value of – 12‰ for grasses, and implies that ≈ 80% of its C was derived from grass (based on the assumption of roughly equal contribution from fungi and lilies and forbs, which the faecal analysis suggests is plausible). The maximum possible contribution of fungi to the C supply of the Rufous Bettong (i.e. assuming no contribution from lilies and forbs) is ≈ 25%. By the same reasoning, the contribution made by grass to the C supply for the Northern Bettong is ≈ 40%, on the assumption that the remainder is contributed equally by fungi and lilies and forbs. The maximum possible contribution from fungus would have been 67%. The significance of invertebrates in the diet of the Northern Brown Bandicoot complicates the picture further for that species, but the clear displacement of its δ13C towards grass suggests that a large proportion, ≈ 50%, of its C was derived from grass.

Discussion

The Northern Bettong clearly depended heavily on fungus for food. This was implied by analysis of faeces from wild-caught animals, and confirmed by stable isotope analysis which showed that most of the nitrogen, and perhaps more than half the carbon, assimilated into body tissue by this species was derived from hypogeous fungi. A high degree of fungus consumption has also been recorded (by faecal analysis) in other bettongs and potoroos (Johnson 1994b; Claridge & May 1994). Fungus made a modest and seasonally variable contribution to the diet of the Rufous Bettong, with faecal analysis indicating a peak in consumption during the wet season. However, stable isotope analysis suggested that Rufous Bettongs derive comparatively little of their dietary nitrogen from fungus. Both faecal analysis and stable isotope analysis indicated that mycophagy was of little significance to the diet of the Northern Brown Bandicoot. Our results agree with previous studies of bandicoot diets (e.g. Quin 1988; Claridge et al. 1991) in demonstrating the importance of invertebrates together with the marginal value of fungi for this species, but differ in showing that unlike bandicoots in southern forests (i.e. the Long Nosed Bandicoot, Perameles nasuta, and Southern Brown Bandicoot, I. obesulus), the Northern Brown Bandicoot eats grass to a substantial extent, especially during the dry season. Again, the significance of grass was confirmed by the stable isotope analysis, which suggested that around half the carbon assimilated into body tissue by Northern Brown Bandicoots was derived from grass.

MYCOPHAGY AND DIGESTIVE STRATEGIES IN SMALL MAMMALS

Most small herbivorous and omnivorous mammals have simple stomachs and, if they possess a capacity for microbial fermentation, it usually occurs in the hindgut (Cork 1994). The lack of extensive microbial fermentation in small mammals is thought to be a consequence of the scaling of gut capacity and energy requirements in relation to body size (Demment & Van Soest 1985): small mammals have low gut capacity relative to body mass; this limits their ability to retain ingesta in the gut for long enough to allow microbial fermentation to take place; and in addition their high mass-specific energy requirements demand a high rate of assimilation of energy. Therefore, small herbivorous mammals are obliged to feed selectively for high-quality items and often have mixed diets (Foley & Cork 1992). Under these circumstances, foregut fermentation is considered to be positively disadvantageous because it results in fermentation of highly digestible cell contents, which could more efficiently be absorbed directly across the small intestine (Van Soest 1982; Cork & Foley 1991). Hindgut fermentation avoids this cost, while still allowing some fermentation of less digestible material further along the gut (Cork 1994).

Cork & Kenagy (1989a) and Claridge & Cork (1994) show that sporocarps of hypogeous fungi have a high energy and nitrogen content, but that much of this energy and nitrogen is locked up in cell wall constituents and requires microbial processing before it can be used by mammals. Rat-kangaroos have an enlarged, elaborate forestomach in which microbial processing takes place (Hume 1982). Our results for the Northern Bettong, together with Claridge & Cork’s (1994) feeding trials on the Long-nosed Potoroo Potorous tridactylus, show that foregut fermentation allows rat-kangaroos to extract energy and nutrients from hypogeous fungi with a high degree of efficiency. Rat-kangaroos, most of which weigh between 1 and 2 kg, are much smaller than most foregut fermenters such as sheep, deer and kangaroos. Their nutritional ecology is therefore of general interest as it seems that the match of their digestive adaptations with the chemical and ecological properties of hypogeous fungi represents the very lowest threshold of body size at which foregut fermentation is a viable ecological strategy. Bandicoots have simple stomachs, with capacity for microbial fermentation in the hindgut only (Hume 1982), and in these respects resemble most other herbivorous/omnivorous small mammals. One of the supposed disadvantages of hindgut fermentation is the limitation it places on the assimilation of nitrogen that has been processed by microbes: foregut fermenters can absorb this microbial nitrogen across the small intestine, but hindgut fermenters, unless they are coprophagic or caecotrophic, lose much of it in faeces (Foley & Cork 1992). Rat-kangaroos possess a further advantage over many hindgut fermenters, which is that an important source of microbial protein comes from urea which is endogenously recycled from the blood into the forestomach (Wallis 1990). Thus, urea recycling enhances the efficiency of nitrogen use and enables the animals to feed on foods which are low in nitrogen. This may be an important factor governing the ecology of rat-kangaroos as nitrogen is often a limiting nutrient in the diets of herbivores, and its digestibility and retention is crucial (Loeb, Schwab & Demment 1991). In contrast to rat-kangaroos, many small hindgut-fermenting mammals are known to select diets that are high in protein and low in fibre, which reduces their dependence on microbial fermentation (Miller, Xia & Norrie 1990; Campbell & Macarthur 1996), although exceptions to this trend are known (e.g. Vickery et al. 1994). Such foods should be rich in a source of readily available nitrogen which can be readily absorbed without microbial processing. This may explain the limited use of fungus by bandicoots, and their extensive consumption of invertebrates.

The digestive system of the Rufous Bettong is similar to that of the Northern Bettong, yet in our study the Rufous Bettong used fungus to a much lesser extent than the Northern Bettong. One hypothesis that might explain differences in the use of fungi by the two species is that a partitioning of resources has enabled their coexistence. However, if this were the case, we might expect a substantial overlap in the distribution of the two species. This hypothesis is unlikely as the natural range of two species shows almost no overlap, despite similarities in their habitat requirements (Carpenter, Gillison & Winter 1994). A more likely hypothesis is suggested from differences in body size of the two species. Rufous Bettongs are, at 3 kg, about two to three times heavier than Northern Bettongs, and it may be that they lie closer to some threshold of body size above which foraging on hypogeous sporocarps ceases to be energetically efficient. Hypogeous sporocarps are small (usually less than 10-g dry mass) and the little that is known of their patterns of fruiting suggests that they occur patchily and often ephemerally (Johnson 1994a). They are, however, readily detectable because many of them produce powerful aromas when mature, and these aromas are demonstrably attractive to small mammals (Donaldson & Stoddart 1994). Therefore a mammal that passes close to a patch of hypogeous sporocarps will quickly become aware of it, but it may then have a large area to search before locating another patch. We speculate that a mammal of the Northern Bettong’s size might be able to satisfy its daily requirements by locating just one or a small number of patches of sporocarps, but a larger species such as the Rufous Bettong with greater absolute food requirements may need to search for a larger number of sporocarps, the patchy distribution of which would result in low foraging efficiency. This argument may explain why the Rufous Bettong devotes most of its feeding to grasses and roots, which are much more abundant and evenly distributed than hypogeous sporocarps. Confirmation of the hypothesis is needed with experiments comparing the availability, food preference and nutritional value of these food items to both species.

All three mammal species in this study relied on grass to a substantial degree. This is also of comparative interest, as our study animals are much smaller than typical grazers. There is extensive evidence that suggests that grazing is generally feasible for large herbivores only, because of the very extensive microbial processing that is required to convert grass into chemical forms that can be assimilated by mammals (Demment & Van Soest 1985). In response to these constraints, our study species appear to graze very selectively. In most samples, more than 70% of grass fragments belonged to one species, Cockatoo Grass Allopteropsis semilata. This species has a swollen stem base. Observation of grazed plants suggested that these stem bases were eaten in preference to leaves, and that the animals somehow separated coarse fibres from them before ingesting the less fibrous and liquid fractions. In the field it was common to see a chewed bolus of coarse fibres lying beside a grazed plant. The three mammals also seemed to select swollen stem bases when eating lilies, but apparently did not separate coarse fibres as they did with Cockatoo Grass.

The behaviour of ingesting just the soluble components of grass may have led to underestimation of its importance to the diet through faecal analysis. At the dry site (site B), both the Northern Bettong and Rufous Bettong showed a distinct reduction in stable isotope ratios for N and an increase in C ratios from the dry through to the wet season, indicating a decline in the use of fungus and an increase in the importance of grasses over this period. However, faecal analysis suggested the reverse trend was true for both species. An explanation for this might be that both bettongs take advantage of grasses when grass biomass increases substantially during the wet season, to maximize either energy, nutrient or water gain while minimizing overall fibre intake, which could potentially delay food retention times and digestion rates.

INTERPRETATION OF STABLE ISOTOPE VALUES IN FUNGI

Sporocarps of ectomycorrhizal fungi are known to be enriched in 15N relative to other forest ecosystem components (Gebauer & Dietrich 1993). Part of the explanation for the elevated δ15N values in fungi relative to plant items may lie in differences between the sources of nitrogen in ectomycorrhizal fungi and plants which possess either no mycorrhiza or vasicular arbuscular (VA) mycorrhizal symbionts (most grasses, forbs and lilies; Allen et al. 1995). Non-mycorrhizal plant roots and VA mycorrhizal fungi can absorb only inorganic forms of nitrogen from the soil, such as nitrate and ammonia. In contrast, ectomycorrhizal fungi can potentially utilize organic forms of nitrogen such as free amino acids, proteins and peptides (Abuzinadah & Read 1989; Read, Leake & Langdale 1989; Marschner & Dell 1994). Utilization of organic N occurs through the secretion of extracellular acid proteinases which degrade organic compounds into simple and polymeric N sources that can be assimilated by the fungi (Maijala, Fagerstedt & Raudaskoski 1991). Thus, ectomycorrhiza provide host plants with access to N sources that are normally unavailable to non-ectomycorrhizal infected plants. This may be an important factor influencing the δ15N values in ectomycorrhizal fungi as all the major pathways of N loss from the soil (denitrification, ammonia leaching and nitrate leaching) discriminate against 15N, leaving organic components of the soil enriched in 15N (Turner, Bergersen & Tantala 1983; Nadelhoffer & Fry 1988; Schulze, Chaplin & Gebauer 1994). Thus, differences in δ15N values between ectomycorrhizal fungi and plants with VA mycorrhiza or non-mycorrhizal plants probably reflect which components of the N pool are utilized. Similar patterns of 15N enrichment have been observed between saprophytic and autotrophic plants with differences due to the assimilation of organic sources of N which are available only to saprophytes (Virginia & Delwiche 1982).

Hogberg et al. (1996) found an 15N enrichment of between 1·7 and 2·0‰ in the fungal sheaths of ectomycorrhizal fungi when compared with non-mycorrhizal roots. Based on the higher percentage N of fungal sporocarps, they suggested that fungal material should be 3–11% more enriched than plant material. Differences between δ15N values in fungi and plant items observed in this study are consistent with this prediction.

It is not clear whether the large differences observed in δ15N between food items in this study are solely due to differences in sources of N uptake, or whether other processes, such as isotope fractionation during the assimilation of N, also influence δ15N values. An experimental study by Hogberg et al. (1994) using 15N-labelled NH4+ showed that differences in the δ15N of ectomycorrhizal and non-ectomycorrhizal plants were due to variations in N uptake, and that isotope fractionation appeared to be unimportant. Given this, it appears that δ15N values provide a useful marker for measuring the organic component and source of N uptake in ectomycorrhizal fungi in the field.

In addition to soil composition, the relative differences in 15N ratios of hypogeous fungi and non-ectomycorrizal plants will be influenced by features such as plant life form (Virginia & Delwiche 1982), vertical depth of plant roots in soil profile (Gebauer & Schulze 1991), ability of plants to fix nitrogen (Peoples et al. 1991) and the extent of mineralization processes (Read et al. 1989; Handley & Raven 1992) which are in turn influenced by soil properties, rainfall, fire and biotic interactions within the soil. We predict that high δ15N values in ectomycorrhizal fungi should occur in forests where the soil is rich in organic N such as many eucalypt and conifer forests (Marschner & Dell 1994). Many of these forests are nutrient limited (Schulze et al. 1994) because rates of decomposition are low because of either low temperatures or low moisture (Van Cleve et al. 1991). In such forests, particularly those with acid organic soils (Abuzinadah & Read 1989), ectomycorrhiza are expected to provide plants with an important means of N acquisition, and in addition, an important food source for mycophagous mammals.

Conclusion

Stable isotope analysis provided a useful quantitative tool for validating the seasonal importance of hypogeous fungus and grasses in the diets of three mycophagous marsupials. Values of stable isotopes largely supported the quantitative conclusions of faecal analysis, suggesting the two techniques can be used in combination to overcome several of the limitations inherent in each technique. The major advantage of stable isotope analysis over faecal analysis was that it enabled us to evaluate the nutritional importance of hypogeous fungi in terms of nitrogen and carbon assimilation, relative to other food groups. We believe that δ15N measurements offer a useful marker for future studies on the nutritional ecology of a wide range of mycophagous mammals.

Acknowledgements

This study was conducted with the financial support of the Queensland department of Environment and Heritage and the Wet Tropics Management Authority. We thank Michael Sutton for carrying out the stable isotope analysis. Dr John Winter provided valuable advice on the planning of the study, and he and Dr Andy Gillison (CSIRO, Atherton) assisted in the selection of study sites. We also thank Damian Milne and Lisa Pope for help with field work.

Ancillary