Modelling and analysis of Cladophora dynamics and their

Modelling and analysis of Cladophora dynamics and their relationship to local nutrient sources in a nearshore segment of Lake Ontario

A completion report on University Sponsored Research Agreement #4253501: “Field Assessment and model validation of a coupled hydrodynamic-water qualityCladophora growth model to evaluate the importance of local point sources of nutrients to algal runs at Pickering”

(June 2008 - July 2009)

Prepared by: Luis F. Leon, Ralph Smith, Sairah Mailkin, David Depew and Robert E. Hecky October, 2009

2

Executive Summary The objectives of the project were to (1) determine the contribution of local nutrient sources to Cladophora growth near the Pickering Nuclear Generating Station (PNGS), (2) assess the contribution of locally-grown algae to fouling problems at PNGS, (3) determine the role played by the thermal bar, and (4) arrive at suggestions for potential control of the Cladophora problem. These objectives were addressed by developing and validating a 3D hydrodynamic and water quality model to simulate conditions affecting Cladophora dynamics, and coupling it to a model that could predict Cladophora growth and detachment. Validation was performed with reference to water quality measurements at >20 sites in the vicinity of PNGS from May through October of 2007 and 2008. Benthic algae (mostly Cladophora) were surveyed with hydroacoustics. The observations provided a view of two contrasting years, one with very little local runoff (2007) and one with unusually high summer runoff (2008). The model proved capable of predicting the values of environmental variables most relevant to Cladophora dynamics with sufficient accuracy to make it useful for the addressing the objectives. Model predictions of Cladophora distribution and dynamics were consistent with independent measurements by hydroacoustics and with records of algal runs at PNGS. The model appeared to capture the dynamics surrounding individual algal run events realistically. Model results and acoustic surveys identified a major potential source area for Cladophora to the east of PNGS. Field observations and model results in the dry (2007) and wet (2008) summer suggested local nutrient sources, which are primarily Duffins Creek and the Duffins Creek Water Pollution Control Plant (DCWPCP), were relatively small influences on conditions for Cladophora growth. Duffins Creek had a very localized effect on suspended sediments but negligible effect on key nutrients and light penetration over the study area. DCWPCP had more effect on nutrients but the impact was still relatively localized. The validated model supported the conclusion that nutrient inputs from Duffins Creek are not an appreciable influence on Cladophora growth in the area. Model results show practically the same growth for the scenario when the temperature discharges from the nuclear plant was considered ambient. Inputs from the DCWPCP at the 2007 loading levels do stimulate Cladophora growth according the model but other factors determine whether the algae will reach nuisance levels or not. Cladophora growth was large enough to be problematic in 2007 but not in 2008, even though inputs from DCWPCP and PNGS were essentially the same in both years. Measured and modeled estimates of Cladophora biomass accumulation in the study area, proximate to PNGS, were far more than the amount impinging so could easily account for the amount harvested at the plant. The modeled dynamics around the biggest impingement event of 2007 were consistent with a nearby origin. There was no evidence that far-field sources must be invoked to account for the impinging algae.

3 The thermal bar was outside the study area by the beginning of the study seasons (late April-early May). Its involvement could not be directly determined but would have to operate by longer-range effects (over time and space) that were not observable in this study. The study results pointed to non-local and/or in-lake processes as key to development of nuisance Cladophora in the study area. Restriction of the dominant local sources would not eliminate problems. Key processes likely include lake exchange processes that replenish nutrient supply in the nearshore and dreissenid mussel activity that helps maintain water clarity and nourishes Cladophora by providing regenerated nutrients on the bottom, where the algae live. There should in theory be some lower value for lakewide nutrient (specifically P) concentrations that would prevent accumulation of Cladophora to nuisance levels. With further work the model developed here could be used to estimate the loading targets for Lake Ontario that would be necessary to reach this point. The model could also be used to produce more regionally-scaled estimates of target loadings that would be more specific for the Lake Ontario north shore, within Canadian jurisdiction.

4

Table of Contents

Executive Summary

2

Introduction

5

Methods Water column sampling and analytical methods Benthic surveys

7 8

Results 1. Water Quality observations

Site-specific comparisons Seasonal averages by depth contour interval Discharge and water quality comparisons for the study area 2. Benthic plant surveys

9 9 13 17 18

3. Model Simulations for 2007 and 2008

Met forcing Model Setup Thermal structure & validation with Thermistor-Chain Data Description of Loads and Scenarios Validation with Water Quality Data Cladophera Growth Simulations

26 28 30 33 39 42

Synthesis and Conclusions

56

Recommendations

61

5

Introduction Lake-wide phosphorus (P) concentrations in Lake Ontario have decreased over recent decades due to restrictions on P loading mandated under the Great Lakes Water Quality Agreement (GLWQA). Offshore concentrations of P have in recent years been at or below the targets set under the GLWQA. One goal of the GLWQA was to eliminate nuisance growth of the benthic algal Cladophora, which inhabits mainly the shallower and rocky parts of the nearshore zone. For a period of time the diminished P concentrations in Lake Ontario were accompanied by a decrease in reports of Cladophora fouling problems. Since the mid-1990’s P concentrations and loadings have remained low but incidence of Cladophora fouling has increased. Surveys in the 2000’s have found high levels of Cladophora biomass accumulation. One manifestation of this apparent resurgence of Cladophora has been recurrent fouling of the water intake structure at the Pickering Nuclear Generating Station (PNGS) on the north shore of Lake Ontario. This expensive problem at the PNGS was the impetus for the work described here. The north shore of Lake Ontario is a mostly open coastline that is strongly affected by lake hydrodynamics. This coastal zone, especially the western part that includes the PNGS area, is also a complex area in terms of the many and varied anthropogenic influences at play. Most catchments in the western part are moderately to heavily urbanized, and population growth over recent decades has been considerable. This has an effect on the quality of tributary water inputs to the coastal zone, and also generates a need for wastewater and stormwater disposal that could provide nutrients for Cladophora growth. At the same time, there is evidence that nuisance growth of Cladophora is quite a general phenomenon over much the north and south coasts of Lake Ontario and is not restricted to the immediate vicinity of known point sources of nutrients. In analyzing problems at a specific site such as PNGS there are unresolved questions about the provenance of nuisance algae (local growth vs longer-range sourcing) and the role of local nutrient sources in generating problematic levels of algal biomass. The work described in this report was designed to address such questions with particular reference to the PNGS. This report builds on previous investigations that were conducted in 2006 and 2007 with support from Ontario Power Generation (OPG), Region of York and Region of Durham. The report “3D Hydrodynamic and Ecological Modelling” submitted to OPG, York and Durham by University of Waterloo in 2008 (Leon, Smith and Hecky) describes results of the previous investigations. The present report deals with the final development and application of a model (Higgins et al. 2005a) based on pioneering work by Auer and Canale (1982 a,b) to allow analysis of the provenance of, and controls on, nuisance Cladophora growth in the PNGS vicinity. In this application, the Cladophora Growth Model (CGM; Higgins et al. 2005a, 2006) is linked to a three dimensional model for water quality variables (ELCOM-CAEDYM, Robson and Hamilton 2004, Leon et al. 2006). This report also deals with results of an additional year of field investigations (2008) that provided further opportunity for model validation and testing of hypotheses about environmental controls on the algal problem as manifested at PNGS.

6 The approach taken in this work is to use a 3D hydrodynamic model, coupled with an ecological model for major water quality (WQ) variables, to define the conditions for growth and detachment of Cladophora in the vicinity of PNGS. This coupled 3D model is the ELCOM-CAEDYM (ELCD) model package developed by the Centre for Water Research (CWR) at the University of Western Australia. It has been refined for application in the Great Lakes, and specifically Lake Ontario, by the present research team. A significant part of the work reported here was to implement ELCD for Lake Ontario, including the nearshore zone proximate to PNGS. Only with such a model can the simultaneous influences of local nutrient inputs and the diffusive/advective effects of the lake hydrodynamics be properly evaluated. A further step forward that was supported by this project was to couple ELCD with a model to describe the growth and detachment of Cladophora. This Cladophora Growth Model (CGM) was first developed at U. Waterloo for application in Lake Erie and was first coupled with ELCD to simulate Cladophora in the east basin of Lake Erie. The ELCD-CGM coupled model has been refined under this and previous research agreements with OPG and municipal partners for use in the PNGS area. When properly validated, such a model provides the best means to establish quantitative relationships between nutrient sources and Cladophora abundance. It is the tool needed to evaluate environmental influences on algal growth, likely sourcing areas and transport pathways for detached algae, and potential control options. A key area of model development was to implement a nesting of ELCD to provide sufficient spatial resolution for meaningful simulations in the nearshore zone. This requires spatial resolution on the order of 50 m and it is not feasible to simulate the whole lake at such a scale. Following previous experience in Lake Erie, the current investigators defined a nearshore region around PNGS to simulate with high spatial resolution. This high definition region interacts across open model boundaries with whole lake simulations at a coarser (2 km) scale. This is intended to allow realistic simulations within the nearshore at manageable model run times. At the time of the previous report, we were obtaining success in ELCD simulations using 2007 observations. Results with ELCD-CGM were at a preliminary stage and a major focus of this report is to show the results obtained by application of the model in both 2007 and 2008. The results have benefited from the inclusion of loading data that were not available at the time of the last report and by a fortuitous contrast in weather and hydrology between 2007 and 2008. An important step in modeling-based analysis is validation. To this end, WQ variables in the water column were measured every 2-3 weeks from late April to mid-October. These measurements provided not only for model validation but also for a direct assessment of influences on WQ conditions in the study area. Measurements made in 2007 were described in our previous report (Leon et al. 2008). Here we include results for 2008 and make comparisons between years. The dry summer of 2007 and wet summer of 2008 provided an excellent natural experiment to assess the importance of variations in local catchment runoff.

7

These WQ observations allowed for testing and validation of ELCD, but we also surveyed benthic plant abundance using hydroacoustic methods to assess the performance of the coupled ELCD-CGM. Limited results for 2007 were available at the time of the last report. Here, we present a more extensive body of measurements for 2007 plus results for 2008. With model validation providing encouraging results, this report will provide some analysis of potential influences on Cladophora abundance in the vicinity of PNGS and the provenance of algae arriving at PNGS. The role of the major local nutrient sources (Duffins Creek and DCWPCP) will be assessed. The influence of thermal effluent from PNGS will also be analyzed. Comparisons against biomass collections on the intake structure of PNGS will be used to further assess model performance and to analyze specific incidents of algal impingement to diagnose likely sources of algae (local vs far-field). Modeled and measured benthic plant distributions are presented and will be analyzed to assess the likelihood that algal impingement is a function of local growth.

Methods Water column sampling and analytical methods Sampling began as early as practical in the spring (late April in 2008, early May in 2007) and continued to October. At each station, CTD profiles were taken using aYSI-6600 profiler. Photosynthetically active radiation (PAR) profiles were measured with a LICOR quantum sensor at 0.5 m to 1 m intervals from the surface to the lake bottom. At stations sampled by UW (all stations in 2007 and most in 2008), a 6 L Niskin bottle was used to collect water at a depth equal to 50% of the mixed layer depth (determined from CTD cast). At deeper stations, when thermal stratification was evident, a sample from below the thermocline was also taken when time permitted. At some stations in 2008, sampling followed a similar protocol but was conducted by TRCA personnel with samples transferred to UW personnel at the field site. Water was transferred to covered carboys and stored in coolers until transported to the laboratory (Univ. Waterloo). Samples for total dissolved P (TDP) and soluble reactive P (SRP) were filtered through a 0.2 um polycarbonate filter, particulate P (Part P) by filtering 500 mL of lake water onto acid soaked (5% HCl ~ 4 hr) Whatman GF/F filters (nominal pore size ~ 0.8 um). Total P (TP) and all composite fractions (TDP, SRP, Part P) were determined according to Stainton (1977). Samples for other particulate nutrients (carbon – Part C, nitrogen – Part N) were determined by filtering 500 mL of lake water onto pre-combusted (500 C ~ 4hr) GF/F filters and assayed using a CEC-440 Elemental Analyzer (Exeter Analytical, MA). Phytoplankton chlorophyll a was measured using a Turner Designs 10-AU fluorometer (Smith, et al 1999). Additional ions (NO3, Cl-, SO4-) were determined using ion chromatography (Dionex DX 500, Dionex AS17 and AG17 guard column). Ammonium was determined by florometry on a Turner Designs TD-700 fluorometer (Holmes, et al 1999). Total suspended solids (TSS) were determined by filtering 2 to 5 L of lake water onto pre-combusted (500 C for 4 hr) pre-weighed GF/F filters, drying at 65 C to a constant weight and re-weighing. AFDW was determined after combustion at 500 C for 4 hrs.

8

Benthic surveys Hydroacoustic survey methods for benthic plant growth, and particularly Cladophora, have formed the core subject of the PhD thesis of David Depew (Depew 2009). A full discussion of the methodology and its validation is beyond the scope of this report but Depew’s thesis gives a comprehensive treatment. Acoustic surveys to assess the spatial patterns of Cladophora sp. abundance were conducted using a BioSonics® DTX system with a 430 kHz 10.2° beam width (algal characterization) and a 120 kHz 7° beam width (substrate characterization) single beam echo sounders. Both transducers were set to ping at 5 Hz, with pulse lengths of 0.1 ms (430 kHz) and 0.4 ms (120 kHz) using the software Visual Acquisition 5.0 (BioSonics Inc, Seattle, WA). Analysis of the 430 kHz acoustic data was performed using a graphical user interface (GUI) in Matlab v7.2 (Depew 2009). Data from the 120kHz transducer was used for substrate classification classification with supplementary observations by video camera and grab samples. Surveys for benthic plant biomass were performed on a bi-weekly schedule (weather permitting). Acoustic surveys were performed by traversing transects parallel to shore, spaced ~ 75-100m apart. This provided optimal coverage of depths between shore and 10m. Average boat speed was ~ 2.1 to 2.3 m sec-1 to minimize excessive surface noise near the transducers and bottom signal loss. To completely cover the survey area, ~14-15hrs running time were required. During the Pickering surveys in 2007, algal material was harvested by divers using only snorkeling gear. This occurred at all UW transects (4 in total) at depths of 2m and 5m. quadrats (0.0625 m2) were sampled at pre-marked locations using weighted floats. Within each quadrat, triplicate measures of algal bed height were made using a ruler, and algal coverage was estimated. Algal biomass was then harvested by hand using fine mesh bags. A second quadrat was sampled for tissue (for later nutrient analysis). Additional measures of bed height and algal cover were made at various locations and depths to compare against acoustic measures. Additional diving was performed at 16 Mile Ck (Oakville) in mid July using SCUBA. The harvesting procedure was similar to that employed using snorkeling, however an airlift was used to harvest algal material into the mesh bags. We completed 10 dives between 9am and 6pm. Validation of acoustic estimates of plant cover, height and biomass are challenging even where restrictions on SCUBA diving are not an issue. A core issue is the uncertainty in matching the area censused by the acoustic beam compared to that sampled by independent means, although there are other difficulties as well. Depew (2009) provides a full discussion of validation work and conclusions, which derive from work at several locations additional to Pickering. Based on the frequency of the acoustic beam and associated physical considerations, plant height must be at least 7.5 cm for detection, but this is contingent on the nature of the bottom substrate; highly structured bottom material (e.g. jumbled boulders) sets up return echoes that can interfere with detection or cause false positives. Echoes from fish require informed operator judgement during signal processing if spurious results are to be avoided.

9 Translating height and cover detection thresholds into biomass values and thresholds entails further uncertainty associated with definition of biomass density of the plant canopy. However, given a high degree of cover (approaching 100%) and a relatively simple bottom morphology then the height detection threshold translates to a biomass detection threshold of approximately 50 mg m-3 dry weight. It is generally considered that algal nuisance and shoreline amenity impairment become serious when biomass approaches or exceeds 100 g m-2 (Higgins et al. 2005a, b). The methodology is therefore capable of usefully identifying times and places of problematic algal growth. Calibration against independent biomass measurements (e.g. by divers) shows a significant relationship when biomass is above the effective detection threshold of 50 g m-2. In theory acoustic signals can distinguish between different plant growth forms (e.g. filamentous Cladophora vs more robust macrophyte forms) but for this report we depend on independent observations and samples to interpret the type of benthic plant growth occurring in different areas.

Results 1. Water Quality observations Water quality observations for 2007 were reported previously and the emphasis here is on the additional observations in 2008 and comparisons between the two years. Site-specific comparisons In this complicated study area, one way to better understand the role of local nutrient sources is to select stations for comparison. TRCA/UW site 4100 (adjacent to Duffins Creek) and site 32000 (most removed from any local sources) were selected to provide a measure of tributary input influence. UW site 61000, proximate to the DCWPCP diffuser (sampled only in 2008) was selected as a reference location where higher nutrient concentrations could be expected and indeed were observed. The sites are shown circled on figure 1. We note here that the diffuser is intended to provide extremely rapid dilution and major perturbations of nutrient concentrations over an appreciable area are not expected. However, dilution is not infinite and, depending on prevailing hydrodynamics, limited areas of elevated concentrations can be expected. Perhaps more importantly, the nutrients from the diffuser do enter the study area so even if an area of elevated concentration is not detected the inputs are contributing to nutrient supply in the study region. Modelling work described below is used to characterize the size of this input compared to the total supply available to support algal growth in the study area.

10

Figure1. Sampling stations for 2008 with special comparison stations circled.

Chloride concentrations near Duffins Creek (Figure 2) were relatively high on the first sampling date in 2007 but were lower and similar to the least impacted site for the remainder of the observation period. Over the observation period there was no significant difference between the sites. In 2008, concentrations were variable but were significantly higher near Duffins Creek than at the least impacted site over the observation period. These observations are consistent with a greater influence of Duffins Creek in the wetter year (2008). Concentrations were also high near the diffuser, indicating that our site was close enough to capture the localized effect of its plume on most sampling days.

11

Figure 2. Chloride dynamics at selected stations in 2007 (a) and 2008 (b). Unlike chloride, TSS (Figure 3) was significantly higher near Duffins Creek than at the least impacted site in 2007, although highly variable. It was even more elevated in 2008, consistent with greater runoff in the wet year. The higher TSS near Duffins in 2007, despite the lack of a corresponding elevation of chloride, points to resuspension as an additional influence of importance. The site near the diffuser in 2008 did not detect any significant elevation of TSS.

Figure 3. Total suspended solids dynamics at selected stations in 2007 (a) and 2008 (b). Nitrate (Figure 4) is a useful nutrient tracer because it is associated with tributary sources and is an essential plant nutrient in the lake ecosystem, but it is available in relatively high concentration and so has a degree of conservative behavior as well. In 2007, there was no significant difference in nitrate near Duffins Creek compared to the least impacted site, suggesting the creek had very limited influence on inorganic nutrient variability in the study area. In 2008, there were days of slightly higher nitrate near

12 Duffins but the average concentration was not significantly different from the least impacted site even in this year of higher summer runoff. We were able to detect significantly higher concentrations near the diffuser.

Figure 4. Nitrate dynamics at selected stations in 2007 (a) and 2008 (b) TP was not significantly different near Duffins Creek (Figure 5) compared to the least impacted site in 2007 or 2008, suggesting that the tributary loading and resuspension effects that were evident in the chloride and TSS concentrations were not accompanied by a detectable impact on phosphorus. In both years there were some dates with elevated concentrations near Duffins Creek but these were not sufficiently large or sustained to create a significant difference when averaged across the observation period. Concentrations were higher at the site very close to the diffuser.

Figure 5. Total phosphorus dynamics at selected stations in 2007 (a) and 2008 (b) SRP (Figure 6), like TP, did not differ between the Duffins Creek and the least impacted site. Compared to the Great Lakes Water Quality Agreement target (dashed line), the concentrations at both sites were usually below the target. Concentrations were much higher close to the diffuser, confirming that we were capable of detecting elevated concentrations if they should occur.

13

Figure 6. Soluble reactive phosphorus dynamics at selected stations in 2007 (a) and in 2008 with diffuser site excluded (b) or included (c). Seasonal averages by depth contour interval Another approach to analyzing potential nearshore sources of nutrients is to organize the data into categories based on station depth and thus proximity to shore. This has the advantage of using the power of the larger data set and allows for more powerful statistical tests (ANOVA). In this analysis, the site very close to the diffuser was excluded because its apparent area of influence was small but it exerted high leverage on the statistical results. In the following graphs, solid stars denote significant (p<0.05) seasonal differences, while lightly shaded stars indicate nearly significant (0.15>p>0.05) differences. Broken arrows indicate nearly significant (0.15>p>0.05) and solid arrows significant (p>0.05) differences between depth categories and the direction of the trend. Chloride was higher in spring than summer at the shallower depth contours in 2007, and at some intermediate depths in 2008. These differences can be understood as reflecting the different magnitude and seasonality of discharge from Duffins Creek in the two years. There was some suggestion of a gradient of concentration across depths in 2007 but not 2008. We might have expected higher concentrations in 2008 than 2007, at least in the shallower depth categories, but this was not observed.

14

Figure 7. Chloride averaged by season (spring vs summer) and depth contour for the area. TSS showed significant differences between depth categories in both years, with higher concentrations at shallower sites (Fig. 8). Seasonal differences also occurred, especially in 2008, with higher concentrations in summer than spring. The differences among depths were consistent with expected effect of runoff and resuspension, but higher concentrations in summer, at least at the deeper sites, would seem to imply some additional influence on TSS. Average concentrations did not differ significantly between years.

Figure 8. Total suspended solids averaged by season (spring vs summer) and depth contour for the study area. Nitrate concentrations (Figure 9) did not differ among depth categories in either year, and there was no significant difference between years in average concentrations. Concentrations were systematically lower in summer than spring in both years, and this difference was significant for some depth intervals.

15

Figure 9. Nitrate averaged by season (spring vs summer) and depth contour for the area. Like nitrate, TP (Figure 10) did not differ between depth categories in either year. The average concentration was higher in 2008 than 2007 but the difference was not significant. Concentrations were systematically higher in spring than summer, especially in 2007, and the seasonal differences were significant for some depth intervals.

Figure 10. Total phosphorus averaged by season (spring vs summer) and depth contour for the study area. SRP (Figure 11), like nitrate, showed systematically lower concentrations in summer than spring, although the difference was only significant or nearly significant for some of the depth categories and then only in 2007. There were no significant differences between depth categories. Spring SRP was higher in 2007 than 2008 but the difference was not significant for either spring or summer averages. The somewhat surprising lack of difference between years in essential inorganic nutrients, especially SRP, could be a result of biological demand acting to deplete available nutrients to similar levels

16 regardless of variations in local loading. In this section we use planktonic chlorophyll a as an indicator of one potentially important biological sink for nutrients.

Figure 11. Soluble reactive phosphorus averaged by season (spring vs summer) and depth contour for the study area Chl a was higher in summer than spring in every depth category in 2007 and most depth categories in 2008 (Figure 12), and the difference was significant for some depth intervals in both years. There was no significant difference among depth categories in either year. Spring Chl a concentrations were higher in 2008 than 2007, while summer concentrations were higher in 2007 than 2008. These differences were not, however, significant. The Chl a measurements did not support the idea that higher phytoplankton biomass in summer of 2008 might have prevented the appearance of higher SRP or nitrate concentrations in that summer of high runoff.

Figure 12. Chlorophyll a averaged by season (spring vs summer) and depth contour for the study area.

17

Discharge and water quality comparisons for the study area To summarize the patterns observed in some of the key water quality variables we calculated average concentrations for the study area (excluding the site close to the diffuser and including all sites with station depths of 12 m or less) for each sampling date and in each year. The averages, along with discharge from Duffins Creek, are plotted in Figure 13. The averages are for samples from the top 5 meters but the water column was well-mixed on almost all occasions so the averages are similar to column-averaged concentrations.

Figure 13. Summary of key observations in 2007 (A, C, E and G) and in 2008 (B, D, F, and H), with water quality variables averaged for all sites with depths shallower than 12 m except for the diffuser site (excluded).

18

The higher summer discharge of Duffins Creek in summer of 2008 compared to summer of 2007 is quite apparent. Despite this, there was no corresponding summer increase of SRP, TP or nitrate in 2008. Except for one high TP value at the end of the observation period in 2007, concentrations were very similar (and not significantly different) between years. The high TP value in 2007 obviously did not correspond to any runoff event at Duffins Creek and is not easily explained but almost all other nutrient measurements suggest remarkable similarity between the two years. There was some suggestion of higher TP at the beginning of the observation period in 2008, likely reflecting the somewhat higher spring discharge in 2008 vs 2007 and the earlier start to the observations. This was not, however, reflected in SRP or nitrate. Chl a appeared to follow a different pattern of temporal variation in 2007 than 2008, but average concentrations were not significantly different. Chl a concentrations were generally low, in line with GLWQA targets and not very different from values that might be expected well offshore. 2. Benthic plant surveys Hydroacoustic survey methods for benthic plant growth, and particularly Cladophora, have formed the core subject of the PhD thesis of David Depew. A full discussion of the methodology and its validation is beyond the scope of this report but Depew’s thesis gives a comprehensive treatment. Here we simply note that validation in numerous Great Lakes locations supports the belief that the methodology is capable of detecting Cladophora reliably when biomass approaches or exceeds 50 g m-2. It is generally considered that algal nuisance and shoreline amenity impairment become serious when biomass approaches or exceeds 100 g m-2. The methodology is therefore capable of usefully identifying times and places of problematic algal growth. Calibration against independent biomass measurements (e.g. by divers) shows a significant relationship when biomass is above the effective detection threshold of 50 g m-2. In theory acoustic signals can distinguish between different plant growth forms (e.g. filamentous Cladophora vs more robust macrophyte forms) but for this report we depend on independent observations and samples to interpret the type of benthic plant growth occurring in different areas. The primary products of acoustic signal processing are estimates of plant cover and plant height. Biomass can be estimated from these two quantities if a value for biomass density is known or assumed. Depew’s thesis describes the derivation of biomass density values for Cladophora and these have been used to generate the biomass estimates shown here. As an example of the component measurements, we show measurements collected near the seasonal peak of Cladophora abundance in 2007. Surveys were generally conducted only to the 10-12 meter depth contour. Algal abundance is usually less at greater depths and not readily assessed with this methodology. Plant cover on July 25, 2007 ranged from near zero (purple) to 50 (green) to 100 (red) percent (Fig. 14, below). An extensive area of high cover was observed west of the PNGS, adjacent to Frenchman’s Bay. That area is characterized by predominantly fine

19 grained substrates and independent observations indicated that the plants were mainly vascular macrophytes and not Cladophora. East of PNGS a number of patches of high cover were observed, and these mostly corresponded with hard substrates. Independent observations indicated the plants were mainly Cladophora. The considerable areas of canopy height greater than 20 cm in the Cladophora-dominated parts of the study area were not exceptional compared to other locations with Cladophora fouling problems in the Great Lakes or in Lake Ontario specifically. Canopy heights exceeding 50 cm have been detected and validated by diver observations along the Oakville shoreline of Lake Ontario in 2006 and 2007, for example (Depew 2009).

Figure 14. Benthic plant cover (%) estimated by acoustic surveys, July 25, 2007. Purple is near zero, green is approximately 50, and red is 100% cover

20 The distribution of plant height on July 25, 2007 ranged from near zero (purple) to 40 (blue) to 90 (green) cm (Fig. 15). The area of vascular plant dominance to the west of PNGS had, as might be expected for such plants, a relatively tall canopy. There was also a small patch of vascular macrophytes immediately adjacent to the PNGS intake structure. Elsewhere, the patches of high plant (mostly Cladophora) cover (Fig. 14) were associated with smaller canopy heights in the 20-30 cm range (Fig. 15).

Figure 15. Benthic plant canopy height estimated by acoustic surveys, July 25, 2007. Purple is near zero, blue is approximately 40, and green is approximately 90 cm height. Biomass of benthic plants on July 25, 2007 ranged from near zero (purple) to 90 (blue), 180 (green) and 320 (orange) g m-2 dry weight equivalent (Fig. 16). The high percent cover and canopy height give high predicted biomass in the patches of vascular macrophytes west of PNGS and adjacent to the intake structure. Patches of biomass greater than 100 g m-2 were evident east of PNGS, in areas dominated by Cladophora. Snapshots of biomass at different survey times during 2007 (Fig. 17-21, all on same scale as Fig. 16) give an impression of the seasonal dynamics. We comment here on areas thought to be Cladophora, not vascular macrophyte, dominated and located generally

21 east of PNGS. On June 11 (Fig. 17) only a few small patches had biomass approaching 90 g m-2 (blue). By June 22 (Fig. 18) there were numerous patches with biomass on that order. By July 17 (Fig. 19), small patches of biomass approaching 180 g m-2 were appearing, and they broadened over the subsequent week leading to July 25 (Fig. 16). By Aug. 8 (Fig. 20) some areas began to show a decrease in estimated biomass while others increased. Indications thereafter were of generally decreasing biomass at most locations.

Figure 16. Benthic plant biomass estimated by acoustic surveys, July 26, 2007. Purple is near zero, blue is approximately 90, green is approximately 180 and orange is approximately 320 .g m-2.

22

Figure 17. Benthic plant biomass estimates for June 11, 2007 (same scale as Fig. 16).

Figure 18. Benthic plant biomass estimates for June 22, 2007 (same scale as Fig. 16).

23

Figure 19. Benthic plant biomass estimates for July 17, 2007 (same scale as Fig. 16).

Figure 20. Benthic plant biomass estimates for Aug. 8, 2007 (same scale as Fig. 16).

24

By July 17 (Fig. 19), small patches of biomass approaching 180 g m-2 were appearing, and they broadened over the subsequent week leading to July 25 (Fig. 16). By Aug. 8 (Fig. 20) some areas began to show a decrease in estimated biomass while others increased. Indications thereafter were of generally decreasing biomass at most locations. Observations in 2008 indicated considerably less biomass development than in 2007 (Fig. 21a-c). Except for the area west of PNGS where vascular macrophytes were abundant, estimated biomass was generally near the threshold for reliable detection even in midsummer. Only small, isolated patches surpassed 50 g m-2. Quantification can be difficult even when biomass is more than 50 g m-2 so these surveys do not provide a very robust estimate for biomass. Diver surveys conducted by OME in 2008 indicated considerable Cladophora biomass but we do not have those data for comparison as yet. A variety of evidence, including these acoustic surveys, points to lower algal biomass in 2008 than 2007 but quantifying the difference is at present difficult.

Figure 21a Benthic plant biomass estimates for July 7, 2008.

25

Figure 21b Benthic plant biomass estimates for July 28, 2008.

Figure 21c Benthic plant biomass estimates for August 20, 2008.

26

3. Model Simulations for 2007 and 2008 with ELCOM-CAEDYM + CGM Met forcing Figure 22 & 23 shows the met forcing data used for the 2007 and 2008 simulations as time series with hourly values. In order to compare the main meteorological differences between the two years, daily averages for temperature, wind speed/direction and daily incoming solar radiation were calculated. Figure 24 shows a comparison plot for the main weather drivers. Table 1 presents a summary, which includes a splitting of springsummer values, where it can be seen that 2008 was not only a wetter year than 2007, but was also colder and cloudier.

Figure 22. Meteorological forcing used in Lake Ontario for the 2007 model simulations (top to bottom: air temperature, wind speed/direction, solar radiation and humidity)

27

Figure 23. Meteorological forcing used in Lake Ontario for the 2008 model simulations (top to bottom: air temperature, wind speed/direction, solar radiation and humidity)

Table 1. Main Met Forcing (2007 vs. 2008) Air Temperature (°C) spring summer Wind Speed (m/s) spring summer Solar Radiation (W/m2) spring summer

2007 15.8 12.9 19.7 4.0 3.4 4.0 309.4 367.9 324.4

2008 14.2 8.6 19.5 4.2 4.1 3.6 245.9 259.8 307.5

Diff 07-08 (1.5) (4.4) (0.1) 0.2 0.7 (0.4) (63.4) (108.1) (16.9)

28

Figure 24. Meteorological yearly comparison: dry-2007 vs. wet-2008 in Lake Ontario (top to bottom: air temperature, wind speed and solar radiation)

Model Setup As in previous runs, the model setup with a 2x2 km grid resolution, 50 vertical layers and a time step of 5 minutes was used for the 2007 and 2008 simulations. Animations were created for temperatures, water quality parameters and main circulation. The links to the animations are also in the web site in WMV (Windows Media Video format): http://www.science.uwaterloo.ca/~lfleonvi/ontmodel/movie_page.html.

29 Figures 25 and 26 show the mean circulation patterns for the two years at a lakewide scale and for the nearshore domain. It can be noted that mean circulation features, such as the main gyros around the Pickering plant, are quite similar for both simulations, which can be expected as the main climate drivers (in particular wind speed and direction) are not that much different between the two years.

Figure 25. Surface mean circulation patterns in Lake Ontario (top 2007, bottom 2008).

30

Figure 26. Mean circulation patterns in Lake Ontario nearshore domain in the vicinity of Pickering (top 2007, bottom 2008)

Thermal structure & validation with Thermistor-Chain Data In general the thermal structure is properly modeled as was shown with the 2004-2005 simulations presented in previous reports. As an example for the 2007 simulation, Figure 27 shows three calculated temperature profiles for sites located arbitrarily in the west, central and east regions of Lake Ontario. From the data collected in the nearshore region of the PNGS, two of the seven thermistor-chains are located in offshore waters (deeper than 25m). The output file of the model was modified to include these two locations as additional profile grid cells. Figure 28 shows the comparison between measured and calculated profiles for TC1 and TC2 (the two offshore sites). As expected, the model performs quite well and the evident presence of upwelling of cold water is reproduced in the model. Taking into account that ELCOM is being revised for the mixing scheme, this result, despite some underestimation of the temperature values, is consistent with previous runs (i.e. 2004-05).

31

Figure 27. Calculated profiles for 2007 at different locations in Lake Ontario

32

Temperatures (°C) : Measured vs Calculated : (2m depth) 28 LONT TC2 : 2007 ELCOM @ (24,31) LONT TC4 : 2007 24

ELCOM @ (13,68) LONT TC6 : 2007 ELCOM @ (28,40)

20

16

12

8

4 121

152

183

214 Day of the year

245

276

Figure 28. Temperature profiles and time series comparisons at selected TC sites

33 - Description of Loads and Scenarios Once the model is properly setup, it can be used to analyze influences on Cladophora growth from different conditions, including any water treatment plant outfall and thermal discharges. A full simulation for 2007 removing the discharge from the Duffin Creek Water Pollution Control Plant (WPCP) and a second scenario assuming ambient water temperature discharges from the Pickering Nuclear Generating Station (PNGS) were prepared to compare Cladophora growth on the vicinity of the plant and the relative effect for the different scenarios. 2007a – refers to the 2007 simulation with no flow/load discharges from the diffuser of the WPCP. It is worth to mention that the flow and therefore load contributions from Duffins Cr. in 2007 were, for practical purposes, negligible in comparison with 2008.. 2007b – refers to the 2007 simulation including flow/load discharges from the WPCP and Duffins Cr. (same as above). 2007c – refers to the loading conditions of the 2007b simulation, but with the thermal effluent from the PNGS using ambient temperatures estimated from the lake wide run, thus eliminating the temperature plume from the plant effluents. 2008 – refers to the 2008 simulation including flow/load discharges from the WPCP and Duffins Cr.

Figure 29. Watershed and WPCP inflows considered in the simulations for 2007 & 2008 (simulation period in between green brackets).

34

When available, daily flows for the stream gauge station 02HC049 Duffins Creek at Ajax were obtained from the Water Survey of Canada (http://www.wsc.ec.gc.ca/) from the section Data Products & Services or calculated from river levels and rating curves in collaboration with MTRCA. Effluents from the WPCP and the nuclear facility were provided by the Duffins Creek WPCP Operational Data (Region of Durham) and Ontario Power Generation (OPG) respectively. Figure 29 shows the time-series for the flows used in the 2007 and 2008 simulations. At the time of the simulations, operational data for the WPCP was not available for 2008, so the 2007 discharge data was used in both years. Loads were calculated using event mean concentrations from historic data provided by MTRCA, from the longer term provincial water quality monitoring network for Duffins, combined with the regular monthly water quality field data collection using a composite automatic sampler in 2007 & 2008. Figures 30 to 34 show mean concentrations and daily loads for TP, SRP, NO3, NH4 and TSS used in the simulations.

Figure 30. Watershed & WPCP discharges for TP

35

Figure 31. Watershed & WPCP discharges for SRP

Monthly Mean Concentrations @ Duffins Creek & WPCP

100.00

Duffins07

Duffins08

WPCP

NO3 (mg/l)

10.00 1.00 0.10 0.01 jan

feb

mar

apr

may

jun

Month

jul

aug

sep

oct

nov

dec

Daily Loads @ Duffins Creek & WPCP

10000.00

NO3 (kg/day)

1000.00

100.00

10.00

1.00 Jan

Feb

Mar

Apr 2007

May

Jun

Jul 2008

Aug

Sep

Oct WPCP

Figure 32. Watershed & WPCP discharges for NO3

Nov

Dec

Jan

36 Monthly Mean Concentrations @ Duffins Creek & WPCP

100.00

Duffins07

Duffins08

WPCP

NH4 (mg/l)

10.00 1.00 0.10 0.01 jan

feb

mar

apr

may

jun

Month

jul

aug

sep

oct

nov

dec

Daily Loads @ Duffins Creek & WPCP

10000.00

NH4 (kg/day)

1000.00

100.00

10.00

1.00 Jan

Feb

Mar

Apr

May

Jun

2007

Jul

Aug

Sep

Oct

2008

Nov

Dec

Jan

WPCP

Figure 33. Watershed & WPCP discharges for NH4

Monthly Mean Concentrations @ Duffins Creek & WPCP

10000.0

Duffins07

Duffins08

WPCP

TSS (mg/l)

1000.0 100.0 10.0 1.0 jan

feb

mar

apr

may

jun

Month

jul

aug

sep

oct

nov

dec

Daily Loads @ Duffins Creek & WPCP

10,000,000

TSS (kg/day)

1,000,000 100,000 10,000 1,000 100 Jan

Feb

Mar

Apr 2007

May

Jun

Jul 2008

Aug

Sep

Oct

WPCP

Figure 34. Watershed & WPCP discharges for TSS

Nov

Dec

Jan

37 As a summary, figure 35 shows the total yearly values and the percentage attributable to each of the sources. For all the parameters, except for suspended solids, the WPCP contributions range from 65 to 90+ percent of the total yearly load. In contrast, for TSS, the discharge from Duffins Cr. in 2008 accounts for almost 85% of the total sediment yield. High values of TSS result in higher turbidity and less light penetration.

1550

500

1682

285 103

40

284 102

Figure 35. Total yearly loads and percentage of the distribution according to the source.

30 400

µg/l

20

300 µg/l 200

10

0

100 WPCP

Duff-2008

Lake-2008*

Duff-2007

Lake-2007*

TP [µg/l]

1.195

0.182

11.740

0.031

10.560

SRP [µg/l]

0.597

0.020

1.779

0.003

1.921

NO3 [µg/l]

5.55

1.88

284.64

0.86

283.58

NH4 [µg/l]

5.39

0.14

103.27

0.05

102.36

0

TSS [mg/l]

WPCP

Duff2008

Lake2008*

Duff2007

Lake2007*

29

340

1682

24

1550

Table 2–Figure 36. Equivalent concentrations for different sources. To establish a reference framework and in order to have some perspective of magnitudes and the importance of each source’s contributions to the nearshore area in the vicinity of the Pickering nuclear facility, the total loads were used to estimate total concentrations by considering a nearshore domain volume of 70,128,000,000 m3. Table 2-Fig.36 shows the equivalent concentrations in the nearshore for the different sources. This volumetric conversion allows for some interesting analysis to identify some of the environmental

38 influences of the different sources (e.g. sewage diffuser) in the study area. For example the yearly TP load from the WPCP yields a concentration of 1.2µg/l, which represents at most ±10% of the equivalent range for in-lake concentrations in 2007-08 (10.6-11.7µg/l). Equivalent concentrations coming down from Duffins Cr., despite that being larger in 2008 than in 2007, are much smaller in comparison the WPCP values. Nitrate and TSS contributions from the local sources (Duffins and WPCP) are also small compared to lake concentrations. The SRP contribution from the WPCP accounts for less than one third of the equivalent in lake concentrations. However, SRP cycles much more rapidly so this mode of comparison does not give an accurate impression of how SRP loads from the WPCP compare to the supply from lake exchange processes. For such comparisons, we must turn to the model, which can account for advection, dilution and cycling processes more accurately. Considering on how to simulate the discharge of the WPCP in the nearshore model, the model calls for input of the proper loading in order to achieve the best spatial distribution of the calculated parameters that will drive the algal growth. In principle, the model setup was not designed for a detailed plume study. However when initially faced with the question of how to include the diffuser discharge we did had access to previous modeling results which faced exactly with that issue, in particular the use of a near field diffuser model CORMIX (CH2M, 2005). Based on such results and in consultation with CWR, we expect a good proxy for the diffuser discharge will be to release the effluent in the model at a layer (9m depth) that produces, with its thickness, a dilution factor close to the diffuser designed ratio of 20:1 in the near field. In principle it might be possible that by releasing at a volumetric cell in the model could miss some additional dilution due to the turbulent effect of the discharge in the first few meters. It is worth noting that in preliminary tests we simulated near filed uniform release (using an average 20:1 dilution ratio as the "minimum for effective dispersal of municipal effluents" - MOE) at all vertical layers, taking care to conserve mass (total load), which by the way is what CORMIX approximately predicts to be achieved in the near field (first 100-200m - but have to note that a main assumption is that the receiving water velocity is constant, which might get compromised by the active dynamic domain that includes frequent current reversals, calm periods, etc.). With this kind of vertical uniform release the resulting "plumes" were even higher than the ones we are actually getting with the selected release just in a volumetric cell, verifying that provides the same 20:1 dilution ratio instantly, and let the hydrodynamic code take over, including any temperature buoyancy effects. Additionally, in contrast to the above plume study, that didn't have available any field water quality data, we have samples at a site near the diffuser. For example in Fig. 6 around day 175 (late June), when the WPCP release rate of SRP for June is 340µg/l, SRP is 19µg/l near the diffuser giving a dilution ratio of 17.8 (and the site might be a bit farther than the 'near field domain'). This supports the fact that the 20:1 we are inducing at the release is adequate, or even a bit more conservative than initially thought. The model predicts SRP vertical integrated concentrations for that site in late June on the range of 0.015 to 0.016mg/l.

39

- Validation with Water Quality Data In order to verify the water quality component, the field samples at TC2 were used to compare against model output for the lake wide (2km) run. Plots are shown in Figure 37 where the measured water quality parameters (TChla, TP, SRP, TN, SiO2, TSS) are compared against the same state variables calculated with ELCOM-CAEDYM.

Figure 37. Water quality comparison at TC2 (lake wide run) for 2007

40

Figure 38. Water quality comparison at uw2-3 station (nearshore domain) for 2007

From the output plots, we conclude that the model performs quite well for most of the variables especially SRP and TSS that will drive the CGM as they determine the availability of the critical limiting nutrient for Cladophora and the transparency of the water which determines light availability to this benthic plant. Some fine-tuning of initial conditions for SRP and TSS was required to better calibrate the model output. At the same time, the offshore TC2 site is also a Time Series point for ELCOM. Unfortunately, the deep TC2 site was dropped from the field campaign for 2008, so there is not an equivalent plot for the offshore output in 2008 (this will be compensated by comparing time-series plots for both years versus the depth averages for all the sample sites presented in the data section of this report).

41

With the output from the nearshore water quality simulations (2007a with no diffuser and 2007b with the WPCP discharge), time series at the closest site to the diffuser (site uw23) were extracted for the main WQ parameters. Figure 38 shows that estimated values are usually in the same range as the measurements, and the seasonal dynamics of Chl a, SRP and silicate are largely captured. Disagreements also occur, mainly late in the series and after the main season of Cladophora growth. Results are similar between scenarios (i.e. plus or minus diffuser), but the simulation with the diffuser included shows higher variability which can be attributed to the passage of the plume being advected back and forth around the sampling station as can be seen in figure 39 (this effect is less noticeable in sites farther away from the diffuser).

Figure 39. Simulated and measured SRP in sites near and far from the diffuser discharge. Additional comparisons between 2007 & 2008 were presented in the field data section from this report. Using the same plot, the time-series model output was extracted for the two runs (both including diffuser discharge, so the output will be from the 2007b and 2008 simulations). Figure 40 shows the overlay of the modeled output using the field data plot (Figure 13). This comparison again shows that the model is producing results in the correct range, especially for the key variables TP and SRP and bearing in mind that measurements at such low SRP concentrations are subject to some analytical uncertainty.

42

2007

2008

Figure 40. Nearshore time-series model output: left-2007, right-2008; lines: TP (blue), SRP (red), Chla (green) and NO3 (purple). Comparison against averages from sites within the 2 to 12 m contour in study area – plot borrowed from field data section. - Cladophera Growth Simulations As mentioned above, in order to evaluate the local effect of the WPCP diffuser, output was prepared, as spatial snapshots in the nearshore domain, for the 2 different scenarios: 2007a & 2007b. Figure 41 shows an earlier onset of predicted growth with the diffuser as well as patches of elevated SRP. Figure 33 shows that predicted algal biomass reaches similar but somewhat more extensive levels with the diffuser later in the season.

43

Figure 41. Cladophora growth model output including T, ExtC and SRP panels for Jun 11th, 2007. Left 2007a simulation (no diffuser); Right 2007b simulation (diffuser).

Additionally in Figure 42, the snapshot for Jul 17th shows the acoustic estimates of Cladophora and a panel for SRP showing the extent of the plume in that particular day. We note that despite the fact that this study did not contemplate any simulation of the diffuser near field, in the SRP panel in the figure, with an estimated average discharge of SRP (see loads section) from the diffuser in the range of 0.8mg/l and considering the 20:1 dilution ratio, the highest expected cell value is 0.04mg/l which is consistent with the range shown in the vertical integrated values plot for SRP [0-0.025 mg/l].

44

Figure 42. Cladophora growth model output including acoustic surveys estimates for Jul 17th, 2007. Top model output panel 2007a simulation (no diffuser); Bottom model output panel - 2007b simulation (diffuser).

Site-specific estimates from the acoustic surveys were extracted for all the available sampling dates and compared with the model time series output at those sites. Figure 43 and 44 show this comparison for transects 2 and 4. Assuming that the hydroacoustic survey estimates are reliable, it seems that the model is capturing the variations of algal growth among depth contours quite well. The predicted biomass time series are mostly in the right range but may be showing different dynamics from the observations. In particular, the model predicts sloughing and re-growth at some sites that is not clearly supported by the surveys, but it is possible that the limited temporal resolutions of the surveys simply failed to document the dynamics properly. As for the contrasting differences in growth between 2007 and 2008, figures 45 to 48 compare a sequence of different time snapshots (early, mid and late season). Care was taken to extract midday values and the colder-cloudier 2008 shows in all the plots as a noticeable decrease in bottom PAR values.

45

Figure 43. ELCD+CGM Nearshore (Clad @ transect 2)

Figure 44. ELCD+CGM Nearshore (Clad @ transect 4)

46

2007

2008

Figure 45. Comparison of Cladophora growth between 2007 & 2008 (14-Jun, early season).

47

2007

2008

Figure 46. Comparison of Cladophora growth between 2007 & 2008 (21 Jun- mid season).

48

2007

2008

Figure 47. Comparison of Cladophora growth between 2007 & 2008 (31 Jul-mid season).

49

2007

2008

Figure 48. Comparison of Cladophora growth between 2007 & 2008 (10 Aug-late season).

Anecdotal evidence indicates that a large Cladophora detachment event occurred in the area near Pickering in Aug 9, 2007. Analysis of the model output around the time of the event, shows a remarkable coincidence in the timing of the algal detachment with the simultaneous occurrence of favourable strong currents that might transport the detached algae to PNGS. Figure 49 shows the mean circulation patterns at the time of the event, it can be seen that a significant east to west current is present the day of the event in contrast with very calm waters prior and after the sloughing event. Figure 50 shows the mean biomass levels estimated with the model for a few days before the event compared with what biomass was left few days after.

50

Figure 49. Mean circulation prior, during and after the sloughing event of Aug 9, 2007

51

Figure 50. Cladophora growth (mean values at the bottom layer) prior and after the sloughing event of Aug 9, 2007

52 Further output post-processing was required in the attempt to arrive at more objective evaluations and comparisons of the different scenarios. So far, considering that the sloughing and transport portion of the model needs further research and validation to be able to follow the cloud of detached algae, we will based our comparisons on the growth of Cladophora, evaluated as estimated mean concentrations and maximum reaches within the study domain. This will give comprehensive estimates of Cladophora growth in the vicinity of the PNGS and verify that the amount growing in the vicinity of the PNP is enough to explain the amount of biomass collected at the cooling water intake. Figure 51 is a time series of the dry weight mass collected at the screen house of plant B at Pickering. The total yields for 2007 & 2008 are 63 and 22 metric tons respectively. Figure 52 shows mean biomass from the field acoustic survey and Figure 53 the modeled means for the two years. Dry weight estimates at screenhouse "B"

30

120

Dry Mass (tons)

100

95

20

80

15

62

61

53

10

63

56

60 40

37

36

22

5 0 1999

20

Yearly total (tons)

102

25

0 2000

2001

2002

2003

2004

2005

2006

2007

2008

OPG Algal_Bins (1999-2008)

Figure 51. Dry weight mass collected at the screen house of plant B at Pickering

2007 Mean Biomass @ Pickering from acoustic survey

Biomass (gD/m2)

350

357

40

300 250

30

200 20

150 100

10

50 0

0 11-Jun

22-Jun

17-Jul

25-Jul

David Depew et al. 2007

Figure 52. Dry weight mass estimated at Pickering from acoustic survey

8-Aug

CumBioMass (tons DW)

400

50

53

Figure 53. Cladophora Growth (2007 & 2008 simulated means)

54 To evaluate the effect of the temperature discharge from the nuclear plant the additional scenario output in Figure 54 compares the base case simulation (2007b) against ambient temperature discharge from Pickering (2007c). Results show exactly the same growth and the only noticeable difference is in the averaged temperature panel.

Figure 54. Cladophora Growth (2007b vs. 2007c simulated means)

55 From the model estimates and using mean values and transect extractions from the spatial distribution, it is possible to evaluate total mean growth in the domain area for the time period simulated. Figure 55 shows transect mean values for the four tested scenarios and the total amounts estimated from the spatial distribution. It can be noted that for the 2007c thermal simulation the total amount of 387 gDM/m2 overlaps with the 2007b result, which is practically the same as the 384 gDM/m2. . Clad_Means (distance from west to east) 175 2007a

150

2007b

gDM/m2

125

2007c 2008

100 75 50 25 0 0

1

2

3

4 5 km (from W to E)

6

7

8

9

Clad_totals (distance from west to east) 20 2007a 2007b

tons DM

15

2007c 2008

10

5

0 0

1

2

3

4 5 km (from W to E)

6

7

8

Model Estimated Mean Growth (East of Pickering)

CumBioMass (tons DW)

Figure 55 Transect means and total growth values for the four tested scenarios

9

56

Synthesis and Conclusions Objective 1: Determine the contribution of local nutrient sources to Cladophora growth near PNGS. The project has revealed some influence of certain local sources on water quality and Cladophora growth but it failed to identify catchment runoff as a significant influence or to show that current levels of loading from DCWPCP might be the key to nuisance growth of Cladophora. Our model predicted that algal growth to high levels could be supported without any local inputs (e.g. in 2007). Observations and modeling in 2008 also showed that Cladophora growth could be reduced below nuisance levels even when local catchment inputs were increased and loadings from DCWPCP were maintained. These results point to the importance of in-lake nutrient sources and of other (nonnutrient) aspects of the environment in regulating nuisance algae. This project has explicitly examined only Duffins Creek and the DCWPCP as local sources, but these are believed to be the dominant ones. Data on contributions from storm drains made available from TRCA showed that on the scale of the study area, the inputs were even smaller than those from Duffins Creek. While there could be differences in quality of P or other nutrients arriving from such sources, compared to the creek or the diffuser, this would seem unlikely to make up for the much smaller total magnitude of storm drain loading. Tributaries other than Duffins Creek are at the extreme margins of the study area or outside it. DCWPCP proved to be a much larger source than Duffins Creek and would still be dominant even if one or two additional tributaries were included. We believe we have included sufficient of the local sources to assess their importance. Furthermore, the field observation program was designed to seek evidence of nearshore enrichment and should reveal the influence of sources that might be affecting the study area. Future work could use the modeling approach developed and validated here to encompass additional sources if that seemed to be a priority. The summer of 2007 was very dry so local runoff effects should have been minimal. By contrast, 2008 had record rainfall for the month of July and a very wet summer overall. If local catchment runoff were important in stimulating higher nutrient concentrations we should have seen higher concentrations in 2008 than 2007 and a stronger gradient of concentrations with proximity to shore. Neither was observed, and evidence for gradients in concentration with proximity to shore was largely lacking in both years. Nutrient and Chl a concentrations were largely in a similar range to offshore waters, albeit somewhat higher in spring, while P and Chl a concentrations mostly met the GLWQA targets for Lake Ontario. Depending on the mode of data analysis, the nutrient input from DCWPCP could have enriched waters at intermediate distances from shore and concealed a gradient in catchment contributions. By isolating stations for comparison, based on their proximity to the creek, the diffuser, or neither, we sought to test this possibility. We still did not see convincing evidence for any substantial nutrient enrichment from Duffins Creek. The DCWPCP diffuser, by contrast, was easily detectable as a feature in the nutrient distribution field but dilution is rapid with distance from the diffuser. At the

57 same time, some of the apparent dilution can, in the case of limiting nutrients like P, reflect biological utilization. Analysis based purely on observation of spatial distributions cannot truly resolve the role played by individual point sources in the highly dynamic lake environment. The coupled ELCD-CGM model is capable of predicting the role played by DCWPCP and Duffins Creek because it takes account of the simultaneous processes of advection, dilution (i.e. mixing) and biological utilization. However its predictions are only as credible as far as the model validation and our acceptance of its logical structure extend. Perfect agreement between model simulations and observations can never be expected, for many reasons, even if the model is quite correct. In the present case, agreement with the WQ variables most important to Cladophora (SRP, TP, light penetration and temperature) was very good. Little tuning of parameters was done when implementing ELCD for this study site; values derived from previous work in Lake Erie were used with little modification. This, and the fact that the model appeared to agree reasonably well with WQ observations in both 2007 and 2008, without any parameter adjustment between years, suggests that the ELCD predictions are reasonably robust. The hydrodynamic predictions of ELCD have also proven themselves in a variety of comparisons against temperature and current velocity measurements both in the current study and a similar one in Lake Erie. As a tool for predicting the environmental conditions for algal growth, and analyzing the role of individual inputs, we believe ELCD has proved to be reasonably reliable. The coupled ELCD-CGM model is to our knowledge still unique among environmental models in using the open boundary approach and driving a spatially-resolved model for benthic algae growth. Validation of its predictions of Cladophora dynamics is a much more difficult challenge than validating the WQ predictions of ELCD because quantitative measurements of benthic algae are arduous and imprecise. Measurements are prone to large sampling error because Cladophora is typical of many benthic organisms in having a very patchy distribution, and physical collection from the hard substrates it favours is not easy to do in a fully quantitative and reproducible manner. The acoustic methods that we mainly relied on overcome some of these problems but raise others, notably the difficult problem of translating acoustic estimates of cover and canopy height into corresponding biomass values. We therefore have to be cautious in making any claims about the accuracy of the ELCD-CGM predictions. We can say that the model captures the major patterns of seasonal, inter-annual and depth related variability as far as we can discern them, and any evidence for differences in the details of the dynamics is inconclusive in light of the uncertainties associated with the observations. It was encouraging that an entirely independent index of algal abundance in the area (algal impingement at PNGS) supported both our model and observations in their conclusion that algal biomass was much less in 2008 than 2007. Of the local influences examined (thermal effluent, Duffins Creek, and DCWPCP) only DCWPCP had an appreciable effect on seasonal accumulation of algal biomass in the study area according to the model. Removing its influence decreased the predicted biomass accumulation by about 10% in 2007, a year of severe algal fouling. This would

58 still leave us in the range associated with problems of impingement and shoreline fouling. By contrast, predicted biomass accumulation was about 30% lower in 2008 than 2007 even with the inputs from DCWPCP and the elevated inputs from Duffins Creek. While truly local nutrient sources were not identified as the key to understanding problematic algal growth in the study area, the circulation patterns indicated by the hydrodynamic model are important to consider. The simulations reflect the reality that the coastal boundary layer of the lake is dominated by along-shore flows and often has limited exchange with truly offshore waters (Rao and Schwab 2007). Within our study area, the thermal effluent of the PNGS has an additional steering effect on circulation that likely helps promote local detention of along-shore transport. In this study, we lacked true offshore reference points that could be used with confidence to assess how different the coastal boundary waters that we sampled might be from true offshore waters. However, autumn observations did reveal SRP concentrations that were very high compared to typical offshore concentrations and indeed quite high compared to expectations for oligotrophic lake systems. These observations were made in both 2007 and 2008. They are consistent with an enrichment of the coastal boundary layer along the north shore of Lake Ontario, resulting from many inputs east and west of the study area and not just within it. We would suggest that the enrichment is important in supporting algal growth and is manifested clearly once the intense biological nutrient demand of spring and summer is relaxed by the declining light and temperature of autumn. For most of the observational period it is masked by the seasonal sequestration of nutrients in Cladophora and other biota (e.g. dreissenid mussels). The in-lake nutrient supply that emerges as the dominant consideration in our modeling of algal growth, is not necessarily from truly offshore sources, but may rather derive from medium to long distance transport along the northern coastal zone. This hypothesis is amenable to further investigation but cannot be properly tested with the observations collected in this study. Objective 2: Assess the contribution of locally-grown algae to impingement at PNGS Subject to the uncertainties in modeling and observations noted above, both simulations and measurements of algal distributions indicated a substantial area of high seasonal Cladophora accumulation east of the PNGS. Estimated production of algal biomass in the area was far greater than the mass of material impinging at PNGS. Allowing that only some fraction of the impinging material is actually Cladophora it would appear that the potential for local production is ample to account for the amount appearing at PNGS. Analysis of the largest impingement event in 2007 indicated a coincidence in timing of algal detachment in the area proximate to PNGS, and simultaneous occurrence of favourable currents to conduct the material to PNGS. These observations do not conclusively establish that algal impingement at PNGS is, in general, the product of local growth but they strongly suggest that there is no need to look further afield to explain the problems.

59 The circulation patterns at the time of the 2007 impingement event analyzed here were not unusual for the study area, suggesting that suitable currents frequently exist to transport local algae to PNGS. The biomass levels estimated for the study are also not unusual for the north shore of Lake Ontario, suggesting that development of sufficient local algae can be expected routinely. Objective 3: Assess the role of the thermal bar This study was not able to directly study the effects of this important physical phenomenon because the feature had moved outside the study area before the observational program began. It can be concluded that the bar could not act to confine local inputs any more tightly than the outer bounds of the study area during our season of observation. It might have exerted a confining effect on a somewhat broader coastal band and we might then expect a marked spring elevation in WQ variables associated with runoff and effluents. This was not obvious in our data. Spring time concentrations were higher than summer for many variables, but this is a rather normal seasonal pattern in temperate lakes even where a thermal bar is not having any effect. The spring concentrations were not greatly elevated over typical offshore concentrations and P was not in the range normally associated with eutrophication problems. The bar may nonetheless play a role by helping to confine inputs closer to shore earlier in the season. This would be earlier than the period of high Cladophora nutrient demand so if it were fueling the growth of the benthic algae we would need to envisage some type of storage mechanism. Such a mechanism might be provided by particle sedimentation, likely aided by dreissenid mussel filtration. Deposition/filtration of particulates from the water column could enrich the benthic habitat. Nutrients released through subsequent decomposition of such material could help fuel growth of Cladophora when light and temperature conditions improve later in the season, and the algae have developed a significant storage capacity in their own tissues. The processes would have to work in such a way that the particulates deposited very early in the season are retained long enough in the nearshore that their nutrients would still be available later on, when the algae start to grow. Dreissenid mussels could be important in such detention of particles and nutrients. The present study is unable to provide insight into this possibility but the ELCD-CGM model is amenable to further development that would allow us to test how such a mechanism might work. Objective 4: Potential control options The results of this study strongly suggest that further restriction of local nutrient inputs to the study area would not eliminate Cladophora problems. This is a somewhat disappointing but not completely surprising conclusion. Cladophora is a problem along many shorelines, including ones that have no point source or tributary inputs nearby. An extreme example is an offshore shoal in eastern Lake Ontario studied by D. Depew. Luxuriant algal biomass developed there despite a complete lack of local tributaries or outfalls (D. Depew PhD thesis, U. Waterloo). However, Cladophora does not reach nuisance levels everywhere in the Great Lakes or comparable waters, suggesting that

60 there might be some measures that could lead to improvements. Further work would be needed to evaluate the possibilities, and some of these are considered in the Recommendations section below. It is also possible that the failure of the current model to incorporate some mechanisms that might be important, such as dreissenid mussel activity and the thermal bar, has led us to underestimate the potential for improvements through restrictions of local nutrient inputs. The model appears to perform well but the general agreement with observations may conceal a significant inaccuracy in the functional relationships. Evaluating this possibility is again a matter for further study. If problems cannot be eliminated through further control of local nutrient inputs, it would still appear prudent to consider carefully before increasing any inputs. The model predicts that algal biomass accumulation is sensitive to P inputs, that is, the algal biomass development is to some extent P-limited. Other factors must be at play as well, because 2008 had much less algae despite similar or slightly higher (due to the relatively small Duffins Creek and storm drain sources) inputs than in 2007. Although algal problems have been chronic in the area for the past decade there has been substantial inter-annual variation with 2008 being exceptionally lacking in algal problems. In the scope of this study we cannot diagnose the causes for these variations but they do suggest that in some years there is potential for more algae to develop if the nutrient supply is sufficient. The mechanisms associated with some of the spatial variations of algal biomass, and of the major difference between 2007 and 2008, are difficult to quantify with the present results and scope of work. Similar modeling work in Lake Erie (Higgins et al. 2006) has shown that seasonal patterns of algal growth are strongly influenced by temperature, while depth-related variations are strongly related to light availability. Temperatures and incident PAR were lower in 2008 than 2007 and both could work to reduce algal biomass accumulation. Some fine-scale variation in biomass in temporal snapshots was likely due to differences in the onset of sloughing. However, disentangling the joint influences of light and temperature in the context of the inter-annual differences observed here would entail a sophisticated sensitivity analysis that is beyond the terms of reference for this project. Likewise, a detailed analysis of the sensitivity of sloughing times would require a fairly extensive set of model simulations and analysis of large amounts of model output. Such analyses are feasible and would be valuable but would entail new work. We have concluded that restriction of nutrient inputs in the study area holds little promise of eliminating the algal problem because in-lake sources are sufficient to produce problematic growth. At the same time, P concentrations in Lake Ontario are already relatively low and prospects for lowering them further may seem poor. Prospects for management of the algal problem through control of P may therefore not seem encouraging. However, we did find that elimination of the DCWPCP inputs (admittedly a radical scenario) did decrease predicted seasonal accumulation of algal biomass by about one third of the estimated difference between 2007 and 2008. That is, removal of the diffuser input decreased predicted biomass by 10% whereas predicted biomass in 2008 was about 30% lower than in 2007. This diffuser effect was predicted even though we estimated that DCWPCP contributed only a small fraction of the total P in the study area. The vast majority of the P is supplied through lake exchange, but the predominant

61 pattern of circulation and exchange in large lakes (including Lake Ontario) is along-shore rather than cross-shore. We might then ask what the implications of a broad campaign to further reduce P inputs to the northern coastal zone of Lake Ontario as a whole might be. Effective elimination of the algal problem at PNGS may not require an extreme reduction in algal biomass; modeled biomass accumulation in 2008 was only 30% less than in 2007 but resulted in essentially no impingement problems at PNGS. This may partly reflect fortuitous timing of algal detachment and contemporary circulation patterns in 2008 that could have spared PNGS, but anecdotal reports suggest that shoreline fouling over extensive areas was much less severe in 2008. The available evidence seems to suggest that the apparently modest reduction of algal biomass in 2008 was sufficient to alleviate the worst fouling problems. Considering as well that other lakes with superficially similar nearshore P concentrations (e.g. central Lake Huron, Lake Simcoe) have little Cladophora problem, it may not take a dramatic reduction in P inputs and in-lake concentrations to have a useful effect.

Recommendations The ELCD-CGM model appears to be a powerful and unique tool for analyzing the provenance of, and controls on, nuisance Cladophora. The model and further improvements upon it seem to have great potential for application to a number of initiatives that could improve, or at least limit, the algal fouling situation at PNGS and nearby shorelines. 1. Proposed or inadvertent changes of P inputs into the study area from external sources (tributaries, drains, wastewater outfalls) should receive careful study to ensure they do not exacerbate the Cladophora problem in this P-limited system. ELCD-CGM seems like a valuable device for such assessments. 2. The potential for Cladophora control via further restriction of P loading into the northern coastal zone of Lake Ontario, not just the current study area, should be evaluated. If the algal problems in the study area are indeed a regional rather than local issue, and we would suggest this to be the case, then information needs to be amassed to support a call for regional-scale action. The model framework developed here provides a suitable tool for assessing algal responses to altered levels of in-lake P availability. An approximate assessment of the potential for success with regional-scale loading changes could be obtained fairly quickly by forcing the model with results from a whole-lake, mixed reactor, model to estimate how in-lake P concentrations might change in response to changes in loading. Even more simply, it could be useful to do a simple sensitivity analysis with arbitrary, hypothetical, in-lake concentrations. A truly satisfactory answer, strong enough to support advocacy for regulatory action, would really need to resolve the spatial variability in the lake and analyze the current loads to the lake comprehensively. This would involve a significant amount of work to incorporate many more individual sources of loading in the model. Model runs would need to be extended to estimate the ultimate effect of loading changes. Such work seems entirely feasible but would constitute a new and substantial research project.

62

3. Comparative studies, including application of this modeling approach, should be undertaken at comparable sites that do not currently have a Cladophora problem to assess whether differences in P supply are sufficient to explain why algae do not reach problem levels and to examine possible alternative controls. Suitable sites can be found in Lakes Huron and Simcoe. Significant work would need to be done to set up the model for those additional sites. Much of the necessary work is planned or underway at Lake Simcoe as part of other research projects. An initiative along these lines could benefit from collaboration with those other projects. This work should include an examination of the bioavailability of P. Conventional measures (TP, SRP) may conceal important variations in the effectiveness of loading from different sources. 4. Further development of ELCD-CGM to include dreissenid mussels should be pursued in order to make the mechanistic basis of the model more authentic. In its current form the model uses simple formulations to describe removal of phytoplankton and other particles from the water column and to characterize the release of nutrients from decomposing material. These formulations do not make the role of the mussels explicit and will not capture the effects of their variable abundance and activity across seasons and locations. The model appears to reproduce most of the observable patterns in the major response variables but may be doing so by inappropriate functional processes. This has the potential to distort its predictions of how the system would response to any given change, such as altered inputs from point sources or a change in the physical environment. Some of this work is underway with modest support through an NSERC CRD that was enabled by the industry and municipal funds directed to the current project but it is likely to require further investment. 5. Additional work targeting the detachment and transport processes that are critical in making Cladophora a nuisance is needed. The current model formulation treats detached Cladophora as a diffusible substance, and this makes predicted transport quite unrealistic. A combination of experimental field work and model development work would be needed to address this current deficiency. Success with a better model formulation would greatly strengthen inferences about provenance of fouling algae at any given point of impact.

Acknowledgments To all our colleagues that at one point in time shared data and collaborated with the project: Ram Yerabundi, Sue Watson, Bill Booty and Bill Schertzer (NWRI); David Depew, Ted Ozersky, Sairah Malkin and Adam Houben (UW); Gary Bowen (TRCA) and Todd Howell (MOE). Finally the support from Carol Gregoris, Bob Hester and Wayne Burchat (OPG) is greatly apreciated.

63

References Auer, M.T. and Canale, R.P. 1982a. Ecological and mathematical modeling of Cladophora in Lake Huron: 2. Phosphorus uptake kinetics. J. Great Lakes Res. 8:84–92. Auer, M.T. and Canale, R.P. 1982b. Ecological and mathematical modeling of Cladophora in Lake Huron: 3. The dependence of growth rates on internal phosphorus pool size. J. Great Lakes Res. 8:93–99. CH2M, 2005. Final Report: “Duffin Creek Water Pollution Control Plant Environmental Assessment”; prepared for the municipalities of York & Durham. Depew, D.C. 2009. Cladophora growth in littoral environments of large lakes: Spatial complexity and ecological interpretations. Ph.D. thesis, Biology Department, U. Waterloo, Sept. 2009. Higgins, S.N., Hecky, R.E., and Guildford, S.J. 2005a. Modeling the growth, biomass, and tissue phosphorus concentration of Cladophora glomerata in eastern lake erie: Model description and field testing. J. Great Lakes Res. 31: 439-455. Higgins, S.N., Howell, E.T., Hecky, R.E., Guildford, S.J., and Smith, R.E.H. 2005b. The wall of green: The status of cladophora glomerata on the northern shores of lake erie's eastern basin, 1995-2002. J. Great Lakes Res. 31: 547-563. Higgins, S.N., Hecky, R.E. and Guildford, S.J. 2006. Environmental controls of Cladophora growth dynamics in eastern Lake Erie: application of the Cladophora Growth Model (CGM). J. Great Lakes Res. 32: 629-644. Holmes, R.M., Aminot, A., Kerouel, R., Hooker, B.A., and Peterson, B.J. 1999. A simple and precise method for measuring ammonuim in marine and freshwater ecosystems. Can. J. Fish. Aquat. Sci. 56: 1801-1808. Leon, L., Smith R.E.H. and Romero, J. 2006. Lake Erie hypoxia simulations with ELCOM-CAEDYM. In Proceedings of 3rd Biennial meeting of Int. Environ. Model. Software Soc., July 2006, Burlingyon, VT, USA pp 1-6 Rao, Y.R. and Schwab, D.J. 2007. Transport and mixing between the coastal and offshore waters in the Great Lakes: a review. J. Great Lakes Res. 33: 202-218. Robson, B.J. and Hamilton, D.P. 2004. Three-dimensional modelling of a Microcystis bloom event in the Swan River estuary, Western Australia. Ecol. Mod. 174: 203-222. Smith, R.E.H., Furgal, J.A., Charlton, M.N., Greenberg, B.M., Hiriart, V., and Marwood, C. 1999. The attenuation of ultraviolet radiation in a large lake with low dissolved organic matter concentrations. Can. J. Fish. Aquat. Sci. 56: 1351-1361. Stainton, M.P., Capel, M.J. and Armstrong, F.A.J. 1977. The chemical analysis of fresh water. Can. Fish. Mar. Serv. Misc. Spec. Publ. 25, 180 pp. Fisheries and Oceans, Canada.