Impact of Oligotrophication, Temperature, and Water Levels on Walleye Habitat in the Bay of Quinte, Lake Ontario

Transcription

1 Transactions of the American Fisheries Society 133: , 2004 Copyright by the American Fisheries Society 2004 Impact of Oligotrophication, Temperature, and Water Levels on Walleye Habitat in the Bay of Quinte, Lake Ontario C. CHU* AND C. K. MINNS Great Lakes Laboratory for Fisheries and Aquatic Sciences, Fisheries and Oceans Canada, Bayfield Institute, Post Office Box 5050, 867 Lakeshore Road, Burlington, Ontario L7R 4A6, Canada J. E. MOORE JEMSys Software Systems, Inc., 22 Marion Crescent, Dundas, Ontario L9H 1J1, Canada E. S. MILLARD Great Lakes Laboratory for Fisheries and Aquatic Sciences, Fisheries and Oceans Canada, Bayfield Institute, Post Office Box 5050, 867 Lakeshore Road, Burlington, Ontario L7R 4A6, Canada Abstract. Recent environmental changes in the Bay of Quinte, Lake Ontario, have coincided with a decline in the stocks of walleye Sander vitreus. Suitable habitat supply was estimated in three sections of the bay during the summers of to assess its role in the decline. An empirical model was developed to predict suitable habitat area for walleyes based on their preferences for cool water and low light intensity. The results indicated that lack of suitable light limits walleye habitat in the bay. Walleye habitat in the shallow upper bay has decreased at the rate of 34 ha/year since the invasion of dreissenid mussels in 1994, while that in the middle and lower bays has remained abundant. Walleye stocks and suitable habitat in the upper bay have both declined since the early 1990s. However, this pattern has not been consistent through time and suggests that other factors have also affected the Bay of Quinte walleye population. The analyses developed here can be used as a tool to enhance the assessment of walleye habitat dynamics in the Bay of Quinte and allow us to examine the impact of oligotrophication on the habitat of an important recreational and commercial species. The lower Great Lakes (Lakes Erie and Ontario) have undergone a series of trophic state changes in the last 150 years. Eutrophication in Lake Ontario began in the mid-1800s with the settlement of Europeans and was exacerbated by the use of phosphate-based detergents in the 1940s (Schelske 1991). In response to this, the Great Lakes Water Quality Act was enacted in 1972 to control phosphorus loadings and nuisance algae in the lower lakes and maintain oligotrophy in the upper lakes (Hartig et al. 1991). As a result, phosphorus loadings in Lake Ontario decreased from 14,000 tons/year in to 500 tons/year in 1981, and mid-lake chlorophyll concentrations decreased 20% from 1981 to This has led to meso-oligotrophic conditions in Lake Ontario today (Johengen et al. 1994). However, the shift in the trophic status of Lake Ontario cannot be attributed to the reduction in phosphorus loadings alone. Dreissenid mussels * Corresponding author: chuc@dfo-mpo.gc.ca Received May 21, 2003; accepted December 8, 2003 were first reported in Lake St. Clair in 1988 and quickly spread to nearby freshwater systems (Bailey et al. 1999). Densities are highest in the nearshore regions of the Great Lakes; in Lake Ontario, densities are highest in the western basin and the Bay of Quinte along the northeastern shore (Bailey et al. 1999; Nicholls et al. 1999). The mussels have since induced a biological oligotrophication by filtering phytoplankton and other suspended materials from the water. In the Bay of Quinte, point source phosphorus loadings have declined from 214 kg/d in to kg/d in (Millard, unpublished data), and there have been decreases in chlorophyll concentrations, primary production, and phytoplankton biovolume (Nicholls et al. 1999). The yearly overall mean water temperature in the bay throughout the ice-free season is 17.9 C, but this has varied by as much as 3 C since 1972 (Casselman et al. 1999). These changes have certainly altered the physical habitat of the bay. Using the Bay of Quinte as a case study, we examine the impact of oligotrophication on the 868

2 WALLEYE HABITAT IN THE BAY OF QUINTE 869 FIGURE 1. Map showing the location of the Bay of Quinte in northeastern Lake Ontario, along with those of the surrounding cities; the upper, middle, and lower bays; and the index stations (Belleville [B], Hay Bay [HB], and Conway [C]). habitat of one of the bay s most sought-after species, walleye Sander vitreus. The walleye is a prime candidate for this study because it is a coolwater species that prefers turbid environments (Ryder 1977; Christie and Regier 1988). Adult walleyes spawn in the shallow waters of the upper bay in early spring and migrate to the lower bay and open lake shortly afterwards. Younger walleyes (ages 1 3) remain within the bay year-round. Adults return in the fall and overwinter in the bay. Walleye numbers have been declining since the mid-1990s as water clarity and temperature have increased (OMNR 2002). A habitat supply model was created to assess the effects of changing temperature and light conditions on the physical space available for walleyes in the bay. The model follows the approach developed by Lester et al. (in press) for walleyes in inland lakes of Ontario. In that approach, light and temperature preferences were used to identify suitable habitat areas. Habitat area was then related to sustained yield. This approach was first developed by Christie and Regier (1988), who combined lake temperature measurements, hypsographic data, and functions of temperature-dependent growth to relate thermal habitat areas and volumes to the sustained yield of four fish species. Empirical water level and macrophyte distribution data also were incorporated into our model to further examine the changes in walleye habitat in the Bay of Quinte from 1972 to Study Area The Bay of Quinte is a Z-shaped bay on the northeastern shore of Lake Ontario with a total area of km 2 (Figure 1). The bay is divided morphometrically into three regions, namely, the upper, middle, and lower bays. The upper bay extends from Trenton to Green Point, Ontario, the middle bay from Green Point to Glenora, and the lower bay from Glenora to Lake Ontario. The upper bay covers km 2 and is the shallowest of the three regions. The middle bay covers 49.2 km 2 while the deep lower bay covers 71.8 km 2. Methods Data. The model was designed to assess changes in walleye habitat in the Bay of Quinte from 1972 to Six types of data are required as inputs. These include hypsographic data for each section of the bay, empirical temperature and light data, light extinction coefficients, water lev-

3 870 CHU ET AL. els, macrophyte distributions, and temperature and light suitability curves for walleyes. The hypsographic data are estimates of lake area (and volume) as a function of depth or elevation. These data were obtained with geographical information systems software (SPANS; TYDAC Research, Inc. 1997) using shoreline and depth information from Canadian Hydrographic Service charts of the bay. All temperature, light, and light extinction data were provided by the Department of Fisheries and Oceans (DFO; Millard, unpublished data) and were collected as part of Project Quinte, an ongoing monitoring program that measures the physical, chemical, and biological properties of the bay. The upper, middle, and lower bays were represented by single stations, namely, DFO s index stations at Belleville, Hay Bay, and Conway, respectively (Figure 1). Vertical temperature profiles from surface to bottom were available on a weekly or biweekly basis for the summers (May to October) of (Figure 2). Temperatures ( C) were measured with various thermistor devices over the years. Linear interpolations were used to estimate temperatures when thermal data were lacking. Temperatures in the three bays showed no identifiable pattern of change from 1972 to 2001 but have increased , , and C/ year since 1994 (Figure 2). Solar radiation data (millieinsteins [me] m 2 min 1 ) were collected at the Point Petre lighthouse (43 50 N, W) with a spherical quantum sensor LI-190SA (LI-COR, Lincoln, Nebraska) connected to an integrating data logger. These data are within the photosynthetically available radiation range ( nm) and were available for the open-water seasons. The data were postprocessed (by DFO) into half-hour integrated values covering the daylight period during every day of the open-water season from 1987 to Averages of the empirical data were used to estimate the solar radiation for each day of the open-water season from 1972 to Light extinction coefficients (m 1 ) were available on a biweekly basis from 1972 to 2001 and were calculated as the slopes of straight lines fitted to semilogarithmic plots of light intensity against depth (Figure 3). Light intensities were measured with an underwater quantum sensor (LI-192SA; LI-COR; Millard and Sager 1994). Linear interpolations were used to calculate the light extinction coefficients on days when data were not obtained. Light at depth was calculated from the following equation: A (e ZU) (e ZL) [ (ZL ZU)] 1, where is the light extinction coefficient, ZU is the top of the depth interval, and ZL is the bottom of the depth interval. Light at depth is calculated by multiplying A by the solar radiation data for that date and time. The light extinction data showed a decrease, that is, increase in transparency, since 1972 (Figure 3). Water-level data were available on a daily basis from 1972 to These data were obtained from the Canadian Hydrographic Service station in Kingston, Ontario, and consist of dates and corresponding elevations. Water-level changes were incorporated into the model by increasing or decreasing the surface elevation of the bay. For elevations above the bathymetric datum, hypsographic curves were extrapolated linearly using the slope prevailing in the first few meters below the datum. The water-levels data set indicated a decrease since 1972 (Figure 4). Long-term macrophyte data were used to identify the depths at which macrophyte cover changed throughout three time stanzas in the Bay of Quinte: (pre phosphorus control), (post phosphorus control), and (post dreissenid mussel invasion) (Table 1). These data were collected using point transects and echosounding surveys (Seifried 2002). The surveys were conducted only once a year and were not conducted every year of the period; therefore, the average of the years with data within a time stanza were used to fill in missing data (Table 1). Because walleyes have a preference for hard substrates such as rock and gravel as opposed to densely vegetated areas (McMahon et al. 1984), the depths where macrophytes provided more than 33% cover were treated as exclusion zones. The area or volume of water available from these depths to the shore were excluded from the analysis and in effect decreased the surface elevation of the bay, causing a decrease in habitat area and volume. Suitability. The temperature and light preferences of walleyes were used to generate the suitability curves. Light preferences were provided by Lester et al. (in press) and were developed from data on catch per unit effort (CUE; fish/anglerhour) at different light intensities. The optimal light intensity for walleyes ranges from 8 to 68 lx, with a peak of 28 lx at depths ranging from 2 to 6 m (Lester et al., in press). To be consistent with the format of our solar radiation data, the light preferences were converted from lux to

4 WALLEYE HABITAT IN THE BAY OF QUINTE 871 FIGURE 2. Depth weighted mean temperatures from vertical profiles taken in the upper, middle, and lower bays of the Bay of Quinte from 1972 to Data were recorded on a biweekly basis, and linear interpolations were used to estimate missing data. me m 2 min 1 by means of the following equation: Y 1.08 X/1,000, where X light intensity (lx) and 1.08 converts killilux (X/1,000) to me m 2 min 1 (LI-COR 1985; Figure 5a). Walleye temperature preferences were determined from Christie and Regier (1988) and nearshore walleye catch data (catch/gill net) grouped by temperature strata for in Lake Ontario and the Bay of Quinte (T. Stewart, Ontario Ministry of Natural Resources, unpublished data; Figure 5b). Model. The general logic of the model algorithms is as follows: The bay is divided into three regions upper, middle, and lower. For each re-

5 872 CHU ET AL. FIGURE 3. Light extinction coefficients for the upper, middle, and lower bays of the Bay of Quinte from 1972 to Data were recorded on a biweekly basis, and linear interpolations were used to estimate missing data. gion, the season (May 1 to October 31) is divided into days and the days into hours. Water level and macrophyte changes are included on a daily basis, and the surface elevation of the bay is adjusted accordingly. The region is then divided into 1-m depth intervals using the hypsometric data. For each combination of depth and hour, the model estimates mean temperature and light intensity values from the input data. Given the temperature and light intensity, the suitability functions compute temperature and light suitabilities for each combination of depth and hour. The temperature and

6 WALLEYE HABITAT IN THE BAY OF QUINTE 873 FIGURE 4. Daily water levels in the Bay of Quinte from 1972 to 2001 as recorded at the Canadian Hydrographic Service station in Kingston. TABLE 1. Depths (m) at which macrophyte cover changed, by time stanza and section in the Bay of Quinte. Data were obtained through the use of point transects and echosounding once a year; averages of years with data were used to approximate values for years without data within each time stanza. Time stanza Year Upper bay Middle bay Lower bay Pre phosphorus control Post phosphorus control Post dreissenid invasion

7 874 CHU ET AL. FIGURE 5. (a) Light and (b) temperature suitability curves for walleyes; the abbreviation E stands for einsteins. light suitabilities are then multiplied to give a combined suitability, temperature light. This suitability is taken to represent the overlap between suitable thermal and optical habitats. Optical habitat is the area or volume of water that is suitable for walleyes given their preference for light conditions in the 8 68-lx range. The three suitabilities are then multiplied by the bay area (or volume) represented by the depth interval. This procedure provides three separate estimates of suitable area or volume for each depth interval. These are then summed over depth to give suitable optical habitat area or volume, thermal habitat, and thermal optical habitat. These can then be averaged over longer time periods to provide daily, monthly, and yearly estimates of suitable area or volume for walleyes in each bay. The area or volume estimates can then be summed to calculate the supply in the whole bay. A model simulation using the available data was performed to estimate the changes in walleye habitat in the Bay of Quinte. Christie and Regier (1988) reported that area estimates describe walleye habitat better than volume estimates because of the benthic nature of the organisms. We therefore report the optical habitat area (OHA), thermal habitat area (THA), and thermal optical habitat area (TOHA) in our results. The values of TOHA for each bay were then compared with walleye (age-1 and older) CUE data from 1972 to 2001 (OMNR 2002). Regression analyses were performed to determine the general trend in TOHA in the three bays through time and to compare habitat supply and walleye catch data. These analyses are not absolute because the habitat supplies are modeled parameters and other studies have indicated that time series data are inherently autocorrelated (Hurlbert 1984; Bence 1995). We have therefore employed the linear regressions merely to show the rate of change in habitat supply and get the slope of the relationship between TOHA and walleye catch. Results The results indicated that THA was abundant in all three bays but that OHA appears to be limiting suitable TOHA in the bay (Figure 6). The difference between OHA and TOHA habitats increased from the upper bay to the middle to lower bays. The thermal habitat has decreased in the upper bay but increased in the middle and lower bays (Table 2). The value of OHA has increased in all three bays. That of TOHA has generally increased in all three bays; however, since dreissenid mussels invaded in 1994, TOHA has decreased ha/year in the upper bay and increased and 6.88 ha/ year in the middle and lower bays, respectively. The monthly means for the three bays indicated that lack of OHA limited the maximum TOHA but that TOHA followed thermal changes throughout the season; that is, TOHA increased as the bay warmed in May and June but decreased as the bay cooled in September and October (Figure 7). The upper bay offered the most suitable habitat throughout the summer. Periods of low ( 500 ha) TOHA in the upper bay during the mid-1970s and late 1990s accompanied low walleye catch. However, this trend is not consistent through time; that is, periods of abundant habitat are not always accompanied by strong recruitment in subsequent years (Figure 8). Linear regressions of TOHA and the walleye catch data showed a positive correlation between declining habitat supply in the upper bay and declining stocks since 1994, the slope of the relationship equaling The value of TOHA and walleye catch were negatively related in the middle and lower bays, with slopes of 0.15 and 0.17, respectively. Discussion The results of these analyses provided insight into the complexity of walleye habitat availability

8 WALLEYE HABITAT IN THE BAY OF QUINTE 875 FIGURE 6. Mean annual summer (May October) optical habitat area (OHA), thermal habitat area (THA) and thermal optical habitat area (TOHA) in the upper, middle, and lower bays of the Bay of Quinte (log 10 scale). TABLE 2. Rates of change (ha/year) in optical, thermal, and thermal optical habitats in the three sections of the Bay of Quinte from 1972 to Section Upper bay Middle bay Lower bay Optical habitat area Thermal habitat area Thermal optical habitat area in the Bay of Quinte. Comparison of OHA, THA, and TOHA in the three bays from 1972 to 2001 showed that suitable light conditions limit walleye habitat supply. The increasing difference between OHA and TOHA in going from the upper bay to the middle and lower bays indicated that there was less overlap between suitable thermal and optical habitats in the middle and lower bays. This can be attributed to the shape of the basins and is better explained by keeping in mind the mean monthly

9 876 CHU ET AL. FIGURE 7. Mean monthly summer (May October) optical habitat area (OHA), thermal habitat area (THA), and thermal optical habitat area (TOHA) in the upper, middle, and lower bays of the Bay of Quinte (log 10 scale). summaries of habitat area. The shallow waters of the upper bay require less time to warm to the preferred temperatures for walleyes (18 22 C). This results in considerable overlap between suitable optical and thermal habitat for much of the ice-free season. In the deeper waters of the lower bay, slower warming causes the overlap between suitable optical and thermal habitat to occur later in the season. Therefore, the mean TOHA for the year is noticeably less than the available thermal and optical habitat. Although THA in the upper bay has decreased

10 WALLEYE HABITAT IN THE BAY OF QUINTE 877 FIGURE 8. Mean annual summer (May October) thermal optical habitat area (TOHA) for walleyes in the upper, middle, and lower bays of the Bay of Quinte and walleye population trends from 1972 to ha/year since 1972 and ha/year since 1994, the depth-weighted mean temperatures show no noticeable pattern of change since The depths of macrophyte beds, however, have been expanding since 1994 (Seifried 2002). This suggests that the decrease in THA can be attributed to the expansion of macrophytes in the upper bay. The upper bay is shallow (mean depth, 3.2 m), and macrophytes expanded 1 m deeper after the dreissenids invaded. This removed much of the suitable THA because vegetated areas were treated as exclusion zones for walleyes. This macrophyte effect was not evident in the OHA results because much of OHA occurs in deeper water than the macrophytes. Adult walleyes spawn in the shallow waters of the upper bay in early spring and then migrate to the lower bay and open lake. They return in the fall and overwinter in the bay. Younger walleyes (ages 1 3) remain predominantly within the upper bay year-round (OMNR 2002). Our findings indicated that the upper bay offered more habitat in the spring than the middle and lower bays combined. This suggests that adult walleye habitat use in the spring is consistent with habitat availability. However, the migration of the adults to the open lake and lower bay after spawning suggests that more habitat or prey resources are available in the open water than in the bay. Habitat use by youngof-year and juvenile walleyes is consistent with habitat availability given that the upper bay had the most TOHA of the three bays throughout the summer. The inconsistent relationship between TOHA and walleye catch suggests that other factors are influencing walleye recruitment in the bay. One possible factor is the predator prey dynamics in the bay. Changes in large-species (walleye and lake trout) abundance are often preceded by changes in the small-species assemblage 2 years earlier (Casselman et al. 1999). During the late 1980s and early 1990s, prey production was high and sustained predation pressure from walleyes and other large species. A decrease in water temperature of 2.1 C from 1991 to 1992 led to a decline in smallspecies abundance. This reduced the body condition of many large species and induced higher than normal mortality in walleyes in subsequent years (Casselman et al. 1999). Our study and these findings suggest that habitat and predator prey dynamics may have been working in concert to regulate walleye stocks in the bay since Limitations of the Input Data As aforementioned, annual estimates of the depths of the macrophyte beds were used to limit the amount of suitable walleye habitat in the bay. However, seasonal changes in the macrophyte distributions (growth, senescence, and death) affect how much habitat is available on a daily or weekly basis. This could affect how walleyes use the available habitat in the bay. Future work involving more

11 878 CHU ET AL. frequent sampling of a small area would allow one to test whether daily or weekly macrophyte estimates significantly affect the monthly or annual trends in suitable walleye habitat reported here. Implications for Management and Species Modeling Although this model does not exclusively describe the spawning habitat of walleyes, it does demonstrate that more suitable adult walleye habitat is available in the upper bay than in the middle and lower bays during the spawning season (May and June). This may have implications for the distribution of reproductive adults prior to spawning and can influence the harvest in the upper bay during those months. Fisheries managers in the Bay of Quinte region suggest that to maintain a sustainable population, exploitation must decline in proportion to population declines (OMNR 2002). If it is accepted that the recent declines in walleye stocks are at least partially attributable to declines in available habitat, the model developed in this study can be used to forecast future changes in walleye habitat and stock dynamics. Possible applications include examining the further impacts of dreissenids and water-level regulation. The model indicated that light conditions limit walleye habitat in the Bay of Quinte. The potential future impacts of the dreissenid mussels can be estimated by varying the light data of the model. Because water levels in the Bay of Quinte are regulated, this model can also be used to assess the impacts of different water-level scenarios on walleye habitat. The results of such scenarios can then be used to ensure that water-level regulations are in compliance with the federal Fisheries Act, which requires no net loss of fish habitat (Minns 1997). Acknowledgments We thank Nigel Lester, Tom Stewart, Jim Hoyle and members of the Glenora Fisheries Station for providing important background information regarding walleye dynamics in the Bay of Quinte. We are also indebted to the staff at DFO who collected and analyzed much of the data used to develop this model. Dennis DeVries and two anonymous reviewers provided constructive comments that guided this manuscript in the right direction. This project was funded by The Environment Canada Great Lakes Action Plan and Natural Resources Canada s Climate Change Action Fund (Impacts and Adaptation). References Bailey, R. C., L. Grapentine, T. J. Stewart, T. Schaner, M. E. Chase, J. S. Mitchell, and R. A. Coulas Dreissenidae in Lake Ontario: impact assessment at the whole-lake and Bay of Quinte spatial scales. Journal of Great Lakes Research 25: Bence, J. R Analysis of short time series: correcting for autocorrelation. Ecology 76: Casselman, J. M., K. A. Scott, D. M. Brown, and C. J. Robinson Changes in relative abundance, variability, and stability of fish assemblages of eastern Lake Ontario and the Bay of Quinte: the value of long-term community sampling. Aquatic Ecosystem Health and Management 2: Christie, G. C., and H. A. Regier Measures of optimal thermal habitat and their relationship to yields for four commercial fish species. Canadian Journal of Fisheries and Aquatic Sciences 45: Hartig, J. H., J. F. Kitchell, D. Scavia, and S. B. Brandt Rehabilitation of Lake Ontario: the role of nutrient reduction and food web dynamics. Canadian Journal of Fisheries and Aquatic Sciences 48: Hurlbert, S. H Pseudoreplication and the design of ecological field experiments. Ecological Monographs 54: Johengen, T. H., O. E. Johannsson, G. L. Pernie, and E. S. Millard Temporal and seasonal trends in nutrient dynamics and biomass measures in Lakes Michigan and Ontario in response to phosphorus control. Canadian Journal of Fisheries and Aquatic Sciences 51: Lester, N. P., A. Dextrase, R. S. Kushneriuk, M. Rawson, and P. A. Ryan. In press. Light and temperature: key factors affecting walleye abundance and production. Transactions of the American Fisheries Society. LI-COR LI-193SA spherical quantum sensor instruction manual. LI-COR, Publication , Lincoln, Nebraska. McMahon, T. E., J. W. Terrell, and P. C. Nelson Habitat suitability model: walleye. U.S. Fish and Wildlife Service, FWS/OBS-82/10. 56, Washington, D.C. Millard, E. S., and P. E. Sager Comparison of phosphorus, light climate, and photosynthesis between two culturally eutrophied bays: Green Bay, Lake Michigan, and the Bay of Quinte, Lake Ontario. Canadian Journal of Fisheries and Aquatic Sciences 51: Minns, C. K Quantifying no net loss of productivity of fish habitats. Canadian Journal of Fisheries and Aquatic Sciences 54: Nicholls, K. H., G. J. Hopkins, and S. J. Standke Reduced chlorophyll-to-phosphorus ratios in nearshore Great Lakes waters coincide with the establishment of dreissenid mussels. Canadian Journal of Fisheries and Aquatic Sciences 56: OMNR (Ontario Ministry of Natural Resources)

12 WALLEYE HABITAT IN THE BAY OF QUINTE 879 Lake Ontario fish communities and fisheries: 2001 annual report of the Lake Ontario Management Unit. OMNR, Report , Picton. Available: /report2001.html. (November 2002). Ryder, R. A Effects of ambient light variations on behaviour of yearling, subadult, and adult walleyes (Stizostedion vitreum vitreum). Journal of the Fisheries Research Board of Canada 34: Schelske, C. L Historical nutrient enrichment of Lake Ontario: paleolimnological evidence. Canadian Journal of Fisheries and Aquatic Sciences 48: Seifried, K. E Submerged macrophytes in the Bay of Quinte: Master s thesis. University of Toronto, Toronto. TYDAC Research, Inc Spatial analysis system (SPANS), version 7. Ottawa.