2.1 Study Area

The Brazilian Amazon Costal Zone (BACZ) has an extension of approximately 1200 km, being characterized by two worldwide important geomorphological features: (i) the mouth of the largest river system on the planet (Amazon River); (ii) it is part of the largest continuous extension of mangroves in the world (Prates et al. 2012). In the BACZ the coastline is extremely irregular and jagged, covering 23 estuaries and 30 drainage basins that together drain an area of 330 thousand km² (Martins et al. 2007). In this scenario is located the Caeté Estuary (Figure 1), which is classified as a permanently open, turbid, and shallow ecosystem, with a maximum depth of about 10 m (Dittmar and Lara 2001).

The local hydrodynamic is driven primarily by the tidal regime, but also by local winds and wind-waves (Pereira et al. 2010; INMET 2020). The Caeté Estuary is subjected to semidiurnal tides, with a 4–6 m tidal range during spring tides, and a 2–4 m range during neap tides (Pereira et al. 2010). Fortnightly, during spring tides, the tidal waters reach their highest levels, washing out soil material from the adjacent mangroves. This process leads to an increased concentration of dissolved solutes and nutrients, which are loaded into the estuary (Pereira et al. 2010). The discharge of the Caeté River varies from 180 m³ s⁻¹ during the rainy season to 0.3 m³ s⁻¹ during the dry season (Dittmar 1999), with an annual average of 42.1 m³ s⁻¹ (ANA 2020). Tidal currents, typical of shallow estuaries, can exceed 1.5 m s⁻¹ (Cohen et al. 1999). The salinity of the estuary fluctuates between 0 and 39 (Diele and Simith 2006), mainly influenced by the seasonality of rainfall.

The region's climate is hot and humid, with a period of high precipitation (rainy season) that usually extends from January to July, when total rainfall often exceeds 2.000 mm. The mean annual precipitation is 2321 mm; however, significant deviations from this value due to rainfall anomalies are not uncommon. These deviations have been observed over the past 33 years, during which several droughts linked to El Niño and Atlantic Dipole events, as well as floods caused by La Niña events, were recorded (NOAA 2020). These changes in precipitation patterns (INMET 2020) can significantly influence the availability and temporal distribution of river flow (ANA 2020), as illustrated in Figure 2. Winds blow with a mean intensity of up to 3.0 m s⁻¹, and temperatures are around 26°C–27°C (INMET 2020). During the dry (or less rainy) season, between August and December, monthly rainfall is normally below 100 mm, winds are the strongest, with mean speeds over 4.0 m s⁻¹, and temperatures are around 28°C–30°C (Moraes et al. 2005).

The estuary is surrounded by dense and relatively well-developed mangrove forests, with trees of up to 20 m in height, and a complex network of tidal creeks that link it to the neighboring Taperaçu Estuary (Dittmar and Lara 2001; Monteiro et al. 2016). The input of nutrients and organic matter from the local mangrove controls both primary and secondary production in the Caeté Estuary and adjacent habitats of the Amazon coast (Dittmar and Lara 2001). Due to its biogeochemical, ecological, and socioeconomic importance, the study area was designated as a conservation unit (RESEX Caeté-Taperaçu) by the Brazilian government, created on May 20, 2005, by ordinance 17, decree without number (Brasil 2005).

2.2 Sampling Strategy

To understand the link between the anomalous period of climate (rainfall levels showing reductions equal to or exceeding 60% of historical averages; INMET 2020), hydrodynamics, hydrology, and the attributes of the copepod community in the Caeté Estuary, the spatial and temporal variations in these variables were studied over five sampling campaigns conducted within a one-year period, from June 2013 to June 2014: (i) drought months: 15–16th June, 11–12th October, and 9–10th December 2013; and (ii) postdrought months: 8–9th April and 7–8th June 2014. Each monthly campaign was conducted during the neap tide over a 25-h sampling period at three fixed stations located in the inner (S1), middle (S2) and outer (S3) sectors of the estuary (Figure 1).

Total monthly rainfall and Caeté River discharge data were obtained for the study period and plotted against monthly average precipitation over the past 33 years to highlight the driest months. These data sets are freely available at the Web sites www.inmet.gov.br (Instituto Nacional de Meteorologia-INMET, Tracuateua station: −1.06° S, −46.9° W) and www.ana.gov.br (Agência Nacional de Águas e Saneamento Básico-ANA, Nova Mocajuba station: −1.16° S, −46.5° W).

The oscillations in water levels were measured simultaneously, every 10 min, at the three stations using a bottom-mounted mooring, to which tide gauges were attached. To understand how reductions in the rainfall levels, related to the drought event, affect the oscillations in the hydrological variables, data on salinity, dissolved oxygen (DO) concentrations, temperature, turbidity were measured in situ (every 10 min) at the subsurface (1 m depth) using CTDs (XR-420-RBR) with Dissolved Oxygen (DO) and turbidity sensors. Water samples (500 mL) were collected every 3 h from the subsurface with a 5 L Niskin oceanographic bottle (General Oceanics) for the determination in the laboratory of the hydrogenionic potential (pH), chlorophyll-a (chl-a) and dissolved nutrient concentrations (nitrate: $$$ {\mathrm{NO}}_3^{-} $$$, nitrite: $$$ {\mathrm{NO}}_2^{-} $$$, ammoniacal nitrogen $$$ {\mathrm{NH}}_4^{+} $$$, dissolved inorganic nitrogen: DIN, total dissolved nitrogen: TDN, dissolved silicate: DSi, orthophosphate: $$$ {\mathrm{PO}}_4^{3-} $$$, total dissolved phosphorus: TDP).

Subsurface zooplankton samples were collected at 3-h intervals, in which each sample consisted of a 3 min subsurface horizontal tow of a conical plankton net (200-μm mesh size and 0.50-m mouth diameter) equipped with a mechanical flowmeter (General Oceanics 2030R) to estimate the volume of water filtered through the net. Nine tows per station were conducted each month by using a small powerboat running at an average speed of 1.5 knots, resulting in a total of 135 samples. As soon as they were collected, the samples were transferred to 600 mL plastic bottles and preserved in a 4% buffered formaldehyde seawater solution (sodium tetraborate).

2.3 Laboratory Analysis

The pH was determined by a pH meter (Labmeter, pH 2—Hs—3B). The water samples collected for the determination of chlorophyll-a (phytoplankton biomass) and dissolved nutrient concentrations were vacuum-filtered through glass-fiber filters (Macherey-Nagel GF-5 47 mm, Carvalhaes) and the filters and the water samples were then freeze-dried for further analysis of both variables, respectively. Chlorophyll-a was extracted with 90% acetone v/v and determined spectrophotometrically, according to the protocol of Parsons and Strickland (1963) and UNESCO (1966). Equations were applied to obtain the chlorophyll-a concentrations of each sample. Dissolved inorganic nutrient concentrations were determined by spectrophotometry, following the procedures described by Strickland and Parsons (1972) and Grasshoff et al. (1983). Dissolved inorganic nitrogen (DIN) levels were determined from the sum of nitrate + nitrite + ammoniacal nitrogen.

The preserved zooplankton samples were rinsed to remove the formalin buffer and divided into aliquots between four and 11 times in a Folsom splitter to provide standardized subsamples containing at least 300 copepods, which were then identified (Björnberg 1981; Bradford-Grieve et al. 1999), classified taxonomically (WoRMS 2021), and counted in a gridded Petri dish under a stereomicroscope (Zeiss, Stemi 2000). The counts for each copepod species were multiplied by the subsampling factor (4–11) to obtain an estimate of the total number of individuals in the total sample. The adult copepods were identified to species or to the lowest taxonomic level possible. Juveniles and larval stages (nauplius) were identified to the genus or group level, respectively.

The quantitative data obtained for each sample were used to calculate density (ind m⁻³), relative abundance, and the frequency of occurrence (FO) of each taxon. Species diversity (Shannon 1948), evenness (Pielou 1977), and richness of copepods (total number of species in each sample) was calculated using only adult organisms identified at the species or morphospecies level. The frequency of occurrence of a given species was determined by the proportion of samples in which it was recorded. In the present study, species were considered dominant if they contributed > 30% of the individuals present in a given sample.

2.4 Statistical Analysis

The analysis of variance (ANOVA—one and two-way) and the nonparametric Mann–Whitney U test and Kruskal–Wallis H test were used to compare environmental variables (salinity, pH, dissolved oxygen, temperature turbidity, chlorophyll-a and dissolved nutrients) and copepod biological attributes (species density, diversity, richness and evenness) between spatial (sampling stations) and temporal scales (diel, monthly, seasonal and inter-annual). The one-way ANOVA followed by Fisher's least significant difference (LSD) multiple comparison post hoc tests or Kruskal–Wallis followed by Student–Newman–keuls's multiple comparison post hoc tests were used to discern significant differences between treatments when significant F values were recorded (Zar 2010). Correlations between environmental variables were evaluated using Spearman rank correlation analysis (r_s). All these analyses were run in STATISTICA 8, with α = 0.05.

Principal component analysis (PCA) was used to detect possible relationships between environmental variables, as well as to identify the main sources of variation. For this analysis, the following data were inserted: Rain: rainfall, Flow: flow rate, Sal: salinity, pH: hydrogenionic potential, DO: dissolved oxygen, Temp: temperature (°C), Turb: turbidity, Chl-a: chlorophyll-a, Nitrate: NO₃⁻, Nitrite: NO₂⁻, ammoniacal nitrogen, DIN: dissolved inorganic nitrogen, TDN: total dissolved nitrogen, DSi: dissolved silicate, orthophosphate: $$$ {\mathrm{PO}}_4^{3-} $$$ and TDP: total dissolved phosphorus. The sampling sectors and the months of collection were inserted as supplementary variables in order to investigate the spatial and temporal arrangement of the data. PCA was performed using the statistical program PRIMER, version 6.

The relationships among environmental variables and the main copepod species were assessed using Canonical Correspondence Analysis (CCA), conducted in CANOCO 4.5 (ter Braak and Šmilauer 2002). The significance of the CCA was determined through a Monte Carlo permutation test comprising 499 permutations, followed by a reduced model (p < 0.05). This analysis involved evaluating response (main copepod species) and explanatory (environmental variables) matrices, aiming to ascertain the distinct and shared influences of these factors on the observed variability in the copepod community.

3.1 Shift in Rainfall Levels and Their Effects on River Flow, Salinity, and Chlorophyll-a

In order to present the effects of rainfall anomalies on temporal variations of hydrological variables, Figure 3 shows changes in seasonal rainfall levels and river flow and their consequences on salinity and chlorophyll-a concentrations between 2013 and 2014. These changes characterize both drought and nondrought periods, respectively. The nondrought period of 2014 was characterized by an annual rainfall of 2066 mm, which closely resembled the historical mean for the region (1982–2014 = 2316 mm). However, during the drought of 2013, the annual rainfall was 1612 mm, showing a decrease of 31% compared to the historical mean rainfall. This reduction was even more pronounced in the first semester (rainy season), with a 67% lower in January, and substantial reductions in February (70%), June (42%), March (22%), April (18%), and May (7%) (Figure 3a). During the last months of the drought period (October and December 2013), the recorded rainfall levels were also considerably lower, with reductions of approximately 93% and 99%, respectively, compared to the same months in the historical dataset. This observation highlights the continuation of abnormal rainfall patterns, extending at least partially into the second half of the year (dry season). The Caeté River discharge in 2013 (26.82 ± 26.91 m³ s⁻¹) was significantly lower (ANOVA: F = 36.3, p < 0.0001) than in 2014 (44.54 ± 32.80 m³ s⁻¹), indicating the influence of the precipitation anomaly on this variable (Figure 3a). A strongly positive correlation between these two variables was observed (r_S = 0.79, p < 0.001; Table S1).

By comparing the rainfall levels, river discharge, and water salinity between the two study years, it becomes evident that a 454 mm reduction during the 2013 drought resulted in a significantly decreased riverine discharge (33%) and high salinity levels (41%) compared with the nondrought period of 2014 (p < 0.0001). Salinity anomalies can be observed in Figure 3b and Figure S1, where reduced rainfall levels and river discharges (June 2013) correspond to high salinity and chl-a concentrations (Figure 3c). The salinity and chl-a values exhibited significant negative correlations with freshwater discharge (Table S1).

3.2 Hydrodynamics and Hydrological Variables

Hydrodynamics in the Caeté Estuary is controlled by the river discharge and tides, with water level oscillations (range 1–4 m) typical of neap tide mesotidal conditions. Tide oscillations were marked by accentuated spatial and temporal asymmetry, characteristic of the semidiurnal tidal regime, with a longer ebb and shorter flood phases. This asymmetry was more pronounced during the drought period, which is evident when comparing the month of June between the years 2013 and 2014. In June 2013, with rainfall levels considerably below normal (typical years), the inner-S1 sector of the estuary showed greater differences in the duration of ebb (8 h 20 min–5 h 20 min) and flood (4 h 20 min–4 h 40 min) tides compared to the same month in 2014 (ebb = 6 h 40 min–6 h 40 min; flood = 5 h 30 min–6 h 00 min). A similar pattern was also observed in the middle-S2 sector.

In the present study, no systematic short-term patterns (circadian and tidal cycles) were statistically identified for the hydrological variables. For this reason, the data were pooled as means and standard deviations (±SD) for the spatial and temporal (monthly and seasonal) analysis. The oscillations of the water variables were influenced by physical forces, such as precipitation and river flow, as well as by biological processes (e.g., photosynthesis), which were affected by anomalous rainfall periods.

Seasonally, average salinity fluctuated from 7.03 ± 10.89 in the rainy season to 19.71 ± 5.64 in the dry season, being significantly higher in the latter period (F = 26.5; p < 0.001). Monthly, the values varied from 0.02 ± 0.00 in the inner-S1 sector in April/2014 to 38.19 ± 0.49 in the outer-S3 sector in December/2013 (F = 7.2; p < 0.001) (Figure S1). Negative and significant correlations were observed between precipitation and salinity (r_s = −0.48; p < 0.001), and between river discharge and salinity (r_s = −0.47; p < 0.001) (Table S1).

Similar to the variations in salinity, the pH presented monthly averages varying between 6.26 ± 0.21 in the inner-S1 sector in April/2014 and 8.21 ± 0.04 in the outer-S3 sector in December/2013 (H = 22.1; p < 0.001). DO concentrations were higher in June/2013 in the outer-S3 sector of the estuary (4.91 ± 0.29 mg L⁻¹; H = 19.2; p < 0.001) (Figure S1). Seasonal differences were not observed for DO concentrations (Figure S1), which showed positive and significant correlations with pH and salinity (Table S1). The water temperature remained high and relatively stable (varying only 3.3°C) throughout the study period (Figure S1).

The turbidity values increased significantly (F = 6.7; p < 0.05) from the dry season (138.94 ± 171.39 NTU) to the rainy season (187.05 ± 159.9 NTU). Monthly averages ranged from 17.15 ± 8.76 NTU in the outer-S3 sector in April/2014 to 364.01 ± 130.91 NTU in the inner-S1 sector in October/2013 (F = 9.0; p < 0.0001; Figure S1). Peaks of up to 42.96 ± 14.46 mg m⁻³ were observed for chlorophyll-a concentrations in the inner-S1 sector in June/2013 (Kruskal–Wallis: H = 37.2; p < 0.0001) (Figure S1). A positive and significant correlation (r_s = 0.34; p < 0.001) was recorded between turbidity and Chl-a concentrations (Table S1).

Besides salinity and DO, the dissolved inorganic nutrients (nitrate: $$$ {\mathrm{NO}}_3^{-} $$$, nitrite: $$$ {\mathrm{NO}}_2^{-} $$$, TDN, DSi, orthophosphate: $$$ {\mathrm{PO}}_4^{3-} $$$ and TDP) peaked during the drought months. The nitrate: $$$ {\mathrm{NO}}_3^{-} $$$ (23.82 ± 4.10 μmol L⁻¹, p < 0.0001), nitrite: $$$ {\mathrm{NO}}_2^{-} $$$ (6.26 ± 2.68 μmol L⁻¹, p < 0.01), DIN (25.48 ± 4.58 μmol L⁻¹, p < 0.001), orthophosphate: $$$ {\mathrm{PO}}_4^{3-} $$$ (0.94 ± 0.18 μmol L⁻¹, p < 0.0001) and TDP (1.02 ± 0.17 μmol L⁻¹, p < 0.0001) were highest in December (inner-S1 and middle-S2 sectors), while TDN (223.84 ± 56.85 μmol L⁻¹, p < 0.0001) and DSi (223.84 ± 56.85 μmol L⁻¹, p < 0.0001) concentrations were the highest in October at middle-S2 estuary. By contrast, ammoniacal nitrogen $$$ {\mathrm{NH}}_4^{+} $$$ increased in the postdrought months. Generally, the dissolved nutrients (Figure S2) showed a horizontal gradient, with increasing values observed toward the upstream estuary. This spatial trend was also reported for chl-a, which was confirmed by its positive correlation with nitrate: NO⁻₃ (r_s = 0.43, p < 0.001), DIN (r_s = 0.40, p < 0.001), as well as others nitrogenous and not nitrogenous compounds (Table S1).

The effects of atypical climatic conditions (low rainfall) on hydrological variables are better understood by comparing the month of June between years (2013 and 2014). In this context, an increase in salinity (Mann Whitney: U = 234, p < 0.05), DO (Mann Whitney: U = 201, p < 0.01) and turbidity (ANOVA: F = 18.4, p < 0.0001) was observed in June/2013 (drought event) in comparison with the same month in 2014, when normal rainfall conditions prevailed (Figure S1). A similar variation pattern was recorded for nitrate: $$$ {\mathrm{NO}}_3^{-} $$$ (ANOVA: F = 13.8, p < 0.0001), orthophosphate: $$$ {\mathrm{PO}}_4^{3-} $$$ (ANOVA: F = 9.0, p < 0.01) and chl-a concentrations (Mann Whitney: U = 132.0, p < 0.0001) (Figures S1 and S2), resulting in more eutrophic waters in June/2013 when dilution of city sewage effluents released into the estuary was less effective. This occurred due to a marked decrease in river discharge in June 2013 (36.71 m³ s⁻¹) compared to the same month in 2014 (76.24 m³ s⁻¹).

3.3 Principal Component Analysis (PCA): Seasonal Variations of the Environmental Variables

The application of PCA analysis to environmental data reveals that the first two PCs had eigenvalues greater than one, indicating that all of them were significant. These components accounted for 53% (rainy season) and 67% (dry season) of the variance (Figure 4) and represent a very good description of the environmental structure. The results are presented only for the first principal component (PC1) since it explained the major variability in the sampling months and sectors (supplementary variables). Variables nitrate, orthophosphate, turbidity, DIN, and chl-a (coefficients of 0.341, 0.294, 0.289, 0.287 and 0.258) present the highest positive loads to PC1 of the rainy season ordination plot (PC1 = 30%, Figure 4a,c), while precipitation, river flow, TDN, ammoniacal nitrogen, and DO (coefficients of rainy −0.387, −0.387, −0.250, −0.053 and − 0.031) present negative loads. These variables were spatially linked with the sampling months and made it possible to identify a temporal variation. Samples from June 2013 and April/June 2014 were discriminated and positioned on opposite sides of the PC1 axis, forming two distinct groups: (I) the drought month of June 2013 and (II) the postdrought months of April and June 2014 (Figure 4a). Group I was affected by a reduction in river flow due to low rainfall levels, which was the primary mechanism leading to increased values of many water variables in June 2013, such as salinity, nitrate, orthophosphate, chlorophyll-a (chl-a), among others. An opposite trend was recorded in periods of higher rainfall (nondrought months, group II), leading to decreased values of these variables.

The PC1 of the dry season ordination plot accounted for 39% of the variance (Figure 4b,d), with variables such as DIN, nitrate, TDN, DSi, and turbidity (coefficients of 0.378, 0.375, 0.359, 0.322, and 0.318) being the primary positive contributors to this axis. The salinity, pH, DO, TDP, and river flow (coefficients of −0.329, −0.329, −0.215, −0.056 and −0.033) showed an opposite pattern, contributing negatively. This component revealed a distinct spatial oscillation in environmental variables between the drought months of October and December, leading to the identification of two distinct groups (Figure 4d). Group I, consisting of samples from the inner-S1 and middle-S2 sectors, displayed estuarine characteristics, characterized by low to moderate salinity values and high concentrations of dissolved inorganic nutrients (DIN, nitrate, TDN, DSi). In contrast, group II exclusively comprised samples from the Caeté Estuary mouth, influenced by marine forces that were particularly pronounced during the drought event.

3.4 Copepods

Copepoda was by far the dominant taxon in terms of relative abundance, accounting for 83% of total zooplankton abundance in the drought period and 73% in the nondrought period. A total of 23 copepod species, belonging to three orders—Calanoida, Cyclopoida, and Harpacticoida—were identified (Table 1). The Calanoida order was the most abundant and taxonomically diverse, represented by six families, 10 genera, and 14 species. Among these, the numerically dominant species were Acartia tonsa Dana, 1849, Acartia lilljeborgii Giesbrecht, 1889, Paracalanus quasimodo Bowman, 1971, and Pseudodiaptomus richardi Wright S., 1936 (Figure 5a–f). These species, together with the cyclopoid Oithona hebes Giesbrecht, 1891, which exhibited the highest relative abundances during the rainy months of 2014 at stations S1 and S2, represented more than 90% of total copepod relative abundance in October at the inner-S1 sector (Figure 6). These species presented accentuated spatial and temporal oscillations in terms of density, with significant interactions among sampling sectors v. months and sampling sectors vs. seasons detected for A. tonsa, A. lilljeborgii, and P. quasimodo (Table S2).

A. tonsa showed significant seasonal variation, with density increasing from 135 ± 472 ind m⁻³ in the rainy season to 479 ± 1.996 ind m⁻³ in the dry season (ANOVA: F = 16.4, p < 0.001), when it reached 94% of frequency of occurrence (Table 1). On a monthly scale (Figure 5c), the highest densities of this species were recorded in October in the middle-S2 sector of the estuary (2258 ± 4628 ind m⁻³; Kruskal-Wallis: H = 19.39, p < 0.001), accounting for 42% of total copepod relative abundance (Figure 6). Seasonal variations were also observed for A. lilljeborgii, with density ranging from 117 ± 624 ind m⁻³ in the rainy season to 163 ± 532 ind m⁻³ in the dry season (ANOVA: F = 16.4, p < 0.001), being present in 70% of the samples analyzed. A peak in the density of this species was observed in the outer-S3 sector, in October/2013 (1045 ± 1675 ind m⁻³; Kruskal-Wallis: H = 32.7, p < 0.0001; Figure 5f), contributing 39% of total copepod relative abundance (Figure 6).

The density of P. richardi increased significantly from 127 ± 666 ind m⁻³ in the rainy season to 292 ± 897 ind m⁻³ in the dry season (Mann–Whitney: U = 1698, p < 0.05), occurring in 63% of the total samples (Table 1). Similarly, P. quasimodo density values were higher during the dry season (324 ± 1216 ind m⁻³) compared to the rainy season (45 ± 223 ind m⁻³) (ANOVA: F = 44.1; p < 0.05). In the dry period, this species reached a 91% frequency of occurrence (Table 1). Monthly, the highest densities for both species were observed in the middle-S2 sector in October/2013 (P. richardi = 983 ± 1856 ind m⁻³; Kruskal–Wallis: H = 23.4, p < 0.0001; Figure 5d and P. quasimodo = 1461 ± 2715 ind m⁻³; Kruskal-Wallis: H = 41.0, p < 0.0001; Figure 5e), contributing 18.1% and 26.9%, respectively, to total copepod relative abundance (Figure 6).

An opposite pattern was observed for O. hebes, which showed higher density during the rainy season (19 ± 85 ind m⁻³; Mann–Whitney: U = 1520.5, p < 0.01), when it occurred in 74% of the samples analyzed (Table 1). Its highest density values were observed in the middle-S2 sector in June/2013 (74 ± 67 ind m⁻³; Kruskal–Wallis: H = 22.4, p < 0.001), representing a relative abundance of 7.3% (Figures 5b and 6).

Considering the five dominant species, an accentuated decline in the mean density of A. tonsa, as well as that of A. lilljeborgii, P. richardi, and P. quasimodo was recorded during the postdrought months (Figure 5), with values below 10 ind m⁻³ for all these species in April and June of 2014 at the inner-S1 and middle-S2 sectors, respectively (Figure 5), which reflected the increase of freshwater discharge from the Caeté River (see Figure 3a). In addition, when comparing the densities of these species in June/2013 (P. richardi = 375 ± 1126 ind.m⁻³, A. lilljeborgii = 348 ± 1056 ind.m⁻³, A. tonsa = 315 ± 782 ind.m⁻³ and P. quasimodo = 122 ± 378 ind.m⁻³) and June/2014 (P. richardi = 4 ± 14 ind.m⁻³, A. lilljeborgii = 1 ± 4 ind.m⁻³, A. tonsa = 45 ± 105 ind.m⁻³ and P. quasimodo = 7 ± 15 ind.m⁻³), the highest significant values (p < 0.05) were recorded in 2013 (drought event, Figure 5) when river discharge was approximately 52% lower than in 2014. Still comparing the June samples (2013 and 2014), an increase of approximately 93% in the density of P. richardi ovigerous females was observed during the drought event in 2013. This pattern suggests that (i) the high primary biomass (chlorophyll-a concentrations) observed in June/13 favored the development of organisms of these species; (ii) the lower volume of precipitation and river flow recorded in 2013 due to the occurrence of the drought event reflected in the increase in salinity to values within the optimal tolerance of its populations, reflecting their preferences for more saline waters. The influence of the drought event on the dynamics of the species of Acartia and P. quasimodo were confirmed by the positive and significant relationships observed between salinity and density of A. tonsa (r_s = 0.72, p < 0.0001), A. lilljeborgii (r_s = 0.75, p < 0.0001) and P. quasimodo (r_s = 0.67, p < 0.0001). The population of P. richardi was directly correlated with the concentrations of chlorophyll-a (r_s = 0.37, p < 0.0001). Comparisons of copepod densities between the present study period (drought event) and a number of other coastal environments under neutral and anomalous rainfall conditions are shown in Table S3.

On a spatial and temporal scale, the Caeté Estuary was considered a heterogeneous estuary in terms of total copepod density, with peaks (p < 0.05) observed in the middle-S2 sector, in October/13 (5722 ± 9153 ind m⁻³), when the density of some of the main species increased considerably (Figure 5a–f). Diversity ranged from 0.46 ± 0.18 bits ind⁻¹ in the outer-S3 sector in April/14 to 2.15 ± 0.33 bits ind⁻¹ in the outer-S3 sector in December/13, and species richness of the copepod community (3.00 ± 1.22 in the inner-S1 sector, in April/14 to 10.11 ± 2.42 in the outer-S3 sector, in October/13) varied significantly between seasons, as well as between months and sectors of study (p < 0.05), showing a spatial and temporal heterogeneity (Figure S3). These patterns of variations were not observed for evenness (Figure S3), which oscillated between 0.19 ± 0.05 and 0.85 ± 0.12 in the respective outer-S3 and inner-S1 sectors of the estuary in April/14. Overall, diversity and evenness decreased with the increased relative contribution of the dominant species to total copepod abundance (Figure 6), showing a lower value in April/14 (in the middle-S2 sector) during the nondrought period.

The environmental preferences of the dominant species selected for the Caeté Estuary during the rainy and dry seasons explored by Canonical Correspondence Analysis—CCA, are shown in Figure 7.

During the rainy season (Figure 7a), the first two CCA axes explained 80.6% of the total variance (axis 1 = 63.2%; axis 2 = 17.4%; p = 0.001 for both axes) (Figure 11a). Species-environment correlations were high for the two observed axes (r = 0.946 and r = 0.756, respectively; p = 0.001 in both cases), indicating a strong relationship between environmental variables and the distribution of main species. The first canonical axis, which explained more than 60% of the variance, was strongly correlated with salinity, and to a lesser extent, with nitrite-NO₂⁻, dissolved oxygen (DO), and rainfall. The second axis correlated with river flow, nitrate-NO₃⁻, turbidity, and chlorophyll-a concentrations. The reduction of freshwater discharge, as a result of the low volume of rainfall in the study period of 2013, was the main factor responsible for the separation of the upper and intermediate sectors (S1 and S2) of the June samples (2013 and 2014), while the grouping of samples from the lower sector (S3) was associated with the strong marine influence in this portion of the estuary. When precipitation and flow levels were normal, there was an association between samples from the upper and intermediate sectors of April and June 2014, characterized by low salinity values (Figure 7).

The CCA plotted with the dry season samples (Figure 7) showed spatial and temporal homogeneity, with the first two axes explaining 95.7% of the total variance (axis 1 = 79.7%; axis 2 = 16%; p = 0.001), which presented the respective correlations: r = 0.976, p = 0.001 and r = 0.593, p = 0.001. The first axis showed a strong correlation with turbidity, nitrate-$$$ {\mathrm{NO}}_3^{-} $$$, and chlorophyll-a concentrations. Salinity, nitrite-$$$ {\mathrm{NO}}_2^{-} $$$, and river flow were correlated with the second axis. The upper sector (S1) differed from the lower sector (S3), as it is strongly influenced by river discharge, responsible for the marked reduction in salinity during the rainy season in this portion of the estuary. In Figure 7, it is possible to observe the distribution of the main copepod species according to the best environmental conditions for their occurrence.

4.1 Influence of Climatic Variability on Copepod Populations

Periods of droughts and floods in the Amazon region are increasingly severe (Lewis et al. 2011; Marengo et al. 2013; Pereira et al. 2018). Despite this, few studies have focused on the consequences of anomalous patterns in rainfall and river discharge on the characteristics of these coastal waters, and the impacts of this variation on the local plankton community. Anomalous reductions in rainfall on the Amazon coast directly affect river discharge in natural systems in the region (Marengo et al. 2013), including dozens of estuaries, causing reductions in freshwater input, modifying the supply and dynamics of nutrients and consequently the phytoplankton development (Pereira et al. 2018; Oliveira et al. 2022), affecting subsequent trophic levels. As water declines in water bodies, ecological processes are changed and the biota can be drastically affected, a factor which may alter the pattern of recruitment, impair or facilitate the reproduction and development of certain zooplanktonic organisms, such as copepods, which play an important role in the link between primary producers and a large part of fishing resources in the estuaries (Munk et al. 2003; Marques et al. 2007; Primo et al. 2009).

In Amazonian estuaries, the high river discharge during the rainy season is substantial (Pamplona et al. 2013; Pereira et al. 2010; Atique et al. 2017), reducing the salinity of coastal waters in the region, including those of the Caeté Estuary (Pereira et al. 2010; Monteiro et al. 2016; Atique et al. 2017). Thus, the effect of the drought event appears to be greater during the rainy season, possibly because 80%–90% of the total annual precipitation occurs during this period. In this study, the low levels of rainfall and river discharge unusually recorded low in June 2013 resulted in more saline and eutrophic waters. In fact, the water was more saline, alkaline, and oxygenated, and the concentrations of dissolved nutrients and chlorophyll-a were much higher (over 40%) in June 2013 compared to June of the following year (nondrought), reflecting lower rainfall and river discharge conditions (less than 49% and 52%, respectively) compared to the same period in 2014. Overall, then, during the rainy season under typical conditions, these waters were less saline than those recorded during the drought event, a period in which peaks of salinity around 30 were observed in the outer-S3 sector of the estuary, values 10% and 30% higher than those observed in 2006 (Atique et al. 2017) and 2014 (present study).

According to Asp et al. (2018), in the Caeté Estuary, a zone of maximum turbidity can be observed in the middle-S2 sector, and its high values were mainly provided by mangrove runoff and the resuspension, transportation, and redistribution of large amounts of fine sediments and mangrove materials. In this case, the turbidity observed in the present study appeared to originate from lithogenic and biological materials. This variable represents the main physical factor regulating the availability of sunlight for the estuary's primary producers (Burford et al. 2012; Andrade et al. 2016) and may limit the productivity of the phytoplankton, although in the present study it seems that it was not the case. One of the highest chlorophyll-a concentrations was recorded during the period of high turbidity in June 2013 (inner-S1 sector). Conversely, the high chlorophyll-a value obtained in October 2013 (inner-S1 sector) coincided with relatively low turbidity levels, and in some months (e.g., April 2014), low values of chlorophyll-a and turbidity were recorded simultaneously. Thus, in the Caeté Estuary, turbidity seemed to be less important than nutrient concentrations in the determination of phytoplankton biomass. However, it is also relevant to comment that the presence of resuspended microphytobenthos associated with mangrove and wetland areas, as observed for other tropical (Matos et al. 2011; Pamplona et al. 2013; Oliveira et al. 2022; Queiroz et al. 2022) and subtropical estuaries (Pereira-Filho et al. 2001), may have also contributed to the increase of chlorophyll-a values during the study period.

The mean phytoplankton biomass (based on chlorophyll-a concentrations) was similar to those of estuarine areas with high phytoplankton production and nutrient loads from anthropogenic sources (Mallin et al. 2005; Guenther et al. 2015). As in other coastal ecosystems in the Amazon (Matos et al. 2011; Pamplona et al. 2013), the highest concentrations of chlorophyll-a were recorded in the rainy season, related to high river discharge and increased erosion of local mangroves by rainfall. The peak in chlorophyll-a concentrations and an increase in nitrate and orthophosphate concentrations during the drought event resulted in more eutrophic waters in June 2013, when dilution of sewage effluents released into the inner sector of the estuary was less effective. According to data previously reported for the Caeté Estuary (Monteiro et al. 2016), a mean monthly river discharge below 45.0 m³ s⁻¹ can be considered to be the threshold limit to converting river flow into a mechanism that significantly increases the concentrations and consequently the impacts of released effluents. The effects of this lower dilution of discharged effluent are more pronounced in the region surrounding the urbanized zone of the study area (Bragança city, inner-S1 and middle-S2 sectors; Figures 5 and 6), where 90% of the local population is concentrated and a public sanitation system is absent. These facts explain the peaks of chlorophyll-a, $$$ {\mathrm{NO}}_3^{-} $$$, $$$ {\mathrm{NO}}_2^{-} $$$, TDN, DSi, orthophosphate, and TDP observed in the drought months in the inner sectors. Thus, our results suggest a high vulnerability of this estuarine system to climatic anomalies, such as the positive dipole in the Tropical Atlantic and the El Niño event in the Tropical Pacific, which induce droughts in the study area through large-scale ocean–atmosphere connections (see Marengo et al. 2013) and directly affect local environmental variables.

The use of univariate and multivariate analyses allowed us to distinguish three distinct ecological areas based on their biological composition and their relationship with hydrologic variables (see PCA and CCA analysis). The inner sector (Sector 1) of the studied estuary was marked by reduced salinity levels. The community in this area was dominated by juvenile stages of Pseudodiaptomus, with peaks up to 259 ± 543 ind m⁻³ observed. The second sector was defined as a transition area influenced by tides and by freshwater flow in periods of high precipitation. This area is mainly characterized by estuarine and coastal species, particularly during the dry season and drought periods when river flow decreases. The last sector comprised the mouth of Caeté Estuary (outer sector), where water circulation is predominantly tide-dependent (Atique et al. 2017).

The main factor driving the observed trend in copepod densities in the Caeté Estuary was the reduction in river flow caused by the drought event of 2013, which led to decreased precipitation. This period was associated with an increase in copepod density, with prevalence of estuarine and coastal species, such as the adults of A. tonsa, A. lilljeborgii, P. quasimodo, and P. richardi and O. hebes. These higher densities were probably influenced by the combined effects of an optimum salinity level and the high availability of food (phytoplanktonic biomass) during this period. Despite the significance of phytoplanktonic carbon for zooplankton, it is important to highlight that microphytobenthos could also play a significant role in the food chain of shallow inshore waters within small estuaries. This role was demonstrated through stable isotope studies conducted by Chew et al. (2007) and Giarrizzo et al. (2011). The microphytobenthos, characterized by their substantial abundance and productivity, hold nutritional significance for micro-, meso-, and macrofaunal grazers within shallow aquatic ecosystems (Aberle-Malzahn 2004). Consequently, these microorganisms could potentially act as important contributors to the food web of the Caeté Estuary.

According to CCA analysis, the spatial distribution of copepod assemblages has been demonstrated to correlate with the extent of seawater intrusion. This correlation arises due to osmoregulatory effects that influence their ecological tolerance for salinity and reproductive capabilities, such as hatching success, egg production, and nauplii development. This pattern is not unique and has been observed in various tropical, subtropical, and temperate estuaries worldwide (Ara 2004; Menéndez et al. 2012; Lu et al. 2019). In the outer-S3 sector of the estuary, the Copepoda group was predominantly composed of A. tonsa, A. lilljeborgii, and P. quasimodo species, which are characteristic of high salinity waters (Cervetto et al. 1999; Primo et al. 2009; Magalhães et al. 2015). The river flow held particular significance for the dominant species, Acartia and P. quasimodo, as these copepod species are broadcast spawners (Huys and Boxshall 1991), releasing their eggs freely into the water column. Consequently, these eggs are more susceptible to advective transport (Kimmel et al. 2006), influencing population dynamics and potentially leading to the removal of other developmental stages (resulting in decreased abundance) from the estuarine waters. During the drought event, it appears plausible for these species to progress from egg to adulthood without being washed out of the estuary. As river flow exhibits a decreasing trend, corresponding variations in environmental variables such as salinity arise (Marques et al. 2007; Leite et al. 2016). The high salinity levels observed during dry periods result from anomalously low rainfall years, and this relationship is confirmed by a significant correlation between salinity and precipitation. Salinity has a strong influence on copepod distribution (Marques et al. 2007; Barros et al. 2019; Andrade et al. 2016; Leite et al. 2023), emerging as a pivotal factor governing copepod distribution in coastal zones directly influenced by freshwater inputs (Primo et al. 2009; Atique et al. 2017; Calliari et al. 2019; Andrade et al. 2022).

The high degree of phenotypic plasticity observed in A. lilljeborgii populations in response to variations in salinity allowed its survival during both dry and rainy seasons. Ara (2001) studying the estuarine region of Cananéia (São Paulo, Brazil), documented the preference of these organisms for waters with salinities exceeding 17, reporting higher density values during the dry season. High densities of this species were also associated with the impact of the drought event in the neighboring Taperaçu Estuary (2012/2013; Leite et al. 2016), characterized by reduced rainfall and a prevailing polyhaline–euhaline regime. Besides these factors, food availability could also contribute to the observed increase in A. lilljeborgii density in the June 2013 samples (during the drought event) in the current study. This increase might be attributed to the higher phytoplanktonic biomass (chlorophyll-a), which was 28.3% greater compared to that recorded in the same analyzed sector in 2014. The peak of A. lilljeborgii observed in June 2013 can be attributed not only to the aforementioned factors but also to the high concentrations of organic matter originating from neighboring mangroves, primarily in the form of detritus. This detritus serves as an important energy source for these copepods, given that A. lilljeborgii can assimilate between 13% and 40% of the organic carbon derived from the mangrove ecosystem (Schwamborn et al. 2002).

The drought and nondrought periods also affected the populations of P. richardi and O. hebes, mainly the first species mentioned. In June 2013 (drought event), the samples of the middle-S2 sector were dominated by the species P. richardi, with density ~90 times higher than the maximum recorded during the nondrought period (considering the same month studied, June 2014). This period was also marked by the increase in egg production (over 80%) of these species, as observed by Andrade in this same environment (unpublished data). In tropical environments, P. richardi and O. hebes actively feed on phytoplankton and constitute a major component of the zooplankton during the periods of high food abundance (Magalhães et al. 2006). In addition, they are considered excellent indicators of oligo-mesohaline waters (Sterza and Loureiro-Fernandes 2006; Correia 2018). The high-density peaks reached by P. richardi were associated with: (i) the opportunistic eating habits in situations of high concentration of food, (ii) the ability to consume cells of a wide dimensional spectrum (Kaminski et al. 2009), and (iii) the high production of eggs in periods of moderate salinity.

Dominant during most of the year in many tropical and subtropical coastal regions of South America (Vega-Pérez and Hernandez 1997; Rosa et al. 2016), P. quasimodo is a marine-euryhaline species that occurs primarily in more saline waters of tide-dominated environments (Lopes et al. 1998). A similar pattern was recorded in the present study, with the highest densities coinciding with the period of greatest salinity. Sterza and Loureiro-Fernandes (2006), in a study of the zooplankton community of the estuarine complex of Vitória Bay in Espírito Santo, Brazil, found that the abundance of P. quasimodo varied positively with salinity, indicating that the presence of this species in this ecosystem depends on recruitment from adjacent coastal zones, as observed here in the Caeté Estuary. This would explain the low density of P. quasimodo during the nondrought (2014; present study), as well as in La Niña periods in Amazonian estuaries (see Andrade et al. 2016).

The reduction in the density of these species during the rainy season was determined by their ecological affinity with environments characterized by moderate–high salinity levels and suggests, at least for A. tonsa and A. lilljeborgii, diapause egg production, which is used as a survival strategy in response to seasonal and longitudinal fluctuations in ecological conditions, as observed previously in temperate coastal and marine ecosystems (Hoffmeyer et al. 2008 and references therein), where the eggs were maintained in a resting state by environmental cues such as salinity and temperature. Magalhães et al. (2015) have suggested that the benthic diapause phase could be one of the principal bases of the seasonal replacement of Acartia species at the neighboring Taperaçu Estuary, in addition to the resumption of growth in the remaining adult population under favorable conditions. Adult Acartia congeners have distinct seasonal and spatial distribution patterns, but nauplii of all species survive well at higher salinities (Chinnery and Williams 2004).

As in the present study, the influence of climatic phenomena on the copepod community dynamics has been monitored in different regions of the world. Marques et al. (2007), for example, observed that the severe drought recorded in 2004 and 2005 was responsible for spatial shifts in the Mondego Estuary (Portugal) regarding the zooplankton community and interannual variability, with a replacement of the freshwater community by one predominantly dominated by estuarine organisms. The occurrence of such an estuarine community contributed to the increase in zooplankton density, which was ascribed to the estuarine species A. tonsa during the period of low freshwater flow. Higher chlorophyll-a levels were observed during the less productive (reduced upwelling) El Niño 2010 period in the Eastern tropical Pacific off central Mexico and were likely due to reduced grazing activities and increased ammonium availability through increased zooplankton metabolism, a period in which copepodites and adults of the carnivorous Euchaetidae (marine copepod) dominated (Kozak et al. 2018). The composition of the copepod species also changed substantially between El Niño 1997–1998 and La Niña 1998–1999 in Baja California (Mexico), following trends in oceanographic conditions. The warm period (El Niño event) was characterized by a community rich in equatorial, tropical/subtropical, and warm-temperate cosmopolites but almost lacking in subarctic copepods. The strong influence of equatorial water in this period occasioned the record abundance of Subeucalanus subtenuis (Giesbrecht, 1888), an equatorial species in oceanic eutrophic waters (Jiménez-Pérez and Lavaniegos 2004). A database synthesized from 19 oceanographic expeditions (1963–89) indicated that temporal patterns in copepod species diversity in the Tropical Atlantic were strongly influenced by climatic events. To assess species diversity, the Shannon diversity index was used. The abrupt increase in community diversity was mainly caused by climatic effects, such as the El Niño event. These conditions cause a greater breakdown of the thermohaline structure, increasing nutrients and chlorophyll-a, although the phytoplankton community was characteristic of tropical oligotrophic waters (Piontkovski and Landry 2003).

The only work published in the Amazon region which is related to the influence of the 2011 La Niña event (anomalous increase in rainfall) on the hydrological variables and copepod community structure in the tropical shallow Taperaçu Estuarywas developed by Andrade et al. (2016). These conditions reflected an extremely high recruitment rate of copepod nauplii, with a peak in density overlapping that of the adult forms.

The differences observed in the population density of copepods collected in the Caeté Estuary, when compared to those obtained in other tropical coastal ecosystems (see Table S3) were related to three main factors: (i) the mesh size used to capture zooplankton; (ii) salinity range; (iii) food quality and availability. In particular, the sampling method and the time taken to collect plankton samples may also have greatly influenced these results, since a 200 μm mesh size, such as those used in the present study, is commonly more effective in capturing adult individuals from A. tonsa, A. lilljeborgii, and copepods of the genus Pseudodiaptomus, allowing the escape of larvae (nauplii), as well as the early stages of juveniles (CI-CIII) of these organisms. However, it must be considered that the results obtained were similar to those observed in other studies performed in the region and that the main species of copepods were well represented by adult specimens, although the values obtained for the nauplii and juveniles of some of these species, as previously mentioned, were in fact underestimated.

4.2 Influence of Climate Variability on Diversity

As a consequence of the variability of copepod populations, the indices of species diversity (H′), equitability (J′) and richness also differed. Seasonally, the dry season samples presented higher diversity. The low river flow of the Caeté in this period, as well as during the 2013 drought event, allowed a major penetration of sea water and an increase in the occurrence of marine and coastal species, mainly in the outer-S3 sector of the estuary, resulting in higher diversity of copepods. In almost all estuaries, the greatest species diversity occurs near the mouth (outer reach) since that diversity is enhanced through exchanges between the opportunistic estuarine populations and coastal populations, characterized by the presence of large consumers (e.g., copepods; Lam-Hoai et al. 2006; Primo et al. 2009; Leite et al. 2016; Barros et al. 2019; Andrade et al. 2022). In this study, in nondrought periods, the diversity was lower. Miranda et al. (2005) reported that the decrease in diversity may be related to either an increase in environmental constraints, which act as niche filters, or to a local heterogeneity loss in terms of habitat and resource, allowing the survival of only some closely related species with particular common biological attributes. Conversely, according to the same authors, high habitat heterogeneity and environmental conditions promote higher diversity levels because this enhances the coexistence of either taxonomically diverse species with contrasting ecological requirements or ecologically similar, congeneric species being either adapted to slightly different niches or able to avoid direct competition.

In general, it was possible to observe that salinity presented an important role in the fluctuations in density and diversity of copepods collected in the Caeté Estuary, since the variations normally observed for these variables impose physiological limitations on biota, which are only tolerated by species that have developed efficient osmoregulatory mechanisms to tolerate such conditions (Magalhães et al. 2015; Calliari et al. 2019). The estuarine circulation patterns certainly exerted an additional selective pressure on the populations of copepods studied. These populations, as well as others that make up the zooplankton community, respond to patterns of spatial, monthly, and seasonal variation imposed by estuarine ecosystems, conditioning the dynamics of resident and transitory species of marine and limnetic origin (Kennish 1990; Atique et al. 2017; Isinibilir and Dogan 2020). In the Caeté Estuary, although no different patterns were observed in the composition or distribution of the species identified in short time scales (between tides or between day and night), spatial, monthly, and seasonal differences were evident, demonstrating the effect of anomalous precipitation oscillations and fluvial discharge, and consequently salinity, on variations in the dynamics of zooplankton in the estuaries of the Amazon region, and especially in the studied estuary. In addition to these factors, food availability, predation, natural mortality, and vertical migration may also have influenced the population dynamics of copepods, although these factors appear to have a secondary role in the succession processes of these organisms in the studied area.

Future studies on estuary conservation and management should focus on biodiversity preservation while maintaining habitat heterogeneity, as diversity decreases with increasing environmental constraints. Salinity significantly influences copepod density and diversity, requiring regulation, where possible, of freshwater inputs and estuarine circulation patterns to ensure ecosystem stability. Long-term monitoring of spatial, monthly, and seasonal variations is essential for understanding zooplankton dynamics, with copepods serving as bioindicators of ecosystem health. However, data obtained alone unfortunately do not allow us to assess the implications of these changes for the conservation and management of estuarine ecosystems, since the control of these environments—especially open ecosystems—does not allow human interference, as could occur in closed or semi-open estuaries. However, we can suggest significant implications for the abundance and distribution of local fish and invertebrates of commercial importance, improving the knowledge and management of similar ecosystems in tropical estuaries worldwide. Future studies focusing on conservation efforts should consider connectivity between resident and transient species, while also addressing trophic interactions such as predation and food availability. Adaptive management strategies that integrate these factors can help sustain estuarine ecosystems against climate anomalies and anthropogenic pressures.