Study Area
The Rio de Janeiro state presents a unique set of lagoon systems, most of them along the east coast (Fig. 1a). These systems are confined to a narrow coastal plain approximately 120 km long and 10 km wide (Ekau and Knoppers 1999). The lagoons are delimited on the continent by a continuous mountainous relief (Serra do Mar) and protected from the sea by a sandy bar that gives them a similar typology. The essential differences are related to the climate of the region, which varies from humid tropical to semi-arid, and the intensity of anthropogenic impacts on their catchment basins (Knoppers and Kjerve 1999). For this study were selected the data of BF from five Rio de Janeiro lagoons that represent a latitudinal and climate gradient.
• Itaipu Lagoon – located between latitudes 22°57ʹ S to 22°58ʹ S and longitudes 043°01ʹW to 043°03ʹW in Niterói city, Rio de Janeiro state. (Fig. 1). The lagoon covers an area of 1.2 km2 and has a water depth between 0.2 and 2.0 m. The climate is warm and humid with a rainy season in summer (December to March), dry season in winter (June to September) and an average rainfall between 1,000 and 1,500 mm/year (Barbiére and Coe-Neto 1999). Itaipu Lagoon is connected to the Atlantic by the Tibau Channel which is an artificial channel that provides an open tidal inlet (Salvador et al. 2002). The lagoon is affected by a microtide effect that has an average height of 0.71 m and the width can increase up to 10 m during the tide of syzygy (Fig. 1E).
• Maricá-Guarapina Lagoon System (MGLS) – located between latitudes 22°52’ to 23°00’ S and longitudes 43°00’ to 42°45’ W) in Maricá city, Rio de Janeiro. The MGLS has a circular shape and is inserted into a kind of deformed rocky amphitheatre with its boundaries determined by the Atlantic Ocean and the Itacoatiara and Negra hills (Perrin, 1984). It is composed by four lagoons comprising a total area of 43.2 km2, which are currently distributed in: Maricá (20.5 km2), Barra (11.7 km2), Padre (1.6 km2) and Guarapina (9.4 km2) which is connected with the ocean by Ponta Negra Channel (Oliveira et al., 1955). The drainage basin consists of three main sub-basins: Vigário River Basin; Ubatiba River, which mouth is located in Maricá Lagoon and Caranguejo River Basin which mouth is located in Guarapina Lagoon (Comitê Gestor da Região Hidrográfica da Baía de Guanabara e dos Sistemas Lagunares de Maricá e Jacarepaguá, 2006). Barbiere (1985) described the bathymetry of the MGLS: Maricá Lagoon has a maximum depth of 2.0 m and a mean depth of 1.0 m; Barra and Padre lagoons represent the shallower part of this system, since 70% of their areas have depths of ≈ 0.5 m; Guarapina is the deepest lagoon of this system, with depths ranging from 2.0 to 5.0 m. The climate of the MGLS region is tropical humid to semi humid, with tropical air masses from oceanic and continental origins Barroso-Vanacôr et al., 1994). It is characterized by the mean annual temperature of ≈ 23°C and the precipitation between ≈ 1100 and ≈ 1500 mm/year (Fig. 1D).
• Saquarema Lagoon System (SLS) - Is located between latitudes 22°55’ to 22°56’S and longitudes 42°35’ to 42°29’W in Saquarema city, Rio de Janeiro. It has an area of 21.2 km2, which extends approximately 11.8 km along the coast with an average depth of less than 2.0 m (Belart et al., 2018; Dias et al., 2017). This system is composed of four large connected lagoons (Fig. 1): Urussanga (12.6 km2), Jardim (2 km2), Boqueirão (0.6 km2) and Saquarema (6 km2). The climate in SLS is sub–humid, with prolonged periods of drought, and high temperatures (Carmouze & Vasconcelos, 1992). The climate in SLS is also highly influenced by Mato Grosso Hills with sub-humid conditions and prolonged periods of drought (Belart et al. 2017). The Mato Grosso Hills effects primarily the rivers that discharge into the Urussanga Lagoon (Mato Grosso, Tingui, and the Jundiá rivers) (Carmouze 1994). This complex ecosystem is connected to the Atlantic Ocean through an artificial channel called Barra Franca channel located in Saquarema lagoon (Fig. 1C).
• Araruama Lagoon – located between longitudes 42°00′W to 42°25′W and latitudes 22°49′ to 22°57′S delimited by five cities, Araruama, Arraial do Cabo, Cabo Frio, São Pedro d’Aldeia and Iguaba Grande in Rio de Janeiro (Fig. 1A). This lagoon comprises approximately 200 km2 (André et al., 1981) and presents salinity ranges from 35 to 56% throughout the year, making this the largest permanent hypersaline lagoon in the world (Coutinho et al., 1999). The local climate is classified as an equatorial desert, according to Kottek et al. (2006). The area displays annual temperatures around 25°C and a rainfall rate of about 1500 mm/year, reaching the highest values between November and March. The average annual rainfall is less than the average annual evaporation, which promoted high salinity values (Dias and Kjerfve, 2009). The lagoon has only one main fluvial channel, while some intermittent tributaries are observed during the rainy season (Barroso, 1987; Kjerfve et al., 1996). The average depth of the lagoon is 2.5 m, but shallow areas ranging from 0.5 to 1.5 m and depth up to 17 m are also present (Souza et al., 2003). The Itajuru Channel is the lagoon's only communication with the ocean, located in the eastern portion of Araruama Lagoon and crossing the urban perimeter of Cabo Frio and São Pedro d’Aldeia. This channel is very narrow (5.5 km length x 180 m width x 3 m depth) and does not allow for efficient water exchanges with the ocean, leading to a long Araruama Lagoon water residence time of around 84 days (Primo and Bizerril, 2002; Valentini et al., 2002).
• Vermelha Lagoon - located in the Massambaba Environmental Protection Area (22˚55’S to 42˚25’W) in Araruama city, Rio de Janeiro. It has 4.3 km in length, 10.88 km perimeter, 0.75 km maximum width and covers an area of ca. 2.5 km2 with the maximum depth of 2 m. The region presents relatively low rainfall (854 mm/year), high annual evaporation rates (between 1,200 and 1,400 mm), average temperature of 23˚C. The climate is classified as an equatorial savanna (Aw—precipitation less than 60 mm) in accordance with Kottek et al. (2006) with annual temperatures around 25˚C. Vermelha Lagoon is not receiving freshwater directly from rivers and streams. This ecosystem is connected to the Araruama Lagoon through an artificial channel inside the salt pan area (Laut et al. 2017). Vermelha Lagoon is the most hypersaline coastal lagoon in Brazil (recording salinities of about 60 on average; maximum 120) and its name is related to the huge proliferation of microbial mats and the purplish bacteria that give a reddish appearance to bottom sediment (Silva e Silva et al., 2004) (Fig. 1B).
Sediment and Water Sampling Method
The present study used the database of the Laboratório de Micropaleontologia da UNIRIO – LABMICRO from which 116 samples were selected. They were collected through the Rio de Janeiro coastal lagoons and originally studied by Raposo et al. (2018) in Itaipu lagoon, Laut et al. (2022) in Marica-Guarapina Lagoonal System, Belart et al. (2018) in Saquarema lagoon, Laut et al. (2019) in Vermelha lagoon and Laut et al. (2024) in Araruama lagoon (Fig. 1). To investigate the factors that drive the distribution of benthic foraminifera metacommunities, both physicochemical data (salinity (Sal), temperature (T), dissolved oxygen (DO), and pH) and sedimentary data (total organic carbon (TOC), total sulphur (TS), carbohydrates (CHO), lipids (LIP), and proteins (PTN)) were selected for analysis.
Foraminiferal Analysis
The sampling, processing and screening methodology followed a standardized methodology proposed by Schönfeld et al. (2012) and it happened as follows: The first upper centimetre of sediment (50 ml) was recovered in triplicate and stained with Rose Bengal plus alcohol solution to stain the living specimens and washed over sieves with mesh openings of 500 and 63µm. The residual fraction in each sieve was dried at 50°C, and the foraminiferal specimens were concentrated by flotation in trichloroethylene (Martins et al. 2015a). All living foraminiferal tests were picked, identified and counted under stereoscopic microscope at 80x magnification. The number of specimens found in the three replicates was then averaged. The generic taxonomical classification of Loeblich and Tappan (1987), and specific concepts of Boltovskoy et al. (1980) and Raposo et al. (2018) and Belart et al. (2018) were followed. After identification, the names of species were checked using the platform available online, WoRMS (World Register of Marine Species; Hayward et al. 2016).
Metacommunity Analysis
Metacommunity structure
To identify the response of the species to unmeasured environmental characteristics (= latent environmental gradient) affecting community structuring, we applied the elements of metacommunity structure (EMS) approach (Leibold and Mikkelson, 2002). This approach uses a Correspondence Analysis (CA) to ordinate the communities by reciprocal averaging, maximizing the proximity of species with similar distributions, as well as the proximity of sites with similar species compositions. To identify these proximities, we considered the abundances of the species on the EMS analyses, discarding those species represented by singletons (one individual on the entire sample), as these species don't provide much information regarding metacommunity structure (Presley et al., 2011). We applied the EMS framework (Presley et al., 2010) with 1,000 simulated null matrices generated by maintaining the species richness of the sites constant and filling species ranges based on their marginal probabilities (“r1” method of null model randomization). This protocol allowed us to discriminate among the metacommunity structures presented by Leibold & Mikkelson (2002), and we investigated the metacommunity structure existent on the latent environmental gradients represented by only the first axis of the CA.
Three interrelated metacommunity metrics were used to verify the structure: coherence, turnover, and boundary clumping. Their values and statistical significance determine the metacommunity's type within the 14 categories defined by the EMS framework (Presley et al. 2010). Different combinations of metric values indicate different underlying processes shaping the metacommunity structure.
Predictor variables
We created spatial variables to represent spatially structured processes operating in the lagoons (Peres-Neto and Legendre, 2010), in broader scales (e.g., unmeasured environmental factors or the different hydrodynamic regimes of the Lagoon’ compartments). To obtain these variables we built Moran’s eigenvector maps (MEMs) from a Principal Coordinates of Neighbour Matrices (PCNM) based on a network connecting all the sampling stations following the Gabriel graph criterion (Dray et al., 2006). Only those eigenvectors which represented significant spatial structure (Moran’s I > 0.1) were selected to be used in posterior analyses (Diniz-Filho et al., 2012), thus resulting in 25 MEMs.
To summarize the environmental variation existing in the lagoons and reduce the number of predictors we performed a Principal Component Analysis (PCA) (Legendre and Legendre, 1998) in R environment (R Core Team, 2017) using the four hydrological parameters (SAL, T, DO and pH) and five sedimentological descriptors of organic enrichment (TOC, TS CHO, LIP, PTN). Whenever necessary, we log-transformed environmental variables to closer them to a normal distribution and/or to improve linearity in relationships. We applied the Kaiser-Guttman criterion (Legendre and Legendre, 1998) to select the number of PCA axes which would be used in further statistical analyses.
Data analysis
We applied a two-stage approach to identify the effects of spatial structure, hydrological and organic matter determinants over foraminifera community structuring at Rio de Janeiro lagoons. First, we calculated Pearson correlation coefficients between the scores obtained for the two axes of the EMS analysis and each of the predictor variables (the spatial structure represented by the MEMs and the environmental gradients represented by the axes of the PCA) to identify the variables that could explain the latent environmental gradients. Only the positive MEMs were used in this analysis as we were interested in the broad-scale spatial structure affecting the metacommunity.
On the second stage, we used variation partitioning (Borcard et al., 1992) to quantify the relative contribution of spatial, hydrological and organic matter components on the BF metacommunity. We applied a partial Redundancy Analysis (RDA) (Legendre and Legendre, 1998) to decompose the variation in the relative abundance and distribution of foraminifera explained by single and shared fractions of three components: spatial, hydrological, and organic matter. Before executing this analysis, we log-transformed the abundance of foraminifera species and some environmental variables. In order to avoid model overfitting due to the high number of variables, we selected the environmental and spatial variables applying a forward selection procedure (Blanchet et al., 2008). A variable would be selected if it explained variance was significant (p < 0.05) and the adjusted coefficient of determination (R²) accumulated by the selected variables did not exceed the adjusted R² calculated using all explanatory variables.
We executed all analyses on R version 3.4.3 (R Core Team, 2017). The package ‘metacom’ (Dallas, 2015) was used to perform the EMS analysis and to obtain the scores of the first two axes of the CA, which represents the latent environmental gradients. The packages ‘spdep’ (Bivand et al., 2015) and ‘adespatial’ (Dray et al., 2018) were used to obtain MEMs. Principal component and variation partitioning analyses (PCA and RDA) were performed with the package ‘vegan’ (Oksanen et al., 2020; see https://cran.r-project.org/package=vegan) The species represented by singletons (that is, those that only appear once) were discarded from analyses.