Draft Hyperpycnal flows control the persistence and flushing of hypoxic high conductivity bottom water in a High Arctic lake

In the deepest portions of many lakes, zones of high-conductivity bottom water (HCBW) depleted in dissolved oxygen (DO) are present. HCBW and DO are important for determining benthic diversity and abundance, nutrients, and contaminant cycling and understanding the long-term evolution of lakes. We investigate the persistence and removal of HCBW and DO replenishment in a High Arctic lake using physical properties and flow velocity data along with hydrometric and suspended sediment inflow data over a 4 year monitoring period (2007–2010). HCBW was removed in 2007 and 2008 but largely remained in 2009 and 2010. Catchment disturbances in 2007 increased suspended sediment concentrations (SSC) in the inflowing river in 2007 and 2008. In the later two years of monitoring (2009 and 2010), fluvial sediment availability relaxed to pre-disturbance levels. High SSC in 2007 and 2008 caused by landscape disturbances formed sustained river-generated hyperpycnal flows during the snowmelt period that are linked to HCBW remo...


Introduction
Many polar lakes have saline or hypersaline bottom waters, which have profound physical and biological implications (Gibson et al. 2002;Van Hove et al. 2006).In freshwater Arctic lakes, seasonal accumulation of solutes in the bottom waters, here referred to as high conductivity bottom water (HCBW), often occur as a shallow bottom layer, where the solutes are removed or diminished by mixing with inflowing water under ice, or subsequent to ice-off, by convective mixing.HCBW can 1) greatly affect lake dynamics by inhibiting mixing with the overlying water column, 2) promote anoxia in bottom waters, 3) increase water residence time, 4) control water quality through suspension or deposition of metals and methylation of mercury (Bergmann and Welch 1985;Welch and Bergmann 1985;Eckley and Hintelmann 2006), 5) affect biota such as amphipods and char, and 6) help preserve annual lamina in lake-bottom sediments, by limiting bioturbation and bottom currents, which is vital to varve-based paleoenvironmental reconstructions in lakes (Lamoureux and Bradley 1996;Besonen et al. 2008;Jenny et al. 2016).
Potential processes for mixing HCBW in ice covered lakes over short time periods include the development of a nearshore thermal bar where circulation occurs, double diffusion, and hyperpycnal flows (Baehr and DeGrandpre 2002;Bergmann and Welch 1985;Imboden and Wüest 1995;Welch and Bergmann 1985).The causes of preservation and mixing of HCBW are poorly understood and varied (Van Hove et al. 2006).
Lakes have the potential to evolve towards increasing salinity, stability, and isolation of HCBW; alternatively, they can increasingly be flushed and mixed (Dugan and Lamoureux 2011; Van D r a f t Hove et al. 2006).It is particularly important to understand the controls on HCBW formation and loss given that climate change will affect ice cover, inflowing water quantity/quality, including sediment flux, dissolved load, and loads of nutrients and pollutants as well as water temperature (Prowse et al. 2011).The most sensitive HCBW are likely those with low concentrations of dissolved solids relative to overlying water.
Here, we investigate the dynamics of HCBW using four years of measurements from the West Lake and West River at Cape Bounty, Melville Island, Arctic Canada (Fig. 1).This paper aims to improve our understanding of the processes responsible for the preservation and removal of HCBW, and specifically aims to: 1) describe the temporal evolution of HCBW over a four year period; 2) examine the role of landscape disturbances and related sediment loading on the removal of HCBW; 3) determine the physical limnological processes responsible for the preservation or removal of the HCBW; and 4) evaluate the role of projected climate change on the future evolution of HCBW.This study provides a comprehensive temporal source-to-sink perspective on the evolution of HCBW in a freshwater lake, and provides insights into the interactions between lakes and catchment processes, including permafrost degradation and disturbance.

Site Description
Cape Bounty is located on south-central Melville Island, in the Canadian High Arctic (74°55'N, 109°35'W, Fig. 1).An 8.0 km 2 catchment area drains into 1.3 km 2 West Lake in a single major inflowing channel (Fig. 1).The Cape Bounty region is composed of rolling hills with up to 100 m of relief.Slopes are generally gentle and are covered by unconsolidated glacial and marine sediments (Hodgson et al. 1984) Continuous permafrost underlies the landscape to hundreds of metres depth, and active layer depth ranges from 0.5 to 1.0 m.The watershed underwent extensive permafrost disturbance in July 2007 that resulted in over 75 slope failures (active layer detachments (ALDs) (Lamoureux and Lafrenière 2009)).Dissolved solid fluxes and sediment availability increased postdisturbance (Lafrenière and Lamoureux 2013;Lewis et al. 2012).The West River is the primary fluvial input into the West Lake.Runoff is generated by snowmelt, and significant rainfallgenerated runoff events are comparatively rare (Favaro and Lamoureux 2014).Discharge rises quickly to peak flow in June, then recedes as the snow supply wanes, and reaches baseflow in July and August.Flow then ceases for the long winter (Lewis et al. 2012).
West Lake has a maximum depth of 34 m near the front of the West River delta.Swath bathymetric investigations of the lake revealed the presence of bedforms related to hyperpycnal flows (Normandeau et al. 2016).These hyperpycnal flows occur mostly during the snowmelt season, when discharge and suspended sediment concentrations (SSC) are high (Cockburn and Lamoureux 2008).The mooring used in this study is located 470 m from the river mouth while the bedforms related to higher energy hyperpycnal flows are located within 200 m of the river mouth.
The lake largely contains fresh water and is cold monomictic.Lake level varies by less than 1 m per year according to seasonal inflows.Ice cover reaches a maximum thickness of 1.8-2.3m in late winter.Beginning in June, a narrow open water moat forms around the perimeter of the lake, and the lake outlet opens.Lake ice clears in late July or early August, but can be perennial in cooler years.A thick lake ice cover was present during the collection of all the data in this study.

6
HCBW in West Lake is typically present before the snowmelt season begins.Its formation is potentially due to exclusion of salts from the ice cover or from groundwater inputs (Dugan and Lamoureux 2011).HCBW form in a 0.5-2.0m-thick zone in the deepest portion of the lake during winter.At 0.5 m thick, the HCBW covers an area of 41 300 m 2 (0.041 km 2 ) and a volume of 34 300 m 3 while at 2 m-thick, it covers an area of 75 750 m 2 (0.076 km 2 ) and a volume of 122 400 m 3 (Fig. 1C).Bottom water conductivity is high (specific conductance greater than about 75 µS/cm) relative to the overlying water column, but is insufficient to be considered brackish (Ibanez et al. 2010).In some years, HCBW disappear from the lake bottom during the snowmelt season, and in some years it persists; the reasons for this variation are explored in this paper.

Materials and Methods
To investigate the controls on HCBW, stations recorded river discharge, conductivity, suspended sediment concentrations, and meteorology.In the lake, CTDs (Conductivity, Temperature/Turbidity, Depth recorders) were used to obtain water column profiles and also moored in the deepest portion of the lake within HCBW to record continuous near-bottom conditions (Fig. 1).An acoustic Doppler current profiler (ADCP) was also moored in the same location for two snowmelt seasons.Methods and results presented here pertain to the 2007-2010 monitoring seasons.

Hydrometeorology
The hydrometric station is located 0.2 km from the West Lake inlet on the main river channel (Fig. 1B).Hydrometric methods have been published elsewhere (e.g.An ISCO 3700C automatic pump sampler obtained 500 mL samples of river water every three hours.Water samples were volumetrically filtered through pre-weighed 1.0 µm glass fibre filters, which were freeze dried and reweighed to determine suspended sediment concentration (SSC).
Sediment fluxes (suspended sediment discharge, SSQ) were calculated from SSC and river discharge.
Calculation of salinity for low conductivity lake water is problematic (Boehrer and Schultze 2009).However, conductivity is sufficiently low and invariable at Cape Bounty that it has a minor influence on density (Results section).Uncertainty in the calculation of density is minimized by our emphasis on density differences rather than absolute values.Salinity was calculated using PSS-78 (Fofonoff and Millard 1983), and density was calculated using equations suitable for fresh water (Boehrer and Schultze 2009).
Meteorological variables were measured 2.8 km from the lake at the MainMet station at 85 m asl (Fig. 1B), where air temperature, precipitation, and wind speed and direction are measured (please refer to Lewis et al. 2012 for more details).

Lacustrine monitoring
Lake water temperature, turbidity, specific conductance (SpC), and dissolved oxygen (DO) were logged using RBR XR-420 CTDs.From 2008 to 2010, a CTD was moored with sensors 0.2 m above the lake bottom (z-0.2;Fig. 1C) within the zone of HCBW, and recorded data every minute.A second CTD was used for profiling from the ice cover 2 m away, and logged every Conductivity was measured with an AMT 3 electrode sensor, with an accuracy of ±3 µS cm -1 .
Specific conductance was calculated using 25 o C as a reference temperature (Boehrer and Schultze 2009).Water density was calculated using the techniques described in the hydrometeorology section.In 2007, one CTD was used; it was moored at the lake bottom and temporarily removed for water column profiling.
Bottom currents within and above the HCBW was measured using moored Nortek Aquadopp acoustic ADCP adjacent to the moored CTD.ADCPs were available for the 2008 and 2010 monitoring seasons.In 2008, a 2 MHz system was used (2.9 m profile, deployed from 4 June to 16 July), and a 1 MHz system was used in 2010 (5.9 m profile maximum, deployed from 12 June to 14 July).In both years the ADCP was deployed horizontally on the lake bottom with upwardlooking transducers.The bottommost current velocity measurements were between 0.16 and 0.475 m above the lake bottom (the sum of the instrument diameter and the blanking distance; Nortek AS 2013).Depending on the deployment, cell size was between 0.03 and 0.05 m, and measurements were obtained every two minutes, with a 60 second averaging interval.The ADCPs were used in an extended velocity range mode that allows very low velocity flows to be recorded.In practice, resolution is about 0.001 m/s (Siegel, personal communication 2009).
However, accuracy depends on the strength of acoustic signals reflected by suspended particles in the water, which is measured as signal amplitude.Accuracy also depends on the change between emitted and received acoustic signals, referred to as correlation.Data with low D r a f t 9 amplitude and correlation were removed, which most often occurred in the upper part of the profile early in the monitoring season.Remaining data were post-processed to extract flow magnitude and direction.The sign of the current magnitude was processed to have a positive velocity if the current was flowing away from the main river input, and a negative velocity if the current was flowing towards the river.Time series of ADCP and hydrological datasets were analysed with cross wavelet transforms, and wavelet coherence (Grinsted et al. 2004;Torrence and Compo 1998).
A Kemmerer sampler was used to obtain 500 mL lake water samples for determination of lake water SSC in 2009.Lake instrumentation was typically removed in mid-to-late July when the ice cover became unsafe for travel.

Results
Data from 2007 and 2009 are presented on the same figure (Fig. 2), as are data from 2008 and 2010 (Fig. 3).Grouped years have similar lengths of monitoring seasons, and ADCP data are available only for 2008 and 2010.

Hydrometeorology
The length of the hydrologic monitoring period typically exceeded the length of the limnologic monitoring period due to deteriorating ice cover (Table 1).The hydrometeorology of the 2007 to 2010 snowmelt seasons is summarized below, and is described in detail elsewhere (Lafrenière and Lamoureux 2013;Lamoureux and Lafrenière 2009;Lewis et al. 2012).

D r a f t 10
In the third week of July 2007, sustained warm air temperatures and a thickened active layer resulted in extensive ALDs in the West River catchment (Lamoureux and Lafrenière 2009).The ALDs occurred at a time when the West River was in a period of baseflow and the thickening active layer is expressed as increasing river water conductance in July 2007 (Fig. 2C, E).The 2007 season was also unusual in that two rain events transported large volumes of sediment in West River, in part due to ALD that elevated late-season SSC (Dugan et al. 2009; Lamoureux and Lafrenière 2009) (Fig. 2D).
The 2008 freshet was the first sustained period when sediment derived from ALDs could be transported in large volumes.Sediment flux in this year was high, as was river water conductivity (Table 1; Fig. 3) although runoff was not unusual due to limited winter snowpack.
Runoff and sediment flux in 2009 were influenced by several large precipitation events that transported large volumes of sediment.However, these events occurred in mid-to-late July, after limnologic monitoring ended (Lewis et al. 2012).It was otherwise a year with very little discharge or sediment flux, due to cool conditions in June and limited catchment snow (Table 1).
In 2010, no elevated mid-to-late season SSC or ionic concentrations were recorded before limnologic monitoring ended.High SSC and runoff were restricted to the snowmelt period in June.

Limnology
In the early snowmelt season, water column temperatures are cold near 0°C under the ice cover and warm slightly at the lake bottom (Fig. 4A).HCBW are present in the deepest 1.5 to 2 metres of the water column, and are depleted in DO.Water temperatures are highest within the HCBW D r a f t 11 zone.Later in the summer, CTD casts show that the lake water is isothermal and isohaline (Fig. 4).
In 2007 and 2008, HCBW was removed, and in 2009 and 2010, HCBW largely persisted during the monitoring period (panel A in Figs 2, 3).When HCBW was present, SpC reached approximately 125 µS/cm.When HCBW was absent, bottom water SpC was c. 60 µS/cm.DO concentration at the lake bottom was inversely related to the presence or absence of HCBW.
An ADCP was moored in the deepest portion of the lake, adjacent to the moored CTD in 2008 and 2010 (Fig. 3 panel B, and Fig. 5).Near-bottom currents were higher and more variable in 2008 than 2010 (2008 average = 6.0 x 10 -3 m/s, standard deviation = 5.3 x 10 -3 ; 2010 average = 4.4 x 10 -3 m/s, standard deviation = 3.1 x 10 -3 ; Fig. 3B).Currents within HCBW was of opposite sign (direction) compared to currents in the immediate overlying water (Figs 4, 5).When HCBW was absent, currents were unidirectional throughout the measured water column, but varied in direction over time.
Bottom current periodicities were investigated towards the top of HCBW at 1.5 m above the lake bottom (Fig. 3G).Wavelet coherence was strongest when the inflowing river was at its most active, and the only persistent and statistically significant periodicities was c. 1 day.Similar strong diel periodicities were found closer to the lake bottom in both 2008 and 2010 (at 0.48 m above lake bottom; not shown).In 2007, contributions to river water density from SSC (Fig. 2D) are particularly evident and long-lived in July (Figs 2F).In 2008, river water density consistently exceeded lake water density (Fig. 3F), and suspended sediment (Fig. 3D) was primarily responsible for elevating river water density relative to lake water density.

River and lake water density
In 2009, the freshet was relatively muted due to sustained cool air temperatures and limited snowpack in June.Unlike 2007 and 2008, post-meltwater elevated fluvial SSC did not occur during the 2009 monitoring period (Figs. 2 and 3).During the 2010 freshet, SSC contributed substantially to river water density (Fig. 3F).However, after the freshet, SSC was low (Fig. 3D), and variability in river water density was controlled more by dissolved solids and water temperature (Fig. 3F; gray and black lines converge in late June 2010).In this sense, 2010 river water density was more similar to 2009 than the 2007 and 2008 seasons.
Variability in lake bottom water density is not evident on the same ordinate scale as river water density (Figs 2, 3F).Water samples and turbidity records from the lake bottom show that SSC was low there (<0.05 g/L, 2009 season, n=6) and varies less than fluvial SSC.The year with the largest bottom water temperature variability was 2007, when total warming was about 3.5 o C.
Temperatures in West Lake bottom waters contribute up to about 0.07 g/L of density variability.
The elevated conductivity in HCBW also contribute to density, adding up to about 0.03 g/L compared to overlying water.D r a f t 13

Discussion
Processes capable of generating currents that could remove HCBW in ice-covered lakes include wind-generated oscillations of the ice cover, heat transfer from bottom sediments to water, convective currents generated by solar radiation penetrating through ice to lake water, and hyperpycnal flows generated by high SSC in rivers (Malm et al. 1998).ADCP data show that the currents in West Lake are very likely river-generated, since currents begin shortly after river discharge begins (Figs.2B, 3B), and wavelet analysis shows that diel periodicities dominate (Fig. 3G).These diel periodicities typically occur when significant river discharge or sediment fluxes occur, and are absent when the river is quiescent.Similar daily periodicities are also found in the river discharge and SSC records.Conversely, records of wind velocity do not have a similarly consistent diel periodicity.Except for a thin moat surrounding the lake, the ice cover was solid, and could not significantly oscillate.Moreover, the seiche period in West Lake is on the order of several minutes, making seiche-generated currents an unlikely source for the higher velocity currents with daily periodicities.Interestingly, results show that near-bottom currents also form within HCBW; however, their sign is opposite to overlying currents, indicating that a convective cell is likely generated within HCBW (Figs. 4, 5) by overlying currents.The data suggest that the persistence or loss of HCBW through a snowmelt season is controlled by the balance between the density of the inflowing river water, mostly related to SSC, and the density of the HCBW.When river water density exceeds lake water density for long periods of time, river water can plunge, forming hyperpycnal flows, and displace HCBW.Conversely, when river water density is less than HCBW density, interflows and overflows tend to form and preserve the HCBW.

The persistence of HCBW in 2009 and 2010
In 2009, HCBW persisted throughout the snowmelt season (Fig. 2A).During September 2008, a within-lake event (likely a subaqueous slump) elevated lake turbidity by suspending very fine grained sediment throughout the 2009 snowmelt season (Dugan et al. 2012).It was hypothesized that the elevated lake turbidity increased lake water density sufficiently to isolate the bottom from hyperpycnal flows.However, water samples show that lake SSC reached a maximum of 0.05 g/L at this time, and this additional sediment (Fig. 2F, green dashed line) was a small contributor to density relative to fluvial SSC.An alternative hypothesis for the persistence of HCBW in 2009 is proposed, where relatively low river water density is the primary control on HCBW persistence, rather than increased lake water density.By 2009, catchment sediment D r a f t availability had decreased following the 2007 landscape disturbances, and the slump-derived sediment had settled out of suspension in the lake (Dugan et al. 2012;Lewis et al. 2012).
Similarly, in 2010, HCBW largely persisted throughout the monitoring period (Fig 3F).River water density exceeded HCBW density during freshet.However, after freshet, river water density was consistently less than HCBW density, which promoted HCBW persistence throughout the summer.
In both 2009 and 2010, river water density did exceed lake water density at times of peak river flow during the snowmelt season, when river SSC was high.Yet, HCBW was not removed in these periods.Several hypotheses can be invoked for this.First, relatively coarse grained sediments transported during freshet are deposited quickly, close to the delta front.During freshet, West River median suspended particle size is composed of grains up to a maximum of about 70 µm (very fine sand) (McDonald and Lamoureux 2009).Fluvial suspended particle size decreases following freshet, and remains in the fine to medium silt fraction during baseflow (10-15 µm).Coarse sediments are deposited quickly at the delta front, so much of the suspended sediment moved during freshet would not contribute to the balance between river and lake water density (Cockburn and Lamoureux 2008).Conversely, silts would remain in suspension for longer, and would be more likely to contribute to the formation of hyperpycnal flows that could remove HCBW.Additionally, the behaviour of hyperpycnal flows downslope can vary depending on its dilution due to sediment deposition and entrainment of surrounding waters (Mulder et al. 1998).Therefore, the density of hyperpycnal flows can be reduced as the flow evolves downslope (e.g., Middleton 1966).Second, HCBW appear to be removed by persistent and continuous hyperpycnal flows (2007 and 2008).Conversely in 2009 and 2010, hyperpycnal flows were less sustained.Third, the pathways of hyperpycnal flows down the delta could D r a f t impact HCBW persistence.Hyperpycnal flows on delta fronts do not follow the exact same pathways downslope every time they are triggered (e.g., Kremer et al. 2015).Multiple gullies on the delta front may distribute hyperpycnal flows in different directions in a given year (Normandeau et al. 2016), which would not allow them to necessarily reach the HCBW zone.

Catchment controls on increased sediment availability
HCBW persistence or loss appears to be linked to sustained elevated fluvial SSC, which Landscape modification thus had an impact on the persistence or removal of HCBW during the four year monitoring period.

Persistence of HCBW and projected climate change
Knowing that the persistence or removal of HCBW is affected by the balance between river and lake water density, it is important to consider how this balance might change in the future.In D r a f t particular, climate change will likely alter lake and river water density by altering water temperature, conductivity, and SSC (e.g., AMAP 2011).Each of these controls is discussed below.

Water temperature
Currently, West Lake bottom water temperatures vary from slightly above freezing to about 5 o C (in exceptionally warm summers).The range of water temperatures straddles the temperature of maximum water density.Importantly, the rate of change of density with temperature is low in this range (Fig. 7).On the other hand, river water temperature is much more variable, and is usually warmer.Recorded river water temperatures reached 15 o C in several years.As water temperature increases beyond 4 o C, the rate of change of density increases (Fig. 7).Increasing water temperature from 4 to 15 o C decreases water density by about 0.88 g/L (at 75 µS/cm).
Water temperature will likely increase in the future, both in rivers and in lakes (AMAP 2011).
Air temperature at the site is expected to increase by 3 to 5 o C by 2100 relative to 2010, depending on the emissions scenario and global climate model (Lewis and Lamoureux 2010).To estimate the effect this change might have on river water temperature and water density, the range was simply added to the maximum river water temperature recorded in 2008.The result is less than 0.75 g/L reduction in water density (Fig. 7).This is likely an overestimate since ground ice melt and increased baseflow will likely moderate river water temperatures (Bolduc 2015).
The lake is unlikely to warm by a similar amount due to thermal inertia of the lake and isolation of the lake from the atmosphere by persistent ice cover.Even if the lake does warm significantly, the current water temperature range is such that the rate of change of density with temperature is small (Fig. 7).Therefore, river water will become increasingly less dense with climate warming, D r a f t D r a f t 19 in most years.However, measurements of SSC indicate that this high turbidity is the result of suspension of very fine particles (<0.050 g/L) and the corresponding density effect is minimal.
Hence, even with these unusual perturbations to the lake sediment system, maximum lacustrine SSC remains typically several orders of magnitude less than maximum fluvial SSC.Hence, fluvial SSC should increasingly promote the formation of hyperpycnal flows and the flushing of HCBW, regardless of thermally-derived and solute-derived contributions to water density.In a warmer climate, the removal of HCBW is expected to be more frequent because of the increasing probability of hyperpycnal flows occurring in the lake.

Implications and conclusions
At West Lake, near-bottom currents are caused by hyperpycnal flows generated by river inflow.
Sustained lake hyperpycnal flows replace hypoxic HCBW with lower conductivity, oxic water.
Results indicate that removal of HCBW occurs after freshet, when fluvial SSC is elevated.
HCBW persisted in years when river SSC was low.
The importance of changing bottom oxygen conditions for aquatic and biogeochemical processes in the lake is substantial (Jenny et al. 2016).Aquatic biota, especially amphipods and juvenile char that use the bottom waters for feeding and to avoid predation are likely to face additional ecological stress.A reduced duration of HCBW and corresponding bottom hypoxia would also likely alter phosphorus availability and reduce Hg methylation, with implications for potential reduction of bioaccumulation of this contaminant through the food web.The biogeochemical impacts of reduced bottom water anoxia are also likely to alter microbial populations, with D r a f t 20 impacts on carbon cycling and greenhouse gas emissions from the lake.While further studies are necessary to define these effects, the transition to reduced HCBW and associated anoxia duration is likely to substantial limnological impacts (Jenny et al. 2016).
Finally, these results have significance for the paleoenvironmental record in lakes.Oxygen content in bottom water is a control on bioturbation, and hyperpycnal flows have the potential to erode and redistribute sediment.HCBW may therefore be important for the preservation of fine sedimentary structures such as clastic varves (e.g.Cockburn and Lamoureux 2008).HCBW evolution may also be recorded in paleoenvironmental archives as changes in sedimentary structure and/or geochemical changes in redox-sensitive elements.
SSC is currently the dominant control on water density, and can control the balance between river water density and lake water density.In the future, likely changes in water conductivity, temperature, and SSC will make fluvial SSC an increasingly dominant control over this balance.
It is therefore likely that hyperpycnal flows will be increasingly common, and that HCBW will be increasingly removed.This anticipated direction of change is likely applicable to other high latitude lakes with HCBW.

D r a f t
water was much more variable than lake water density, primarily due to the large but variable contributions from fluvial SSC.River water density was calculated both considering density from only SpC and temperature (ρ(C,T)), and from SpC, temperature, and SSC (ρ(C,T,S); Figs.2F, 3F).
HCBW persisted during the monitoring period in2009 and 2010, but not in 2007 and 2008.The persistence or loss of HCBW appears to be strongly affected by high fluvial SSC during the summer.HCBW was removed in years where fluvial SSC contributed greatly to fluvial water density, and persisted in years where fluvial SSC was low.For example, in 2007, fluvial water density exceeded HCBW density for almost all of the lake monitoring season.(Fig.2F).Fluvial water density above HCBW density in July (Fig.2F), which then led to hyperpycnal flows and the removal of the HCBW.The same processes occurred in 2008, where mid season elevated fluvial SSC increased fluvial water density sufficiently to exceed lake water density for the entire monitoring season.However, if fluvial SSC is not included in density calculations, fluvial water density is consistently less than lake water density in 2008 (Fig.3F), supporting the role played by hyperpycnal flows in bottom mixing.
increases river water density relative to lake water density.During the 2007 to 2010 monitoring period, elevated SSC was mainly caused by ALDs.Disturbances increased sediment availability in 2007 and 2008, which facilitated the generation of hyperpycnal flows in the lake.Increased sediment availability was particularly evident after freshet at low discharge, when water interacted with thawed soil, and when water transport times were slower.Following the exhaustion of sediment related to the major 2007 disturbances, fluvial SSC decreased and only reached pre-disturbance levels in 2009 and 2010 (Fig. 6).Warm summer temperatures in 2007 thus had an indirect impact on the removal of HCBW by creating disturbances, which increased SSC in West River, which in turn favoured the generation of hyperpycnal flows in the lake.

Figure 2 :
Figure 2: Physical conditions near the bottom of West Lake and in West River in 2007 and 2009:

Figure 3 :
Figure 3: As in figure 2, except for 2008 and 2010.An ADCP was moored in West Lake in these

Figure 4 :
Figure 4: Representative water column profiles from 2010 in the deepest part of West Lake (Fig.

Figure 5 :
Figure 5: Profiles of currents near the lake bottom in (A) 2008 and (B) 2010.

Figure 6 :Figure 7 :
Figure 6: Relationship between West River discharge (Q) and suspended sediment concentration

Figure 1 :Figure 2 :Figure 3 :Figure 4 :Figure 5 :Figure 6 :Figure 7 :
Figure 1: Maps of (A) Cape Bounty on Melville Island in the Canadian High Arctic, (B) the West Lake watershed at Cape Bounty and (C) West Lake.The topographic contour interval in 'B' is 10 m, with the lowest contour level near the West Lake shoreline at 10 m asl.Bathymetric contours intervals are 5 m and are from Normandeau et al. (2016).The estimated extent of high conductivity bottom water (HCBW) is indicated.99x52mm (300 x 300 DPI) Prostrate shrub and grass tundra vegetation covers much of land surface and the Arctic climate is severe: only June, July, and August have mean monthly air temperatures above 0°C.
. Lithology is primarily clastic sedimentary (sandstone and

Table captions Table 1 .
Summary of hydrologic data from West River 2007-2010.Fluvial data are summarized for the entire hydrologic monitoring period, and for the portion overlapping with the limnologic monitoring period.

Table 1 .
Summary of hydrologic data from West River 2007-2010.Fluvial data are summarized for the entire hydrologic monitoring period, and for the portion overlapping with the limnologic monitoring period. https://mc06.manuscriptcentral.com/asopen-pubs