Abstract

Due to their potential significance as indicators of ecological health, the biogeochemical cycling of carbohydrates and uronic acids was investigated in the Trinity River Texas, during 2000–2001. Concentrations of dissolved organic carbon (DOC), total carbohydrates (TCHO), polysaccharides (PCHO), monosaccharides (MCHO), uronic acids (URA), as well as of oxygen, suspended particulate matter, nutrients and trace metals (Cu, Pb, Cd) were assessed at various stages of discharge. TCHO/DOC ratios, as well as nutrient and hydrogen ion concentrations, were inversely related to temperature, which suggests that biological processes in Lake Livingston, the largest freshwater reservoir along the Trinity River, are not only regulating nutrient concentrations but also the preferential degradation of carbohydrates over that of bulk DOC. However, uronic acids were selectively preserved during this temperature controlled biological process, as is evident from the positive correlation of URA/TCHO ratios and temperature. Thus, uronic acids are more refractory compounds than bulk TCHO. Significant correlations between TCHO and dissolved Cu, Pb and Cd suggest that their pathways and cycles are linked through common sources or removal processes.

Article Outline

1. Introduction

2. Materials and methods

2.1. Sample collection and measurement

2.2. Analysis of carbohydrates and uronic acids

3. Results and discussion

3.1. Discharge and ancillary hydrographic data

3.2. DOC and carbohydrates

3.3. Effect of biological activity on CHO and URA concentrations

3.4. Relationship between trace metals and carbohydrates

4. Conclusions

Acknowledgements

References

1. Introduction

Carbohydrates are among the most important biopolymers in freshwater systems (Thurman, 1985, Buffle, 1990, Sigleo, 1996 and Mannino and Harvey, 2000). Previous research has shown that the composition of carbohydrates in fresh waters is similar to that in seawater (Repeta et al., 2002). However, only a minor fraction of carbohydrates is characterized at the molecular level. For example, glucose, galactose, arabinose, xylose, mannose, fucose, and rhamnose, identified by HPLC or GC–MS, are the main components of carbohydrates in rivers and lakes (Sweet and Perdue, 1982, Cheng and Kaplan, 2001 and Repeta et al., 2002). Since it has been shown that acid polysaccharides such as uronic acids (URA) can be excreted by algae and bacteria in response to low nutrient or high metal stress (Costerton, 1984 and Leppard, 1993), they can also play a significant role in heavy metal detoxification in aquatic environments (Leppard, 1997). Hung et al. (2001) reported that the concentrations of dissolved URA, acid polysaccharides with a carboxylic group, can make up a significant fraction of carbohydrates, i.e., up to 12% in the waters of the Gulf of Mexico. Recent studies have shown that URA compounds also exist in both suspended and sinking particles as well as in sediments (Bergamaschi et al., 1999 and Hung et al., 2003), but their biogeochemical cycling in river, estuarine and oceanic environments is still poorly understood.

Recent studies have shown that trace metals in natural waters are complexed by natural organic ligands (van den Berg et al., 1987, van den Berg et al., 1991, Kozelka and Bruland, 1998, Muller, 1999 and Tang et al., 2002) and the distribution of trace metals in natural waters is strongly affected by the production of organic ligands (Sholkovitz, 1976 and Santschi et al., 1997). Some organic ligands, such as phytochelatins, thiols and extracellular polysaccharides, have been found in metal stress experiments (Leppard, 1993, Leppard, 1997, Ahner and Morel, 1995, Leal et al., 1999 and Rozan et al., 2000), but the chemical compositions of organic ligands in natural waters are largely unknown. In order to better understand the processes that control the biogeochemical cycling of carbohydrate and URA compounds, and their interaction with trace metals, concentrations of carbohydrates and URA were measured during the calendar years 2000–2001 in the Trinity River, Texas. The Trinity River is the primary freshwater input source to Galveston Bay, contributing, on average, approximately 83% of the gauged input. Lake Livingston, the second largest lake in Texas, with a 84,800-acre (34,300-ha) reservoir, is located around 80 miles (130 km) north of the mouth of the Trinity River. Lake Livingston’s dam was built in 1969 to modulate the flow of the Trinity River. Detailed information about organic carbon is reported separately (Warnken and Santschi, 2004).

2. Materials and methods

2.1. Sample collection and measurement

Samples were collected from the Trinity River (29°48′N, 94°45′W, Fig. 1), Texas, using clean Teflon bottles during the period beginning September 2000 through August 2001. After collection, samples were immediately brought to the lab and filtered through GF/F filters. Carbohydrate samples were stored in a freezer (−20°C) until analysis. Water temperature, conductivity and dissolved oxygen (DO) were measured during the sampling period. DO was measured using a dissolved oxygen electrode sensor (YSI 90 MEA). Additionally, samples were collected for the analysis of suspended particulate matter (SPM), dissolved organic carbon (DOC), nutrient, and trace metal concentrations. DOC samples were collected by filtration of 50 ml aliquots of river water using a 25 mm GF/F filter, precombusted at 450 °C for 5 h and the filtered solution was measured by catalytic high temperature combustion using a Shimadzu TOC 5000 analyzer (Guo and Santschi, 1997). Nutrients were analyzed by a flow-injection spectrophotometric method (Grasshoff et al., 1983). Dissolved trace metal (Cu, Cd and Pb) concentrations were measured by on-line ICP-MS (Warnken et al., 2000). Briefly, river samples were collected by ultra-clean techniques, filtration using acid-cleaned prefilters and filters (0.45 μm, MSI) in acid-cleaned Teflon bottles. These samples were stored in a cooler in double plastic bags (the inner bag acid-cleaned) and stored in a refrigerator until measurement (Warnken, 2002). The daily discharge of the Trinity River was recorded by the United States Geological Survey (2002).

2.2. Analysis of carbohydrates and uronic acids

Dissolved carbohydrates, including total carbohydrates (TCHO), monosaccharides (MCHO), and polysaccharides (PCHO) were analyzed using a modification (Hung et al., 2001) of the TPTZ method (Myklestad et al., 1997). Briefly, the concentrations of TCHO were determined as follows: 3 ml of river sample and 0.3 ml of 1 N HCl were added to a 10 ml glass ampoule. The sealed ampoules were hydrolyzed at 150°C for 1 h. Then, 1 ml of hydrolysate, after neutralization with sodium hydroxide, was pipetted to a dark glass bottle containing 1 ml of 0.7 mM potassium ferricyanide, and the well-mixed solution was placed in a boiling water bath for 10 min. After that, 1 ml of 2 mM ferric chloride solution and 2 ml of 2.5 mM TPTZ (2,4,6-tripyrridyl-s-triazine, Sigma) solution were added and mixed on a Vortex mixer. The absorbance was measured at 595 nm. The reagent blank in Milli-Q was subtracted before calculating the concentration of the monosaccharide (e.g., glucose). The MCHO concentration in samples was directly measured without hydrolysis. The PCHO concentration in samples was equal to the difference between TCHO and MCHO ([PCHO] = [TCHO] − [MCHO]). The concentrations of TCHO, MCHO and PCHO are expressed as μM C.

The concentrations of dissolved uronic acids (URA) were analyzed according to Filisetti-Cozzi and Carpita (1991) and Hung and Santschi (2001). Briefly, 10 ml of river water were concentrated to 0.4 ml using lyophilization and was pipetted to a glass vial. Then, 40 μl of sulfamic acid was added and the solution was stirred on a vortex mixer. Subsequently, 2.4 ml of 75 mM sodium tetraborate in concentrated sulfuric acid was added to the vial and heated at 100°C for 10 min in a boiling water bath. After cooling, 30 μl of 0.15 % m-hydroxydiphenyl was added and the absorbance was measured at 525 nm. Glucuronic acid was used as a standard compound. For the control experiment, a UV-irradiated river water sample (24 h) was used, following the same procedures as for the samples. The river water blank was subtracted from the measured concentrations and the concentration of uronic acids were reported as total uronic acids (URA, μM C).

3. Results and discussion

3.1. Discharge and ancillary hydrographic data

Discharge rates, water temperature, conductivity, concentrations of dissolved oxygen and SPM in the Trinity River are shown in Table 1 and Fig. 1 and Fig. 2. Trinity River discharge was low during the months of September and October of 2000, but showed a high variability beginning in November 2000. A high discharge event began in mid-February 2001, with a crest in mid-March that eventually declined by the end of April. Discharge rates remained below average throughout the remainder of the study period, with the exception of an additional high discharge event during the month of June, associated with tropical storm Allison. This storm made landfall over Galveston, Texas, on June 8, 2001, and dumped up to 30 inches of rain, flooding the city of Houston and its surrounds based on the rainfall information from the National Weather Service Forecast Office (www.srh.noaa.gov/hgx/projects/allison01.htm).

Water temperature ranged from 7.5 to 31.6°C, reflecting a remarkably seasonal variation (Fig. 2A). The low value was recorded just after one of the coldest Decembers on record for this region of Texas. Elevated temperatures during the summer result in increased respiration rates, leading to oxygen undersaturation, while low temperatures in winter are the cause of reduced photosynthesis and respiration rates and increased gas solubility, leading to higher dissolved oxygen concentrations. Dissolved oxygen concentrations varied by a factor of two, and ranged from a summertime low of 6.15 mg l⁻¹ to a high value of 12.1 mg l⁻¹ (Table 1) in winter. A simple linear correlation to temperature was not observed during the study period, suggesting that oxygen concentrations were not simply controlled by equilibration with ambient atmospheric oxygen concentrations, but were modified further by in situ bacterial respiration and phytoplankton photosynthesis. For instance, Trinity River water was supersaturated or at saturation during the winter months and became increasingly undersaturated (as low as 83%) with increasing temperature during the summer months of both 2000 and 2001.

The concentrations of SPM ranged from 13 to 219 mg l⁻¹ (Table 1), with an extremely high value of 292 mg l⁻¹ appearing in June, 2001, caused by Tropical Storm Allison. Conductivity showed a general decrease during the study period, with values as high as 454 μS cm⁻¹ measured in September 2000. Generally, Trinity River water had increased and highly variable conductivity levels at low discharge rates, possibly due to evapotranspiration effects (Oktay et al., 2001) and reached higher but nearly constant conductivity levels at intermediate and high discharge rates. Observations in a Wisconsin watershed (Milwaukee River) by Shafer et al. (1997) for stream flows up to 50 m³ s⁻¹ were similar to ours, in that a significant linear decrease in conductivity was observed with increasing discharge.

The pH of Trinity River water ranged from a high value of 8.22 to a low of 7.56 (Table 1) remaining on the alkaline side of neutral throughout the study period. Generally, the pH is controlled by the CO₂ dissolution of CaCO₃, producing Ca²⁺ and 2 , the major component of alkalinity at the pH of 7.9 ± 0.3 in Trinity River water. This equilibrium with CaCO₃ is the main reason why Ca concentrations in Trinity River water are constant (Warnken, 2002).

A strong inverse relationship between dissolved oxygen concentrations and those of nitrate (Fig. 3A, R = 0.90; n = 24; p < 0.001), and, to a lesser extent, phosphate (Fig. 3B, R = 0.70; n = 24; p < 0.001) and urea (not shown), was observed throughout the study period, suggesting in situ biological control. For instance, during the winter, oxygen concentrations would have been least affected by biological processes, resulting in modest oxygen supersaturation and high dissolved nutrient (i.e., nitrate, phosphate) concentrations. During the summer months, disequilibria between respiration and photosynthesis would have resulted in the modest dissolved oxygen undersaturation, which increased with increasing temperature, and decreasing dissolved nutrient concentrations.

N:P ratios of between 30 and 50 were determined for the winter months (Fig. 3C). However, strong removal of total inorganic nitrogen (TIN), likely in Lake Livingston and the other upstream reservoir lakes, resulted in N:P ratios of approximately one during the summer months in both 2000 and 2001 (Fig. 3C). Lake Livingston has a significant effect on phosphorus and nitrogen in the lower watershed, accounting for a 60% decrease in phosphorus and a 30% decrease in nitrogen or a 2:1 decrease in phosphorus relative to nitrogen, which helps to explain the observations during winter (GBNEP, 1994). In other words, the average concentrations of TIN and P in Trinity River samples are 45 and 1.87 μM, respectively. The ratio of TIN/P in the winter should be 42 due to a 30% and 60% decrease in N and P, respectively. Nitrate was the nitrogen species present at the highest concentration during much of the year, followed by ammonium, urea and nitrite, respectively. Only at elevated temperatures (low nitrate concentrations) did ammonia reach levels similar to or greater than those of nitrate. A source of nitrate could be from soil leaching. However, ammonium and urea are derived mostly from washout of fertilizers and other biological nitrogen sources from agricultural soils. In summer time (at elevated temperatures), agricultural activity could result in higher ammonium and urea discharges and recycling of biological nitrogen sources. Normalized carbohydrates (e.g., TCHO and URA) also have a striking seasonal variation (Fig. 2D), similar to that of the nutrients. Thus, the concentrations of nutrients may be an important factor that affects the distribution of autochthonous and allochthonous sources of dissolved carbohydrates in the Trinity River.

3.2. DOC and carbohydrates

Temporal variations of DOC, MCHO, PCHO, TCHO and URA concentrations in the Trinity River are shown in Fig. 2. DOC concentrations exhibited low but variable concentrations at low discharge, increasing with increasing discharge, and ranging from 350 to 560 μM. The range of our DOC data is in agreement with previous investigations from the Trinity River mouth (Benoit et al., 1994, Guo and Santschi, 1997 and Tang et al., 2001). Most importantly, DOC was inversely related to conductivity (Warnken and Santschi, 2004), suggesting a watershed influence on the transport pathway of DOC, demonstrating the different effects of lower ionic strength surface runoff, enrichment by chemically eroded major ions, both coupled to the retention of DOC compounds during percolation through the saturated zone (see Table 2).

Concentrations of MCHO in the Trinity River showed large irregular variations, ranging from 27 to 112 μM C. PCHO concentrations were much lower than those of MCHO, accounting for only 10–20% of TCHO. MCHO was thus a major component of TCHO, ranging from 80% to 90% of TCHO. This high proportion of MCHO to TCHO is in contrast to that found in marine environments, where PCHO compounds are the major fraction of TCHO (Pakulski and Benner, 1994, Hung et al., 2001 and Hung et al., 2003). For instance, Hung et al. (2003) reported that PCHO is about 3–4 times higher than MCHO in the Gulf of Mexico. This difference between riverine and marine systems is due to in situ production of high molecular weight PCHO compounds by phytoplankton or bacteria in pelagic systems, while in river systems, most carbohydrates as MCHO originate from decomposition products of soil litter and plant leaves that are leached into surface waters (Hedges et al., 1994).

The temporal variation in TCHO/DOC ratios was inversely related to water temperature, with maximum ratios (>30%) in winter time and low ratios (on average, 13%) in summer time (Fig. 2D). In addition, the temporal variation of URA/TCHO roughly paralleled that of water temperature, i.e., was lower (on average, 14%) in the winter and higher (on average, 26%) in the summer.

In order to compare our data to that from other rivers and lakes, it is necessary to take into account geographic locations (mainly latitude and climate conditions). Table 3 lists relevant literature data on carbohydrates from rivers and lakes, indicating that the ratio of total carbohydrates to total organic matter ranges from 5% to 72% of DOC. Since different analytical methods for measuring total carbohydrates appear to give different results, one can only compare data acquired using similar methods. With that in mind, our data are in good agreement with those measured by spectrophotometric methods (Buffle, 1990, Wilkinson et al., 1997, Paez-Osuna et al., 1998 and Guéguen et al., 2004), but appear higher than data obtained using other methods (Ding et al., 1999, Guo et al., 2003, Benner and Opsahl, 2001 and Bianchi et al., 2004). For example, TCHO concentrations, determined as the sum of individual identifiable monosaccharide compounds determined by HPLC-PAD or GC–MS methods appear to be significantly lower. The significantly higher values obtained by the spectrophotometric technique, when compared to chromatographic methods are, however, to be expected, since a large fraction of the carbohydrate-type structures in natural organic matter has not been characterized at the molecular level by current chromatographic methods (Benner and Kaiser, 2003). In addition, humic substances (mainly humic and fulvic acids) also contain carbohydrate structures that are determined by the TPTZ method (Buffle, 1990), but not yet adequately by chromatographic techniques. In fact, we have tested the reaction of humic matter for TPTZ method and the result is 8:10:100 for humic acid:fulvic acid:glucose in terms of carbon. It supports that humic matter also provides a part of the TPTZ chemical reaction.

URA concentrations in the Trinity River also showed large variations, ranging from 5.2 to 25.7 μM C, without any marked seasonal variation in absolute concentrations (Table 3). However, relative amounts, as a percentage of TCHO, were strongly influenced by temperature. The average concentration of URA in the Trinity River was two to three times higher than in Galveston Bay (Hung et al., 2001), which receives its water mainly from the Trinity River. For example, these authors reported URA concentrations in Galveston Bay that ranged from 1.0 to 8.3 μM C. However, it is likely that some of the URA in the Trinity River were contained in humic acids, as glucuronic acids have been documented as parts of humic acids (Buffle, 1990). The URA/TCHO ratios in the Trinity River range from 0.1 to 0.3 and URA/DOC ratios from 0.01 to 0.05. Because data of URA and total acid polysaccharide (APS) concentrations in aquatic systems are scarce, it is difficult to directly compare our data with literature data. Buffle (1990) reported that URA made up 0.01–0.05 of soil organic matter and Bergamaschi et al. (1999) reported that uronic acids normalized to organic carbon in sinking and sedimentary particles ranged from 0.01 to 0.1. Recently, Aluwihare et al. (2002) reported that URA, mainly glucuronic and galacturonic acids, accounted for less than 2% of OC in the high molecular weight fraction. This comparison demonstrates, however, that our values of URA fall within the range of previously reported data.

3.3. Effect of biological activity on CHO and URA concentrations

The majority of riverine DOC compounds have been shown to be easily degradable by microorganisms (Benner and Opsahl, 2001, Hedges et al., 1997 and Raymond and Bauer, 2001). Microbial activity increases with increasing water temperature, preferentially utilizing specific and more labile DOC compounds, e.g., structural carbohydrates (Wilén et al., 2000). Sweet and Perdue (1982) observed during a short-term investigation (5 months) that dissolved polysaccharide concentrations in rivers and lakes of Oregon appeared to decrease from winter to spring. They attributed this decrease to microbial activity. Findlay et al. (1991) reported that bacterial abundances and production were significantly correlated with temperature in the Hudson River with higher bacterial abundances and production in summer. Leff and Meyer (1991) measured the bioavailability of DOC along the Ogechee River, Georgia, and the results showed that bioavailability of DOC changes temporally, resulting in lower concentrations at high discharge (March) and higher values at low discharge (January). These authors concluded that the causes for these changes were related to the composition and physiological state of the bacterial assemblage, as well as the nature of the DOC. They also pointed out that both organic matter and bacteria are also washed into the river from the floodplain during high discharge rates but many of these bacterial cells might be inactive.

We did not measure bacterial production and bacterial biomass during this investigation, but based on previous research, we postulate that the higher temperatures resulted in higher bacterial activity and abundance. Indeed, the negative linear relationship between in situ water temperature and the TCHO/DOC ratios, which coincides with that between O₂ and nutrients, suggests selective biodegradation of more labile organic carbon compounds (i.e., carbohydrates) at elevated temperatures (Fig. 4A) that also coincide with increased oxygen consumption and nutrient uptake. The most important known compounds of DOC in freshwaters, humic acids also contain a significant fraction of carbohydrates (e.g., Buffle, 1990; and references therein). Despite the documented degradability of the carbohydrate fraction of DOC, DOC concentrations did not correlate with temperature. The cause is unlikely related to the small temperature difference between the region where production and degradation processes might be most active, i.e., Lake Livingston, and 80 miles (130 km) downstream where it was recorded, as climatic differences are very small (www.weather.com/weather/climatology/monthly).

Carbohydrate concentrations are a function of the rates of production and degradation processes. TCHO compounds in the Trinity River, like in most rivers, have several sources: soil erosion of humic substances and in situ production, especially in reservoir lakes such as Lake Livingston, as well as point source discharges from waste water treatment plants. While soil erosion is most active at high discharge rates, treatment plant discharges are independent of discharge rates and temperature making this a relatively constant source. Our results suggest that carbohydrate concentrations decrease in response to temperature-dependent uptake and degradation reactions by bacteria and photochemical processes, leaving more refractory compounds behind. Recent work by Repeta et al. (2002) has suggested that organic carbon in freshwaters mostly consists of terrestrially derived humic substances and in situ microalgal production, with degradation resistant compounds being transferred into the ocean. However, one of the carbohydrate compound classes, URA, seems to be more resistant to degradation.

This interplay between production and degradation, resulting in the formation of refractory URA compounds, is exemplified by the positive correlation between water temperature and URA/TCHO (but not URA/DOC) ratios in our data set, with a positive intercept (Fig. 4B). The positive intercept suggests that about 8 ± 2% of TCHO consisted of pre-existing URA compounds. This either indicates that extra URA compounds are produced in situ at elevated temperatures, or that they are more refractory than more labile bulk TCHO compounds. While it is possible that bacterial activity, which increases with temperature (Shiah and Ducklow, 1997), can produce certain uronic acids, such as mannuronic acid and glucuronic acid (Kenne and Lindberg, 1983), the data would suggest biomass build-up at elevated temperatures being less important than increased degradation of bulk carbohydrates. This is supported by the significant negative correlation between TCHO/DOC and temperature. Thus, this positive correlation between URA/TCHO and temperature more likely indicates that URA compounds are more stable compounds than the neutral carbohydrates that make up the bulk of TCHO. Indeed, recent research has also shown that uronic acids exist in the water column, as well as in sediment trap material and in sediments (Bergamaschi et al., 1999, Hung et al., 2001 and Hung et al., 2003).

3.4. Relationship between trace metals and carbohydrates

It is well known that a large fraction of trace metals in aquatic environments may be complexed by natural organic ligands (Donat et al., 1994, Tang et al., 2001 and Ndung’u et al., 2003). However, it is difficult to directly measure the type of organic ligands binding trace metals in natural waters. In watersheds, both metals and potential organic ligands can have similar sources: waste water treatment plants and soil erosion. Any correlation between organic ligands and individual trace metals (e.g., Cu, Pb, and Cd), and potential ligand compounds that are present in constant proportion to some major fractions, such as TCHO (Fig. 5A–C) suggests that these trace metals and organic ligands have similar sources, e.g., soil leaching or waste water treatment plant discharges. The significant correlation found during this study between URA and dissolved Cu (p = 0.02) and Pb (p < 0.1) (Fig. 6A and B) could likewise indicate similar sources. While less likely here, it could also indicate trace metal stress that can cause the secretion of phytochelatins, thiols and extracellular polysaccharides (such as uronic acids) (Leppard, 1993, Leppard, 1997, Ahner and Morel, 1995, Leal et al., 1999 and Rozan et al., 2000). Furthermore, aquagenic or pedogenic sources of uronic acids might be distinguishable from the analysis of individual acid polysaccharide compounds. In the future, rather than an inferred relationship between URA and trace metals, a more direct approach to analyze organic-metal complexes such as hyphenated techniques, i.e., HPLC-ICP-MS, is needed to determine directly the organic matter components responsible for trace metal binding.

4. Conclusions

A seasonal investigation of carbohydrate species and related chemical parameters at one station in the lower reaches of the Trinity River shows that biological activity and photo-chemical reactions may significantly affect the production and consumption of organic matter in freshwater systems, especially in rivers with reservoir lakes. The significant and negative correlation between TCHO/DOC and temperature suggests that carbohydrate compounds are more easily utilized by bacteria or decomposed by photo-chemical reactions than bulk DOC. Conversely, the positive correlation between URA/TCHO and temperature indicates that uronic acids (URA), which are generated by microbial activities, are more resistant to degradation than bulk carbohydrates. More research is needed to test the relative degradability of individual acid polysaccharides, such as mannuronic acid, glucuronic acid, and galacturonic acid, etc., as compared to labile sugars. Concentrations of both total carbohydrates and URA showed significant correlations with those of selected trace metals (Cu, Pb and Cd), suggesting that they may have similar sources (e.g., washout from agricultural fields and waste water treatment effluents) and/or removal processes (e.g., scavenging by coagulating exopolymeric substances).

Acknowledgements

We thank Scott Bean for his assistance in carbohydrate analysis. This work was funded, in parts, by NSF (OCE-9906823 to P.H.S. and OCE-0351559 to P.H.S. and C.C.H.), and the Texas Institute of Oceanography.