Phosphorus loading declined between the 1970s and the 1990s, leading to oligotrophication of the offshore waters of Lake Ontario during that time period. Using lake-wide data from the intensive field years of 2003 and 2008 and from available long-term data sets on several trophic state indicators (total phosphorus [TP], soluble reactive silica [SRSi], chlorophyll a and Secchi disc transparency [SDT]), we tested the hypothesis that oligotrophication of the offshore waters of Lake Ontario has continued in the 2000s. Significant differences between 2003 and 2008 include higher spring (April) TP, SRSi, and SDT in 2008, lower summer (July–August) SDT in 2008, higher summer chlorophyll a in 2008, and lower fall (September) TP, SRSi, and chlorophyll a in 2008. The decline in SRSi from spring to summer was greater in 2008 than in 2003. Change point and regression analyses on the long-term data revealed no trend in spring TP since 1996, in summer chlorophyll a since 1994, in spring SDT since 1998, in spring SRSi or SRSi decline from spring to summer since 1999, or in summer SDT since 2001. Neither the comparison of the 2003 and 2008 surveys nor the analysis of the long-term data supported our hypothesis of continued oligotrophication of the offshore of Lake Ontario in the 2000s.
Lake Ontario, one of the Laurentian Great Lakes, is a large (18,960 km2), deep (mean depth 86 m; maximum depth 244 m) lake bordering the Province of Ontario to the north and west, and New York State to the south and east. The lake has undergone drastic ecological change in response to anthropogenic stressors (Mills et al., 2003). Cultural eutrophication resulting from phosphorus inputs from the watershed led to the 1972 Great Lakes Water Quality Agreement (GLWQA; revised in 1978, and amended in 1983, 1987 and 2012) between the United States and Canada, an agreement that demonstrates a commitment from both countries to address problems related to the chemical, physical, and biological integrity of the Great Lakes basin ecosystem. A major aim of the GLWQA was to reverse the effects of cultural eutrophication (Vollenweider et al., 1974; Schelske, 1991). A phosphorus abatement program was implemented, and target phosphorus concentrations were established for each of the Great Lakes. Implementation of GLWQA led to a decrease in phosphorus concentrations and to the oligotrophication of offshore waters in the Great Lakes (Environment Canada and the United States Environmental Protection Agency, 2014) including Lake Ontario (Millard et al., 2003; Mills et al., 2003; Munawar and Munawar, 2003; Holeck et al., 2008; Dove, 2009). Oligotrophication is defined here as the combined ecological response of a lake to decreased nutrient loading (Anderson et al., 2005), which may include decreased algal production, decreased secondary production, and increased water transparency (Mills et al. 2003). The target TP concentration of 10 μg l-1 for Lake Ontario's offshore waters was attained in 1986 (Mills et al., 2003) and has remained below that level since that time (Dove, 2009; Holeck et al., 2013).
However, problems related to eutrophic conditions, particularly nuisance Cladophora growth and cyanobacteria blooms in embayments and shoreside (depth <1.2 m) areas, have returned (Makarewicz et al., 2012). This has led to a re-evaluation of targets for phosphorus loadings as called for in the GLWQA Protocol signed in 2012. At the same time, continued oligotrophication in the offshore is a concern to fisheries managers since, among the possible impacts, decreased phosphorus concentrations have been implicated in the decline of Alewife (Alosa pseudoharengus) and the multi-million dollar sport fisheries in Lakes Michigan and Huron (Bunnell et al., 2014, Barbiero et al., 2009, 2012). These processes — eutrophication of the nearshore and oligotrophication of the offshore waters — may be the combined response of an overall decline in phosphorus loading to the lake coupled with increased localized phosphorus input to some nearshore areas and an increased abundance of Dreissena Mussels that led to a retention of phosphorus in the nearshore (Mills et al., 2003; Hecky et al., 2004; Makarewicz et al., 2012).
Simultaneous nearshore eutrophication and offshore oligotrophication present a special problem for lake managers since measures taken to reverse one of those processes could exacerbate the other. For example, reducing phosphorus inputs to the lake might improve water quality in the nearshore but could cause phosphorus declines in already nutrient-poor areas of the offshore. It is therefore important to evaluate if the oligotrophication of the offshore waters that was initiated by the GLWQA has continued in the offshore of Lake Ontario in the 2000s. Here, we evaluate several water quality parameters to help determine if the process of oligotrophication that began in the 1970s has continued in the offshore waters of Lake Ontario.
Materials and methods
During 2003 and 2008, the United States Environmental Protection Agency's (EPA) R/V Lake Guardian and the Canadian Coast Guard Ship Limnos performed three lake-wide surveys assessing the lower trophic levels in Lake Ontario (Figure 1, the Lake Ontario Lower food web Assessment [LOLA] project). Timing of the spring cruises was similar for the two years (4/28 – 5/1, 2003 and 4/21 – 4/24, 2008), but the timing of the summer and fall cruises differed both by date and by the seasonal development of surface temperature (summer 7/20–7/26, 2003 and 8/10, 8/11, 8/19–8/22, 2008; fall 9/2–9/5, 2003 and 9/21–9/25, 2008; Figure 2). Both the summer and fall cruises in 2008 were up to 3 weeks earlier than the corresponding 2003 cruises. Station locations and depths are in Rudstam et al. (2012).
Data collected include total phosphorus (TP), soluble reactive silica (SRSi), chlorophyll a, and Secchi depth transparency (SDT). TP, SRSi, and chlorophyll a were measured from integrated water samples in the epilimnion. The samples were collected either with an integrated tube (Limnos) or by pooling several discrete Niskin bottle samples (Lake Guardian) collected through the epilimnion. During stratified periods, samples were collected from the surface to a depth of 1 m above the top of the thermocline (defined as the depth where the rate of temperature decline increases relative to the mixed layer). During spring isothermal conditions, samples were collected from the surface to 20 m depth or to two meters above the bottom (for stations <20 m depth).
TP and SRSi were processed using an autoanalyzer at the Environment Canada laboratory in Burlington, Ontario (2003 samples) and at the EPA certified laboratory at SUNY-Brockport (J. Makarewicz, 2008 samples) using standard methods (APHA, 1998; Wetzel and Likens, 2000). Total phosphorus was determined using the ammonium molybdate – stannous chloride method after preservation with 1 mL 30% H2SO4 (per 100 mL sample) and persulfate digestion (APHA, 4500-P D). For SRSi, water was filtered through a 0.45-μm membrane filter and SRSi concentration was determined by the heteropoly – blue method (APHA, 4500-SiO2 D). Chlorophyll a was determined by acetone extraction after filtration through GF/C (nominal pore size 1.2 μm) glass fiber filters followed by spectrophotometry (2003) or fluorometry (2008, Turner 10-AU unit). Values for total chlorophyll a were not corrected for phaeophytins. Detection limits were TP: 0.2 μg l−1 (2003), 1.2 μg l−1 (2008); SRSi: 20 μg l−1 (2003), 50 μg l−1 (2008); chlorophyll a 0.5 μg l−1 (2003 and 2008). When concentrations were below the detection limit for any parameter, the detection limit for that parameter was used to calculate means and variability. When detection limits varied between years, the higher detection limit was used to calculate means for comparison between years. The mean of the deviations (difference between duplicates/mean) was 20% or less with the exception of TP in 2003 (48%). Replicates were averaged and mean values were used in all subsequent analyses. More details and all data are available electronically through the Knowledge Network for Biocomplexity (Rudstam et al., 2012).
2003 vs 2008
Samples were obtained from at least 8 and up to 17 stations depending on the season and year (Figure 1). For each parameter (TP, SRSi, chlorophyll a, and SDT) t-tests with unequal variance were used to compare 2003 and 2008 data. Differences were considered significant at the p < 0.05 level. Significance values were not corrected for multiple comparisons (see discussion in Gotelli and Ellison  about the issues associated with applying Bonferroni adjustments for multiple tests).
Comparison with long-term data
Four existing Lake Ontario water quality data sets were used to compare trends to the LOLA data: the Department of Fisheries and Oceans Canada (DFO) Bioindex Program (1981–1995) sampled two stations (midlake station 41 and Kingston Basin station 81) on a biweekly basis from the beginning of April until the end of October (Johannsson et al., 1998; Johannsson, 2003; Millard et al., 2003); the Environment Canada Great Lakes Surveillance Program (EC-GLSP; 1969 – present) conducts whole-lake water quality surveys in April and August approximately every second year (Dove, 2009); the United States Environmental Protection Agency (EPA) Great Lakes National Program Office's (GLNPO) limnology program (1986 – present) samples eight offshore stations in April and August of each year (Great Lakes Environmental Database [GLENDA]); and the US Biomonitoring Program (US-BMP, 1995 – present), which is a collaboration between the New York State Department of Environmental Conservation (NYSDEC), the United States Geological Survey (USGS), the United States Fish and Wildlife Service (USFWS) and Cornell University, samples between 3 and 21 offshore stations 1–3 times per year (Holeck et al., 2013). Detailed methods for the long-term data can be found in Dove et al. (2009, EC-GLSP), Holeck et al. (2013, US-BMP), Johannsson et al. (1998, DFO Bioindex Program), and EPAs GLENDA database (GLNPO, 2012). To analyze trends when data were available from more than one sampling program, we calculated a simple mean of offshore data from those sampling programs. Portions of these data series have been published (Hall et al., 2003; Holeck et al., 2008; Dove, 2009). Here we extend the analyses to 2011 (2010 for SRSi) and combine data from different programs. Change point analyses were performed on long-term data using Change-Point Analyzer version 2.3 (Taylor Enterprises, 2003). Change point analysis detects changes in time-ordered data and provides confidence levels and intervals for each change. Time trends were assessed using linear regression on 1) all available years, and 2) years after the most recent change point for each parameter.
In 2008, offshore TP concentrations were below the target of 10 μg l−1 established in the GLWQA (Table 1). Spring TP was significantly higher in 2008 than in 2003 in the offshore. Summer and fall TP were significantly lower in 2008 compared to 2003. The long-term trend of spring TP indicates a decline from 1970 until the mid-1990s, with no significant trend since that time (Table 2; Figure 3). A change point analysis on spring TP showed a break point in 1996, representing a change in the mean spring TP concentration from 9.9 μg l−1 for the period 1985–1995 to 6.1 μg l−1 for the period 1996–2011. There has been no further decline in spring TP since 1996 (r2 = 0.01, n = 16, p = 0.71).
Although offshore spring chlorophyll a concentrations were similar in 2003 (1.35 μg l−1) and 2008 (1.26 μg l−1), summer chlorophyll a levels in 2008 (3.19 μg l−1) were about twice the values measured in 2003 (1.48 μg l−1, Table 1). Fall offshore chlorophyll a levels were higher than summer values in 2003 and lower than summer values in 2008. Offshore epilimnetic chlorophyll a concentrations declined significantly from 1981 to 2011 (Figure 4; r2 = 0.15, n = 31, p = 0.029). However, chlorophyll a concentrations are more variable among the different data series than TP, and there is greater interannual variability. Sources of variability in chlorophyll a include methodological differences among laboratories as well as difference in timing of sampling. For the long-term dataset, there was a change point in chlorophyll a in 1994 (Table 2) and no trend thereafter (r2 = 0.03, n = 18, p = 0.51).
Mean SDT in the offshore was greater in the spring of 2008 (14.9 m) than in 2003 (10.0 m, Table 1). Mean summer SDT in the offshore was shallower in 2008 (6.8 m) than 2003 (8.9 m) which was consistent with the difference in chlorophyll a (higher chlorophyll a = lower SDT). Long-term SDT has increased in Lake Ontario in both spring and summer (Table 2; Figure 5). SDTs from the 2000s are roughly twice those measured in the 1980s (Figure 5), tracking a substantial increase in water transparency in Lake Ontario. Change point analyses indicate a break in spring SDT in 1998 and in summer SDT in 2001, with higher SDT recently. Spring SDT shows a marginally significant increase since 1998 (r2 = 0.19, n = 12, p = 0.10), but summer SDT indicates no recent increase (2001–2011; r2 = 0.04, n = 10, p = 0.59).
Soluble reactive silica concentrations were higher in the spring than summer, an observation consistent with the typical seasonal pattern in Lake Ontario (Millard et al., 2003). Offshore spring SRSi concentrations were significantly lower in 2003 (793 μg SiO2 l−1) than in 2008 (868 μg SiO2 l−1) although the difference was less than 10% (Table 1). Offshore summer SRSi values decreased from spring values (see above) to 190 and 155 μg l−1 in 2003 and 2008, respectively. Fall SRSi concentrations were significantly lower in 2008 than in 2003 (Table 1). The increase in spring SRSi from 2003 to 2008 is consistent with the long-term trend (Figure 6). Spring silica concentrations have increased significantly between 1986 and 2010 (r2 = 0.70, n = 23, p < 0.0001). There has been no concomitant increase in summer silica concentrations (r2 = 0.01, n = 23, p = 0.58). The decline in SRSi in the upper waters between spring and summer (defined as silica “utilization” by Mida et al., 2010) has increased in Lake Ontario over the last 25 years (r2 = 0.62, n = 23, p < 0.0001; Figure 6, Table 2). There is a change point in silica “utilization” in the year 1999 (Table 2), and no significant trend thereafter (r2 = 0.04, n = 12, p = 0.52).
The LOLA 2003 and 2008 programs were designed to investigate the degree of change in lower trophic levels in Lake Ontario. Here we compared nutrients (P and Si), chlorophyll a, and water transparency measures from the LOLA 2003 and 2008 programs to assess changes in water quality and trophic state of the lake between these two years. We also compare the results of the LOLA surveys with the long-term data available through the different agencies around the lake. We were particularly interested in whether the oligotrophication of the offshore described previously for the period from the 1970s to the early 2000s (Mills et al., 2003; Holeck et al., 2008; Dove, 2009) has continued during the most recent decade. We used total phosphorus (TP), chlorophyll a, and Secchi disc transparency (SDT) as indicators of water quality and trophic state (Carlson, 1977; Wetzel, 2001). These indicators are typically correlated, at least in systems where primary production is phosphorus-limited. Other companion papers in this volume present data on phytoplankton biomass and productivity (Munawar et al., 2015), zooplankton biomass and community composition (Rudstam et al., 2015) and benthos abundance (Birkett et al., 2015).
Contrary to our expectations, the results indicate no further oligotrophication in the offshore in the 2000s. The comparisons of trophic levels indicators measured during the LOLA 2003 and 2008 surveys suggest a more mesotrophic offshore in 2008 than in 2003. Spring TP, summer chlorophyll a and spring to summer silica “utilization” were higher in 2008. Summer and fall water clarity (SDT) were lower in 2008. Only spring SRSi and SDT suggested continued oligotrophication, and these indicators are discussed further below. Higher summer chlorophyll a values in our data are consistent with independent measurements by Twiss et al. (2012). Howell et al. (2012) and Makarewicz et al. (2012) also measured relatively high chlorophyll a levels in 2008 at nearshore locations on both the north and south shores of the lake (north shore average 3.1μg l−1; south shore average 2.9 μg l−1). The three long-term monitoring programs (EPA-GLNPO, EC-GLSP, and US-BMP) also show higher summer chlorophyll a levels in 2008 than in 2003 (Figure 4), although the difference between the two years is smaller in these three data series. Our results are also consistent with the assessment of Munawar et al. (2015) based on phytoplankton biomass composition and primary production data. According to the scale of Munawar and Munawar (1982), Lake Ontario was mesotrophic in summer 2008 (2–3 g m−3) (Munawar et al., 2015). These combined results suggest that Lake Ontario's offshore waters were more mesotrophic in 2008 compared to 2003, which does not support continuing oligotrophication of Lake Ontario's offshore in the 2000s.
Although the long-term phosphorus and chlorophyll a data do not support a trend toward mesotrophy, they do suggest that oligotrophication did not continue past 1999. Offshore spring TP concentrations leveled off after 1995 (Table 2; Figure 5) and have remained between 7 and 10 μg l−1 throughout the 2000s. Likewise, other indicators (spring and summer STD, summer chlorophyll a, spring SRSi) showed no significant change with time since the last change point (Table 2). These change points all occurred between 1994 and 2001. Similarly, Holeck et al. (2013) found no significant trends in spring TP in Lake Ontario's nearshore from 1995–2012 in data from the US-BMP. Likewise, Reavie et al. (2014) did not detect a decline in phytoplankton biovolume from 2001 to 2011 in the GLNPO – EPA data, and there was no decline in satellite derived surface chlorophyll concentrations in the 2000s (Watkins et al., 2013, Barbiero et al., 2014). The change point in TP in 1996 coincides with the time Zebra Mussels (Dreissena polymorpha) became firmly established in nearshore areas and Quagga Mussels (D. rostriformis bugensis) began expanding into offshore areas of Lake Ontario (Mills et al., 1999; Watkins et al., 2007; Birkett et al., 2015). However a link between the expansion of Mussel biomass and an end to oligotrophication is in contrast to the hypothesized increase in retention of phosphorus in the nearshore by these Mussels (Hecky et al., 2004). The apparent end to the process of oligotrophication in Lake Ontario's offshore may be attributable to the fact that phosphorus concentrations have stabilized there.
Analysis of the long-term Lake Ontario silica data showed a significant increase in spring silica concentrations but not in summer concentrations (Table 2; Figure 6). Dove (2009) warned that increasing spring silica concentrations signal a declining diatom population, since Schelske et al. (1986) noted an inverse relationship exists between spring silica concentrations and diatom biomass in the Great Lakes. However, there was no further increase in spring SRSi after 1999, and epilimnetic silica was depleted by the summer throughout the time series. Because silica is taken up by diatoms between early spring (April) and summer (July–August), the decline in silica concentration between spring and summer has been used as an indicator of diatom abundance (Conley et al., 1993; Schelske et al., 1986; Mida et al., 2010). In this context, silica utilization in Lake Ontario increased in the long-term data series as there was an increase in spring SRSi without a concomitant increase in summer SRSi levels. Thus, the SRSi utilization data suggest no further decline in diatoms through the 2000s and may even indicate an increase in diatoms over time. These results are consistent with observations in Lake Ontario of higher diatom biomass in 2008 than in any other year previously measured (1970, 1978 and 2003; Munawar et al., 2015), and with observations of no trend in diatom biovolume from 2001 to 2011 in the EPA-GLNPO data (Reavie et al., 2014).
Secchi depth transparency (SDT) is the only indicator of water quality and lake trophic state presented here that suggest continued oligotrophication of the offshore in the 2000s. Both spring and summer SDT were higher in the 2000s than in previous decades (spring mean SDT 8.2 m prior to 1998 and 13.6 m thereafter; summer mean SDT 5.1 m prior to 2001 and 8.4 m thereafter), there was a marginally significant increase in spring SDT since 1998 (p = 0.10, Table 2), and spring SDT was significantly greater in 2008 than in 2003. Note though that summer and fall SDT declined from 2003 to 2008. Further, SDT is also affected by various inorganic particles in the water column and may therefore not always be a good indicator of lake trophic state. For example, Secchi depth was shallower in the fall of 2008 (5.0 m) than in the summer, despite lower fall chlorophyll a concentrations (1.7 μg l−1), likely due to carbonate precipitation (whiting event) observed at the time (Peng and Effler, 2011; Watkins et al., 2013).
The lack of continuing oligotrophication in Lake Ontario in the last decade is likely due to Lake Ontario's location downstream from Lake Erie. The Niagara River, which connects the two water bodies, contributes approximately 27% of the phosphorus load in Lake Ontario (Chapra and Dolan, 2012), and Lake Erie TP has been increasing over the last several years (Environment Canada and the US Environmental Protection Agency, 2014; Dolan and Chapra, 2012; Scavia et al., 2014). If this trend continues in Lake Erie, we may expect increases in phosphorus concentrations in Lake Ontario in the future. This has implications for ecosystem structure and function, and fisheries management.
Comparisons of lake-wide surveys in of Lake Ontario in 2003 and 2008 showed higher phosphorus and chlorophyll a concentrations in 2008 than in 2003. Long-term data series indicate change points in Secchi depth, total phosphorus, chlorophyll a, spring silica, and silica utilization in the mid-to late-1990s, significant oligotrophication prior to that time, and no significant changes since. Our analysis suggests that the oligotrophication of the offshore waters of Lake Ontario has not continued into the 2000s.
We thank the crews of the USEPA R/V Lake Guardian and CCGS Limnos as well as the participants in the LOLA research cruises and contributors of the long-term data sets used for comparisons with the LOLA surveys. We thank Michael Twiss, Edward Mills, Ora Johannsson and the anonymous reviewers for their helpful comments, and Alice Dove for her valuable contributions to earlier versions of this manuscript. The opinions presented herein are those of the authors and do not necessarily represent the position of EPA or USGS. This is contribution 1900 of the USGS Great Lakes Science Center. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. or Canadian Governments.
This work was funded by EPA Grant 97220700-0 to Cornell University as part of the Great Lakes Restoration Initiative and a grant from the International Joint Commission. Further support is provided by the agencies collaborating in the lower trophic level assessments of Lake Ontario (EPA GNLPO and Region 2, NYSDEC, USGS, USFWS, DFO, EC, OMNR).