1 Introduction

Riverine supply of many elements, including carbon, nitrogen and phosphorus of largely terrestrial origin, is important to the steady-state chemistry of the oceans (Wachholz et al., 2023; Bauer et al., 2013). Metabolism in temperate coastal systems is associated with the magnitude of nutrient loads (carbon-C, nitrogen-N, and phosphorus-P) derived from the Earth to the Ocean (Wachholz et al., 2023; Tanioka et al., 2022; Liu et al., 2020; Causse et al., 2015; Eyre & Ferguson, 2002); however, in tropical systems, there are still uncertainties about the impact that these loads generate in the coastal region. Some authors indicate that this is mainly due to the interaction of multiple physical and biogeochemical factors (Causse et al., 2015; Jani & Toor, 2018; Eyre & Ferguson, 2002). Riverine nutrient loading, C, N, and P, is increasing in many catchments worldwide due to changes in land management, farming practices, and increasing urbanization (Gruber and Galloway, 2008; Blaen et al., 2017).

These changes are mainly intensified by the increase in land use cover and the increase in the population density rate that provides high rates of remaining organic load to rivers mainly in cities with low percentage of sewage treatment. These changes may also alter the rates of C, N, and P available for aquatic metabolism in tropical rivers.

According to Peacock et al. (2022), ecosystems and processes are dynamically linked through the transfer of water, solutes, and particles from the headwaters to the sea, and by interactions between different elements in biogeochemical cycles, particularly C, N, and P. Consequently, changes in the ecological status and management of terrestrial systems have the potential to alter the condition and function of downstream coastal and freshwater ecosystems, and the goods and services they provide, in ways that cannot be easily predicted from the study of single systems or individual processes. Several studies in the last decade have suggested that element ratios (stoichiometry) may provide a more effective indicator of ecosystem status than individual nutrient concentrations (Jarvie et al., 2018; Peacock et al., 2022; Stutter et al., 2018; Wachholz et al., 2023). According to Islam et al. (2019), changes in the stoichiometry of organic matter are associated with nutrient cycling and ecosystem biogeochemical processes. Thus, the transformations of N and P in organic forms reflect the assimilation and dissimilation of these important nutrients. Organic matter is initially synthesized in aquatic ecosystems mostly through algal photosynthesis and microbial activity, and the C:N:P ratio is well known to be close to the Redfield ratio (106:16:1). However, during the decomposition of organic matter, the behavior and the rates of mineralization of C, N, and P can vary. It is known that the mineralization rate of organic P is much faster than the mineralization rate of organic C and N, resulting in the deviation of C:N:P ratio from the Redfield ratio (Islam et al., 2019; Jeanneau et al., 2018). Another factor that affects C:N:P ratio of organic matter in rivers is the input of allochthones organic matter via wastewater, livestock, agriculture, etc.

The study region includes large coastal cities with high population density rates and regions of high agricultural activity in inland municipalities. In this paper, we explore the hypothesis that, globally, the stoichiometric proportions of C, N, and P are altered as agriculture and urbanization intensify. In doing so, we aim to make connections between increases in these elements (fluxes and concentrations) in sources of input and changes in nature. We used public data from sixty tropical rivers (drainage area: 20 km2 to 6.6 × 105 km2) to explore variations in C, N, and P and the potential impact of anthropogenic activities over a decade on freshwater bodies.

2 Methods

2.1 Study Area

The study region has an estimated area of 564,760 km2 which represents 6.7% of the area of the Brazilian territory. This region corresponds to the state of Bahia with a population of 14,930,634 million people, a demographic density of 24.8 inhab km−2, and is divided into 417 municipalities (IBGE, 2010). The hydrographic network of the region covers the geographic coordinates -46.6 to -37.3ºW longitude and -18.3 to -8.5ºS latitude. This region includes two of the seven major Brazilian hydrographic regions (Eastern Atlantic and São Francisco) with a drainage area of 1,014,667 km2. This region also includes major river basins such as São Francisco, Contas, Paraguaçu, Itapecuru, Grande, and Jequitinhonha (Fig. 1 and Table 1), with a strong amplitude between drainage areas (between 20 km2 and 600,000 km2). Through this hydrographic network, 453 fluviometric stations measure flows and limnimetric quotas, administered by governmental entities such as ANA (National Water Agency), CODEVASF (São Francisco Valley Development Company), CRA (Regional Administration Council of the State of Bahia), and INGÁ (Water Management Institute). Mean river discharges in the region vary strongly from small streams (< 1.0 m3 s−1) to large rivers (> 1,000 m3 s−1).

Fig. 1
figure 1

Study region (blue polygon) and hydrographic network (gray polygon) including monitoring stations (red triangles). The colors in the study region correspond to the Koppen climate classification

Full size image
Table 1 General characteristics of watersheds. Climate types are represented in Fig. 1. All data refer to the period 2008–2018
Full size table

Another network managed by INMET (National Institute of Meteorology) collects meteorological data through 29 stations in this region. This network shows an amplitude of rainfall in the region that oscillates between 400 and 1300 mm per year (INMET, 2018). Additionally, water quality data were collected at 143 stations by INEMA (Institute of Environment and Water Resources) throughout the Bahia-Brazil state region (INEMA, 2018). These parameters are indicated in Table 2.

Table 2 Analytical methodologies of the variables used in the monitoring for the period 2008–2018
Full size table

In the northeast region of Brazil, there are five climatic types according to the Köppen climate classification (Noriega & Araujo, 2014). The climatic types present in the study region are Af, Am, As, Aw, and BSh. These range from humid equatorial climates (tropical rainforests) to semi-arid types (Fig. 1). According to EMBRAPA (Empresa Brasileira de Pesquisa Agropecuária) estimates from Santos et al (2018) publications, the soils that prevail in the region are latosoils, argisoils, and neosoils, representing ~ 70% of the total (Fig. 1S).

2.2 Data Sources

Data on physical and chemical parameters were obtained bimonthly by INEMA (www.inema.ba.gov.br) for the period 2008–2018 and analyzed for Total Organic Carbon (TOC), Total Dissolved Nitrogen (DIN), Total Nitrogen (TN), Total Dissolved Phosphorus (DIP), Total P (TP) and Total Alkalinity (TA). Dissolved Inorganic C (DIC) was calculated from pH, alkalinity, and water temperature measurements using the mStatGraph v1.7 software (Varona et al., 2023). Total carbon (TC) was calculated as the sum of TOC and DIC. The organic particulate forms of C (POC), N (PON), and P (POP), were obtained through the differences between the dissolved and total forms. Additionally, rainfall data were obtained from INMET (www.inmet.gov.br) for the same study period. River discharge records were obtained from ANA (www.ana.gov.br). Demographic data (population density) were obtained from the IBGE (www.ibge.gov.br) and associated with the watersheds and corresponding municipalities. The spatial distribution of mineral resources (limestone, silica, and sedimentary) was obtained from the updated GLORICH database. The analysis methodologies of the physical–chemical parameters are shown in Table 2.

Total Alkalinity [mg l−1], as CaCO3 by titration method was used to measure the total alkalinity of water, which is the water's ability to neutralize acids. The precision of this method is ± 0.1 mg l−1 for samples with total alkalinity of up to 100 mg l−1 and ± 0.2 mg l−1 for samples with total alkalinity greater than 100 mg l−1. Samples were sampling in polyethylene bottles and store at a low temperature.

Temperature [ºC]: the precision of this method is ± 0.1 ºC for samples with temperatures up to 40 ºC and ± 0.2 ºC for samples with temperatures greater than 40 ºC. pH: the precision of this method is ± 0.02 pH units for samples with pH up to 7.0 and ± 0.03 pH units for samples with pH greater than 7.0. Total Nitrogen [mg l−1]—Method 4500-N: this method was used to measure the total nitrogen in water, which is the sum of all forms of nitrogen in water (NO3− + NO2− + NH4+). The precision of this method is ± 0.2 mg l−1 for samples with total nitrogen up to 10 mg l−1 and ± 0.5 mg l−1 for samples with total nitrogen greater than 10 mg l−1. Total nitrogen was determined through the oxidative digestion of all digestible nitrogen forms to nitrate, followed by quantitation of the nitrate using an in-line persulfate/UV digestion (4500-N/B) of Standard Methods.

Dissolved Inorganic Phosphorus [mg l−1]: this method was used to measure the dissolved inorganic phosphorus in water, which is the form of phosphorus that is more readily available to phytoplankton’s. The precision of this method is ± 0.05 mg l−1 for samples with dissolved inorganic phosphorus up to 0.5 mg l−1 and ± 0.1 mg l−1 for samples with dissolved inorganic phosphorus greater than 0.5 mg l−1.

Total Phosphorus [mg l−1]: this method was used to measure all forms of phosphorus in water. The precision of this method is ± 0.2 mg l−1 for samples with total phosphorus up to 10 mg l−1 and ± 0.5 mg l−1 for samples with total phosphorus greater than 10 mg l−1.

Total Organic Carbon [mg l−1]: this method is used to measure the total organic carbon in water, which is the sum of all the organic carbon in water (DOC and POC). The precision of this method is ± 0.5 mg l−1 for samples with total organic carbon up to 100 mg l−1 and ± 1 mg l−1 for samples with total organic carbon greater than 100 mg l−1. The sample is acidified, purged to remove inorganic carbon, and oxidized with persulfate in an autoclave at temperatures from 116 to 130 ºC. The resultant carbon dioxide (CO2) is measured by nondispersive infrared spectrometry.

The sampling frequency of water samples for the analysis of physical and chemical parameters was bimonthly for each monitored station between 2008 and 2018. All samples were collected at the surface of each water body (< 1 m).

The weighted average element/compound concentration was calculated by the following equation:

$$Ci=\frac{\sum_{i=1}^{n}Cith*Qith}{\sum_{i=1}^{n}Qith}$$
(1)

where Ci is the weighted average element/compound concentration [mg l−1], Ci, is the concentration for the ith measurement [mg l−1] and Qi is the discharge of the river in the day of the ith measurement [m3 s−1]. The load of C, N and P for each sampling station was estimated from the expression: Load-CNP (tons per year) = Discharge (m3 per year) × Concentration (g/m3) / 1,000,000.

Descriptive Statistics

The mean, standard deviation, minimum, maximum, and percentages were computed. Nonparametric statistics: the Mann–Whitney test was used to verify significant differences between two sample sets. The statistics for the time series were computed using the Mann–Kendall trend test. In the Mann–Kendall test, a trend is considered negative or positive, indicating a decrease or increase in the attributes of the historical series analyzed, if the Mann–Kendall (Kendall’s tau) score is negative or positive, respectively. In addition, the trend indicated by this methodology is considered significant when the p-value has a value less than α = 0.05. All statistical analyses were performed using mStatGraph v1.7 software (Varona et al., 2023).

3 Results

3.1 Rainfall and Fluvial Discharge

The analysis of rainfall for the period 2008–2018 did not show significant differences (Mann–Whitney; p > 0.05; α = 0.05) with the historical period (30 years). Additionally, trend analysis (Mann–Kendall test) showed no statistically significant trend (Fig. 2S).

The rainfall regime of the coastal or coastal sector (E) and the rainfall regime of the interior region of the state of Bahia (W) were analyzed through a 30-year series (1960–2010) of data obtained from meteorological stations in the study region and we concluded that the wet period includes the months of April, May and June on the coast, and November, December, and January in the interior. The dry period includes the months of August, September, and October on the coast, and June, July, and August inland (Fig. 2S).

Fluvial discharges ranged from 0.5–1654 m3 s−1 with maximum values observed in the São Francisco River. Other important river discharges were observed in the Carinhanha (146 m3 s−1), Jequitinhonha (144 m3 s−1) and Rio Grande (104 m3 s−1) rivers. The average value for the observed period was 49 m3 s−1 for the 60 rivers studied (Table 1).

3.2 C-N-P Spatial Distribution of Observations

According to the observations of the period, TOC varied between 2.7–73.0 mg C l−1, with an accumulation of values in the range of 10–20 mg C l−1 (Fig. 2a). The mean value of all sampling stations was 13.0 ± 10 mg C l−1, while spatial averages showed maxima in the Peixe de Baixo, Salitre, Real, and Jequiezinho rivers. The lowest mean value was 2.7 mg C l−1, observed in the Itaguari River (Table 1). The TA values observed over the period showed a wide range of values (1.80—372.0 mg TA l−1); however, the highest frequency of data was concentrated in the 0–20 mg TA l−1 range (Fig. 3S). The mean value obtained for TA was 53.0 ± 66 mg TA l−1. According to the spatial analysis, TA showed high mean values in the rivers Real (372.0 mg TA l−1) and Jequiezinho (329.0 mg TA l−1) respectively. Low values of TA were observed in the Paraguaçu (1.8 mg TA l−1) and Formoso (2.7 mg TA l−1) rivers, respectively (Table 1 and Fig. 2b). The observed inorganic/organic C ratio (TA/TOC) was 0.66 for all sampling stations.

Fig. 2
figure 2

Spatial distribution of the mean observed values per sampling station of TOC (a), TA (b), TN (c), and TP (d) in the studied rivers. The color bar indicates the concentration in mg l.−1

Full size image

TN observations showed concentrations between 0.9 and 32.0 mg N l−1 in the studied rivers. The highest frequency of data was concentrated in the range of 2–3 mg N l−1, equivalent to 75% of the recorded observations (Figs. 2c and 3Sc). The mean value obtained for the 60 rivers was 2.5 ± 4 mg N l−1; whereas the spatial means in the rivers showed a maximum mean of 32 mg N l−1 in Rio Real and a minimum mean of 0.9 mg N l−1 in Rio de Janeiro (Fig. 3Sc).

Fig. 3
figure 3

C-N-P fluxes for the studied rivers between 2008 and 2018. a Carbon (Tg C yr−1), b Nitrogen (Tg N yr−1), and (c) Phosphorus (Tg P yr.−1)

Full size image

TP observations showed a mean concentration of 0.2 ± 0.5 mg P l−1 (Fig. 2d); while the histogram showed a higher frequency of data in the range of 0.0—0.4 mg P l−1; however, values > 9 mg P l−1 were also recorded (Fig. 3Sd). The spatial averages showed a maximum average of 4.6 mg P l−1 in the Real River in the northern region of the state (Fig. 2d).

3.3 C-N-P Fluxes

Estimates of the C fluxes (Tg yr−1) of the studied rivers are shown in Fig. 3a. The annual total C load (TC) was estimated at 32.7 Tg C yr−1 (0.0327 Pg C yr−1), where DIC accounted for 69% of TC (22.7 Tg C yr−1) and DOC contributed 22% (6.9 Tg C yr−1), whereas particulate forms showed 8% for POC (2.50 Tg C yr−1) and 1% for PIC (0.6 Tg C yr−1). Estimated total N (TN) loads amounted to 1.6 Tg N yr−1 (Fig. 3b); distributed between DIN with 44% of TN (0.7 Tg N yr−1), DON with 25% of TN (0.4 Tg N yr−1), and PN with 31% (0.5 Tg N yr−1) respectively.

P loads amounted to 0.07 Tg P yr−1 (Fig. 3c); where DIP accounted for 29% of the TP (0.02 Tg P yr−1), DOP accounted for 14% of the total (0.01 Tg P yr−1), and PP accounted for 57% of the TP (0.04 Tg P yr−1) respectively.

3.4 C-N-P Trends

The observed yearly fluxes were estimated for TOC, TA, TN, and TP, indicating a statistically significant increment of TOC for the period 2008–2018 (Mann–Kendall test; p = 0.0006; α = 0.05). The slope of this series indicated an increase of 0.001 g C m−2 yr−1 for the TOC time series (Fig. 4b). TA did not show a significant trend (Mann–Kendall test; p > 0.05; α = 0.05); however, it recorded high loads (bubble size in Fig. 4a), especially in the years 2013 and 2017. TN and TP also showed no statistically significant trends (Mann–Kendall test; p > 0.05; α = 0.05). The annual TN rate indicated a mean value of 0.03 ± 0.007 g N m−2 yr−1; while TP showed a mean value of 0.02 ± 0.006 g P m−2 yr−1 (Fig. 4c and d). The mean annual TOC rate for the period indicated a value of 1.4 ± 0.15 g C m−2 yr−1; whereas TA showed a lower mean annual rate than TOC (0.6 ± 0.04 g C m−2 yr−1).

Fig. 4
figure 4

Yearly fluxes (g m−2 yr−1) of TA (a), TOC (b), TN (c), and TP (d), between 2008 and 2018 for the study region. The size of the bubbles refers to the annual loads (Tg yr−1) for TA, TOC, TN and TP

Full size image

3.5 C-N-P Fluxes and Climate Types

The C-N-P fluxes for the climate types were grouped into five clusters (Fig. 1) and are shown in Fig. 4S. The flux results for the climatic types showed significant differences (Kruskal–Wallis test; p < 0.5; α = 0.05) between the different climatic types of the study region; it seems evident that the Am and BSh types exports more inorganic (DIC + PIC) and organic (DOC + POC) carbon (Fig. 4Sa, Sb, and Sc). The inorganic form showed higher fluxes compared to the organic forms. Type Am had an average flux of 88.0 ± 80 g C m−2 yr−1; whereas climate type BSh had a flux of 42.0 ± 10 g C m−2 yr−1 (Fig. 4Sa). DIC fluxes accounted for > 90% of inorganic forms in the Am and BSh climate types. For organic forms, DOC fluxes represented the highest percentages in climate types Am (70%) and BSh (74%).

About N fluxes, Fig. 4S shows the Am and BSh climate types with higher rates. The Am type prevails over the BSh type with a 2.3 times higher rate (Fig. 4Sd). Inorganic nitrogen fluxes (average = 2.5 ± 11 g N m−2 yr−1) represent almost twice the organic nitrogen fluxes (average = 1.3 ± 5 g N m−2 yr−1). The organic/inorganic ratio was estimated at O/I = 0.6 for all climate types (Fig. 4Se and Sf). The main organic influence is the As type (O/I = 0.64) and the main inorganic influence is the Am type (I/O = 0.5).

P fluxes were higher in the Am, BSh, and Af climate types, respectively (Fig. 4Sg). The mean inorganic P flux (0.06 ± 0.2 g P m−2 yr−1) was slightly higher than the organic P flux (0.04 g P m−2 yr−1), respectively (Fig. 4Sh and Si). The ratio of organic/inorganic forms was O/I = 0.75; it was observed that in the Am and As types these ratios are slightly higher, indicating greater organic influence.

3.6 C-N-P Stoichiometry

Table 3 shows the percentages of the main constituents and their average stoichiometric ratios for the sixty rivers studied.

Table 3 Mean stoichiometric constituents samples collected from sixty rivers refer to the period 2008–2018
Full size table

Using the conceptual ternary diagram, we plot the stoichiometric C/N/P ratios for all studied rivers (Fig. 5a). The results showed P depletion relative to C and/or N in 39 of the sixty rivers (65%) indicated in the ternary diagram (20–80%), 17 rivers indicated P and/or N depletion relative to C, three rivers showed a balance between C, N, and P (central sector of the diagram), and one river (ITM) with C and P co-depleted. These results show an excess of C and N relative to P. This was further supported by the fact that most catchments had TOC values > 50% (Table 3; Fig. 5a).

Fig. 5
figure 5

Classification of different classes in ternary plot. According to Smith et al. (2017); (i) values in the central portion of the graph would be “balanced” relative to each other; (ii) values to the right of the yellow line would be P depleted (i.e., < 20% TDPR); (iii) values to the left of the black line would be C depleted (i.e., < 20% TOC); and (iv) values below the cyan line would be N depleted (i.e., < 20% TDNR). The corners would represent codepletion. Values with a “balanced” Redfield ratio (i.e., 106 C/16 N/1 P) would appear at the coordinates 33.3% TDC, 33.3% TDNR, and 33.3% TDPR; a Stoichiometric ratios of C-N-P at all rivers; b stoichiometric ratios at all stations in hydrographic basins categorized according to land-use; and (c) stoichiometric ratios at all stations categorized according to population density

Full size image

Here, we only include DIN and DIP as Nr, and Pr due to the lack of monitoring data for organic N and P fractions. However, our conclusions on the dominance of TOC, Nr not change, even in a scenario with high concentrations of TOC and reactive N (Table 3).

Stoichiometric ratios were also analyzed according to land use in the study region (Fig. 5b). The results showed P depletion in most monitoring stations (83%) of the sixty rivers (133 stations). Agricultural activity, savannah features and the urban area spatially dominate the study region associated with P depletion (Fig. 5b). The spatial patterns of the classes suggest that watershed landscape features exert a first-order control on average C/N/P reactive rates. Natural features also show P depletion; however, they can also be associated with N depletion. Anthropogenic influence through population density was included in the stoichiometric analysis (Fig. 5c), indicating P depletion in regions with low and high population density. Low population densities spread across the diagram; whereas high population densities remain in the region of depleted P and N (see Fig. 5c; 20–80% Nr and 0–20% Nr respectively).

The mean ratios shown in Table 3 indicated a mean value of N/P = 89:1 for all rivers in the study region, indicating the high loading of N relative to P.

4 Discussion

The analysis of the data series obtained through the proposed methodology (time series: 10 years; frequency: bi-monthly) represented an important source of information for obtaining trends in C-N-P elements and also for identifying the current state of water quality in the 60 river systems. We believe that the geographical coverage of the systems analyzed represents a good distribution in the tropical region from the coast to the interior of the continent. This coverage includes 10º of latitude and 5 different climatic regions. Some limitations were identified (more frequent sampling, time series > 10 years) that could have provided more information; however, we believe that these results give us a good overview of trends and the current state of C-N-P concentrations and fluxes.

The analysis of the climatological series showed a well-defined pattern between the coastal and inland regions of the continent. Additionally, the trend analysis (Mann–Kendall test) showed no significant differences (p > 0.05) in rainfall in the study region for the 30-year period. These climate patterns from 2 different regions were also observed by Molion and Bernardo (2002) and Silva et al. (2012). Thus, we can suggest that the controlling factors of C-N-P variations are associated with anthropogenic origin, such as: agriculture, domestic and industrial sewage and livestock, mainly. According to Drake et al. (2021), Amazon River DOC concentrations ranged from 2.86 to 5.23 mg C l−1, with a mean of 4.19 mg C l−1; whereas in our study, DOC ranged from 4 to 21 mg C l−1, with a mean value of 7.6 mg C l−1. For DIN, Drake et al. (2021), reported a range of 0.08—0.28 mg N l−1; whereas the DIN in our study corresponded to a much larger range (0.02 to 15.8 mg N l−1). The DIP reported for the Amazon River (Drake et al., 2021), ranged from 0.006 to 0.02 mg l−1; this range is lower than that observed in our study (0.002—1.1 mg P l−1). The C-N-P concentrations of our study region have higher concentrations than those reported by Drake et al (2021) in the Amazon River. These high concentrations are significant and indicate a warning sign in the flux estimates for this study region.

To analyze the inputs of C-N-P exported by the sixty rivers to the coastal region we included the fluxes of these loads. The export of fluvial C from land to the ocean is an important interface between two of the largest carbon reservoirs in the world. Recent syntheses have provided revised estimates for the riverine transport of C that river water annually provides 0.80—1.33 Pg of C to the world’s oceans (Huang et al., 2012), in which organic carbon (TOC) accounts for about 45% (Cai, 2011; Meybeck, 1982). In the tropical region (30ºN – 30ºS), the fluvial input of C delivers approximately 0.53 Pg carbon to the estuaries annually. Of this, 0.21 Pg C is dissolved inorganic carbon (DIC), 0.14 Pg C is dissolved organic carbon (DOC), 0.05 Pg C is particulate inorganic carbon (PIC), and 0.13 Pg C is particulate organic carbon (POC) (Huang et al., 2012).

Global annual export of total N (TN), P (TP) and organic C (TOC) from rivers is estimated to be 44.9 Tg N, 9.04 Tg P, and 317 Tg C respectively.

Estimates of the C fluxes (Tg yr−1) of the studied rivers are shown in Fig. 3a indicated the annual total C load (TC) of 32.7 Tg C yr−1 (0.0327 Pg C yr−1), where DIC accounted for 69% of TC (22.7 Tg C yr-1), and DOC contributed 22% (6.9 Tg C yr−1). According to our results, the DOC represented ~ 27% of DOC export by Amazon River (average = 25.0 Tg C yr−1; Moreira-Turcq et al., 2003; Huang et al., 2012; Drake et al., 2021). When compared to tropical rivers in the Americas (Huang et al., 2012), exported DOC represented approximately 12% of the total.

Estimated total N (TN) loads amounted to 1.6 Tg N yr−1 (Fig. 3b), where DIN represented 44% of TN (0.7 Tg N yr−1). This flux is like that reported by Drake et al. (2021) in the Amazon River (0,82 Tg DIN yr−1). According to Mayorga et al. (2010), modeled N fluxes for South American rivers represent ~ 8.0 Tg N yr−1. These authors also indicate DIN fluxes between 40—570 kg km−2 yr−1 for the study region, whereas our fluxes indicated a mean value of 196 kg km−2 yr−1.

P loads amounted to 0.07 Tg P yr-1 (Fig. 3c); where DIP accounted for 29% of the TP (0.02 Tg P yr−1). This DIP load is approximately one third of the load reported by Drake et al. (2021) for the Amazon River (0.063 Tg P yr−1). Modeled yearly fluxes of DIP were reported by Mayorga et al. (2010), indicating a range of 2–5 kg km−2 yr−1; whereas our study indicated a mean value of 7.9 kg km−2 yr−1.

A temporal analysis of these fluxes between 2008—2018 showed a significant increasing trend for TOC (Fig. 4b). According to Mayorga et al. (2010), yearly DOC fluxes range between 250—4000 kg C km−2 yr−1 for the study region, whereas the observed range of DOC varied between 12—3300 kg C km−2 yr−1. The slope of this series indicated an increase of 0.001 g C m−2 yr−1 for the TOC time series (Fig. 4b). This rate is indicative of the fact that the values observed in this study increase in accordance with the growth and expansion of urban and agricultural areas.

We analyzed the produced and remaining organic loads (that which reaches the receptor aquatic body) for the 2008–2018 series of all municipalities (N = 417). Trend analysis (Mann–Kendall test; α = 0.05) showed 85% of municipalities (N = 354) with a positive trend (p < 0.05) and 63 municipalities with a negative remaining organic load (15%). These results are also associated with the population density observed for the study period (Fig. 6a).

Fig. 6
figure 6

Trends of organic load in 417 municipalities of the study region (a); variations of anthropogenic (km2) and natural (km2) expansion (b); categorized distribution of land-use in the study region (c); and distribution of anthropic and natural areas (d). L1 = Artificial area; L2 = Agricultural area; L3 = Managed Pasture; L4 = Mosaic of occupations in forest area; L5 = Silviculture; L6 = Forest vegetation; L7 = Wetland; L8 = Grassland vegetation; L9 = Mosaic of occupations in grassland area; L10 = Continental water body; L11 = Coastal water body and L12 = Uncovered area

Full size image

The high rate of positive trends in the remaining organic load is indicative of urban and agricultural expansion in the region (Fig. 6b). The organic load entering the rivers is increasing progressively, mainly due to poor public sanitation policies. In the study region only 19 municipalities have a remaining organic load of < 40% (acceptable % of discharge into the water body); this represents < 5% of the municipalities in the region (N = 417).

Land use and cover were analyzed for the period 2008–2018 (Fig. 6b, c and d). The results indicated a trend of increasing anthropized area of 4% for this period, whereas the natural area decreased by 3.1% for the same period. As observed in Fig. 6a, the average C load represents approximately 15 times the load of natural origin (30.7 × 106 t yr−1). Similar flux increases can also be observed for N and P respectively. In this sense, we should consider the C-N-P stoichiometry that also indicated a P depletion; thus, the high organic load (positive TOC trend) reflects the urban and agricultural expansion in most municipalities and rivers of the region. The C-N-P loads reported here represent an important contribution to the regional and global balance of these macronutrients in tropical rivers.

4.1 Implications for Freshwater Management

The results of the study presented, which analyzed the relationships between the elements carbon (C), nitrogen (N) and phosphorus (P) in tropical river ecosystems, have significant implications for freshwater management and conservation efforts in these areas. In this analysis, we will explore the main implications of each element and propose actions for different sectors.

4.1.1 Implications of Carbon (C)

C Cycle

C dynamics in tropical rivers are complex and influenced by various factors, such as vegetation, microbial activity and sedimentation. Changes in the C cycle can affect water quality, primary productivity and the structure of aquatic communities.

Greenhouse Gas Emissions

Tropical rivers can be significant sources of greenhouse gases such as carbon dioxide (CO2) and methane (CH4). Inadequate water and soil management can increase greenhouse gas emissions, exacerbating climate change.

Carbon Sequestration

Tropical rivers can also act as carbon sinks, storing carbon in sediments and plant biomass. The protection and sustainable management of river ecosystems can contribute to mitigating climate change.

4.1.2 Implications of Nitrogen (N)

Eutrophication

N excess in rivers can lead to eutrophication, a process that causes excessive algae growth and reduced oxygenation of the water. Eutrophication can have negative impacts on water quality, aquatic fauna and flora and human use of water.

Nitrate Pollution

Nitrate, a form of N, can contaminate groundwater and represent a risk to human health. Appropriate fertilizer and wastewater management is crucial to prevent nitrate pollution.

N Cycling

N is an essential element for life and plays an important role in the primary productivity of aquatic ecosystems. Understanding the N cycle is fundamental to the sustainable management of water resources.

4.1.3 Implications of Phosphorus (P)

Nutrient Limitation

P is often a limiting nutrient in tropical rivers, meaning that its availability controls the growth of algae and other aquatic plants. Changes in P levels can affect the structure of aquatic communities and primary productivity.

P Pollution

P excess in rivers can lead to eutrophication, with the same negative consequences as those mentioned above for N. Proper management of fertilizers, detergents and sewage is essential to prevent P pollution.

Bioavailability of P

The form and bioavailability of phosphorus in rivers are important for the assimilation of this element by aquatic organisms. Understanding these aspects is crucial for the sustainable management of water resources.

4.2 Actions for freshwater management and conservation

Implementation of integrated water resources management policies: Policies should consider the biogeochemical cycles of C, N and P and promote the sustainable use of water.

Improving Water Quality Monitoring

Collecting and analyzing data on C, N and P levels in rivers is essential for understanding the dynamics of these elements and identifying potential problems.

Restoring Riparian Areas

Restoring degraded areas along rivers can help filter nutrients and reduce water pollution.

Developing Sustainable Agricultural Practices

Practices such as agroecology can help reduce the use of fertilizers and soil erosion, reducing the amount of nutrients entering rivers.

Promoting Environmental Education

Raising awareness about the importance of fresh water and the challenges of managing it is fundamental to changing behavior and adopting sustainable practices.

Scientific Research

More research is needed to better understand the biogeochemical cycles of C, N and P in tropical rivers and to develop effective solutions for freshwater management and biodiversity conservation.

5 Conclusions

The results obtained from the time series of hydrochemical and demographic variables in 60 tropical rivers showed that the potential impact of anthropogenic activities over a decade on freshwater bodies. The trend analysis showed no significant differences in rainfall in the study region for the 30-year period. Thus, we can suggest that the controlling factors of C-N-P variations are associated with anthropogenic origin, such as: agriculture, domestic and industrial sewage and livestock, mainly.

The annual total C load (TC) was estimated at 32.7 Tg C yr−1, where DIC accounted for 69% of TC and DOC contributed 22%. When compared to tropical rivers in the Americas, exported DOC represented approximately 12% of the total.

The inorganic forms of DIN and DIP represented the main components of TN and TP, respectively. DIN fluxes were like other tropical rivers (196 kg km−2 yr−1), while TP showed higher fluxes than those reported by other studies for tropical areas.

The stoichiometry showed P depletion relative to C and/or N in 39 of the sixty rivers (65%). This was further supported by the fact that most catchments had TOC values > 50% (C/N/P = 100%). A significant trend was found for TOC.

Land-use and cover at period 2008–2018 indicated a trend of increasing anthropized area of 4%, whereas the natural area decreased by 3.1%. The organic load trend analysis showed 85% of cities with a positive trend, this high rate in the remaining organic load is indicative of urban and agricultural in the region.

The results obtained for the C-N-P elements in this study contributed to characterizing the current state of water quality in 60 rivers, their fluxes, and their trends. This has generated important information that can be used in the future studies at other authors aimed at complementing time series and their associations with anthropogenic factors.