Niche differentiation among neotropical soldierless soil-feeding termites revealed by stable isotope ratios

Abstract

Termites represent one of the most abundant belowground animal taxa in tropical rainforests, where their species richness is much higher than in any other ecosystem. This high diversity in soil ecosystems is however difficult to explain by classical Hutchinsonian niche theory, as there is little evidence for spatial or temporal separation between species. Using δ¹³C and δ¹⁵N isotopic ratios, we tested if resource partitioning along the humification gradient occurs in neotropical soldierless termites of the Anoplotermes-group. Two distinct sites were investigated to check if interspecific differences are transposable between sites. Significant differences in δ¹⁵N were found between species of the Anoplotermes-group. Although some species displayed higher intersite δ¹⁵N variation than others, species-average δ¹⁵N values for both sites were highly correlated, showing that sympatric soldierless soil-feeding termites feed on distinct components of the soil. Our data also suggest that some species are more likely to shift along this gradient than others, in response to overall habitat conditions or to the presence of competitors. Feeding niche differentiation can therefore account for the high species richness and diversity of soldierless soil-feeding termites in neotropical rainforests.

Introduction

Termites are major decomposers of organic matter in tropical ecosystems (Holt and Lepage, 2000), where they constitute a large part of the animal biomass (Fittkau and Klinge, 1973, Martius, 1994, Eggleton et al., 1996). Termite species have been distributed among “feeding groups” or “functional taxonomic groups”, mainly based on the microhabitat in which foragers are found and observations of feeding habits and gut content (Eggleton et al., 1996, Bignell and Eggleton, 2000, Donovan et al., 2001, Gathorne-Hardy et al., 2002, Davies et al., 2003a). For instance, Donovan et al. (2001) recognized four feeding groups corresponding to an increasing humification of the feeding substrate. Wood feeders constitute groups I (non-Termitidae) and II (Termitidae), whereas soil-feeding Termitidae constitute groups III and IV. Anatomical features (sclerotized enteric valve and absence of mandible ridges) distinguish group IV from group III. Presumably, group III species feed at the soil–wood interface, whereas group IV species feed on more heavily mineralised soil (Donovan et al., 2001).

Soil-feeding taxa constitute a major part of termite species richness in tropical rainforests, especially in Africa and South America (review in Davies et al., 2003a). In the Neotropics, the Anoplotermes-group (subfamily Apicotermitinae), characterized by the absence of soldiers, is especially abundant and diversified. This group comprised, respectively, 34%, 45% and 30% of all termite species (33%, 35% and 34% of all encounters) recorded by ground transect sampling in a French Guiana, a Panamanian and a Central Amazonian rainforest (Davies et al., 2003b, Roisin et al., 2006, Ackerman et al., 2009). However, in spite of their richness and abundance, their biology remains poorly known. All are presumably interface- or soil-feeders (feeding group III or IV), but their precise food requirements remain largely unknown. Laboratory studies are hindered by the extreme difficulty in maintaining these termites under artificial conditions.

Usually, stable carbon isotope ratios (¹³C/¹²C) of animals are similar or slightly enriched compared to their diets, whereas stable nitrogen isotope ratios (¹⁵N/¹⁴N) (hereafter expressed as δ¹³C and δ¹⁵N, see Methods) increase by 1.5–3‰ for each trophic level (McCutchan et al., 2003, Vanderklift and Ponsard, 2003, Hood-Nowotny and Knols, 2007). The picture is more complex in forest soils, where humification gradients are characterized by a gradual enrichment in ¹⁵N resulting from a series of interrelated mechanisms, still incompletely understood, involving microorganisms and plant roots (Högberg, 1997, Kramer et al., 2003, Dijkstra et al., 2008). Therefore, the older and deeper the organic matter, the higher its δ¹⁵N value (Piccolo et al., 1996, Martinelli et al., 1999, Krull et al., 2002, Billings and Richter, 2006, Hyodo et al., 2006, Hyodo et al., 2008). δ¹³C values also increase down the soil profile but to a lesser extent and in a less consistent way (Tiunov, 2007). The low degree of enrichment in ¹³C across trophic levels and down the soil profile makes the δ¹³C ratio less useful than the δ¹⁵N to infer the trophic structure of soil communities (Ponsard and Arditi, 2000), but the δ¹³C may still help in distinguishing between trophic chains based on C3 versus C4 plants (Smith and Epstein, 1971, Boutton et al., 1983).

Stable isotope studies of termites revealed that δ¹⁵N increases along the sequence wood → soil–wood interface → soil feeders (Tayasu et al., 1997, Tayasu et al., 1998, Tayasu et al., 2002a). In the Mbalmayo forest, Cameroon, δ¹⁵N values are below 8 for wood feeders, but above 12 for mineral soil feeders (Tayasu et al., 1997). The same tendency is present in termites of the Darwin area, Australia, but with globally lower δ¹⁵N values: wood feeders below 3, soil feeders above 5.5, interface feeders in between (Tayasu et al., 1998). In both sites, the δ¹⁵N of wood-feeding termites is almost equal to the average δ¹⁵N of their food source. Stable carbon isotopes show lesser discriminatory value than nitrogen, but δ¹³C may separate grass- from wood-feeders (Tayasu et al., 1998, Tayasu et al., 2002b). These results are consistent with the classification into feeding groups proposed by Donovan et al. (2001), although there is ample overlap in δ¹⁵N values between group I (non-Termitidae, all wood-feeding) and group II (wood-feeding Termitidae), and between group III and group IV (alleged interface- and soil-feeding Termitidae) (Eggleton and Tayasu, 2001). Although these studies encompassed a broad array of genera, they provided very few data on intrageneric variation and possible species-specific patterns. Nevertheless, they demonstrated the suitability of stable isotope analysis to situate termite niches in a gradient of humification.

The high diversity of decomposer animals in belowground ecosystems is difficult to explain by classical Hutchinsonian niche theory, as there is little evidence for spatial or temporal separation between species. A recently proposed mechanism to explain the diversity of detritivores is resource partitioning along the decomposition gradients of their food sources, revealed by δ¹⁵N ratios (oribatid mites: Schneider et al., 2004; springtails: Chahartaghi et al., 2005, Hishi et al., 2007). Likewise, we hypothesized here that the high species richness of the Anoplotermes-group in neotropical rainforests could be explained by a differentiation of their feeding niches, which should be revealed by species-specific δ¹⁵N values. A special feature of termites is that isotope effects (the difference in δ¹⁵N between consumers and their diet) seem very variable, ranging from −1.6 to +8.8‰ (Tayasu et al., 1997). These authors suggested that fixation of atmospheric N₂ pulls this effect down for wood-feeding species, while selection of specific food particles from the soil pushes it up for soil feeders. As a consequence, the range of δ¹⁵N values within a termite assemblage may greatly exceed that of their macroscopically identifiable food sources.

Considering that different sites are likely to display different overall isotopic signatures (Piccolo et al., 1996, Martinelli et al., 1999, Powers and Schlesinger, 2002), we carried out our investigations in two distinct forests. Even though the isotopic baseline might be different and influence the isotopic composition of soil-feeding termites, we expected that interspecific differences should be transposable between sites if trophic niche differentiation is an important structuring factor for the termite assemblage.