INTRODUCTION
Understanding how communities are structured in an ecosystem is a major theme in ecology (Hairston et al. Reference HAIRSTON, SMITH and SLOBODKIN1960, Odum Reference ODUM1969). Determining the trophic position of consumers is essential in understanding community structure and in predicting the effect of the loss of species on an ecosystem's properties (Duffy et al. Reference DUFFY, CARINALE, FRANCE, MCINTYRE, THEBAULT and LOREAU2007, Thebault & Loreau Reference THEBAULT and LOREAU2003). Over the decades, carbon (N) and nitrogen (N) stable isotope techniques have been used to examine the flow of energy and materials in communities and the trophic positions of consumers in food webs, because they have advantages that enable time-integrated estimation of trophic positions of a consumer (Fry Reference FRY2006, Post Reference POST2002). The approach is dependent on the assumption that the stable C isotope ratio (δ13C) of a consumer is almost identical to that of its diet, while the stable N isotope ratio (δ15N) is increased, about 3‰, compared with the diet (DeNiro & Epstein Reference DENIRO and EPSTEIN1978, Minagawa & Wada Reference MINAGAWA and WADA1984).
Because trophic enrichment (Δδ13C and Δδ15N), that is, the differences in the isotopic ratios between a consumer and its diet, is key for estimating trophic position (Post Reference POST2002), an increasing number of studies have explored factors influencing trophic enrichment using laboratory feeding experiments. These studies have revealed that trophic enrichment is influenced by various factors, such as the tissues and organs used for analyses (Tieszen et al. Reference TIESZEN, BOUTTON, TESDAHL and SLADE1983), quality of the diet (Oelbermann & Scheu Reference OELBERMANN and SCHEU2002), developmental stages (Doi et al. Reference DOI, KIKUCHI, TAKAGI and SHIKANO2007), and biochemical form of N excretion and feeding habits (McCutchan et al. Reference MCCUTCHAN, LEWIS, KENDALL and MCGRATH2003, Vanderklift & Ponsard Reference VANDERKLIFT and PONSARD2003). The stable isotope technique has been applied successfully to studies of food webs primarily in aquatic ecosystems, such as marine and lacustrine systems (Grey & Jones Reference GREY and JONES2001, Yoshii et al. Reference YOSHII, MELNIK, TIMOSHKIN, BONDARENKO, ANOSHKO, YOSHIOKA and WADA1999).
In terrestrial ecosystems, the technique has been used to examine the feeding habits of a single taxon of organisms, such as earthworms (Briones et al. Reference BRIONES, BOL, SLEEP, SAMPEDRO and ALLEN1999), bats (Herrera et al. Reference HERRERA, GUTIERRES, HOBSON, ALTUBE, GIAZ and SANCHEZ-CORDERO2002), and ants (Davidson et al. Reference DAVIDSON, COOK, SNELLING and CHUA2003), trophic interactions in some groups of invertebrates in experimental plots (Ostrom et al. Reference OSTROM, COLUNGA-GARCIA and GAGE1997, Wise et al. Reference WISE, MOLDENHAUER and HALAJ2006), and food webs among soil invertebrates (Illig et al. Reference ILLIG, LANGEL, NORTON, SCHEU and MARAUN2005, Ponsard & Arditi Reference PONSARD and ARDITI2000, Scheu & Falca Reference SCHEU and FALCA2000, Schmidt et al. Reference SCHMIDT, CURRY, DYCKMANS, ROTA and SCRIMGEOUR2004) and vertebrates (Ambrose & Deniro Reference AMBROSE and DENIRO1986, Bocherens & Drucker Reference BOCHERENS and DRUCKER2003). Meanwhile, it is well recognized that terrestrial ecosystems have characteristics distinct from aquatic ecosystems, such as the dominance of the detrital (below-ground) food webs over the grazing (above-ground) food webs, in terms of energy and material flows (Swift et al. Reference SWIFT, HEAL and ANDERSON1979, Wardle Reference WARDLE2002). Additionally, earlier studies revealed some mechanisms associated with the below-ground system that can influence the isotopic signatures of soil organisms, such as 13C enrichment in basidiomycetes (Hobbie et al. Reference HOBBIE, MACKO and SHUGART1999, Kohzu et al. Reference KOHZU, MIYAJIMA, TATEISHI, WATANABE, TAKAHASHI and WADA2005). In fact, it is known that detritivores tend to have higher δ13C than plant substrates (Ponsard & Arditi Reference PONSARD and ARDITI2000), probably because they selectively feed on the fungal tissues and the 13C-enriched components of the plant materials, such as cellulose (Pollierer et al. Reference POLLIERER, LANGEL, SCHEU and MARAUN2009). In the above-ground system in tropical and temperate forests, it has been reported that δ13C in tree leaves shows a vertically increasing trend, from the forest floor to the canopy (Garten & Taylor Reference GARTEN and TAYLOR1992, Hanba et al. Reference HANBA, MORI, LEI, KOIKE and WADA1997), which could influence δ13C in consumers. Despite these characteristics of terrestrial ecosystems, few studies have examined how terrestrial food webs, including various organisms above and below ground, can be depicted using C and N isotopic signatures.
We measured δ13C and δ15N of diverse organisms (14 orders, ≥24 families), classified into four trophic groups (detritivores, herbivores, omnivores, and general predators), in a tropical rain forest in Sarawak, Malaysia. We hypothesize that (1) the general predators are similar in δ13C signatures to the detritivores, because of the significance of the below-ground system; (2) the herbivores are more similar in δ13C signatures to the canopy leaves than the understorey leaves, because the most primary production occurs in the canopy layers (Osada et al. Reference OSADA, TAKEDA, FURUKAWA and AWANG2001).
MATERIALS AND METHODS
Study site and sample collection
Sampling was conducted in Lambir National Park (4°2′N, 113°5′E, 20–150 m asl), Sarawak, Malaysia, mostly in June 2005. Additional samples were collected between November 2004 and January 2006. Mean annual rainfall is about 2700 mm, with no distinct dry season (Nakagawa et al. Reference NAKAGAWA, TANAKA, NAKASHIZUKA, OHKUBO, KATO, MAEDA, SATO, MIGUCHI, NAGAMASU, OGINO, TEO, HAMID and LEE2000). The consumers used in this study included various taxa (14 orders, ≥24 families) and were classified into taxonomic groups at various levels (Appendix 1). The taxonomic groups were assigned a priori into four trophic groups, i.e. detritivores, herbivores, omnivores and predators, according to generally known feeding habits (Davies Reference DAVIES1988, Nakagawa et al. Reference NAKAGAWA, HYODO and NAKASHIZUKA2007, Payne et al. Reference PAYNE, FRANCIS and PHILLIPS1985, Price Reference PRICE1997, Smythies Reference SMYTHIES1999). They were further classified into vertebrates and invertebrates. The invertebrates were collected haphazardly from the forest floor. Some invertebrates were also collected from canopy layers using a walkway and a crane tower system (Yumoto & Nakashizuka Reference YUMOTO, NAKASHIZUKA, Roubik, Sakai and Hamid2005). The vertebrates were collected using traps: wire-mesh traps (15 × 12 × 30 cm) with bait for small mammals, such as rats, mice, squirrels and tree shrews (Nakagawa et al. Reference NAKAGAWA, HYODO and NAKASHIZUKA2007), and mist nets for bats and birds (Fukuda et al. Reference FUKUDA, TISEN, MOMOSE and SAKAI2009, Kamoi Reference KAMOI2007). The clipped toes of rats, mice, squirrels and tree shrews, the hairs of bats, and the feathers of birds were used for stable isotope analyses. The δ13C and δ15N values for small mammals have already been reported (Nakagawa et al. Reference NAKAGAWA, HYODO and NAKASHIZUKA2007). Plant materials and soils were also collected for comparison. Fresh leaves were randomly sampled from trees at a height of less than 2 m (understorey leaves) and from the canopy layers (canopy leaves). Litter, dead wood and soil (0–10 cm depth) were also sampled randomly from the forest floor.
The samples were dried at 60 °C for 24 h. Collected invertebrates were kept in a freezer for 24 h to terminate their activity, before they were dried in a drying oven. The whole bodies or legs of invertebrates and toes of the small mammals were ground into powder using a mortar and pestle prior to the analyses. The leaves, litter and dead wood materials were ground using a ball mill. The soil samples were sieved with a 2-mm mesh and then treated with 0.5 M HCl overnight to remove inorganic carbon.
Stable C and N isotope analyses
For stable C and N isotope analyses, the samples were placed in tin capsules. Stable C and N isotope ratios were measured using a mass spectrometer (Deltaplus XP, Germany), coupled with an elemental analyser. The precision of the on-line procedure was better than ± 0.2‰ for both isotope ratios. The natural abundances of 13C and 15N are expressed in per mil (‰) deviation from international standards: δ13C or δ15N = (R sample/R standard –1) × 1000, where R in δ13C or δ15N is 13C/12C or 15N/14N, respectively. Pee Dee belemnite and atmospheric nitrogen were used as the international standards for carbon and nitrogen, respectively.
Statistical analyses
To examine differences in each isotope ratio among the plant and soil samples (canopy and understorey leaves, litter, soils), we used analysis of variance (ANOVA), with the isotope ratio as the dependent variable and the sample type as the independent variable. The relationship between average δ13C and δ15N values of the consumers in each taxonomic group was examined using a simple linear regression. To test if the isotopic compositions of consumers differ between the canopy and the understorey layers, we used t-test for the taxonomic groups that could be collected in both layers (i.e. leaf beetles, homopterans and spiders). In addition, to examine differences in isotopic signatures among the trophic groups, we used a linear mixed model (Grafen & Hails Reference GRAFEN and HAILS2002). Each isotope ratio was used as the dependent variable, and the trophic group, the taxon at the level of vertebrates vs. invertebrates (taxon (VR vs. IN)) (as fixed effects) and the taxonomic group (as a random effect) were used as the independent variables. Taxon (VR vs. IN) was nested within trophic group, and taxonomic group was nested within both trophic group and taxon (VR vs. IN). In the model, the least-squares means, which are within-group means appropriately adjusted for the other effects, were calculated. We used the Tukey–Kramer HSD test to examine differences among trophic groups. The statistical analyses were performed using JMP statistical software (version. 5.1.2 for Macintosh, SAS Institute, Cary, NC, USA).
RESULTS
δ13C and δ15N in plant materials and soil organic matter
δ13C and δ15N differed among plant materials and soil organic matter (F4,66 = 22.9, P < 0.0001 for δ13C; F4,66 = 32.4, P < 0.0001 for δ15N; Figure 1a, b). Soil organic matter and dead wood had significantly higher δ13C than litter. The δ13C of canopy leaves was intermediate between them. Understorey leaves had significantly lower δ13C than the rest. Soil organic matter had significantly higher δ15N signatures than canopy and understorey leaves or the litter, which were also more enriched than dead wood.
δ13C (a) and δ15N (b) (mean ± SE) of canopy and understorey leaves, litter, and woody litter in Lambir National Park, Sarawak, Malaysia. Means marked with the same letter do not differ, according to the Tukey–Kramer HSD test at P = 0.05 following ANOVA. CL, canopy leaves; UL, understorey leaves; L, litter; WL, woody litter; Soil, soil (0–10 cm).
δ13C and δ15N of consumers and the difference between canopy and understorey layers
δ13C and δ15N of consumers varied from –32.3‰ (average value of scale insect) to –23.0‰ (white-rumped shama), and from –1.6‰ (average value of fungus-growing termite) to 7.2‰ (white-rumped shama), respectively (Figure 2, Appendix 1). There was a significant relationship between average values of δ13C and δ15N in each taxonomic group (r2 = 0.445, n = 39, P < 0.001).
Scatter plot of δ13C and δ15N of the taxonomic groups, leaves, litter, and soil (mean ± SE) in Lambir National Park, Sarawak, Malaysia. The taxonomic groups examined are listed in Appendix 1. The δ13C and δ15N of canopy and understorey leaves, litter, woody litter and soil are also presented for comparison.
δ13C did not differ significantly between the two layers for the leaf beetles (canopy, –26.5‰ ± 0.3‰, n = 3; understorey, –27.5‰ ± 1.7‰, n = 6), and spiders (canopy, –26.8‰ ± 0.9‰, n = 13; understorey, –27.2‰ ± 1.4‰. n = 15). However, there was a significant difference in δ13C of homopterans between the two layers (canopy, –26.7‰ ± 1.3‰, n = 10; understorey, –29.0‰ ± 2.6‰, n = 7, P < 0.05). δ15N did not differ for leaf beetles (canopy, 1.7‰ ± 0.6‰, n = 3; understorey, 1.2‰ ± 1.4‰. n = 6), homopterans (canopy, 1.7‰ ± 2.5‰, n = 10; understorey, 1.9‰ ± 2.3‰. n = 7) and spiders (canopy, 3.0‰ ± 1.8‰, n = 13; understorey, 3.9; ± 1.7‰. n = 15).
δ13C and δ15N of consumers of four trophic groups
δ13C of consumers differed between trophic groups (F3,34 = 3.83, P < 0.05, Figure 3a). Predators had higher δ13C (–25.4‰ ± 0.5‰; least-squares mean ± SE) than herbivores (–27.8‰ ± 0.6‰). Omnivores (–25.7‰ ± 0.7‰) and detritivores (–26.6‰ ± 0.7‰) had intermediate levels. Taxon (VR vs. IN) had significant effects on δ13C (F3,34 = 5.36, P < 0.01), and vertebrates had higher δ13C than invertebrates in the same trophic group.
δ13C (a) and δ15N (b) (least-squares mean ± SE) of trophic groups in Lambir National Park, Sarawak, Malaysia. DT, detritivores; HB, herbivores; OM, omnivores; PD, predators; VR, vertebrates; IN, invertebrates. The least square means of the trophic groups (av) are indicated by closed circles, and those of the invertebrates and vertebrates of each trophic group by open circles. Note that the least-squares mean of the detritivores is identical to that of the invertebrates, because there were no detritivorous vertebrates. The least-squares means of each trophic group with the same letter do not differ, according to the Tukey–Kramer HSD test at P = 0.05, following analysis of a linear mixed model.
δ15N of consumers differed between trophic groups (F3,34 = 9.38, P < 0.001; Figure 3b). Predators (5.1‰ ± 0.5‰) had significantly higher δ15N than omnivores (2.5‰ ± 0.7‰), herbivores (2.1‰ ± 0.6‰), and detritivores (1.4‰ ± 0.7 ‰). Taxon (VR vs. IN) also influenced δ15N (F3,34 = 4.47, P < 0.001), and vertebrates had higher δ15N than invertebrates in the same trophic group. Consequently, predators showed elevated δ13C and δ15N (Figure 2, 3a, b). Comparison of δ13C and δ15N in consumers and plant materials showed that the consumers had higher δ13C and δ15N than canopy leaves and the herbivores showed δ13C similar to the canopy leaves rather than the understorey leaves.
DISCUSSION
We demonstrated that there is a significant correlation between δ13C and δ15N of consumers for each taxonomic group. The values differed significantly among the trophic groups and the values in vertebrates were significantly higher than those in invertebrates. Predators had δ13C closer to those of detritivores and omnivores, rather than that of herbivores. It is well recognized that δ13C can be used to trace energy and material flows in food webs, because the C isotopic signature does not change through trophic interactions (DeNiro & Epstein Reference DENIRO and EPSTEIN1978, McCutchan et al. Reference MCCUTCHAN, LEWIS, KENDALL and MCGRATH2003). In addition, most primary production enters the below-ground system without being consumed by herbivores in the above-ground system (Swift et al. Reference SWIFT, HEAL and ANDERSON1979, Wardle Reference WARDLE2002). In this light, the present results suggest that most of the predators examined in this study are more dependent on detritivores than on herbivores. The omnivores may also depend to some extent on detritivores as a food source. In temperate regions, it is known that predators can switch temporally from detritivores to herbivores as prey according to their relative abundance and availability (Birkhofer et al. Reference BIRKHOFER, WISE and SCHEU2008). This is unlikely the case in this study, because central South-East Asia, including the study site, is one of the wettest and most aseasonal climates of any tropical region (Whitmore Reference WHITMORE1984).
Although the significance of detritivores in supporting predators in terrestrial ecosystems has been discussed (Polis & Strong Reference POLIS and STRONG1996), few studies have actually examined it: field observations in a tropical paddy-field (Settle et al. Reference SETTLE, ARIAWAN, ASTUTI, CAHYANA, HAKIM, HINDAYANA, LESTARI, PAJARNINGSIH and SARTANTO1996), in tundra ecosystems (Oksanen Reference OKSANEN, Gange and Brown1997), and experimental manipulations in agroecosystems (Halaj & Wise Reference HALAJ and WISE2002) and forest ecosystems (Miyashita et al. Reference MIYASHITA, TAKADA and SHIMAZAKI2003). Our results support the fundamental importance of the linkage of below- and above-ground food webs in terrestrial ecosystems (Scheu Reference SCHEU2001). Based on δ13C values, an earlier study suggested that detritivores played an important role in supporting predators in a temperate rice field during the rice growing season (Park & Lee Reference PARK and LEE2006). Thus, we propose that δ13C of predators can be used as an indicator of the dependency of predators on below-ground food webs.
The enrichment of 13C in detritivores relative to plant substrates has been reported in a single taxon, as well as in entire soil food webs (Hishi et al. Reference HISHI, HYODO, SAITOH and TAKEDA2007, Ponsard & Arditi Reference PONSARD and ARDITI2000, Spain & Reddell Reference SPAIN and REDDELL1996, Tayasu et al. Reference TAYASU, ABE, EGGLETON and BIGNELL1997). Although the underlying mechanisms remain unclear, this may reflect the increase in δ13C in soil organic matter through microbial activity, as found for δ15N (Nadelhoffer & Fry Reference NADELHOFFER and FRY1988) and the detritivores’ selective feeding on the organic matter enriched in 13C (Pollierer et al. Reference POLLIERER, LANGEL, SCHEU and MARAUN2009). In particular, it has been reported that basidiomycetous fungi, which play an important role in decomposition processes (Swift et al. Reference SWIFT, HEAL and ANDERSON1979), are about 3‰ more enriched in 13C than their substrates (Hobbie et al. Reference HOBBIE, MACKO and SHUGART1999, Kohzu et al. Reference KOHZU, MIYAJIMA, TATEISHI, WATANABE, TAKAHASHI and WADA2005). Indeed, high δ13C was observed in a fungus-growing termite (Macrotermes malaccensis), which depends on symbiotic basidiomycetes as a food source to some extent (Hyodo et al. Reference HYODO, TAYASU, INOUE, AZUMA and KUDO2003), as well as a fungus beetle (Appendix 1).
Canopy leaves had higher δ13C than understorey leaves, and herbivores had δ13C similar to those of canopy leaves, rather than understorey leaves (Figure 3), suggesting that most herbivores examined were sustained by production in the canopy layers. This is likely consistent with most of the litter being yielded by the canopy layers in tropical forests (Osada et al. Reference OSADA, TAKEDA, FURUKAWA and AWANG2001). The increasing trend in δ13C in leaves with height in the forest is caused by various factors, such as the source of CO2, coming either from the atmosphere (about –8‰) or from the forest floor by soil respiration (about –26‰), and differences in stomatal conductance, due to water and light conditions (Dawson et al. Reference DAWSON, MAMBELLI, PLAMBOECK, TEMPLER and TU2002).
Note that homopterans were more enriched in 13C in the canopy layers than in the understorey, which likely corresponds to the higher δ13C of the canopy leaves, although the difference in δ13C between canopy and understorey layers was not observed in the leaf beetles and spiders. In addition, the leaf beetles, weevils and homopterans were relatively enriched in 13C compared to the other invertebrate herbivores. The high δ13C of the predators might be explained by their use of these 13C-enriched herbivores. The use of radiocarbon (14C) should allow the quantification of the dependency of predators on detritivores through the measurement of the 14C contents of detritivores and herbivores; detritivores should have older carbon than herbivores (Hyodo et al. Reference HYODO, TAYASU and WADA2006). Also, it should enable us to estimate the dependency of herbivores on canopy leaves instead of understorey leaves, because canopy leaves should have the same 14C concentration as that of atmospheric CO2 in the year of sampling (i.e. 0 y), whereas the 14C concentration of understorey leaves should indicate a mix of CO2 from the atmosphere and that respired from soil organic matter (>0 y).
As for the δ15N in consumers, the value increased as trophic level increased: predators had higher δ15N than detritivores, as well as herbivores and omnivores. The δ15N of the vertebrate predator, tree shrews (Tupaia spp.), which have previously been reported in this site (Nakagawa et al. Reference NAKAGAWA, HYODO and NAKASHIZUKA2007), were comparable to the other predatory mammals, such as bats. The average Δδ15N observed between predators and detritivores was consistent with a value reported previously, 3.4‰ ± 1.1‰ (Minagawa & Wada Reference MINAGAWA and WADA1984). Meanwhile, earlier studies revealed that δ15N in soil invertebrates also increased along with humification of diets (i.e. more humified and decomposed organic matter) (Hishi et al. Reference HISHI, HYODO, SAITOH and TAKEDA2007, Hyodo et al. Reference HYODO, TAYASU, KONATE, TONDOH, LAVELLE and WADA2008, Tayasu et al. Reference TAYASU, ABE, EGGLETON and BIGNELL1997). Indeed, the detritivores that feed on humified organic matter, such as soil-feeding termites (D. nemorosus) and a rhinoceros beetle, had high δ15N (6.7‰ and 4.7‰, respectively; Appendix 1), which are comparable to the ratios in the predators. Thus, predation on detritivores feeding on humified diets can increase the δ15N in such predators (and omnivores). This might explain the high δ15N in vertebrate predators and omnivores. On the other hand, invertebrate predators might depend not on such detritivores, but on those feeding on less humified organic matter, like surface litter layers. These possibilities could be tested by measuring the diet ages of the consumers based on the radiocarbon analysis.
In addition to trophic group, taxon (VR vs. IN) had a significant influence on δ13C and δ15N. The enrichment in 13C in vertebrates should reflect the fact that the tissues used for the isotopic measurements tended to be more enriched in 13C, by ~2‰, than the muscles (DeNiro & Epstein Reference DENIRO and EPSTEIN1978, Hobson et al. Reference HOBSON, SCHELL, RENOUF and NOSEWORTHY1996, Tieszen & Boutton Reference TIESZEN, BOUTTON, Rundel, Ehleringer and Nagy1989). Vertebrates also had higher δ15N than invertebrates of the same trophic group. The likely explanation is that vertebrates used food sources with higher δ15N than did invertebrates of the same trophic group. For example, vertebrate omnivores may have consumed a higher proportion of prey to plant materials, compared with invertebrates. Vertebrate predators may have preyed on invertebrate predators, as well as herbivores and omnivores. In particular, vertebrates may have used detritivores that feed on humified organic material to a higher extent than the invertebrates, as discussed above. Another explanation may be differences in tissues used for the δ15N measurements between the vertebrates (feathers for birds, hairs for bats, and toe-tips for small mammals) and the invertebrates (whole bodies or body parts), as well as in physiological processes between vertebrates and invertebrates, because several factors influencing Δδ15N are known to be related to taxonomic identity (Vanderklift & Ponsard Reference VANDERKLIFT and PONSARD2003).
Overall, based on the δ13C and δ15N in consumers, we suggest that the predators examined in this study depended largely on detritivores for their energy and materials. Thus, general predators may interconnect the above- and below-ground food webs. The predation on the detrivores and the further 13C enrichment in the hair and feather tissues of vertebrate predators likely account for the observed correlation between δ13C and δ15N in each taxonomic group. We also suggest that herbivores derive most of their C from forest canopy layers. In this study, we could not examine important consumers in tropical food webs, such as amphibians, reptiles and large invertebrates (e.g. scorpions), the latter of which may feed on vertebrates (Kupfer et al. Reference KUPFER, LANGEL, SCHEU, HIMSTEDT and MARAUN2006, Reagan et al. Reference REAGAN, CAMILO, WAIDE, Reagan and Waide1996). In addition, it is well known that the isotopic signatures of consumers within a trophic group and even within a taxonomic group can vary (Ponsard & Arditi Reference PONSARD and ARDITI2000, Scheu & Falca Reference SCHEU and FALCA2000), For example, it has recently been reported that δ15N ranges between 7.5‰ to 16.5‰ for termite species of the genus Anoplotermes in a French Guianan forest reserve (Bourguignon et al. Reference BOURGUIGNON, SABOTNIK, LEPOINT, MARTIN and ROISIN2009). Regarding the invertebrate herbivores, vascular and non-vascular feeders might have different N isotopic signatures due to the different Δδ15N (McCutchan et al. Reference MCCUTCHAN, LEWIS, KENDALL and MCGRATH2003, Spence & Rosentheim Reference SPENCE and ROSENTHEIM2005), although they did not clearly differ in this study. These factors might affect the present conclusion. As such, further studies are necessary to examine more taxa with higher taxonomic resolution to confirm the isotopic pattern found in this study.
ACKNOWLEDGEMENTS
We thank Josef Kendawang (Forest Department, Sarawak) and Lucy Chong (Sarawak Forestry Corporation) for permission for research in Sarawak. We also thank Tohru Nakashizuka (Tohoku University) and Het Kaliang (Sarawak Forestry Corporation) for kind arrangement to conduct the field sampling, Koji Kawamura (Centre INRA d'Angers) for statistical advice and Takuo Hishi (Kyusyu University), Alexander Kupfer and an anonymous reviewer for helpful comments. This study was supported by Research Institute for Humanity and Nature, Japan (P3–1, P2–2 and 3–5), by Grant-in-Aids (no. 17405006 to T.I., and no. 16405009 to Y.T.) and partly by Special Coordination funds for Promoting Sciences and Technology from the Japanese Ministry of Education, Science, and Culture. F.H., T.M. and M.N. were supported by the Research Fellowships of the Japan Society for the Promotion of Science (JSPS) for Young Scientists.
δ13C and δ15N of various organisms in Lambir National Park, Sarawak, Malaysia. n = number of replicates examined for isotopic compositions.
