Trophic Ecology of Some Common Juvenile Fish Species in Mtwapa Creek, Kenya

The trophic status of common fish species in Mtwapa creek on the Kenyan coast was studied. Both the qualitative and quantitative spectra of the diets of these fish species were investigated. It was found that the eight most abundant fish species, Sardinella gibbosa, Pellona ditchella, Spratelloides delicatilus, Atherinomorous lacunosus, Gerres oyena, Secutor insidiator and Leiognathus equula, consumed principally copepods, while Selar crumenophthalmus fed mainly on fish scales. Polychaetes were an important diet for Gerres oyena and Leiognathus equula. Spratelloides delicatilus was a carnivore feeding only on zooplankton and zoobenthos, and had the lowest diet diversity (H’ = 0.40). The rest were omnivorous. Sardinella gibbosa, P. ditchella, L. equula, Sec. insidiator, G. oyena and Sel. crumenophthalmus fed on phytoplankton, zooplankton, zoobenthos and detritus with a relatively higher diet diversity (H’ = 0.68–0.96). Atherinomorous lacunosus did not take detritus in its diet and it had a lower diet diversity compared to the other omnivores (H’ = 0.47). The feeding niches of the examined species were highly overlapped. The Morisita feeding niche overlap values ranged between 0.84 and 1.00. Most of them were generalized opportunistic feeders except for S. dilicatilus, which exhibited a specialised feeding strategy. Overall, this study showed that the fishes had a flexible diet, which relied greatly on the prevailing biotic and abiotic conditions in their habitats.


INTRODUCTION
In the various studies aimed at understanding the feeding regimes, food preference, migrations, growth and breeding patterns of fish, diet has been found to be an important factor, especially in governing their growth, condition, fecundity and migration patterns (Rao, 1974).
Food habits of fish provide essential information for bionomic studies of single species, comparing related or coexisting species, describing assemblages and the analysis of ecosystem energetics and regulation (Sheldon & Meffe, 1993). The analysis of stomach contents in fish is a common method for investigating the diet of fish, and thus describing food chains and webs shared by different species. Such studies also reveal interactions among species i.e. among potential competitors and between predators and their prey (Rice, 1988). The composition of food items in the fish guts also provides information about the habitats in which the fish feed. The nature of this ingested food depends primarily on the mode of feeding, morphology of the feeding apparatus and the time of feeding. The composition and amounts taken also depend on selectivity and the availability of the prey (Pillay, 1952).
In studies on diets of fish along the Kenyan coast (Kimani et al., 1996;De Troch et al., 1998;Wakwabi, 1999), both micro-and macro-fauna were found to the principal diet items. The four trophic guilds identified were: omnivore, piscivore, zooplanktivore and benthic carnivore (Wakwabi, 1999). The omnivores had the lowest feeding niche while the zooplanktivores had the highest. In this paper, we describe the trophic ecology of some common fish species in Mtwapa Creek using analytical and graphical techniques that summarise broad patterns of feeding biology.

Study area
This study was carried out in Mtwapa Creek (3º 55' S, 39º 45' E), situated 25 km north of Mombasa along the Kenyan coast ( Fig. 1). Fish samples were collected from three sampling stations located within the creek. Shallow waters at the entrance and towards the inner parts characterize the creek. The middle section of the creek, especially near the bridge, is narrower and deeper. The creek receives freshwater from a seasonal river (River Mto Mkuu/ River Luadani) whose annual mean discharge is 0.3 m 3 /s (Magori, 1997). A dense mangrove cover of mainly Rhizophora mucronata on extensive mud banks lines the edges of the creek. There is little obvious industrial development around Mtwapa creek except for the intensive boating activities. The creek is relatively eutrophic (Mwangi et al., 2001) with evidence of seasonal water contamination attributed to river discharge, surface water runoff and raw sewage disposal, especially near Shimo la Tewa prison.

Fish sampling
Fish were caught using a simple, 20-m-long, 2-mwide beach seine-net with a 20-mm stretched mesh size. With the help of an outboard powered boat, the net was laid perpendicular to the shore starting from the shoreline into the waters. It was then dragged in a semi-circular manner shoreward and hauled onto the beach. Three tows were made at each station. The catches from the three tows at each study area were pooled to constitute the fish community at the station. The fish samples were sorted and identified to species level using identification keys from Brock (1978), Fischer & Bianchi (1984) and Smith (1986). Fishing schools of Atherinomorus lacunosus and Sardinella gibbosa were often found close to each other. Standard body length was measured to the nearest millimetre. All fish caught were immediately fixed in 10% formaldehyde and later preserved in 70% ethanol.
The fish caught were randomly subsampled and all were juveniles (Table 1). Fish for stomach content analyses were weighed and later dissected to open the viscera. Fish stomachs were carefully severed from the oesophagus to the last portion of the intestine. Each stomach was opened and its contents carefully removed and weighed to the nearest 0.0001g. Stomach contents were fixed in 4% buffered formaldehyde and preserved in 70% ethanol before further analysis. During analysis the contents were emptied into a Petri dish and examined under a binocular microscope. They were sorted through and identified as much as possible to the lowest taxonomic levels and counted. In order to minimise the underestimation of small and soft prey, utmost care was given to the identification of small fragments. For partially digested food items, only the undigested portion of the organisms, usually the head, was used for identification and counts. All the identified prey items were counted and used for subsequent dietary analysis.
Various indices described below were applied to the weights and counts data to determine the food habits of these fishes. Fish stomach fullness index (FI) as a measure of feeding intensity (Hureau, 1969) was obtained as Weight of ingested food FI = x 100 Weight of fish The percentage frequency of occurrence of each dietary category in the stomach of all fish (%F) and the relative contribution of each dietary category to the numbers (%N) of the total food items in the stomach of each fish were calculated (Berge, 1979) as where n f i = number of fish with prey item i in the stomach contents and N f = total number of fish examined, and % N = (n i / N tot ) x 100 where n i = number of individuals of prey item i, found in the stomach contents of each species, and N tot = total number of prey items encountered in the stomach contents of each species.
The Shannon-Weiner diversity index (H') (Washington, 1984) was calculated to determine fish diet diversity: H' = n log n -Â (Ø log Ø) / N where n = number of species in a sample, Ø = number of individuals in a species, and N = total number of individuals in a sample.
One step towards an understanding of community organization is to measure the overlap in resource use among the different species. The most common resources measured in order to calculate overlap are food and space (or microhabitat). The simplified Morisita index proposed by Horn (1966) as a similarity index was used to measure the feeding niche overlap between the eight fish species that were selected for diet analysis in this study, i.e. C H = 2 ÂP ij P ik / Â P 2 ij + Â P 2 ik where C H = Simplified Morisita index of overlap between species j and species k.; P ij P ik = proportions of resource i of the total resources used by the two species (i= 1, 2 … n), and n = total number of resource states. The feeding strategy of each species was determined by Tokeshi analysis (Tokeshi, 1991). Mean individual feeding diversity (ID) and population feeding diversity (PD) were determined using the following equations. ID = (-∑ P ij ln P ij ) / N PD = ∑ P i ln P i where N = the total number fish examined. P ij = the proportion of prey type i in the jth fish. P i = the proportion of prey type i in the entire fish population.
The ID was plotted against PD to indicate the feeding strategy of the species.

Diet composition
The dietary status of eight most common fish species is summarized in Table 1 and Fig. 2. All prey items which accounted for less than 5 % N in the diet of each of the fish species were grouped together and categorised as "others". Comparison of the dietary distribution of the prey across all the eight species by Friedman's ANOVA showed a significant difference among all species in the proportions of each eaten prey (P<0.05). The frequencies of occurrence of each prey type were all significantly different among the eight species (P<0.05). There was a positive correlation between the frequency of occurrence (%F) and the percentage abundance (%N) contributed by each prey item eaten by all the eight species (r = 0.57, P<0.05).

Spratelloides delicatilus
An average of 20 prey categories were identified in the stomach contents (Table 1). Copepods (51.8% N) and brachyuran larvae (44.2% N) were the dominant dietary items (Fig. 2). Copepods, brachyuran larvae and gastropod larvae were encountered in 100, 100 and 87.2 %F respectively of all examined stomachs for this species. The absence of phytoplankton and detritus in the diet of this species is noted.

Sardinella gibbosa
Average of 26 prey categories were present in the stomach contents of Sardinella gibbosa (Table 1). Numerically, copepods were the most important prey item (41.6%N). Other important prey items were the hyperrids, mysids and cladocerans that accounted for 33.5%N, 10.3%N and 5.3%N respectively (Fig. 2). The prey items with the highest frequency of occurrence (%F) were the phytoplankton, detritus, copepods, mysids, and brachyuran megalopas, which occurred in 100, 100, 100, 73.3, and 63.3 % respectively in the examined stomachs.

Selar crumenophthalmus
An average of 20 prey categories were encountered (Table 1). This species mainly fed on fish scales (49.3%N), nematodes (21.1%N) and rotifers (7.3%N) (Fig. 2). Other prey items were eaten but in very low quantities. Detritus, phytoplankton and fish scales occurred in 100% of the examined stomachs while rotifers occurred in 55.5%.

Feeding strategies
Using the Tokeshi graphical analysis method on feeding strategies, Spr. delicatilus was clearly separated from the other seven species (Fig. 3). It was noted that for Spr. delicatilus, both individual and population feeding diversities were low, indicating a specialised feeding strategy. The other seven species were characterised by high individual (ID) and population feeding diversities (PD), suggesting a more generalist and opportunistic feeding strategy. Species with low ID and PD are specialised feeders (Tokeshi, 1991), while those with high ID and low PD are generalists. Species with high ID and high PD are generalists with a homogeneous feeding pattern.

DISCUSSION
In this study, two trophic categories were identified among the eight species of fish that were examined.  (1998) and Wakwabi (1999), it was noted that low stomach fullness index and very low diet diversity are characteristic of planktivorous species, while benthivores tend to have intermediate fullness indices with very diverse diets. James (1988) reviewed the diet of some commercially important clupeids and found that most of them were omnivorous microphagists that derived the bulk of their energy from zooplankton.
Their feeding modes have also been described from laboratory and field studies (Blaxter & Holliday, 1958). In addition to filter and particulate feeding, some species are capable of capturing floating prey (James, 1987). Many filter-feeders display tendencies towards detritivory, utilizing the high detrital loads and large biomass of unicellular benthic algae present in the shallow coastal and estuarine waters they inhabit (Darnell, 1958). The diets of Sar. gibbosa and P. dichella examined in this study support this view. Flexibility in the feeding behaviour of these species enables them to efficiently utilise the available food.
The diet diversity of omnivorous fishes was high compared to that of the carnivorous fish. This was because the omnivores fed on a wider range of food items than the carnivore (Spr. delicatilus). Note that the fish species classified as carnivorous mostly fed on zooplankton and zoobenthos. Atherinomorous lacunosus, which did not feed on detritus, also had relatively low diet diversity. The high diversity of prey items found in the diets of the omnivorous group suggests that they are more generalist predators.
This study also shows that the investigated fish species foraged on a wide variety of prey organisms with a high level of interspecific dietary overlap. The examined fish species were caught from the same habitat. Therefore, selection of common habitats resulted in high dietary overlap and possible interspecific competition for available food resource. Similarly, high dietary overlap may also indicate low competition, if resources are not limited or carrying capacities are not reached (Tokeshi, 1999), since the ratio of demand to supply is of vital concern in the relationship between ecological overlap and competition (Pianka, 1994). Hence, the low diet niche overlap values present in G. oyena/Spr. delicatilus, G. oyena/A. lacunosus, L. equula/Spr. delicatilus and L. equula/A. lacunosus occurred because G. oyena and L. equula utilised more diverse food resources including endobenthos, which A. lacunosus and Spr. delicatilus did not ingest.
Seven of the examined fish species exhibited a generalist feeding strategy. Spratelloides delicatilus, which had a diet specialised on copepods and brachyuran larvae was the only specialist. The species had the lowest mean individual and population diet diversity as shown by Tokeshi's method. Atherinomorous lacunosus, although exhibiting a generalist feeding strategy, had a relatively low individual and population diet diversity compared to the other generalists, because unlike the other species, it did not take detritus in its diet. The high feeding niche overlap between Spr. delicatilus and A. lacunosus observed in this study may also be attributed to utilisation of the same food resource.

CONCLUSIONS
Although Sar. gibbosa and A. lacunosus shared common near-shore microhabitats and most of the time foraged together, they utilised its spatial and nutritional resources of the environment differentially by differentiating their feeding niche breadth and selecting different preys. Sardinella gibbosa had a higher diet breadth than A. lacunosus, since it utilised detritus as a dietary component.
Site-specific features of particular habitats and their associated predator and prey assemblages determine feeding patterns in fish. The choice of prey by the examined fish species seemed to depend upon food availability, and hence they can also be referred to as opportunistic feeders. Fish species with a generalised and flexible feeding behaviour are less vulnerable to changes in abundance or availability of particular prey categories, giving them a competitive advantage over specialised feeders.