Introduction

The microphytobenthos (MPB), an assemblage of unicellular photoautotrophs inhabiting the euphotic layer of aquatic and marine sediments, is an under-explored functional group in mangrove ecosystems. Most studies of MPB have been done on unvegetated tidal flats or intertidal areas without a complete canopy (e.g., saltmarsh), but rarely on mangrove forests (Heip et al. 1995; MacIntyre et al. 1996; Kwon et al. 2020). This bias may have originated from the early paradigm that mangrove ecosystems are supported by a “detritus-based food chain” (Odum and Heald 1975) based on vascular plant production, and the assumption that the light-limited environment under the mangrove canopy is not conducive to MPB growth (Alongi 1994). However, a recent study by Kwon et al. (2020) discovered that MPB biomass under the mangrove canopy was higher than that on the nearby tidal flat. Similarly, Underwood (2002) observed dense biofilms on mangrove sediments comparable to those on nutrient-rich temperate estuarine sediments, and suggested that the mangrove canopy might foster higher MPB biomass by dampening their exposure to extreme environmental conditions.

Although not as well-studied as their counterparts on tidal flats, MPB have been demonstrated to play functional roles in the mangrove ecosystem. Evidence of varying contributions (due to methodological issues, for example) of MPB to the diets of mangrove macrofauna have emerged from stable isotope studies (e.g., Shahraki et al. 2014; Gao and Lee 2022). There is also evidence of the influence of MPB on sediment-air CO2 fluxes in mangrove forests (Ouyang et al. 2017; Chen et al. 2019), for instance, through the ability of MPB-dominated biofilms in physically reducing CO2 efflux (Lovelock 2008; Leopold et al. 2013, 2015; Bulmer et al. 2015). These findings imply that MPB do have a potentially notable presence and may play important roles in mangrove ecosystem processes, which need to be better assessed based on data from a wider geographic coverage and broader environmental gradients.

The photosynthetic activity of mangrove MPB has seldom been directly measured, instead mostly indirectly inferred from studies of their primary production, such as those measuring O2 flux (e.g., Alongi 1994). Alongi (1994) reported that mangrove MPB showed lower gross primary production than tidal flat MPB, and attributed it to the lower MPB biomass (due to reduced irradiance levels) inside the mangrove forest. These findings, however, may not sufficiently convey MPB’s photosynthetic performance, as their abundance data may not necessarily translate to productivity. Kwon et al. (2020) not only found mangrove MPB biomass exceeded that of tidal flats, but also the primary production of the mangrove MPB community was either similar to or lower than those of tidal flats. They further determined that the biomass-specific primary production (i.e., photosynthetic activity) of mangrove MPB was also lower, prompting the speculation that acclimation to lower light levels might have influenced mangrove MPB’s cellular photosynthetic performance. This finding conflicts with an earlier report that mangrove MPB have photosynthetic performance comparable to tidal flat MPB (Underwood 2002). How the photosynthetic performance of mangrove and tidal flat MPB compare warrants clarification. Its seasonality should also be examined, since there is evidence of positive temperature effects on MPB photosynthetic performance (e.g., Morris and Kromkamp 2003).

MPB-derived carbon (C) is excluded in the current global C budget of mangroves (e.g., Bouillon et al. 2008a; Alongi 2014) due to the lack of knowledge on MPB abundance and productivity, and how they participate in mangrove C cycling.While vegetation is the key factor driving the variation in MPB biomass (Kwon et al. 2020) in coastal wetlands, it is unknown if mangrove productivity affect mangrove-derived C. MPB have been found to regulate the quality and quantity of sediment organic matter in mesocosm experiments (Hardison et al. 2013). It is well known that MPB can assimilate CO2 from the atmosphere for growth, and also transport bicarbonate into cells using CO2 concentration mechanisms (Borowitzka et al. 2015). There is experimental evidence that photosynthesis in estuarine MPB is limited by inorganic carbon availability (Vieira et al. 2016). Both MPB and dissolved inorganic carbon in tidal marshes had low δ13C values (Curry et al. 2020), suggesting a possible linkage between the two. However, it is unknown if MPB can assimilate dissolved inorganic carbon (e.g., bicarbonate) from mangroves (e.g., produced via underground respiration) and if so, how this process is achieved, and its implications for C fluxes in mangrove ecosystems. It is also unknown if high plant density, as a proxy for productivity, is coupled with mangrove-derived dissolved inorganic carbon.

A few studies have been conducted on the mangrove MPB community (e.g., Sylvestre et al. 2004; Liu et al. 2013; Chen et al. 2019; Benny et al. 2021) and photosynthetic activity (Alongi 1994; Underwood 2002; Kwon et al. 2020), but none have presented a comprehensive view of the mangrove MPB community ecology and photosynthetic performance with respect to their counterparts on adjacent habitats (e.g., unvegetated tidal flats), as well as their role in ecosystem carbon flow. This study assesses the biomass, richness, and composition of these MPB communities over different seasons in two subtropical mangrove forests, where considerable variations in environmental conditions exist. The photosynthetic performance of mangrove MPB was also measured seasonally using pulse-amplitude modulated (PAM) fluorometry and compared with that of the MPB on the adjoining tidal flat. Additionally, we conducted a series of isotope-labelling mesocosm experiments to assess the coupling between MPB and mangrove in sediment and porewater carbon dynamics, and the general trophic role of MPB in mangrove ecosystems.

We hypothesize that a) MPB biomass and richness in mangroves are not lower than those on adjoining tidal flats regardless of seasons; b) different genera of MPB are present in mangroves and on tidal flats, and also under different environmental conditions; c) mangrove MPB are different from tidal flat MPB in that they are acclimated to lower light levels; and d) MPB production is tightly coupled with mangrove-derived dissolved inorganic carbon. The findings of this study will shed light on the significance of MPB as a functional assemblage in mangrove ecosystem processes.

Materials and Methods

MPB Community Ecology

Sample Collection

All sample collections and field measurements were conducted in two mangrove forests in Hong Kong. Hong Kong is influenced by a subtropical monsoonal climate with a considerable difference in air temperatures between summer and winter, offering an opportunity to examine the temporal dynamics of MPB community ecology in the mangrove forests versus that on the adjoining tidal flats. MPB were collected from the Mai Po Nature Reserve (MP) and Ting Kok wetlands (TK) in Hong Kong, for the determination of biomass (using chlorophyll a concentration as a proxy), taxonomic richness, and composition.

MP (22.4951°N, 114.0339°E) is located in the north-western part of Hong Kong, in the lower eastern Pearl River Estuary. The site receives brackish water (salinity range 0.2 to 24.4 ppt), has a fine-textured substrate (42.1% silt/clay, Tong et al. 2006), and is dominated by dense stands of the mangroves Kandelia obovata, Aegiceras corniculatum, Avicennia marina, and Acanthus ilicifolius. In contrast, TK (22.4686°N, 114.2184°E) is located in the north-eastern part of Hong Kong, away from the influence of the Pearl River, and therefore has saline waters (salinity range 21.3 to 33.3 ppt) and sandy substrates (12.8% silt/clay, Tong et al. 2006). The mangrove trees in TK are shorter (3 m) than those in MP (6 m), with Kandelia obovata and Aegiceras corniculatum as the dominant species. Both MP and TK have an unvegetated tidal flat located immediately adjacent to the mangrove forest, and a roughly equal tidal range of 2.7 m. For each of MPB biomass and composition analyses, MPB samples were collected randomly, at least 5 m apart, by scra** the top 0.5 to 1 cm of the sediment with a trowel. Ten samples were collected from each location (i.e., TK Mangrove, TK Tidal Flat, MP Mangrove, MP Tidal Flat) over nine occasions in summer (June–September 2021), and the same was done over five occasions in winter (December 2021–March 2022). However, the actual numbers of replicates were often less than 10, as the recovery rate of MPB was lower than expected in some samples, and these samples were omitted from the analyses.

Environmental Parameters

The following environmental parameters were measured during each MPB sample collection: temperatures of surface soil and porewater, pH and salinity of porewater, and irradiance level. Temperature of the surface soil was measured with a glass thermometer inserted to the top 1 cm of the sediment. Porewater was gathered by digging into the sediment until a pool of water was formed, where temperature reading was taken in situ with a glass thermometer. 50 ml of porewater sample was collected from the pool using a syringe and brought back to the laboratory for the measurement of pH with a pH meter (3540 Bench Combined Conductivity/pH Meter, Jenway, UK) and salinity with a hand-held refractometer (HI96822 Seawater Refractometer, Hanna Instruments, USA) within 24 h. Irradiance level was measured in situ with a photosynthetically active radiation (PAR) meter (Active Eye Quantum PAR Meter, HydroFarm, USA), by taking the average of 12 readings obtained over one minute.

MPB Biomass

The concentration of chlorophyll a ([Chl a]) was used as a proxy for MPB biomass. For the calculation of [Chl a] μg g−1 dry sediment of an MPB sample, 1 g of homogenized sediment was transferred into a pre-weighed crucible and dried at 60 °C for 48 h. Another 5 g was placed in 10 ml of 90% acetone and left in darkness at 4 °C overnight to extract the chlorophyll a pigments. Samples were centrifuged (4 °C, 3720 rpm, 5 min), and the absorbance of the supernatant was measured at wavelengths of 664 nm, 665 nm, and 750 nm with a spectrophotometer (U-5100 Ratio Beam Spectrophotometer, Hitachi, Japan) before and after the addition of 30 μl of 0.1 M HCl. [Chl a] in µg L−1 was derived with Lorenzen’s equation (Lorenzen 1967) and then converted to [Chl a] in µg g−1 dry sediment. The sampling groups in this analysis are: Summer TK Mangroves (n = 9), Summer TK Tidal Flat (n = 9), Winter TK Mangroves (n = 10), Winter TK Tidal Flat (n = 10), Summer MP Mangroves (n = 10), Summer MP Tidal Flat (n = 10), Winter MP Mangroves (n = 10), and Winter MP Tidal Flat (n = 10).

MPB Taxonomic Richness and Assemblage Composition

MPB was extracted using the density separation method (de Jonge 1979), by centrifuging 10 ml of the sample with colloidal silica (LUDOX TM-50, Sigma-Aldrich) diluted to a density of 1.34 g ml−1 at 4 °C and 4000 rpm for 10 min. The lighter particles that were concentrated at the top of the suspension after centrifugation, which included MPB, were siphoned out and rinsed with MilliQ water multiple times to remove the colloidal silica as much as possible, and then sieved through a 500 µm mesh to remove any remaining detritus. After checking the density of MPB in each sample, the MPB composition experiment was conducted by viewing the samples contained in a Sedgewick rafter cell at 400X under a phase-contrast microscope (Primovert Compact Inverted Microscope, ZEISS, Germany). Between 100 and 600 MPB cells were counted per sample. The MPB cells were identified to genus level as much as possible. The MPB taxonomic richness (at genus level) of each sample was measured using the method of rarefaction and extrapolation with Hill numbers (Chao et al. 2014). The asymptote of the rarefaction and extrapolation curve of a sample reflected the estimated number of genera of MPB present. The sampling groups in this analysis are: Summer TK Mangroves (n = 6), Summer TK Tidal Flat (n = 7), Winter TK Mangroves (n = 6), Winter TK Tidal Flat (n = 8), Summer MP Mangroves (n = 8), Summer MP Tidal Flat (n = 10), Winter MP Mangroves (n = 10), and Winter MP Tidal Flat (n = 9).

MPB Photosynthetic Performance

MPB photosynthetic performance was measured using an Imaging-PAM MAXI chlorophyll fluorometer (Walz, Germany), equipped with a red light (650 nm) version of the LED-array-illumination unit IMAG-MAX/LR. The productivity measurements were made for MPB samples collected from TK one day in summer (July 2022) and one day in winter (February 2023). On each occasion, MPB were collected randomly, at least 5 m apart, from the mangrove forest and tidal flat during low tide in the morning, and processed immediately in the laboratory to avoid potential alteration to the MPB’s photosynthetic performance. The sampling groups in this analysis are Summer TK Mangrove (n = 5), Summer TK Tidal Flat (n = 5), Winter TK Mangrove (n = 5), and Winter TK Tidal Flat (n = 5). 3 ml of the sample was sieved through a 500 µm mesh to remove the macrobenthos and macroalgae. The sieved sample was filtered through a 20 µm polycarbonate filter paper (GE Water and Process Technologies, USA), which was carefully placed on a non-reflective plastic sheet and then transferred to the Imaging-PAM for measurement. After five minutes of dark adaptation, the Gain setting was adjusted such that the basal Ft value was 0.1. The Gain setting used in this set of experiment ranged between 3 and 15, while the settings for Measuring Light Intensity and Frequency remained constant at 20 and 1, respectively. An exception occurred for one winter mangrove sample, where the Measuring Light Intensity was set at 12, and both Gain and Measuring Light Frequency at 1, to obtain a basal Ft value of 0.1. For the construction of rapid light curve (RLC), the sample was exposed to incremental irradiance levels (E) of 1, 7, 17, 27, 42, 62, 87, 117, 152, 192, 237, 287, 342, 402, 467, 537, 617, 707, 807, 932, and 1082 µmol photons m−2 s−1 for 10 s each. At the end of each light step, a saturation pulse was applied, and the maximum fluorescence yield in the light-adapted state (Fm’) and the fluorescence yield in the light-adapted state just before the saturation pulse (F) were recorded (Heinz Walz GmbH 2019). The fluorescence values were retrieved from the ImagingWinGigE software (Walz, Germany) for RLC construction, by calculating the rETR at each irradiance level (Perkins et al. 2010a):

$${\text{rETR}}= \frac{{{\text{Fm}}}^{\prime}-{\text{F}}}{{{\text{Fm}}}^{\prime}}\times {\text{E}}$$
(1)

where

rETR = :

relative electron transport rate (unitless)

Fm’ =:

maximum fluorescence yield in the light-adapted state (unitless)

F =:

fluorescence yield in the light-adapted state just before the saturation pulse (unitless)

E =:

incident irradiance level (µmol photons m−2 s−1)

As rETR declined at high irradiance level, the RLC was fitted into the model by Platt et al. (1980):

$${\text{rETR}}={{\text{P}}}_{{\text{s}}} \times \left(1-{{\text{e}}}^{- \frac{\mathrm{\alpha E}}{{{\text{P}}}_{{\text{s}}}}}\right)\times ({{\text{e}}}^{- \frac{\mathrm{\beta E}}{{{\text{P}}}_{{\text{s}}}}})$$
(2)

which allows characterisation of light response by estimating the maximum light utilisation coefficient (α), the maximum photosynthetic activity (rETRmax), the light saturation index (Ek), the irradiance level at which photosynthetic activity is maximum (Em), and the index of photoinhibition (Eb) (Talling 1957; Platt et al. 1980; Sakshaug et al. 1997). Eb obtained from this equation was used to characterize the photo-downregulation activity. α was obtained directly from the model fitting, while the other parameters were calculated as follows:

$${{\text{rETR}}}_{{\text{max}}}={{\text{P}}}_{{\text{s}}}\times \left(\frac{\mathrm{\alpha }}{\mathrm{\alpha }+\upbeta }\right)\times {\left(\frac{\upbeta }{\mathrm{\alpha }+\upbeta }\right)}^{\frac{\upbeta }{\mathrm{\alpha }}}$$
(3)
$${{\text{E}}}_{{\text{k}}}=\frac{{{\text{rETR}}}_{{\text{max}}}}{\mathrm{\alpha }}$$
(4)
$${{\text{E}}}_{{\text{m}}}=\frac{{{\text{P}}}_{{\text{s}}}}{\mathrm{\alpha }}{{\text{log}}}_{{\text{e}}}(\frac{\mathrm{\alpha }+\upbeta }{\upbeta })$$
(5)
$${{\text{E}}}_{{\text{b}}}=\frac{{{\text{P}}}_{{\text{s}}}}{\upbeta}$$
(6)

For each sample, an RLC was constructed and subsequently fit to the Platt et al. (1980) model using Excel Solver. Fittings were good in all cases (R2 > 0.85, p < 0.001).

Coupling Between Dissolved Nutrients, MPB and Mangrove – An Enrichment Mesocosm Experiment

Experimental Set-up and Design

Sediments were collected from the mangrove forest at TK. Mature propagules of Kandelia obovata fallen on the sediment surface (~1 mm depth) were collected from the same site. After returning to the laboratory, sediments were passed through a 1 mm mesh after air-drying to obtain homogenized sediments and to exclude macrobenthic animals. Sieved sediments were moved to fill plastic containers (71 cm × 54 cm × 38 cm) to a depth of about 15 cm. The containers had inlets at the upper side of one wall and outlets at the lower side of another wall close to the sediment surface, and were controlled by timers to maintain semi-diurnal tidal cycles, creating a tidal mesocosm that simulated field mangrove conditions. The mesocosms were established in an outdoor area with little disturbance. The source water was mixed freshwater and seawater pumped from Tolo Harbour. The salinity of the source water was maintained at about 10 to facilitate propagule growth. Propagules of Kandelia obovata were planted in tidal mesocosms following a factorial design, i.e., enrichment (yes/no) × seedling density (high/low) × enrichment time (i.e., every 10 days over one month). There are three replicates for each treatment of enrichment and seedling density, respectively. Sampling was done repeatedly for each mesocosm over the enrichment time. Seedling density included two levels, i.e., four or eight seedlings per mesocosm. The enrichment experiment was started when propagules grew to seedlings with three or more fully expanded leaves. The experiment was designed to test if significant coupling of C and N flow occurred between mangrove seedlings and the MPB with the use of isotopically labelled tracers. Mangrove seedlings were foliar labelled with enriched C and N stable isotopes (Putz et al. 2011). A 97atom% 13C and 2atom%15N urea solution was prepared by dissolving 250 mg 99atom% 13C and 5 mg 98atom%15N urea (Sigma Aldrich, Vienna, Austria) in 125 ml MilliQ water. To facilitate the contact of the C and N enriched solution with the seedling leaves, 30 µl of wetting agent was added. For the control solution, 255 mg unenriched urea (Sigma Aldrich) was dissolved in 125 ml MilliQ water with 30 µl of wetting agent. Small amounts of enriched and unenriched urea solutions were applied by brushes daily to the leaves of mangrove seedlings in each of the paired treatment (n = 6) and control mesocosms (n = 6), respectively.

Sample Collection and Analysis

Leaf, root, sediment and porewater samples were harvested from the mesocosms and sampled every 10 days. During the low tide period, small amounts of leaves and roots were cut from the seedlings by scissors. Surface sediments were collected by a small shovel. Porewater sample was collected using a 50 ml syringe after being purged three times from a drilled hole in the sediment, and then stored in 40 ml gas-tight high-density polyethylene glass vials pre-combusted at 450 °C for three hours and poisoned with HgCl2 solution. Separate scissors, shovels and glass vials were used for enriched and unenriched mesocosms to avoid cross-contamination. After returning to the laboratory, roots were rinsed in MilliQ water to remove sediments on the surface. Leaves, roots and small amounts of sediments were dried at 60 °C until constant weight. The MPB were separated from the rest of the sediments following the method as described earlier. Their C and N contents, δ13C and δ15N were measured by a EuroVector Elemental Analyser - Nu Perspective IRMS at The University Hong Kong, with iACET standards used for quality control check. For the δ13C analysis of porewater dissolved inorganic carbon (DIC), 40 ml of porewater was passed through a glass fibre filter (Whatman GF/F filter, 0.7 mm pore size, 47 mm diameter). Porewater DIC and total nitrogen (N) concentrations were measured by a Shimadzu TOC analyser. For the measurement of porewater δ13C-DIC, the vials were injected with N2 gas to displace 10 ml porewater and generate a 10 ml headspace, and then 2 ml phosphoric acid (> 99% purity) was injected to the vials via needles. The samples were mixed and kept at 4 °C overnight. Headspace air samples in the vials were drawn and the δ13C-CO2 was measured by a Picarro G2201-i Cavity Ringdown Spectroscopy analyser (Picarro Inc., USA) as described by Ouyang et al. (2021) the following day. The δ13C-CO2 values presented hereafter are the δ13C-DIC of porewater samples.

The %atom in excess of C and N for plant, sediment and MPB were calculated as:

$$\mathrm{atom}\;^{13}\mathrm{C\; \% \;in\; excess}=\mathrm{atom}\;^{13}\mathrm{C\; \% \;in\; enriched\; sample}-\mathrm{atom}\;^{13}\mathrm{C\; \% \;in\; control}$$
(7)
$$\mathrm{atom}\;^{15}\mathrm{N\; \% \;in\; excess}=\mathrm{atom}\;^{15}\mathrm{N\; \% \;in\; enriched\; sample}-\mathrm{atom}\;^{15}\mathrm{N\; \% \;in\; control}$$
(8)

Specific uptake rate (V, h−1) of the samples was calculated as:

$${\text{V}_\text{c}}=\mathrm{atom}\;^{13}\mathrm{C\; \% \;in\; excess}/(\mathrm{atom}\;^{13}\mathrm{C\; \% \;enrichment}\times \mathrm{enrichment\; time})$$
(9)
$${\text{V}_\text{N}}=\mathrm{atom}\;^{15}\mathrm{N\; \% \;in\; excess}/(\mathrm{atom}\;^{15}\mathrm{N\; \% \;enrichment}\times \mathrm{enrichment\; time})$$
(10)

Statistical Analysis

As the MPB biomass data violated the assumptions for normality (verified by Shapiro-Wilk normality test) and homoscedasticity (verified by Levene’s test), Kruskal-Wallis test and post hoc Dunn’s test Holm-adjusted P-values were used to identify differences among sampling groups. Principal component analysis (PCA) was used to visualise the patterns of environmental factors and MPB biomass in the sampling groups; here, the chlorophyll a data was log-transformed to satisfy the assumption of normality. Spearman’s correlation was used to determine the relationship between MPB biomass and the environmental parameters. The rarefaction and extrapolation with Hill numbers for estimating MPB genus richness were performed using the iNEXT package in R (Chao et al. 2014; Hsieh et al. 2020; R Core Team 2020). Difference in MPB richness among sites, habitats, and seasons was then tested with three-way analysis of variance (ANOVA). Permutational multivariate analysis of variance (PERMANOVA) was conducted to ascertain the differences in MPB assemblages between sampling groups, which was visualized with the canonical analysis of principal coordinates (CAP) plot. Similarity percentages (SIMPER) analysis identified the genera typifying each sampling group (Clarke and Gorley 2015). The distance-based linear modelling (DistLM) sequential test was used with the step-wise method and adjusted R2 to identify all environmental parameters that imposed a significant influence on the MPB assemblage, and then visualized with the distance-based redundancy analysis (dbRDA). All multivariate analyses were performed on the PRIMER v7 software package (Anderson et al. 2008; Clarke and Gorley 2015).

Regression analysis was used to examine the relationship between (1) δ13C of MPB and porewater DIC concentrations as well as porewater δ13C-DIC; (2) δ13C of MPB and δ13C of sediments or sediment C/N or MPB C/N, VC and porewater DIC concentrations; and (3) δ15N of MPB and MPB C/N. The assumption of normality was tested by the Shapiro-Wilk normality test before regression analysis (α = 0.05). Two-way ANOVA was performed to examine the difference in (1) atom 13C % and 15N % in excess, Vc and VN across seedling density and enrichment time which is a repeated measure, (2) δ13C and δ15N of MPB across the factors ‘enrichment’ and ‘enrichment time’, and (3) the light response parameters between seasons and habitats. For (1) and (2), sample types included roots, leaves, porewater, sediments and MPB. Where there was a significant treatment effect, Tukey’s HSD test was used to further examine the significant difference. Before ANOVA, the assumptions of normality and heterogeneity were tested by the Shapiro-Wilk normality test and Levene’s test, respectively. Data were transformed (e.g., log-transformation) if the assumptions could not be met. When transformed data could not meet the assumptions, Kruskal-Wallis test was used, followed by the Mann-Whitney test. Paired-sample t-test was performed to examine the difference in porewater δ13C-DIC, δ13C and δ15N of different sample types. The light response parameters were also analysed using Welch’s t-test for differences between habitats within a given season. Data were presented as mean ± standard error (SE).

Results

MPB Community Ecology

The patterns of environmental parameters and MPB biomass among the sampling groups are visualized with PCA (Fig. 1). The first two PCs (PC1 and PC2) explained a cumulative 65.0% of the variation, with PC1 accounting for 42.4% and PC2 22.6%. Summer and winter samples were clearly separated along PC1, which suggests a negative correlation of MPB biomass with porewater and surface soil temperatures. Along PC2 were gradients of porewater salinity, porewater pH, and irradiance level.

Fig. 1
figure 1

Results of principal component analysis (PCA) of five measured environmental factors (porewater salinity (ppt), porewater temperature (°C), porewater pH, surface soil temperature (°C), and irradiance level (µmol m−2 s−1)) and MPB biomass expressed as chlorophyll a concentration (µg g−1 dry sediment) for all sampling groups combined

Full size image

The Spearman’s correlation analysis supported a negative correlation between MPB biomass and surface soil temperature (rs(76) = -0.580, P < 0.001), porewater temperature (rs(76) = -0.519, P < 0.001), porewater pH (rs(76) = -0.234, P = 0.039), as well as irradiance level (rs(76) = -0.233, P = 0.040), but no correlation was identified for porewater salinity (rs(76) = -0.046, P = 0.687). There was no evidence that MPB biomass in the mangroves was lower than the tidal flats in any occasion (Fig. 2a). Notably, mangrove MPB biomass was found to exceed tidal flat’s in TK winter (8.3 ± 1.0 µg g−1 dry sediment in mangroves and 3.8 ± 0.3 µg g−1 dry sediment in tidal flats, z = 2.852, Pholm = 0.046).

Fig. 2
figure 2

Biomass and generic diversity of the MPB community of each sampling group. a Biomass represented by chlorophyll a concentration (µg g−1 dry sediment); b MPB genus richness estimated by rarefaction and extrapolation with Hill numbers

Full size image

The estimated median number of genera of MPB in each sampling group varied between 23 and 45 (Fig. 2b). The three-way ANOVA (Table 1) revealed no difference in MPB genus richness across all sites, habitats, and their interactions, including those across seasons. A main effect of seasons was found (F1,56 = 7.896, P = 0.007), where the MPB genus richness was higher in winter (39.1 ± 2.4) than in summer (29.5 ± 2.1). The PERMANOVA main test results indicated distinctions in the MPB genus composition among all habitats, sites, and seasons, as well as all interactions of the factors (Table 2). The post hoc pairwise test also indicated distinctions among all analogous sampling groups (Table 3). Generally, habitat differences (i.e., between tidal flat and mangrove) were smaller than those between seasons (summer and winter) while there were consistent differences between MP and TK (Fig. 3).

Table 1 Results of the three-way ANOVA of MPB genus richness in different seasons, sites, and habitats
Full size table
Table 2 PERMANOVA main test results for MPB genus assemblage
Full size table
Table 3 PERMANOVA post hoc pairwise test results for MPB genus assemblage
Full size table
Fig. 3
figure 3

A canonical analysis of principal coordinates (CAP) plot illustrating the separation of MPB assemblages at different sampling locations in different seasons. Representation of the data is based on Bray-Curtis similarity indices of square root-transformed MPB genus counts

Full size image

The genera typifying the MPB assemblage of each sampling group are given in Supplementary Table 1. Pinnularia was the dominant genus in MP, regardless of habitats and seasons. In contrast, TK habitats were dominated by Navicula, Nitzschia, Pinnularia, and Amphora to different extents. Another notable difference was with the occurrence of the cyanobacteria Anabaena/Nostoc as a typifying genus in three out of four sampling groups of MP, yet was absent in those of TK. Overall, Pinnularia, Navicula, and Nitzschia were among the typifying genera in the mangroves and tidal flats of TK and MP in both seasons.

The DistLM sequential test indicated that all five environmental factors (mean values and ranges specified in Supplementary Table 2) contributed to the variation in MPB assemblages among the sampling groups (Table 4). Porewater salinity had the greatest contribution (12.35%), followed by porewater temperature (8.21%), irradiance level (3.94%), porewater pH (2.64%), and surface soil temperature (2.40%), together explaining 23.47% (adjusted R2) of the overall variation in the MPB assemblages. The two axes of the dbRDA plot (dbRDA1 and dbRDA2) explained 21.7% of the variation in MPB assemblages (Supplementary Fig. S1). Most variation was along dbRDA1 (13.1%), representing the salinity gradient and to a lesser extent, irradiance level.

Table 4 Sequential test results of the distance-based linear modelling (DistLM)
Full size table

MPB Photosynthetic Performance

For both seasons, the RLC for tidal flat was markedly taller and wider than that of mangrove, with rETR approaching zero at a higher irradiance level (Fig. 4). Furthermore, the curves of summer MPB were taller and wider than those of winter MPB for both habitats, implying that the MPB had higher photosynthetic capacities in summer.

Fig. 4
figure 4

Rapid light curves (RLC) of MPB collected from TK mangrove and TK tidal flat in summer and winter, fitted to the model by Platt et al. (1980). Error bars represent ± SE

Full size image

These distinctions could be quantified with the light response parameters (Supplementary Fig. S2). In summer, the tidal flat MPB exhibited higher (by ~30%) Ek, rETRmax, Em, and Eb than mangrove MPB, but not α (Table 5). An equal α implied that mangrove MPB had equal photosynthetic efficiency as tidal flat MPB at limiting irradiance levels. MPB from the two habitats in winter also showed equivalent α, as well as all other light response parameters (Table 6). This might be due to the influence of the outliers among the winter tidal flat samples. The two-way ANOVA identified main effects of season and habitat for most light response parameters, but no interaction effect (Table 7). The exception lied with α, where a main effect was found for seasons (F1,16 = 59.910, P < 0.001) but not for habitats (F1,16 = 1.820, P = 0.196).

Table 5 Results of Welch’s t-test for the light responses exhibited by MPB collected from TK mangrove and tidal flat in summer
Full size table
Table 6 Results of Welch’s t-test for the light responses exhibited by MPB collected from TK mangrove and tidal flat in winter
Full size table
Table 7 A summary of two-way ANOVA results for each light response parameter
Full size table

Mesocosm Experiment: Isotopic Enrichment of Samples

δ13C and δ15N of different sample types were significantly higher in enriched mesocosms than unenriched ones (Table 8). Specifically, MPB δ13C and/or δ15N values were significantly higher in enriched (δ13C = -20.6‰ ± 1.0‰, δ15N = 27.7‰ ± 4.7‰) than unenriched (δ13C = -26.5‰ ± 0.2‰, δ15N = 0.3‰ ± 1.2‰) mesocosms (Paired-sample t test, n = 18, P = 1.7 × 10–5 (δ13C), P = 1.8 × 10–5 (δ15N)). Porewater δ13C-DIC was very significantly higher in enriched (12.2‰ ± 1.4‰) than that of unenriched mesocosms (-0.5‰ ± 0.4‰) (Paired-sample t test, n = 18, P = 2.7 × 10–7). Similarly, leaf δ13C and/or δ15N values were significantly higher in enriched (δ13C = 22.4‰ ± 11.8‰, δ15N = 766.9‰ ± 94.1‰) than unenriched mesocosms (δ13C = -27.8‰ ± 0.4‰, δ15N = 6.8‰ ± 0.3‰) (Paired-sample t test, n = 18, P = 5.3 × 10–4 (δ13C), P = 3.2 × 10–7 (δ15N)). Root δ13C and/or δ15N values were significantly higher in enriched (δ13C = -18.3‰ ± 2.4‰, δ15N = 99.1‰ ± 27.2‰) than unenriched mesocosms (δ13C = -25.4‰ ± 0.3‰, δ15N = 6.2‰ ± 0.3‰) (Paired-sample t test, n = 18, P = 8.7 × 10–3 (δ13C), P = 3.3 × 10–3 (δ15N)). Sediment δ13C and/or δ15N values were also significantly higher in enriched (δ13C = -10.1‰ ± 2.5‰, δ15N = 35.9‰ ± 7.0‰) than unenriched mesocosms (δ13C = -22.0‰ ± 0.7‰, δ15N = -15.6‰ ± 2.2‰) (Paired-sample t test, n = 18, P = 2.8 × 10–4 (δ13C), P = 1.6 × 10–6 (δ15N)).

Table 8 Differences in δ13C and/or δ15N of porewater DIC, MPB, leaves, roots and sediments between enriched and unenriched samples
Full size table

Relationships Between δ13C, δ15N, Vc and VN of MPB and Porewater/Sediment Properties

Our results showed that δ13C of MPB was related to porewater DIC and sediment stoichiometry. There was a significant relationship between δ13C of MPB and porewater DIC concentrations as well as δ13C-DIC (R2 = 0.54, n = 18, P = 0.0028, Fig. 5). Further, porewater DIC was a much more important variable than porewater δ13C-DIC in explaining the variance in δ13C of MPB (46.1 vs 8.3%). Moreover, δ13C of MPB had significant relationships with δ13C of sediments (R2 = 0.34, n = 18, P = 0.018, Fig. 6a), sediment C/N (R2 = 0.3, n = 18, P = 0.027, Fig. 6b), as well as MPB C/N (R2 = 0.52, P = 0.001, Fig. 6c). There were also very significant relationships between δ15N of MPB and MPB C/N (R2 = 0.49, n = 18, P < 0.01, Fig. 6d), and between MPB Vc and porewater DIC concentrations (R2 = 0.5, n = 18, P = 0.001, Fig. 6e). There was a significant exponential relationship between MPB VN and porewater total N concentrations (R2 = 0.41, n = 18, P = 0.011, Fig. 6f).

Fig. 5
figure 5

The relationship between δ13C of MPB and porewater DIC concentrations and DIC δ13C

Full size image
Fig. 6
figure 6

Relationships between δ13C, δ15N, Vc and VN of MPB and porewater or sediment properties. TN denotes total N. The shaded area represents the 95% confidence intervals of the regression relationships

Full size image

Variation in Atom 13C% and 15N% in Excess, Vc and VN with Seedling Density and Enrichment Time

Atom 13C% and 15N% in excess, Vc and VN of some sample types varied with enrichment times while all did not vary with seedling density. Atom 13C% in excess of MPB varied significantly with enrichment times (ANOVA, F-value = 7, P = 0.0097, Fig. 7), as did atom 15N% in excess (ANOVA, F-value = 12.43, P = 0.001), Vc (ANOVA, F-value = 12.36, P = 0.001) and VN (ANOVA, F-value = 13.81, P = 0.0008). In particular, atom 13C% and 15N% in excess, Vc and VN of MPB measured during the second enrichment was significantly higher than those measured during the first (Tukey’s HSD test, P = 0.0007) and third enrichment (Tukey’s HSD test, P = 0.02) but the latter two were not significantly different (Tukey’s HSD test, P = 0.19). Similarly, atom 13C% in excess of roots varied significantly with enrichment times (ANOVA, F-value = 5.88, P = 0.017), as did atom 15N% in excess (ANOVA, F-value = 5.88, P = 0.02), Vc (ANOVA, F-value = 5.8, P = 0.017) and VN (ANOVA, F-value = 5.79, P = 0.017). Atom 13C% and 15N% in excess, Vc and VN of roots measured during the third enrichment was significantly higher than those measured during the first (Tukey’s HSD test, P = 0.022) and the second enrichment (Tukey’s HSD test, P = 0.04) while the latter two were not significantly different (Tukey’s HSD test, P = 0.94). There were no significant differences for other comparisons with the factor ‘enrichment time’ (Kruskal-Wallis test, P = 0.44). Further, atom 13C% and 15N% in excess, Vc and VN of all sample types did not vary with seedling density (Paired-sample t test, P > 0.05).

Fig. 7
figure 7

Differences in atom 13C% and 15N% in excess, Vc and VN among different sample types and enrichment times. a and b atom 13C% and 15N% in excess, c and d Vc and VN. L, MPB, R and S denote leaves, MPB, roots and sediments, respectively. Groups with significant differences were labelled by different letters

Full size image

Discussion

Mangrove MPB are Abundant and Different from Tidal Flat MPB

This study provides novel insights into the characteristics of MPB inhabiting the mangrove forest, which are often overlooked. We found that (1) the MPB biomass in mangroves was the same, if not higher, than on the adjoining tidal flats regardless of seasons; (2) the MPB generic richness was equal between mangroves and tidal flats, regardless of seasons; (3) the MPB generic compositions varied with environmental conditions, including a change in seasons; and (4) the photosynthetic performance of mangrove MPB was characteristic of acclimation to lower light levels.

A higher MPB biomass in mangroves than the adjacent tidal flats was also reported by Kwon et al. (2020) in Cambodia and Australia, and by Chen et al. (2019) in southern China. These, along with the findings of this study, serve as evidence against the notion that light availability is the major limiting factor of MPB biomass in mangrove sediment (Alongi 1994). There are several explanations for MPB’s ability to thrive under lower light levels inside the mangrove forests. At the cellular level, MPB allegedly change their pigment composition to adapt to the local light regime (Jesus et al. 2009), or they may possess the ability to expand their photosynthetic unit, and therefore the chlorophyll a content, under low light, as has been reported in the microalgae Nannochloropsis sp. (Fisher et al. 1998). Physically, mangrove trees would dampen tidal action and trap sediment (Kathiresan 2003; Lee et al. 2014), consequently reducing erosion and enhancing the deposition of MPB on the forest floors, as well as providing a stable substrate for continual growth. Furthermore, since the photosynthetic activity of suspended MPB was found to saturate at relatively low light levels of 30 to 360 µmol m−2 s−1 (MacIntyre et al. 1996), the tidal flat environment may not be as ideal as previously assumed. Exposure to intense irradiance may cause photo-downregulation, or, if prolonged, photoinhibition, where irreversible damage is made to the photosynthetic pigments of the MPB. The MPB have well-known adaptive features to avoid photo-downregulation and photoinhibition, by a combination of vertical migration within the sediment column (Blanchard and Cariou-Le Gall 1994; Serôdio et al. 1997, 2008; Underwood et al. 1999; Underwood and Kromkamp 1999; Perkins et al. 2010b) and non-photochemical quenching (NPQ) of light energy (Lavaud et al. 2007; Serôdio et al. 2008). The mean irradiance levels in the mangrove forests of MP (60.0 ± 11.3 µmol m−2 s−1) and TK (596.0 ± 113.7 µmol m−2 s−1) lied closer to the favorable range than those of the tidal flats, which frequently surpassed 1000 µmol m−2 s−1 in both seasons, suggesting that the mangrove canopy might in reality promote MPB growth by alleviating the light stress. On the tidal flat, the MPB might encounter additional stressors, such as frequent salinity changes and desiccation during low tide (Underwood 2002), also extreme temperature fluctuations that can exacerbate the photoinhibitory effects (Bártolo et al. 2023). As a result, the photosynthate of tidal flat MPB might be spent on processes that counteract these stressors, leaving less energy for cell growth and division.

It is well documented that the MPB are dominated by pennate diatoms (Serôdio et al. 1997; de Jonge et al. 2012; Serôdio and Paterson 2021), which was also demonstrated in the present study. In this study, centric diatoms and cyanobacteria occurred occasionally, and chlorophytes less often. Dinoflagellates were present but rare. The MPB generic makeup was significantly different across all habitats, sites, and seasons, and porewater salinity, porewater temperature, irradiance level, porewater pH, and surface sediment temperature (in decreasing order of importance) were identified as significant contributors to such differences. The literature reports on several other factors known to influence MPB assemblage, including sediment grain size (Underwood 2002), sediment resuspension (Underwood 1994), as well as tolerance to stressors like ammonia, sulphide, and anaerobiosis (Admiraal and Peletier 1979). Availability of sediment organic matter (Underwood and Kromkamp 1999; Yang et al. 2003; Virta et al. 2019) and nutrients such as nitrogen (Underwood et al. 1998; Underwood and Kromkamp 1999; Virta et al. 2019), phosphorous, and silicate (Pan et al. 1996; Du et al. 2017; Virta et al. 2019) are likewise determinants of MPB community composition. The common occurrence of cyanobacteria in MP but not in TK might be due to the lower total nitrogen to total phosphorus ratio in MP waters (~10; vs ~17 in TK) (Smith 1983; Levich 1996).

Tidal flat MPB generally showed higher Ek, rETRmax, Em, and Eb values than mangrove MPB, which was characteristic of high-light acclimated MPB assemblages (Underwood 2002; Perkins et al. 2006). A high rETRmax protects the MPB against photodamage to their photosystem (Ralph et al. 1999), which may be particularly beneficial for tidal flat MPB. This high level of photosynthetic capacity may not translate to a high MPB biomass, however, possibly due to the diversion of energy to counteract stressors on the tidal flat. The finding of mangrove MPB exhibiting a lower rETRmax is consistent with the previous observation of a lower productivity in mangrove MPB (Kwon et al. 2020). A lower Eb of mangrove MPB suggested that these MPB experienced more intense photo-downregulation than their counterparts on the tidal flat, which also implied a capacity for stronger NPQ activity in the mangrove assemblage. Indeed, diatoms inhabiting an environment with unstable light exposure (i.e., the occasional shading of light by the mangrove canopy) are found to exhibit stronger NPQ at inhibitory irradiances than those with more stable light exposure (Lavaud et al. 2007). These results demonstrated that MPB from different habitats (i.e., mangrove forests and tidal flats) were well acclimated to the local light levels. One deviation from the literature was with α. While MPB assemblages adapted to lower light generally show a higher α (Morris and Kromkamp 2003; Perkins et al. 2006), in this study, the mangrove and tidal flat MPB had equivalent α values, suggesting that the two assemblages had similar photosynthetic efficiency at limiting irradiances.

These light response measurements may not directly reflect the MPB’s in situ performance, because there are several ways in which they may be influenced by the in situ conditions. Firstly, any shift in the MPB assemblage resulting from changes in environment conditions, including a change in seasons, may alter these measurements as photosynthetic performance is unique to each MPB species (Barranguet et al. 1998; Underwood and Kromkamp 1999; Underwood et al. 2005; Vieira et al. 2013). Secondly, each MPB cell’s position inside the sediment may have an impact on the measured light responses due to the optical characteristics of the non-algal materials in the sediment, such as light scattering, light absorbance, and fluorescence emission (Underwood and Kromkamp 1999; Underwood 2002). These effects would be further complicated by the sizes of these substances and the mineral content of the sediment (Blanchard and Montagna 1995; Serôdio et al. 1997). Moreover, there are evidence of environmental influences on MPB’s light response measurements, in which rETRmax and Ek are affected by total inorganic nitrogen, phosphate, silicate (Underwood 2002), and iron (Geider et al. 1993) concentrations, and α by phosphate concentration (Underwood 2002). Furthermore, it has been shown that MPB’s Ek (Serôdio et al. 2005) and other photophysiological parameters, namely the effective quantum yield of PSII (ΔF/Fm’) and the absorption spectrum of PSII (σII) (Frankenbach et al. 2020), vary rapidly in response to the incident irradiance level, changing the other light response parameters as a result. This may be especially relevant as the MPB encounter periods of submergence. Hence, another photosynthetic performance analysis done during the submergence period would help to construct the full picture. Nonetheless, we expect that the light response measurements presented in this study would reflect the general trends of those in situ (i.e., tidal flat MPB consistently showing light response parameters characterizing higher light acclimation than mangrove MPB), despite that these parameters may vary to different extents depending on the environmental circumstances.

Coupling Between Mangrove-Derived Nutrients and MPB

The results of the mesocosm experiment suggested the presence of a flow pathway of MPB nutrient acquisition via the leaf-root-porewater-MPB continuum. While microbial breakdown of DOM to dissolved nutrients may exists, this process occurs mainly in the porewater. δ13C and/or δ15N values of leaves, roots, porewater and MPB were all higher in 13C- and 15N-enriched mesocosms than those in unenriched mesocosms. Only mangrove leaves were 13C- and 15N-enriched directly in the enriched mesocosms. Therefore, other components (including roots, porewater and MPB) could only be 13C- and 15N-enriched via the leaf-root-porewater-MPB continuum. This was corroborated by the indirect 13C-enrichment of roots and sediments observed in another mesocosm experiment (Ouyang et al. 2018). Further, δ13C and/or δ15N values of MPB were closely and negatively related to the DIC concentrations and δ13C values of porewater. In the mesocosms, MPB utilization of sediment porewater DIC resulted in decreased DIC concentrations and thus lower porewater DIC δ13C. Where δ13C values of MPB were high, DIC δ13C values were also high and vice versa (Curry et al. 2020). δ13C and δ15N values of MPB were also closely and negatively related to MPB C/N. An increase in MPB C/N might be due to greater C assimilation from higher DIC, which could decrease MPB’s affinity for inorganic C and lower the expression of CO2 concentrating mechanisms (Raven et al. 2011; Reinfelder 2011), subsequently resulting in lower MPB δ13C values. Higher MPB δ13C values were also observed in estuaries with higher local DIC concentrations and concomitant absence of CO2 concentrating mechanisms (Vieira et al. 2016).

The results of this study showed that specific N uptake rates of MPB were positively related to porewater TN concentrations, which was consistent with the view that MPB was mainly supported by nutrients in porewater (Brito et al. 2010). Labile fractions of dissolved organic N (e.g., urea and dissolved free amino acids) could be important sources of N for MPB (Tuchman et al. 2006). It has been shown that algae uptake could account for ~55 to 90% of dissolved organic N uptake by MPB and inorganic N (i.e., ammonia) is a more preferred N source than organic N (Sundbäck et al. 2011). MPB largely incorporates dissolved inorganic N at the sediment surface when ammonium concentrations are the highest (Longphuirt et al. 2009). Further, δ13C values of MPB were negatively related to sediment C/N. Algae uptake of dissolved organic N may be dominant in the photic zone of the sediments (Veuger and Middelburg 2007). Greater assimilation of sediment N resulted in lower sediment N content and higher C/N. Greater assimilation of sediment N was also concomitant with the assimilation of sediment C, which had much lower δ13C values, confirming the findings of Bouillon et al. (2008b).

We found that component 13C and/or 15N of MPB decreased in the sequence of leaves, roots, sediments, porewater DIC and MPB in the enriched mesocosms. Leaves were directly enriched by 13C and 15N, which were transferred to roots, exuded in porewater, and subsequently assimilated by MPB. During this process, porewater served as a media for the tight coupling between MPB and mangrove-derived nutrients. Our findings suggested that MPB in mangroves could obtain mangrove-derived nutrients via the leaf-root-porewater-MPB continuum, challenging the existing paradigms that MPB would only obtain nutrients from porewater alone (Brito et al. 2010; Sundbäck et al. 2011). Our results finds the C and N acquisition pathway of MPB which is also applicable to real field conditions. However, the in situ conditions are much complicated with changing environmental parameters. In situ specific C and N uptake rates of MPB may be different from mesocosm experiments, and remain to be tested in future field experiments.

This study presents novel perspectives on the abundant and rich assemblage of MPB inside the mangrove forest, as well as how these assemblages are highly dynamic in terms of adaptation to environmental conditions, while distinct from tidal flat MPB in terms of generic make-up and photosynthetic performance. Mangrove MPB’s high abundance and ability to maintain high productivity under low irradiance levels allude to their potential as a significant contributor to mangrove trophodynamics. Furthermore, this study finds a new pathway of the leaf-root-porewater-MPB continuum by which MPB can obtain C and N. Therefore, mangrove MPB serve as a driver of biogeochemical cycling of C and N which has been ignored in the past global conceptual models sha** C and N cycling in mangroves. We highlight that MPB should be included in the nutrient cycling process of mangroves.