Juvenile Penaeid Shrimp Density , Spatial Distribution and Size Composition in four adjacent habitats within a Mangrove-Fringed Bay on Inhaca Island , Mozambique

The effects of habitat characteristics (mangrove creek, sandflat, mudflat and seagrass meadow) water salinity, temperature, and depth on the density, spatial distribution and size distribution of juveniles of five commercially important penaied shrimp species (Metapenaus monoceros, M. stebbingi, Fenneropenaeus indicus, Penaeus japonicus and P. semisulcatus) were investigated during a high shrimp recruitment peak lasting from January to June 2002, in four contiguous habitats within a non-estuarine mangrove bay at Saco da Inhaca, Inhaca Island, Southern Mozambique. A total of 14,976 specimens representing the five species were collected by means of a 1m beam-trawl fitted with a 2 mm mesh net with a cod-end. Every two weeks corresponding with spring tides on three consecutive nights, three trawls of 50 meters each were carried out each night in each habitat at 1.40–2.15 a.m. after the daily spring tide high water peak. Species distributions among the four habitats during the six months sampled showed significant differences in habitat preference (p>0.001). Fenneropenaeus indicus, M. stebbingi and P. japonicus dominated the catches in sand flats with densities of 27 ± 0.94, 18 ± 0.98 and 7 ± 0.76 shrimp /50 m, respectively. Metapenaeus monoceros dominated catches in mudflats at a density of 21 ± 0.78 /50 m followed by M. stebbingi with 13±1.2 /50 m. Penaeus semisulcatus was almost exclusively captured in seagrass meadows at densities of 8± 0.57 /50 m. Only the Metapeneaus species showed a wide range of habitat utilization. The differences in habitat preference emphasises spatial partitioning between shrimps, reducing competition. Multiple linear regressions of juvenile shrimp densities for each species with water depth, salinity and temperature showed some significant effects. Fenneropenaeus indicus density showed a significant negative relationship with salinity and water depth and a positive relationship with temperature and turbidity. Metapenaeus stebbingi numbers showed a significant positive correlation with increasing salinity and temperature, but a negative one with depth. Penaeus japonicus numbers were significantly related to decreasing salinity and depth. Penaeus semisulcatus abundance was significantly related to decreasing temperature, and increasing turbidity and depth. More than 50% of the total shrimps captured had carapace length of ≤ 3 mm, substantiating the nursery function of the different habitats sampled in the study.


INTRODUCTION
All penaeid shrimps that comprise the bulk of the fisheries are believed to have a similar life cycle with the following developmental sequences: planktonic larvae, with several naupliar, protozoeal, mysid and postlarval stages, followed by benthic juvenile and adult stages (Dall et al., 1990).The greatest differences between the larval, juvenile and adult stages are attributed to habitat preferences, either demersal or pelagic and predominantly estuarine, inshore or offshore (Dall et al., 1990).Because the various life stages of penaeids often occupy different habitats, migration is a necessary component of their life cycles.
Studies on penaeid shrimp populations have demonstrated different migration patterns.Larvae and postlarvae migrate from the spawning grounds to the nursery grounds, juveniles migrate out of the nursery areas, and adults migrate to deeper water; spawning migrations are therefore common in some species (Garcia & Le Reste, 1981;Dall et al., 1990).
Habitat requirements for juveniles have been described by different authors for different species and may include associations with macrovegetation, or sediment type.The habitats preferred by penaeid species are diverse and include mangroves, seagrasses and macroalgae, as well as estuaries, mudflats and sand flats (Hughes, 1966;Macnae, 1974;Staples, 1979;Staples et al., 1985;de Freitas, 1986;Robertson & Duke, 1987;Primavera & Lebata, 1995).Habitat choice by juvenile shrimp has been reported to be speciesdependent, varying from a specific nursery habitat to more than one habitat type (Coles et al., 1987;Dall et al., 1990).
A considerable amount of information concerning the temporal and spatial abundance of penaeid juvenile shrimps is available in the literature (see Staples et al., 1985;Dall et al., 1990;Subramaniam, 1990;Haywood et al., 1995;Primavera, 1998;Vance et al., 1998).However, only a few studies have been undertaken on the small-scale, short-term variation within different habitat types to simultaneously assess the relative abundance of juveniles as well as shrimps size and abundance (Vance et al. 1996, Kangas & Jackson, 1998;Rönnbäck et al., 1999;2002).. Knowledge of the early development of penaeid shrimps in the eastern Africa region is relatively scarce and this limits geographical comparisons.Shrimp fisheries are important in the region and this lack of information also impedes the development of effective shrimp fishery management protocols.
In Mozambique, only four studies have attempted to assess juvenile penaeid shrimp distribution patterns and seasonality in the marginal areas of Maputo Bay (de Freitas, 1986) and Inhaca Island (Hughes, 1966;Abdurremane, 1998;Rönnbäck et al. 2002;Macia, unpublished data).Generally these studies were performed in different periods of the day and tidal height.Hughes (1966) and de Freitas (1986) sampled penaeid shrimps during low tide while Abdurremane (1998) sampled during the peak of high tide.Despite the differences in the sampling periods, their results were quite similar in terms of the species of shrimps they found, with a slight differences in the presence of certain species in the catches.Rönnbäck et al. (2002) assessed the habitat preferences for juvenile penaied shrimps in mangroves based on sampling habitats during high tide periods, when shrimps can actively choose between the different habitat types.These authors found distinct habitat preferences between F. indicus and M. monoceros; F. indicus preferred mangrove habitats over adjacent sand flats whereas M. monoceros exhibited preference for sand flats rather than mangrove intertidal forest.However, species like P. japonicus, P. semisulcatus and M. stebbingi were not well represented in their catches, but were reported by Macia (unpublished data) to occur in relatively high numbers in the adjacent intertidal sand flat.
The study reported here expands this work to other habitats such as the mudflats, the creeks, and seagrasses in order to assess how and in what proportions and sizes, postlarvae and juveniles of all the species occurring in the area may utilize these habitats, as well as which environmental factors (such as temperature, salinity, turbidity and depth) influence their abundance.The study was carried out during the on-going ebb tide because shrimp catches were sufficiently high for population studies near the mangrove area at Saco da Inhaca.The study of penaeid shrimps in the mangrove-adjacent areas is important for coastal resources management planning.

Study Area
Saco da Inhaca is a small mangrove-fringed bay located at southern Inhaca Island (26º07S, 32º56E´) (Fig. 1).Detailed descriptions of its general ecology, fauna and flora of the Island can be found in the literature (Hughes, 1966;Macnae & Kalk, 1969;de Boer and Longamane, 1996;Guerreiro et al., 1996;de Boer 2000;Rönnbäck et al., 2002).The climate of Inhaca is characterized by two seasons, hot and rainy (September-March) and warm and dry (April-August) (Macnae & Kalk, 1969).The tides of Inhaca Island are semi-diurnal with a mean spring and neap tidal range of 2.2 m and 0.7 m respectively (de Boer et al., 2000).The whole western coast of Inhaca Island is shallow, sheltered and gently sloping, showing large extensions of sand and mud flats, seagrass beds, coral reefs and mangrove forests in the northern and southern part (Saco da Inhaca).
According to de Boer & Longomane (1996), Saco da Inhaca (a small circular cove) occupies a total area of 66 ha and comprises several closely related habitats adjacent to the mangrove fringe and is described in relation to sediment type and degree of exposure.Saco da Inhaca is connected to Maputo Bay and the Indian Ocean at high tide but at low tide 94% of its intertidal area is exposed (de Boer, 2000) thus preventing circulation with Maputo Bay, the major shrimp trawling area, and the Indian Ocean.During low tide most of the swimming fauna are expected to follow the water withdrawal and juvenile shrimps concentrate in the creeks, channels and water flanks (Hughes, 1966;personal observ.).
The study was conducted in four different habitats adjacent to the fringing mangrove forest (Fig. 1).The sandflat and the creek habitats were closer to the mangrove fringe while the mudflat habitat was located in the central part of the cove and the seagrass further away in the channel in a subtidal area.The creek habitat is lined by Rhizophora mucronata and patches of Avicennia marina and exhibits bottom sediment of coarse sand.The sand flat habitat lies adjacent to the A. marina fringe and is characterized by sediment of coarse sand and mangrove debris.The mudflat habitat occurs in the central part of the bay and is characterized by a relatively high percentage of very fine sediment.The seagrass meadows occur along the subtidal channel at the entrance of the bay and are characterised by the occurrence of Cymodocea serrulata, patches of Thalassodendron ciliatum, and Halodule wrightii.Additional characteristics of the sampled area are described by de Boer (2000) and are supplied together with the results on the environmental parameters measured in the area (Table 1).

Sample Collection
Quantitative samples were collected fortnightly at the four defined habitats (Fig. 1) during three consecutive nights, as the shrimps are typically more active at night (Vance 1992).Collection were by means of a 1-m (1 x 0.25 m mouth aperture) beam trawl equipped with 2-mm mesh size net at both body and cod-end.The inferior part of the net aperture had a tickler chain (Kuipers et al. 1992) to ensure effective sweeping of the bottom.A small (4 m) boat with a 25 HP outboard engine was used to pull the net.On each sampling three trawls were performed along 50-m long prefixed transects parallel to the coastline.Trawl duration was for one minute at a velocity of 3 km per hour.Samples were collected in each of the selected habitats 1:40 to 2:15 hours after the high water peak when the water depth was about 1 m.The time span of 35 minutes did not interfere with the sampling period between the habitats and was due to differences in tidal height.Preliminary samples showed that at about 1 m depth the abundance of juvenile shrimps was high and diverse near the mangrove fringe.Despite attempts to standardize depth among the habitats it was almost impossible to get all sampled transects at a similar depth due to the gently sloping topography of the habitats specifically at the seagrass meadows.The time elapsed from the first to the fourth sampled habitat was about one hour and thirty minutes.This time allowed the habitats to be sampled when water depth was more-or-less similar.The approximately similar depth in all habitats and the 4 kg of extra weight attached to the frame of the beam trawl enabled efficient sweeping of the bottom sediment.Bottom water temperature, salinity, turbidity and water depth were measured before each trawl using a salinityconductivity-temperature meter, a turbidity meter and a hand-held portable depth sounder.Shrimps were preserved in 4% formaldehyde solution, diluted with seawater, and returned to the laboratory for examination.All shrimps were identified using Joubert (1965) and Kensley (1972) and carapace length (CL) was measured to the nearest 0.1 mm using a micrometer adjusted to the occular of a stereomicroscope for small specimens (<10 mm), and vernier scale callipers were used for larger specimens.

Data analysis
Environmental factors such as temperature, salinity, turbidity and water depth were compared statistically within and between habitats using twoway ANOVA.Shrimp catches were expressed as total catches per habitat and as density per 50 m 2 for the whole period of the study.Catch rates (densities) of all captured species: F. indicus, M. stebbingi, M. monoceros P. japonicus and P. semisulcatus were compared statistically within and between habitats.Repeated two-way ANOVA, after logarithmic log (x+1/2) transformation of catch data, was used to compare environmental factors and shrimp catches between habitats and replicates.Cochran's C Hartley F max test was used when necessary to test for homogeneity of variance before performing the ANOVA.Tukey's (HSD) multiple comparison test was used to identify differences among the mean densities and the environmental parameters (temperature, salinity, water depth and turbidity).Shrimp catches were also tested using a multiple-regression-based approach.Accordingly, in order to assess the relative importance of each environmental parameter as a factor that might influence shrimp catch, all species and each of the four environmental parameters analysed were included in a backward stepwise multiple regression.
Mean CLs were calculated for each species per habitat and sampling date.Difference in mean CL together with spatial distribution and density might help identify ontogenetic trends in habitat preferences and is an important tool in assessing growth, mortality and migration periods (Staples, 1979).Shrimps were grouped in 1-mm class intervals and numbers for each size interval expressed as a percentage of the catch in each habitat.Shrimps with 1 to ≤3mm CL were referred to as postlarvae and over 3 mm CL as juveniles.Postlarvae and juvenile shrimps were separated into groups and their numbers expressed as a percentage of the total catch in each habitat for each of the five species captured.No morphological, behavioural or physiological changes were considered to distinguish postlarvae from juveniles (Young & Carpenter, 1977;Forbes & Benfield, 1986;Haywood et al., 1995).

Physical conditions
The mean values of environmental parameters such as temperature, salinity, turbidity and depth are summarized in Table 1.Bottom water temperature of the four habitats sampled between January and June varied between 18 and 31 °C.The water salinity had a narrow range compared to the temperature range: 33.3 to 36.6 ppt.The water turbidity ranged from 0.09 to 0.16 g/l while water depth ranged from 0.7 to 1.4 m.
Significant differences were found in bottom water temperature, salinity and depth between the habitats (Table 2), but none between water temperature, salinity and turbidity within the habitats (Table 2).The Tukey's HSD multiple comparison test showed that no differences in bottom water temperature were detected between sand flats, mudflats and creeks (P 1 = P 2 =P 3 > 0.05), but differences were encountered between all three habitats (sand flat, mudflat and creek) and the seagrass meadows (P 1 <0.001;P 2 =0.004;P 3 <0.001).No differences in water salinity between sand flat, mudflats and creek habitats were found (P 1 = P 2 =P 3 >0.05)but these habitats differed significantly to seagrass in terms of salinity (p 1 =<0.001;p 2 =0.002 and p 3 <0.001).Water depth was also significantly lower in sand flat, mudflat  and creek habitats (0.7-1.1m) compared to seagrass meadows (0.9-1.4 m, p 1 =p 2 =p 3 <0.001).Turbidity was slightly higher in the creek compared to the other habitats.However, no differences were reported in the analysis (Table 2).

Shrimp catches and abundances
Four hundred and thirty-two (432) trawls were performed during the study period at the four habitats around Saco da Inhaca corresponding to a trawled area of 21,600 m 2 .A total of 14,976 specimens belonging to five species of penaeid shrimps were captured: Fenneropenaeus indicus, P. japonicus, P. semisulcatus, and Metapenaeus stebbingi.One unidentified species of Metapenaeus was also found in the seagrass habitat but not included in the statistical analysis owing to the very low numbers collected.
Fenneropenaeus indicus, M. monoceros and M. stebbingi were the most common species captured during the study period, accounting for about 80% of the total catch (Fig. 2).In terms of distribution between the four habitats, F. indicus, M. stebbingi and P. japonicus were significantly more abundant in the sand flat habitat than in the other three habitats while M. monoceros was dominant in the mudflat.Penaeus semisulcatus was almost exclusively captured in the seagrass habitat.
Figure 3 shows shrimp densities throughout the study period.Different mean catch values can be depicted for each of the shrimp species for the study period.The shrimp mean densities per 50 m 2 were 27± 0.75 for F. indicus, 25± 0.45 for M. monoceros, 21± 0.31 for M. stebbingi, 7±0.31 for P. japonicus, and 10± 0.81 for P. semisulcatus for the whole study period.Significant statistical differences were found in shrimp abundance across the habitats (Table 3).
Shrimp abundance differed between habitats.Tukey's HSD multiple comparison test showed that F. indicus abundance differed between sand and creek, between sand and mud and between sand and seagrass (p 1 =p 2 =p 3 <0.001),but no significant differences were found between seagrass and creek or between seagrass and mud (P 1 =P 2 >0.05).
Metapenaeus monoceros also showed significant differences between habitats, with shrimp abundance in the sand flat significantly different from mud (P 1 <0.001) and from creek (P 2 < 0.001) as well as between mud and creek (P 3 < 0.001).ANOVA showed that M. stebbingi also differed significantly in abundance between sandflat and creek (P < 0.001) and between sandflat and seagrass (p < 0.001), between creek and mud (P < 0.05) as well as between seagrass and creek (P < 0.01) but no differences were noted between sand flat and mud (p>0.05).For P. japonicus Tukey's HSD test also showed significant differences between sandflat and mud, between sandflat and creek, between mudflat and creek and between creek and seagrass (P 1 =P 2 =P 3 =P 4 <0.001) but no differences were detected between mud and seagrass (p>0.05).Penaeus semisulcatus was more restricted to seagrass meadows and very few specimens were captured in the other habitats.Its high density in seagrass differed significantly from its lower densities in each of the other three habitats (p<0.001).

Catches and environmental parameters
Multiple regression runs on shrimp catches (logtransformed data) and some of the environmental parameters provided useful data for the interpretation of the potential causes of variation in shrimp abundances between the four studied habitats (Table 4).Fenneropenaeus indicus abundance was strongly related to decreasing water depth and salinity and significantly related to increasing temperature and turbidity.Depth and salinity variation was negatively related to P. japonicus while no relationship was found with temperature and turbidity.Penaeus semisulcatus abundance was negatively related to temperature but positively to water depth and turbidity.The abundance of M. monoceros was significantly related only to turbidity, while M. stebbingi abundance was positively related to both temperature and salinity but negatively to depth (Table 4).

Shrimp sizes
A considerable number of very small shrimps dominated the catches throughout the study period.
Cumulative length frequency histograms per habitat and species are presented in Figure 4 a & b.For P. japonicus and M. monoceros the number of specimens within the 3 mm CL group were higher than for any other size group in all habitats.
For F. indicus the 3 mm CL class was most numerous in creeks and sand flat but in mudflat and seagrass the most abundant size group was 6 mm.For M. stebbingi catches of size class 3 mm occurred in creeks, sand flat and mudflat.In the seagrass the modal size class was 5 mm.Penaeus semisulcatus showed highest abundance in class 3 mm CL in the seagrass and no consistent mode could be identified in other habitats as too few specimens were captured (Fig. 4 a, b).There were distinct differences in the size distribution patterns of the species between the four habitats (Fig. 5).
Mean carapace length of all juvenile shrimps captured varied between species and between the four sampled habitats ranging from 5.95 mm to 8.42 mm CL.Among the species captured in all habitats, F. indicus showed the largest mean CL (8.42±0.73CL) and M. stebbingi the shortest mean CL (5.9±0.55CL).The remaining two species, P. japonicus and M. stebbingi had a mean CL of 6.67± 0.53 and 5.95±0.55,respectively (Fig. 5).
Fenneropenaeus indicus, P. japonicus, and P. semisulcatus showed similar associations between mean CL and habitat.Smaller shrimps were captured in sand flats and creeks while bigger sizes were caught in the mud and seagrass habitats.Metapenaeus monoceros, and M. stebbingi showed a consistent aggregation of smaller sizes in mud, sand and creeks (areas closer to mangroves) while the bigger shrimps were dominant in the seagrass meadow (Fig. 5).
The relative abundance of postlarvae also showed a distinct distribution with habitat.The majority of F. indicus, P. japonicus, M. monoceros and M. stebbingi specimens captured were postlarvae (sizes 1-≤ 3mm) (Table 5).In general, postlarvae were more common in the sand flat, mudflats and creeks for all species except P. semisulcatus, which occurred almost exclusively in the seagrass.Fifty three per cent of total species caught during the study period were postlarvae and the remaining percentage were juveniles of dominant sizes (4-8 mm CL).Fenneropenaeus indicus postlarvae and juveniles occurred in almost similar proportions in the sand flat, but in other habitats juveniles were more common but found in negligible abundances (Table 5).Metapenaeus monoceros, both postlarvae and juveniles, were most abundant in mudflat habitats followed by sand flat where juveniles were higher in abundance.A small fraction was distributed between creek and seagrass habitats.Metapenaeus stebbingi was the only species occurring in similar abundance in mudflat and sandflat.It also occurred in a fairly high percentage in the creek but in negligible numbers in seagrass.Penaeus japonicus postlarvae and juveniles were numerous in the sand flat and creeks but they were captured in low numbers in the mudflat and seagrass.Penaeus semisulcatus postlarvae and juveniles occurred basically in the seagrass habitat, but the juveniles were 4 times more abundant than postlarvae (Table 5).

Shrimp catches and distribution among the habitats
All species captured in the study area are of economic importance and comprise the same species captured in Maputo Bay by commercial trawling fisheries.Metapenaeus stebbingi comprised 32% of the catches while M. monceros was 29%, and F. indicus, 23%.Of the remaining 16%, 9% were P.japonicus and 7% were P. semisulcatus.This pattern of juvenile abundance does not reflect the pattern of abundance found for the adult population in the shrimp fisheries at Maputo Bay where F. indicus and M. monoceros comprise 80% of the total catches (Sousa, 1985;1987;Macia, 1990).This result suggests perhaps that the mortality of M. stebbingi before reaching the fishing areas is higher compared with the other two species, or perhaps this species prefers areas that are not fished commercially.However, it is also important to note that this result only refers to Saco Bay, which is one of the potential nursery areas for commercial shrimps.
There are many other nursery areas around Maputo Bay such as estuaries and mangrove habitats that were not accounted for in this study but contribute to the commercial shrimp fishery.On the other hand the differences in the recruitment peak may be another source of differences in abundance between these species.Fenneropenaeus indicus, during the period of the study, seemed to have peaked before the first January sampling month, suggesting that a great proportion of recruiting post-larvae and juveniles were missed.Penaeid shrimps were not evenly distributed among the sampled habitats in spite of their adjacent locations.This suggests different habitat preferences between the species and that the  nursery habitat requirement for the juvenile penaeid species is not necessarily the same as for the adults.This result also emphasises spatial partitioning among the species captured, which reduces competition for space and food.In general, the shrimps were significantly more abundant in mud flats and sand-flats compared to other habitats.However, some species appear to be more widespread, foraging over much wider areas.This is the case for M. stebbingi, M. monoceros and to a lesser extent for F. indicus and P. japonicus.In contrast, P. semisulcatus showed a very restricted distribution pattern, suggesting habitat specialization from postlarvae to larger juvenile stages.With respect to substrate type, the distribution pattern of the five species sampled is similar to those reported by Hughes (1966), de Freitas (1986) and Abdurremane (1998) despite differences in the periods of tides when samples where collected.Hughes (1966) and de Freitas (1986) found F. indicus in muddy areas within the mangrove swamps and channels, which might indicate a preference for mangrove areas.Abdurremane (1998) could not find F. indicus in her study and suggested that this species may enter the mangrove forest during high tide and is thus unavailable in daylight catches.
The distribution of juvenile shrimps within the mangrove forest and adjacent sand flat at Saco da Inhaca, has simultaneously been studied by Rönnbäck et al. (2002).Fenneropenaeus indicus and M. monoceros were found to make extensive use of the mangroves, but F. indicus was the most abundant species in the mangroves compared to intertidal sand flats while M. monoceros preferred sand flats.Their study showed a higher preference of F. indicus for mangrove habitats in comparison to sand flat habitat.The preference of this species for mangroves was more pronounced as individuals grew bigger than 4 mm.As larger shrimp are more powerful swimmers, the trend shows a more active choice of refuge.This may explain why Abdurremane (1998) did not catch F. indicus when using a sampling net of 5 mm aperture mesh size.Staples (1979) reported that P. merguiensis, a species closely related to F. indicus, became unavailable for capture at high tide in an Australian estuary as a result of migration towards the mangrove areas, where predation pressure is reduced.This finding may explain why F. indicus abundance increases near the fringing mangrove during the ebb tide.
Metapenaeus stebbingi was the only species that was widespread over the habitats examined.Surprisingly this species was not reported by Hughes (1966) in his study at Saco da Inhaca, perhaps due to the period of the year when the samples were taken.Its pattern of distribution confirms the findings of de Freitas (1986) and Abdurremane (1998) for bare sand substratum, the firm muddy sand, mud and among the mangrove pneumatophore zone fringing the bay.Penaeus japonicus was mostly found in the sand flat and mudflat.Forbes & Benfield (1986), reported P. japonicus to occur in soft bottom and turbid water habitat in St. Lucia in South Africa.Hughes (1966) also did not report this species at Saco da Inhaca and de Freitas (1986) suggested that the species identified by Hughes as P. semisulcatus, occurring in the shallow pools on the intertidal flats adjacent to the Dotilla sand flats, was probably P. japonicus since it was uncharacteristic for this species to be abundant in the sand flats.The results of this study are consistent with de Freitas's observation and revealed that the most common species captured in the sand flats as well as in the "Dotilla sand flats" (Hughes, 1966) was P. japonicus and probably not P. semisulcatus, since during the whole sampling period few specimens were captured in any other habitat but seagrass.However, the period of the year and time of sampling could be another reason affecting these comparisons.Penaeus semisulcatus in this study was strongly associated with seagrass meadows as reported from Australia (Staples et al., 1985, Loneragan et al. 1998).Similar results are reported by de Freitas (1986) for Maputo Bay and by Abdurremane (1998) for Saco da Inhaca.This study therefore supports the nursery function previously attributed to the seagrass meadows.
In terms of the density of shrimps in general, the values obtained in this study are comparable to densities obtained from other geographic areas.(Staples et al., 1985, Haywood et al., 1995;Primavera, 1998;Kangas & Jackson, 1998).However, care should be taken when comparing catches at different locations and times of sampling, which may cause differences in shrimp abundances due to tidal height.Comparing the shrimp densities of F. indicus and M. monoceros captured by stake nets (Rönnbäck et al., 2000) with this result, the overall mean value obtained by these authors in the same sand flat for F. indicus and M. monoceros were of the same magnitude as the present study, but the mean values within the fringing mangroves were eight times greater for F. indicus.This result emphasizes the importance of mangrove habitats for F. indicus compared to the other species occurring at Saco da Inhaca.The comparison of these studies also showed that M. monoceros was more common in the sandflats compared to the mangrove area, which together with the results from this study suggests that the mudflats are relatively more important for this species than other habitats.

Size distribution
The results in this study revealed that mangroveadjacent habitats at Saco da Inhaca are highly utilized by postlarvae and juvenile penaeid shrimps during the early stage of their life cycles.Catches were entirely dominated by shrimps of carapace length smaller than 5 mm (Fig. 4a,b) and shrimp found in the majority of habitats were not bigger than CL 8 mm, which stresses the importance of this area as a potential nursery for penaeid species.Metapenaeus monoceros showed a wider range of size distribution within the four studied (Fig. 4a,b) while Penaeus semisulcatus was the only species with fairly high numbers of juveniles compared to postlarval shrimps in the seagrass.
The size-frequency distribution reported by de Freitas (1986) in the coastal areas of Maputo Bay close to Maputo city for all species studied was beyond the sizes found in this study and ranged from 16-45 mm for the five species studied.Abdurremane (1998) reported size frequencies ranging from 5 to 17 mm which was much closer to the results obtained in this study, despite the lack of postlarvae in her study.The differences encountered in the size ranges between the three studies are probably due to mesh size differences or more likely due to differences in nursery function.
With the exception of P. semisulcatus all shrimp species with carapace length ≤ 3 mm occurred in higher numbers in the sand flat, mudflat and creek.It is clear that not all shrimp postlarvae settle in all sampled habitats at the same magnitude.It seems therefore, that once in the nursery area postlarvae select their preferred habitat for settlement.However, some species are more restricted to one or two main habitats where catches were higher.All shrimps of the same genus showed alternatively similar patterns of distribution of mean carapace lengths over the habitats.Metapenaeus monoceros and M. stebbingi were similar in their preference and behaviour, as well as in morphology.
Bigger shrimps of the genus Penaeus were found in the mud and the seagrass beds and smaller shrimps were found in the sand flat and creeks while Metapenaeus species were much bigger in the seagrass and much smaller in the other three habitats (Fig. 5).As the shrimps grow bigger they apparently move to deeper waters and the mud and seagrass are the deepest areas of the Saco.They act as transition habitats for these shrimps.
One of the factors that may contribute substantially to the pattern of distribution of smaller sizes in close proximity to mangroves could also be the water current.Smaller shrimps are expected to be less active swimmers than bigger shrimps, therefore may be a tendency for net transport to the mangrove edges and mud flats (Vance et al., 2002;Rönnbäck et al., 2002;Loneragan et al., 1998).Data collected during 14 months near the mangrove forest (sand flat habitat) at Saco da Inhaca, also found that sizes smaller than 3 mm CL dominated the catches (Macia, unpublished data).Haywood et al. (1995) reported considerably higher densities of postlarvae and juveniles of P. semisulcatus in seagrass meadows at the Gulf of Carpentaria in Australia.Loneragan et al. (1998) have also shown the same pattern of distribution and abundance of postlarvae and juveniles of tiger shrimps (P.semisulcatus); however, considerable numbers of postlarvae were found in seagrass together with larger-sized juveniles in the seagrass beds.In this study, this species showed a high number of postlarvae and juveniles occurring together (98%) in the seagrass habitat with only the remaining 2% percent of specimens captured in the other habitats.

Environmental parameters
There are many environmental parameters governing shrimp distribution in coastal areas.In this study salinity, temperature, turbidity and depth varied significantly between the habitats.The seasonal variation of these factors has been shown to influence shrimp abundance in the area as previously reported by Macia (unpublished) and other authors such as Dall et al. (1990) and Vance et al. (1996) for Australian shrimps.However, in this study sediment type and depth differences are most responsible for differences in shrimp abundances between habitats.
The strong positive correlation encountered between P. semisulcatus and depth could partially explain the low abundance in the other three habitats compared to seagrass areas.On the other hand, the high density of F. indicus in the sand flats reflected the movement of this species from mangrove forests due to depth reduction.Therefore, its abundance does not seem to reflect a preference for the habitat, but perhaps its transition to the waterways in the bay caused by the ebbing tide.It is important therefore, to understand the shrimp dynamics in the area so as to interpret results correctly.
Other studies have emphasized the role of environmental parameters in determining the distribution of juvenile shrimps in coastal areas.Penn (1984) reported that F. indicus prefers soft sediment over hard.Grady (1971) found that the greatest catches of penaeid shrimps in the Gulf of Mexico were made over bottoms with fine substrata, either sand-silt or sand-silt-clay.Branford (1981) showed that P. semisulcatus did not have a clear preference for different sizes of sediment, and Moller & Jones (1975) found P. semisulcatus burrowing in a range of different sediment particle sizes.Penn (1981) explained variation in abundance of P. esculentus-a species closely related to P. semisulcatus-in Australia as probably being due to depth variation.Penaeus japonicus is known to be inhibited from burrowing in fine sand and fine gravel thus their preferred particle range is from 500µm to 2µm (Otazu-Abril & Ceccadi, 1981).This finding may explain why P. japonicus was found in low quantities in the mud habitat.Substrate itself has other properties that may influence penaeids that burrow into it, for example, organic content that can affect feeding Metapenaeus monoceros and M. stebbingi also showed a high abundance in the mud-flat habitat and sand-flat.Previous studies found that M. monoceros (Hughes 1966;de Freitas 1986) and other related species, e.g.M. ensis (Staples et al., 1985, Sasekumar et al., 1992), are widely distributed among coastal habitats and this may be associated with their ability to burrow.Joshi et al. (1979) showed that M. monoceros preferred mud substrates.and this preference may be associated with the predation avoidance or food availability.Macia et al. (2003) also found that predation of M. monoceros by Terapon jarbua in aquaria experiments was reduced on mud substrate compared to sand and shell debris substrate types.It has been reported that the amount of organic content constitutes an important factor determining penaeid shrimp abundance (Dall et al., 1990).In an attempt to assess the carbon sources of juvenile shrimps occurring at Saco da Inhaca, Macia (unpublished) found that F. indicus and M. monoceros isotope signatures were highly correlated to sediment signatures where they were collected.The results from this unpublished study also revealed that M. monoceros has a wide range of carbon signals which supports its wide distribution in the field.

CONCLUSIONS
The differences in habitat preference by penaeid shrimp species seen from this study emphasise their spatial partioning to reduce competition for space and food.Previous studies have reported a similar findings (Staples et al., 1985;de Freitas, 1986;Primavera, 1998;Rönnbäck et al., 2002) and a tidal migration has been described for the shrimp Palaemon peringueyi between seagrass beds and channels (Emmerson, 1987).This study also showed that areas adjacent to mangroves such as the intertidal sand flat, mudflat, creek and seagrass were crucially important for all commercial species captured in the area.Even species associated with mangroves, species such as F. indicus, utilize the adjacent habitats during low tide.This finding is of crucial importance for management and the protection of nursery areas.Mangrove conservation should therefore include the adjacent areas.Therefore, the utilization of adjacent habitats during low tide needs to be addressed in future studies, to determine whether juvenile shrimps leave the Saco area completely during low tide.
Penaeus semisulcatus was influenced significantly by seagrass presence and depth variation.Water depth and sediment type, among others, were the factors most responsible for differences in shrimp abundances between habitats.However, water currents, as suggested by Vance et al. (2002), Loneragan et al. (1998) and Dall et al. (1990) may also have an important influence on shrimp abundance.
Size-frequency distributions among habitats show that the proportions of smaller shrimps are greater closer to the mangrove areas than further away from the mangroves.Bigger shrimps occur in the subtidal areas, which probably act as a transition area during migration periods.

Fig 1 .
Fig 1. Map of Saco da Inhaca showing the sampling locations

Fig. 2 .
Fig. 2. Total numbers of penaeid juvenile shrimp species caught at each of the sampled habitats at Saco da Inhaca throughout the period from January-June 2002.

Fig. 4b .
Fig. 4b.Length-frequency distributions (%) of M. stebbingi, F. indicus and P. semisculcatus caught in the four habitats during the study period at Saco da Inhaca

Figure 5 .
Figure 5. Mean carapace length (± S.E.) of the juvenile shrimp species captured in different habitats throughout the study period.