INTRODUCTION
Spatial and temporal changes in physical–chemical environmental characteristics strongly influence the structure of fish assemblages within estuarine ecosystems (Pessanha & Araújo, Reference Pessanha and Araújo2003), and other factors, such as predation and competition relationships, seem to act at a small scale (Kennish, Reference Kennish1990; Jung & Houde, Reference Jung and Houde2003). In these environments, the fauna is highly dynamic due to interactions between species-specific physiological limitations and life strategies. The salinity, temperature, dissolved oxygen, turbidity and nutrient concentrations are the main abiotic factors controlling the abundance, distribution and composition of the fish communities in tropical and subtropical estuaries (Blaber, Reference Blaber2000; Rueda & Defeo, Reference Rueda and Defeo2003). Thus, an investigation of the environmental affinities of species and of their distribution in space and time is a basic step toward conservation and sustainable use planning. This is especially relevant for Baía da Babitonga because the estuarine area and surrounding mangrove forests in this region have been assigned a high-priority status for conservation, and management measures, such as the establishment of a marine reserve, are currently being developed (MMA, 2007).
The role of environmental variables on the structure of fish assemblages in Brazilian estuaries is still poorly understood (Garcia et al., Reference Garcia, Vieira and Winemiller2001; Araújo et al., Reference Araújo, Azevedo, Silva, Pessanha, Gomes and Cruz-Filho2002; Bouchereau & Chaves, Reference Bouchereau and Chaves2003; Barletta et al., Reference Barletta, Barletta-Bergan, Saint-Paul and Hubold2005; Chagas et al., Reference Chagas, Joyeux and Fonseca2006; Azevedo et al., Reference Azevedo, Araújo, Cruz-Filho, Pessanha, Silva and Guedes2007). Published studies mainly deal with demersal species that live in deep areas (>3 m), leaving the environmental affinities of species that predominantly inhabit shallow water areas unclear. As elsewhere, logistical and financial constraints have traditionally impeded the determination of the environmental factors and interactions that most influence the distribution and structure patterns of Brazilian intertidal fish communities. Consequently, studies on fish assemblages in shallow water areas have minimized the spatial dimension in their analyses, obscuring patterns, variability scales and the interpretation of causal effects (Jung & Houde, Reference Jung and Houde2003).
Baía da Babitonga is a subtropical estuary located near the southern extreme of the Brazilian zoogeographical province (sensu Briggs, Reference Briggs1995). It offers a particularly good opportunity for analysing the effects of environmental factors on species relative abundance and fish community structure in the intertidal areas of the south-west Atlantic Ocean. Physically, its shore is dominated by low energy shallow water areas (< 1.5 m), with gradual spatial changes in environmental conditions along the bay main axis. In its inner zone, which is more influenced by continental discharge, salinity and transparency are relatively lower than in the outer zone, nearer the sea. The bay ichthyofauna consists of approximately 134 species, mainly marine and estuarine-dependents (IBAMA, 1998; Corrêa et al., Reference Corrêa, Pinheiro, Almeida, Kersten, Lienstadt, Vollrath, Cremer, Morales and Oliveira2006; Gerhardinger et al., Reference Gerhardinger, Marenzi, Hostim-Silva and Medeiros2006). Fishing pressure appears to have risen continuously since the 17th Century following the arrival of the Azoreans. In 1998, there were 33 landing areas, and 1089 registered fishermen obtained their livelihood or supported their income from artisanal fishing or tourism related to recreational fishing (IBAMA, 1998). Recently, signs of overfishing and environmental changes have been reported by most fishermen (Gerhardinger et al., Reference Gerhardinger, Marenzi, Hostim-Silva and Medeiros2006). Considering its capacity for exporting and dissolving nutrients, the bay is highly susceptible to contamination by organic and industrial waste. Among the impacts that have already been observed, nitrogen concentrations are currently much above normal due to anthropogenic enrichment (Mizerkowski, Reference Mizerkowski2007).
This study provides information concerning the intertidal fish assemblage structure in Baía da Babitonga throughout its extension. The central aims are to test whether fish assemblage structure changes between the inner and outer sectors of the bay and among the seasons of the year, and to analyse the role of environmental characteristics in these changes. Additionally, this study aims to address the following questions: (1) what is the spatial and temporal similarity in species relative abundance and in composition?; (2) which of the measured environmental variables has the greatest influence on the assemblage structure?; and (3) how is the abundance of dominant species related to environmental variables?
MATERIALS AND METHODS
Study area
The Baía da Babitonga (26°02′–26°28′S; 48°28′–48°50′W) is located in Santa Catarina State, south Brazil (Figure 1). It is divided into three main water bodies: the bay itself, which provides access to the Atlantic Ocean, and two divergent waterways located in its inner area, the Linguado Channel and Palmital River. The bay is an estuarine area of approximately 130 km2, with an average depth of 6 m. The maximum depth is 28 m in the access channel to the international harbour of São Francisco do Sul, on its southern shore. The length of the bay is 20 km, and its width varies from 1.5 km at the inlet to the sea to 5 km in its inner portion. In addition to anthropized areas, its margins are covered by Atlantic Forest, mangroves (6200 ha) and salt marsh banks (mainly Spartina densiflora Brong.) along sandy beaches, rocky formations and extensive tidal flats. The bay sediment is mainly composed of sand varying from very coarse to very fine, but with the very fine type predominating. According to the Köppen–Geiger classification, the region has a humid-subtropical climate (Cfa) with year-round precipitation and a drier winter (Peel et al., Reference Peel, Finlayson and McMahon2007). The estuary is under a microtidal system with a semidiurnal regime and tide amplitude of 1.30 m. The main river into the bay, Palmital River, receives untreated domestic sewage and industrial waste from the city of Joinville (population 429,000).
Fig. 1. Geographical location of Baía da Babitonga, showing adjacent water bodies (Palmital River (A) and Linguado Channel (C)), the city of Joinville (B), the international harbour of São Francisco do Sul (D) and the position of the thirteen sampling stations.
Data collection
Sampling was conducted during daylight hours on eight occasions (October and November 2007, January, February, April, May, July and August 2008) at thirteen stations distributed on the shoreline along a 21-km estuarine gradient. All thirteen stations were always sampled on the sample day. A beach seine net (15 m × 2 m; 2.5 mm mesh size) was hauled for 30 m parallel to the coastline at a maximum depth of 1.3 m. The unit effort is, thus, defined as one 30-m haul. Though it was strictly controlled to better standardize effort, the net aperture varied slightly between hauls (mean ± SD: 12.6 ±1.5 m). However, no significant relationship between the number of captured individuals and the net aperture was found (linear regression: r2 = 0.04, F = 1.12, P = 0.301). At each sampling station, a single haul of approximately 376 m2 was conducted, representing 0.004% of the bay area being sampled each day. This sampling protocol resulted in collection of 104 samples (1 haul × 13 sampling stations × 8 occasions). All fish caught were kept on ice and later frozen at the laboratory.
Salinity (refractometer), temperature (°C; mercury thermometer), pH (digital portable pH meter-206), transparency (cm; measured with a Secchi disc at a maximum distance of 50 m offshore) and depth (cm; measured with a ruler at the net extremity most distant from shore) were measured at each station on each occasion. Monthly rainfall data were obtained from the meteorological station of the Universidade da Região de Joinville (UNIVILLE), located near the estuary (26o15′19″S–48o51′36″W; altitude 20 m).
Fish classification and enumeration
Captured individuals were identified to the lowest taxonomic level possible following Figueiredo & Menezes (Reference Figueiredo and Menezes1978, Reference Figueiredo and Menezes1980, Reference Figueiredo and Menezes2000) and Menezes & Figueiredo (Reference Menezes and Figueiredo1980, Reference Menezes and Figueiredo1985), or by specialists, separated according to taxa and counted. Fish species were classified into the estuarine use functional guilds defined by Elliott et al. (Reference Elliott, Whitfield, Potter, Blaber, Cyrus, Nordlie and Harrison2007): (i) marine migrants, species that spawn at sea and always enter estuaries in large numbers, particularly as juveniles; (ii) marine stragglers, species that spawn at sea and enter estuaries in low numbers, occurring most frequently in outer areas where the salinity is around 35; (iii) estuarine residents, species capable of completing their entire life cycle within the estuary environment; (iv) estuarine migrants, estuarine species in which the larval stages of their life cycles are completed outside the estuary, and/or are also represented by small marine or freshwater populations; and (v) freshwater migrants, freshwater species found regularly and in moderate numbers in estuaries whose distribution can extend beyond the oligohaline section of these systems. Species classification into functional guilds was based on the information provided by Chaves et al. (Reference Chaves, Bouchereau and Vendel2000), Garcia & Vieira (Reference Garcia and Vieira2001) and Barletta et al. (Reference Barletta, Amaral, Corrêa, Guebert, Dantas, Lorenzi and Saint-Paul2008). Fish nomenclature follows Eschmeyer (Reference Eschmeyer2008) and Figueiredo et al. (Reference Figueiredo, Salles and Rabelo2010).
Late larvae and early juveniles from some abundant taxa (e.g. non-identified Engraulids, Mugil spp. Linnaeus 1758 and Eucinostomus spp. Baird & Girard, 1855) were not identified to species level due to the impossibility of recognizing diagnostic features in small individuals. Data from the collection of these taxa were not included in either the statistical analyses of the community structure or in the calculation of estuarine use ecological guilds (except Mugil spp. for guilds), because these species have different habitat preferences and life strategies (IBAMA, 1998; Corrêa et al., Reference Corrêa, Pinheiro, Almeida, Kersten, Lienstadt, Vollrath, Cremer, Morales and Oliveira2006; Pessanha & Araújo, Reference Pessanha and Araújo2003). Mugil sp. refers to the undescribed species commonly known under the invalid name Mugil gaimardianus (Menezes et al., Reference Menezes, Buckup, Figueiredo and De Moura2003).
Statistical analyses
Seasonality was estimated from temperature and salinity data, and estuarine sectors were defined based on salinity data. In both cases, similarity matrices were used calculated from the Euclidean distance among samples (Q-mode; Legendre & Legendre, Reference Legendre and Legendre1998). These matrices were submitted to cluster analysis to generate graphic representations and identify groups of occasions (e.g. seasons) and sampling stations (e.g. sectors). Differences in physical–chemical characteristics among seasons and sectors that were identified in the cluster analyses were tested by permutational multivariate analysis of variance (PERMANOVA), in which seasons and sectors were the factors. The Euclidean distance between samples computed from the environmental matrix was used in this analysis (Anderson et al., Reference Anderson, Gorley and Clarke2008). Non-parametric analysis of variance (np-ANOVA) was applied using the same distance data to individually test each environmental variable in relation to all factors included in the PERMANOVA. When the null hypothesis was rejected, comparisons of means among groups were made using a permutational Student's t-test (Anderson et al., Reference Anderson, Gorley and Clarke2008).
To verify whether the fish assemblage structure changed according to seasons and sectors, a bifactorial PERMANOVA was conducted on both the quantitative (abundance of each species per sample) and qualitative (presence/absence) data. The similarity matrices were built using the Bray–Curtis coefficient (quantitative) or Sorensen index (qualitative) (Anderson et al., Reference Anderson, Gorley and Clarke2008). In all PERMANOVAs, np-ANOVAs and Student's t-tests, 5000 permutations were performed.
The similarities in species abundance and presence/absence among occasions and stations were analysed through cluster analysis using the unweighted pair group method with arithmetic mean (UPGMA). The indices used to build the similarity matrices were the same as those used in the PERMANOVA. The species and their respective percentages of contribution to the mean similarity in the groups defined by the cluster analyses on abundance data were identified by similarity percentage analysis (SIMPER; Clarke & Warwick, Reference Clarke and Warwick2001).
A canonical correspondence analysis (CCA) was used to assess the relationships between the most abundant fish species in the assemblage (>0.1% of total abundance) and the characteristics of the environment. In this type of analysis, a multiple linear regression is conducted between a matrix of species abundance in each sample (variable answers) and a matrix of environmental variable values (exploratory) (Legendre & Legendre, Reference Legendre and Legendre1998). Only the environmental variables indicated by randomization Monte Carlo test to significantly and independently (P < 0.05 after 1000 runs) explain part of the variation in the biotic data were included in the model. The species Anchoviella lepidentostole (Fowler, 1911) was not included because it occurred in only one sample.
Before all analyses, the environmental variables were centred (mean = 0) and standardized (SD = 1) to put them on the same scale, and the abundance data were transformed in log10(x + 1) to approximate the normal distribution and to minimize the dominant species effect (Legendre & Legendre, Reference Legendre and Legendre1998).
RESULTS
Environmental parameters
The bay was dominated by marine water with a high salinity (mean and range: 27, 9–36), alkaline pH (7.8, 7.1–8.5), low transparency (90, 20–220 cm) and moderate temperature (22, 17–29°C).
The salinity and temperature constrained the sampling occasions to cluster into three seasons: a transition season (October and November 2007); a warm/wet season (January, February and April 2008); and a cold/dry season (May, July and August 2008) (Figure 2A). The environmental characteristics differed among seasons (PERMANOVA: F2,98= 20.8, P < 0.001), and, with the exception of depth, all characteristics differed individually (P < 0.05) among seasons. The cold/dry season presented the lowest temperature (mean ± SD: 19 ± 1oC) and pH (7.7 ± 0.1) and the highest salinity (32 ± 3) and transparency (99 ± 40 cm). Conversely, the highest temperatures (25 ± 2oC) and pH (8 ± 0.4) and the lowest salinity (23 ± 6) and transparency (82 ± 36 cm) occurred during the warm/wet season (Figure 3).
Fig. 2. Dendrograms based on mean monthly salinity and temperature values (A) for the eight sampling occasions and mean salinity values (B) at the thirteen sampling stations, using Euclidean distance. The groups defined were labelled as: (A) I, warm/wet season; II A, cold/dry season and II B, transition season; (B) I, outer sector and II, inner sector.
Fig. 3. Spatial and temporal variation of environmental parameters (salinity, temperature (°C), transparency (cm), pH, depth (cm) and rainfall (mm)) measured between October 2007 and August 2008 at thirteen stations along Baía da Babitonga. The values refer to mean ± SD, except for rainfall (accumulated value for the month of sampling at the meteorological station). Seasons with the same letter are not significantly different from each other (permutational Student's pair-wise test).
Two sectors were defined within the bay based on the salinity at sampling stations: an outer sector (Stations 1 to 6) and an inner sector (Stations 7 to 13) (Figure 2B). The innermost Station, 13, is highly influenced by continental drainage and, consequently, had lower salinity levels that separated it from all of the other sampling stations in the cluster analyses. However, due to its geographical proximity, it was considered to belong to the inner sector. Salinity, transparency and depth tended to decrease from the outer to the inner portion of the bay, while temperature showed the opposite trend. The pH presented no spatial pattern (Figure 3). The habitat characteristics differed between sectors (PERMANOVA: F1,98 = 30.2, P < 0.001). Individual analyses showed differences in salinity (mean ± SD: 31 ± 4 outer sector; 24 ± 6 inner sector; P < 0.001), transparency (102 ± 42 cm outer sector; 79 ± 25 cm inner sector; P = 0.001), temperature (21 ± 3oC outer sector; 23 ± 3oC inner sector; P < 0.001) and depth (85 ± 31 cm outer sector; 48 ± 20 cm inner sector; P < 0.001). The differences between the sectors were independent of seasonality (PERMANOVA: F1,98 = 1.3, P = 0.243), except for transparency (P = 0.050) and salinity (P = 0.021).
Fish assemblage composition
A total of 71,085 individuals from 70 taxa (65 species) distributed into 30 families were captured (Table 1). Late larvae and juveniles of Engraulidae, Eucinostomus spp. and Mugil spp. dominated the assemblage in abundance, comprising 62% of the total number of collected fish. Of the remaining, the 25 most abundant taxa represented over 36% of the total abundance and were further analysed in relation to the environmental variables.
Table 1. Mean catch per unit effort (CPUE) and percentage of occurrence (%) per sector and season for fish species collected in Baía da Babitonga. Ecological guilds are also indicated: MM, marine migrant; MS, marine straggler; E, estuarine; EM, estuarine migrant; FM, freshwater migrant. Codes used in the canonical correspondence analysis for the 25 most-abundant species are presented beside the species name.
Twenty-five taxa can be considered marine stragglers, 25 marine migrants, 12 estuarine residents, 5 estuarine migrants and 1 freshwater migrant (Table 1). Estuarine resident taxa were the most abundant (22.4% of total abundance), followed by marine migrants (19.7%), marine stragglers (4.9%), estuarine migrants (4.7%) and freshwater migrants (0.1%). Among the estuarine fish, the most abundant species were Atherinella brasiliensis (Quoy & Gaimard, 1825) (12.9%) and Anchoa januaria (Steindachner, 1879) (8.9%). Mugil spp. (14%) had the greater contribution among the marine migrants, Harengula clupeola (Curvier, 1829) (3.2%) among the marine stragglers and Anchoa tricolor (Spix & Agassiz, 1829) (2.8%) and Sphoeroides greeleyi Gilbert, 1900 (1.6%) among the estuarine migrants. Odontesthes bonariensis (Valenciennes, 1835) was the only representative of the freshwater fauna, corresponding to only 0.1% of the total abundance.
Spatial and seasonal changes
Significant differences in assemblage structure were found among the bay sectors (PERMANOVA using relative abundance: F1,98 = 10.0, P < 0.001; PERMANOVA using presence/absence: F1,98 = 10.5, P < 0.001). These differences were independent of seasonality in both quantitative (PERMANOVA: F2,98 = 1.3, P = 0.102) and qualitative (PERMANOVA: F2,98 = 1.2, P = 0.243) analyses. The thirteen sampling stations clustered into three groups connected with 53 and 63% of similarity for abundance and presence/absence, respectively. Overall, the stations clustered according to an estuarine gradient in which central stations (4–9, group II) separated the outer-most (1–3, group I) from the inner-most (10–13, group III) stations (Figure 4A, B). The SIMPER analysis computed a 62% mean similarity among the stations of group I, with the greater contributions coming from the marine stragglers H. clupeola and Trachinotus carolinus (Linnaeus, 1766), the marine migrant Oligoplites saliens (Bloch, 1793) and from the estuarine migrant A. tricolor. Stations from group II were linked with 60.9% of the mean similarity, with the greater contributions coming from the estuarine-resident A. brasiliensis and A. januaria, and the estuarine migrant S. greeleyi. In group III, the mean similarity among the stations was 66.8%, with the species most important for the individualization of this group being the same as those of group II, reinforced by the estuarine resident Citharichthys spilopterus Gunther, 1862 and the marine migrant Diapterus rhombeus (Curvier, 1829) (Table 2).
Fig. 4. Dendrograms based on the abundance (A) and presence/absence (B) of fish species collected at thirteen stations in Baía da Babitonga. Each object corresponds to the sampling station (1 to 13) and sector (inner; outer) where the samples were collected.
Table 2. Percentage of contribution of the six most important species identified by the similarity percentage analysis as responsible for the similarity within the groups of sampling stations and sampling occasions defined by the cluster analysis.
Seasonal changes in the fish assemblage structure were detected based on abundance (PERMANOVA: F2,98 = 5.5, P < 0.001) and presence/absence data (PERMANOVA: F2,98 = 6.3, P < 0.001). In the paired tests, all stations differed from each other (P < 0.001) for both quantitative and qualitative data. The eight sampling occasions (months) clustered into three groups of 52% or greater similarity in species relative abundance. The warm/wet season months (January, February and April) and the first cold/dry season month (May) defined group I. The cold/dry months presenting the lowest temperatures and the highest salinity (July and August) were segregated into group III. The transitional months (October and November) aggregated into group II (Figure 5A). These seasonal changes were well supported by the SIMPER analysis. A mean similarity of 54.9% was found among group I months, 66.9% among group II months and 69.8% among group III months. In all three groups, the estuarine-resident A. brasiliensis and estuarine-migrant S. greeleyi, which were common to abundant year-round, were among the three most important species contributing to within-group similarity. The other species are T. carolinus for the warm/wet season (group I), T. carolinus and Micropogonias furnieri (Desmarest, 1823) for the transition season (group II) and O. saliens for the cold/dry season (group III) (Table 2). With respect to the qualitative data (presence/absence), the last transition month (November), all warm/wet months and the first cold/dry month (May) were united in a single group (group I), presenting 62% similarity. The remaining two cold/dry months were separated from the rest, presenting 73% similarity between them (group II), and October remained isolated (Figure 5B).
Fig. 5. Dendrograms based on the abundance (A) and presence/absence (B) of fish species collected on eight occasions in Baía da Babitonga. Each object corresponds to the month and season (transition; warm/wet; cold/dry) of the year in which the samples were collected.
Species–environment relationships
After the Monte Carlo randomization test, the CCA evidenced significant associations between environmental characteristics (depth, salinity, transparency and temperature) and the abundance of 25 taxa (P = 0.001). However, only 20.7% of the variation in species abundance was explained by the four selected axes. Species distribution was unrelated to the pH (P = 0.282), and this factor was, therefore, not included in the analysis. The first CCA axis explained 9.2% of the variation in species abundance and was positively correlated with depth and salinity. Axis 2 explained 7.1% and was strongly correlated to temperature (negatively) and transparency (positively). Axes 3 and 4 presented lower contributions to the variability explained by the analysis (Table 3).
Table 3. Results of the canonical correspondence analysis (CCA) performed between environmental variables and the 25 most-abundant fish species of Baía de Babitonga.
Sectors were clearly distributed along the first axis, with outer sector samples on the positive side and those of the inner sector on the negative side. Seasons were distributed along the second axis, with transition and cold/dry samples on the positive side and the samples from warm/wet season on the negative side (Figure 6).
Fig. 6. Ordination diagram for the canonical correspondence analysis showing the association of the 25 most abundant fish species with selected environmental variables (represented by vectors). Species names are codified according to Table 1. Samples are codified according to sector (1, outer; 2, inner) and seasons (T, transition; W, warm/wet; C, cold/dry). Species inside the circle showed a low association with environmental parameters, while species having a stronger relationship with a particular parameter are located outside the circle.
Species were distributed throughout the plane defined by axes 1 and 2 according to their affinities to the abiotic parameters included in the analysis. Thus, the species that showed a low or variable association with those parameters are located close to axes origins, while species having a stronger relationship with one or the other parameter are located farther from the origins. Pomatomus saltatrix (Linnaeus, 1766) and Menticirrhus littoralis (Holbrook, 1847) were strongly associated with high transparency, high salinity, high depth and low temperature (i.e. outer stations and cold/dry season; upper right quadrant of Figure 6). On the other hand, D. rhombeus and Genidens genidens (Cuvier, 1829) were typical of low transparency, low salinity, low to medium depth and high temperature samples (i.e. inner stations and warm/wet season; lower left quadrant of Figure 6). Oligoplites saurus (Bloch & Schneider, 1801) and Trachinotus falcatus (Linnaeus, 1758) were strongly associated with higher depths (i.e. outer stations; lower right quadrant), while Cathorops spixii (Agassiz, 1829), Sphoeroides testudineus (Linnaeus, 1758), S. greelyi and M. furnieri were weakly associated with lower depths (i.e. inner stations; upper left quadrant of Figure 6).
DISCUSSION
Salinity and depth exhibited a strong gradient along the estuary, generated by freshwater, sediment and organic matter influx into the inner portion of the bay. In addition, the protection of the bay against the predominant swells (from the south-east quadrant), together with the low tide amplitude, favours the formation of extensive tidal flats near the mouth of small perennial rivers, especially in the inner bay area. The temperature varied according to the depth and proximity to the ocean, being lower in the deeper outer stations and higher in the shallower inner ones. The salinity gradient was most pronounced during the warm/wet season. The observed seasonal patterns for physical–chemical features of the water follow the rainfall regime of the region, with higher levels of precipitation in summer (January, February and March) and lower levels in winter (July, August and September) (Mizerkowski, Reference Mizerkowski2007).
The shallow-water fish faunas of Baía da Babitonga and Lagoa dos Patos, south of Brazil (Garcia & Vieira, Reference Garcia and Vieira2001), Chesapeake bay, USA (Jung & Houde, Reference Jung and Houde2003) and Embley estuary, Australia (Barletta & Blaber, Reference Barletta and Blaber2007) share a number of features, such as a large number of marine species and an absence or scarcity of freshwater taxa. In fact, the dominance (in richness) of assemblages by marine species seems to be a general feature of the tropical and temperate estuaries of the western Atlantic (Vieira & Musick, Reference Vieira and Musick1993). Overall, the species adapted to complete their life cycles within tropical and temperate estuaries actually represent a small percentage compared to the marine stragglers and migrant taxa, which use them seasonally (Day et al., Reference Day, Hall, Kemp and Yañez-Arancibia1989). However, although marine species dominated the fish estuarine assemblage composition, estuarine fish were numerically more abundant in both Baía da Babitonga and Lagoa dos Patos (Garcia & Vieira, Reference Garcia and Vieira2001). In the Embley estuary, Australia, where salinity is relatively uniform, fish biomass is dominated by marine species, whereas in areas influenced by freshwater influxes from Amazonian rivers in northern Brazil, the assemblages are dominated in biomass by estuarine species (Barletta & Blaber, Reference Barletta and Blaber2007; Giarrizzo & Krumme, Reference Giarrizzo and Krumme2008). The distribution and abundance of ecological guilds within estuaries are determined primarily by the hydrological features and by the habitat availability found at each location (Barletta & Blaber, Reference Barletta and Blaber2007), and the scarcity and/or low abundance of freshwater species can be explained by the relatively high salinity level in marine-dominated estuaries.
Variation in species abundance and composition occurred across the bay in association with environmental characteristics, suggesting that these observations may be related to species-specific environmental preferences and habitat use strategies. The fish fauna of the outer sector, which is under a stronger marine influence, exhibited a greater abundance of juveniles of marine migrant species (e.g. Oligoplites saliens, Oligoplites saurus and Pomadasys corvinaeformis (Steindachner, 1868)) and marine stragglers (e.g. Trachinotus carolinus, Trachinotus falcatus, Menticirrhus littoralis and Harengula clupeola). The estuarine residents and migrant species (e.g. Anchoa januaria, Atherinella brasiliensis, Sphoeroides greeleyi and Citharichthys spilopterus), and the marine migrant Diapterus rhombeus which tolerate low salinity levels dominated in the inner estuary. Another few abundant species that were not identified as important in the characterization of each sector by the SIMPER analysis also contributed to the differentiation of the fauna. For example, the marine species Pomatomus saltatrix, Opisthonema oglinum (Lesueur, 1818), Paralichthys orbignyanus (Valenciennes, 1839) and Citharichthys arenaceus Evermann & Marsh, 1900 were found exclusively in the outer sector, while gobies were more abundant or only found in the inner bay. Despite small seasonal changes in species spatial distribution, the sectors defined according to salinity corresponded satisfactorily to the ichthyofauna distribution within the bay.
Evidence for the spatial and temporal partitioning of estuaries among abundant fish species that are either transitory or resident in these ecosystems have been provided for both shallow (Pessanha et al., Reference Pessanha, Araújo, De Azevedo and Gomes2003) and deeper areas (Chagas et al., Reference Chagas, Joyeux and Fonseca2006). In Baía da Babitonga, seasonality was more pronounced in relation to species abundance than to ichthyofauna composition (presence/absence). In qualitative terms, seasonal differentiation resulted from the occurrence of a few species restricted to the warm/wet season months (e.g. Oligoplites palometa (Cuvier, 1833) and D. rhombeus) or the cold/dry season months (e.g. P. saltatrix and Cosmocampus elucens (Poey, 1868)). The occurrence of some species extended from the last transition season month (November) throughout the warm/wet season months (e.g. Mugil curema Valenciennes, 1836, Lagocephalus laevigatus (Linnaeus, 1766) and Chloroscombrus chrysurus (Linnaeus, 1766)). Other species occurred from this summer period to the first cold/dry season month (May) (e.g. Achirus lineatus (Linnaeus, 1758), Caranx hippos (Linnaeus, 1766), Sardinella brasiliensis (Steindachner, 1789), Ctenogobius boleosoma (Jordan & Gilbert, 1882) and P. corvinaeformis), contributing to the similarity among seasons found in the assemblage composition.
Asynchronic peaks in the dominant species abundance were responsible for the seasonal changes seen in the assemblage structure. The results of the present study are consistent with information widely reported about the ichthyofauna of shallow areas (Pessanha & Araújo, Reference Pessanha and Araújo2003; Nanami & Endo, Reference Nanami and Endo2007; Araujo et al., Reference Araujo, Rosa, Musiello, Ripoli and Krohling2008), i.e. that seasonal changes in species relative abundance result from variation in reproduction periods and in subsequent recruitment because most of the captured individuals in these environments are at the juvenile stage. The use of shallow areas by young-of-the-year is related to the increase in growth promoted by high food abundance and the decrease in mortality due to lower predation (Whitfield, Reference Whitfield1999; Layman, Reference Layman2000). In Baía da Babitonga, as elsewhere, some species use shallow waters only at the beginning of their life and migrate towards deeper areas after the juvenile phase (e.g. H. clupeola and Mugil spp.; Pessanha & Araújo, Reference Pessanha and Araújo2003), contributing to the seasonal dynamics of the ichthyofauna. However, although the assemblage structure changed seasonally according to the abundance peaks of species (e.g. O. saliens, M. littoralis and Lycengraulis grossidens (Agassiz, 1829) in the cold/dry season; Micropogonias furnieri and Stellifer rastrifer (Jordan, 1889) in the transition season), it was dominated by the estuarine taxa A. brasiliensis and S. greeleyi throughout the year. Adaptive features, such as extended spawning periods (Schultz et al., Reference Schultz, Favaro and Spach2002; Favaro et al., Reference Favaro, Lopes and Spach2003) and tolerance to environmental variation, seem to contribute to the extensive distribution and high abundance of these species.
Although shallow areas may be considered safer (Whitfield, Reference Whitfield1999), the abundance of most species was positively correlated with depth, which was the factor responsible for the higher explicative power for the assemblage structuring. Based on these results, we can hypothesize that the benefits offered by shallowness, which are widely accepted (Whitfield, Reference Whitfield1999; Layman, Reference Layman2000), can be outweighed by higher risks in very low water levels, making the fish avoid these environments. Chagas et al. (Reference Chagas, Joyeux and Fonseca2006) reported that depth effects on the fish assemblage of Baía de Vitória, south-east of Brazil, may be caused by a series of factors that include hydrostatic pressure and the reduction of predation risk through access to vertical dimensions. Additionally, we suggest that fish species distribution in the shallowest areas is influenced by more extreme temperature, greater risk of predation by aerial visually-oriented predators (i.e. birds), greater susceptibility to wave stress and an increased possibility of getting trapped when the tide retreats. Species especially correlated with shallow areas, such as Sphoeroides testudineus and S. greeleyi are, hypothetically, protected by their tetraodontoxins (Matsumura, Reference Matsumura1995). This would make them less susceptible to predation, thus facilitating their occupation of a niche released by other taxa.
Salinity played an important role in structuring the fish assemblage of Baía da Babitonga, similar to previous findings (Pessanha et al., Reference Pessanha, Araújo, De Azevedo and Gomes2003; Barletta et al., Reference Barletta, Amaral, Corrêa, Guebert, Dantas, Lorenzi and Saint-Paul2008). The gradient present was sufficient to affect species relative abundance and locally determine the assemblage composition, even, in the absence of a proper liminic zone. Opportunistic euryhaline marine species have a lower ability to osmoregulate at low salinity, which does not allow them to penetrate deeper into estuaries (Rueda & Defeo, Reference Rueda and Defeo2003). Thus, there is probably a physiological barrier to the occupation of the inner bay by typical marine species such as T. carolinus and P. saltatrix. In addition, many species present preferences for distinct salinity levels during their ontogenetic development (Marshall & Elliott, Reference Marshall and Elliott1998). For example, the late larva and early juveniles of Eucinostomus spp. were more abundant in the outer sector, whereas congeneric individuals at late juvenile stages predominated in the inner sector of the estuary, which is a pattern previously described for Baía de Sepetiba, Rio de Janeiro (Araújo & Santos, Reference Araújo and Santos1999). Contrarily to marine taxa, estuarine species are not adapted to high salinity conditions (Whitfield, Reference Whitfield1999). Atherinella brasiliensis, a species typical of estuarine systems in south and south-east Brazil (Ramos & Vieira, Reference Ramos and Vieira2001; Pessanha et al. Reference Pessanha, Araújo, De Azevedo and Gomes2003), was extremely abundant between salinities of 20 and 29 and was little represented at stations having lower or higher values.
The CCA results showed a secondary contribution of transparency and temperature in species distribution. Protection from visually oriented predators and increases in food availability are the two main factors contributing to the importance of turbidity for small, juvenile or small-sized adult fish (Cyrus & Blaber, Reference Cyrus and Blaber1992; Whitfield, Reference Whitfield1999). However, Johnston et al. (Reference Johnston, Sheaves and Molony2007) could not find any evidence that turbidity influenced species distribution in four tropical estuaries and, thus, stated that support to validate these theories is still inconclusive. Temperature was negatively correlated with transparency and salinity, mainly because of the increase in pluviosity during the warm/wet season. Thus, while salinity and depth (and possibly other factors, such as sediment type) may be the main factors spatially structuring the assemblages, temperature seems to have effects on the temporal scale (Marshall & Elliott, Reference Marshall and Elliott1998; Rueda, Reference Rueda2001). Migration, reproduction and recruitment processes are directly related to seasonal variations in temperature (and photoperiod). Such processes are the dominant factors influencing the temporal distribution of fish in the Humber estuary, England (Marshall & Elliott, Reference Marshall and Elliott1998) and, most probably, in Baía da Babitonga.
As has been suggested for estuaries in general (Whitfield, Reference Whitfield1999), the physical environmental characteristics (in decreasing order of importance: depth, salinity, transparency and temperature) predominantly influence the spatiotemporal distribution of species in Baía da Babitonga. However, the amount of variation unexplained by the abiotic variables included in the CCA model was high (73%). This is a common situation and, for example, only 39.3% of the assemblage variation was explained in Baía de Sepetiba, south-east of Brazil (Pessanha et al., Reference Pessanha, Araújo, De Azevedo and Gomes2003) and 18.4% in the Humber estuary, UK (Marshall & Elliott, Reference Marshall and Elliott1998). Many other biotic and abiotic factors, such as substratum type (Rueda, Reference Rueda2001), pollution (Whitfield & Elliott, Reference Whitfield and Elliott2002), habitat availability (Barletta & Blaber, Reference Barletta and Blaber2007), bay margin usage (Tong, Reference Tong2001), competition, predator–prey interactions and food availability (Kennish, Reference Kennish1990), may concomitantly exert some control over species distributions.
Some of the complex and poorly understood relationships between the fish species that live in estuarine shallow areas in south Brazil and the environment have been clarified, such as the small-scale depth effect and salinity preferences, although the influence of many other factors that may regulate distribution and abundance remain to be investigated. We suggest that further research aimed at analysing the relationships existing between species and environmental characteristics should attempt to better isolate these variables during sampling, to avoid the confounding effects of multicollinearity (see Mac Nally, Reference Mac Nally2000).
ACKNOWLEDGEMENTS
We would like to thank the friends and volunteers who helped us during fieldwork and at the laboratory, especially Lilyane Santos, Daliana Bordin, Andréia Schwingel, Bianca Budel and André Cattani. We are also very thankful to José Figueiredo (Museu de Zoologia, Universidade de São Paulo) for his assistance in engraulid identification and to José Maria Conceição (Universidade da Região de Joinville) for providing the boat used during sampling expeditions. The first author was financially supported by Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) with a post-graduate scholarship.
