Introduction

Stream ecosystems are dependent on the characteristics and changes of the watershed environment. Studies have been conducted to determine the effects of changes in the watershed environment, especially the increase in human activity, on the river ecosystem (Clément et al., 2017; Li et al., 2018; Yirigui et al., 2019).

Several evaluation methods, namely, the Bellan’s pollution Index (Bellan, 1967), the Saprobic Index (Zelinka & Marvan, 1961), the Shannon-Wiener Index (Shannon & Wiener, 1963), and the Index of Biotic Integrity (Clark et al ., 2003) based on characteristics, such as, water quality measurement (dissolved oxygen (DO), chemical oxygen demand (COD), biological oxygen demand (BOD), pH, and electric conductivity (EC)) and biological community species composition, biomass, and production, are used to investigate the status of stream ecosystems. However, biological methodologies (e.g., diversity index and species number), as well as chemical and physical approaches, are insufficient for assessing the health of stream ecosystems (An et al., 1992; Karr, 1981).

Since the ecosystem metabolism is an integrated measure of the gross primary production and ecosystem respiration, the metabolic balance provides basic information for understanding the overall ecosystem status (Young & Matthaei, 2008). In a stream ecosystem, gross primary production uptakes CO₂ and releases O₂ , whereas ecosystem respiration is a reverse process in principle. CO₂ in water exists in various forms, such as, HCO₃^-, CO₃^2-, CO₂ and their compounds, depending on the pH condition, and they are collectively known as dissolved inorganic carbon (DIC). Therefore, in stream ecosystems, DIC is directly related to biological photosynthesis and organic matter degradation. However, the DIC concentration in stream water can also be changed by chemical weathering and degassing.

Information regarding which of these is the major mechanism is estimated by determining the correlation between the DIC concentration change and other accompanying variables. However, more direct information can be obtained using the stable carbon isotope ratio of DIC (δ¹³C-DIC) (Nagata & Miyajima, 2008). For example, if photosynthesis, which is a biological process, is active in the water body, the DIC concentration will decrease and δ¹³C-DIC will increase (Finlay, 2004; Herczeg, 1987; Hollander & KcKenzie, 1991; Quay et al., 1986; Stiller & Nissenbaum, 1999; Wang & Veizer, 2000). On the other hand, it is expected that the DIC concentration will increase and δ¹³C-DIC will decrease in a respiration dominated waterbody (Atekwana & Krishnamurthy, 1998). These are explained by the isotope fractionation in the photosynthesis and respiration processes.

Rubisco enzyme fixes carbon in photosynthesis, adding CO₂ to a five-carbon compound to form a six-carbon sugar. The lighter carbon (¹²C) reacts faster in this kinetic reaction with bond formation and the kinetic isotope effect is -29 per mill (Fry, 2006). Therefore, the kinetic isotope effect increases δ¹³C value for DIC. However, the isotope fractionation by respiration is very insignificant (Nagata & Miyajima, 2008). Hence, δ¹³C of organic substances used as a substrate for respiration is generally very low (-25~-30‰) and δ¹³C of the decomposition product, DIC, has a low value.

Increase in human activity in the watershed affects streams in two ways, namely, nutrient pollution and organic pollution. In stream ecosystem dominated by nutrient pollution, active photosynthesis decreases the DIC concentration and increases the δ¹³C-DIC. However, in stream ecosystem dominated by organic pollution, active respiration increases DIC concentration and decreases δ¹³ C-DIC.

Despite this theoretical establishment, variation patterns of the δ¹³C-DIC values in actual streams are diverse and complex. Previous studies have shown that in some cases, δ¹³C-DIC increased in the river flowing direction (Finlay, 2003; Telmer & Veizer, 1999) and decreased in some other cases (Aucuor et al., 1999), and these were mostly explained by the geological and hydrological characteristics of the watershed (Aucour et al., 1999; Das et al., 2005; Telmer & Veizer, 1999).

Furthermore, it has been shown that if artificial land use (agriculture, urban) at the watershed increased, the DIC concentration in the river would increase too (Barnes & Raymond, 2009). Such an increase might be due to the inflow of high concentrations of DIC in the watershed or might be due to the inflow of organic matter, which would accelerate organic decomposition in the river.

Therefore, in this study, we tested the suitability of DIC concentration and δ¹³C-DIC values as indicators for monitoring the health of stream ecosystems by analyzing the watershed environment, water quality variables, and the DIC concentration and δ¹³C-DIC values of the stream water of two streams, namely, the Yasukawa stream and the Adokawa stream, which have watershed environments with contrasting artificial impacts.

Materials and Methods

Watershed status

This study was conducted at the Adokawa stream and the Yasukawa stream, two river streams located in the same river system with contrasting watershed environments. Adokawa stream has a length of 52 km and a watershed area of 305 km², and Yasukawa stream has a length of 62 km and a watershed area of 380 km² . The size of the two streams is roughly similar but the population size differs (8,000 and 120,000 people for the Adokawa stream and the Yasukawa stream, respectively).

The research stations for each stream are shown in Fig. 1. Field measurements and sampling were conducted at 14 stations for the Adokawa stream on May 28 and 29, 2004, and 23 stations for the Yasukawa stream on September 27 and 28, 2004. We referred to Kobayashi et al . (2009) for the geographic and demographic features.

Catchment area

Yasukawa stream was divided as upstream (St. Y1–Y7), midstream (St. Y9–Y11), and downstream (St. Y12-Y23) with catchment area ranges of 1.2–70 km², 119–152 km², 283– 380 km², respectively. Adokawa stream was divided as upstream (St. A1–A6), midstream (St. A7–A9), and downstream (St. A10–A14) with catchment ranges of 0.3–66 km², 156–183 km², and 281–305 km², respectively (Table S1, Table S2).

Yasukawa stream branches Y8’ (47.5 km²), Y15’ (11.2 km²), and Y18’ (23.4 km²) had catchment areas similar to the upstream catchment area size, and Y13’ (117 km²) had

Fig. 1.

Map of study sites in the Yasukawa stream and the Adokawa stream draining the Lake Biwa basin. Sampling stations are indicated by solid circles (•). Branches are indicated by a prime added to the site code. Background satellite image reproduced from ESRI.

a catchment area similar to that of the midstream.

Adokawa stream branches had a catchment area similar to that of the upstream.

Population density

The population density increased noticeably (0–321 people/km²) at Yasukawa stream along with the increase in the flow distance, and three of the four branches, excluding Y8’ (67 people/km²), showed high population densities, similar to that of the downstream watershed (198–540 people/ km²) (Table S1, Table S2). On the other hand Adokawa stream, including the branches, showed a decreasing trend in the downstream direction (4–43 people/km²) (Fig. 2).

Land use

The forest percentage along the Yasukawa stream watershed significantly decreased, following this order, upstream (90.9 ± 3.6%), midstream (78.6 ± 5.0%), and downstream (61.0 ± 1.8%) (Table 1, Fig. 3), but, the urban and rice field percentages increased (Table 1). Branches Y8’ (80.8%) and Y15’ (78.3 %) showed forest percentages similar to the upstream and midstream watershed, respec tively. Y13’ (54.0%) and Y18’ (41.4%) showed the highest urban (Y13’ = 6.6%, Y18’ = 10.1%) and rice field (Y13’ = 27.2%, Y18’ = 27.5%) percentages among all stations (Table S1).

Fig. 2.

Longitudinal changes of watershed features in the Adokawa and Yasukawa streams.

Fig. 3.

Box plots of longitudinal changes of watershed features in the Adokawa and Yasukawa streams.

The forest percentage at the Adokawa stream watershed was highest (96.6 ± 2.6%), and the percentages rice field (0.02 ± 0.01%) and urban (0.36 ± 0.22%) activities were similar or less than those upstream of the Yasukawa stream (Table 1, Fig. 3).

All chemical variables excluding the DIC were measured by Kobayashi et al. (2009). DIC concentrations were measured with a TOC-5000A total organic carbon analyzer (Shimadzu) (Kim et al., 2006; Maki et al., 2010).

The δ¹³C values of DIC were determined using the headspace equilibration method, using an online system consisting of a gas chromatograph (GC-6890, Thermo Fis-

Fig. 4.

Longitudinal changes of environmental variables in the Adokawa and Yasukawa streams.

her), combustion furnace (Combustion III, Thermo Fisher), and isotope ratio mass spectrometer (252, Finnigan MAT) (Maki et al., 2010; Miyajima et al., 1995).

Isotope ratios were obtained relative to a high-purity CO₂ reference gas and were determined in standard δ notation as the difference in parts per thousand (‰) rela- tive to international standards (Pee-Dee Belemnite):

δX = [(Rsample/Rstandard)-1] × 10³,

where, δ X is δ¹³C, and Rsample and Rstandard are ¹³C/¹²C ratios of the sample and the standard, respectively

Fig. 5.

Box plots of longitudinal changes of environmental variables in the Adokawa and Yasukawa streams.

(Maki et al., 2010; Miyajima et al., 1995).

Results

Chemical variables of stream water

pH

From the Yasukawa stream, it was observed that pH increased following this order: upstream (7.9 ± 0.2) < midstream (8.0 ± 0.2) < downstream (8.2 ± 0.9), but there was no significant difference (Fig. 4, Fig. 5, Table 1). In Adokawa stream, the difference in pH (7.4 ± 0.4) between stations was smaller than in Yasukawa stream (Fig. 4, Fig. 5, Table 1).

EC

EC is widely used as an indicator of the effects of human activities. EC in the Yasukawa stream significantly increased in the direction of upstream (69 ± 17 μS/cm), midstream (129 ± 14 μS/cm), and downstream (193 ± 37 μS/cm) (Fig. 4, Fig. 5, Table 1), whereas, in the Adokawa stream, low EC level (52 ± 13 μS/cm) was maintained, similar to that of Yasukawa stream upstream (Fig. 4, Fig. 5, Table 1).

DIC

The DIC concentration in Yasukawa stream significantly increased (Fig. 5) in the following direction: upstream (272 ± 89 μM) < midstream (553 ± 109 μM) < downstream (690 ± 37 μM). However, Y18’ showed the highest value (885 μM).

In all sections of the Adokawa stream, a low DIC concentration (249 ± 76 μM) was recorded, similar to that of Yasukawa stream upstream (272 ± 89 μM).

δ¹³C-DIC

In the Yasukawa stream, δ¹³C-DIC significantly decreased in the following direction: upstream (-8.68 ± 2.3 ‰) > midstream (-10.95 ± 1.21 ‰) > downstream (-12.13 ± 0.43 ‰).

In all sections of Adokawa stream, low δ¹³C-DIC level (-9.50 ± 2.54 ‰) was recorded, similar to that of Yasukawa stream upstream (-8.68 ± 2.3 ‰).

Correlation analysis

Catchment area vs. other variables

The catchment areas at each station of Yasukawa stream and Adokawa stream were positively correlated with the percentages of rice field and urban activities, and negatively correlated with the percentage of forest. When the mainstream data of Yasukawa stream only were analyzed, the population density, EC, and DIC were positively correlated with catchment area, and negatively correlated with δ¹³CDIC. Theses correlations increased further when Y1 was excluded (Table 2, Table S3).

In the case of the Adokawa stream, catchment area of the mainstream data was positively correlated with the percentage of rice field and negatively correlated with the percentage of forest at a significant level (Table 2, Table S4).

Population density vs. other variables

The population densities at each station of the Yasukawa stream were positively correlated with the percentage of rice field and urban activities, and negatively correlated with the percentage of forest. When the mainstream data of the Yasukawa stream only were analyzed, the catchment area, EC, and DIC were significantly correlated to the population density, and the correlations further increased when Y1 was excluded. In this case, in addition to the percentage of forest, δ¹³C-DIC also became a variable with a negative correlation (Table 2, Table S3).

In the case of the Adokawa stream, population densities were not significantly correlated to the other variables (Table 2, Table S4).

EC vs. other variables

The EC at each station of the Yasukawa stream were positively correlated with the population density, DIC, percentages of rice field and urban activities, and negatively correlated with the percentage of forest. When the mainstream data of the Yasukawa stream only were analyzed, the catchment area was significantly correlated to the EC, and the correlations increased further when Y1 was excluded. In this case, in addition to the forest percentage, δ¹³C-DIC also became a variable with a negative correlation (Table 2, Table S3).

In the case of the Adokawa stream, only pH was significantly positively correlated with EC (Table 2, Table S4).

DIC concentration vs. other variables

The DIC at each station of the Yasukawa stream was positively correlated with the population density, EC, the percentage of rice field and urban activities, and negatively correlated with the percentage of forest. When the mainstream data of the Yasukawa stream only were analyzed, the catchment area have a significant correlation with DIC, and the correlations increased further when Y1 was excluded. In this case, in addition to forest percentage, δ¹³C-DIC also became a variable with a negative correlation (Table 2, Table S3).

In the case of Adokawa stream, pH and EC were significantly positively correlated with EC (Table 2, Table S4).

δ¹³C-DIC vs. other variables

The δ¹³C-DIC at each station of the Yasukawa stream was positively correlated with the forest percentage, and negatively correlated with the population density, DIC, the percentage of rice field and urban activities. When the mainstream data of the Yasukawa stream only were analyzed, the catchment area also became a variable with a negative correlation (Table 2, Table S3).

In the case of the Adokawa stream, δ¹³C-DIC was not significantly correlated to other variables (Table 2, Table S4).

Discussion

The DIC concentration and isotope ratio in the flow-direction of a stream can be altered by different mechanisms, such as, air-water CO₂ exchange, kinetic isotope effect of CO₂ dissolution, isotope conversion and equilibrium with CO₂ , CO₂ consumption by photosynthesis, and CO₂ generation by respiration (Nagata & Miyajima, 2008).

In the present study, the DIC concentration between Y1 and Y2 increased, and δ¹³C-DIC sharply increased because of the CO₂ evasion mechanism of CO₂ over-saturation due to high CO₂ in the soil. The δ¹³C of dissolved CO₂ was ≥ 8 per mill lower compared to the co -existing HCO₃^-. Therefore, the δ¹³C of the residual DIC increased due to the degassing of dissolved CO₂ (Nagata & Miyajima, 2008). This phenomenon was also confirmed in the change between A1 and A2, the uppermost stations of the Adokawa stream.

On the other hand, the change in the DIC concentration was insignificant between Y3 (271.4 μM) and Y4 (263.7 μM), and between Y5 (324.2 μM) and Y6 (334.7 μM), but, δ¹³C-DIC increased. This is likely because of the isotope fractionation caused by the active photosynthesis between these stations. Kobayashi et al. (2009) reported that the concentration of chlorophyll a in epilithon at these stations increased more rapidly at Y4 (0.792 μg cm^-2) and Y6 (2.366 μg cm^-2) compared to Y3 (0.263 μg cm^-2) and Y5 (2.573 μg cm^-2), respectively, implying a relatively more active photosynthesis.

However, while the overall DIC concentration increased in the river flowing direction in the Yasukawa stream, δ¹³C-DIC showed an inversely proportional decreasing trend (Fig. 2, Fig. 6). Furthermore, the DIC concentration was significantly positively correlated with the BOD and DOC concentrations (Fig. 7) due to active respiration of living organisms in the stream water. The δ¹³C of CO₂ generated by respiration was almost identical to the δ¹³C-DIC of carbon that constituted organic matter, which became food, or decaying organic matter, and generally had a value between -25 and -30 per mill (Clark & Fritz, 1997; Nagata & Miyajima, 2008).

Therefore, as shown by the high level of correlations with the population density, land use, EC, BOD, and DOC in the watershed, the increase in human activity in the watershed led to an increase in the supply of organic matter to the stream. δ¹³C-DIC demonstrated a possibility that the DIC concentration could have increased because of actively decomposing them.

However, as shown by Kobayashi et al. (2009) in a study

Fig. 6.

Relationship between the DIC concentration and carbon stable isotope ratio of DIC (δ¹³C-DIC).

Fig. 7.

Relationships between the organic matter (DOC, BOD) and DIC concentration.

which investigated nutrient pollution (nitrogen and phosphorus inventories) and organic pollution (DOC, BOD) by comparing the upstream and downstream of the Yasukawa stream, there was no clear evidence which proved that organic matter used in respiration inside the stream was derived from other sources due to the organic pollution or that the organic matter produced was related to nutrient pollution. Nevertheless, since the DIC consumed during the internal production process by the nutrient pollution and the DIC produced by the decomposition of autochthonous organic matter theoretically offset each other, the net increase in the DIC would be due to the decomposition of allochthonous organic matter in addition to autochthonous organic matter.

However, in the Adokawa stream, there was almost no change in the DIC concentration and δ¹³C-DIC in the river flowing direction. Furthermore, the DIC concentration showed no significant correlation with the BOD and DOC concentrations (Fig. 7). The Adokawa stream showed very little changes in population density, land use, and EC in the flowing direction of the stream, which might explain the negligible effects of human activity in the watershed. Consequently, there was almost no change in the DIC concentration and δ¹³C-DIC (Fig. 6).

It is interesting to note that DOC and BOD concentrations showed a significant positive correlation in the Yasukawa stream (r² =0.60, p<0.0001, n=23), whereas in the Adokawa stream, this was not the case (r² =0.04, p<0.50, n=14) (Fig. 8).

Fig. 8.

Relationship between the DOC and BOD concentration.

In addition, in the Yasukawa stream, both BOD (r²=0.31, p<0.05, n=23) and DOC (r²=0.71, p<0.0001, n=23) showed a significant positive correlation with DIC, whereas in the Adokawa stream, this was not the case (BOD: r²=0.13, p<1.752, n=14; DOC: r²=0.04, p<0.54, n=14) (Fig. 7).

This is because labile organic matter contributed to the increase of organic matter in the Yasukawa stream, which was significantly affected by human activities, but in the Adokawa stream, where the percentage of forest was significant, refractory organic matter contributed to the increase in DOC.

In stream ecosystems, the impacts of the watershed might change quantitatively and qualitatively in the flowing direction of the stream. Nevertheless, the metabolism is in equilibrium in the ecosystems that stably accommodate environmental changes of watersheds. In other words, gross primary production and ecosystem respiration are in equilibrium at the ecosystem level, and as shown in the case of the Adokawa isstream, the variation range of DIC concentration and δ¹³C -DIC remained stable. On the other hand, as shown by the large changes in the DIC concentration and δ¹³C-DIC in the river flowing direction of the Yasukawa stream, the environmental changes of watersheds exceeded the accommodating capacity of the river ecosystem.

Therefore, this study suggests the possibility of using DIC concentration and δ¹³C-DIC as indicators for monitoring whether ecosystems with different watershed characteristics can accommodate the effects of artificial changes.

Conflict of Interest

The authors declare that they have no competing interests.

Acknowledgments

This research was supported by a grant from the 5th National Ecosystem Survey of National Institute of Ecology (NIE), funded by the Ministry of Environment (MOE) of the Republic of Korea (NIE-A-2021-01).