Lotic food webs mainly derive their energy from allochthonous and autochthonous (algae) organic sources (e.g. Vannote et al., 1980; Thorp and Delong, 1994; Ishikawa et al., 2010). Food webs in adjacent habitats and ecosystems may be strongly linked by fluxes of nutrients, detritus, or organisms (Polis et al., 1997). The spatial and temporal scales of connections between food webs were known only fragmentary knowledge. Such information is crucial to understanding the processes that control the dynamics of local populations, communities, and ecosystems, as well as the fate of fluxes of materials and individuals across habitat, ecosystem, and landscape boundaries (Polis et al., 1997; Laurance et al., 2001). The mobility of limiting nutrients, organic matter, and organism characteristic of lotic environments suggests that such linkages are especially strong in river food webs. For example, detrital carbon inputs from terrestrial to stream food webs are important in many stream ecosystems (Hynes, 1975). Organic carbon sources for lotic food webs may be derived from multiple sources, including local algal production and particulate and dissolved organic matter transported from upstream habitats. Consequently, the spatial scales of trophic interactions in rivers are largely unknown (Cooper et al., 1997). Furthermore, relatively few studies have been conducted on food webs related to spring ecosystems. Spring streams are closed ecosystems and arise from seepage areas; in addition, primary production is relatively high in these small spring streams, which are exposed to sunlight, algae-rich, and contain adequate concentrations of nitrogen and phosphate (Allan, 1995).

The aims of this study were to identify the trophic pathways from organic matter to macroinvertebrates in terms of consumers, using stable isotopes (carbon and nitrogen); and to characterize the food webs in an intermittent stream system of the Echi River in Japan.

The mean δ¹³C and δ¹⁵N values of macroinvertebrates (ranging from -28.7 to -19.4‰ and 2.3 to 9.9‰, respectively) differed among stations and species (Table 1). The consumers such as Psephenoides spp., Ecdyonurus levis, Ephemera strigata, Rhyacophila nigrocephala, and Davidius lunatus were common to both ES 1 and ES 2. The mean δ¹³C values of these species showed gradually increasing pattern from ES 1 to ES 2 (Psephenoides spp.: -23.1 to -21.2‰, Ecdyonurus levis: -28.7 to -25.1‰, Ephemera strigata: -25.6 to -23.3‰, R. nigrocephala: -23.8 to -21.1‰, and Davidius lunatus: -24.5 to -23.9‰). Although the mean δ¹⁵N values of Psephenoides spp., Ecdyonurus levis, and Ephemera strigata did not differ among stations (ranging from 2.3 to 2.7‰, 4.5 to 4.9‰, and 5.2 to 5.9‰, respectively), those of R. nigrocephala and Davidius lunatus displayed increasing from ES 1 to ES 2 (5.4 to 6.9‰ and 5.2 to 6.1‰, respectively). Psephenoides spp. and were also observed at ES 4. The mean δ¹³C values of Psephenoides spp. at ES 2 were relatively higher, whilst those of R. nigrocephala at ES 4 were relatively higher than at the other stations.

The δ¹³C values of Ecdyonurus levis and Lymnaea auricularia little differ at ES 2 and ES 3 (ranging from -26.1 to -24.6‰ and -22.0 to -21.7‰, respectively), whilst the mean δ¹⁵N values of these species tended to decrease from ES 2 to ES 3 (4.9 to 2.5‰ and 5.6 to 2.5‰, respectively). The δ¹³C and δ¹⁵N values of Paraleptophlebia chocolata showed increasing pattern from ES 3 to ES 4 (-23.9 to -19.4‰ and 3.1 to 6.1‰, respectively). Moreover, the δ¹⁵N values of macroinvertebrates at ES 3 (ranging from 2.4 to 3.1‰) were relatively lower, whilst those at ES 4 (ranging from 4.2 to 9.9‰) were relatively higher than at the other stations. We used data of Shin et al. (2013) in relation to carbon and nitrogen stable isotope ratio for SPOM, BPOM, and periphyton, in order to reveal relationship between macroinvertebrates and their food sources (Fig. 2).

Fig. 2. 
Mean carbon and nitrogen stable isotope plots of macroinvertebrates and their potential food sources at the respective stations along the Echi River. The square frames indicate the classification of aquatic insects according to functional feeding groups. The open triangles indicate ES 1, gray triangles indicate ES 2, open squares indicate ES 3, and black squares indicate ES 4. SPOM: suspended particulate organic matter, BPOM: benthic particulate organic matter, 1: Rhyacophila nigrocephala 2: Ecdyonurus levis 3: Paraleptophlebia chocolata 4: Ephemera strigata 5: Lymnaea auricularia 6: Psephenoides spp.; 7: Davidius lunatus 8: Hexatoma spp.; 9: Stenelmis spp. SPOM, BPOM, Periphyton data from Shin et al. (2013).

The potential food sources of macroinvertebrates included suspended particulate organic matter (SPOM), benthic particulate organic matter (BPOM), and periphyton (Allan, 1995). Commonly, even in the mean δ¹³C value of 0.8‰ and mean δ¹⁵N value of 3.4‰ in the isotopic fractionation (DeNiro and Epstein, 1978; Minagawa and Wada, 1984; Post, 2002) between macroinvertebrates and their food sources (SPOM, BPOM and periphyton) were considered, our results showed slightly difference between macroinvertebrates and their food sources. In this study, δ¹³C value of macroinvertebrates showed more than 5‰ of difference with periphyton of potential food sources(Fig. 2). The large difference between macroinvertebrates and periphyton can be explained by below reasons.

First, the macroinvertebrates has body with an exoskeleton made of chitin during growth. Lee et al. (2002) reported that water mite which sucks only body blood from organisms will assimilate the organic matters the value of which is higher than the whole body of the prey organism. In this study, δ¹³C value of macroinvertebrates mainly feeding on periphyton can be lower than those of food sources because chitin is isotopically lighter than muscle in organisms (Schimmelmann and DeNiro, 1986).

Second, Ephemeroptera may have feeding habit as scraper, as well as collector-gatherer according to around food environment (Shin et al., 2011; Merritt and Cummins, 1996).

Relationship between macroinvertebrates and their food sources are shown in Fig. 2. Although the δ¹³C and δ¹⁵N values of SPOM and BPOM did not differ between ES 1 and ES 2, those of Psephenoides spp., and Ephemera strigata in the upper reaches changed in parallel with those of periphyton in situ (Fig. 2). Moreover, the δ¹³C and δ¹⁵N values of Paraleptophlebia chocolata in the lower reaches were also changed in parallel with those of periphyton. This indicates that these macroinvertebrates assimilated the food sources having much more periphyton than POM. However, the δ¹³C and δ¹⁵N values of Ecdyonurus levis were close to those POM at each station (ES 1, ES 2, and ES 3). Merritt and Cummins (1996) reported that Ecdyonurus levis belong to either the scraper or collectorgatherer group. However, our results indicate that may be closer to the collector-gatherer group than the scraper group. Moreover, they reported that R. nigrocephala, Davidius lunatus, and Hexatoma spp. belongs to the predator group. The δ¹³C values of Davidius lunatus were similar to, and particularly appeared to change in parallel with, those of Ephemera strigata, suggesting that the larvae of Davidius lunatus may feed on Ephemera strigata in the upper reaches. Furthermore, the δ¹³C values of Hexatoma spp. at ES 4 were similar to those of Stenelmis spp., indicating that the larvae of Hexatoma spp. may feed on Stenelmis spp. larvae at ES 4. These results were similar to the result of the Inukami River reported by Shin et al. (2011), who studied on variation in trophic pathways and food web characteristics in an intermittent stream system. The δ¹³C values of macroinvertebrates and periphyton tended to increase from the upstream to the downstream reaches (Fig. 2). They (Shin et al., 2013) reported that the water temperature and pH at ES 4 showed the relatively higher (WT 28.3℃, pH 8.6) than those at the other stations (WT 23.7 to 24.9℃ pH 7.1 to 7.2). Presumably, these results suggested that the enriched δ¹³C values at ES 4 reflect a shift from CO₂ to bicarbonate as a substrate for algal photosynthetic activity according to these parameters.

The δ¹⁵N values for Ecdyonurus levis and Lymnaea auricularia considerably decreased from ES 2 to ES 3 (Table 1). The δ¹⁵N values of their potential food sources at ES 3 ranged from 0.3 to 2.2‰ (Fig. 2), indicating that relatively lower stable isotope signal of Ecdyonurus levis and Lymnaea auricularia reflected those of its diet in situ. In general, denitrification between ES 2 and ES 3 may be show heavy δ¹⁵N values of periphyton and then also macroinvertebrates. However, ES 3 influenced by seepage water showed relatively lower δ¹⁵N values of macroinvertebrates according to its diets. These results suggested that the diminished δ¹⁵N values of the same food sources originated from periphyton, which assimilated by light δ¹⁵N associated with relatively low ammonium concentration (Shin et al., 2013), may directly influence low trophic level of macroinvertebrates. Shin et al. (2013) reported that although ES 3 was characterized by groundwater derived from seepage water, the physicochemical parameters did not differ among stations. Probably seepage water would be little influenced by the industrial activities such as the inflow of contaminants from the surrounding watershed (Verbanck et al., 1989), and consequently ES 3 showed relatively lower chloride, sulfate, and sodium ions (Shin et al., 2013).

In conclusion, our results revealed that the macroinvertebrates and periphyton tended to get enriched δ¹³C values from the upper to lower reaches in relation to the physicochemical parameters such as water temperature and pH for longitudinal distribution in the Echi River, except for ES 3. Thus, in this study, habitat such as the seepage water region in streams reflects the change in habitat of organisms. The stable isotopic analysis of macroinvertebrates may be use as indicator to reflect stream states in relation to the trophic pathways according to in situ food source in the dry or wet stream.