Over the past 40 years, hydroelectric, agricultural and urban watershed development and changing hydrology have transformed Lake Winnipeg into a highly eutrophic reservoir with annual outbreaks of widespread surface algal blooms, shoreline and net fouling, and concerns with intermittent cyanotoxin production. To provide a better understanding of the magnitude of these changes and the major causes, we examine long-term increases in phytoplankton biomass and shifts in phytoplankton species dominance in the context of both in-lake and watershed processes. We compare phytoplankton and water quality data from early (1969) and recent (1994–2007) lake-wide surveys, and information from paleolimnological analysis of sediment cores and satellite remote sensing. Our results demonstrate a recent and dramatic rise in severe algal blooms and increased dominance of cyanobacteria beginning in the mid-1990s, coincident with a large increase in phosphorous loading to the lake. Distinct increases in sediment core accumulation of nutrients and chlorophyll, cyanobacteria and diatom microfossils coincided with hydroelectric and agricultural development, increased Red R discharge and shifts in water transparency patterns across the lake. There has been a dramatic increase in phytoplankton biomass, accompanied by marked shifts in seasonal community composition. Spring diatoms blooms are of shorter duration and increasingly dominated by more eutrophic diatom taxa while summer blooms show reduced taxonomic diversity and an increased predominance of nitrogen-fixing cyanobacteria. Satellite images showed annual development of vast summer surface blooms, mainly in the north basin, with chlorophyll highest in regions of relatively low suspended sediment concentration and high transparency. There is an increasing dominance of potentially toxic cyanobacteria taxa and high levels of microcystins in nearshore samples of surface blooms. The combined effects of nutrient increases, algal species shifts and toxin production represent a potential threat to the sustainability of ecosystem function and productivity.
Lake Winnipeg (LW), Manitoba, is the world's 11th largest lake. Now considered as one of the North American Great Lakes, it is perhaps the one of greatest concern. However, it differs from the five Laurentian Great Lakes (LGLs) in several important ways. It is generally shallower (zmax , south basin = 12 m; zmax , north basin = 18 m) and even in the deeper northern basin most of its surface area corresponds to the LGL zone technically classified as inshore (defined by the 20m isobath; SOLEC, 2008).
With a catchment area ratio approximately 10x larger than even the more eutrophic lower LGLs (Figures 1 and 2), LW is perhaps more vulnerable to increased nutrient loading due to cultural development in, and increased runoff from its watershed (Webster et al., 2008). In fact, it has become more eutrophic in recent years, largely due to the increased nutrient loading from the watersheds of the Red and Winnipeg Rivers (R) which since 1994 have together contributed over 80% of total phosphorous (P) loading to LW (State of the Lakes, SOL 2011, in press). The Red R in particular sustains year-round P concentrations (>200 mg P m−3; Jones and Armstrong, 2001) an order of magnitude higher than the other major tributaries to the lake. Because P concentration in the Red R is positively correlated with discharge a near-doubling of annual runoff due to recent changes in basin precipitation and discharge patterns in the watershed has caused total P loading from this river to more than double since the early 1990s (McCullough et al., 2011 in press).
Along with nutrients, the Red R also delivers a constantly high input of suspended sediment, so that although P and nitrogen (N) are highest in the south basin, the phytoplankton assemblage there is limited by low light in the water column from fully exploiting these resources. Sediment deposition occurs as water moves northward through the lake, so that there is more light (as indicated by higher Secchi depth readings; Salki et al., 2006) and consequently higher productivity in the north basin. This manifests as widespread surface blooms which have increased both frequency and severity in the north basin since the mid-1990s. These blooms are dominated by filamentous, heterocyst-bearing cyanobacteria taxa, which fix significant levels of atmospheric nitrogen each summer, fuelled by light and the high ambient P concentrations (Hendzel et al. 2006). Thick windblown mats and scums of this bloom material have fouled beaches and recreational areas in recent years, attracting media and public attention. Of greater concern are high levels of cyanotoxins that have been measured in several surveys since 1999 (SOL, 2011, in press).
Anecdotal evidence from the Red River Expedition (Huysche, 1871) and from early scientific expeditions on the lake indicates that surface blooms in LW are not limited to the late-20th century. A micrograph published by Lowe (1924) of a sample from a mid-summer algal population shows mostly centric diatoms (Stephanodiscus and Aulacoseira species) and some large dinoflagellates (Ceratium furcoides) and few cyanobacteria (Anabaena). Nonetheless, Lowe also reported common thick scums of Anabaena flos-aquae. Likewise, Bajkov (1930, 1934) described cyanobacteria (mainly Aphanizomenon and several species of Anabaena) blooms ‘as thick as carpets’ in some areas of the lake. These incidental reports indicated at least periodic bloom development signifying that the lake was likely mesotrophic even early in the 20th century. However, sediment records (Kling, 1998) which indicate that productivity has increased dramatically over the last half-century, and a 25 year series of satellite images (SOL, 2011, in press) which shows an increase in the frequency of very widespread surface blooms beginning in the mid-1990s, provided supporting evidence for reports of fishermen and other longtime residents that past bloom-forming events were less frequent than the almost annual large blooms that have been observed in recent years.
Given the recent onset of these recurrent and severe blooms and potential for a decline in ecosystem integrity, there is a need for scientifically-based management of the lake. This requires an understanding of long-term changes in phytoplankton species and biomass and their relationship with in-lake processes, meteorological and hydrological events and drainage basin characteristics and activities. However, few quantitative early data exist, with the exception of some extensive water quality surveys in 1969 (which did not include an assessment of the biota; Brunskill et al., 1979) and an early assessment of the benthos (which recorded a diverse assemblage of over 400 species; Flanagan et al., 1994). To address this data gap, we have sampled phytoplankton during spring, summer and fall lake-wide surveys three times annually since 2002, and related this to reconstructed LW paleolimnology of nutrients and algal microfossils from sediment cores. We combined these data collected at discreet dates and sites with satellite imagery to obtain a better understanding of spatial and temporal bloom dynamics from large-scale remote surveillance mapping. This paper presents a synopsis of major changes in the phytoplankton, comparing early assemblages (prior to 2000) to those observed from two of the recent surveys from 2003 and 2007. The focus is on important changes in algal community structure and dominant cyanobacteria implicated in bloom formation, N2 fixation and toxin production.
Paleolimnological data presented here are from analysis of a representative core taken in the central north basin of LW in 1994 (53°27.29’N 98°21.95’WPM) using algal microfossil methods described by Kling (1998) and dated using lead-210 and cesium methods described by Wilkinson (1985). Spring, summer and fall phytoplankton surface (0–1 m depth) samples were collected on surveys of up to 60 stations in the north and south basins. Phytoplankton samples were preserved immediately with Lugol's iodine. All taxonomic data were determined by visual counts made under an inverted microscope using the standard Utermöhl technique and reporting composition and biomass (Findlay and Kling, 1996). Satellite-derived maps shown in this paper are from from images recorded by MERIS (the European Space Agency's Medium Resolution Imaging Spectrometer). Chlorophyll biomass in these maps was estimated from MERIS data by the fluorescence line height method (calibrated for LW using in-situ samples) which provides a measure of upwelling chlorophyll-a fluorescence at 681 nm (McCullough, 2007).
Results and Discussion
A sediment core, dated according to methods described in Wilkinson (1985), retrieved in 1994 (Figure 3) shows a distinct increase in P, carbon (C) and chlorophyll-a since the late 1950s, which follows rapidly increasing agricultural development in the watershed, hydro-electric development on the Saskatchewan R (1960s) and lake level regulation (1976). These sediment core nutrient data are consistent with a LW nutrient model (McCullough et al., 2011, in press) indicating that P concentration in the lake, although partly a function of increasing anthropogenic loading to the watershed, also varies closely with discharge from the Red R watershed. The model indicates variability but no long term trend in P loading to the lake from the 1960s until a sudden and dramatic rise in the 1990s (Figure 4) that coincides with a regime shift in the Red R to a period of higher annual mean discharge and more frequent flooding. The P-shift indicated in this modeled record is consistent with survey data which indicates that the basin-weighted mean P concentration in LW was 34 mg m−3 in 1969, and less in the early 1990s, but has been from 38–60 mg m−3 in every mid-summer whole lake survey since 1998 (McCullough et al., 2011, in press). Since 1969, there has been a change in water transparency in the north basin which Salki et al. (2006) explained as a consequence of the construction of upstream reservoirs on the Saskatchewan R which traps both suspended clay and silt, and nutrients derived from the river's watershed. We postulate that this resulted in decreased light limitation, and hence, increased phytoplankton utilization of the P transported from the south basin which coincides with increased nutrient concentrations in the core (Figure 3) and a parallel increase in non-siliceous and siliceous algal microfossils (Figure 5).
Although seasonal and spatial phytoplankton surveys in 1969 showed a maximum recorded biomass of 10,000 mg m−3, the basin-wide summer (Jul–Aug) average biomass was less than 2000 mg m−3 (Figure 6) and comprised a diversity of taxonomic classes. In marked contrast, since 1999, total biomass has exceeded 40,000 mg m−3 in several individual samples, and there has been a progressive increase in mean biomass reaching over 8000 mg m−3 by 2007, and this has been accompanied by a shift to a predominance of cyanobacteria during the summer.
In recent years (data on file, SOL, 2011, in press; Kling et al., 2007), typical spring survey phytoplankton assemblages have been dominated by the several species of centric diatoms in the genera Aulacoseira, Stephanodiscus and Cyclostephanos although the planktonic araphid species in the genera Asterionella, Fragilaria, Diatoma and Tabellaria have also been present. In summer, cyanobacteria have developed high biomass in both basins. The south basin may become dominated by non-N2 fixing cyanobacteria (several species of Microcystis and Planktothrix suspensa) by early July. In the north basin, surface blooms of N2-fixing taxa develop, and have at times been observed in satellite images to cover as much as two-thirds of the north basin. Depending on climatic conditions, in some years a mixed cyanobacteria bloom develops in the either basin in mid- to late July (rarely June), usually dominated by the Aphanizomenon flos aquae complex but sometimes by Anabaena flos aquae, A. lemmermannii and A. mendotae). Usually by fall, however, the cyanobacteria consist almost totally of the Aphanizomenon flos aquae complex. These cyanobacterial-dominated assemblages persist in some years as late as November, although more commonly centric diatoms reappear in the fall.
Remote sensing has provided an effective method to monitor the development of algal and cyanobacterial blooms across the entire lake, and has demonstrated that vast surface blooms of cyanobacteria (often 10,000 to 15,000 km2 in extent) now develop in the north basin in most years (SOL, 2011). An example is provided in Figure 7 A, which shows turbidity and phytoplankton distributions in Lake Winnipeg on 8 August 2008. At the time of this image, suspended solids concentration would have ranged from <2 g m−3 north of Long Point (black regions in NW north basin in Figure 7 A) to >30 g m−3 in bays along the southwest side of the Narrows (white in Figure 7 A; very bright white indicates clouds). The light tones in the northern north basin indicate phytoplankton surface blooms. By quantitative analysis of the satellite data, chlorophyll biomass exceeded 50 mg m−3 (the upper limit of determination by the method) throughout most of the north basin, including areas not covered by surface mats of phytoplankton (Figure 7 B). In situ data indicates that cyanobacteria dominated the plankton community at the time. The lower chlorophyll concentration in the plume of the Saskatchewan R. (black regions along the NW shore and N of Long Point) has been observed in most satellite-derived chlorophyll maps of the lake. Likewise, chlorophyll biomass is often lower in the most turbid (white or very light grey in Figure 7 A) regions in and just to the north and west of the narrows region of the lake. However, in the more nutrient-rich south basin, high chlorophyll biomass co-exists with moderately high turbidity.
Recent surveys have shown an increasing predominance of potentially toxic cyanobacteria taxa (Figure 6). Toxins (microcystins and others) have been detected in phytoplankton net hauls and whole water samples as well as from surface blooms since 1999, particularly in samples collected nearshore. Jones et al. (1998) recorded microcystins in algal blooms in the south basin as early as 1998. Bloom material sampled off Elk Island in the south basin on 26 September 2001 with biomass of ∼40,000 mg m−3 had >200 μg l−1 of microcystins (Herbert and Kling data on file; SOL, 2011, in press; Kling et al., 2007).
There is clear paleological evidence of recent increases in the primary productivity in LW. Survey data support this paleological evidence and further show that it has been accompanied by a shift in phytoplankton population structure consistent with severe eutrophication. Diatom blooms have been greatly reduced in duration and relative abundance during the open water season since 1969. Since 1999, very high surface phytoplankton biomass, often comprised of an excess of 90% cyanobacteria, have developed in mid-late summer. This increase in biomass and shift to cyanobacterial dominance is concurrent with a modeled dramatic increase in P concentrations that took place in the lake in the mid-1990s. Significant toxin levels have been measured in some samples from surface blooms. This has important implications for nearshore food webs, wild life and humans (SOL, 2011, in press). The combined effects of nutrient increases, phytoplankton species shifts and toxin production represent a potential threat to the sustainability of ecosystem function and productivity in Lake Winnipeg.
We dedicate this paper to Dr. Jack Vallentyne for his enthusiastic support of the baseline whole lake surveys of Lake Winnipeg conducted by the Fisheries Research Board of Canada in 1969. Without his leadership and foresight, this historic study of Lake Winnipeg organized and coordinated by Drs. Greg Brunskill and David Schindler may never have taken place. We also acknowledge the support over the years from Fisheries and Oceans Canada, Manitoba Conservation and Manitoba Water Stewardship, the Geological Survey of Canada, Environment Canada, the Canadian Space Agency and the Lake Winnipeg Research Consortium.