Hydrological conditions and water quality
Mean annual SLR outflow at Québec averaged 12,090 m3 s−1 for the 1995–2011 period which is very close to the value of 11,880 m3 s−1 obtained for the same period using the alternative method of Bourgault and Koutitonsky (1999). Mean monthly discharge showed strong seasonal variations with maximal values in the spring (April–May) and minimum discharge in summer (August–September) (Fig. 3). Lake Ontario discharge to the SLR showed little seasonal (e.g. monthly) variability and, on an annual basis, constituted the bulk of total freshwater outflow at Québec (Fig. 3b). In contrast, Ottawa River discharge exhibited high inter-annual and seasonal variability. Seasonal variations in SLR discharge at Québec were largely driven by the Ottawa River (min–max daily discharge over the sampling period: 1520–2360 m3 s−1) and other tributaries which together accounted for 28% (August) to 55% (April) of total discharge at Québec.
Waters originating from Lake Ontario exhibited very low concentrations of SPM, DOC, TP and chlorophyll a (Table 1). In comparison, consistently higher concentrations of all substances (except DON and NO2–NO3) were recorded at Québec, including SPM (30-fold), TP (2-fold), TDP (2-fold), NH3 (3-fold), DOC (1.3-fold) and chlorophyll a (2.6-fold). High concentrations measured at the mouth of the Ottawa River partially explained the recorded increases at Québec (Table 1).
Temporal variations of C, N, and P concentrations
The temporal variations of concentrations of each water quality variable differed markedly among sites as a function of the morphology of their watershed. At Wolfe Island, none of the water quality variables showed a correlation with discharge owing to the very stable outflow of Lake Ontario. Instead, concentrations of most variables fluctuated on a seasonal and/or inter-annual basis (Table 2) consistent with the cycle of vertical mixing/stratification and year-to-year variations of Lake Ontario waters. In contrast, river discharge and seasonal variations were strong drivers of SPM, DOC and nutrient concentrations in the Ottawa River and at Québec, indicative of the influence of seasonal variations in runoff and temperature over the watershed. Accounting for differences in temporal variations allowed us to select the optimal model to calculate annual loads of each substance at each site.
Time series of NO2–NO3 and TP over several years exemplify the contrast in seasonal/inter-annual patterns at our three sites (Fig. 4; Table 2). At the outlet of Lake Ontario, concentrations of both NO2–NO3 and TP showed seasonal variations with minimum concentrations over summer months and maximum concentrations in winter. In addition, winter and spring (February to May) maxima tended to rise over the last years of the NO2-NO3 and TP series at Wolfe Island, justifying our use of season and year to model their loads at the Lake Ontario outlet (Table 2). For the 1995–2001 period, mean monthly NO2–NO3 concentrations were consistently <0.4 mg N L−1 whereas concentrations >0.4 mg N L−1 occurred for 10 of the 15 months between January 2010 and March 2011(Fig. 4a). For TP, mean monthly concentrations remained <10 µg P L−1 for the 2001–2004 period but rose >20 µg P L−1 for 12 of the 15 months between January 2010 and March 2011 (Fig. 4b). At the mouth of the Ottawa River (Carillon), discharge explained most of the variations in SPM (Spearman r = 0.51, N = 882, p < 0.0001), DOC (r = 0.44, N = 172, p < 0.0001), and TN (r = 0.36, N = 43, p = 0.02) concentrations, whereas concentrations of dissolved inorganic nitrogen species (NO2–NO3 Fig. 4c, and NH3) and chlorophyll a fluctuated seasonally (Table 2). Neither TP nor TDP followed any specific seasonal or long-term (1995–2011) temporal trend (Table 2).
At the SLR outlet at Québec, positive correlations between river discharge and DOC (Spearman r = 0.59, N = 406, p < 0.0001), and TN concentrations (r = 0.51, N = 35, p < 0.0001) were observed, whereas models including discharge and season accounted for most of the temporal variations of SPM, TP, TDP and NO2–NO3 (Table 2). Overall, regardless of major hydrological and morphological differences among sites, all dissolved nitrogen forms, including NH3 (Table 2) and NO2–NO3 (Fig. 4e), followed clear seasonal variations with highest concentrations observed over the coldest months and lowest concentrations in late summer. In contrast, chlorophyll a concentrations were maximal during the spring and summer at all sites. No significant long-term trend was observed for any of the water quality variables (Spearman rank-order correlation).
C, N, and P loads from urban sources
Wastewaters from the 6 largest physical–chemical WWTP represented the treated domestic effluents from about 2.6 million persons—about 4 times the population whose wastewaters were treated in the 44 small WWTP using other treatment methods (Table 3). Treated wastewater discharge and annual loads of SPM, TKN and DOC from the 6 large physical–chemical WWTP were about 2–10 times higher than for the 44 small WWTP. Large physical–chemical plants were more effective than small WWTP in removing TP, whose load was only twice that originating from the 44 small plants. As expected, TKN (i.e. organic N + NH3) represented the bulk of nitrogen loads in treated municipal waters whereas NO2–NO3 only represented a minor component (Table 3).
Loads at St. Lawrence River inlets and outlet
Calculation of annual flux of SPM, C, N, and P from Wolfe Island and Carillon further emphasized the major differences in morphology and watershed characteristics between Lake Ontario and the Ottawa River. In spite of a discharge 3.6 times higher at Wolfe Island than at Carillon, the SPM load originating from Lake Ontario was 4 times smaller than for the Ottawa River (Table 4). Loads of DOC, TP, TDP, NH3 and chlorophyll a originating from the Great Lakes were only about twice those of the Ottawa River. In contrast, NO2–NO3 loads from Lake Ontario were 5 times higher than from the Ottawa River, contributing 58% of the load at Québec. Loads originating from the 50 municipal WWTP were generally small in comparison with those from the main river stem and the Ottawa River with the notable exception of NH3, which represented 51% of the load measured at Québec (Table 4).
At Québec, SPM exhibited a 10-fold increase in comparison with inflowing sources, whereas DOC (1.4-fold), TN (1.5-fold) and TP (2-fold) showed a lesser enrichment (Table 4). Loads of planktonic chlorophyll a were 2.3-fold higher at Québec than at the two major inlets. This information indicates that over its 550 km-long course, the St. Lawrence is enriched by diffuse nutrient sources, including tributaries, erosion and atmospheric sources, which reflect land-use through their individual and area/specific loads.
Diffuse sources and area-specific C, N, and P loads
In spite of the large population inhabiting the Laurentian Great Lakes watershed (≈34 million persons) and intensive land-use, area-specific loads of SPM (0.13 t km−2 year−1), C, N, and P (750, 165 and 4 kg km−2 year−1, respectively) flowing into the SLR at Wolfe Island were very low, reflecting the large proportion of the watershed covered by the lakes themselves acting as a sedimentation basin for particles. Ottawa River area-specific output was about 2-3-fold higher than the Great Lakes for DOC (2650 kg C km−2 year−1), TN (236 kg N km−2 year−1) and TP (11 kg P km−2 year−1) but about 20-fold higher for SPM (2.99 t km−2 year−1, Table 5). Area-specific TN and TP measured at Québec (260 and 12 kg km−2 year−1, respectively) were slightly higher than output values for the Ottawa River watershed.
Load values from Wolfe Island and Carillon contrasted sharply with the inflow from tributaries draining the north and south shores of the SLR (Patoine 2017, Table 5). Taken together, these tributaries annually brought large loads of SPM (about 1.9 million t), DOC (0.46 million t), TN (53,000 t) and TP (3300 t) to the SLR. Area-specific carbon loads between 2.36 (Richelieu River) and 5.90 (Jacques-Cartier River) t C km2 year−1 reflected the proportion of forested watershed and land use. N/C ratio also provided a distinction between largely forested (N/C < 0.1) and intensively farmed watersheds (N/C > 0.1). Farming induced important differences in area-specific loads of SPM and nutrients among tributaries. South shore tributaries flowing through the fertile, intensively farmed (cultivated watershed area: 22–61%, Patoine 2017) St. Lawrence River lowlands contributed about 3 times the loads of TN as those draining less intensively farmed (cultivated watershed area: <1–17%, Patoine 2017) watersheds from the north shore. Of all the tributaries documented, the Yamaska (55% cultivated area) exhibited the highest area-specific output of SPM (42 t SPM km−2 year−1), TN (1579 kg N km−2 year−1) and TP (90 kg P km−2 year−1). Most major south shore tributaries (all but the Richelieu River) exhibited elevated area-specific output of SPM (>20 t SPM km−2 year−1), TN (>500 kg N km−2 year−1) and/or TP (>40 kg P km−2 year−1). In comparison, only two north shore tributaries (Du Nord and L’Assomption rivers) were similarly enriched (Table 5). Noticeably, small tributaries (<1000 km2) from both shores of the SLR carried the highest area-specific loads of SPM, C, N, and P (Table 5) since those watersheds were located almost entirely on the SLR lowlands and supported the highest proportion of farmlands (up to 82% of watershed area).
Annual atmospheric deposition of C, N, and P over the wetted St. Lawrence River surface from Wolfe Island to Québec (2637 km2) were estimated to be in the order of 4480, 1530 and 40 tonnes of C, N, and P, respectively (Table 6), assuming a high retention rate of nutrients and carbon into seldom flooded watershed area. Given a 64% NO3 composition (Ouimet and Duchesne 2009), atmospheric N deposition translated into a direct addition of about 980 tonnes of NO3 to the SLR (0.7% of annual NO2–NO3 load at Québec).
Mass-balance assessment and relative importance of individual sources compared to within-river C, N, and P processing by primary producers
The assessment of mean annual loads originating from various sources (Table 6) allowed us to determine their relative contribution to the total annual flux of SPM, C, N, and P into the SLR estuary at Québec which we present in decreasing order of magnitude. Shoreline and riverbed erosion accounted for the largest part of SPM (65%) and TP (29%) flux to the SLR estuary. By virtue of its large discharge, inflow from Lake Ontario carried 42% of DOC and 47% of TN loads at Québec. The Ottawa River was also a large source of SPM and DOC (8 and 28% of load at Québec, respectively). Although they represented much smaller cumulated watershed areas than the Ottawa River, tributaries draining the north and south shore (6% of watershed area each) contributed equivalent or higher proportions of SPM (8 and 25%, respectively), TN (7 and 13%) and TP (9 and 18%) as originated from the Ottawa River. Treated effluents from 50 municipal WWTP outflowing directly into the SLR contributed a small fraction of DOC (3%), TN (6%) and TP (6%) loads at Québec. Atmospheric deposition represented <1% of total annual C, N, and P flux at Québec. Unaccounted loads represented −4% of SPM, 5% of DOC, 11% of TN and −17% of TP load at Québec (Table 6, line K), revealing a higher level of uncertainty for the nutrient budgets than for SPM and DOC. The negative balance observed for SPM (−4%) and TP (−17%) suggested an overestimation of internal sources, possibly linked to our use of a fixed proportion (65% of SPM load at Québec) for erosion. The positive balance for DOC (5%) and TN (11%) was indicative of unaccounted internal sources such as small and ungauged tributaries, unconnected residences and industries, urban sewer storm overflow and/or other sources.
Annual C, N, and P turnover through primary production of aquatic plants (phytoplankton, epiphytes, submerged and emergent macrophytes) estimated for the entire wetted SLR surface represented a substantial fraction of the overall flux at Québec. Annual aquatic plant production fixed 277,000 t of C and assimilated 49 000 t of N and 7 000 t of P (Table 6, Line L), representing 20, 18 and 58% of annual loads at Québec, respectively. These percentages reveal that riverine plant production taking place over the growth season (May–Sept.) can significantly modify the seasonal nutrient and organic carbon flux to the estuary.
The observation that seasonal growth of aquatic plants could modify the seasonal flux of C, N, and P from the SLR to its estuary was examined further through the monthly SPM, C, N, and P net flux at Québec (Fig. 5) which revealed the major influence of seasonal and inter-annual discharge variations. All water quality variables exhibited peak loads in April, coincident with the spring freshet, followed by receding values during the summer growth season down to a minimum in September which coincides with the time of minimum discharge. Net flux of all substances rose again in the fall corresponding to rising discharge and reduced biological activity. Strong inter-annual variability was also observed for all substances with lower than average net flux over years of low discharge (2010) and maximum flux during high flow years (2008) (Fig. 5). For both TP and DIN, negative net monthly flux were observed for summer months (June–September) over years of low discharge, indicating the SLR acted as a nutrient sink under lowest water level conditions.