Although there has been reduction in nutrient loadings from external sources, high nutrient levels and a prolific algal growth remain major stresses, affecting the water quality in Cootes Paradise, a coastal wetland at the western tip of Lake Ontario. It has been documented that internal loading, resulting from sediment P release, may be a significant contributor of the total P input to some lakes. To assess the importance of internal loadings from sediments to Cootes Paradise, nutrient fluxes from sediments at three locations were estimated. The investigated sites were representative of different sedimentary environments, including sites receiving effluents from a sewage treatment plant and combined sewer overflows. An unaffected site in the main body of the marsh was included for comparison. The results derived from the Fickian diffusion model indicate that sites receiving outfalls from the sewage treatment plant and combined sewer overflow had the highest nutrient fluxes. These two sites had also the highest ammonia-nitrogen fluxes. The lowest flux was estimated at the unimpacted site located in the main body of Cootes Paradise. The differences in nutrient fluxes appear to be attributable to spatial heterogeneity of bottom sediments. Sites which had the highest sediment phosphorus concentrations had the steepest porewater nutrient gradients and the highest fluxes. The results suggest that sediment phosphorus geochemistry is important in regulating the phosphorus concentrations in porewater, which consequently, governs the phosphorus fluxes from sediments. The data suggest that sediment may be an important source of nutrients in areas containing nutrient contaminated sediments and release of nutrients from these sediments may delay the recovery of the marsh even after the reduction of the external phosphorus loading.
Wetlands are important links between the terrestrial and aquatic ecosystems as they attenuate the flow of nutrients and toxicants to receiving waters and provide a habitat for many species of fauna and flora. Many Great Lakes wetlands receive a broad variety of contaminant loads from various sources, including agricultural and urban runoff, municipal waste waters and effluents from Combined Sewer Overflows (CSOs). In spite of their nutrient attenuation capacity, excessive loadings of nutrients have disrupted their natural cycles, resulting in accelerated eutrophication. It has been suggested (Peverly, 1982) that nutrient retention by wetlands depends on nutrient history and water movement through the system. There is, therefore, a need for better understanding of wetland processes which affect the assimilative capacity of wetlands.
Cootes Paradise Nature Sanctuary, a coastal wetland at the Western tip of Lake Ontario was once a mesotrophic aquatic system (Chow-Fraser et al., 1998). The marsh has been considerably degraded due to the excessive sediment and nutrient inputs from Dundas Sewage Treatment Plant (STP), marsh tributaries and CSOs (Hamilton Harbour RAP, 1992). Various strategies were initiated to improve the water quality in the marsh and restore the native vegetation. Exclusion of carp from the marsh was one of the initiatives undertaken, as it was suggested that resuspension of sediment and uprooting of plants by carp was largely responsible for the marsh degradation. Reduction in nutrient loadings from point sources, through upgrades in Dundas STP and reduction in CSO events, was another initiative in the restoration efforts. Although some improvement in water quality was evident, the response of the marsh to the restoration efforts has been varied. For instance, in 2001 the total phosphorus concentration in the water column averaged at 210 μ g l−1 (T. Theÿsmeÿer, Royal Botanical Garden, 2001 unpubl. data). This concentration is substantially higher than 70 μ g L−1, the target for P concentration set by the Hamilton Harbour Remedial Action Plan (RAP).
The ability of sediments to sequester P from the water column has long been recognized. However, whether sediments act as a P source or sink is determined by sediment composition and limnological conditions. Numerous studies (Williams et al., 1980; Fox, 1993; Lijklema, 1993; Sondegaard et al., 1993) have shown that some forms of sediment P are more readily exchangeable than others. The major processes which control the particulate/dissolved P equilibrium include adsorption/desorption, precipitation/dissolution, ligand exchange and enzymatic hydrolysis. The dominant transport mechanism from sediments to overlying water are diffusion, resuspension (by wind), bioturbation and gas ebulition.
Studies of shallow lakes (Larsen et al., 1981; Auer et al., 1993; Phillips et al., 1994) have shown that sediment may be an important source of phosphorus to the overlying water, with the P release as high as 278 mg P m−2 d−1. Although many estimates of P flux from sediments have been reported in the literature (Moore et al., 1991), the majority of them are results of laboratory studies obtained under the confined laboratory conditions (Holdren and Armstrong, 1986; Nurnberg, 1987; Auer et al., 1993; Kelton et al., 2004), or were obtained from the mass balance calculations (Preskot and Tsanis, 1997; Nichols, 1999). There is a broad range of values reported in the literature ranging from 0.1 × 10−4 to 0.1 × 10−1 g cm−2 y−1 (Lerman, 1978). To better predict the success of remediation in Cootes Paradise, reliable information on the sediment P release rates is essential.
To assess the importance of internal loadings of nutrients from sediments in Cootes Paradise, a nutrient loading budget has to be developed. This requires quantification of nutrient loadings from all sources, including reflux from sediments. The aim of this study was to estimate the in situ nutrient release rates from sediments to overlying water and to establish the relationship between nutrient fluxes and sediment composition.
Materials and methods
Cootes Paradise Nature Sanctuary is a 250 ha coastal wetland, adjacent to Hamilton Harbour, at the Western tip of Lake Ontario. The marsh is 4 km long and its maximum width is 1 km (Painter et al., 1989). The surface area and the volume of the marsh vary according to the water level fluctuations. The mean depth of the marsh is approximately 0.7 m (Chow-Fraser, 1999), but the water levels vary substantially. Interannual mean water levels fluctuated 1 m over the past two decades (Chow-Fraser et al., 1998). Three main tributaries drain into Cootes Paradise. The largest one is the Spencer Creek, a 43.5 km watercourse, which drains 79% of the 290.9 km2 watershed (Chow-Fraser, 1999). Two smaller creeks, Borer's Creek and Chedoke Creek, drain 6.9 and 9.4%, respectively of the remaining watershed (Chow-Fraser, 1999).
Three sites (Figure 1), representative of different sedimentary environments (Lee, 2001), were selected for the study of nutrient fluxes from sediments. One of the sites, denoted as CC, was situated near the mouth of the Chedoke Creek which conveyed the effluent of the CSO and upstream Kaydrage Park Landfill site (RBG, 2001). The second site, BH, was located near the Royal Botanical Garden (RBG) boathouse in the north-eastern corner of the main body of Cootes Paradise. Finally, the third site, WP, was located at the West Pond, which is a 9 ha receiving water body of the Dundas STP (RBG, 2001). Water from the West Pond passes to the main body of Cootes Paradise through a remnant dredged channel lined with willows on both banks. At each site, sediment porewater profiles were obtained by deployment of interstitial water (porewater) samplers and sediment cores, and water column samples were collected (circles in Figure 1). To investigate the spatial variability of P in sediments, surficial sediments were collected from six sites (triangles in Figure 1) distributed over the entire area of Cootes Paradise. Site selection was guided by the results of Lee (2001) to represent different depositional environments, hence sediments with different geochemical characteristics.
Sampling and analytical procedures
Interstitial water was sampled using acrylic porewater samplers (peepers) fitted with the inert 0.45 μ m polysulfone membrane (Gelman Scientific, Inc.). Peepers were filled with oxygen-free double distilled water (Azcue and Rosa, 1996). Prior to deployment, samplers were kept for several days in oxygen-free water maintained by bubbling with N2 gas. Details concerning their preparation are given in Rosa and Azcue (1993) and Azcue and Rosa (1996). A subsample of water, used for assembling and storage of peepers, was kept as a procedural blank for analysis to monitor for any possible contamination. The P concentration of the procedural blank was 0.0105 mg l−1. On 24 July 2001, several assembled samplers were placed in the sediments and allowed to equilibrate for a two-week period at each of the three locations, indicated as full circles in Figure 1. Porewater samples were withdrawn from individual cells immediately after retrieval using disposable syringes, transferred to plastic tubes containing appropriate sample preservative, and stored at 4°C until analyses. Sampling was completed within 10 min of retrieval of peepers from the sediments. Separate sets of samples were used for nutrient and metal analyses. The samples used for nutrient analyses were stored in vials containing 10 μ l of 7% H2SO4, and the samples used for metal analyses were preserved with 50 μ l of ultrapure Seastar concentrated HNO3.
The porewater samples were analyzed for nutrients and major ions by the National Laboratory for Environmental Testing (NLET) in Burlington ON, using Environment Canada Protocol (1979). An ascorbic acid technique was used to determine soluble reactive P (SRP), which includes mostly H2PO4− and HPO42 − species, denoted further as PO4-P. The O-tolidine method was used to determine ammonia nitrogen (NH3-N) from ionized (NH4+) and un-ionized (NH3) species. Major ions (Fe, Mn, Ca) were measured with an inductively coupled Ar plasma emission spectrophotometer (ICAP-ES). To aid with the explanation of the geochemical processes controlling the nutrient concentrations in porewater, supporting data on alkalinity, DOC, pH and Eh were obtained. Not all samples were analyzed for all the parameters. Dissolved organic carbon (DOC) was determined at 5 cm intervals using an Infrared Beckman Detector, Model 880. The remaining parameters, alkalinity, pH, and Eh were measured randomly. Alkalinity of porewater samples was determined in the field using standardized H2SO4 and a Hach digital titrator. Oxidation-reduction potential (Eh) and pH were determined in porewater immediately after retrieval in the field. A glass combination electrode calibrated against standard buffers (4 and 7) was used for pH measurements and a Pt/Ag/AgCl combination electrode calibrated against standardized (KCl/K4Fe(CN)6/K3Fe(CN)6) solution was used for Eh measurements.
Water column samples were collected at each site at the time of peeper deployment and retrieval. The samples were transported to the laboratory in 2-l bottles in a cooler. The bottles were shaken and subsamples were taken for pH and conductivity measurements. Separate subsamples were taken for total suspended solids (TSS), total P (TP) and total Kjeldahl, nitrate + nitrite and ammonia nitrogen determinations. Additionally, two ∼ 100 ml filtered samples were obtained using 0.45 μm filter paper for total filterable P (TFP) and filterable total Kjeldahl, nitrate + nitrite and ammonia-N. The subsamples taken for TP and TFP analyses were acidified with 1 ml of 30% H2SO4. In addition, depth and temperature of the water column were measured at the time of the peeper retrieval. Because of drought conditions, there was an approximately 15 cm drop in water level between the deployment and retrieval of peepers.
Sediment cores were also collected from all three sampling sites (Figure 1) at the time of peeper retrieval. The cores were transported to the National Water Research Institute where they were refrigerated and sliced into 1-cm sections to a depth of 20 cm and into 2-cm sections further down. The sectioned sediments were frozen in polystyrene vials and subsequently freeze-dried for chemical analysis. Homogenized sediments were used for all analyses. The surficial sediments were collected using a miniponar dredge. The top 5 cm of sediment from each site was stored in a polystyrene vial for analysis. The processing of collected sediment was described above. Water content and porosity of sediment was determined gravimetrically. Loss on ignition (LOI), a measure of the organic matter content, was determined gravimetrically after igniting the dry sediment at 550°C in a muffle furnace for 2 h. The resulting weight loss, expressed as a percentage of the dry material, was taken to represent the loss due to ignition of organic material. Total phosphorus content and distribution of P among various fractions were determined in all surficial and core sediments. Total phosphorus concentrations were determined on ignited sediments using a 16-hr 1 N HCl extraction. Three operationally defined forms of P comprising TP were determined by sequential extraction of Williams et al. (1976). They include non-apatite inorganic P (NAI-P), apatite P (AP) and organic P (OP), each of which can be interpreted in a depositional context. The NAI-P fraction includes orthophosphate adsorbed on Fe and Al oxides, Fe and Al minerals such as vivianite or variscite, and Ca-P minerals other than crystalline apatite (Williams et al., 1980). The NAI-P fraction is generally considered to be a measure of the maximum particulate P that can be rendered soluble during diagenesis. Apatite P includes P bound in crystal lattices of apatite grains and is generally considered biologically inert. This form of P is abundant in detrital particles. Organic P, which is considered to represent refractory organic P, was calculated by difference between the TP and the sum of the NAI-P and AP concentrations. This form includes P associated with carbon atoms in C─O─P and C─P bonds.
A multi-parameter HydrolabTM DataSonde, Series 3, was placed at the boathouse (BH) site to check if anoxic conditions occurred near the sediment-water interface. The water depth at the time of deployment was approximately 0.55 m and the distance of sensors from the sediment surface was about 10 cm. The Datasonde, which was calibrated before and checked after deployment, was installed on 24 July 2001 and took measurements in 1-h intervals. The DataSonde was retrieved on 8 August 2004, at the same time as peepers.
Data analysis and flux calculation
Statistical analysis was done for each sampling site separately to determine the relationships between nutrients and major ions (Fe, Mn and Ca). Simple correlation and multiple regression analyses for porewater samples were done using PO4-P concentrations as a depended variable and Fe, Mn and Ca concentrations as independent variables. Due to the log-normal distribution of the investigated parameters, the analyses were done on the log transformed data.
Diffusional flux calculation, assuming Fickian diffusion, was used to estimate the nutrient fluxes from sediments in the present study. The P fluxes across the sediment-water interface were calculated from the porewater data using solute gradient (dC/dZ)Z = 0 across the interface and the Fick's First Law:Li and Gregory (1974). The coefficient D is temperature dependent and increases with rising temperature, resulting in seasonal differences in diffusional flux (Li and Gregory, 1979). The value of the diffusion coefficient, DS, in P flux calculation is 231.5 cm2 y−1. The following assumptions were made for calculations: 1) viscosity and coupling effects are negligible, 2) there is no solid phase precipitation or biological uptake of the dissolved species near the sediment water interface, 3) concentration gradients are linear, so that (dC/dZ)Z = 0 is equal to Δ CvΔ Z (Lerman, 1979). The concentration gradients (dC/dZ)Z = 0 were found by fitting the porewater data to a linear regression equation Y = A + BZ, where Y is the concentration of the solute at the depth Z (Z = 0 at the sediment-water interface), B is the calculated slope of the line (dC/dZ) and A is a constant calculated from the linear regression. Comparable calculations were carried out for the release of the NH3-N using Ds = 624.4 cm2 y−1.
The physical parameters of the water column presented in Table 1 indicate hot summer conditions, which prevailed in the marsh at the time of our survey. Because there was no substantial change in specific conductance and pH of water samples between July and August, average values are presented in Table 1.
The chemical parameters in the water column measured at the time of peeper deployment and retrieval (Table 2) provide a snapshot of the nutrient status of the water column at the time of deployment and retrieval of the peepers. As expected, the highest nutrient concentrations in the water column were observed in the West Pond (Table 2), where extremely high NH3-N concentrations were measured in August. Under the hot summer conditions, much (9–14%) of the NH3-N in the water column would have been present as toxic unionized ammonia (NH3). In July, the NO3 + NO2 concentrations reached 15.7 mg l−1 in West Pond, concentrations substantially higher than those measured at the other two sites. As shown in Table 2, TFP accounted on average for about 19% of the TP in the main water body of the marsh. The proportion of TFP was, however, substantially higher (average 42%) in the West Pond, where massive benthic algae mats were observed during the large part of the summer (RBG, 2001).
The oxygen concentrations above the sediment-water interface, measured by the DataSonde, showed a strong diurnal variation (Figure 2). Because there was a problem with the initial data, only the data past the fifth day are shown in Figure 2. Although the oxygen concentrations were dropping steadily, at no time did the concentrations drop much below 4 mg l−1. This indicates that at no time over a two-week period anoxic conditions were present at 10 cm above the sediment-water interface at the BH site. The oxic conditions above the sediment-water interface were likely sustained by frequent mixing of the water column caused by wind action. The site was located in the main body of water of shallow depth (0.4 m), so the water column was highly exposed to the wind.
Porewater composition and nutrient release rates
The porewater profiles of major dissolved constituents are shown in Figure 3. Although the concentrations of all dissolved constituents increase with depth, there are noticeable inter-site differences between the porewater profiles and maximal values. As seen from Figure 3, the highest porewater concentrations of phosphorus were measured at the West Pond (WP) site, where the maximum concentrations reached about 7.5 mg PO4-P l−1. The site adjacent to the Chedoke Creek (CC) had the second highest P concentrations with maximum values slightly exceeding 5 mg PO4-P l−1. The lowest P concentrations were measured at the Boathouse (BH) site, where the maximum values hovered around 1 mg PO4-P l−1. The WP site, with the highest P concentration had the steepest porewater P gradient, followed by the CC site. Conversely, the BH site with the lowest porewater P concentration had the smallest P gradient.
A similar trend was also observed for the NH3-N. The CC site had the highest NH3-N concentrations, followed by the WP site. As for P, the BH site had the lowest NH3-N concentrations. At the CC site, the porewater reached concentrations of nearly 65 mg NH3-N l−1 at 40 cm below the sediment surface.
A significant (p = 0.001) correlation was observed between the NH3-N and metal ion (Fe, Mn and Ca) concentrations at all three sites (Table 3), suggesting deamination or ammonification and reductive dissolution of Fe and Mn oxides due to increasingly reducing conditions within the sediment column. This is confirmed by decrease in Eh values from positive values (+200 mV, 3 cm above the sediment-water interface) to −140 mV in deeper porewater. There was a noticeable increase in porewater alkalinity with depth at the CC site, where the alkalinity increased from 169 mg CaCO3 l−1 at 9 cm above the sediment-water interface to 897 mg CaCO3 l−1 at 30 cm below the sediment surface. Somewhat lesser increase in alkalinities was observed at WP and BH sites, with the highest alkalinities reaching 450 and 360 mg CaCO3 l−1, respectively. The alkalinity increase is likely due to denitrification, Fe, Mn reduction and CaCO3 dissolution. The significant (p = 0.001) correlation between NH3-N, Ca, Fe and Mn (Table 3) reflects the concurrent occurrence of these geochemical processes. The PO4-P concentrations were strongly correlated with Fe, Mn and Ca (Table 3) at CC and WP sites, however the relation between PO4-P and major ions was not that close at the BH site. As shown in the next section, the weaker correlation between PO4-P and major ions is due to P speciation in sediment. Incorporation of three variables (Fe, Mn, and Ca) into multiple regressions improved correlation coefficients and resulted in following relationships for CC, BH and WP sites, respectively:
Dissolved organic carbon concentrations increased with depth from 7 mg l−1 in water above the sediment-water interface to 13 mg l−1 at 12 cm below the sediment surface and kept increasing to 30 mg DOC l−1 at 42 cm below the surface at the CC site. However, the steady increase in the DOC concentrations with depth was lacking at the BH and WP sites. In general, the porewater DOC concentrations at the BH site were low (7–10 mg l−1); higher DOC concentrations (14-17 mg l−1) were measured at the WP site.
Phosphorus release rates, calculated as upward diffusive fluxes, at each site are shown in Table 4. The results show that there are substantial (nearly 20 fold) differences in diffusive fluxes of P from sediments, resulting from large spatial variation in sediment porewater chemistry. The estimates of P fluxes are highest (5.3 mg m−2 d−1) at the West Pond site, followed by the site adjacent to the Chedoke Creek (4.4 mg m−2 d−1) and lowest at the Boathouse site (0.3 mg m−2 d−1).
Fluxes 29.3 and 23.9 mg m−2 d−1 of NH3-N were calculated for CC and WP sediments, respectively. No calculation was done for BH sediments, as there was a poor fit for the NH3-N concentration. However, because of the gentle gradient of this parameter (Figure 3), the fluxes at this site are not expected to be high.
The LOI, water content and porosity of sediments from all three locations are shown in Figure 4. Sediments from WP had the highest porosity and the highest content of organic matter. As shown by the LOI values (Figure 4), a sharp increase in organic matter content is observed at about 12 to 13 cm below the sediment surface at the WP site. The LOI increase is coupled with the marked drop in the TP concentrations (Figure 5) at this depth, largely because of an abrupt drop in NAI-P concentrations (Figure 6). Interestingly, the OP concentrations, reflective of refractory terrigenous organic matter, increase at this depth from ∼ 300 mg kg−1 to ∼ 500 mg kg−1 at 25 cm below the sediment surface (Figure 6). This increases the OP contribution to the TP from ∼ 20 to ∼ 50%.
The sediments from the CC site (Figures 5 and 6) had the second highest TP concentrations. The lowest TP concentrations (Figures 5 and 6) were measured at the BH site. The NAI-P, which includes the most readily soluble forms of P, accounted for 50% of the TP in the top 12 cm of WP sediments (Figure 6), while below 13 cm this form of P contributed only 21% to the sediment TP content. The NAI-P constituted about 47% of the TP in sediment core collected at the CC site. The proportion of the TP in the NAI-P fraction (45%) was lower in the top 20 cm than that in deeper sediments, where the NAI-P accounted for more than 50% of the TP. The sediments from BH site, with the lowest TP concentrations had the lowest proportion (22%) of TP in the NAI-P form (Figure 6). The TP concentrations of surficial sediments from six stations varied from 891 to 1655 mg P kg−1 dry weight (Table 5). Likewise, the distribution of P among various fractions (Table 5) varied and showed a trend similar to that observed in core sediments, where the P apportionment shows that sediments which had the highest P content had also the greatest NAI-P component.
A mass balance study of Cootes Paradise (Prescott and Tsanis, 1997) suggested that reflux from sediments may be equivalent to as much as 57% of the total P loadings or 45 mg P m−2 d−1. The reflux estimate of that study includes diffusion, wind resuspension and bioturbation. Similar P release rates were reported for Lake Sobygard, a shallow lake in Denmark by Sondergaard et al. (1990). Since Cootes Paradise is shallow and largely well mixed, it is unlikely that anoxic conditions would occur in open water for any significant length of time, and anoxic release from sediments would not likely take place except for areas with restricted water circulation. This includes areas such as West Pond, Westdale Cut and Chedoke Inlet. The transport of P from sediments into the overlying water would likely occur from readily exchangeable and highly mobile P in porewater through diffusion and resuspension (Bostrom et al., 1982). While the estimates of P reflux from sediments were derived from the mass balance calculation, no validation of this parameter was performed and the reflux of P from sediments of Cootes Paradise was identified as a data gap (Prescott and Tsanis, 1997).
There are various methods to estimate the solute fluxes from sediments. Three of them, namely direct flux, diffusional flux and seepage flux measurements were employed by Matisoff and Eaker (1992) in their study of the Lake Erie coastal wetlands at Old Woman Creek. The seepage flux method does not measure chemical changes directly; rather, it calculates the solute transport associated with the vertical component of the groundwater advection (Matisoff and Eaker, 1992). A study measuring seepage fluxes at Cootes Paradise has recently been initiated and is presently ongoing (C. Ptacek, Environment Canada, Burlington, ON, Canada, pers. comm.). A flux box was employed by Matisoff and Eaker (1992) to directly measure the total flux across the sediment-water interface. However, because this measurement is dominated by rapid reactions occurring at the interface (Matisoff and Eaker, 1992), the method provides a flux estimate at the time of measurement. Diffusional flux measures concentration differences between the porewater and the overlying water and provides an estimate of a longer term flux associated with sediment burial (Matisoff and Eaker, 1992). For this reason, diffusional flux calculation was used in the present study.
As shown in the previous section, the current study indicates that there are large spatial differences in P release rates (Table 4) from sediments. Such large differences are plausible, as a broad spread of values (0.27–270 mg m−2 d−1) was reported for the freshwater sediments by Lerman (1979). A current laboratory study of microbial community at Cootes Paradise (Kelton et al., 2004), although overestimating the P release rates, also suggested that the P release rates were highly variable (0.96–28.28 mg P m−2 d−1). The estimated fluxes were highest (Table 4) for sediments containing highest concentrations of dissolved P (Figure 3), which appeared to be related to sediment P pool, particularly to the most labile form of sediment P, the NAI-P. As the close correlation between phosphorus (PO4-P) and major ions indicates (Table 3), Fe, Mn, and Ca associated P is easily released into the interstitial water during the early stages of diagenesis and is readily available for release into the overlying water. The form of dissolved P most readily utilized by the aquatic biota in the water column is PO4-P. Therefore, it is evident that the NAI-P is the source of the exchangeable P in sediment. This conclusion is consistent with the conclusion of Sondergaard et al. (1993), who found that most of P lost from sediment was derived from the NaOH fraction and some from the organically-bound fraction. The results suggest that sediment geochemistry, including forms of P and LOI (Figures 5 and 6), is an important factor regulating the PO4-P concentrations in porewater. The nutrient enriched sediments deposited in areas receiving the outfalls of the STP and CSOs are reservoirs of a large pool of nutrients which are rendered soluble during the early stages of diagenesis. Release of P from sediments with a history of high nutrient loadings such as those in West Pond has been suggested by Schindler et al. (1977). Although near-surface sediment P concentrations were highest at the WP site, they drop abruptly at about 12 or 13 cm below the surface. The NAI-P and other P form distribution as well as the LOI profile in this core suggest changes in the character of the deposited organic matter. Organic matter below 13 cm is more refractory and suggests the presence of vascular plant material rather than phytoplankton and benthos. This explains the absence of a significant DOC increase with depth in deeper porewater at the WP site. The observed trend in the sediment organic matter suggest a presence of a cattail dominated marsh, which was present in the area of the West Pond until the late 1960s to early 1970s. Prior to this time, water flowed into Cootes Paradise via Desjardin Canal. The canal wall broke in late 1960s and early 70's, since when water flows into the West Pond. A similar trend associated with disappearance of a cattail marsh due to a fire was observed at the Sanctuary Pond of Point Pelee Marsh (Mayer et al., 1999). Assuming linear deposition rates and knowing the depth of recent sediments, the calculated rate of sediment deposition in the West Pond is ∼ 0.4 cm y−1. This is somewhat higher than 0.24 cm y−1, the intermediate sediment accumulation rate estimated for Cootes Paradise by Prescott and Tsanis (1997). A higher accumulation rate is expected because the West Pond is the receiving water body for the Dundas STP and storm sewer.
Sediments with the lowest P fluxes (BH) from the BH site had the lowest proportion of the TP in the NAI-P form (Figure 6), which comprises Fe and Mn bound P and Ca associated P other than crystalline apatite. This sediment P form is most readily soluble under anoxic conditions, as those found in sediments, and as already discussed, is a principal contributor to dissolved P pool in porewater. The low proportion of this sedimentary P form explains the weaker correlation (Table 3) between the PO4-P and major ions (Fe and Ca) in porewater at this site. The results clearly show that the sediments with the highest TP concentrations and greatest proportion of the NAI-P, have the largest dissolved P pool.
The present investigation shows that there are hot spots in Cootes Paradise, where sediment may be an important source of nutrients to the water column. For instance, calculation using the estimated flux indicates that West Pond sediments may contribute 0.47 kg P d−1 to the water column, assuming similar composition of sediments throughout the West Pond. This compares with the Dundas STP P loading of 7.15 kg d−1, calculated from the mean (May-September) effluent P concentration and average flow of 0.19 m3 s−1 (T. Theÿsmeÿer, Royal Botanical Gardens, Ontario, Canada, 2001 unpubl. data). Although the Dundas STP has a certificate of approval (CA) to discharge effluent with P concentrations up to 0.500 mg l−1, it is capable, under the optimal operating conditions, to remove P to concentrations as low as 0.250 mg l−1. Assuming this concentration and an average daily flow of 0.21 m3 s−1 (a value more realistic for all seasons), the average calculated loadings of P from the STP are about 4.5 kg d−1. These loadings are lower than the loadings calculated for the four month summer period, which may reflect the worse case scenario. Assuming a mean flux of 1 mg P m−2 d−1 over the entire area (250 ha) of Cootes Paradise, the calculated internal loading of P from sediments is 2.5 kg d−1. Although the net internal loading of P from sediment to the water column would be somewhat lower due to sedimentation, the results show that sediment is an important contributor of dissolved P in Cootes Paradise. The internal loading from sediment is particularly important in the areas with the restricted water circulation such as West Pond, where an abundant growth of benthic algae was observed during the summer. The release of P from sediments may be further exacerbated if periods of anoxia occur from time to time in these parts of the marsh. Enhanced release of P from sediments encourages the growth of benthic algae (Phillips et al., 1994), such as that observed in West Pond.
While higher temperature in the summer enhances the diffusional fluxes, the above flux estimates may be conservative and probably underestimate the quantities of P released from sediments, as they consider diffusion under quiescent conditions only. Wind, which has not been taken into account, enhances release of P and other nutrients from sediments by stirring of sediment, thus reducing the distance at the sediment-water interface over which P diffuses (Andersen, 1974). Stirring of sediment by wind also exposes the sediments to chemical changes which may promote increased phosphorus release (Bostrom et al., 1982) and allows direct mixing of porewater and overlying water, thereby increasing P concentration in the water column. There is also evidence (Sondergard et al., 1993; Phillips et al., 1994) that bioturbation enhances P release from sediments.
Several studies (Sondergard et al., 1993; Phillips et al., 1994) have shown that P may be released from sediments long after the external loadings of P are reduced. Significant release of P from sediments apparently occurs (Holdren and Armstrong, 1986; Phillips et al., 1994) if dissolved P in sediment exceeds the molar equivalent of dissolved Fe. Our porewater data (Figure 3) indicate a substantial excess of dissolved P (25 μ mol l−1) in comparison to molar Fe concentration (5 μ mol l−1) at the sediment-water interface at the West Pond (WP) site and suggest a strong release of P from sediment. The porewater data (36.8 μ mol P l−1 and 24.5 μ mol Fe l−1, at the interface) suggest that the rate of P release would be somewhat lower at the CC site and no P release would occur at the Boathouse (BH) site where the molar concentration of dissolved Fe exceeds that of P. Thus, sediment release of P may hinder recovery of Cootes Paradise following any further reduction of nutrient loadings from the external sources and further delay the restoration efforts.
Since nutrient fluxes are highly dependent on the sediment nutrient pool, spatial differences in sediment geochemistry constitute an important factor in overall internal loadings of nutrients. An earlier sediment survey (Lee, 2001) of Cootes Paradise has shown that there is a large spatial variation in sediment geochemistry and our results (Table 5 and Figure 5 and Figure 6) support it. Thus, a large spatial grid needs to be investigated to address the spatial heterogeneity of sediments and provide meaningful estimates of fluxes for quantifying the overall loadings from sediments.
The authors gratefully acknowledge the dedicated assistance of Aaron Lawrence and the capable field assistance of Mike Mawhinney and Ken Hill. The field equipment was prepared and calibrated by Sergio Krickler and Bob Rowsell. The efforts of Graphic Arts Unit in production of final figures and the logistical support of the Royal Botanical Gardens staff, namely, Tÿs Theÿsmeÿer is greatly appreciated. Professor Peter Lee of the Lakehead University in Thunder Bay kindly shared the information with us and provided access to sediment data. We also thank Scott Brown, Jim Maguire and the anonymous reviewers for valuable comments and constructive criticism. The financial support of the Regional Municipality of Hamilton Wentworth is greatly appreciated.