Hamilton Harbour was identified by the International Joint Commission as a problem area in 1978 and later identified as an Area of Concern in 1985. In response to its Remedial Action Plan, the harbour has been systematically monitored since 1987. In this study, we present the long-term water quality record (1987 to 2012) of Hamilton Harbour, focusing on the parameters and seasonal intervals of particular importance to the Remedial Action Plan beneficial use impairments. The long-term summer records showed that total phosphorus concentrations decreased in the first decade and have remained relatively unchanged since 1998, while an increasing trend in Secchi disc depth was observed until 2005 only to be reversed since then. No significant changes in chlorophyll a concentrations have been observed since 1987 despite significant changes in total phosphorus. Hypoxia in the hypolimnion of Hamilton Harbour remains a common occurrence and despite long-term trend improvements there has been little change over the last decade. Spring conditions have also changed, and higher conductivity values and chlorophyll a concentrations have been measured in recent years. A strong correlation between spring hypolimnetic dissolved organic carbon concentrations and dissolved oxygen depletion rates was found suggesting the organic material load to the harbour is an important controlling factor for hypoxia. Finally, the data suggest that the hypolimnetic accumulation rate of soluble reactive phosphorus in the summer has increased since 1987, most notably since the early 2000s. However, it is only since 2008 that this translated into an increasing trend in hypolimnetic soluble reactive phosphorus concentrations. These data suggest that conditions have recently changed in Hamilton Harbour and that sediment phosphorus release may delay water quality improvements for many years following reductions in total phosphorus loadings. Further research is needed to understand the implications to the Remedial Action Plan water quality goals.
Hamilton Harbour is one of the 43 Areas of Concern (AOC) that was identified in the 1987 revision of the Great Lakes Water Quality Agreement (GLWQA) (Charlton, 2001). The Remedial Action Plan (RAP) for Hamilton Harbour identified a number of beneficial use impairments (BUIs) in the harbour, including the water quality specific BUIs: Eutrophication or Undesirable Algae and Beach Closings. The monitoring program for Hamilton Harbour was initially designed to specifically address the Eutrophication or Undesirable Algae BUI which identified four interlinked key parameters for which criteria needed to be set to achieve an ultimate remediation goal (Rodgers et al., 1992). These parameters were: (1) total phosphorus (TP) concentrations, which drive, (2) algal biomass (measured as chlorophyll a concentrations), which in turn affects (3) water clarity (measured as Secchi disc depth), which ultimately controls (4) the amount of available aquatic plant habitat (Janus, 1987; Painter et al., 1990).
Within this context, a water quality monitoring program was designed to measure nutrient concentrations, algal biomass and water clarity in the harbour. Additionally, profiles characterized the vertical structure of the water column and the oxygen conditions in the hypolimnion. Water quality measurements to assess the Eutrophication and Undesirable Algae BUI were initially sampled at one center station chosen as a representative of overall harbour conditions and samples were collected at 1m below the surface and 1 to 2 m above bottom sediments (Rodgers et al., 1992). The temporal component of this monitoring program recommended samples be collected on a weekly basis during the months of June to August to address the water quality criteria monitoring in the harbour. In addition to this compliance period, sampling events have occurred in the months flanking this period as often as possible given variable logistical and resource constraints over the years (Hiriart-Baer et al., 2009).
In the mid-1990s, the basic framework of this monitoring program was enhanced to include multiple sampling depths within the water column at the center station. The purpose of these additions was to capture the vertical variability in nutrient and biomass concentrations, information that could not be extracted from the basic profiling units available at the time (M. Charlton, Environment Canada, Burlington, ON, Canada, pers. comm.). This exercise illustrated that the vertical distribution of nutrients and algal biomass were not always uniform within the harbour epilimnion, nor were they consistently higher or lower at 1m compared to other depths. In 2008, this discrete depth sampling approach was enhanced by an integrated water sample with the intent of capturing average conditions within the epilimnion of the harbour and relating those to external nutrient inputs (Hamilton Harbour Remedial Action Plan, 2010b).
Since the beginning of the Hamilton Harbour RAP (HH RAP) in 1987, a number of additional stations have been sampled over the years, although not systematically, to assess the horizontal and vertical variation in water quality conditions around the harbour. The information from this exercise showed that while absolute water quality conditions differ spatially, quantitative shifts are consistent in trend direction among all stations confirming that changes measured at the center station reflect changes across the harbour (Hiriart-Baer et al., 2009).
In the late-2000s the monitoring program was once again modified, this time, to include additional physical, chemical and biological parameters. In the discrete water samples, dissolved organic carbon (DOC) and dissolved inorganic carbon (DIC) were included to capture the role of carbon inputs on the oxygen dynamics in the harbour hypolimnion while particulate organic carbon (POC) and particulate organic nitrogen (PON) were included as coarse biological indicators of plankton nutrient status. In addition, new sensors have been added to the profiling units to better inform on the underwater biomass distribution and light climate.
This article presents an update on the spatio-temporal water quality trends in Hamilton Harbour, and examines the parameter relationships that drive the remediation agenda of Hamilton Harbour.
Materials and methods
Hamilton Harbour is located in the western end of Lake Ontario (Figure 1). It receives municipal and industrial sewage as well as urban and agricultural runoff delivered from streams such as Red Hill Creek and Grindstone Creek. While water quality has been monitored in Hamilton Harbour since the late 60s, it is only since 1987 that a comprehensive sampling program with regular sampling of center station has occurred. Historically, 270 stations have been sampled in Hamilton Harbour by Environment Canada, for various projects or programs (Figure 1). Today 4 stations are sampled systematically, center station (station 1001) and the three corners of the harbour (stations 9030, 9031 and 9033) to capture both the temporal and spatial variability in harbour water quality conditions. For the purposes of this analysis, data from station 1001 between the years 1987 and 2012, with the exception of 1993, were used to evaluate temporal variability in water quality targets of interest to the HHRAP.
At station 1001, epilimnion (1 m depth and starting in 2008, additional integrated water samples) and hypolimnion (19 to 22 m depth) water samples were collected using a Van dorn water sampler. Water column profiles (Seabird [1989 to 1992], Hydrolab™ H2O [1993 to 2001], YSI 6600 [2002 to 2012]) of oxygen, temperature, pH and conductivity were also taken and Secchi disc depth was determined. Weekly calibrations of profiling equipment and method validation and comparison have been carried out over the 25 year period of analysis, assuring the quality and consistency of the data. Water samples from the surface and hypolimnion waters were returned to the laboratory, samples requiring filtration were filtered on the same day and all samples submitted for various nutrient analyses (phosphorus, nitrogen and carbon) and chl-a. All these analyses were carried out by the CALA accredited National Laboratory for Environmental Testing in Burlington, Ontario (Environment Canada, 1994). Phosphorus (P) concentrations were determined by colorimetry (stannous chloride-molybdate complex) on unfiltered (TP) and filtered (TDP, SRP) lake water samples following acidic persulfate digestion (TP and TDP, only). NH3-Tot concentrations were determined by colorimetry (indophenol blue) on filtered lake water samples and NO3/2 concentrations were determined by colorimetry (azo dye) following a copper-cadmium column reduction (American Public Health Association, 2005). Chl-a concentrations were determined by spectrophotometry following an acetone extraction (Unesco, 1969). Temperature (Temp), dissolved oxygen (DO), pH and conductivity (Cond) measurements at each depth were extracted from the profiles.
Dissolved oxygen depletion and soluble reactive phosphorus accumulation rates
Hypolimnetic DO depletion rates were calculated by selecting the lowest DO concentration in the hypolimnion from each profile collected at station 1001 between 1989 and 2012. The value selected was between 1 m above the sediment to the bottom of the thermocline to minimize errors associated with resuspension due to profiler disruption of the sediments. Using these data, a scatter plot of the lowest DO over time was created in OriginPro 9.0. For each year, a linear regression was then fit through all available data points from the highest selected DO, typically in April or May, to the lowest selected DO concentration, typically in July or August, to determine the slope or DO depletion rate over time. Due to Lake Ontario intrusions, increases in DO were often observed and at times DO depletion rates could not be calculated for the entire stratified period. In these cases, the depletion rates calculated at the beginning of the stratified period were used in the analyses.
Similarly, SRP accumulation rates were calculated for station 1001 between 1987 and 2012 from SRP concentrations measured at 19 to 22 m depth during the stratified period only. Over the years, sample depths have varied due to modifications in the monitoring program, although since 1995, 19 m has been consistently sampled. Using these hypolimnetic concentrations, a scatter plot of SRP values over time was created and a linear regression was fit for each year to determine the slope of the line, interpreted here as an SRP hypolimnetic accumulation rate. During Lake Ontario intrusion events, determined by the DO concentrations, SRP concentrations were excluded from the analyses.
The analysis of the long-term temporal trends was conducted by segmented (piecewise) linear regression using the SegReg free software program (www.waterlog.info/segreg.htm). Since the collection of integrated epilimnetic water samples extends no further then 2008, only the 1m discrete water sample was considered for long-term trend analyses. The segmented regression approach introduces a breakpoint in the data, where appropriate, allowing for a broken or discontinuous line across all data points. In SegReg a total of 7 models can be fit to the data, for example, no breakpoint, two sloped lines or two parallel lines with different Y intercepts. The model of best fit, as well as the breakpoint, is selected by SegReg according to the amount of variance explained by the model and a significance test. For this analysis, the data were separated into one of two seasons: (1) early-season, including data between the months of March and June; and (2) mid-season, encompassing data between the months of June and September. Water quality targets for Hamilton Harbour are based on the relevance of each parameter with respect to the identified BUI, including the seasonal time frame when the respective targets are to be met. Epilimnetic TP and chl-a concentrations as well as water clarity targets were established for the summer growing season (mid-season) and the epilimnetic unionized ammonia (NH3-un) concentration target was established for the pre-growing season (early-season), when concentrations are expected to peak. For each season, data were assessed visually for normality using probability distribution plots and log transformed where necessary to meet the normality assumption of the parametric segmented regression test. All chemical parameters including chl-a required transformation with the exception of NO3/2, whereas conductivity, pH and temperature did not.
A specific model-2 linear regression termed the reduced major axis method was used to determine the relationship between hypolimnetic DO depletion rates and spring hypolimnetic DOC concentrations since both variables are measured with error. The slope was considered significantly different from zero by examining the confidence intervals of the parameter estimate. The correlation coefficient between these and other variables was determined by Pearson correlation. Where appropriate a LOWESS line of best fit was used to demonstrate patterns in the data. The LOWESS line of fit is a model-free locally weighted sequential smoothing approach to identify trends in data (Cleveland, 1979).
Mid-season water quality status and trends
Mid-season surface (1m) TP concentrations have significantly declined since 1987, although, most of the improvements occurred prior to 1999 (Table 1; Figure 2a). Over the last 15 years there have been no significant changes and in 2012 the surface TP concentrations averaged 39.7 ± 3.4 µg l−1, well above the TP target of 20 µg l−1. While the historical record, reflects TP concentrations at 1m depth the integrated epilimnetic water sample is the delisting criterion. The relationship between the discrete and integrated mid-season samples since 2008 shows that these parameters are highly correlated (r = 0.72), the slope of the line is significantly (p < 0.001) different from one and the 1m samples are significantly higher (3.7 ± 0.8n = 140 µg l−1; p < 0.001) than the integrated samples. Despite these differences, the relationship between the discrete and integrated samples provides evidence that the observed long-term temporal trends in the surface water samples reflect the long-term integrated epilimnetic trends.
Mid-season surface water chl-a concentrations have shown no significant change since 1987 and seasonal mean concentrations continue to remain above the water quality target of 10 µg l−1 (Table 1; Figure 2b). The discrete and integrated water samples do not appear to be as comparable as was observed for TP concentrations, so it is unclear whether the long-term trends observed in the surface waters are reflective of the epilimnetic long-term trends.
Mid-season water clarity, measured as Secchi disc depth, has improved since 1987; however, as for TP concentrations, the improvements have not been continuous (Table 1; Figure 2c). An increasing trend in water clarity was observed until 1997 and since then there has been no improvement and Secchi disc depths remain below the target of 2.5 m.
Early-season water quality status and trends
Early-season NH3-un concentrations have decreased since 1987 (Table 1; Figure 3a). Despite a peak in 2010, the water quality criterion of <0.02 mg l−1 was met in 2011 and 2012, and the long-term trend suggests that epilimnetic NH3-un is on the decline. Epilimnetic NH3-tot concentrations have been steadily decreasing since 2003 (Table 1; Figure 3b) following a long period of no change. NO3/2 concentrations showed no change until 1997 when they began to increase in the harbour epilimnion (Figure 3c).
Additional changes in early-season water quality conditions were observed for pH and conductivity; both have increased over historical values since the early 2000s (Table 1; Figure 4). On the other hand, there have been no significant concomitant long-term trends in early-season chl-a concentrations; instead, these data have shown high inter-annual variations. Changes were found in early-season epilimnetic temperature, TP and SRP concentrations which have all shown recent increasing trends since 2007, 1996 and 2000, respectively (Table 1).
Hypolimnetic water quality status and trends
Mid-season, when the water column is stratified, hypolimnetic anoxia remains a chronic issue in the harbour (Figure 5a). Typically, hypolimnetic intrusions from Lake Ontario provide an influx of DO to the harbour one to three times during the stratified period (data not shown); however, the benefits of these intrusions are typically short lived (1 to 2 weeks) and anoxic conditions quickly return. Since 2007, DOC concentrations have been measured alongside other water quality parameters and a strong relationship has been observed between spring hypolimnetic DOC concentrations and seasonal oxygen depletion rates during thermal stratification period (Figure 5b).
Further changes that have occurred in the harbour hypolimnion in recent years are increases in TP and SRP (Figure 6). Of most concern are the increases in SRP since 2007. Between 1987 and 2007, average mid-season hypolimnetic SRP concentrations were 7.7 ± 0.8 µg l−1 while in 2012 these concentrations were 48.3 ± 6.4 µg l−1. Similarly, the ratio of SRP to TP has increased since 2007. Hypolimnetic SRP accumulation rates have increased since 1987, and it would appear that since the year 2000, the rate of increase in SRP accumulation rates has more than doubled (Figure 7).
Large improvements in TP concentrations occurred early on in the remediation process of Hamilton Harbour, with concentration reductions of ca. 50% between 1987 and 1997. However, since then, little to no improvement in TP concentrations has been achieved. This is not entirely unexpected since TP loading estimates for Hamilton Harbour have not greatly decreased since 1992 (Hamilton Harbour Remedial Action Plan, 2010a; Vogt, 1998). in fact, P loads to Lake Ontario as a whole have been increasing since 1999 (Dolan and Chapra, 2012). Between 1999 and 2011, inclusively, discharge to Hamilton Harbour inferred from Grindstone Creek discharge data (Environment Canada, 2014b) increased nearly four-fold from 6205 to 20110; these large changes in hydrological flows may be negating the gains that have been achieved through the technological improvements at wastewater treatment plants (WWTP) and best management practices in the watershed. While hydrological loads are unpredictable, point sources of nutrients to Hamilton Harbour account for an average of 60% of the TP loads (Hamilton Harbour Remedial Action Plan, 2010a; Vogt, 1998) and management of these sources remains important. WWTP upgrades alone are expected to reduce target TP loads by ca. 60% and these reductions are anticipated to improve water quality conditions in the harbour (Hamilton Harbour Remedial Action Plan, 2007, 2010b; Ramin et al., 2012).
Chlorophyll a concentrations, on the other hand, have remained essentially unchanged since 1987 despite the large initial decreases in TP. In freshwater ecosystems, P is often invoked as the factor limiting algal growth (Schindler, 1977); however, it is not the only driver of primary productivity. Research has suggested that in Hamilton Harbour, light attenuation (limitation) may be a more important driver for chl-a production relative to the supply of nutrients, particularly P (Pemberton et al., 2007). This is consistent with the findings of Hiriart-Baer et al. (2009) that showed P limiting growth typically occurs under higher mean light and higher temperature conditions. Given TP concentrations rarely fall below 30 µg l−1, it is reasonable to assume that nutrient supply typically outweighs nutrient demand in this system. Thus, reducing P inputs remains an effective strategy to achieving HH RAP goals despite the uncertainty and variability associated with the other factors in determining algal growth.
Since 1997, there has been no significant improvement in water clarity which may be in part related to an increase in particulate material. Good correlations, as inferred by Pearson correlations, exist between Secchi disc depths and particulate phosphorus (PP; r = −0.49), particulate organic carbon (POC; r = −0.52) and particulate organic nitrogen (PON; r = −0.48) concentrations all surrogates for organic biomass. While segmented regression analysis did not reveal a recent increase in PP concentrations, the only long-term record for particulates available, LOWESS showed an increasing trend since 2005 (data not shown). Other factors such as DOC concentrations may be playing a role in water clarity; however, no long-term record is available.
Unionized NH3 (NH3-un) concentrations have improved significantly over the years, however, there have been some years when concentrations have been unusually low (e.g. 2002) or high (e.g. 2010). NH3-un concentrations depend on NH3-tot concentrations, water pH and temperature and the long-term trends in NH3-un are in fact congruent in time with early season pH and temperature (data not shown); for example, seasonal pH means were lowest in 2002 and highest in 2010. While extremes in temperature were not observed in those years, early-season temperatures have been increasing since 2007 (Table 1) likely in response to increased spring (March, April and May) regional air temperatures which have been observed since 1959 (Environment Canada, 2014a). Increased surface water temperatures can result in earlier growing seasons and increased water column stability which can lead to increases in spring phytoplankton blooms (Stainsby et al., 2011) and associated increases in pH (Kalff, 2001). Continued increases in surface water temperatures and pH may ultimately lead to increases in NH3-un despite decreases in NH3-tot.
Recent declines in NH3-tot concentrations may be related to more recent increases in NO3/2 as a result of increased wastewater treatment efficiencies for nitrogen through the conversion of NH3-tot to NO3/2. Indeed, the relationship between NH3-tot and NO3/2 since 2003, when NH3-tot began to decrease, has a Pearson correlation coefficient of −0.568. However, other sources of NO3/2 in addition to wastewaters, including fertilizers from agricultural runoff may be leading to the nearshore enrichment of NO3/2, particularly in proximity to rivers and streams, a phenomenon common within the Great Lakes basin (Howell, 2012). Furthermore, increases in NO3/2 concentrations in the Great Lakes appear to be a regional phenomenon (http://www.epa.gov/grtlakes/monitoring/limnology/index.html, accessed 3 February 3 2014).
Following the onset of stratification, oxygen depletion begins rapidly and hypoxic conditions typically persist until the fall turnover (Hiriart-Baer et al., 2009). While intermittent increases in DO concentrations often occur at center station due to hypolimnetic intrusions of Lake Ontario (Yerubandi et al., 2016), these reprieves are short lived and hypoxia reestablishes itself within one or two weeks. Since 2008, samples for DOC concentrations have been collected as part of the regular monitoring program. A strong relationship between hypolimnetic DO depletion rates and spring (pre-stratification) hypolimnetic DOC concentrations has been observed with the data collected to date. This strongly suggests that regardless of epilimnetic biomass production, respiration of the bacterial biomass feeding on DOC is an important factor dictating oxygen depletion rates in the hypolimnion although other factors such as nitrification or sediment oxygen demand may also be playing a role (Hamilton Harbour Remedial Action Plan, 2007; Hiriart-Baer et al., 2009).
Despite chronic anoxia, P release from the sediment has not been considered an issue. Historically, the redox potential across the sediment-water interface had not been considered low enough to allow for iron reduction, a necessary step in the release of sediment bound P (Barica, 1989). However, hypolimnetic TP and SRP concentrations have been increasing during the stratified period since 2008 and sediment iron concentrations have been declining steadily since the 1990s (Hamilton Harbour Remedial Action Plan, 2010b). Given the ratio of SRP to TP has increased concurrently, it is reasonable to suggest that the observed increases in SRP are due to sedimentary P release rather than increased resuspension events. This is further substantiated by the increase in SRP accumulation rates calculated for the stratified period.
Water quality conditions have improved significantly in Hamilton Harbour since the late 1980s and further ameliorations are anticipated as both point and non-point nutrient source abatement strategies are implemented. Our data also suggest that additional chemical changes have occurred in Hamilton Harbour as evidenced from the recent increase in P release from the sediments. Water bodies, like Hamilton Harbour, that have been subjected to excess inputs of P over many years have inevitably accumulated a large reservoir of this nutrient in their sediments (Kalff, 2001). Under chronic anoxic conditions, sediments become a source of P and can continue to do so many years after external P loading reductions. Surface sediment P concentrations are high in Hamilton Harbour (average in 2014 at center station: 2.8 mg P g−1 DW, unpublished data) and it is likely that sediment P release will delay the response of Hamilton Harbour to the planned WWTP improvements. Further research into the sediment and hypolimnetic chemical conditions is needed to characterize the net phosphorus release rate from the sediment and its implications to the surface water quality conditions.
We would like to thank Environment Canada Technical Operations and all the summer students throughout the years who have helped collect and process all the water quality data from Hamilton Harbour.