Lake Winnipeg supports the largest commercial fishery on Canadian Prairies. It has been influenced by a variety of environmental forces and anthropogenic activities. To gain a better understanding of recent changes in nutrient status of the lake, it is important to reconstruct its previous history from sedimentary records. Lacustrine sediments are known to be an important sink of many dissolved and suspended substances, including phosphorus, hence, they provide a permanent historical record of changes occurring in the lake. These changes may be induced by natural factors or by anthropogenic activities in the watershed. Phosphorus profiles from dated sediment cores collected in 1999 and 1994 from the South Basin of Lake Winnipeg were investigated to determine phosphorus enrichment in recent sediments. To interpret the nutrient status and depositional conditions responsible for the trends in total phosphorus, three operationally defined forms of phosphorus (P) were determined: non-apatite inorganic P, apatite P, and organic P. Significant increases in sediment phosphorus concentrations were observed in the uppermost 20 cm of the cores and several anomalies were observed at depth. A doubling in total phosphorus relative to aluminum over the last fifty years is largely due to increases in the non-apatite inorganic fraction, suggesting that much of sedimentary phosphorus increase is attributable to changes in the nutrient status of the water column related to anthropogenic inputs. Organic phosphorus exhibits a subtle increase in the upper 20 cm of the gravity cores, likely due to increases in the primary productivity of the lake. Except for the slight increase in deeper sediments, apatite phosphorus, which is thought to be of detrital origin, remained fairly constant over the length of the cores. Anomalous spikes in phosphorus concentrations deeper in the cores, comprised mainly of the non-apatite inorganic phosphorus fraction, likely resulted from natural variation in local oxidizing conditions, possibly induced by changes in water circulation and/or changes in sediment deposition rates due to climatic variation. The results of this investigation contribute to increased understanding of the depositional history of phosphorus in the lake over the last millennium.
Lake Winnipeg is a major feature on the North American landscape. The lake, with an area 25% larger than Lake Ontario (Todd, 1996), is crucial for hydroelectric, fisheries, and tourism industries. Like other Great Lakes, Lake Winnipeg is exposed to a diversity of anthropogenic impacts including increased nutrient and contaminant inputs, increased fishing pressures, altered water budget, water level regulation, and exotic species invasion. In spite of several limnological studies (Brunskill et al., 1980; Hecky et al., 1986; Patalas and Salki, 1992) and fisheries studies (Franzin et al., 2003) over the last few decades, Lake Winnipeg remains one of the most poorly understood great lakes of the world with respect to its food web structure and function. In 1994, the Lake Winnipeg Project, a joint research program of the Geological Survey of Canada (GSC), Manitoba Energy and Mines, the Freshwater Institute (FWI) in Winnipeg and several university researchers was initiated to enhance the understanding of regional geological history of Lake Winnipeg (Todd and Lewis, 1996). In response to the 1997 Red River flood, a large-scale research program was undertaken by several federal and provincial agencies (Simpson et al., 2003) to assess the flood history and basin evolution of the lake. Sediment coring in South Basin of the lake was carried out on both occasions. Because sediments accumulate materials over time, they are excellent chronicles of the lake's history, providing insights into issues such as nutrient status, climate history and shoreline erosion.
Lacustrine sediments are known to be important storage compartments of many dissolved and suspended substances, including phosphorus (P), hence they provide a valuable record of changes occurring in the lake, whether induced by natural factors or by anthropogenic activities in the watershed. Phosphorus concentrations in sediments, in combination with other geochemical information, are especially useful for interpreting the nutrient status of the lake over its history (Massaferro et al., 1999; Yiyong et al., 2001). In particular, sediment P speciation is valuable for the understanding of past nutrient histories of lakes (Williams and Mayer, 1972; Williams et al., 1976a; Sondergaard et al., 1996, 1993). Sequential extractions (Williams et al., 1976b; Psenner et al., 1984; Sondergaard et al., 1996) have been used to accomplish differentiation of forms of phosphorus in sediments. For example, (Williams et al. 1976a, b) and Williams and Mayer (1972) used sequential extraction to subdivide total sedimentary phosphorus (TP) into three major categories: non-apatite inorganic phosphorus (NAI-P), apatite P (AP), and organic P (OP). As each form represents a unique source, sediments containing sequences of these forms can be used to delineate past conditions; using this approach these investigators were able to interpret trophic changes in the histories of Lakes Erie and Ontario.
The purpose of this study was to examine changes in sediment P concentrations to augment the understanding of its depositional history in Lake Winnipeg over the last millennium. This was accomplished by interpreting the distribution of forms of P in their depositional contexts.
Lake Winnipeg, located in central North America, is the fourth largest lake in Canada and tenth largest freshwater lake in the world, having an area of 24,387 km2 (Herdendorf, 1990). The lake, which is a remnant of the prehistoric glacial Lake Agassiz, is situated in southern Manitoba, Canada, north of the city of Winnipeg (population ca. 620,000; Evans, 2000). The lake straddles the contact between Paleozoic carbonate rocks to the west and Precambrian igneous and metamorphic rocks of the Canadian Shield to the east. A smaller south basin, approximately 90 by 40 km, is separated from a north basin, that measures approximately 240 by 100 km, by a connecting passage known as The Narrows. The glacially-scoured depression in the rocks underlying the lake, with a depth of 50 to 100 m, is largely filled with fine-grained glacial Lake Agassiz sediments, overlain by up to 15 m of post-glacial Lake Winnipeg sediments (Todd and Lewis, 1996). The bathymetry of the lake is shallow and flat, with depths averaging 9 m in the south basin and 16 m in the north basin. The lake has a catchment area extending from the Rocky Mountains to very near Lake Superior, including the basins of the Winnipeg, Saskatchewan, Red, Dauphin and numerous smaller rivers. The south basin receives 80% of its annual water supply from the Winnipeg River, which drains the Precambrian Shield and 20% from the Red River, carved in glacial till (Evans, 2000). The Assiniboine River joins the Red River at the city of Winnipeg. Because much of the drainage basin of the Red River is located in agricultural area, it supplies approximately 70% of the total P load to the south basin (Brunskill et al., 1980; Patalas and Salki, 1992; Evans, 2000; Bourne et al., 2002). Outflow from the lake is through the Nelson River, which flows north to Hudson Bay. Postglacial uplift has caused the north end of the basin to rise relative to the south, causing a gradual rise of lake levels that continues at present (Lewis et al., 2000). The mean annual temperature of the region, recorded at Gimli, is +1.4°C, although the continental climate of the region ranges from hot summers to very cold winters. The mean annual precipitation is 50.8 cm. The lake is ice free from early May to late November (Ice Thickness Climatology, 1992) and ice cover from November to May is consistently thick and continuous. The lake waters are poorly stratified to unstratified (Hecky et al., 1986) and during the summer months water temperatures range from +17°C in the north basin to +19.5°C in the south basin (Todd, 1996), hence bottom sediment temperatures are equally warm in mid-summer. The lake supports a commercial fishery, it is a major destination for recreation, and lake levels are regulated to optimize hydroelectric power generation on the Nelson River.
Sampling and analytical methods
The investigated sediment cores include two gravity cores (cores 4 and 8) collected on the 99-900 cruise and a box core (NAM-7) obtained on the 94-900 cruise, both from the south basin of Lake Winnipeg (Figure 1).
Gravity coring was conducted in calm weather on August 24, 1999 from the Canadian Coast Guard Ship (CCGS) Namao. Fifteen gravity cores were obtained using a coring apparatus that consisted of a 100 kg weight over a 2 m steel pipe with a 10 cm diameter plastic liner tube. The lengths of cores 4 and 8 were 162 and 158 cm, respectively. The cores were analyzed for physical properties and paleomagnetic chronology at the University of Rhode Island, where selected gravity cores were split in half. Later at the GSC Atlantic, the entire working half of cores 4 and 8 were sliced into 1-cm sections, transferred into vials and refrigerated. At the FWI in Winnipeg, a 1 cc subsample was removed for pollen and textural analysis, and the remaining material was freeze dried and pulverized. Dry, homogenized sediments were used for all analyses described herein.
Ten box cores, one of which is examined here, were collected on a cruise of the CCGS Namao in August 1994. The box corer, which consisted of a steel cube with movable base, obtained a half-meter by half-meter cubic sample. Following retrieval of the apparatus, the top of the corer apparatus was removed to expose the surface of the sediment. The enclosed sediments were then carefully subsampled using a tube with a diameter of 10 cm and a length of 30 to 50 cm, with the aid of a vacuum pump to ease the tubes into the sediment without disturbance. Four tubes, approximately 30 cm in length, were obtained from each box core. The cores were extruded using a Teflon plunger, and as sediment emerged from the tube into a clear plastic ring, 1-cm slices were cut off with a stainless steel slicer. The sliced samples were sealed in Whirlpak bags and refrigerated until transfer to the FWI for storage at 4°C. Prior to analysis, the samples were frozen and freeze-dried.
Major and trace element concentrations, including P, Al and Mn, were measured on 99-900 gravity cores 4 and 8 (Simpson and Thorleifson, 2003) with an inductively coupled plasma emission spectrometer (ICP-ES), following digestion using nitric, hydrofluoric and perchloric acids. Results of these P analyses are used here for comparison with TP estimated by ignition, while the results of Al analyses are used for the determination of a P-enrichment factor. Mn concentrations obtained from these analyses are used to examine the covariance of TP and Mn to explain the P trends in deeper sediments. To confirm the P profile, TP content was also determined by ignition at 550°C and subsequent 16 h extraction with 1 N HCl. The coefficient of variation of this method, determined from 60 analyses was less than 5%. Generally, a good agreement was found between the two methods, although TP values determined by ignition were consistently lower than those determined by ICP-ES, the differences being attributed to P occluded in lattices of silicate minerals. This form of P is, however, not environmentally significant and the difference was less than 10%.
Non-apatite inorganic P and AP at selected intervals were determined by sequential extraction described by Williams et al. (1976b). Organic P was determined by the difference between the TP and the sum of the NAI-P and AP concentrations. The NAI-P fraction includes orthophosphate adsorbed on Fe and Al-oxides, Fe and Al minerals such as vivianite or variscite, and Ca-P minerals other than crystalline apatite (Williams et al., 1980). The concentration of NAI-P is generally considered to be a measure of the maximum particulate P that can be rendered soluble during diagenesis (Logan et al., 1979a, b). The AP includes P bound in crystal lattices of apatite grains and is generally considered biologically inert. This form of P is abundant in detrital particles. The OP includes forms of P associated with carbon atoms in C─O─P and C─P bonds.
Recent chronology for the cores discussed here was obtained by Pb-210 and Cs-137 analysis at the FWI (Wilkinson and Simpson, 2003). Chronology for the past 1000 years was supplemented by radiocarbon dating (Telka, 2003) and paleomagnetic analysis (King et al., 2003).
Results and discussion
Twentieth century trends
Total P concentrations in NAM-7 exhibit a gradual increase from about 20 cm, which according to Pb-210 chronology corresponds to the 1930s, to the sediment-water interface (Figure 2). More noticeable increases, at about 12 cm, correspond to the 1960s. The deviations from a smooth trend line may likely be attributed to variation in P loading, resulting from high and low water yields of the Red River. Recent studies (Stainton et al., 2003) have shown that the loading of TP to Lake Winnipeg is closely related (r2 = 0.965) to water yields and the Red River is the major P and sediment source (Bourne et al., 2002). A spike (Figure 2) corresponding to 1976 is likely a result of the very high water yield (222 m3 sec−1, mean annual flow) in 1975, followed by low water yields (51.7 and 43.6 m3 sec−1, mean annual flows) in 1980 and 1981, respectively. Similar trends in TP were also observed in the uppermost 20 cm of cores 4 and 8. However, the trends were more distinct in core 4 and NAM-7 than those observed in core 8. The TP concentrations in the surficial sediments, were almost 20% (200 mg kg−1) lower in core 8 than those in core 4, probably the result of dilution of P bearing autochthonous lacustrine material by the suspended load of largely inorganic sediment (Brigham et al., 1996) associated with the Red River flood event of 1997. Alternatively, deeper mixing of sediment and local disturbances may be contributing factors responsible for the observed differences.
Phosphorus forms also exhibit trends in concentration with depth (Figure 3). In the NAM-7 core, the NAI-P, that includes P from mineralized organic matter, increases from 15 cm to the sediment-water interface, suggesting increasing input of readily available P forms probably from increased productivity. While at a depth of 15 cm the NAI-P comprises about 36% of the TP, the surficial sediments contain as much as 62% of the TP in this form. Below this depth the TP concentrations and the P forms distribution appear to be comparable to Red River sediments. Similar NAI-P trends were observed in the uppermost 20 cm of cores 4 and 8, although the differences in core 8 were less pronounced. While the NAI-P accounts for as much as 40 to 45% of the TP in the surficial sediments in core 4, only 30 to 35% of TP is in this form in surficial sediments of core 8. The proportions and absolute concentrations of various P forms in core 8 were very similar to those observed in suspended sediments from the Red River (Brigham et al., 1996), thus supporting the concept that TP concentrations in core 8 are controlled to a greater extent by the composition of incoming suspended sediment from the Red River. This suggests that even the small distance between the cores is significant enough to be affected by variable distribution of incoming sediments from the Red River.
Within the top 40 cm of cores 4 and 8, OP exhibits minor up-core increases in concentration. However, this trend is not evident in core NAM-7. Increases in OP can only be attributed to increases in the proportion of organic matter in the sediments. The increased organic matter contribution may be a result of the higher productivity in Lake Winnipeg, resulting from increased nutrient loadings. Some of the sediment OP would be of autochthonous origin (derived from the lacustrine organic matter), while some OP would be brought in with the allochthonous organic matter from the watershed. Phosphorus associated with autochthonous organic matter is generally less resistant to diagenetic alteration under oxidizing conditions and would be readily mineralized, hence, determined in the NAI-P fraction. Schelske and Hodell (1995) have noted a similar relationship between NAI-P and primary productivity in Lake Erie. This may explain the relatively uniform profile of OP, with concomitant greater decrease of the NAI-P with depth in core NAM-7. Alternatively, increase of NAI-P in recent years may also be attributed to increased loadings of available P from agricultural and hydraulic activities in the watershed. In contrast, OP incorporated in allochthonous (terrestrial) organic matter is more refractory (less easily degraded) and will be determined in the OP fraction. The decreasing OP trend with depth is particularly evident for the longer cores (cores 4 and 8), where the OP concentrations drop from about 200 mg kg−1 in surficial sediments to half of their values (∼100 mg kg−1) in the deeper sediments (150 and 146 cm, respectively; Figure 4).
Apatite P, which is largely of detrital and erosional origin, shows small down-core increases in all three cores, likely due to the diagenetic formation of apatite in the deeper sediments (Williams and Mayer, 1972). Sediment pore water is generally supersaturated with respect to this mineral because of high concentrations of Ca and PO4, resulting from dissolution of calcium carbonate and reductive dissolution of iron oxides containing adsorbed orthophosphates (Matisoff et al., 1980; Azcue et al., 1996). Apatite is a thermodynamically stable phase in calcareous sediments and its formation is largely controlled by the reaction kinetics (Matisoff et al., 1980). Stumm and Leckie (1970) showed that under normal sedimentary conditions, the formation of hydroxylapatite is very slow, hence, this process could be responsible for the observed small down-core increases in AP. Alternatively, a shift in the supply of detrital material may have caused the observed trend. The down-core increase in apatite P is consistent with a down-core increase in Mg and Ca in cores 4 and 8 (Simpson and Thorleifson, 2003). Magnesium and Ca are proxies for carbonate content in the sediments (Henderson and Last, 1998). Henderson and Last (1998) observed up-core decreases in detrital carbonate over the past 4000 years since south basin Lake Winnipeg sedimentation began. Since the shoreline sediments are richer in carbonate material, and the relative contribution of shoreline erosion has been decreasing as transgression of the lake has progressed over the past 4000 years (Lewis et al., 2000), the influx of detrital carbonate to the lake has been gradually decreasing. The proportions of fluvially derived material, as well as autochthonous lacustrine material have, however, increased during this period. This results in lower sedimentation of detrital carbonates and consequently, in reduced in situ dissolution of carbonates (Kemp et al., 1976). A similar trend was also observed in Lake Erie sediments (Kemp et al., 1976).
Normalization for a conservative element, such as Al, is generally used to assess the P enrichment of recent sediments. The sediment P enrichment was estimated by calculating an Enrichment Factor (EF). Using this approach, the P concentrations in core NAM-7 were normalized for Al concentrations, and compared with corresponding ratios for deeper sediments. This procedure is commonly used to detect changes in sediment P concentrations associated with mineralization of organic matter and dissolution of carbonates and was recommended by Kemp and Thomas (1976) to facilitate separation of soil P associated with Al from the excess P derived from anthropogenic sources (Horowitz, 1991; Kemp and Thomas, 1976). The following equation was used in calculations:Figure 5. Enrichment factor values of ∼ 1 indicate no enrichment, whereas EF values > 1 suggest enrichment. A steady increase in EF to a factor of two was observed in the top 20 cm, providing evidence for input of P from anthropogenic activities.
The P enrichment in recent sediments and the observed trends in sedimentary P forms are likely attributable to higher algal productivity, resulting from enhanced P loadings to Lake Winnipeg in recent years (Bourne et al., 2002). These in turn, may be attributed to increase in loadings from anthropogenic sources, which coincide with increases in anthropogenic activities in the watershed. Anthropogenic sources of nutrients include fertilizers and animal wastes from agricultural activities, sewage treatment plant effluents from urban centers, and industrial effluents. Fertilizer usage in Canada has increased significantly since the 1960s, although most of this increase is from nitrogen fertilizers in the Prairie region. According to Statistics Canada (1977) and CFI (1977–2000), a steady, approximately 4.5-fold increase in fertilizer use in Manitoba, and a nearly 3-fold increase in fertilizer P input occurred between 1967 and 1993 (Figure 6).
Excessive loading of nutrients from fertilizers, in addition to other watershed loadings, can stimulate algal growth that leads to deterioration of the lake's trophic status.
Pre-twentieth century trends
In cores 4 and 8, OP concentrations remain constant (131 ± 6 and 141 ± 8 mg kg−1, respectively) with depth, but increase moderately (50–100 mg kg−1) from a depth of approximately 20 cm to the surface (Figure 3). Increases occurring in close proximity to the onset of intense agriculture (early 1930s; Brunskill et al., 1980) suggest that increases in OP could be the result of anthropogenic activity rather than a natural gradual shift in organic matter supply to the lake.
In addition to increasing P concentrations in the top 20 cm, noticeable peaks in TP were observed in the deeper sediments of the two cores with longer records (Figure 7). However, while three prominent peaks in TP concentrations were observed in the deeper sediments of core 4 (60, 95 and 121 cm), similar P maxima were noticeably absent from the deeper sediments of core 8. According to the chronology model (King et al., 2003; Telka, 2003; Wilkinson and Simpson, 2003), the peaks correspond to approximate ages of 1650 A.D., 1400 A.D. and 1225 A.D., respectively. These peaks are dominantly comprised of increases in the NAI-P form, with little change in the concentrations of OP and AP forms (Figure 4). The P-rich peaks in the lower part of core 4 are also enriched in Mn (Figure 8; Simpson and Thorleifson, 2003), suggesting that the layers were formed under strongly oxidizing conditions, marking an intense episode of mineral precipitation. The precipitated P would increase both the absolute NAI-P concentrations in these layers and the relative contribution of this P form to the TP content. The strong association of NAI-P with Mn suggests that similar processes controlled the precipitation of both. The conditions controlling formation of these layers were likely localized, as the layers are not present at the same depths in both cores 4 and 8. Similar P enrichment concomitant with increases in Mn concentrations was observed in Lake Ontario sediments west of the mouth of Niagara River (Manning and Mayer, 1987). It was suggested that the formation of these layers was related to the current flow pattern. A similar cause may be responsible for the layers of enhanced P and Mn concentrations in Lake Winnipeg. Alternatively, the enriched layers may mark pre-historic intervals of little or no deposition at this site allowing upward migration of Mn and P from reduced sediments and chemical precipitation near the well oxidized surface of the sediments. Reduced burial rates relative to upward diffusion rates will allow these enriched layers to be deposited and then preserved. Depositional conditions favorable for the formation of these enriched layers would likely occur during dry cycles, which alternated with wet cycles throughout the history of Lake Winnipeg.
Chronological data reveal small differences in the sedimentation rates between the two sites. The estimated sedimentation rates in core 4 are slightly higher than those in core 8, providing for higher resolution. Because of the relatively low sedimentation rates in both cores, mixing of surface sediments is an important process affecting the distribution of particles and associated contaminants (Robins, 1982) including P upon their deposition. Sediment P profiles reveal a greater depth of the surficial mixed layer (12–13 cm) in core 8 than in core 4 (∼ 6 cm), suggesting higher ratios of mixing rates to sedimentation rates in core 8 than those in core 4. The higher rates of mixing would more effectively smooth out short-term variations over the mixed layer. This, together with the lower P concentrations observed in core 8 may have caused attenuation of P and Mn peaks.
The trend of increasing AP concentrations with depth in cores 4 and 8 continues over the length of these cores (Figure 4) and, as suggested, is likely the result of the autochthonous apatite P formation and/or shifts in the supply of detrital material. In total, approximately a 10% difference is observed between the AP concentrations of surficial and deep sediments.
The sedimentary phosphorus profiles in the South Basin of Lake Winnipeg show a steady increase in TP concentrations, due largely to recent increases in the NAI-P concentrations. Increases in the NAI-P concentrations suggest that the TP increases are the result of anthropogenic activities in the watershed. The most noticeable increase in P concentrations occurs at depths corresponding to the 1960s, a time of increased usage of fertilizers in the Prairies and the resulting enhanced nutrient loading to the lake. When normalized for a conservative element, Al, the uppermost sediments are enriched in P by a factor of up to two relative to background levels, supporting the premise of the input of P from anthropogenic sources. The subtle increases in OP in recent sediments are likely due to increases in the lake's primary productivity, related to the anthropogenic influx of nutrients. The small increase of AP in deeper sediments may be attributable to the diagenetic formation of apatite and/or to a shift in the supply of detrital carbonate. The three prominent peaks in TP concentrations observed in the deeper sediments of core 4 (60, 95 and 121 cm), but absent from core 8, are likely the result of P and Mn co-precipitation under oxidizing conditions and negligible sedimentation rates. The presence of such layers in only one of the cores suggests that the phenomenon is localized.
We would like to thank Aaron Lawrence and Karuna Ramakrishnan for sample analyses. We also thank the two reviewers for their salient comments and suggestions that improved the final version of the manuscript. This is an NWRI Contribution No. 02-385 and an excerpt from the Geological Survey of Canada, Open File 4196.