1 Introduction

Mangrove forests along the tropical and subtropical coasts have high levels of biological productivity (Alongi, 2009) and rich diversity of benthic fauna as an important component of the ecosystem (Chen et al., 2013). Man-grove plants provide organic matter and nutrients critical to the benthic fauna primarily through litterfall (Kieckbusch et al., 2004; Chen et al., 2013). In addition, the com-plex aboveground structures of mangrove plants provide the physical structure for benthic fauna to inhabit and pro-tect them from environmental stresses and predators (Chen et al., 2013).

The benthic fauna plays an important role between the mangrove detritus at the base of the food web and the consumers at higher trophic levels in mangrove ecosystems (Macintosh, 1980). The knowledge of the importance of autotrophic sources in supporting the faunal community is fundamental for understanding the relationship between fauna and flora in different environments, which is vital for the development of conservation priorities and effec-tive management policies for ecosystems (Connolly et al., 2005).

Mangrove-derived organic carbon generally contributes a majority of food sources for benthic fauna within man-grove forests (Giarrizzo et al., 2011; Abrantes et al., 2015). In addition to mangrove materials, macrobenthos can also utilize other organic matters, such as autochthonous micro-and macrophytobenthos, and tidally imported allochtho-nous sources, including plankton-or seagrass-derived or-ganic matters (Bouillon et al., 2004; Kieckbusch et al., 2004; Thimdee et al., 2004). Sesarmid crabs, as the most common crustacean group in mangrove forests, are main-ly dependent on the mangrove leaf litter; however, they also utilize sediments and invertebrates to supplement their food sources (Lee et al., 2001; Chen et al., 2016). Fiddler crabs in mangrove forests, represented by Uca spp., are mainly deposit feeders (Hsieh et al., 2002; Meziane and Tsuchiya, 2002). Therefore, the food sources of benthic fauna in mangrove forests are suggested to be partly re-lated to the composition of the sedimentary organic mat-ter in mangroves (Bouillon et al., 2004). Where the man-grove forest receives substantial exogenous carbon sources, such as plankton, macroalgae and seagrass, these sources also play important roles in sustaining secondary produc-tion (Bouillon et al., 2002; Kieckbusch et al., 2004; Thimdee et al., 2004).

Indonesia, with a total mangrove area of 31100 km², represents the global hotspot in terms of mangrove extent (Giri et al., 2011), while the trophic relationship between mangroves and benthic fauna is currently poorly under-stood in this area. In this study, the primary food sources of epifauna and sediment carbon sources were investi-gated in an oceanic mangrove forest in North Sulawesi, Indonesia. Recent studies have found that the mangrove sediments in this area were rich in organic carbon (Murdiyarso et al., 2015; Chen et al., 2018), and the sediments mainly consisted of mangrove detritus (Chen et al., 2018). Whether the mangrove-derived organic carbon could make an important contribution to the food sources of epifauna in the oceanic mangrove forest was further tested in this study. We hypothesized that the mangrove-derived orga-nic carbon comprises the majority of the sediment organic carbon and the mangrove carbon generally provides ma-jor food sources for epifauna.

2 Methods 2.1 Study Area

The study was conducted in a mangrove forest located in Kema (1°22'56.47 〞N, 125°5'43.65〞E) in North Sula-wesi, Indonesia (Fig. 1). The mangrove forest, dominated by Rhizophora apiculata Bl., extends about 200 m from its land fringe to the sea fringe. Seagrass meadow dominated by Enhalus acoroides (Linnaeus f.) Royle, 1839 and Tha-lassia hemprichii (Ehrenberg) Ascherson, 1871 occurs at the sea fringe of the forest (Chen et al., 2017). This area has a typical equatorial climate, and the mean tempera-tures at sea level are nearly uniform, varying by only a few degrees throughout the year, from 20 to 28℃. Tides in this area are mixed and mainly semidiurnal and fluctuate slightly, with an annual tidal range of 2.4 m (Chen et al., 2017).

Samples were collected in August 2015 (dry season). The individuals of epifauna on the mud surface, in burrows and prop root gaps near the ground, and on the trees were collected. In the sampling area, surface sediment samples (6 replicates) were collected to a depth of 5 cm using hand-held PVC corers (inner diameter 70 mm). The epifauna and sediment samples were kept in a cool box and transported to the laboratory. The epifauna were identified and divi-ded into four feeding guilds, including carnivorous, omnivorous, phytophagous, and planktophagous (Lai and He, 1998; Vannini et al., 2001; Alfaro, 2008; Poon et al., 2010; Yu et al., 2016; Quan et al., 2018).

We defined the potential primary sources for sediment carbon and the food sources of epifauna as mangrove-de-rived organic carbon, imported seagrass-derived organic carbon and suspended particulate organic matter (SPOM) by tides in this study. Macroalgae was not taken into ac-count as a source because it was not observed within the forest. Senescent leaves of R. apiculata, representing the initial stage of leaf material entering the food web, were collected from 9 different trees by shaking the trunk, and five leaves from each tree were combined into one sample. To collect the SPOM, seawater that is far away from man-grove and seagrasses was collected in a plastic bottle. After passing through a 75 μm mesh, it was filtered through a precombusted 0.7 μm GF/F filter (Whatman) in the labo-ratory. For seagrass, leaf samples were collected in three replicate quadrats of 625 cm² at the sea fringe of the man-grove forest, and the tissues were cleaned to remove epi-phytes and sands.

2.2 Sample Analysis

For crabs, the muscle tissue of the whole body was used for isotopic analysis after removal of the gut and intesti-nal system. The muscle tissues of mollusks were removed from their shell and analyzed together. For the small snail Optediceros breviculum, five or six individuals were pool-ed into one sample. All fauna samples were ground into fine powder after freeze drying. Plant tissues were dried at 60℃ for 24 h and the sediment samples were air-dried after removing the visible animals, plant residues and stones (Chen et al., 2017). For isotopic analysis, all ground sam-ples of epifauna, seagrass leaves and sediments were placed into silver cups, acidified with diluted HCl (5%) and then oven dried at 40℃ to remove the carbonates. Filtered samples of SPOM for isotope analysis were acidified by fuming overnight over 1 mol L^-1 HCl to remove inorganic carbonates. No acidification was done for mangrove sam-ples, because no carbonate was expected to be present in the leaves.

The carbon and nitrogen isotopic compositions of the samples were measured using a Thermo Flash EA 1112 HT-Delta V Advantages system. The stable isotopic composi-tions were reported in the δ notation (‰) as the ratio of the heavy (¹³C or ¹⁵N) to the light (¹²C or ¹⁴N) stable iso-tope in the sample (R_sample) relative to that of a standard (R_standard), i.e., δX = 1000 × [(R_sample/R_standard)-1], with the standard being Vienna Peedee Belemnite (VPDB) for car-bon and atmospheric N₂ for nitrogen and R = ¹³C/¹²C or ¹⁵N/¹⁴N.

2.3 Statistical Analyses

The differences in carbon and nitrogen isotopic compositions among the three primary food sources were com-pared using the Kruskal-Wallis nonparametric test, per-formed using IBM SPSS Statistics (version 22.0, Chicago, IL, USA). The potential contributions of the three primary sources to sediment carbon composition and epifauna diet were estimated using a Bayesian stable isotope mixing model, SIMMR. The SIMMR model required the input of values representing isotopic shifts between consumers and their potential food sources (trophic enrichment factors, TEFs). The TEF of δ¹⁵N was 3.3‰ ± 0.26‰ for the car-nivorous group and was 2.2‰ ± 0.3‰ for the other three feeding groups; for δ¹³C, the TEF was 1.3‰ ± 0.3‰ for all feeding groups (McCutchan et al., 2003; Park et al., 2017).

3 Results 3.1 δ¹³C and δ¹⁵N Values of the Primary Sources and Sediment

Three primary food sources can be clearly distinguished in the dual-isotope signatures (Table 1). Mangrove leaves had the most negative δ¹³C value (-30.11‰) among the three primary sources, while the most enriched δ¹³C was measured in seagrass leaves. Mangrove and seagrass leaves had the highest δ¹⁵N values (about 4‰). Mangrove sedi-ment was more δ¹³C-depleted than the other two alloch-thonous sources (seagrass leaves and SPOM), and its δ¹³C value was closer to that of mangrove leaves than to the values of these two sources. The SIMMR model analysis results showed that mangrove-derived carbon contributed 64.90% of the organic carbon in the benthic sediment.

3.2 δ¹³C and δ¹⁵N Values of Epifauna in the Mangrove Forest

A total of 18 epifauna species (some individuals were identified only to genus) were collected from the Kema mangrove forest, including 6 crustacean species, 3 bivalve species and 9 gastropod species (Fig. 2 and Table 2). These species were classified as carnivorous (4 species), omni-vorous (3 species), phytophagous (7 species), and plank-tophagous (4 species).

The epifauna had a broad range of δ¹³C values, from -26.10‰ to -13.80‰, and the δ¹⁵N values varied from 1.44‰ to 9.46‰ (Fig. 2). The most depleted δ¹³C, about -26‰ for the two sesarmid crabs, was closer to the man-grove signal than those of other primary sources, and it was 4‰ more enriched than mangrove signal. The three bi-valve species, i.e., Anadara antiquata, Anadara spp. and Callista erycina, had the highest δ¹³C values (about -14‰) and were about 2‰ more depleted than the seagrass sig-nal. The three littorinid species showed little variability in δ¹⁵N value, ranging from 1.44‰ to 1.60‰, and their δ¹³C values lied between those of the mangrove and SPOM signals. The δ¹³C and δ¹⁵N values of the epifauna did not differ significantly among the four feeding guilds.

3.3 Potential Food Sources of the Epifauna

The SIMMR mixing model estimations showed a dis-similarity in the composition of primary food sources as-signed to different epifauna species (Fig. 3). Overall, man-grove-derived carbon predominantly comprised the food sources of epifauna of variable feeding guilds, including Epixanthus dentatus, Parasesarma semperi, Penaeus sp., Optediceros breviculum, Terebralia sulcata and Littora-ria spp. The highest proportional contribution of mangrove was 71.7% for P. semperi, with E. dentatus following. For the three littorinid species and T. coarctata, mangrove-de-rived carbon and SPOM made a similar contribution to their food sources. Seagrass was found to dominate the food sources of three planktophagous species, A. antiquata, Ana-dara sp. and C. erycina.

4 Discussion 4.1 Sources of Sediment Organic Carbon

Mangrove detritus is generally the major source of the accumulated carbon in mangrove sediments, while allo-chthonous organic matter may provide high OC inputs to the sediments when a high rate of input from riverine or tidal sources occurs (Jennerjahn and Ittekkot, 2002). The surface sediment in the Kema mangrove had a δ¹³C value close to that of the mangrove signature which is about 2‰ more enriched in δ¹³C than the mangrove leaves, while the sediment organic carbon was primarily comprised of the autochthonous mangrove source. The more depleted δ¹³C and higher proportional contribution of mangrove to sediment organic carbon measured in this study and in another oceanic mangrove in North Sulawesi (Chen et al., 2018) than those in other estuarine mangrove forests (Xue et al., 2009; Ranjan et al., 2011), which may suggest a more important role of autochthonous mangrove source in sedi-ment carbon accumulation in the oceanic mangrove for-ests.

Although seagrass meadow occurred at the seaward fringe of the mangrove forest, seagrass-derived carbon had a much lower contribution than that of mangrove carbon. This was similar to our previous finding that mangrove-derived car-bon accounted for > 80% of the sediment organic carbon at another oceanic mangrove adjacent to seagrass meadow in Wori, North Sulawesi (Chen et al., 2018). A recent study also suggested that mangrove forests could export carbon to their adjacent seagrass meadows, and mangrove detri-tus contributed a higher proportion than that of autoch-thonous seagrass to the sediment organic carbon in tropi-cal areas (Chen et al., 2017).

4.2 Contribution of Mangrove-Derived Carbon to the Food Sources of Epifauna

Numerous studies have demonstrated that in mangrove forest which receives allochthonous organic matter inputs, the mangrove materials have limited importance to the diet of consumers associated with the mangrove habitat (Bouillon et al., 2002; Kieckbusch et al., 2004; Thimdee et al., 2004). For instance, the major food source of the benthic fauna in an estuarine mangrove was attributed to SPOM rather than to mangrove tissues (Yang et al., 2017). However, higher uptake of mangrove carbon by benthic consumers is usually accompanied by a decrease in the importance of allochthonous carbon input in the sediment (Bouillon et al., 2002). The present study found that the mangrove-derived carbon was the major food source for the majority of the epifauna collected in Kema mangrove forest, where the sediment carbon was mainly comprised of mangrove source. This was consistent with a previous finding that mangrove carbon could be an important source for consumers in a habitat where other primary sources are much less available (Giarrizzo et al., 2011). Al-Maslamani et al. (2013) also suggested that the fauna resident within an arid mangrove forest with a small amount of inwelling carbon were mainly dependent on local retained nutrients (Al-Maslamani et al., 2013). The contribution of mangrove carbon could be up to 84% for prawn tissues in a Malaysia mangrove forest where the sediment mainly contained mangrove detritus (Chong et al., 2001).

The δ¹³C and δ¹⁵N values of the epifauna collected in this study were not clearly distinguished among the four feeding guilds, and mangrove-derived carbon could be the major food sources for epifauna belonging to different feeding guilds. In addition to those phytophagous species, e.g., sesearmid crabs which mainly feed on mangrove tis-sues (Chen et al., 2014), some epifauna of other feeding guilds are deposit feeders (Hsieh et al., 2002; Meziane and Tsuchiya, 2002) and could uptake mangrove detritus during scraping materials from the sediment surface. We therefore suggest that mangrove-derived carbon can directly (via grazing plant tissues) or indirectly (via deposit feedings) make an important contribution to the food sources of epifauna in Kema mangrove forest where the allochtho-nous input of organic carbon is low.

In this study, benthic diatoms were not taken into ac-count, although they have also been recognized as an im-portant primary food source for the benthic fauna (Bouil-lon et al., 2002). Considering the benthic diatoms had higher δ¹³C values (-20.6‰ -12.1‰) than the mangrove signa-ture (Newell et al., 1995; Bouillon et al., 2002), the pro-portional contribution assigned to the mangrove source in this study might be underestimated. The food sources of benthic fauna will be investigated in the future consider-ing benthic diatoms as one potential food source.

4.3 Food Sources of Different Faunal Groups

The highest contribution of mangrove-derived carbon to the food source was found in P. semperi crabs, con-sistent with previous findings that sesarmid crabs relied on mangrove source (Kristensen et al., 2010). However, sesarmid crabs must use supplementary N sources, inclu-ding sediments and invertebrates, as mangrove leaves are high in the C:N ratio and the crabs are capable of utilizing carbon but not nitrogen from leaf litter (Lee, 1997; Thongtham and Kristensen, 2005). The sediment in the Kema mangrove forest had a δ¹³C signature similar to that of mangrove leaves, and the P. semperi crabs were 1.3‰ more enriched in δ¹³C signature than mangrove sediment, sug-gesting that the crabs could also use other sources for food. A recent study found that macroalgae could even domi-nate the food source of sesarmid crabs over mangrove tis-sues when the macroalgae was present in the forest (Gao et al., 2018).

Littorinid snails are generally herbivores, grazing on mi-croalgae, filamentous algae, bacteria, and mangrove tis-sues. Their dietary differences appear to be related to food composition and abundance on different parts of the man-grove trees where they attach rather than active food se-lection (Alfaro, 2008). The δ¹³C values of the three lit-torinids in this study varied within a narrow range from -25.09‰ to -24.14‰, indicating a similarity in their diets, which were mainly comprised of mangrove and SPOM. There was a difference in food source between two pota-midid snails, P. cingulata and T. sulcata, with T. sulcata relying more on the mangrove source. However, T. sul-cata collected from an estuarine mangrove in Vietnam showed a δ¹³C value close to those of sources other than mangrove plants (Tue et al., 2012).

Previous studies showed that the δ¹³C value of fiddler crabs, representing another important faunal group, was generally higher than that of the mangrove signature (Rodelli et al., 1984; Kon et al., 2009), suggesting that they selectively feed on δ¹³C-enriched sources and were much less dependent on mangrove source than sesarmid crabs. A study conducted in estuarine mangrove forest showed that mangrove source represented < 25% of the primary food source of Uca sp. (Le et al., 2017), and a negligible contribution of mangrove was detected in the diet of U. maracoani crabs in Curuçá Estuary in Brazil (Giarrizzo et al., 2011). However, mangrove-derived carbon com-prised 24.0% – 52.9% (with a mean of 39.2%) of the food source of T. coarctata in the Kema mangrove forest, close to the contribution of SPOM. Because fiddle crabs are generally deposit feeders (Hsieh et al., 2002; Meziane and Tsuchiya, 2002) and mangrove source could be utilized during their scraping sediment materials rich in mangrove source. Kon et al. (2009) also found that Uca crabs de-pended on mangrove detritus for food in a mangrove fo-rest where the isotopic compositions of the sediment organic carbon were very close to the mangrove leaf values.

4.4 Seasonal Variation in the Food Sources of Epifauna

The faunal community shows seasonal changes in com-position and distribution (Al-Maslamani et al., 2013; Zou et al., 2018), and the feeding behaviors of epifauna might vary with their life stages (Tue et al., 2012). It has been found that the δ¹³C of epifauna was significantly and ne-gatively correlated with their body sizes and that their diet could present an ontogenetic increase in the contribution of mangrove detritus (Tue et al., 2012). In this study, we did not address the seasonal variation in the food sources of epifauna and the samples of each faunal species were combined together for analysis without considering their body sizes. We considered that this would not prevent us from detecting the role of mangrove-derived carbon as a major food source of epifauna in this study. The benthic en-vironment in the Kema mangrove forest did not signifi-cantly vary across seasons (Sprintall et al., 2003). Bouillon et al. (2004) has suggested that the lack of considera-tion of seasonal variation would not limit the assessment of the general patterns of food sources of benthic fauna in mangrove forests. Nevertheless, we are proposing more de-tailed studies to compare the food sources of benthic fauna in mangrove forests with different allochthonous carbon loadings, and their seasonal variation in composition and size (i.e., life stage) conditions will be considered.

5 Conclusions

The present study investigated the primary food sources of epifauna in a tropical oceanic mangrove forest. The results showed that the benthic sediment mainly consisted of mangrove detritus, and a majority of the collected epi-fauna species were more dependent on mangrove source than on other allochthonous sources, such as seagrass and phytoplankton. The results suggested that mangrove-de-rived carbon contribute significantly to the food sources of the epifauna inhabiting tropical oceanic mangrove fo-rest where the allochthonous input of organic carbon is low.

Acknowledgements

This study was supported by the National Key Tech-nologies Research and Development Program of China (No. 2017YFC0506101), the China-ASEAN Maritime Fund 'Monitoring and Conservation of the Coastal Ecosystems in the South China Sea', and the National Natural Science Foundation of China (No. 31101876). The authors are grateful to Mr. Asep Rasyidin, Mr. Mochtar Djabar and Dr. Xijie Yin for their assistance with field sampling and laboratory analysis.