We identified two categories of leaf-eating crabs according to their feeding habits: (1) species active outside their burrows during low tide; and (2) species spending most of their time inside burrows. For the first category, we used Parasesarma affine, a common sesarmid associated with mangrove forests in Hong Kong, as the animal model. This crab is distributed in the mid intertidal zone, spending most of their foraging time (45~83% of observation time) consuming mangrove leaves or scraping the surface sediment during the low tide (Kwok, 1999). For the second category, we used Neosarmatium smithi as the model species, which is a large species (maximum carapace width about 5 cm) specifically inhabiting the high-intertidal Bruguiera sexangula mangrove forests on tropical Hainan Island, China. Feeding activities of N. smithi in the field monitored using an infrared camera for 24 hours during the active season of the crabs (June 2016) within the site where the crabs were sampled in this study. It showed that they spent most of their time in the burrows (75~85% of observation time) even without any disturbance and only 15 to 25% of observation time searching and walking. Neither of the two species showed readiness to consume animal material when provided under captive conditions in this study.

Individuals of P. affine of carapace width (CW) ranging from 11 to 20 mm were collected from the mangrove forest (Kandelia obovata) at Mai Po (22°29’38”N,114°01’56”E), Hong Kong. Individuals of N. smithi of CW ranging from 20 to 25 mm were collected from the Bruguiera sexangula forest at Dongzhai Bay (19°57’01”N,110°34’45”E), Hainan Island. Only male crabs were collected for experiment to reduce the effect on the crab populations as female crabs were often pregnant, and the female crabs may also have different physiology during egg production. All the crabs, fresh litter (yellow leaves), fallen calyx, and MPB (surface sediment) were collected from the same sites in the mangrove forests. Crabs for the feeding experiment were allocated to individual containers and starved for two days and “field” samples were rinsed with MilliQ water and sacrificed by freezing. Sediments were collected by scraping the top 1 cm surface layer, homogenized with a hand trowel before incubation with isotopically enriched media or used for MBP extraction.

MPB was extracted from sediment by density gradient centrifugation in colloidal silica (Bui and Lee, 2014). The sediment sample was washed through a series of decreasing mesh sizes: 2-mm, 500μm, 250μm, 45μm and 5μm. The material retained on the 5μm sieve was then transferred to 50ml falcon tubes. After settling at 4°C overnight, the clear supernatant was removed carefully to avoid disturbing the sediment at the bottom. 23ml of colloidal silica (LUDOX™) solution (1.340g ml^-1 density) was then added to each tube containing the sample, mixed and centrifuged (4000 rpm at 4°C) for 10 mins. The entire top layer (MPB was in this layer) was isolated and washed with MilliQ water and then transferred into a tin capsule and dried for stable isotope analysis.

Two enrichment solutions were prepared following Putz et al. (2011) with some modification: ¹³C-urea solution and ¹⁵N-urea solution by dissolving 100 mg 99-atom% ¹³C urea and 2 mg 98-atom% ¹⁵N urea in 50 mL MilliQ water, respectively. 12.5µl of wetting agent was added to each solution for good contact of the labeling solution with leaf surface. The two urea solutions were applied with a small paint brush on the upper and lower surfaces of the mangrove leaves on two Kandelia obovata trees in the field. Two trees separated by >100 m were selected for the two respective enrichments to avoid cross-contamination. Ten branches (each with 10 to 15 leaves) on each tree were selected for enrichment. Labelling was applied once a day over five consecutive days for ¹⁵N enrichment and 15 days for ¹³C enrichment.

Surface sediment from the study sites was incubated with F/2 medium (Guillard and Ryther, 1962; Guillard, 1975). The N source (NaNO₃) in F/2 medium was replaced by ¹⁵NH₄Cl (99-atom% ¹⁵N), incubated for 10 days with an irradiance at 150 to 200 μmol m^-2 s^-1 (same level as in the field), and ambient outdoor temperature at 25°C to 35°C. ¹³C-enriched MPB preparation followed the same method for ¹⁵N-enriched MPB, with the N source being NaNO₃ and NaH¹³CO₃ (99-atom% ¹³C, 2.5mM) was added to F/2 medium. MPB incubated with non labelled F/2 medium was used as a control (IMPB).

C and N content as well as stable isotope values (δ¹³C, δ¹⁵N and δ³⁴S) of leaf, crab, and MPB samples were measured with a Thermo Analytical elemental analyzer, Flash EA 1112 Series coupled via a ConFlo IV interface to a Thermo Delta V Plus isotope ratio mass spectrometer. Stable isotope ratios are expressed as δ values (in per mil notation, ‰) relative to conventional standards (Vienna Pee Dee Belemnite for C and atmospheric N₂ for N), according to:

where X = ¹³C, ¹⁵N or ³⁴S, and R = ¹³C/¹²C, ¹⁵N/¹⁴N or ³⁴S/³²S. Measurement precision was better than 0.3‰ for δ¹³C, δ¹⁵N and δ³⁴S.

Sediment inside crab burrows in the field and faeces of crabs fed yellow leaves (for 30 days, 60 days and 210 days) were sampled for functional bacteria analysis. Using qPCR and amplicon sequencing technology, the special functional genes could be identified directly, and the universality and diversity of information for key functional genes was also provided. Here our target functional gene was the N-fixing gene nif H. According to the trait of the amplification area, based on the IonS5TMXL sequencing platform, a small fragment library was constructed for single-end sequencing. By cutting and filtering reads, clustering operational taxonomic units (OTUs), and annotating species and abundance analysis, the microbial community structure of the samples was assessed, and the difference among samples was further compared by alpha and beta diversity analyses.

N mass balance was assessed for the field population of P. affine to address the question of whether sole dependence on MPB N could sustain the crab’s need. The sesarmid crabs usually collect MPB at the surface sediment using their chelae, with a length of the dactyl and propodus at about 4 mm. However, they usually do not insert the whole dactyl and propodus into the sediment, so we assumed the depth of the sediment they could have access to MPB was 2 mm. MPB biomass (g m^-2) in the top 2 mm surface sediment was estimated by the concentration of chlorophyll-a (by aqueous acetone extraction method, Johan et al., 2014), assuming a carbon to chlorophyll-a ratio (C: Chl-a) of 40 (De Jonge, 1980). The amount of MPB N available (g m^-2) was determined according to their biomass and N content (%) obtained from elemental analysis.

The amount of N needed (g m^-2 day^-1) for P. affine growth was estimated according to their tissue N turnover rate (NTR):

where DW = total dry weight of crabs m^-2 (g DW m^-2); N% = N content of crab tissues; and D = turnover time (days) for tissue N, estimated to be 70 days for sesarmid crabs (Bui and Lee, 2014).

In the K. obovata mangrove forest at Mai Po, the mean population density of P. affine was 1.39 individuals m^-2, with the mean carapace width (CW) and the mean dry weight at 18.3 mm and 1.5 g m^-2, respectively (Lee and Kwok, 2002). The N content of the crab tissues was about 6.5% of dry mass estimated according to the equation %N = 7.536 – 0.059 CW established for the congener P. messa (Robertson, 2021). Therefore, assuming that no other significant N loss pathways exists, the amount of N needed for sustaining the biomass of P. affine at field density was taken to be their tissue N turnover rate (NTR). This demand was then compared with the amount of N available from MPB in the top 2 mm of the sediment.

Five feeding experiments (E1 to E5) were designed to test the hypotheses H1 and H2. Experiments E1~E3 were aimed to test H1, experiments E4~E5 were to test H2, with details of each experiment summarized in Table 1 . For E1~E4 and treatments T2 and T3 in E5, the crabs were kept in natural atmosphere. For the treatment T1 in E5 crabs were fed in an air-tight container (size: 70.5cm × 48 cm ×38.5 cm) with an atmosphere enriched with ¹⁵N₂, and 1L ¹⁵N₂ was added to the container at the beginning of experiment (δ¹⁵N of the atmosphere in the experimental container was about 3100‰) and was replenished with the same volume after each weekly sampling. All the crabs were sacrificed by freezing and the muscle tissue was taken from their claw for stable isotope analysis.

One-way ANOVA followed by post hoc Tukey’s test were used to test differences (α=0.05) in stable isotopic values among different groups or treatments. Nonparametric Kruskal-Wallis test followed by all pairwise multiple comparisons was conducted if the assumption of homogeneity of variance was violated. All data analyses were performed using SPSS 28.0.

After incubation for ten days with F/2 medium (N: ¹⁵NH₄Cl), the δ¹⁵N level of MPB reached 262.2 ± 51.3 ‰ (mean ± SD). Tissue δ¹⁵N value of crabs fed ¹⁵N-enriched MPB increased significantly with experiment time, from 12.6 ± 0.1‰ (mean ± SD) at the beginning to 183.1 ± 89.2‰ (mean ± SD) at the end of experiment, which was almost approaching the enrichment level of the MPB ( Figure 1 ).

Crabs on a sole diet (leaf or MPB) showed a higher mortality than crabs fed a mixture of MPB and leaves ( Figure 2 ). δ¹⁵N values separated crabs from food sources, while δ³⁴S values were able to separate crabs on different food types. The δ³⁴S value of the crabs fed both MPB and yellow leaves (M+YL) was intermediate between those fed MPB (M) and yellow leaves (YL) ( Figure 3A ), so the contribution of each of the two foods was approximately 50%. For δ¹³C there was significant difference between crabs and foods, but no significant difference among crabs fed different foods ( Figure 3B ).

For crabs fed ¹³C-enriched leaf + ¹⁵N-enriched MPB (treatment 1), the tissue δ¹³C values increased with leaf δ¹³C, whereas the δ¹⁵N values increased with MPB δ¹⁵N from week 1 to week 4 (W1 to W4, respectively). There were significant differences between the start date (δ¹³C= -24.9 ± 06 ‰, δ¹⁵N = 13.5 ± 0.2 ‰), and end date values (δ¹³C = -17.2 ± 2.7 ‰, δ¹⁵N = 174 ± 77.1 ‰) for both δ¹³C and δ¹⁵N (p<0.05). For crabs fed ¹³C-enriched MPB + ¹⁵N-enriched leaf (treatment 2), the δ¹³C and δ¹⁵N of crab tissue did not show any significant change ( Figure 4 ).

Conversion from the chlorophyll a content in the top 2 mm of surface sediment (9.42 mg m^-2) using a 40x conversion factor (De Jonge, 1980) suggests that the C biomass of MPB was 377 mg C m^-2. As the C/N of MPB is ~15, the amount of MPB N in the top 2 mm of the surface sediment was 25.1 mg N m^-2. Based on the NTR formula, the amount of N needed for sustaining the biomass of P. affine at field density was 1.5 g m^-2 × 6.5%/70 days = 1.4 mg N m^-2 day^-1. The N requirement for crab sustenance is therefore 5.58% of the total MPB N standing crop in the surface sediment.

There is a significant difference in δ¹³C and δ¹⁵N values as well as their C and N contents (p<0.05) between the calyx and yellow leaves of B. sexangula. The calyx was more depleted in ¹⁵N than the yellow leaf, but less depleted in ¹³C ( Figure 5 ). The calyx has a higher N content (0.83 ± 0.13%) and lower C content (C/N = 48.3) than the yellow leaf (0.44 ± 0.08%) (C/N= 99.5). After feeding for 60 days, the group that was supplied with only yellow leaves showed a significant increase in δ¹⁵N compared with those before feeding (p<0.05), but no change in δ¹³C value. The group supplied with only calyx showed no significant change in both their δ¹³C and δ¹⁵N values. The group fed with yellow leaves and supplemented with calyx (YL+C7) showed a significant change in both δ¹³C and δ¹⁵N values (p<0.05): lower δ¹⁵N but higher δ¹³C values, with the change in the YL+C7 group being much larger than that of the YL+C14 group ( Figure 5 ).

After feeding on yellow leaves only for 60 days, the δ¹⁵N value of N. smithi showed a significant increase of 1.51‰ compared with that before feeding (p<0.001). However, 210 days later, the δ¹⁵N of N. smithi decreased by approximately 2‰. The δ¹³C of N. smithi showed no significant change with feeding time ( Figure 6 ).

The crabs started to become enriched from 6.3 ± 1.4 ‰ (mean ± SD) in the field to 25.6 ± 16.1‰ (mean ± SD) after three weeks in a ¹⁵N-enriched atmosphere. At the fourth week, their tissue δ¹⁵N had enriched significantly (p<0.05) to 67.0 ± 33.4‰ (mean ± SD) compared with the start date values ( Figure 7A ). The crabs fed normal mangroves in an unenriched atmosphere showed no significant change in their δ¹⁵N after four weeks ( Figure 7B ). The crabs fed enriched mangrove leaves (δ¹⁵N (mean ± SD) = 3395 ± 2464 ‰) in an unenriched atmosphere also got enriched significantly (p<0.05) compared with the start date values after feeding for four weeks, and their δ¹⁵N enriched from 6.3 ± 1.4 ‰ (mean ± SD) before feeding to 166.7 ± 71.4 ‰ (mean ± SD) at the end of the experiment ( Figure 7C ), δ¹⁵N of crabs fed only unenriched yellow leaves changed to 7.4 ± 1.1 ‰ (mean ± SD) after four weeks.

Genomic analysis of functional bacteria groups showed a high relative abundance of nitrogen-fixing bacteria in the living environment (sediment from crab burrows) (~20%) of N. smithi, also their fecal material (~10 to 15%) ( Figure 8 ). And also a high relative abundance in the fecal material of the crabs fed only mangrove leaves in experiment 4.

In our study, the higher mortality of crabs on a sole diet (MPB or mangrove leaves) than on a mixed diet (both MPB and mangrove leaves, Experiment 2) suggests that P. affine achieved better physiological performance on a mixture of available food items. The δ³⁴S value of the crab tissues suggested that the contributions of the two food sources were each ~50%. Results of our enrichment experiment (Experiment 1) indicate that P. affine consumed MPB as a N source, which is the first study to provide direct evidence of MPB dependence using ¹⁵N-tracer. The results of experiment 3 in a differential labelling design using dual isotope tracers (¹³C-MPB + ¹⁵N-leaf and ¹³C-leaf + ¹⁵N-MPB) also suggest that P. affine relied on mangrove leaves as their C source and MPB as their N source. The N mass balance calculations indicate that sole dependence on MPB N is possible to sustain the crab’s need. As the N requirement for crab sustenance at field density and biomass is only 5.58% of the total MPB standing crop, it is possible for MPB to recover to original concentration quickly enough to sustain future extraction. In the field, P. affine spends 45%~ 83% of their time scraping material from the surface sediment (Kwok, 1999). All these findings suggest that consumption of MPB could be one way P. affine meets its N needs. However, there are still questions concerning the efficacy of this approach. In order to obtain this amount of N from MPB consumption, each crab needs to consume or process ~110 ml day^-1 of surface sediment, which is quite impossible to ingest as it is >100 times the stomach volume, unless they can separate the MPB from sediment effectively before ingestion. The mechanism by which the crab may extract MPB effectively from the sediment to make MPB a feasible food source is worth further investigation.

Several studies have highlighted the prominent trophic roles of MPB in coastal ecosystems (Miller et al., 1996; O'Meara et al., 2017; Pinckney, 2018; Hope et al., 2020). Even in the high Antarctic ice sheets with little light availability (<0.1%), MPB account for a significant proportion of the marine primary production (Dayton et al., 1986; Lohrer et al., 2013). Earlier data from stable isotope and other tracers suggest an important contribution of MPB to mangrove macrobenthos as a carbon source (Alongi, 2009; Oakes et al., 2010). However, only recently did Kristensen et al. (2017) first suggest that MPB was an important N source using the IsoConc mixing model on isotopic data to assess the food partitioning of mangrove leaf-eating crabs. These authors also suggested that mangrove litter was the dominant carbon source (contribution ratio >50%) for all the leaf-eating crabs.

Previous studies, however, have noted that several mangrove crab species spent most of their feeding time scraping material from the mud surface, e.g. Parasesarma messa (Robertson, 1986), P. erythodactyla (Camilleri, 1992), P. bidens and P. affine (Kwok and Lee, 1995), P. guttatum (Skov and Hartnoll, 2002), and Neosarmatium meinerti (Dahdouh-Guebas et al., 1997). These crabs occur in different intertidal zones and are associated with different mangrove communities. These crabs may rely on MPB as their N source to different extents, responding to local conditions such as forest light penetration and sediment nutrient levels that influence MPB abundance. The trophic role of MPB as the principal N source for leaf-eating sesarmid crabs in general needs to be further assessed in the future using a wider range of species covering different geographic and biogeographic locations.

Study of foraging behavior using remotely operated infrared cameras showed that this species spent the majority of time inside their burrow. In our observation, N. smithi did not feed on surface sediment. Our data agree with previous studies that the congener N. trispinosum actively collects and promptly feeds on mangrove leaf litter freshly available on the forest floor, and spends 97.5% of time remaining inside its burrow (Harada and Lee, 2016). This behaviour also excluded the possibility that they meet their N needs through consuming MPB. The stable isotope values also showed that the crabs assimilated N from the calyx food. Therefore, problems associated with low-nutrient vascular plant diets may be ameliorated by selecting higher-quality food items such as calyx, when they become available, which may be only for restricted periods. For example, calyx of B. sexangula are only available during the blossom season, which only lasts for less than 1 month (personal observation).

The δ¹⁵N value of the muscle tissue of N. smithi increased significantly after feeding on yellow leaves for 60 days. Previous studies showed that animals feeding on a diet with a low nitrogen content might recycle their internal N to sustain their metabolism and excrete the isotopically lighter ¹⁴N, which leads to an increase in ¹⁵N in their tissues (McCutchan et al., 2003; Herbon and Nordhaus, 2013). ¹⁵N enrichment in Daphnia magna also occurs when individuals suffered from starvation (Adams and Sterner, 2000).

Our results indicate that N deficiency occurred on N. smithi after feeding only on leaves after 60 days. After continued feeding on leaves only for 210 days, the δ¹⁵N value decreased significantly, suggesting that the N supply may be sourced from a lighter N source, which could be atmospheric N₂ (δ¹⁵N =0). Genomic analysis of functional bacteria groups showed a high relative abundance of nitrogen-fixing bacteria in their living environment (sediment from crab burrows), also the crab faecal material. The results of the crab feeding experiment on a sole diet of mangrove leaves in ¹⁵N-enriched air (experiment 5) showed that crab tissue became enriched from the third week of experiment, suggesting that N fixed from the atmosphere may be assimilated by the crabs through ingesting the bacteria directly or the organic matter they produced. However, the actual mechanism and processes need to be clarified in the future. Recently, nitrogen fixation was also found in the intestine of two sesarmid crabs (N. smithi and Episesarma versicolor) and their habitat sediments (by determining the nitrogenase activity using acetylene reduction activity) but no evidence for the contribution of the fixed N to the crabs’ tissues was provided (Tongununui et al., 2021). Crabs feeding on ¹⁵N-enriched mangroves leaves also became enriched in ¹⁵N after feeding for four weeks. All these findings suggest that N. smithi can get their N from both mangrove leaves and N-fixation by the associated microbes, and they can also meet their N need by selecting high-quality food sources when available.

The ecological role of crabs in mangrove ecosystems has been addressed in many studies (Lee, 1998; Welsh, 2003; Werry and Lee, 2005; Kristensen, 2008; Laverock et al., 2011; Gilbertson et al., 2012). The trophic significance of leaf-eating crabs is two-fold: (1) removal of leaves from the forest floor and therefore reduction in tidal export to the estuary, leading to N retention; and (2) processing of leaf litter to fine sized particles, which can more readily be mineralized by microbes and utilized by detritivores (Werry and Lee, 2005). Leaf-eating crabs digesting/assimilating mangrove detritus also play an important role in acting as a trophic intermediate between hard-to-digest detritus and higher-level consumers in mangrove ecosystems, as they are preyed upon by higher consumers such as fish (Sheaves and Molony, 2000; Lee, 2008; Kawaida et al., 2021; Robertson, 2021).

Previous studies have demonstrated that animals can play integral roles in the storage and remineralization of elements, thus they also play an important role in the biogeochemical cycles of both terrestrial and aquatic ecosystems (Augustine and McNaughton, 2006; Vanni et al., 2006; Vaughn, 2010; Coetsee et al., 2011). Strategies by which some species meet their nutrient requirement on low-nutrient diets may mediate a keystone role in nutrient cycling and overall ecosystem stoichiometry (Small et al., 2011). Like other organisms, leaf-eating crabs often face stoichiometrically imbalanced diets as mangrove leaves are high in C and low in N. They can meet their nutrient demands by both pre-ingestion (e.g. food selection) and post-ingestion regulation (e.g. egestion or excretion). These stoichiometric regulations influence the amount of nutrients retained or released by the individual organismal-level processes, such as food choice or the selective release of nutrients, ultimately driving higher ecosystem-level dynamics, such as ecosystem elemental cycling (Schade et al., 2011; Sperfeld et al., 2016). The strategy by which sesarmid crabs meet their N needs is important, as it can affect the growth rate and population size of these primary nutrient processors at community levels, ultimately affect trophic efficiency and nutrient fluxes in ecosystem levels.

The strategy by which leaf-eating crabs meet their N needs represents a significant contribution both to the growth of crabs and to their ecosystem functions of processing biologically significant elements such as carbon and nitrogen. We hypothesize that the consequence of this consumer-driven nutrient dynamic (CND) is more rapid nutrient turnover and higher primary productivity in ecosystems with large populations of leaf-eating crabs. CND also has been identified as an important function contributing to ecosystem services (Cardinale et al., 2012; Doughty et al., 2016; Atkinson et al., 2017). This role of mangrove leaf-eating crabs in regulating essential nutrients cycling has important implications for community structure and ecosystem functioning (Allgeier et al., 2013). Further work on the effects of leaf-eating crabs on ecosystem-level nutrient dynamics will help develop a comprehensive and mechanistic understanding of the functional role of sesarmid crabs, and how their diversity and abundance may influence mangrove ecosystem characteristics such as stability and resilience.