The effect of spring water on the growth of a submerged macrophyte Egeria densa

The study site is located in an upstream reach of the Kurohashi River (35°06′57.0 N, 136°07′11.2E–35°07′04.2 N, 136°06′51.1E), a spring-fed stream that extends for 5.5 km. Its catchment area is 6.9 km² and the average river bed slope is approximately 1/1,000. The study site stretch is 0.8 km long, with a stream width ranging from 1 to 2 m. The water depths are 0.05–0.20 m and 0.10–0.70 m at times of no and peak vegetation biomass, respectively. The flow velocity is 0.1–0.7 ms⁻¹ during times of no vegetation and between 0.1 and 0.4 ms⁻¹ during peak vegetative biomass. These ranges of water depth and flow velocity allow the submerged macrophytes to grow (Otaki and Ishido 1980). The river bed is composed of stones with gravel in the upstream reach, whereas the mid- to downstream reach is basically clay or partially concrete. As shown in Fig. 1, there is a spring-fed pond in an upstream area, which feeds stream water. The upstream reach also contains several places where spring water gushes from the river bed. The main stream in the study site has seven tributaries without vegetation, one of which supplies spring water to the midstream reach.

The Kurohashi River flows through paddy fields and is used for agricultural purposes as an irrigation channel. Though the stream banks are mostly adjacent to paddy fields, some parts are bordered by riparian forests of bamboos, broad-leaved trees, and cedars in the upstream reach (Fig. 1). The stream bed, as well as submerged macrophytes, is dredged annually to maintain the flow condition of the channel. The most recent dredging was carried out on 5 July 2009. After their removal, however, macrophytes regrow rapidly (Fig. 1).

Field observations were done along the main channel and its tributaries. Plant, sediment, and water samples were collected from the main channels, whereas physical and chemical features of the water were measured in observation points of the tributaries channels. In the main channel, sampling points were located approximately every 50 m in summer (June to July) of 2009 (St. 1–18, Fig. 1). Four additional sampling points (St. 4′, 5′, 12′, 14′, Fig. 1) were set in the autumn (September to October) of 2009. Water parameters were recorded from one observation point at each tributary (St. A–G, Fig. 1). At each observation, pH, water temperature, dissolved oxygen (DO), and electric conductivity (EC) was measured by the Onsite Aqua Sensor Integrated System (OYO Corporation) at different times: morning, midday, and afternoon. Averaged values of three were used for analyses.

Macrophytes, water, and bed-sediment samples were collected from each sampling point in the main channel (St. 2–17). Macrophytes and sediment cores were sampled from 30 × 30-cm quadrats in summer on 20 June 2009 and in autumn on 20 September 2009. Sediment-layer thickness was measured at all sampling points. In addition, it was measured in winter on 6 December 2009 and in spring on 9 May 2010. Sediment-layer thickness is defined as the depth between the bed-sediment surface and the underlying concrete river bed or clay or stone layers. When the river bed was composed of stones, because it was difficult to recognize the boundary between the bed sediment and the river bed, sediment-layer thickness was considered as >5 cm.

Macrophytes, water, and bed-sediment samples were stored in cool boxes and immediately brought to the laboratory. In the laboratory, macrophytes and bed-sediment samples were dried at 80 °C using a convection oven, NDO700 (TOKYO RIKAKIKAI Co., Ltd.) for >3 days until no further weight change was detected. Bed sediment fractions of different particle sizes were measured based on the protocol of the American Society for Testing and Materials (ASTM D422-63, 2007). Total nitrogen content (TN) of the bed sediment was analyzed by the Vario Micro Cube (Elementar Co., Ltd.) and total phosphorus content (TP) of the bed sediment was determined with the ascorbic acid method postdigestion using the perchloric acid method (APHA 2005). Biomass of macrophyte samples was measured for each species. Dissolved inorganic nitrogen (DIN) (NH₄-N, NO₂-N, NO₃-N) and phosphate (PO₄-P) of water were analyzed by an automatic analyzer, TRAACS800 (BL TEC K.K.). Ammonium concentrations were determined by the phenate method and nitrite concentrations by the sulfanilamide-NED dihydrochloride colorimetric method (APHA 2005). To analyze nitrate concentrations, nitrate was reduced to nitrite in the presence of cadmium treated with copper sulfate (cadmium reduction method), whereupon nitrite concentrations were determined by the aforementioned colorimetric method (APHA 2005) and phosphate by the ascorbic acid method (APHA 2005). Total dissolved carbon (TDC) and dissolved inorganic carbon (DIC) were determined by the Total Organic Carbon Analyzer (Shimazu Co., Ltd.). Dissolved organic carbon (DOC) was calculated by the difference between TDC and DIC. As carbon is a key factor for photosynthesis, free CO₂, HCO₃ ⁻, and CO₃ ²⁻ concentrations were determined via DIC concentration multiplied by the carbon fraction at a particular pH, which was derived from the following chemical equilibriums as specified by Stumm and Morgan (1996) (Eqs. 1, 2)

The most dominant species in the study site is E. densa. Other species in the site are Elodea nuttalliii St. John, Potamogeton crispus L., and some Chlorophyceae species (Fig. 2). The biomass of E. densa in summer was higher at the midstream reach (St. 6–15, except for St. 9–11), with an average value of 0.799–1.351 kg DW m⁻², in comparison with upstream and downstream reaches (St. 2–5 and 16–17), where the average biomass record was 0.052–0.272 kg DW m⁻². The reach from St. 9–11 has a concrete river bed, and biomass ranged widely from 0.047 to 1.241 kg DW m⁻², which was proportional to sediment-layer thickness (r = 0.71, p = 0.01) (Fig. 3).

The growth rate at the upstream, midstream, and downstream reaches were 0.2–5.1, 2.3–9.4, and 4.7–7.2 g DW m⁻² day⁻¹, respectively. In autumn, pH values of the channel decreased from St. 1–5 (7.37 ± 0.43 to 6.10 ± 0.06), maintained through St. 6–11 (6.10 ± 0.05 to 6.36 ± 0.07), and then again increased gradually from St. 12–18 (6.48 ± 0.08 to 6.90 ± 0.20). Heterogeneity of the pH distribution was more notable in autumn, and the biomass of E. densa at corresponding reaches seemed to follow the pH distribution pattern. The pH of the spring-fed pond (St. A) in autumn was 6.05 ± 0.04 and of the spring water supplied from the tributary to the midstream reach (St. C) 6.30 ± 0.02 (Figs. 2, 3).

Spring water contains lower DO but higher free CO₂ concentrations than the air-equilibrium concentrations. On the one hand, DO concentrations of the spring-fed pond (St. A) and spring water gushing from the river bed (St. C) were 6.88 ± 1.71 mg L⁻¹ (20.4 ± 1.8 °C) and 4.18 ± 0.34 mg L⁻¹ (18.1 ± 1.4 °C), respectively. The air-equilibrium values were 8.84 and 9.18 mg L⁻¹ at 20 and 18 °C, respectively, as reported by Truesdale et al. (1955). On the other hand, free CO₂ concentrations of St. A and St. C were 9.90 and 7.10 at mg L⁻¹, respectively. The air-equilibrium values for CO₂ were 0.60 mg L⁻¹ and 0.64 mg L⁻¹ at 20 and 18 °C, respectively (Weiss 1974).

Eigenvectors of the first three principal components (PCs) derived from PCA are shown in Table 1. PC1 represents the ratio of spring water to water from other sources because PC1 has a strong negative correlation with EC and a strong positive correlation with DOC and PO₄-P. It indicates that EC values of spring water are higher but poor in DOC and PO₄-P concentrations. PC1 also has positive correlation with water temperature, indicating that PC1 is related to spring water.

As shown in Fig. 3a, PC1 score variation over two seasons exceeds that recorded among observation points in the same season. It implies that the balance between the amount of spring water and that from surroundings, such as paddy fields, changed seasonally, and the ratio of spring water to all stream water in autumn was relatively greater than in summer. This might be attributable to no paddy field irrigation in autumn. PC2 represents the component associated with stream water acidity, as it has a strong positive correlation to pH. It has a strong negative correlation with TP content of bed sediment, indicating that PC2 also represents bed-sediment characteristics, especially the chemical aspects. As shown in Fig. 3a, sampling points indicating relatively lower pH of water (St. A3–13) in autumn, when the ratio of spring water to stream water was high, were located closer to springs, as shown in Fig. 1. This implies that stream pH was affected by spring water more strongly in autumn than in summer. PC3 mainly represents land-use type of adjacent areas: riparian forests or paddy fields. As PC3 has strong negative correlation to land-use type, higher PC scores of PC3 indicate paddy fields, whereas lower scores indicate riparian forests. PC3 also has negative correlation with particle diameter of bed sediment and sediment-layer thickness. These two variables correlate because a thin bed consisting of 1-cm-thick fine sand (d₅₀ = 0.41 mm) (ASTM 422-63, 2007) accumulated on the concrete river bed at St. 9–11. Therefore, PC3 also represents bed-sediment characteristic, especially of physical aspects. Furthermore, PC3 has negative correlation to both particle diameter of the bed sediment and land-use type. Positive correlation to TP content of the bed sediment is due to inflow of relatively fine soils with high TP contents from paddy fields (PC3 at St. 7–17 are higher than at others, Fig. 1). The inverse correlation between pH and land-use type is due to the fact that the upstream reach adjacent to riparian forests includes spring water gushing from the river bed (Table 1; Fig. 4).

TP content of bed sediment in both summer and autumn in the reach adjacent to paddy fields was significantly higher than that in the reach adjacent to riparian forests (p < 0.05) (Table 2). Particle diameter of bed sediment in the reach adjacent to paddy fields in autumn, on the other hand, was significantly smaller than that in the reach adjacent to riparian forests (p < 0.05) (Table 2). This implies that particle diameter of bed sediment in the reach adjacent to riparian forests changed from coarse to fine with the sedimentation process (Table 2).

Though the sum of free CO₂ and HCO₃ ⁻ concentrations were similar among E. densa vegetation stands, growth rates of those stands differed significantly (Fig. 5b). E. densa growth rate, however, was correlated to free CO₂ concentration (r = 0.67, p = 0.02). This indicates that E. densa growth rate is influenced by free CO₂ concentration (Fig. 5a).

Table 3 illustrates the Spearman rank correlation coefficients of growth rate to variables and shows that pH and DO had negative correlations and DOC had positive correlation with growth rate. Negative correlations of pH and DO with growth rate imply that E. densa growth rate increases with acidification of water (Table 3).

From summer to autumn (July to September), E. densa growth rate in the midstream reach was relatively higher (0.069–0.282 kg DW m⁻² month⁻¹) than in other reaches. These growth rates are comparable with that reported by Haramoto and Ikusima (1988) (0.102–0.415 kg DW m⁻² month⁻¹) for Lake Kasumigaura of eastern Japan between March and July. Slightly lower values in our study might be due to the difference in growing period and the interruption to growth during the developmental period of E. densa caused by dredging in July (Haramoto and Ikusima 1988). Roberts et al. (1999) reported an E. densa biomass range from 0.400 to 3.000 kg DW m⁻² in summer in the Hawkesbury–Nepean River in Australia. Yarrow et al. (2009) reviewed E. densa biomass production and concluded that it can range from 0.800 to 1.000 kg DW m⁻² during its rapid growth period. The growth rates in our study are in agreement with other studies of E. densa.

E. densa growth rate positively correlates to free CO₂ concentration and negatively to that of HCO₃ ⁻ in water. The change in pH shifts the equilibrium balance of free CO₂ and HCO₃ ⁻ toward either one. In the study site, the sum of free CO₂ and HCO₃ ⁻ concentrations among vegetation stands were similar. Therefore, as either form of DIC concentration becomes higher, the other one correspondingly becomes lower. This implies that if one DIC form has a positive correlation to the growth rate, the other automatically has a negative correlation. Approximately 50% of aquatic plants, including E. densa, can use HCO₃ ⁻ for photosynthesis (Yarrow et al. 2009). However, HCO₃ ⁻ can affect the photosynthesis of aquatic plants only to a limited extent (Vadstrup and Madsen 2006), as HCO₃ ⁻ assimilation requires more energy (Jones et al. 1993). Therefore, CO₂ must be far more important for E. densa growth rate, and it can thus be concluded that the growth rate is affected by free CO₂ concentration. Thus, HCO₃ ⁻ concentration has a negative correlation to the growth rate. This trend showing the relationship between the growth rate and HCO₃ ⁻ concentration may be applicable given sufficient free CO₂ concentration is present in the system. If the water has a free CO₂ deficit, the relationship may become positive with HCO₃ ⁻ (Sand-Jensen and Gordon 1986). Consequently, free CO₂ influences E. densa growth rate and may be one of the key factors for E. densa. Several studies report that macrophyte growth at lower pH is greater than that at higher pH with constant DIC levels (Sand-Jensen 1983; Titus et al. 1990; Titus 1992). This indicates that free CO₂ concentration determines macrophyte growth. Furthermore, the fact that CO₂ diffuses more slowly in water than in air (Lara et al. 2002) promotes the potential for free CO₂ to be the key factor affecting the growth of submerged macrophytes. Regression analysis between E. densa growth rate and free CO₂ concentration indicates that these two components are nearly proportional (Fig. 5). E. densa has a Kranz-less C₄-acid metabolism (Lara et al. 2002). In general, terrestrial C₄ plants have a CO₂ compensation point of 0–5 ppm in the gas phase. However, prediction of the same for submerged macrophytes is difficult, as the value varies from 150 ppm to nearly zero (Bowes and Salvucci 1989). It is also known that the CO₂ compensation point of E. densa changes with temperature (Lara and Andreo 2005). The regression line, of which y intercept is almost zero, indicates that E. densa can use even very low CO₂ concentrations for its growth.

The features of spring water in the study site are rich in free CO₂ concentration and poor in PO₄-P concentration. The values of free CO₂ concentration are 11–16.5 times greater than air-equilibrium concentration, and several streams contain more than ten times greater free CO₂ than atmospheric partial pressure (Madsen and Maberly 1991; Rebsdorf et al. 1991). Lower PO₄-P concentration compared with stream water is due to adsorption by soil particles (Pinoke and Urban 1985; Sakadevan and Bavor 1998). It is considered that spring water increases E. densa growth rate in two ways: One is that spring water itself is generally rich in free CO₂ and therefore directly promotes photosynthesis rate. The other is that if it is acidified, such as when it is affected by carbon acid, it shifts the equilibrium balance of free CO₂ and HCO₃ ⁻ toward a free CO₂-dominant regime in water. This change in equilibrium consequently increases free CO₂ concentration in vegetation stands. Titus et al. (1990) assumed the vigorous macrophyte growth at high DIC levels in low pH lakes on the basis of experimental results. This hypothesis supports the latter idea.

In conclusion, we observed that with a supply of spring water, submerged macrophytes were growing faster in other nearby streams, whereas in the absence of spring water, no colonization was observed. Although a detailed investigation is required, within the scope of the study reported here, it can be concluded that low pH spring water promotes the growth of submerged macrophytes.