Environmental Pollution and Biological Invasions Threaten Native Freshwater Infaunal Bivalves in the Guandu River Basin, Southeast Brazil

Abstract

Freshwater bivalves play essential ecological roles in ecosystems, but they are among the most threatened fauna worldwide. Despite receiving industrial and domestic wastes, the Guandu River is the main source of drinking water for more than nine million people in the Rio de Janeiro metropolitan region. This study aimed to assess how infaunal bivalves respond to water and sediment quality in the Guandu River basin. Samples were collected at 10 sites across reservoirs, lotic, and lentic systems during cold–dry and warm–rainy seasons. Four bivalves were identified: Anodontites trapesialis, Diplodon ellipticus, Corbicula fluminea (non-native), and C. largillierti (non-native). Native species were restricted to two lentic sites at Guandu Lagoon, with the poorest environmental quality, significantly affected by high chlorophyll a and ammonia in the water. In contrast, C. fluminea was widely distributed and more abundant in the basin but restricted to less degraded sites, suggesting a lower tolerance to environmental pollution. Multivariate analyses indicated significant differences in environmental conditions and species–environment correlation. The non-native species spread and poor environmental quality threaten native bivalves in the Guandu River basin, leading them to a local extinction risk. Results highlight the need for effective management and conservation actions to protect biodiversity in tropical river basins.

1. Introduction

Freshwater ecosystems are experiencing a progressive decline in biodiversity [1]. Although freshwater covers less than 1% of the planet’s surface, more than 50% of the human population lives within 3 km of a freshwater body. This human proximity can threaten freshwater species and habitats, including many biodiversity hotspots, through changes in flow, pollution, overexploitation of species, and invasive species. Human activities pose a major threat to the ecological integrity of river ecosystems, as they impact habitats, the quality of water available to the population, and biota [2]. According to the Living Planet Report [3], a significant average decline in the relative abundance of wildlife populations was found, with the greatest losses in the tropical regions of Latin America and the Caribbean, reaching a drop of 94%, with freshwater organisms being the group with the greatest reduction, around 83%. Furthermore, freshwater is a critical resource for living animals and human society.

According to Sustainable Development Goal (SDS) No. 6 proposed by the United Nations [4], it is necessary to provide safe drinking water for more than 800 million people who lack basic services and improve the accessibility and security of services for more than 2 billion people, mainly in Africa and Latin America. In Brazil, the main urban and industrialized areas are located in the states of São Paulo and Rio de Janeiro, which explains why a considerable percentage of the Brazilian population lives in these two metropolises and needs resources from their rivers, such as for water supply, animal and plant production, and hydropower electricity generation (more than 60% of the country’s electricity output in 2023) [5]. The Guandu River is the main river basin in Rio de Janeiro state, which runs through its entire territory and supplies eight municipalities, being the only source of drinking water for more than nine million people in the metropolitan region [6]. Although it is of essential importance, the Guandu River receives untreated industrial and domestic waste daily [7].

Freshwater mollusks are among the most threatened fauna groups in the world [8,9]. Thus, concern about the conservation of natural resources has been increasingly frequent, mainly due to habitat alteration, the introduction of non-native species, and excessive use of natural resources, causing a reduction in diversity [10]. The Asian clam Corbicula fluminea (Muller, 1774) is among the most widespread non-native species in Brazil and the most widely distributed freshwater species in South America [11], which can cause several direct and indirect impacts on macroinvertebrate assemblages through the reduction in available habitat [12]. In addition, filtration by invasive bivalves can negatively affect the recruitment of some species [13]. Thus, these effects can alter the abundance and diversity of native mollusks in freshwater ecosystems in America and Europe [14,15,16,17]. A study on the survey of mollusks in the Guandu River Basin evidenced this pattern, where C. fluminea was the species found in all sampling stations [18]. The main problems faced by the invasion of these organisms are changes in environmental conditions and competition with native bivalves, which reduce their populations. Furthermore, due to the excessive input of waste that the Guandu River receives daily, the presence of several types of contaminants, such as phosphate [19], metals [20], and persistent organic pollutants [7], has been observed along its course. This may interfere with local diversity and abundance of species, as evidenced in other rivers [21,22]. Molluscan distribution has been associated with physical and chemical environmental variables, such as pH [23], conductivity [24], and chlorophyll a [25]. Moreover, other studies have demonstrated the molluscan assemblage responses to organic pollution [26] and contaminants [27], which reduce assemblage richness and consequently interfere with the availability of nutrients.

Bivalves are important in providing ecosystem services, playing an essential role in the food web [10], and are among the most vulnerable groups of invertebrates in freshwater ecosystems [18]. One of the possible reasons may be the synergistic effect between habitat degradation and the loss of larval vectors to colonize new habitats, limiting the ability of species to persist [19] and invasive species [28]. In Rio de Janeiro state, where five native and two invasive species have been recorded, the protected areas are ineffective in protecting native freshwater bivalves, especially members of the order Unionida [29]. Recently, the golden mussel Limnoperna fortunei (Dunker, 1857) [30] was recorded in the Santa Branca Reservoir, part of the Paraíba do Sul River Basin in São Paulo state. This new invasive species in the Paraíba do Sul River basin highlights an enormous risk of new introductions, with potential risks of severe ecological and economic impacts, since part of its waters are diverted to the Guandu River basin through a network of reservoirs, pipes, water tunnels, and pumping stations [30]. For this reason, it is of fundamental importance to assess the distribution of freshwater bivalves throughout a gradient of environmental conditions so that there is increasingly more predictability regarding species conservation. Understanding the distribution of native bivalves in freshwater ecosystems, especially in systems subjected to different environmental conditions, can provide important insights to improve conservation status. Therefore, this study aimed to assess the composition, abundance, and spatial distribution of infaunal bivalves in the Guandu River basin (Rio de Janeiro state, southeast Brazil) and evaluate their responses to environmental variables in the water and sediment. For that, the samplings included sites within three freshwater systems in this river basin (i.e., lentic, lotic, and reservoirs).

2. Materials and Methods

2.1. Study Area

The Guandu River basin has an extension of 108.5 km and an area of 1385 km², extending from the confluence of the Ribeirão das Lajes River with the Santana River to the mouth of Sepetiba Bay (22.55–23.19° S and 43.29–44.28° W). The river flow is mainly made up of the discharge from Ribeirão das Lajes regulated by the Pereira Passos Hydroelectric Plant in Piraí city (Rio de Janeiro state). In its upper course, the Guandu River borders low-income urban areas, such as Japeri, Engenheiro Pedreira, and Parque Cesária. In Japeri city, there is a garbage dump that accumulates waste from the population living in this area, and after precipitation events, residuals are drained into the Guandu River. The contribution to the flow of the Guandu River after the diversions of the Paraíba do Sul and Ribeirão das Lajes rivers allowed the construction of the Guandu Water Treatment Station (ETA-CEDAE) for human supply, in operation since 1955, which currently treats around 47 m³s⁻¹ of water for the consumption of the population from the Rio de Janeiro metropolitan region. Upstream of the ETA Guandu intake, there is the Guandu Lagoon, where the Poços and Ipiranga Rivers flow. Both are heavily polluted by domestic sewage, industrial effluents, and solid wastes from surrounding low-income communities. In the mountainous portion of the Guandu River Basin, there are extensive areas of depleted soils, dominated by erosion processes, which, when sediments are released back into the watercourse associated with sand mining activities, produce fine sediments, although prohibited, to be released [31]. Considering that several municipalities intensely use the Guandu River basin in the Baixada Fluminense (i.e., part of the Metropolitan Region of Rio de Janeiro, composed of 13 municipalities, where more than 3.5 million inhabitants live) and by an industrial complex in the municipality of Itaguaí, it has been chronically and gradually affected by pollutants from wastes of diverse nature. The industrial complex located at Baixada Fluminense has more than 100 active industries distributed across eight municipalities, with a higher concentration of industries in the municipalities of Rio de Janeiro, Itaguaí, and Queimados. Half of the large companies are from the metallurgical sector, and there are also important industries in the brewery, printing, and rubber sectors [32]. The metallurgical sector is made up of medium- and large-sized companies, generally old, with low efficiency in the use of raw materials, with a high potential for waste production and precarious conditions in terms of treatment, thus presenting low environmental performance [31].

2.2. Samplings

The Environmental Institute of the State of Rio de Janeiro (INEA) established a water quality gradient following the environmental resolution [33], which specifies the classification of water bodies and environmental guidelines for their classification. Ten sampling sites were surveyed in the Guandu River basin following the river course, encompassing four municipalities of Rio de Janeiro state (Nova Iguaçu, Japeri, Paracambi, Piraí) and an environmental gradient that included sampling sites submitted to different degrees of impacts and anthropogenic activities [34] (Figure 1). In addition, the samplings comprehended sites with lentic (n = 3; ST1, ST2, and ST4) and lotic waters (n = 4; ST3, ST5, ST6, and ST7) and reservoirs (n = 3; ST8, ST9, and ST10) during two seasonal periods: cold–dry (August 2022) and wet–warm (April 2023).

Bivalve sampling was carried out by active search using a handled metallic scoop in all suitable habitats, including marginal and floating vegetation. The samplings were performed by two researchers for 30 min, totaling 60 min per sampling location, and bivalves were surveyed by hand in sandy and muddy sediments to avoid damage to their shells. The sampling protocol was based on a previous study conducted at the Guandu River Basin [18]. After sampling, bivalves were conditioned in plastic zip bags, identified, kept at ice temperature, and, subsequently, frozen at −20 °C for laboratory procedures. The abundance values represent the sum of specimens collected at each sampling station.

Simultaneously, water variables were assessed using a multiparameter probe (model HI98494, Hanna Instruments, Sat Nusfalau, Romania), and chlorophyll a was measured using a calibrated FluoroSense handheld fluorometer (Turner Designs, San Jose, CA, USA). Thus, seven water variables were assessed during samplings: temperature (°C), conductivity (µS/cm), pH, dissolved oxygen (mg/L and %), total dissolved solids, and chlorophyll a (µg/L). Moreover, water (n = 2) and sediment samples (n = 2) were taken for environmental variable measurements in the laboratory. Surface water (~20 cm) samples were collected using a glass bottle (250 mL), previously cleaned, and kept at ice temperature for nutrient and turbidity analyses within a few hours (1–2 h). Surface sediment (~5 cm) was collected in two replicates manually using a stainless-steel core (80 cm²) at each sampling station. All the samples were immediately packed in plastic zip bags and stored at ice temperature until transport to the laboratory, where samples were frozen at −20 °C for further analysis.

2.3. Laboratory Procedures

Water turbidity was analyzed in triplicate in the samples (n = 2) using a turbidimeter (model TD-300, ±0.5 NTU, Instrutherm, São Paulo, Brazil). Phosphate (PO₄³⁻), ammonia (NH₃), and nitrite ions (NO₂⁻) were analyzed in the water samples (n = 2) from the sampling sites. The method for phosphate analysis is an adaptation of the acid ascorbic method [35] using the Hanna Checker colorimeter for low range (0–2.5 mg/L; HI713) with a resolution of 0.01 mg/L and an accuracy of ±0.04 mg/L. The method for ammonia analysis is an adaptation of the Nessler method [36], using the Hanna Instruments Checker colorimeters (Sat Nusfalau, Romania) for low range (0–3 mg/L; HI700) or high range (0–100 mg/L; HI733). The ammonia colorimetric method for low and high ranges has, respectively, a resolution of 0.01 and 0.1 mg/L and an accuracy of ±0.05 or ±1.0 mg/L. The method for nitrite analysis is an adaptation of the EPA Diazotization method 354.1 using the Hanna Checker colorimeter for low range (0–600 μg/L; HI707) with a resolution of 1 μg/L and accuracy of ±20 μg/L.

The sediments were analyzed on the laser diffraction equipment Malvern Hydro 2000 MU (Lancashire, Great Britain), in which the limits of equipment detection and quantification are 0.01 μm and 0.03 μm, respectively. The equipment employs laser diffraction to analyze particle size distributions. A laser beam is directed through a suspension of particles, and a series of detectors measure the light scattered by these particles. Grain size statistics and parameters were calculated by sampling area using the method of moments [37]. Six grain-size parameters were obtained: mean (Φ), median (Φ), sorting coefficient (Φ), skewness, kurtosis, and kurtosis for a normal distribution. Grain classification was determined based on the literature [38,39]. Sediment samples were homogenized, separated in triplicate, and dried for 48 h at 80 °C. Their moisture content (%) was determined by weight loss after sediment drying, and their organic matter content (mg/g sediment dry weight and %) was determined from the weight loss after ignition at 550 °C for 4 h [40,41]. In addition, carbonates, total organic carbon (TOC), nitrogen (TN), and phosphorus (TP) contents were analyzed in the sediment samples. The carbonate content of sediments was assessed by acid dissolution, with a limit of detection (LD) of 0.1% [42]. TOC and TN contents in the sediment were quantified with an infrared combustion using a CHN Flash 2000 carbon analyzer (Thermo Scientific, Milan, Italy) [43]. The LD for TOC and TN in the sediment was, respectively, 0.05 mg/g (0.005%) and 0.07 mg/g (0.007%), and the limit of quantification (LQ) was 0.15 mg TOC/g (0.015%) and 0.21 mg TN/g (0.021%). TP content was assessed by sediment calcination, followed by acid treatment and photometry analysis in the spectrophotometer Lange DR 5000 (Hach, London, ON, Canada) [44]. The LD for TP was 31 × 10⁻⁶ mg/g (31 × 10⁻⁷%), and the LQ was 93 × 10⁻⁶ mg/g (93 × 10⁻⁷%).

2.4. Statistical Analyses

Principal component analyses (PCA) were applied to the environmental data (water (n = 9) and sediment (n = 10)) after Hellinger transformation, in order to evaluate which surface water (temperature, conductivity, pH, turbidity, chlorophyll a, dissolved oxygen, ammonia, nitrite ions, and phosphate) and sediment variables (grain size parameters—mean, median, skewness, kurtosis—and the content of organic matter, carbonates, total organic carbon, total nitrogen, total phosphorus) mostly contributed to explaining the environmental quality and seasonal patterns. The Broken-Stick criterion was used to retain the PCA significant axes for sampling ordination. A randomization test was requested with 9999 runs, and the seed for the random number generator was set to 4507 for all eigenvalues and eigenvectors.

A constrained redundancy analysis (RDA) was applied to assess the relationship between bivalve species abundance and environmental variables of water (n = 9) and sediment (n = 10). A forward model selection routine was used for RDA to select the independent variables with significant contributions (9999 permutations, p ≤ 0.05) for the best model explanation. The environmental data were Hellinger-transformed to standardize the variable’s units and improve the performance of the RDA linear method [45]. Axes significance was tested by permutation (9999). PCA and RDA analyses were performed in the software PC-ORD 6 and CANOCO 5, respectively.

3. Results

The first two PCA axes were selected by the Broken-Stick (BS) method, with axes 1 and 2 significantly (p = 0.0001) explaining, respectively, 51.73% (BS eigenvalue = 2.93) and 26.43% (BS eigenvalue = 1.93) of the total variance (Figure 2). Both axes 1 and 2 PCA of surface water data clearly evidenced the formation of two groups, distinguishing samples from the dry–cold (blue symbols) and from the wet–warm (red symbols) seasons. PCA axis 1 mostly distinguished samples with distinct water quality indicators, in which samples from ST1 and ST2 were located on the left side of the biplot and correlated with high concentrations of chlorophyll a (r = 0.77) and ammonia (NH₃; r = 0.76), and lower values of dissolved oxygen (r = 0.68), temperature (r = 0.88), and pH (r = 0.82). Except for the sample from ST2 collected during the warm–rainy season, which was on the right side of the biplot and closely associated with the sample from ST10 collected during the warm–rainy season and correlated with high values of turbidity (r = 0.62). PCA axis 2 mostly evidenced seasonal differences among samples from the other sampling sites (ST3–ST10), in which samples collected during the cold–dry season were located on the upper side of the biplot and correlated with high values of dissolved oxygen (r = 0.60), conductivity (r = 0.63), and pH (r = 0.5). Samples collected during the warm–rainy season were correlated with high values of nitrite ions (r = 0.42) and turbidity (r = 0.71).

The mean values of water quality variables measured at sampling sites in the Guandu River basin during the cold–dry and warm–rainy seasons are presented in

Table 1. Water quality in the surface water of sampling sites measured during cold–dry and wet–warm seasons in the Guandu River basin. Measured variables were temperature (°C), conductivity- Cond (µS/cm), dissolved oxygen—DO (mg/L and %), pH, turbidity (NTU), chlorophyll a—Chlo (µg/L), and the nutrients phosphate—PO₄³⁻ (mg/L), ammonia—NH₃ (mg/L), and nitrite ions—NO₂⁻ (mg/L). Data are shown as mean values.

	Variables	ST1	ST2	ST3	ST4	ST5	ST6	ST7	ST8	ST9	ST10
Cold–dry	°C	22.66	23.0	22.03	23.85	22.15	21.89	21.49	22.72	22.34	21.86
	Cond (μS/cm)	180.0	264.0	89.50	88.00	89.00	90.00	89.50	89.00	86.00	97.50
	DO (mg/L)	3.67	1.70	8.36	6.73	9.03	9.58	9.00	7.90	7.63	8.63
	DO (%)	42.0	16.20	95.75	79.20	103.05	709.65	107.05	94.70	90.30	107.15
	pH	6.11	6.93	6.90	6.28	6.68	6.62	6.35	6.52	6.48	7.34
	NTU	1.77	30.68	1.10	0.35	2.71	3.07	3.21	0.43	1.08	3.75
	Chlo (µg/L)	31.0	64.0	1.00	2.00	1.00	2.50	3.00	4.00	1.00	2.00
	PO43− (mg/L)	2.200	1.550	1.500	1.150	1.250	0.060	0.070	0.050	0.065	0.050
	NH3 (mg/L)	4.531	6.372	0.140	0.018	0.049	0.023	0.036	0.00	0.182	0.024
	NO2− (mg/L)	0.000	0.007	0.000	0.000	0.000	0.033	0.049	0.000	0.000	0.168
Wet–warm	°C	27.96	28.22	24.54	25.51	24.94	25.51	25.46	25.45	25.49	24.53
	Cond (μS/cm)	258.0	82.0	75.5	76.5	74.0	88.0	86.5	73.0	75.5	80.5
	DO (mg/L)	3.85	12.48	7.45	5.23	8.35	7.09	6.57	6.45	6.27	4.24
	DO (%)	49.10	159.9	88.95	64.75	100.8	88.00	82.45	80.60	79.10	55.35
	pH	7.05	8.65	6.49	6.39	6.74	6.40	6.37	6.13	6.29	7.33
	NTU	13.46	17.63	20.63	9.10	18.67	23.98	24.19	21.39	21.81	40.78
	Chlo (µg/L)	73.5	29.5	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
	PO43− (mg/L)	1.310	0.300	0.125	0.000	0.055	0.035	0.05	0.045	0.000	0.065
	NH3 (mg/L)	8.826	0.827	0.079	0.225	0.000	0.055	0.109	0.127	0.091	0.091
	NO2− (mg/L)	0.069	0.049	0.035	0.059	0.058	0.066	0.058	0.041	0.046	0.064

. In general, the values of phosphate, ammonia, and chlorophyll a were higher in ST1 and ST2 than in other sampling sites, while dissolved oxygen showed lower concentrations in those sites. Independent of the sampling site in the river basin, water data showed a tendency to increase turbidity and pH, associated with reductions in nutrients (phosphate, ammonia, and nitrite ions) and chlorophyll a during the warm–rainy season (Table 1).

For the sediment data matrix, the first two PCA axes were selected by the Broken-Stick (BS) method, with axes 1 and 2 significantly (p = 0.0001 and p = 0.001, respectively) explaining, respectively, 45.08% (BS eigenvalue = 2.93) and 23.94% (BS eigenvalue = 1.93) of the total variance (Figure 3). Both axes 1 and 2 of PCA of sediment data clearly evidenced the formation of two groups, distinguishing samples from the dry–cold (blue symbols) and the wet–warm (red symbols) seasons. PCA axis 1 grouped samples from various sampling sites, independent of system characteristics (i.e., lotic, lentic, or reservoirs), with a slight tendency to separate by seasons that were mostly distinguished by grain size (median r = 0.85 and mean r = 0.88), content of carbonates (r = 0.88), total carbon (r = 0.79), total nitrogen (r = 0.82), and total phosphorus (r = 0.88). Samples located on the right side of the biplot (e.g., ST5, ST6, ST7, and ST10) were associated with lower values of those variables, while samples on the left side of the biplot (e.g., ST1 and ST2) were related to higher values. PCA axis 2 was mostly correlated with grain-size parameters—positively for skewness (r = 0.82) and sorting (r = 0.75), and negatively for kurtosis (r = −0.81). There was no tendency for group formation by sampling site location, independent of the seasonal period or characteristic of freshwater systems (i.e., lotic, lentic, or reservoir).

Detailed data concerning sediment classification, grain-size parameters, and the content of organic matter, carbonates, and nutrients (organic carbon, nitrogen, and phosphorus) are shown in

Table 2. Grain-size parameters (mean grain size, median grain size, sorting, skewness, and kurtosis), sediment classification (mean, sorting, asymmetry, kurtosis, and Folk’s classification), and sediment content (organic matter (mg/g), carbonates (%), total organic carbon—TOC (mg/g), total nitrogen (TN), and total phosphorus (TP)) of sediments from sampling sites collected during cold–dry and wet–warm seasons in the Guandu River basin, Rio de Janeiro state, Brazil. Sampling sites are presented by codes, as shown in Figure 1, and comprise distinct freshwater systems, such as lotic waters (ST3, ST5, ST6, and ST7), lentic waters (ST1, ST2, and ST4), and reservoirs (ST8, ST9, and ST10).

	Variables	ST1	ST2	ST3	ST4	ST5	ST6	ST7	ST8	ST9	ST10
Cold–dry	Grain size parameters
	Φmean	3.71	4.73	4.11	5.21	2.65	3.14	1.24	2.87	1.98	2.68
	Φmedian	3.51	4.53	3.93	4.90	2.58	3.14	1.29	3.05	1.90	2.84
	Sorting (Φ)	2.10	2.09	1.92	2.05	1.26	3.45	1.57	2.51	1.84	2.95
	Skewness	0.19	0.17	0.18	0.24	0.23	1.73	−0.01	−0.07	0.12	0.03
	Kurtosis	1.15	1.09	1.15	1.03	1.30	0.32	1.20	1.18	1.85	0.77
	Sediment classification *
	Mean	VFS	CS	CS	MS	FS	VFS	MS	FS	MS	FS
	Sorting	VPS	VPS	PS	VPS	PS	PS	PS	VPS	PS	VPS
	Asymmetry	PA	PA	PA	PA	PA	VPA	AS	AS	PA	AS
	Kurtosis	L	M	L	M	L	L	L	L	VL	P
	Folk’s classification	MS	SM	MS	SM	SM	SGS	SGMS	SGMS	SGMS	SGMS
	Sediment content
	OM (mg/g)	0.630	0.370	0.890	0.650	0.330	0.390	0.460	0.660	0.400	0.620
	Carbonates (%)	3.470	5.020	3.760	4.290	0.960	6.640	0.700	2.760	1.210	1.100
	C (mg/g)	59.94	48.38	25.68	15.10	1.560	0.660	1.320	36.53	2.290	11.45
	N (mg/g)	5.000	4.850	2.150	1.580	0.240	0.060	0.170	1.820	0.270	1.000
	P (mg/g)	0.896	2.394	0.922	0.732	0.353	1.460	0.171	0.481	0.316	0.320
Wet–warm	Grain size parameters
	Φmean	0.89	1.81	0.32	0.78	2.55	2.09	1.61	1.37	1.83	2.30
	Φmedian	0.71	2.03	0.23	0.83	2.56	2.12	1.90	−0.36	2.28	1.91
	Sorting (Φ)	1.93	1.99	1.93	1.38	1.83	0.78	2.24	2.88	2.68	2.64
	Skewness	0.19	−0.05	0.37	−0.07	−0.03	0.04	0.01	0.90	−0.04	0.27
	Kurtosis	1.20	0.77	0.75	0.93	1.60	0.91	1.13	0.55	0.64	0.96
	Sediment classification *
	Mean	CS	MS	CS	CS	FS	FS	MS	MS	MS	FS
	Sorting	PS	PS	PS	PS	PS	MS	VPS	VPS	VPS	VPS
	Asymmetry	PA	AS	VPA	AS	AS	AS	AS	VPA	AS	PA
	Kurtosis	L	P	P	M	VL	M	L	VP	VP	M
	Folk’s classification	SGS	SGMS	SGS	SGS	SGMS	S	SGMS	SGMS	SGMS	SGMS
	Sediment content
	OM (mg/g)	0.420	0.940	0.760	0.090	0.670	0.210	0.840	1.160	0.810	0.500
	Carbonates (%)	0.930	2.390	1.150	0.560	1.470	1.530	1.060	1.110	2.530	0.420
	C (mg/g)	2.610	26.54	9.140	0.550	8.800	0.440	3.610	3.660	19.93	5.540
	N (mg/g)	0.290	1.790	0.590	0.000	0.650	0.000	0.290	0.460	1.170	1.210
	P (mg/g)	0.169	1.050	0.274	0.096	0.425	0.287	0.207	0.187	0.396	0.096

* Sediment classification: 1. codes by mean: MS = medium silt, CS = coarse silt, VSF = very fine sand, FS = fine sand, MS = medium sand, CS = coarse sand; 2. codes by sorting: VPS = very poorly sorted, PS = poorly sorted, MS = moderately sorted; 3. codes by asymmetry: PA = positive asymmetry, VPA = very positive asymmetry, AS = approximately symmetric; 4. codes by kurtosis: L = leptokurtic, VL = very leptokurtic, M = mesokurtic, P = platykurtic, VP = very platykurtic; 5. codes by Kolk’s classification: SM = sandy mud, MS = muddy sand, S = sand, SGS = slightly gravelly sand, SGMS = slightly gravelly muddy sand.

.

Four species of infaunal bivalves were found in the Guandu River basin, including two native species—Anodontites trapesialis (Lamarck, 1819) and Diplodon ellipticus (Wagner, 1817)—and two non-native species—Corbicula fluminea and Corbicula largillierti (Philippi, 1844), belonging to the families Mycetopodidae, Hyriidae, and Cyrenidae, respectively. An assembly of 2900 bivalves was collected during this study, among which 1182 bivalves were sampled in the dry period and 1718 individuals in the warm–wet season. The non-native C. fluminea was the most abundant species, totaling 2553 individuals (88%), followed by the other non-native species C. largillierti (9%) and the native species D. ellipticus (2%) and A. trapesialis (0.6%). The distribution of native species was restricted to the sampling sites at the Guandu lagoon (i.e., ST1, ST2, and ST4), despite the presence of the non-native species at ST4 and ST1 (only in the wet–warm season) (Figure 4). For all the other sampling sites, only non-native species of Corbicula were found during the two seasonal periods.

A constrained redundancy analysis (RDA) was applied to summarize the abundance of bivalve species (i.e., response variable) that can be explained by the environmental conditions of the Guandu River basin (i.e., explanatory variables), following a stepwise forward criterion for the selection of environmental variables. RDA axes 1 and 2 showed a species–environment correlation of 49.06% and, respectively, accounting for a percentage variation of species–environment relation of 57.4% (Eigenvalue RDA 1 = 0.49) and 8.33% (Eigenvalue RDA 2 = 0.08) (Figure 5). All the canonical axes of this RDA were statistically significant (Permutation test, Axis 1: pseudo-F = 15.4 and p = 0.004; and Axis 2: pseudo-F = 10.8 and p = 0.002). RDA evidenced spatial segregation of native species occurrence (A. trapesialis and D. ellipticus) in the lentic areas (ST1 and ST2), where high concentrations of chlorophyll a (forward selection, percentage of explanation = 42.6%, p = 0.002) and ammonia (forward selection, percentage of explanation = 15.8%, p = 0.016) in the surface water are the main environmental drivers distinguishing these sampling sites in the Guandu River basin. The non-native species (C. fluminea and C. largillierti) were associated with all the other sampling sites with better environmental quality, independent of freshwater systems (i.e., lentic, lotic, or reservoirs).

4. Discussion

The results of the present study evidenced the struggling conditions faced by native infaunal bivalves in the Guandu River basin, where solely two native species (A. trapesialis and D. ellipticus) that still occur in this basin are threatened by the spread of non-native Corbicula species and the severe pressure of pollutants that degrade the environmental quality of infaunal bivalve habitats. It is important to highlight that the sampling sites still not invaded by the non-native species (ST1 and ST2) comprise a lentic area in the Guandu River basin known as Guandu Lagoon, although these species were recorded at these sites by [18]. This lentic area suffers from the input of chemical contaminants and organic wastes from its tributaries (e.g., Ipiranga River, Cabuçu River, Queimados River), mostly untreated domestic sewage, leading to high levels of dissolved total phosphorus, cyanobacteria, enteric bacteria and cyanobacteria blooms [46,47,48]. Fecal contamination by human and animal sources (e.g., bovine, swine, and equine markers) was detected by host-specific genetic markers in the waters of the Guandu River, and it was associated with Escherichia coli presence, but in compliance with Brazilian guidelines (<1000 MPN/100 mL) [48]. This severe environmental quality in the Guandu Lagoon (ST1 and ST2) was evidenced by multivariate analyses in the present study, in which high concentrations of chlorophyll a and ammonia explained most of the species–environmental relationship for bivalve species. These results suggest a high tolerance of the native bivalves A. trapesialis and D. ellipticus to the poor environmental quality caused by high pollution levels, mostly evidenced by the organic pollution indicators. Simultaneously, it is also expected that a lower tolerance of the non-native Corbicula species will be found in the worst water quality habitats. The species C. fluminea has demonstrated a significant response to ammonia contamination, in which mortality was detected when experimental concentrations of ammonia nitrogen (NH3-N) in the water increased [49]. Besides organic pollutants, chemical contaminants have already been detected in Guandu Lagoon. Recently, the broad-spectrum insecticide Fipronil—a phenylpyrazole used in industrial, residential, and agricultural sectors—was detected and quantified in the Guandu Lagoon (2.41 µg/L) and two other downstream lotic areas of Guandu River (0.343 and 1.79 µg/L) [50]. Surface waters sampled at two sampling sites (very close to ST1 and ST2) in the Guandu Lagoon induced cytotoxic and genotoxic effects in the Allium cepa system [44]. However, the intensity of cytotoxic and genotoxic effects of polluted waters varied between cold–dry and warm–wet seasons, suggesting the effects of tributary river discharges and an increase in contaminated effluents [51].

The native D. ellipticus and the non-native C. fluminea occurred in sympatry only at the sampling site ST4—a lentic area known as Patos Lagoon and characterized as a system with a transitional environmental quality considering its geographical proximity to Guandu Lagoon but, at the same time, favored by the entrance of water flow from the main Guandu River. This better water quality was indicated by the physical and chemical water variables, such as the dissolved oxygen concentration higher than 5 mg/L and intermediate to low nutrient concentrations. For all the other sampling sites in the Guandu River basin, native species occurred in the Guandu Lagoon (ST1 and ST2), or only the non-native species were found together, or only one non-native species (ST3). However, it is not under discussion the dominance of C. fluminea in the abundance of bivalves and its widespread distribution along the Guandu River basin, occupying all the freshwater systems evaluated herein—lentic, lotic, and reservoirs. Possibly, this high number of non-native Asian clams is affecting the water variables, especially turbidity, chlorophyll a, and nutrients, as well as benthopelagic dynamics affecting organic matter and nutrients in water and sediment in the Guandu River basin. Individuals of C. fluminea have the greatest potential among freshwater bivalves to affect sediment and water exchanges due to their filtering activities [52]. Like other invasive bivalves [53], Corbicula spp. can affect the water quality of invaded systems, increasing water transparency and altering nutrient metabolism [54,55,56]. Corbicula fluminea feeds by filtering organic matter in the water column and cycling organic matter in the sediments of the stream bed, which sustains its rapid growth rate [52]. The widespread distribution of C. fluminea in invaded systems resulted in negative ecological and socioeconomic impacts [57]. A great difference between the native and non-native species in the Guandu River basin is undoubtedly related to the life cycle and reproduction. While Corbicula spp. reach ~35 mm and have a short life span of 2–5 years [57]. In addition, C. fluminea has various reproductive strategies, including sexual dioecious lineages or androgenetic cross- and self-fertilizing hermaphrodites conferring a high fecundity for individuals, which is estimated to range from 25,000 to 75,000 veligers during its lifetime [58].

In contrast, the native A. trapesialis and D. ellipticus are larger individuals (reaching 200 mm) that can live for several years. There is little information about the lifetime of South American mussels, but there are records of species that can live 100 years [59]. They also have complex life cycles that include a parasitic larval stage in fishes. In addition, D. ellipticus and A. trapesialis can filter and remove suspended solids, microalgae, and pathogenic bacteria. However, under stressful conditions, such as poor environmental quality, changes in filtration rates may occur, affecting energy balance, growth, and reproduction [60,61]. Therefore, its rapid growth, earlier sexual maturity, short life span, and high fecundity make Corbicula spp. better competitors in invaded systems, facilitating its widespread colonization of new freshwater environments [57,62]. Moreover, C. fluminea can impair the survival of the parasitic larvae (glochidia) of Unionida freshwater bivalves by filtration and biodeposition, negatively impacting and directly affecting the recruitment of juveniles of the native species [63]. Therefore, invasive species may have pushed out the native species to low-quality environments, making the occurrence of native species in other sites difficult. Complementary, ref. [18] advocated for a natural occurrence of large native mussels at the downstream stretches of the Guandu River. The classical theory of the River Continuum Concept [64] points to the occurrence of large mussels at these portions of the rivers due to naturally high levels of nutrients and chlorophyll a.

The multivariate redundancy analysis evidenced a species–environment relationship explained by indicators of water quality (i.e., chlorophyll a and ammonia concentrations) despite evaluating benthic infauna species. This may be explained by lower differences in sediment characteristics among sampling sites and a slight tendency to separate by seasons, even samples from the same site, mostly affected by grain size and content of carbonates, total carbon, nitrogen, and phosphorus. These results suggest a great instability of the benthic compartment between seasonal periods, probably affected by increased input of sediments from adjacent terrestrial sources or tributaries during the rainy season. While the water variables better distinguished the sampling sites along the river basin, especially those located in the Guandu Lagoon (ST1 and ST2), which showed the worst water quality indicators (higher chlorophyll a and ammonia concentrations and lower dissolved oxygen and pH), they also evidenced differences between the seasonal periods. In tropical regions, the warm–rainy season is characterized by higher accumulated precipitation, increasing the input of rainfall and discharge of tributary rivers that also carry contaminants, such as the pesticide glyphosate detected in reservoirs of the Guandu River basin [65]. Independent of the sampling site in the river basin, there was a tendency to increase turbidity and pH in the warm–rainy season, probably related to organic matter and sediment influx from tributaries, associated with lower concentrations of nutrients and chlorophyll a—probably more diluted considering an increase in river flow. Rainfall seasonality can lead to changes in the physical and chemical water variables, such as turbidity, particularly in areas without vegetation and unprotected regions, due to the transport and resuspension of sediments [66]. Water pH may also change due to differences in the concentration and types of organic matter or pollutants present during different seasonal periods [67]. Thus, even for benthic species, the water variables were more effective in indicating differences in the environmental quality of various freshwater systems (i.e., lentic, lotic, and reservoirs) in the Guandu River basin.

5. Conclusions

In conclusion, the spatial distribution of native bivalves is limited to Guandu Lagoon, the area with the worst water quality in the Guandu River basin, and the low abundances found for these species reinforce a local extinction risk of these native species, especially A. trapesialis, which only occurred in this lentic area (ST1 and ST2). In the Guandu Lagoon, the pollution scenario is a major struggle for the native species since the poor environmental conditions can impact their physiological status and reproduction. Several attempts have been made to improve the water quality of the Guandu River, which provides drinking water for more than nine million people in the metropolitan region of Rio de Janeiro [6], such as the use of bentonite clay, oxygen peroxide, phytoremediation, hydrogen-oxidizing bacteria, and magnetic phosphorus removal [68,69,70]. However, besides some treatment compounds that can have undesirable consequences for human health, none of the treatments seemed to be completely effective in water quality management since they are palliative methods. However, improvements in water quality and environmental conditions in the Guandu Lagoon may also favor the spread of non-native Corbicula to this area, where the genus seems to be tolerance limited. The occurrence of C. largillierti at ST3—a lotic sampling site with high saturation of dissolved oxygen in the main Guandu River, geographically close to the Guandu Lagoon—highlights the concern of the potential spread of Corbicula spp. along the river basin. Thus, an additive effect of pollution and invasive species can enhance the risks of local extinctions of Anodontites and Diplodon species in the Guandu River basin. Unfortunately, native freshwater bivalves are not at risk only in the Guandu River basin; native freshwater bivalves are mostly distributed in areas that offer low levels of protection and frequent sympatry with non-native species in the Rio de Janeiro state, which includes the Paraíba do Sul River [28]. Moreover, the recent introduction of the golden mussel L. fortunei in the Paraíba do Sul River, associated with the predictions of an inevitable introduction in the Guandu River [30], represents an additional factor that endangers these native infaunal bivalves in freshwater ecosystems of the Rio de Janeiro state.