Prévia do material em texto
Ecological Modelling 147 (2002) 69–83 An empirical model of the fate of organic carbon in a mangrove forest partly affected by anthropogenic activity John F. Machiwa a,*, Rolf O. Hallberg b a Department of Zoology and Marine Biology, Uni�ersity of Dar es Salaam, PO Box 35064, Dar es Salaam, Tanzania b Department of Geology and Geochemistry, Stockholm Uni�ersity, S-106 91 Stockholm, Sweden Received 14 June 1999; received in revised form 28 April 2000; accepted 27 June 2001 Abstract A model of biogeochemical and hydrological processes that drive organic carbon dynamics of a mangrove ecosystem has been developed. Tidal regime parameters as well as biological factors related to macrofauna and microbiota are described within the model. The model includes three sub-models, each representing a major form of organic carbon in the mangrove ecosystem; litter organic carbon, particulate organic carbon and dissolved organic carbon (DOC). Results from the model reveal that discharge of DOC to the adjacent ocean may be one of the dominant outputs of a mangrove forest. The model confirms the observed data that DOC accounts for 80% of total organic carbon export. In order to balance the standing crop of DOC in the marine fringe zone, the model suggested that 40% is removed in this zone by microorganisms as well as resident and tidal migrant fauna. © 2002 Elsevier Science B.V. All rights reserved. Keywords: Crabs; Dissolved organic carbon; Empirical model; Litter; Mangroves; Particulate organic carbon www.elsevier.com/locate/ecolmodel 1. Introduction Summers and McKellar (1981) made a sensitiv- ity analysis of carbon flows in an estuarine envi- ronment. Their model consisted of 19 compartments within four well-defined subsys- tems interacting through distinct fluxes. The model took into consideration the time variable and could be used for predictive purposes. A model of phosphorus dynamics in a Spartina salt marsh was developed by Pomeroy et al. (1972). The model consisted of five compartments, the external forcing function was the physical removal of Spartina. Lugo and Snedaker (1974) modelled the effects of nutrient runoff on mangrove pro- duction, and the influence of tides on mangrove production and export. Their model also consid- ered the consequences of mangrove succession on nutrient content of surrounding waters. They used non-linear differential equations in simulation of a three-compartment system. State variables which were used in the model were light, tide, and dissolved oxygen. Jacobi and Schaeffer-Novelli (1990) developed a conceptual model on the fate * Corresponding author. Fax: +255-5141-0038. E-mail addresses: jmachiwa@ucc.udsm.ac.tz (J.F. Machiwa), rolf.hallberg@geo.su.se (R.O. Hallberg). 0304-3800/02/$ - see front matter © 2002 Elsevier Science B.V. All rights reserved. PII: S 0304 -3800 (01 )00407 -0 mailto:jmachiwa@ucc.udsm.ac.tz mailto:rolf.hallberg@geo.su.se J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8370 of oil spills in mangroves. Detritus, inorganic and organic particulate matter are state variables or system features that influence oil storage in the system. Tidal flush is the main forcing function affecting residence time of oil constituents. Pilette and Kincaid (1992) used a non-cyclic approach to analyse carbon transfer in a small hypothetical ecosytem. Models derived from empirical data which integrate structural and functional ecosys- tem characteristics are rare. For instance, Wort- mann et al. (1997) modelled the vegetative spread of sea-grass in estuarine environment, and Duarte and Ferreira (1997) simulated macro-algae popu- lation dynamics in a sub-tidal area. Few studies, if any, have used a holistic ecosystem analysis to model structures and energy flows in mangrove systems. Most of the previous works had suffi- cient temporal resolutions, but they were of a limited spatial extent and the coverage was on just few aspects of the ecosystem. Dynamics of macrodetritus and benthic nutri- ents in mangrove ecosystems have been investi- gated by several workers (e.g. Lugo et al., 1976; Boto and Bunt, 1981; Twilley, 1985; Alongi et al., 1989; Lee, 1990; Alongi, 1996). For instance, the influence of tidal regime and benthic organisms on organic matter exchange has been well docu- mented. It has also been reported that the extent of tidal export depends on the amplitude and frequency of tidal inundation. Golley et al. (1962) used mass balance approach to estimate tidal exchange of material between a mangrove stand and adjacent marine biotopes. Cundell et al. (1979) and Robertson (1988) concluded that the amount of organic matter, which was available to coastal and offshore higher trophic level, was dependent on the sum of production, leaching, decomposition and detrital storage in a mangrove system. Invertebrates are an important factor in regulating the magnitude of organic carbon ex- port (Robertson, 1986; Lee, 1989; Robertson and Daniel, 1989; Robertson et al., 1990; Camilleri, 1992; Emerson and McGwyne, 1992). Indeed, the grapsid crabs are more efficient in retention of autochthonous organic matter and litter turnover than purely microbial decay process (Macnae, 1968; Leh and Sasekumar, 1985; Robertson, 1986, 1988). Microbiological degradation of litter pro- duce dissolved organic carbon (DOC), that forms a high proportion of material export from man- grove areas (Twilley, 1985). A review by Muller (1998) discussed the connectivity of structural and functional characteristics of an ecosystem. For a holistic ecosystem comprehension, the author em- phasised analysis of ecological gradients in physi- cal, chemical and biological compartments of the system. In this study, we used a modelling approach to investigate the fate of organic carbon in a man- grove ecosystem. We have shown that, modelling can reliably estimate some parameters that are difficult to quantify under field conditions and can explain processes that can not be well constrained by field measurements (e.g. Muller, 1998). 2. Materials and methods 2.1. Construction of the model In ecosystem simulation, a system comprises several compartments, in this case specific parts of the mangrove forest like the terrestrial fringe (TF) or the crabs. Compartments are connected via flows whereby they act in full or partial depen- dency on each other to contain a certain amount of carbon. Inputs were litter fall in the mangrove area and litter as well as particulate organic car- bon (POC) and DOC from the ocean. The ocean was used as a sink for all outputs from the mangrove forest, Fig. 1. These inputs and outputs are named variables and can be a value according to an equation, which is calculated as the model is run. In our case, we used annual averages accord- ing to our observations. Knowledge of internal processes that are connected with input and out- put, their functions as well as their limitations is of fundamental importance in ecological modelling. In order to develop a model for the fate of organic matter in an ecological system, we first considered a conceptual scheme of the organic carbon dynamics in the forest (Fig. 2). The proce- dure involved in the construction of this ecologi- cal model includes identification of the main compartments, and measurement of their sizes. J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 71 Major fluxes, their magnitudes and directions were also determined. System variables and boundary conditions were considered during the field surveys. Quantitative data regarding various aspects of matter turnover were then accumulated over a period of 2 years (Table 1). The subsystems were the ecologically distinct zones in the man- grove forest: Marine fringe (MF) zone was colonised by a nearly pure stand of Sonneratia alba. Trees are large and widely spaced, they harbour dense beds of oysters (Machiwa and Hallberg, 1995). The Fig. 1. Parent model of the fate of organic carbon indicating flows (Mg org.C year−1) to or fromsub-models and the ocean. J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8372 Fig. 2. Generalised section of Maruhubi forest showing location of the sub-areas in relation to tidal level (HST=highest spring tide; MHT=mean high tide; MT=mean tide level; MLT=mean low tide). Vegetative structure of the forest is represented by relative heights (not to scale) of trees and shrubs. The conceptual diagram of organic carbon dynamics is shown (A= total litter production in each mangrove sub-area; C=grapsid crab density; L= litter standing crop; S=organic carbon content of the sediment; thick arrows indicate direction of flow and magnitude (not to scale) of LOC, POC and DOC). sandy/muddy substrate supports a high popula- tion density of ocypodid crabs and other benthos. Shoreline (SL) zone was composed of mixed vegetation of S. alba and A�icennia marina. Trees were relatively short, with few crabs and oysters. Landward extent (LE) zone was situated land- ward adjacent to the SL. Half of its area com- prised a mixed stand of short A. marina and Ceriops tagal plants. Another half harboured an almost monospecific stand of densely growing C. tagal shrubs. Relatively few ocypodid and sesarmid crabs were found in this zone. For mod- elling purposes, the SL and LE zones were com- bined to form a common SL–LE compartment. This was necessary because of the inherent difficulties of sampling of suspended particulate matter (SPM) and macrodetritus transport in the inner zones of the forest. TF zone had an almost pure stand of large A. marina trees. Mangrove floor macrofauna in- cluded big sesarmid crabs. Each zone was treated as containing a homoge- neously mixed volume of water, with tidal water flow serving as a link for transfer of different forms of organic carbon (org.C) between the compartments. Position of the mangrove zones in relation to the amplitude and frequency of tidal inundation is J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 73 Table 1 Summary of the field data (mean�standard deviation) that were used to construct the empirical model of organic carbon. SL and LE TFMarine fringe Litterfall (g m−2 year−1) 241.4�141.5632.5�364.1 728.9�432.2 14.3�10.6 7.4�5.23.0�1.7Litter standing stock (g m−2) Suspended particulate matter (mg l−1) 5.18�2.793.44�2.56 9.04�3.71Flood tide Ebb tide 4.28�2.241.81�1.74 4.54�2.94 Dissol�ed organic carbon (mg l−1) 1.355�0.482Flood tide 1.920�0.5240.789�0.365 2.005�0.488 2.573�0.3791.276�0.221Ebb tide Lateral transport, litter (g m−1) 63.9�36.214.0�15.3 16.2�23.9Flood tide 37.1�15.3 68.9�66.1Ebb tide 52.5�51.9 10.1Crab litter removal (Mg year−1) 80.2– Depth of water (m) 1.9�0.57Spring tides 0.9�0.4 0.5�0.1 0.9�0.58Neap tides 0.5�0.5 – 95 975 160 385112 980Surface area (m2) Estimated �olume of water (m3) 80 210214 660 80 190Spring tides 22 630Neap tide –101 680 18252375 2128Length of SL (m) shown in Fig. 2. Compartments of the model are related with actual mangrove regimes in Maruhubi mangrove forest as shown in Fig. 3. A total of eight compartments within the three sub- models were used (Figs. 4–6). Initial value of these compartments was zero. A steady state of the eight compartments was usually achieved within ten cycles, where each cycle represents 1 year. Field data has been combined to make an algorithm using Model Maker ver. 3 software. For example, the change with time in the TF compartment of subsystem 2 (Fig. 4) is described as a differential equation. dTF dt =TFlitter−F1−F4−F7 =TFlitter−TF(k1+k4+k7). The variable TFlitter is the calculated mass of organic carbon of annual litter fall based on observed data from the collecting nets. The rate constants k1 and k4 have been chosen to satisfy observed mass flows to Crabs and SL–LE, respec- tively. k7 is chosen to satisfy observed standing crop of TF. Fig. 3. Map of the study area showing sub-areas of the forest: marine fringe (MF) which is mainly colonised by S. alba ; shoreline (SL) composed of S. alba and A. marina ; landward extending (LE) with an almost pure stand of C. tagal as well as a mixed stand of C. tagal and A. marina ; terrestrial fringe (TF) dominated by A. marina. J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8374 Fig. 4. Litter sub-model showing the compartments and sinks of litter (rectangular boxes): MF; combined SL and LE; and TF; crab consumption; and degradation. Fluxes of litter in the forest sub-areas are shown. Model and observed (in parentheses) values are given in Mg org.C year−1. 2.2. Estimation of mangro�e leaf remo�al by crabs Grapsid crabs, especially Neosarmatium mein- erti, were quantitatively the dominant members of the macrofaunal community of the mangrove floor of the LE and TF zones (Machiwa and Hallberg, 1995). Other important crabs included the gecarcinid, Cardisoma carnifex and few Uca spp. Rate of litter removal by the grapsid crabs was estimated by using a laboratory feeding ex- periment. The set-up involved collection of live crab specimens of varying size classes from the forest. Crabs were then fed on senescent A. marina leaves that were about to be shed. Half leaves (split along the petiole and midrib) were introduced into plastic buckets (10 l) in which the crabs were reared; the other halves were retained for dry weight (dw) determination. 2.3. Determination of organic carbon in the forest Litter production was measured using 36 litter traps made of a square wooden frame (0.25 m2 ) and a tapering net (mesh size 2 mm). Six traps were deployed in each sub zone of the forest, and monthly collections were made. Aboveground standing litter biomass in the forest subsystems was determined by collecting all litter in 1 m2 quadrants. Litter was dried to constant weight at 80 °C. A portion was finely ground (�80 mesh), homogenised and litter organic carbon (LOC) was determined as total organic carbon (TOC). A Shimadzu TOC-5000 analyser (equipped with Solid Sample Module SSM-5000A) was used. Floating mangrove litter was collected at five sampling sites along the ocean– forest boundary, three sites at the MF–SL–LE boundary and four J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 75 sites at the SL–LE–TF boundary. The same sites were used for water sampling to the POC and DOC analysis. Macrodetritus was collected with a net (2 mm mesh size) that was set perpendicular against the direction of the tidal flow. The net was kept above the water surface by closely fastened buoys. Macrodetritus was dried to constant weight and TOC was determined as above. Water samples for SPM and DOC analysis were collected at depths of 0.25, 0.5, 1.0, 1.5 and 2.0 m depending on the amplitude of the tide at the specific site. Samples for POC determination were immediately centrifuged (3500 rpm, for 10 min). SPM was washed with distilled water, freeze dried and the weight was determined gravimetri- cally. POC content of SPM was analysed with a Shimadzu TOC-5000 analyser. Samples for DOC analysis were filtered through a 0.45 �m membrane filter (Schleicher & Schuell®). DOC was determined as non-purgeable organic carbon (NPOC) with a Shimadzu TOC- 5000 analyser coupled to an autosampler (ASI- 5000). The methods are described in details in Machiwa and Hallberg (1995) and Machiwa (1999). 2.4. Computation of the field measurements Maruhubi forest system is relatively open to tidal flushing. Mass balance studies are limited by the accuracy of estimation of rates of discharge. Per tidal cycle POC and DOC fluxes were calcu- lated by multiplying average concentration of each zone with estimated net exchange of water between the zones. Values were integrated over a complete tidal cycle (about 12 h) to obtain net Fig. 5. POC sub-model showing compartments and sinks (MF; combined SL and LE; TF; and sedimentation). Fluxes of POC include input of POC from the ocean and standing stocks of POC in the forest zones (MF–POC; SL–LE–POC; and TF–POC). POC to DOC is a POC sink that is further degraded into DOCin the sediment. Model and observed (in parentheses) values are given in Mg org.C year−1. J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8376 Fig. 6. DOC sub-model showing the fate of DOC in the forest zones. Compartments are the Ocean; MF; combined SL and LE; and TF). Fluxes and sinks of DOC (mainly removal in the MF sub-area by organisms) are shown. Model and observed (in parentheses) values are given in Mg org.C year−1. material transport (Spurrier and Kjerfve, 1988). Volume of water was estimated by multiplying the area of each forest zone with the mean depth of water of that zone. The within forest standing stock of DOC and POC were calculated by multi- plying the concentration per unit volume with total volume of water at mean water depth within a given forest zone. Litter transport was estimated by multiplying the mass of trapped litter (per unit length of net), with the length of the boundary line between given sub zones of the forest. It was assumed that cross-sectional variability in litter transport within given boundaries was small. Amount of leaf foraged by a crab was deter- mined after 24 h, as the difference between the initial wet weight of half leaf and the wet weight of the remaining leaf portions. Weight increase of leaf by water absorption due to exposure of meso- phyll tissue was checked using a control for each half leaf. Rate of litter removal by crabs was estimated by multiplying the average per capita leaf consumption with the numerical abundance of crabs in a given forest zone or in the entire forest. 3. Results and discussion 3.1. Leaf litter remo�al by crabs Crab size correlated well (Pearson r=0.81, P=0.027) with the amount of leaf material in- gested per day (Table 2). With an average number of 21 sesarmid crabs per m2, the TF zone sup- Table 2 Consumption of A. marina leaf by the crabs N. meinerti (size classes between 10 and 60 mm) and C. carnifex (size classes between 60 and 80 mm) Size classes of crabs Rate of consumption (Carapace width mm) (mean�SD mg dw crab−1 day−1) 10–19 (n=5) 15.7�4.9 20–29 (n=5) 22.3�8.2 30–39 (n=8) 64.5 �32.8 40–49 (n=6) 89.8�57.6 72.6�41.250–59 (n=4) 60–69 (n=5) 443.0�203.4 70–79 (n=5) 957.4�231.9 J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 77 ported a sesarmid population of 3.36 million. Mean carapace width for the crabs was estimated at 30 mm, a crab of this size would consume about 23.73�12.05 g dw of litter year−1 equiva- lent to 10.59�5.48 g org.C year−1. Similar feed- ing experiments have been conducted, for instance, Leh and Sasekumar (1985) estimated mangrove leaf consumption by Chiromantes ony- chophorum (carapace width 11–16 mm) in the laboratory to be 14.97�7.67 g dw year−1 per crab. The estimate for C. eumolpe (carapace width 9.7–18.4 mm) was 12.78�3.29 g dw year−1 per crab. Emmerson and McGwynne (1992) reported ingestion rates in the laboratory of between 26.65 and 168.27 g dw year−1 per crab for Sesarma meinerti de Man. Rearing of crabs in the labora- tory may underestimate their rate of feeding. Un- der field conditions, grapsid crab removal of litter includes consumption and storage in their burrows. Analytical results showed that A. Marina dry leaves contain �45% orgC. Relevant figures for TF zone (e.g., field estimates of mean crab density and laboratory estimate of crab ingestion rates) were applied to estimate litter removal by crabs. Calculations reveal that sesarmid and gecarcinid crabs consume 498.23�252.95 g dw leaf litter m−2 year−1. This amount was equivalent to 68.6% of litter production in the TF zone, the rate of consumption in the LE zone was equivalent to 14% of litter fall. 3.2. Parent model The conceptual framework for this model al- lows it to be used in an environmental setting similar to Maruhubi site. The assumptions of the model were: the import of litter from other adja- cent marine ecosystems was only via the ocean– MF boundary; there was no net import of non-mangrove litter from adjacent marine ecosys- tems; all the litter that was taken into the burrows by crabs was eventually consumed; the number of crab holes m−2 closely represent the number of crabs m−2. Sub-models were constructed using observations in the field, i.e., organic carbon was exported to the ocean in the form of DOC (62 Mg org.C year−1) and macrodetritus (18 Mg org.C year−1) only; organic carbon was imported to the forest in the form of SPM (7 Mg org.C year−1). Litter (69 Mg org.C year−1) is fragmented into POC, and is exported to the POC sub-model. POC (98 Mg org.C year−1) is degraded into DOC in the forest sub zones and is exported to the DOC sub-model. The sub-models are linked to the parent model as shown in Fig. 1. Raw data of compartments, variables, standing stocks, flows and rate constants is available from the corre- sponding author. The fate of organic carbon in the mangrove forest involved a series of degradation steps from production to the final stage of export to the ocean. Litter production (mainly leaves) is an index of primary production. Leaves form the basis for the organic carbon flux through the mangrove ecosystem. Leaf litter from the man- groves constitute LOC, which is the major organic carbon input of the system. Primary production in the ocean formed another essential input of car- bon as POC. One of the sensitive parameters that controlled DOC fluxes between the mangrove ecosystem and the ocean was tidal amplitude and flow velocity of tidal current for LOC. Hydrody- namics affected the rate of within system organic carbon transport and accumulation. Therefore, changes in the fluxes significantly affected the standing crops of organic carbon in the zones of the mangrove forest. 3.3. Sub-model litter organic carbon Fig. 4 shows annual input (variables), standing crops of litter (compartments) and fluxes of litter in the mangrove zones. Crabs remove litter from the forest sub zones TF and SL–LE (36 and 3 Mg year−1) and bury it in the sediment, where it is degraded. This makes the TF zone especially rich in organic matter. Removal of litter by grapsid crabs was the predominant sink for litter in the TF zone. Crabs dragged leaves down their bur- rows as well as consuming them on the sediment surface (Machiwa and Hallberg, 1995). These crabs fragment leaf litter and return the nutrients as faecal pellets which can be easily degraded by micro-organisms, enhancing POC and DOC pools of the forest (Robertson, 1987). J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8378 Results suggest that sesarmid crabs have a sig- nificant quantitative impact on litter removal (68% of litterfall) from the TF sub-area of the forest. The TF zone had the highest amount of litter input in the system (331.3 g. orgC m−2 year−1). This fraction of litter is available for export only during highest spring tides. Robertson (1991) has reported similar observation. He found that Sesarma messa consumed or buried at least 154 g m−2 year−1 or 28% of the annual leaf fall of 556 g m−2 year−1. His observations were restricted to the low and mid-intertidal forests of Rhizophora spp. that are tidally flushed twice per day. Robertson and Daniel (1989) also estimated that in the high intertidal mangroves dominated by Bruguiera or Ceriops, crabs removed up to 79% of the total annual litter fall. Mean instanta- neous standing stock of leaf litter on the floor where crabs exist was low. About 1% of annual litterfall is removed by in situ microbial decompo- sition per se in intermittently flooded forests (Robertson and Daniel, 1989). Earlier, Valk and Attiwill (1984) estimated the processing (immedi- ate consumption or storage) of freshly fallen litter by the macro detritivores such as grapsid crabs as 40% of total litterfall in the daily inundated man- grove zones. Variation in the litter removal rates reflect the between sites differences in the tidal regimes and food habits of the grapsids. Imbalance between litterfall, standing crop and litter transport was evidentin the SL–LE zone, where litter standing crop of the model (15 Mg org.C) suggested a higher rate of litterfall than the observed value (Fig. 4). Model results showed that the observed litter standing crop (10 Mg org.C) in the SL–LE zone was underestimated by 30%. Discrepancy between the observed and model data was because of the presence of short scrub plants in the SL–LE zone, which were efficient in trapping lateral flux of litter. These shrubs also were a problem in estimation of canopy cover and hanging of litter baskets. SL– LE litter, was a variable that represented litter which was accumulated at the spring high tide mark of the SL–LE zone. This fraction of litter is fragmented by physical factors and decomposed in situ by benthic microbiota. The SL–LE zone was not exporting any litter to other zones. Re- sults from the model (Fig. 4), concluded that in addition to the crabs, the major sink for litter in the SL–LE zone was fragmentation by physical factors and in situ decay (12 Mg org.C year−1). Litter in the MF zone was mainly exported to the ocean (18 Mg org.C year−1). In situ decay removed 9 Mg org.C year−1, and 5 Mg org.C year−1 was transferred to the SL–LE zone. Flows to the degradation compartment were given val- ues to balance the compartments in accordance with observed values of standing crop in the forest sub-areas. These flows, as well as 39 Mg org.C year−1 of faecal material from crabs en- tered the degradation compartment as POC and were further exported to the POC sub-model. Unlike in Florida where almost all litter is flushed (Twilley et al., 1986), at Maruhubi the sesarmid crabs exert a major impact on leaf litter turnover in the landward part of the forest. Simi- lar observations have been reported from tropical Australia, Malaysia and South Africa (Leh and Sasekumar, 1985; Robertson, 1986; Emmerson and McGwynne, 1992). In support of Emmerson and McGwynne (1992), the model results suggest that the TF zone mangroves are ecologically im- portant. TF zone supports a large macrofaunal biomass that process mangrove litter, therefore, it plays a major role in the flow of energy in Maruhubi forest. Results of macro detritus trans- port revealed that tidal export of litter in the TF zone is not so effective. High litter production in the TF zone coupled to high abundance of crabs ensures retention of nutrients in the forest (un- published phosphate data). This possibly creates a large nutrient sink in the TF zone, which was flushed to the marine environment by tidal, storm and ground water flow mostly as DOC. 3.4. Sub-model particulate organic carbon There was a net import of POC from the ocean to the mangrove area, which thus was acting as a filter for the ocean water of the Zanzibar channel, thereby enhancing the environment for the sur- vival of the neighbouring coral reefs. Inputs of POC in the three zones were calculated as directly proportional to the retention time of tidal water in each separate zone, which has been multiplied J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 79 Table 3 Data derived from the model in Mg org.C year−1 unless otherwise noted MF SL–LE TF 9Degradation/fragmentation of litter 12 9 to POC 3Faecal POC (crabs) 36– 414 12Deposition of POC 24Transformation of POC to DOC 16 59 7Flux of DOC to overlying water 88 –22 –Removal of DOC by organisms 5Rate of sedimentation (mm year−1) 1 3 33 43 14Rate of POC mineralization (% year−1) be an overestimation of the real situation, because resident and tidal migrant fauna consume part of the POC. The model also showed that sedimenta- tion in the MF zone was accompanied by high rate of POC mineralization (Table 3). Observed and model data for POC mineralization in the forest sub-areas compares well, except for the TF zone where the observed rate is twice the model rate. Discrepancy between the observed and model values for the TF zone may be explained with the TF zone containing a more pronounced amount of litter than the other zones, while the material used for field observations was only POC. 3.5. Sub-model dissol�ed organic carbon Standing crops and net flows of DOC from the mangrove forest zones to the ocean are shown in this sub-model (Fig. 6). Like in the POC sub- model, all inputs into the three zones were calcu- lated from the retention time of tidal water in each forest sub zone. In order to balance the model with observed standing crops and fluxes, additional DOC was imported from the SED compartments of the POC model. Decay of POC in the sediments of the mangrove zones increased the magnitude of the flow of DOC to the ocean. DOC was exported from the forest as a result of in situ decay of macro-detritus and particulate matter. The model showed an imbalance between DOC production in the forest and export at the MF-ocean water boundary. DOC discharge into the ocean amounted to 62 Mg. year−1, and the DOC mass balance was achieved by removal of DOC from the water column by organisms in the MF sub-area. This amounted to about 22 Mg year−1 which was removed from the water column when the MF zone was inundated. Possi- ble sinks for DOC in the MF zone included removal by macrofauna, bacterial and microalgae utilisation and absorption of dissolved organic nitrogen by mangrove rootlets. Muller (1998) ad- vocates that, in an open system structures are created and can be detected using ecological gra- dients which build up in the system. Camilleri and Ribi (1986) reported that flocculation of DOC and formation of flakes, feeding of the flakes by with the observed standing crop of POC. Accord- ing to the model (Fig. 5), the SPM (7 Mg org.C year−1) import to the forest from the ocean has a high impact on the rate of sediment accretion in the MF zone. POC was deposited in the MF zone during the high tide flow. Most of the POC from the autochthonous suspended matter generated in the SL–LE (10 Mg org.C year −1) was removed from the water column in the TF zone by sedi- mentation (12 Mg org.C year−1). The accumula- tion of organic carbon in SED compartments was used to balance the model with respect to stand- ing crop data and flow rates between the compart- ments of MF, SL–LE and TF. The accumulation in the SED compartments was further enhanced by the import of degraded litter from the litter sub-model, which was distributed between SED compartments in accordance with the inputs to the degradation compartment of this sub-model (Table 3). Material balance of the model sug- gested that litter removed by crabs was turned into faecal POC, which was deposited in the TF sediments. Crab faecal material and SPM con- tribute POC (39 Mg org.C year−1) to the TF sediment (litter sub-model). Calculated values for the rate of sedimentation in the forest sub-areas (Table 3) were derived from the models rate of accumulation of carbon in the sediment of each specific mangrove zone and the average amount of observed carbon in the uppermost 5 cm of the sediment from these zones (Machiwa, 1998). Model data for rates of sediment accretion may J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8380 tidal migrant (e.g. juvenile fish and crustaceans) and resident macrofauna was a prominent DOC sink in mangrove systems. 3.6. Ad�antages of the empirical model In general, the model provides good estimates of field conditions of organic carbon dynamics (Table 4). For instance, the model may answer the following questions: What would happen to the system if clear felling of trees occurs in the MF or TF zones?; What is the implication of clear cut- ting trees in the TF zone to mangrove fauna?; If a large proportion of trees in the MF zone is to be clear cut, the export of litter to the ocean will decrease tremendously. Dynamics of DOC and the deposition of POC will be impacted to a greater extent. Clear cutting of trees in the TF zone will wipe out a large number of litter eating crabs, litter turnover will take a longer time than in the presence of crabs. What would be the impact of sewagedisposal in the TF zone? This was actually the case be- tween 1970s and 1993 when about 5 Mg of wet sewage (mixed with sand) per day was dumped in the TF zone. The model gives an estimate of annual amount of organic carbon accumulated in the sediments of respective area. Numbers are derived from balancing the model data of stand- ing crops and fluxes with observation data. The model allows simulation of the influence of hu- man perturbation of mangrove organic carbon budget. Sensitivity tests reveal that disposal of sewage wastes in the form of POC in the TF zone results in accumulation of POC in this zone only with negligible effects in the ambient areas (com- pare Fig. 7a and b). Alternatively, if sewage is dumped at the SL–LE zone, the rate of degrada- tion is high, and a small accumulation takes place in this zone but again increased accumulation is noted in the TF zone (Fig. 7c). Sewage disposal in mangrove areas can, therefore, be beneficial, haz- ardous or both, depending on a number of fac- tors. Long term discharge of sewage can bring intense anoxic chemical conditions that are offen- sive to the resident and tidal migrant fauna. High organic carbon loading causes de-oxygenation in the sediments of the TF zone, initiating anoxic sediment reactions. In connection with the role of grapsid crabs in the structure and function of mangrove systems, Smith et al. (1991), found that, in plots where crabs were removed, sulphidic con- ditions developed quite rapidly and the ammonia level was high. In addition, total number of ma- ture propagules that were produced was signifi- cantly low at sites where crabs were removed compared to control plots with crabs. Grapsid crabs, therefore, can have profound benefit in sewage contaminated mangrove areas, such as enhancement of aeration and propagule perfor- mance in sulphidic sediments. The modelling approach has attempted to achieve a holistic assessment of the behaviour of organic carbon in the mangrove ecosystem. The model has demonstrated that the flux of organic Table 4 Comparison of some observed mean values and model results of organic carbon (org.C) from macro-detritus, SPM and DOC of the mangrove forest ModelObserved Litterfall in forest zones (Mg org.C year−1) 32.2Marine fringe 32.2 SL and LE combined 10.3 14.6 52.6Terrestrial fringe 52.6 Flows of org.C within and between the forest and the ocean (Mg org.C year−1) Macro-detritus: 17.817.0MF to ocean 5.25.8MF to SL and LE combined 8.08.3TF to SL and LE combined Suspended particulate matter: 7.2Ocean to MF 7.2 6.2MF to SL and LE combined 6.1 9.8 9.6SL and LE combined to TF Dissol�ed organic carbon: MF to ocean 6062.0 25.4 25.4SL to MF LE to SL 11.811.2 8.6TF to LE 7.6 Crab remo�al of litter (Mg org.C year−1): 2.7Combined SL and LE zones 1.9 36.1Terrestrial fringe 36.2 J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 81 Fig. 7. Sensitivity test on annual accumulation of organic carbon in the sediment with varying inputs of sewage in the form of POC. Graph (a) represents steady state with no addition of sewage. Graph (b) represents an input of 20�15 Mg to the SL–LE zone and (c) represents the same input into the TF zone. J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–8382 Fig. 7. (Continued) material from the forest depends on the extent of mangrove cover and the hydrodynamic properties of the system. The model confirms, using field data that a large proportion of mangrove organic carbon is channelled into in situ utilisation by mangroves and juveniles of fish as well as crus- taceans, molluscs, and insects. Considering the constraints of sampling in the field, this model was an attempt to give site-spe- cific description of the dynamics of organic car- bon within the forest and between the forest and adjacent ocean. The empirical model has gener- ated some data that otherwise would be difficult to retrieve by sampling within the mangrove forest (Muller, 1998). Acknowledgements Anonymous reviewers are thanked for their critical comments. SIDA/SAREC East African Programme in Marine Science funded the project. References Alongi, D.M., 1996. The dynamics of benthic nutrient pools and fluxes in tropical mangrove forests. J. Mar. Res. 54, 123–148. Alongi, D.M., Boto, K.G., Tirendi, F., 1989. Effect of ex- ported mangrove litter on bacterial productivity and dis- solved organic carbon fluxes in adjacent tropical near shore sediments. Mar. Ecol. Prog. Ser. 56, 133–144. Boto, K.G., Bunt, J.S., 1981. Tidal export of particulate organic matter from a northern Australian mangrove sys- tem. Estuar. Coast. Shelf Sci. 13, 247–255. Camilleri, J.C., 1992. Leaf-litter processing by invertebrates in a mangrove forest in Queensland. Mar. Biol. 106, 453–463. Camilleri, J.C., Ribi, G., 1986. Leaching of dissolved organic carbon (DOC) from dead leaves, formation of flakes from DOC, and feeding on flakes by crustaceans in mangroves. Mar. Biol. 91, 337–344. Cundell, A.M., Brown, M.S., Stanford, R., Mitchell, R., 1979. Microbial degradation of Rhizophora mangle leaves im- mersed in the sea. Estuar. Coast. Mar. Sci. 9, 281–286. Duarte, P., Ferreira, J.G., 1997. A model for the simulation of macroalgal population dynamics and productivity. Ecol. Model. 98, 199–214. Emmerson, W.D., McGwynne, L.E., 1992. Feeding and assim- ilation of mangrove leaves by the crab Sesarma meinerti de Man in relation to leaf-litter production in Mgazana, a J.F. Machiwa, R.O. Hallberg / Ecological Modelling 147 (2002) 69–83 83 warm-termperate Southern African mangrove swamp. J. Exp. Mar. Biol. Ecol. 157, 41–53. Golley, F., Odum, H.T., Wilson, R.F., 1962. The structure and metabolism of Puerto Rico mangrove forest in May. Ecology 43, 9–19. Jacobi, C.M., Schaeffer-Novelli, Y., 1990. Oil spills in man- grove: a conceptual model based on long-term field obser- vations. Ecol. Model. 52, 53–59. Lee, S.Y., 1989. The importance of sesarminae crabs Chiro- manthes spp. and inundation frequency on the decomposi- tion of mangrove (Kandelia candel (L.) Druce) leaf litter in a Hong Kong tidal shrimp pond. J. Exp. Mar. Biol. Ecol. 131, 23–43. Lee, S.Y., 1990. Primary productivity and particulate organic matter flow in an estuarine mangrove-wetland in Hong Kong. Mar. Biol. 106, 453–463. Leh, C.M.U., Sasekumar, A., 1985. The food of sesarmid crabs in Malaysian mangrove forests. Malayan Nat. J. 39, 135–145. Lugo, A.E., Snedaker, S.C., 1974. The ecology of mangroves. Ann. Rev. Ecol. System 5, 39–64. Lugo, A.E., Sell, M., Snedaker, S.C., 1976. Mangrove ecosys- tem analysis. In: Patten, B.C. (Ed.), Systems Analysis and Simulation in Ecology. Academic Press, New York, pp. 113–145. Machiwa, J.F., 1998. Distribution and remineralization of organic carbon in sediments of a mangrove stand partly contaminated with sewage waste. Ambio 27, 740–744. Machiwa, J.F., 1999. Lateral fluxes of organic carbon in a mangrove forest partly contaminated with sewage waste. Mangrov. Salt Marsh. 3, 95–104. Machiwa, J.F., Hallberg, R.O., 1995. Flora and crabs in a mangrove forest partly distorted by human activities, Zanzibar. Ambio 24, 492–496. Macnae, W., 1968. A general account of the fauna and flora of mangrove swamps and forests in the Indo-West-Pacific region. Adv. Mar. Biol. 6, 73–270. Muller, F., 1998. Gradients in ecological systems. Ecol. Model. 108, 3–21. Pilette, R., Kincaid, D.T., 1992. First flow-thru analysis in ecosystem studies. Ecol. Model. 64, 1–10. Pomeroy, L.R., Shenton, L.R., Jones, R.D., Reimold, R.J., 1972. Nutrient flux in estuaries. In: Likens, G.E. (Ed.), Nutrients and Eutrophication. American Society of Lim- nology and Oceanography Special Symposium. Allen Press, Lawrence, KS, pp. 274–291. Robertson, A.I., 1986. Leaf burying crabs: their influence on energy flow and export from mixed mangrove forests (Rhi- zophora sp.) in northeast Australia. J. Exp. Mar. Biol. Ecol. 102, 237–248. Robertson, A.I., 1987. The determination of trophic relation- ships in mangrove dominated systems:areas of darkness. In: Field, C.D., Dartnall, A.J. (Eds.), Mangrove Ecosys- tems of Asia and Pacific: Status, Exploitation and Manage- ment. Australian Institute of Marine Science, Townsville, pp. 292–304. Robertson, A.I., 1988. Decomposition of mangrove leaf litter in tropical Australia. J. Exp. Mar. Biol. Ecol. 116, 235– 247. Robertson, A.I., 1991. Plant-animal interactions and the struc- ture and function of mangrove forest ecosystems. Austr. J. Ecol. 16, 433–443. Robertson, A.I., Daniel, P.A., 1989. The influence of crabs on litter processing in mangrove forests in tropical Australia. Oecologia 78, 191–198. Robertson, A.I., Alongi, D.M., Daniel, P.A., Boto, K.G., 1990. How much mangrove detritus enters the Great Bar- rier Reef lagoon? In: Choat, J.H., Barnes, D., Borowitzka, M.A., Coll, J.C., Davies, P.J. (Eds.), Proceedings of the Sixth International Coral Reef Symposium, August, 1988, Townsville, Australia, vol. 2, pp. 601–606. Smith III, T.J., Boto, K.G., Frusher, S.D., Giddins, R.I., 1991. Keystone species and mangrove forest dynamics: the influ- ence of burrowing by crabs on soil nutrient status and forest productivity. Estuarine, Coastal and Shelf Science 33, 419–432. Spurrier, J., Kjerfve, B., 1988. Estimating the net flux of nutrients between a salt marsh and a tidal creek. Estuaries 11, 10–14. Summers, J.K., McKellar, H.N. Jr., 1981. A sensitivity analy- sis of an ecosystem model of estuarine carbon flow. Ecol. Modelling 13, 283–301. Twilley, R.R., 1985. Exchange of organic carbon in basin mangrove forests in southern Florida estuary. Estuar. Coast. Shelf Sci. 20, 543–557. Twilley, R.R., Lugo, A.E., Patterson-Zucca, C., 1986. Litter production and turnover in basin mangrove forests in southwest Florida. Ecology 67, 670–683. Valk, A.G.V., Attiwill, P.M.D., 1984. Decomposition of leaf and root litter of A�icennia marina at Westernport Bay, Victoria, Australia. Aq. Bot. 18, 205–221. Wortmann, J., Hearne, J.W., Adams, J.B., 1997. A mathemat- ical model of an estuarine seagrass. Ecol. Model. 98, 137–149. An empirical model of the fate of organic carbon in a mangrove forest partly affected by anthropogenic activit Introduction Materials and methods Construction of the model Estimation of mangrove leaf removal by crabs Determination of organic carbon in the forest Computation of the field measurements Results and discussion Leaf litter removal by crabs Parent model Sub-model litter organic carbon Sub-model particulate organic carbon Sub-model dissolved organic carbon Advantages of the empirical model Acknowledgements References