Canadian Journal of Fisheries and Aquatic Sciences

Influence of glacial flour on the primary and secondary production of Sockeye Salmon nursery lakes: a comparative modern and paleolimnological study

Journal: Canadian Journal of Fisheries and Aquatic Sciences

Manuscript ID cjfas-2018-0372.R1

Manuscript Type: Article

Date Submitted by the 14-Jan-2019 Author:

Complete List of Authors: Barouillet, Cécilia; Queen's University, Biology Cumming, Brian; Queen's University, Biology; Laird, Kathleen; Queen's University, Biology Perrin, Chris;Draft Limnotek Research and Development Inc., Selbie, Daniel; Fisheries and Oceans Canada, Science Branch

Glacial flour, Sockeye Salmon, CLADOCERA < Organisms, Bridge River Keyword: Diversion, HYDROPOWER < General

Is the invited manuscript for consideration in a Special Not applicable (regular submission) Issue? :

https://mc06.manuscriptcentral.com/cjfas-pubs Page 1 of 45 Canadian Journal of Fisheries and Aquatic Sciences

1 Title: 2 Influence of glacial flour on the primary and secondary production of Sockeye Salmon nursery 3 lakes: a comparative modern and paleolimnological study 4 5 6 Authors: 7 Barouillet, Cécilia*,1, Brian F. Cumming1, Kathleen R. Laird1, Christopher J. Perrin2 and Daniel 8 T. Selbie1,3 9 1 Paleoecological Envrionmental Assessment and Research Laboratory, Department of Biology, 10 Queen’s University, 116 Barrie Street, Kingston, ON, Canada K7L 3J9 11 2 Limnotek Research and Development Inc. 4035 West 14 Avenue, Vancouver, BC, Canada V6R 12 2X3 13 3 Fisheries and Oceans Canada, Pacific Region, Science Branch, Cultus Lake Salmon Research 14 Laboratory, 4222 Columbia Valley Highway, Cultus Lake, BC, Canada V2R 5B6 15 Draft 16 Brian F. Cumming: [email protected] 17 Kathleen R. Laird: [email protected] 18 Christopher J. Perrin: [email protected] 19 Daniel T. Selbie: [email protected] 20 21 * Corresponding author 22 Phone: (+1) 343 333 5353 23 Email: [email protected] 24 25

1

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 2 of 45

26 Abstract

27 The increasing rate of glacier retreat, and turbid glacial runoff can have a strong influence

28 on freshwater ecosystems. Seton and Anderson lakes (, Canada) are Sockeye

29 Salmon (Oncorhynchus nerka) nursery systems. Since the 1940s, the Diversion

30 (BRD) introduced glacially-turbid water into Seton Lake. To assess the impact of the BRD on

31 the production of Seton Lake, we combined data from limnological surveys with the analysis of

32 sub-fossil cladocerans and diatoms from sediment cores; using Anderson Lake as a reference.

33 The modern data indicate that the euphotic zone is 14m shallower, and the cladoceran density

34 and biomass are significantly lower in Seton Lake in comparison to Anderson Lake. The paleo- 35 data indicate that following the BRD, theDraft sedimentary fluxes of cladoceran and diatom declined 36 2- to 10-fold in Seton Lake and remained low thereafter. Together our data support declines in

37 primary and secondary producers following the BRD, likely due to changes in light penetration,

38 and/or other indirect influence, and provides insights into the impact of turbid meltwater on the

39 biological production of downstream lakes.

2

https://mc06.manuscriptcentral.com/cjfas-pubs Page 3 of 45 Canadian Journal of Fisheries and Aquatic Sciences

40 Introduction

41 The increased rate of glacier retreat is a direct, visible, and perhaps one of the most

42 widely known outcomes of the rise in global air temperatures (IPCC 2014). The resultant glacial

43 runoff can have important ecological consequences for lakes situated downstream, including

44 through the transport of suspended glacial flour particles, which originate from the erosion of the

45 bedrock by glaciers (Fegel et al. 2016). Suspensions of glacial flour can reduce light penetration,

46 including the penetration of UVR and photosynthetically-active radiation (PAR) in the water

47 column (Rose et al. 2014). Additionally, through the colloidal binding of nutrients (Hodson et al.

48 2004, 2008), glacial flour can become a significant source or sink of biologically-available 49 nutrients (Hodson et al. 2004; Saros et al.Draft 2010; Slemmons and Saros 2012). Together, these 50 changes may limit primary producers and alter rates of primary production. As a result, lakes

51 receiving glacial meltwater may support relatively low abundances of primary and secondary

52 producers (Vinebrooke et al. 2010). Glacial flour can also directly or indirectly alter assemblages

53 within and among plankton communities (Kirk 1991; Saros et al. 2010; Slemmons and Saros

54 2012); influence trophic interactions (i.e. weakening or intensifying predator-prey interactions;

55 Horppila and Liljendahl-Nurminen 2005); and modify the vertical distribution of phytoplankton,

56 zooplankton (Hylander et al. 2011) and planktivores (Gregory 1993) within the water column.

57 Such limnological changes induced by glacial flour can thus result in the modification of energy

58 flows throughout food webs (Koenings et al. 1990; Laspoumaderes et al. 2013). Ecosystem

59 responses to glacial runoff can, however, be complex, as they vary spatially and temporally as a

60 function of the distance to and the magnitude of the glacial source (Saros et al. 2010; Bliss et al.

61 2014).

3

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 4 of 45

62 In North America, many populations of Sockeye Salmon and kokanee (Oncorhynchus

63 nerka) rear in lakes influenced by glacial runoff. While the fish productivity in those turbid lakes

64 tend to be lower than in clear-water lakes (Koenings et al. 1985), the impacts of long-term inputs

65 of glacial flour on O. nerka freshwater productivity are not well known. Sockeye Salmon sustain

66 important commercial and recreational fisheries throughout the North Pacific, and understanding

67 the factors limiting the productive capacity of freshwater nursery ecosystems is critical for

68 effective conservation and management of salmon populations (Stockner and Macisaac 1996).

69 However, salmon populations and their nursery lakes are often exposed to multiple stressors that

70 can include overfishing, habitat destruction, and dams (Nehlsen et al. 1991; Healey 2011; Schoen

71 et al. 2017), making it hard to assess the impact of glacially-turbid water on nursery lakes

72 ecosystem dynamics. As juveniles O. nerkaDraft preferentially feed on cladoceran zooplankton,

73 particularly the energy-rich Daphnia (Mazumder and Edmundson 2002; Scheuerell et al. 2005),

74 the abundance of Daphnia (and other cladocerans such as Bosmina) is a key metric of the food

75 supply for O. nerka (Schindler et al. 2005b). In the absence of long-term monitoring data,

76 paleolimnological techniques (Smol 1992) have been used to reconstruct past ecological and

77 environmental changes in salmon nursery lakes (Finney 2000; Finney et al. 2002; Schindler et al.

78 2005a; Selbie et al. 2007; Gregory-Eaves et al. 2009). For instance, sedimentary cladoceran

79 remains have been used to reconstruct the secondary production over time in order to track

80 changes in predation (i.e. fish vs invertebrate planktivory) and primary production (Jeppesen et

81 al. 2001; Sweetman and Finney 2003; Korosi et al. 2013; Shi et al. 2016). Similarly, multi-

82 trophic paleolimnological analyses of sedimentary remains of diatoms, cladocerans, and stable

83 nitrogen isotopes (δ15N) have been successfully used to understand variations in past salmon runs

84 and the cumulative influences of stressors affecting existent salmon populations and their

4

https://mc06.manuscriptcentral.com/cjfas-pubs Page 5 of 45 Canadian Journal of Fisheries and Aquatic Sciences

85 freshwater habitats, including overfishing, dams, watershed disturbance, and climate (Finney

86 2000; Selbie et al. 2007; Gregory-Eaves et al. 2009). Moreover, the integration of modern and

87 paleolimnological techniques can further help to understand the influence of different factors on

88 the productive capacities of nursery ecosystems to support O. nerka (Selbie et al. 2007, 2011).

89 In British Columbia, the drainage basin hosts 377 individual populations of

90 Sockeye Salmon (Grant and Pestal 2012). Two of these populations are found in Seton and

91 Anderson lakes, two large and deep oligotrophic lakes of similar geomorphological, physical and

92 climatic conditions. These lakes also support the landlocked variant kokanee (O. nerka), also

93 known in Seton and Anderson lakes as “black” kokanee (Moreira and Taylor 2015) or “Gwenis” 94 in St’at’imc First Nations dialect (Geen Draftand Andrews 1961). Since the construction of 95 hydroelectric dams and a water diversion (i.e. the Bridge River Diversion, BRD) between the

96 1930s and 1960s, novel inputs of glacially-turbid water have been introduced to Seton Lake, a

97 historically clear-water system; whereas Anderson Lake, located upstream of the BRD, remains

98 undisturbed. Modern limnological studies have found that the primary and secondary production

99 of Seton Lake is currently lower than in Anderson Lake (Geen and Andrews 1961; Shortreed et

100 al. 2001), and according to the Photosynthetic Rate (PR) Model (Hume et al. 1996; Shortreed et

101 al. 2000), the rearing capacity of both lakes is underutilized (Shortreed et al. 2000). However,

102 existing limnological information on Seton Lake represents only a small portion of the post-

103 diversion period, making it difficult to assess the full impact of hydroelectric development on the

104 limnology and the rearing capacity of O. nerka in Seton Lake.

105 In the present study, we applied modern and paleolimnological techniques to assess

106 changes in the primary and secondary production of Seton and Anderson lakes, and to evaluate

107 the effects of the BRD on food availability for planktivorous fish in Seton Lake. We 5

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 6 of 45

108 hypothesized that the introduction of glacially-turbid water would lead to a decline in the

109 primary (diatoms) and secondary (cladocerans) production in Seton Lake. A sampling design

110 similar to a Before-After-Control-Impact (BACI) layout (Stewart-Oaten et al. 1986) was used to

111 assess the cumulative effects of the water diversion and other regional factors that would be

112 unrelated to the BRD (e.g. climate), but influential upon Seton Lake primary and secondary

113 production, with Seton Lake being the impacted system, and Anderson Lake the reference site.

114 The results from our study are applicable to understanding variability in the food-web production

115 that supports valued fish populations in lakes located downstream from glaciers.

116 117 Materials and Methods Draft 118 Study Lakes

119 Seton and Anderson lakes are large and deep oligotrophic lakes located in the Fraser

120 River drainage basin, within the St’at’imc traditional territory in the central interior of British

121 Columbia, Canada (Fig. 1). Seton Lake (N 50°41.758’ W 122°08.007’) is located to the west of

122 the town of , and Anderson Lake (N 50°38. 089’ W 122°23.577’) flows into the west end

123 of Seton Lake via the (Fig. 1). The geology of the region is composed of a mix of

124 volcanic, metamorphic and sedimentary rocks of Carboniferous to Jurassic origin (Geen and

125 Andrews 1961). Both systems are glacially-formed depressions surrounded by steep mountains.

126 Seton Lake and Anderson Lake have a mean water depth of 85 m and 140 m and a maximum

127 depth of 151 m and 215 m, respectively. The area is characterized by a semi-arid microclimate

128 with periodic influences of coastal weather and cold arctic air during the winter. The region is

129 hot and dry in the summer, with annual precipitation averaging only 300 to 400 mm, and

6

https://mc06.manuscriptcentral.com/cjfas-pubs Page 7 of 45 Canadian Journal of Fisheries and Aquatic Sciences

130 consequently this region is sensitive to sporadic forest fires. The vegetation is dominated by

131 Ponderosa Pine (Pinus ponderosa). Seton and Anderson lakes contain anadromous populations

132 of Sockeye Salmon and resident populations of kokanee (O. nerka). Two populations of Sockeye

133 Salmon spawn in the tributaries of Anderson and Seton lakes, referred to as the Gates Creek and

134 the Portage Creek populations, respectively (Bell 1985). Although inter-annual variability in

135 these lakes is large, salmon populations have averaged ~33,770 ± 26,353 (1SD) of anadromous

136 Sockeye Salmon spawners over the past decade (Fisheries and Oceans Canada, unpubl.).

137

138 History of the Bridge River Diversion

139 The Bridge River Diversion (BridgeDraft River Power Project; BRD) was initiated by the 140 Bridge River Power Company in 1927. It was then taken over by the British Columbia Electric

141 Company, which was nationalized in 1961 to become part of the British Columbia Hydro and

142 Power Authority. The BRD was built to provide hydropower by diverting water from the Bridge

143 River into Seton Lake, a vertical drop of 1067m. The project started in the early 1930s when

144 water from the Bridge River was diverted for the first time into the upper part of Seton Lake at a

145 rate of ~2.7m3·sec-1 (Geen and Andrews 1961). The project stalled during the Great Depression

146 and World War II but was revived between 1946 and 1956 with the construction of the Seton

147 Dam at the outflow of Seton Lake, and the , which flooded the Bridge River

148 Valley and created the Carpenter Reservoir (Fig. 1). This infrastructure allowed the flow from

149 the diversion to increase up to a mean annual flow rate of ~56.6 m3·sec-1 (Geen and Andrews

150 1961). By the end of the construction in 1962, two-thirds of the water coming into Seton Lake

151 originated from the Carpenter Reservoir, with a mean annual inflow rate of ~87 m3·sec-1±12 (data

152 courtesy of BC Hydro). The water from Carpenter Reservoir, rich in glacial flour, originates 7

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 8 of 45

153 from the Bridge Glacier, a ~83 km2 lake-calving glacier which is rapidly retreating (Chernos

154 2014).

155

156 Modern limnological data

157 Modern limnological data were collected in 2014, 2015, and 2016 in order to obtain

158 knowledge of the present-day limnological state of the study lakes. Data were collected at 2

159 monitoring stations along the length of each lake to span a spatial gradient that could result from

160 the observed inflow of glacial flour from the BRD. The limnological data does not allow us to

161 directly assess the impact of the diversion on the primary and secondary production in Seton 162 Lake, as no pre-diversion limnological dataDraft exists for comparison. Instead, the modern data can 163 help us understand the present-day conditions, which provides context for the paleolimnological

164 study.

165 Over the three years of sampling, measurements of photosynthetically-active radiation

166 (PAR), turbidity, nutrient concentrations, rates of primary production (or gross primary

167 production, GPP), and zooplankton net hauls were completed monthly from May through

168 October (Fig. 1). A LiCor LI250A irradiance meter was used to estimate the depth of the

169 euphotic zone by measuring PAR in 1m intervals from the lake surface to a depth where PAR

170 was less than 1% of that at the surface. Turbidity measurements were collected using a Sea-Bird

171 Electronics SBE19plusV2 CTD along the entire water column. Total nitrogen (TN), total

172 phosphorus (TP), and bioavailable forms of nitrogen (NH4-N and NO3-N) and phosphorus

173 (soluble reactive phosphorus; SRP) were measured from water samples collected at a depth of

174 1m, and analysed using standard methods (APHA 2014).

8

https://mc06.manuscriptcentral.com/cjfas-pubs Page 9 of 45 Canadian Journal of Fisheries and Aquatic Sciences

175 Measurement of Gross Primary Production

176 Gross primary production (GPP) was measured by the British Columbia Ministry of the

177 Environment in situ as the amount of 14C incorporated into particulate organic matter (Steeman

178 Nielsen 1952). Discrete water samples were collected for seven depths over the profile of the

179 euphotic zone. At each depth, water samples were transferred directly into two light and one dark

180 glass bottle, which were rinsed three times with the collected water before filling. Each water

14 181 sample was inoculated with 0.185 MBq (5 µCi) of NaH CO3 New England Nuclear (NEC-

182 086H). The cluster of bottles at each depth was incubated for 4-5 hours at the original collection

183 depth in the water column. Following retrieval, 100mL of each sample was filtered onto 0.2µm 184 polycarbonate filters, and stored in the dark.Draft The samples were then treated with 100µL of 0.5 N 14 Ò 185 HCl to eliminate the unincorporated inorganic NaH CO3, and five millilitres of Scintisafe

186 scintillation cocktail. The vials were counted using a BeckmanÒ Model #LS 6500 liquid

187 scintillation counter. Daily rates of primary production were vertically-integrated for each station

188 (Ichimura et al. 1980), and were calculated by multiplying the hourly primary production

189 (Parsons et al. 1984) by the incubation time, and were corrected for solar irradiance. For more

190 details about the method refer to Limnotek (2017).

191 Zooplankton Collection and Analysis

192 Vertical zooplankton hauls from 30 m depth to the surface were collected using a

193 Wisconsin net (153 µm mesh). The total volume of water that passed through the net was

194 recorded, and the samples were preserved in a 10% sugar-buffered formalin solution. Aliquots of

195 known sample volume were separated from each sample using a Folsom plankton splitter, and

196 species were identified and enumerated at 5-100x magnification. Biomass of zooplankton was

9

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 10 of 45

197 determined from length-to-weight regressions (MacCauley 1984) using the average length

198 measurement of 25 random measurements taken for each taxon found in each sample. The

199 number of individuals counted and estimated biomass (µg dry weight) were extrapolated to the

200 total volume and the estimates were then integrated over the sampling depth (30m) to the number

201 of individuals or biomasses found in a 1 m2 (i.e. areal density). The monthly average over the

202 three sampling years of the zooplankton density and biomass was calculated for each of the

203 primary zooplankton groups (i.e. cladoceran, copepods, rotifers).

204 Data analysis was performed using R (v. 3.2.1). For all limnological variables (PAR,

205 turbidity, nutrient concentration, GPP, zooplankton density and biomass), the two monitoring 206 stations within Seton and Anderson lakesDraft were not significantly different (t-test; p>0.05) and 207 therefore, were treated as replicates. The Bartlett and Shapiro tests were used to evaluate whether

208 the modern limnological variables met the assumptions (homoscedasticity and normality) for a

209 parametric test, and the limnological variables that did not meet the assumptions were log-

210 transformed. A t-test was performed on the monthly average of each limnological variable to test

211 for inter-lake differences. Due to the high inter-annual variability of the GPP measurements, a 2-

212 way ANOVA was performed on GPP to test for differences between lakes, and years.

213

214 Paleolimnological Data

215 The paleo-data were used in this study to reconstruct the history of the primary and

216 secondary production of Seton and Anderson lakes, and assess the potential impact of the Bridge

217 River Diversion on the food supply for O. nerka in Seton Lake. The core chronologies relating

218 depth (cm) to sediment age in each core was developed to define the timing of the diversion and

10

https://mc06.manuscriptcentral.com/cjfas-pubs Page 11 of 45 Canadian Journal of Fisheries and Aquatic Sciences

219 calculate fluxes. Core chronologies of the Anderson Lake cores was established using standard

220 methods of radioisotopic dating (Appleby 2002), while for the Seton cores, grain size analysis

221 was completed as a complement of the radioisotopic analyses . This intermediate step was

222 necessary due to the low 210Pb activities found in the Seton Lake cores. The history of the

223 primary and secondary production in the two study lakes was reconstructed using sedimentary

224 sub-fossils of cladoceran and diatom assemblages.

225 Sediment core collection

226 Five gravity cores were retrieved from Seton and Anderson lakes in August 2014 (Fig. 1)

227 using a Glew gravity corer (Glew et al. 2002). Sediment cores from both lakes were retrieved 228 from deep depositional basins, at 203 mDraft and 205 m in Anderson Lake (cores A1 and A2, 229 respectively), and at 128 m, 118 m, and 110 m in Seton Lake (cores S3, S2, and S1,

230 respectively). Cores A1 and A2 were 66.5 cm and 51.5 cm in length, respectively. Cores S3, S2,

231 and S1 were 74 cm, 66.5 cm, and 74.5 cm in length, respectively. Sediment cores S1-S3 were

232 collected along the central axis of Seton Lake to capture a gradient of influence from the BRD,

233 with core S3 located nearest to the discharge of the BRD, and cores S2 and S1 located at

234 increasing distances eastward from the BRD (Fig. 1). The two sediment cores from Anderson

235 Lake were collected along the length of the lake to span a similar spatial gradient as in Seton

236 Lake, and used as a spatio-temporal reference. All sediment cores were sectioned at 0.5 cm

237 intervals into sterile bags, and stored at 4 °C.

238 Grain size analysis & core chronology

239 Sediment samples for grain size analyses were prepared following the methods from

240 Jiilavenkatesa et al. (2001). Approximately, 100 to 150 mg of wet sediment was treated with the

11

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 12 of 45

241 following chemical treatment: ~30 mL of 30% H2O2 for several weeks to digest the organic

242 matter; 2mL of 11M HNO3 for 10 minutes, to digest any remaining organic matter; 40mL of 1M

243 NaOH, placed in a water bath at ~70oC for a minimum of 20 minutes, to remove biogenic silica;

244 and ~8 mL of 0.5M HCl, to help neutralize the solution and flocculate the sediment. Between

245 each chemical treatment, the samples were rinsed three times as follows: the samples were

246 centrifuged at 3500 rpm for 30 minutes to 1 hour (until supernatant clear); the supernatant was

247 aspirated and; the remaining material was re-suspended in 40 mL of deionized water using a

248 vortex agitator. The grain size distribution was analyzed on a Malvern Mastersizer 2000,

249 assuming a reflective index of 1.54. The grain size results were then partitioned into three size

250 categories (clays, <2 µm; silts, 2-63 µm; and sand >63 µm) according to grain size classification

251 developed by Blott and Pye (2001), andDraft expressed as percent of total grain size.

252 Radioisotopic dating (210Pb, 137Cs, 214Pb, and 214Bi) was performed on all sediment cores

253 using an Ortec gamma spectroscopy counter (Schelske et al. 1994) and between 16 and 25

254 samples were analyzed per core. For the two cores from Anderson Lake (A1 and A2), a

255 Constant Rate of Supply (CRS) model was used to estimate the core chronology, based upon the

256 unsupported 210Pb activity in the core (Binford 1990). For the sediment cores from Seton Lake,

257 the 210Pb activity was very low (likely due to the presence of glacial flour in the core tops; Fig.

258 S1), and led to the use of independent markers to develop the chronologies, which included the

259 peak in 137Cs activity and changes in grain-size composition (Figs. S1, S2, S3). The peak in the

260 radioisotope 137Cs was used to identify the year 1963 (corresponding to the fallout maximum of

261 atmospheric nuclear testing; (Appleby 2001), while the sharp transition in grain sizes was used to

262 localize the first elevated inflows of glacially-turbid water coming from the BRD between the

263 late-1940s and the early-1950s. While the constructions of the diversion started in the 1930s, it is

12

https://mc06.manuscriptcentral.com/cjfas-pubs Page 13 of 45 Canadian Journal of Fisheries and Aquatic Sciences

264 only around the late 1940s that the BRD started to divert important volumes of water (~56.6

265 m3·sec-1) down into Seton Lake. The sedimentation rates for the Seton Lake cores were

266 calculated using linear interpolation between the first increase in clay above background levels,

267 and the ca. 1963 horizon (137Cs peak), and from ca. 1963 to the top of each core, which

268 corresponded to 2014, the year of retrieval (Fig. S3). The average sedimentation rates from the

269 Anderson Lake sediment cores were used to estimate the sedimentation in Seton Lake prior to

270 the BRD (Fig. S3).

271

272 Sedimentary cladoceran and diatom analyses

273 The analysis of cladoceran sub-fossilsDraft were prepared following standard methods 274 (Korhola and Rautio 2001). Approximately 1 g of wet sediment was treated with 150 mL of 10%

275 KOH to deflocculate the sediment. The sediment-KOH mixture was then sieved through a 34 µm

276 mesh and backwashed with deionized water into a 12mL glass vial, to which several drops of

277 safranin glycerine solution (dye) and alcohol (preservative) were added. A 50µL aliquot of slurry

278 was deposited on a slide and allowed to dry. This process was repeated as necessary to achieve

279 samples of sufficient concentration for enumeration. Individuals on the entire slide were counted.

280 At high concentrations, a minimum count of 70 individuals per sample could be reached, but at

281 low concentrations a minimum count of 20 individuals per sample were enumerated, which is

282 sufficient to characterize a sedimentary cladoceran assemblages with low diversity (Kurek et al.

283 2010). For some intervals at the top of the Seton Lake cores, the minimum count of 20

284 individuals could not be reached because of the low concentrations, and limitations in

285 concentrating the slurry on the slides, due to high concentrations of clay (see intervals marked by

286 a dot in Fig. 5). Standard identification keys were used to identify the cladoceran remains 13

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 14 of 45

287 (Szeroczyñska and Sarmaja-Korjonen 2007; Korosi and Smol 2012a, 2012b). Bosmina spp. was

288 only identified to the genus level because the head pores were difficult to localize due to the

289 presence of other small perforations.

290 The analysis of diatoms was completed following standard methods (Battarbee et al.

291 2002). Briefly, ~0.2-0.3 g of wet sediment was sub-sampled in 20 mL glass vials. To remove

292 organic matter, a 1:1 mixture by molar weight of concentrated nitric (HNO3) and sulphuric

293 (H2SO4) acid was added to each sample, which were then heated in a hot water bath for 6-7

294 hours. Samples were allowed to settle for 24 hours before the acid above the sample was

295 aspirated, and rinsed with deionized water; this rinsing procedure was repeated until the sample 296 had the same pH as deionized water. SedimentaryDraft concentrations of diatoms were determined 297 using known concentrations of microspheres added to each of the diatom samples (Battarbee and

298 Kneen 1982). Microspheres were then enumerated along with the diatoms.

299 Diatom and cladoceran fluxes were calculated using concentration data and estimated

300 sedimentation rates for each core. The fluxes were expressed as a net accumulation rate (#ind·

301 cm-2· year-1) and calculated by dividing the concentration (#ind·g dry weight-1) by the volume of

302 the sediment sample (cm-3), and by the number of years contained in the sediment sample (0.5

303 cm thickness). While the relative abundance (%) can give a good representation of the

304 cladoceran community composition present in the water column, fluxes give a better estimate of

305 the absolute density of individuals in the water column (Nykänen et al. 2009). However, it is

306 important to note that the calculation of flux data depends on a reliable core chronology with a

307 known estimates of core-specific sedimentation rates. The age-depth model developed for each

308 of the Seton Lake cores provides an average sedimentation rate for the three following periods:

309 prior to the diversion, from the start of the diversion until 1963, and from 1963 to present. The 14

https://mc06.manuscriptcentral.com/cjfas-pubs Page 15 of 45 Canadian Journal of Fisheries and Aquatic Sciences

310 flux data thus provides a sufficient estimate of the magnitude of change in sedimentation rate

311 from before to after the diversion, and can be used in a comparative sense to the changes in the

312 Anderson Lake cores.

313

314 Results

315 Modern limnological data

316 Over the three sampling seasons, from 2014 to 2016, the depth of the euphotic zone was

317 significantly lower while turbidity was significantly higher in Seton Lake compared to Anderson

318 Lake (t-test, p<0.005, Fig. 2), TP and TN were not significantly different between the two lakes

319 (t-test, p<0.05, Fig. 2), when averaged acrossDraft sampling stations. GPP was significantly different

320 between years (t-test, p<0.05) and significantly greater in Anderson Lake than in Seton Lake (t-

321 test, p=0.05). The depth of euphotic zone and turbidity remained stable throughout the sampling

322 seasons in both lakes. The depth of the euphotic zone was on average ~14 m shallower in Seton

323 Lake compared to Anderson Lake, respectively (Fig. 2). The seasonal average turbidity in Seton

324 Lake was 3.3 NTU, and 1.9 NTU in Anderson Lake. The depth of the euphotic zone was

325 negatively correlated with turbidity in Seton Lake (r= -0.83, p<0.05), but was not correlated with

326 turbidity in Anderson Lake (r= -0.21, p>0.05). TP varied between 3 µg·L-1 in June and 2 µg·L-1

327 in October, and TN varied around a mean of 42 µg·L-1 in Seton Lake. In Anderson Lake, TP

328 remained relatively stable around a mean of 2 µg·L-1, and TN reached a maximum in May (76

-1 -1 329 µg·L ) and declined to a minimum in October (40 µg·L ). Ammonium (NH4-N) and SRP were

-1 330 below detection levels in both lakes (detection limits: 5 µg·L ). Nitrate (NO3-N) was low in both

331 lakes, seasonal average (May-Oct) of 7.6 µg·L-1 and 17 µg·L-1 in Seton Lake and Anderson Lake,

15

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 16 of 45

332 respectively; with highest values reached in May in Seton Lake and in Anderson Lake (13.3

333 µg·L-1 and 44 µg·L-1, respectively). Seasonal mean TN:TP (molar ratio) averaged ~37 in Seton

334 Lake and ~54 in Anderson Lake. Gross primary production (GPP) of Seton Lake varied between

335 a maximum of 123 mg C·m-3·day-1 in June and a minimum of 71 mg C.m-3.day-1 in August, with a

336 seasonal average of 96.73 mg C·m-3·day-1. In Anderson Lake, GPP increased from May (96 mg

337 C·m-3·day-1) to September (255 mg C·m-3·day-1), and then declined in October (Fig. 2), with a

338 seasonal average of 157 mg C·m-3·day-1.

339 Modern zooplankton abundance and composition

340 Total zooplankton density and total cladoceran density were 2-fold lower in Seton Lake 341 than in Anderson Lake (t-test, p<0.005; DraftFigs. 2, 3). Cladoceran biomass was also significantly 342 different between the two lakes with a biomass 10-fold higher in Anderson Lake than in Seton

343 Lake. The density and biomass of the copepods (Fig. 3) and the density of the rotifers were not

344 significantly different between the two lakes (t-test, p>0.05). Total zooplankton biomass was

345 highly correlated with cladoceran biomass (r=0.9, p<0.005), while the total zooplankton density

346 was highly correlated with copepod density (r=0.9, p<0.005). In Seton Lake, copepod density

347 and biomass was highest in May, and cladoceran density and biomass was slightly higher in

348 August and September (Fig. 3). In Anderson Lake, copepod density and biomass was higher than

349 in Seton Lake, throughout the sampling season, with a maximum peak reached in May.

350 Cladoceran density and biomass was highest in June in Anderson Lake (Fig. 3).

351 In Seton Lake, the copepod community was numerically dominated by the cyclopoid

352 copepod, Cyclops scutifer Sars G.O (~84% percent of individuals; Fig. S5). The cladoceran

353 community was dominated by Daphnia ambigua Scourfield and Eubosmina longispina Leydig

354 (~76% and ~16%, respectively), and Eubosmina longispina was the only Bosmina sp. found in 16

https://mc06.manuscriptcentral.com/cjfas-pubs Page 17 of 45 Canadian Journal of Fisheries and Aquatic Sciences

355 the modern samples. The rotifer assemblage was co-dominated by Kellicottia sp. and Asplancha

356 sp. (both ~50%). In Anderson Lake, the copepods were also dominated by Cyclops scutifer

357 (~50%; Fig. S5) and the cladoceran assemblage was dominated by Daphnia ambigua and

358 Eubosmina longispina (~68% and ~31%, respectively). The rotifers were dominated by

359 Kellicottia sp. (~98%).

360

361 Paleolimnological data

362 Grain size analysis & core chronology

363 Grain size analysis was used in conjunction with the radioisotope analysis of 137Cs in the

364 development of the core chronology forDraft Seton Lake. The results from the grain size analysis

365 were used as a marker of the completion of the BRD. In Seton Lake, clay (particle size <2 µm)

366 contributions to the sediment matrix began to increase around 50 cm, 44 cm, and 30 cm in cores

367 S3, S2, and S1, respectively, with percent composition of clay rising from ~5% to above 40%

368 (beginning of Zone A in Fig. S2). Clay contributions continued to rise until reaching a maximum

369 of ~80% around 37 cm, 26 cm, and 23 cm in cores S3, S2, and S1, respectively. Thereafter, the

370 contributions of clay to the sediment matrix was variable, but generally remained well above

371 background levels (Fig. S2). In contrast, the cores from Anderson Lake had a high (between 70%

372 and 90%) contributions of silt (i.e. particle of a size between 63-2µm) and low contributions of

373 clay (<1%) to the sediment matrix. Core S2 from Seton Lake was characterized by a discrete

374 sand horizon (63µm ≥) from 27 to 33 cm, which was treated as a single short-term event, and

375 excluded from the age-depth model (Fig. S2).

17

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 18 of 45

376 The unsupported 210Pb activity, which derives from the atmospheric flux (versus

377 supported which derives from in situ decay, see Appleby 2001 for more information) in all cores

378 from Anderson and Seton lakes was generally low, but declined exponentially with depth in each

379 core (A1, r2=0.89; A2, r2=0.90; S1, r2= 0.85, S2, r2= 0.90 and S3, r2= 0.78), suggesting that the

380 cores were minimally disturbed (Fig. S1). The surficial total 210Pb concentration was below 100

381 Bq·Kg-1 in the Seton Lake cores, and just below 350 Bq·Kg-1 in the cores from Anderson Lake. A

382 peak in the 137Cs activity (> 50 Bq·Kg-1) was present at 9 cm and 8 cm in cores A1 and A2

383 (Anderson Lake). In Seton Lake, the peak in 137Cs activity occurred at 38 cm, 25 cm, and 20 cm

384 in cores S3, S2, and S1, respectively (Fig. S1), above the initial increase in clay, which was

385 associated with the time at which an important volume of glacially-turbid water was diverted for

386 the first time into Seton Lake ca. 1940 (DraftFigs. S2, S3). For the cores from Anderson Lake, 210Pb-

387 derived dates overestimated the age of the sediment down core and therefore were anchored to

388 the 137Cs dates in the CRS-model ( i.e. estimation of ca. 1960 from 210Pb was 1 cm below 137Cs

389 estimates; Appleby 2001).

390 Altogether, we were able to develop an age-depth model for the Seton Lake cores using

391 data from the grain size and radioisotope analysis, and for the Anderson Lake cores using the

392 radioisotope analysis. Results of the age-depth models are provided in the Supplemental material

393 (Fig. S4). It is important to note that while the age-depth model does not provide a sample-

394 specific date for each slice of sediment, it provides a reasonable estimate of the magnitude of

395 change of Seton Lake sedimentation rate.

396 Sedimentary cladoceran and diatom analyses

397 Total cladoceran and diatom fluxes in Seton Lake decreased post ca. 1960, whereas

398 fluxes increased or remained stable in Anderson Lake (Fig. 4). During this period, both Bosmina 18

https://mc06.manuscriptcentral.com/cjfas-pubs Page 19 of 45 Canadian Journal of Fisheries and Aquatic Sciences

399 spp. and Daphnia longispina complex fluxes decreased in the Seton Lake cores (Fig. 5). Prior to

400 ca. 1960, core S3 had the lowest cladoceran and diatom fluxes of all cores and experienced an

401 ~2- to 3-fold decline post 1960. Cores S1 and S2 from Seton Lake displayed much higher fluxes

402 prior to 1960, and experienced ~3 to 10-fold declines post-1960 (Fig. 4). In Anderson Lake, the

403 total cladoceran and diatom fluxes remained constant relative to Seton Lake, with low variability

404 until ca. 1970-80, when they increased (Fig. 4), contributed to by an increase of both Bosmina

405 spp. and Daphnia longispina complex (Fig. 5). In both lakes, the concentration (Fig. S6) and flux

406 (Fig. 4) data show similar trends but different magnitudes. Bosmina spp. and the Daphnia

407 longispina complex were the dominant cladoceran groups (>90% cumulative relative abundance)

408 found in the sediments from both lakes (Fig. 5). In Seton Lake, Bosmina spp. was the dominant

409 taxonomic group prior to ca.1945. The relativeDraft abundance of the Daphnia longispina complex

410 then increased between the 1940s and 1950s. This change in relative abundance coincided with

411 an abrupt decrease in the flux of both Bosmina spp. and Daphnia longispina in S2 and S3 (Fig.

412 5). In the Anderson Lake cores, a shift from a Bosmina spp. to Daphnia longispina complex also

413 occurred, but started in the 1920s (Fig. 5).

414

415 Discussion

416 Together, our modern and paleo-data support substantial declines in the primary

417 (diatoms) and secondary (zooplankton) producers in Seton Lake following the establishment of

418 the BRD (Fig. 4). The modern limnological data enabled a direct comparison of the two lakes,

419 and the paleo-data allows a pre- and post-diversion perspective, that is not available through any

420 other means. Our modern limnological data indicate that the euphotic zone of Seton Lake is

421 approximately 10 m shallower than that of Anderson Lake. Similarly, secondary production (i.e. 19

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 20 of 45

422 zooplankton density and biomass) is currently 3- to 10- fold lower in Seton Lake relative to

423 Anderson Lake, and gross primary production was lower in Seton Lake in July and August (Figs.

424 2, 3). Our analysis of the sedimentary remains of diatom and cladoceran sub-fossils allowed the

425 reconstruction of long-term spatial and temporal variation in the primary and secondary

426 production of these two systems over the last ~200 years, in order to determine whether the

427 lower primary and secondary production observed in the modern data between 2014-2016 in

428 Seton Lake is consistent with changes induced by the Bridge River Diversion (BRD). The fluxes

429 of diatom and cladoceran sub-fossils indicate that the production in Seton Lake declined greatly

430 following the establishment of the BRD (Fig. 4). Conversely, the fluxes in the cores from

431 Anderson Lake (reference system) exhibited stable or increasing trends over the last fifty years

432 (Fig. 4), which is consistent with pronouncedDraft regional warming, which is apparent in the mean

433 annual and winter temperatures (BC-MoE 2016). The large decline in primary and secondary

434 production in Seton Lake may be due to direct or indirect mechanisms related to the BRD,

435 including changes in the availability of light and nutrients, and/or modification of the water

436 residence time. Multiple lines of evidence suggest that the intrusion of glacial flour is the

437 primary mechanism by which the BRD negatively affected the production of Seton Lake.

438

439 Effects of glacial flour on lake production

440 The intrusion of turbid water can induce complex modifications to lake ecosystems.

441 Enhanced inputs of glacial meltwater can result in direct and indirect impacts on primary and

442 secondary production through several ecological processes which include changes in light

443 penetration, food-web interactions, nutrient availability, and water residence time (Hylander et

20

https://mc06.manuscriptcentral.com/cjfas-pubs Page 21 of 45 Canadian Journal of Fisheries and Aquatic Sciences

444 al. 2011; Sommaruga 2015). The importance and potential significance of these various factors

445 are discussed below.

446

447 Change in light availability and effects on primary production

448 Glacial meltwater can induce important changes in light penetration, reducing the extent

449 of the euphotic zone. A survey of twenty-two lakes indicated that lakes receiving meltwater rich

450 in glacial flour had the lowest water clarity, along with lakes affected by humic coloration (Irwin

451 1974). Analogous to the findings of this study, our limnological data indicate that the depth of

452 the euphotic zone in Seton Lake is approximately half that in Anderson Lake (Fig. 2). Glacial 453 flour is often the primary factor responsibleDraft for light attenuation in glacially-fed lakes (Rose et 454 al. 2014). The high reflectance and scattering properties of glacial flour (Gallegos et al. 2008)

455 lessens the penetration of photosynthetically active radiation (Rose et al. 2014), thereby reducing

456 the depth at which photosynthesis can occur. Similarly, we found that water turbidity in Seton

457 Lake is 2-fold higher than in Anderson Lake, and measurements of turbidity and depth of the

458 euphotic zone were negatively correlated in Seton Lake (Fig. 2). The shallower euphotic zone in

459 Seton Lake is likely the direct result of the input of glacial flour from the BRD.

460 In lakes influenced by glacial flour, water clarity can become the main factor regulating

461 primary production, influencing plankton community structure and abundance (Sommaruga

462 2015). Joint and Pomroy (1981) found that turbid riverine inputs in the Bristol Channel (UK)

463 decreased primary production by up to 20-fold in regions of high turbidity. Prior to the diversion,

464 Seton Lake received its main inflow from Anderson Lake. Therefore, the changes in light

465 penetration induced by the BRD is likely the primary mechanism responsible for the decline in

21

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 22 of 45

466 the fluxes of diatom and cladoceran observed in the sediment cores (Fig. 4). The paleo-data

467 support a reduction in the abundance of diatom algae in Seton Lake as a result of the BRD,

468 highlighting that the diatom flux declined by 3 to 10-fold in the Seton Lake cores following the

469 establishment of the diversion (Fig. 4). Interestingly, the diatom flux from Seton Lake during the

470 post-diversion period were on average 2 to 3-fold lower than the Anderson Lake flux, whereas

471 the present-day estimates of gross primary production (GPP) show that Anderson Lake has only

472 a slightly higher GPP (p= 0.05; Fig. 2). The discrepancy observed between our limnological data

473 and paleolimnological data may be explained by the fact that the diatoms are not the dominant

474 algal group in these lakes (Limnotek 2017). Nevertheless, the limnological survey shows that the

475 inter-annual variability of GPP is high in both lakes (Fig. 2), and previous surveys have shown

476 that Seton Lake tends to be less productiveDraft than Anderson Lake (Geen and Andrews 1961;

477 Shortreed et al. 2001). As the data from our sediment cores are an integration of several years,

478 our paleo-data may imply that on average Seton Lake primary production is lower than in

479 Anderson Lake, a trend not as apparent in the more limited modern data. Moreover, the paleo-

480 data indicate that the diatom flux in the cores from Seton Lake were higher than Anderson Lake

481 pre-BRD, suggesting that the primary production in Seton Lake was higher than in Anderson

482 Lake before the establishment of the diversion.

483

484 Influence on lake secondary production

485 The overall lower post-BRD zooplankton densities observed in the Seton Lake cores (Fig.

486 2) and the significantly lower cladoceran density and biomass (Fig. 3) could be related to the

487 decline of the primary producers following the establishment of the diversion, and subsequent

488 bottom-up food-web interactions. The reduction of the cladoceran flux in the cores from Seton 22

https://mc06.manuscriptcentral.com/cjfas-pubs Page 23 of 45 Canadian Journal of Fisheries and Aquatic Sciences

489 Lake were coeval with the declines in diatom flux, suggesting that decline of the primary

490 producers, related to the increased turbidity, has negatively impacted the secondary producers in

491 Seton Lake. The adverse effects of turbid glacial meltwater on pelagic primary and secondary

492 producers of lacustrine systems have been previously recorded in paleolimnological studies of

493 the Holocene Late Glacial Maximum and Younger-Dryas (Birks and Birks 2008). The study

494 reported a decline in the phytoplankton and zooplankton abundance, corresponding to the time at

495 which the lakes received inputs of glacial runoff. Similar to these studies, our data indicate that

496 the timing of the decline of cladoceran and diatom fluxes in the sediment corresponded with the

497 novel increases in percent clay contributions to the sediment matrix of Seton Lake (Figs. 3, 5).

498 The currently lower secondary production found in Seton Lake could thus be a result of bottom-

499 up cascading food web effect induced byDraft the decline of lake primary production.

500 The input of glacial flour can also impair the filter-feeding apparatus of Cladocera

501 (Koenings et al. 1990; Kirk 1991). Often similar in size to food particles, filter-feeding

502 zooplankton are susceptible to ingest the suspended particles of clay. Koenings et al. (1990)

503 previously demonstrated that filter-feeding daphnid, bosmid, and taxa from the genus

504 Holopedium did not discriminate against glacial particles in glacially-turbid lakes at a threshold

505 turbidity of ≥5 NTU, which are similar to the turbidity levels reached in the early spring in Seton

506 Lake (Fig. 2). As the cladoceran assemblages from both Seton and Anderson lakes were almost

507 exclusively composed of Bosmina spp. and Daphnia spp. (Fig. 5, Fig. S5), two pelagic filter-

508 feeders commonly found in deep oligotrophic Sockeye Salmon nursery lakes, the input of glacial

509 flour to the system may have been detrimental to the overall cladoceran community. In fact, as

510 seen in Anderson Lake, spring appears to be the period at which zooplankton bloom (Figs. 2, 3).

511 However, corresponding with the high turbidity values, this spring zooplankton bloom is 3-fold

23

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 24 of 45

512 lower in Seton Lake than it is in Anderson Lake (Fig. 2). More importantly, the cladoceran

513 density and biomass remain low during that period in Seton Lake, whereas they increase by 4-

514 fold in Anderson Lake (Fig. 3). Moreover, the competitive abilities of Daphnia can be

515 substantially lessened relative to other zooplanktons in glacial flour-rich waters. In fact, previous

516 laboratory studies found that Daphnia feeding rates was affected to a greater extent compared to

517 copepods when exposed to glacial flour (Hart 1988). The disproportionate susceptibility to

518 glacial flour amongst zooplankton groups may therefore explain the large declines in cladoceran

519 flux post-1940s in the cores from Seton Lake (Fig. 4), the overall lower modern cladoceran

520 density found in Seton Lake relative to Anderson Lake (Fig. 3), as well as the more diverse

521 copepod and rotifer communities found in Seton Lake (Fig. S5). Draft 522

523 Changes in nutrient availability

524 Glacial flour can impact the biological availability of nutrients in freshwaters. Through

525 the sequestration of nutrients by colloidal binding (i.e. Fe-bound and apatite-calcite bound),

526 glacial flour can alter nutrient stoichiometry in lakes located downstream from glaciers, acting as

527 a sink or source of phosphorus (Hodson et al. 2004; Sommaruga 2015). Investigations of the

528 phosphorus content of glacial flour have shown that glacial runoff containing weathered minerals

529 may furnish bioavailable inputs of phosphorus in P-limited freshwater ecosystems, whereas

530 phosphorus adsorption to the minerals tends to occur in ecosystems with higher phosphorus

531 concentrations (Hodson et al. 2004). Oligotrophic lakes can be especially sensitive to the

532 influence of glacial flour on nutrient bioavailability, as slight increases in available nutrients can

533 alter algal assemblages (Hylander et al. 2011; Slemmons and Saros 2012). The similar GPP

534 observed in the spring in Seton and Anderson lakes could therefore suggest that Seton Lake 24

https://mc06.manuscriptcentral.com/cjfas-pubs Page 25 of 45 Canadian Journal of Fisheries and Aquatic Sciences

535 primary production may have benefited from an enhancement of nutrients, thereby compensating

536 for the lower light regime. While the bioavailable forms of phosphorus (SRP) were below the

537 detection levels in Seton and Anderson lakes, the stimulatory effect associated with an increase

538 in bioavailable nutrients can occur before nutrient levels are detectable (Hylander et al. 2011;

539 Slemmons and Saros 2012). As such it is difficult to assess whether the input of glacial flour

540 indeed enhanced the primary production through the transport of nutrients. Nonetheless, while

541 the GPP in Seton Lake remains low throughout the growing season, it increases from June to

542 September in Anderson Lake, and TN:TP molar ratios were above thresholds for P and N co-

543 limitation of autotrophic production in Seton Lake, supporting that phosphorus is the limiting

544 nutrient (Elser et al. 1990; Downing and McCauley 1992). As such, the stimulatory effect of

545 glacial flour is likely limited in Seton Lake.Draft It is also noteworthy that the colloidal binding

546 properties of glacial flour is relative to the type of bedrock underlying the originating glacier

547 (Hodson et al. 2004). The BRD glacial runoff comes from the Bridge Glacier which overlays on

548 Mesozoic and Cretaceous bedrocks rich in quartz diorite (BC digital Geology, Geospatial Data,

549 Government of British Columbia), a silica-rich mineral with low-phosphorus content and poor-

550 binding properties (Porder and Ramachandran 2013). The mineral properties of the glacial flour

551 may thus account for the lack of influence on the phosphorus levels in Seton Lake, and likely did

552 not have a stimulatory effect on primary production.

553 In regions where deposition of atmospheric reactive nitrogen is high, glacial runoff can

554 enrich freshwater ecosystems in nitrate, yet another mechanism by which glacial meltwater can

555 stimulate plankton production (Saros et al. 2010; Slemmons and Saros 2012). Although we did

556 not find data on atmospheric reactive nitrogen deposition in our study region, the concentrations

25

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 26 of 45

557 of TN in Seton Lake were lower than in Anderson Lake discounting any prolonged nitrate

558 enrichment resulting from the input of the glacial flour (Fig. 2).

559

560 Changes in Seton Lake water residence times, water levels, & temperature

561 Water residence time can influence zooplankton community structure and recruitment

562 (DFO 2014). However, the post-diversion water residence time of Seton Lake is relatively

563 constant and long enough (~320 days) to not adversely affect the productivity (i.e. residence time

564 far exceeds biological replacement). In fact, Obertegger et al. (2007) showed that in a lake

565 experiencing large fluctuations in flushing rate (from 23 to 786 days), zooplankton abundance 566 and community structure were significantlyDraft affected by changes in water residence time below a 567 threshold value of 193 days, a much shorter water residence than Seton Lake, post-diversion

568 (~320 days). It is thus unlikely that the large reductions in cladoceran flux, coincident with the

569 onset of the BRD, were the direct result of higher flushing rates in Seton Lake.

570 Likewise, the BRD and the dam constructed at the outflow of Seton Lake could have

571 modified water levels. While we do not know by how much, if at all, the water levels of Seton

572 Lake have changed since the establishment of the hydroelectric management, it is unlikely to

573 have had a significant long-term influence on the biological composition of this deep lake.

574 In fact, littoral cladoceran taxa were virtually absent from the sediment cores collected in both

575 Seton and Anderson lakes, supporting the fact that the littoral habitat is limited in those lakes

576 (Fig. 5).

577 Originating from the Bridge Glacier, the BRD could transport much colder waters into

578 Seton Lake, thereby influencing the primary and secondary production of Seton Lake.

26

https://mc06.manuscriptcentral.com/cjfas-pubs Page 27 of 45 Canadian Journal of Fisheries and Aquatic Sciences

579 Temperature profiles were taken monthly during the three years of limnological sampling at 6

580 stations along the length of Seton Lake and 2 stations in Anderson Lake (Limnotek 2017). The

581 results indicated that there were no differences observed between the limnological stations within

582 Seton Lake, nor between the lakes, thereby providing evidence that the BRD did not significantly

583 change the water temperature of Seton Lake (Limnotek 2017).

584

585 Changes in Seton Lake sedimentation rate

586 Changes in sedimentation rate can influence the accumulation of subfossils in the

587 sediments (Battarbee et al. 2002). The results from the grain size and radioisotope analysis (Figs. 588 S1, S2) demonstrate that the influx of glacialDraft flour increased the sedimentation rate of Seton 589 Lake, which would dilute of the concentrations of cladoceran and diatom microfossils. In order

590 to correct for potential bias associated with changes in sedimentation rate, the paleo-data were

591 expressed as flux. However, flux data depend upon accurate estimates of sedimentation rates,

592 and the linear age-depth models developed for the Seton Lake cores present some challenges.

593 For instance, the linear model assumes a constant sedimentation rate between set ages (Fig. S3).

594 The main factors regulating the sedimentation rate of Seton Lake during the post-diversion

595 period are changes in the water flow coming from the BRD, and the delivery of inorganic

596 material (i.e. amount of glacial flour being discharged by the BRD). The annual inflow coming

597 from the BRD is regulated, and has been relatively constant since its completion (~87 m3.sec-1

598 ±12, data courtesy of BC Hydro). However, as the Bridge Glacier is receding (Chernos 2014),

599 the influx of glacial flour may have varied through time. The grain size analysis indicates that the

600 percent clay contribution during the post-diversion period indeed varies (Fig. S2); however, it

601 remains on average 3 to 4-fold higher in magnitude compared to the pre-diversion period in all 27

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 28 of 45

602 cores from Seton Lake, as such the post-diversion sedimentation rate in Seton Lake remained on

603 average higher than the pre-diversion period. This is further supported by a first order

604 exponential decay of 210Pb in all cores from Seton Lake (Fig. S1), indicating an relatively

605 constant sedimentation rate. Moreover, the grain size analysis did not show any decline in the

606 percent of clay contribution in the top of the Seton Lake cores suggesting that the delivery of

607 glacial flour did not decline yet as a result of the Bridge Glacier retreating. Altogether, annual

608 water inflow and grain size data demonstrate that the main factors regulating the Seton Lake

609 sedimentation rate during the post-diversion period remained, on average, stable.

610 Another potential limitation to our linear age-depth model is the assumption that Seton 611 Lake sedimentation rate was similar to AndersonDraft Lake before the diversion was established. This 612 assumption is supported by the fact that Anderson and Seton lakes share similar morphological

613 properties, and are exposed to similar edaphic and climatic conditions. Various indigenous

614 accounts also refer to a landslide that separated the once single lake into two lakes (i.e. Seton and

615 Anderson lakes) at the Seton Portage about 8,000-20,000 year BP (Edwards 1985), further

616 supporting the assumption that the two lakes likely had a similar sedimentation rate before the

617 diversion was established. Nevertheless, the sedimentation rate can vary substantially within a

618 lake basin. This appears to be the case for the post-diversion period in Seton Lake, as illustrated

619 by the cesium peak (Fig. S1), and it is likely the result of the localized input of glacial flour from

620 the diversion which created a gradient of sedimentation rate within the deep depositional basin

621 (i.e. slower sedimentation rate in the cores located further from the diversion; Figs. S1, S2 & S4).

622 However, the age-depth model for the Anderson Lake cores developed from the radioisotopic

623 methods supports uniform sedimentation rate in the deep basin of Anderson Lake (Fig. S4), and

28

https://mc06.manuscriptcentral.com/cjfas-pubs Page 29 of 45 Canadian Journal of Fisheries and Aquatic Sciences

624 thus, without the influence of the diversion, the Seton Lake pre-diversion sedimentation rate can

625 be assumed to be similar between the three cores.

626 Lastly, our paleo-data suggest that the fluxes of cladoceran and diatom were higher in

627 Seton Lake than in Anderson Lake before the establishment of the diversion, and this pattern was

628 reversed after the diversion was established. Although we do not have limnological data from the

629 pre-diversion period, the post-diversion limnological data, as well as that of previous studies

630 (Geen and Andrews 1961; Shortreed et al. 2001) agree with our paleo-data inferences, indicating

631 that the post-diversion cladoceran community is less abundant and GPP is lower in Seton Lake

632 than in Anderson Lake. Therefore, the magnitude of changes in the cladoceran and diatom fluxes 633 from before to after BRD is likely to beDraft real, and primarily driven by a decline in primary and 634 secondary production rather than an artifact of dilution due to an increase in sedimentation rates

635 in Seton Lake.

636

637 Recent increases in the biological production of Anderson Lake

638 In contrast to Seton Lake, the Anderson Lake cores suggest an enhancement of both

639 primary and secondary production in Anderson Lake over the last fifty years (Fig. 4). Starting

640 between the 1960s and 1980s, the increased primary and secondary production in Anderson Lake

641 could reflect an increase in mean annual temperature. In fact, the central interior of British

642 Columbia has warmed on average 1 °C per century from 1900 to 2013 (BC-MoE 2016). As most

643 of the warming has occurred during the winter months with an increased average winter

644 temperature of 1.6 °C per century (BC-MoE 2016), the increase in production observed in

645 Anderson Lake cores may be related to changes in the seasonality of the lake induced from

29

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 30 of 45

646 warming atmospheric temperatures, including changes in the timing and magnitude of thermal

647 structure of the lake, lake ice phenology, and lake-water temperatures (Weyhenmeyer 1996). As

648 zooplankton density and biomass tended to be higher during the month of May and June in both

649 lakes (Figs. 2, 3), the increase in cladoceran flux observed in the Anderson Lake sedimentary

650 record could reflect longer ice-free periods and subsequent enhancement of the primary

651 production earlier in the spring (Schindler et al. 2005b). However, this hypothesis needs to be

652 further explored, as the diatom flux only increased in one of the Anderson Lake cores.

653

654 Effects of glacial flour on juvenile O. nerka food supply

655 Rates of primary production are Drafthighly related to fish biomass accrual (Downing et al. 656 1990) and the foundation for O. nerka productive capacity models throughout British Columbia

657 (Hume et al. 1996; Shortreed et al. 2000). Subsequent bottom-up control of the zooplankton

658 community strongly mediates trophic energy flows to O. nerka (Hyatt et al. 2004). Our results

659 suggest that the development of the BRD reduced the abundance of cladoceran, an important

660 food supply for O. nerka, relative to conditions prior to hydropower development.

661 The seasonal timing of zooplankton production can be an important factor regulating O.

662 nerka fry and smolt sizes. In lakes affected by glacial turbidity, juvenile Sockeye Salmon tend to

663 be smaller only in cold-water lakes where the peak of zooplankton production is delayed to later

664 in the summer (Koenings et al. 1985). In most Sockeye Salmon nursery lakes, peak fry

665 emergence occurs during the spring (Burgner 1991), a critical time for juvenile salmon to grow

666 and survive in freshwater. Our data indicate that zooplankton density peaks in May in both Seton

667 and Anderson lakes, suggesting that glacial flour does not induce a timing mismatch with

30

https://mc06.manuscriptcentral.com/cjfas-pubs Page 31 of 45 Canadian Journal of Fisheries and Aquatic Sciences

668 foraging potential for juvenile O. nerka (Fig. 2). However, the cladoceran biomass remained low

669 in Seton Lake throughout the growing season while in Anderson Lake, the cladoceran biomass

670 peaked in June (Fig. 3). As cladocerans are considered as an important and efficient conduit of

671 energy to juvenile O. nerka (Shortreed et al. 2001; Mazumder and Edmundson 2002), and have

672 been found to be an important food supply for the juvenile O. nerka of the Seton-Anderson lakes

673 systems the higher cladoceran biomass observed in Anderson Lake suggests consistent foraging

674 opportunities sustaining the food web for O. nerka, whereas the foraging opportunities in Seton

675 Lake are not maintained to the same extent.

676 677 Implications for other glacially influencedDraft fisheries lakes 678 The BRD resulted in a substantial decline in the ecologically important primary (diatoms)

679 and secondary (Cladocera) producers in Seton Lake. Following the establishment of

680 hydroelectric management, the bioavailable nutrient concentrations in Seton Lake and the

681 enhanced flushing rates were unlikely to have had a measurable influence on the lake production.

682 Rather, the reduction of light penetration and the shallower euphotic zone in Seton Lake,

683 consistent with the influx of the glacial flour-rich meltwater, is likely the primary mechanism

684 that triggered the observed declines in production. The presence of glacial flour in the system

685 may have further contributed to the reduction of cladoceran filter-feeders in Seton Lake, thus

686 affecting the composition of the zooplankton community. Overall, our results suggest that the

687 long-term input of glacial flour can substantially reduce the primary and secondary production of

688 O. nerka nursery lakes. As no data exist on juvenile O. nerka population abundance and size

689 from the pre-diversion period, the direct impacts of glacial water intrusions on growth and

690 survival dynamics in populations of O. nerka in Seton Lake cannot be directly assessed. 31

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 32 of 45

691 Nonetheless, the reduction of primary and secondary production in Seton Lake following the

692 BRD implicates large-scale changes to energy flow within the O. nerka food web.

693 Lake responses to glacial retreat may be region specific, presenting a spatio-temporal

694 dimension (Baraer et al. 2012; Bliss et al. 2014). Referred to as the peak water concept the early

695 stage of glacial retreat typically results in a peak of glacial meltwater runoff, followed by an

696 extended period of decline in discharge (Baraer et al. 2012). While most glaciers in British

697 Columbia may have passed this early stage and downstream rivers may experience a decline in

698 annual glacial runoff (Bliss et al. 2014), the upper Bridge River, which drains the Bridge Glacier,

699 has experienced continuous increase in flow from the Little Ice Age to the present (Moyer et al. 700 2016). As glacial retreat is accelerating,Draft the influences of turbid meltwater may thus vary greatly 701 among many lakes exposed to these upstream geomorphic processes. For instance, during the

702 early stage of glacial retreat, in the absence of potential compensatory mechanisms that enhance

703 upward trophic energy flows, such as changes in diel vertical migration of O. nerka permitting

704 longer daily foraging windows (Gregory 1993; Liljendahl-Nurminen et al. 2008; Hansen and

705 Beauchamp 2015), and/or reduced predation from piscivores (De Robertis et al. 2003; Pangle et

706 al. 2012), the expected increases in glacial flour delivery to O. nerka nursery lakes, and other

707 lakes experiencing similar influences, likely decrease their basal productive capacities to support

708 fish populations. This pattern may be slowly reversed at a later stage of glacial retreat as glacial

709 runoff declines (Bliss et al. 2014; Moyer et al. 2016).

32

https://mc06.manuscriptcentral.com/cjfas-pubs Page 33 of 45 Canadian Journal of Fisheries and Aquatic Sciences

710 Literature Cited

711 American Public Health Association (APHA), American Water Works Association (AWWA) 712 and Water Environment Federation (WEF). 2014. Standard methods for the examination 713 of water and wastewater. Accessed: March 2014 714 Appleby, P.G. 2001. Chronostratigraphic techniques in recent sediments. In Tracking 715 Environmental Change Using Lake Sediments. Edited by W.M. Last and J.P. Smol. 716 Kluwer Academic Publishers, Dordrecht. pp. 171–203. doi:10.1007/0-306-47669-X_9. 717 Appleby, P.G. 2002. Chronostratigraphic Techniques in Recent Sediments. In Tracking 718 Environmental Change Using Lake Sediments. Edited by W.M. Last and J.P. Smol. 719 Kluwer Academic Publishers, Dordrecht. pp. 171–203. doi:10.1007/0-306-47669-X_9. 720 Baraer, M., Mark, B.G., McKenzie, J.M., Condom, T., Bury, J., Huh, K.-I., Portocarrero, C., 721 Gómez, J., and Rathay, S. 2012. Glacier recession and water resources in Peru’s 722 Cordillera Blanca. J. Glaciol. 58(207): 134–150. doi:10.3189/2012JoG11J186. 723 Battarbee, R.W., Jones, V.J., Flower, R.J., Cameron, N.G., Bennion, H., Carvalho, L., and 724 Juggins, S. 2002. Diatoms. In Tracking Environmental Change Using Lake Sediments. 725 Edited by J.P. Smol, H.J.B. Birks, W.M. Last, R.S. Bradley, and K. Alverson. Kluwer 726 Academic Publishers, Dordrecht. pp. 155–202. doi:10.1007/0-306-47668-1_8. 727 Battarbee, R.W., and Kneen, M.J. 1982. The use of electronically counted microspheres in 728 absolute diatom analysis. Limnol.Draft Oceanogr. 27(1): 184–188. 729 doi:10.4319/lo.1982.27.1.0184. 730 BC-MoE. 2016. Indicators of climate change for British Columbia 2016 Update. British 731 Columbia Ministry of Environment, Victoria. 732 Bell, L.M. 1985. A Fish passage problem at the Seton hydroelectric project in southwestern 733 British Columbia. Can. Water Resour. J. 10(1): 32–39. doi:10.4296/cwrj1001032. 734 Binford, M. 1990. Calculation and uncertainty analysis of 210Pb dates for PIRLA project lake 735 sediment cores. J. Paleolimnol. 3(3). doi:10.1007/BF00219461. 736 Birks, H.J.B., and Birks, H.H. 2008. Biological responses to rapid climate change at the Younger 737 Dryas—Holocene transition at Kråkenes, western Norway. The Holocene 18(1): 19–30. 738 doi:10.1177/0959683607085572. 739 Bliss, A., Hock, R., and Radić, V. 2014. Global response of glacier runoff to twenty-first century 740 climate change. J. Geophys. Res. Earth Surf. 119(4): 717–730. 741 doi:10.1002/2013JF002931. 742 Blott, S.J., and Pye, K. 2001. GRADISTAT: a grain size distribution and statistics package for 743 the analysis of unconsolidated sediments. Earth Surf. Process. Landf. 26(11): 1237–1248. 744 doi:10.1002/esp.261. 745 Burgner, R.L. 1991. Life history of Sockeye Salmon (Oncorhynchus nerka). In Pacific Salmon 746 Life Histories, Groot C. and Margolis L. UBC Press, Vancouver, BC. pp. 3–117. 747 Chernos, M. 2014. The relative importance of calving and surface ablation at a lacustrine 748 terminating glacier: a detailed assessment of ice loss at Bridge Glacier, British Columbia. 749 University of British Columbia. Available from 750 https://open.library.ubc.ca/cIRcle/collections/24/items/1.0166946. 751 De Robertis, A., Ryer, C.H., Veloza, A., and Brodeur, R.D. 2003. Differential effects of turbidity 752 on prey consumption of piscivorous and planktivorous fish. Can. J. Fish. Aquat. Sci. 753 60(12): 1517–1526. doi:10.1139/f03-123.

33

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 34 of 45

754 DFO. 2014. Technical review of the effects of the site C Clean Energy Project on fish and fish 755 habitat of the , British Columbia. Canadian Science Advisory Secretariat 756 Science Response 2014/011. 757 Downing, J.A., and McCauley, E. 1992. The nitrogen: phosphorus relationship in lakes. Limnol. 758 Oceanogr. 37(5): 936–945. doi:10.4319/lo.1992.37.5.0936. 759 Downing, J.A., Plante, C., and Lalonde, S. 1990. Fish production correlated with primary 760 productivity, not the Morphoedaphic Index. Can. J. Fish. Aquat. Sci. 47(10): 1929–1936. 761 doi:10.1139/f90-217. 762 Edwards, I. 1985. Short portage to Lillooet. Cold Spring Books, Mission, B.C. 763 Elser, J.J., Marzolf, E.R., and Goldman, C.R. 1990. Phosphorus and nitrogen limitation of 764 phytoplankton growth in the freshwaters of North America: A review and critique of 765 experimental enrichments. Can. J. Fish. Aquat. Sci. 47(7): 1468–1477. doi:10.1139/f90- 766 165. 767 Fegel, T.S., Baron, J.S., Fountain, A.G., Johnson, G.F., and Hall, E.K. 2016. The differing 768 biogeochemical and microbial signatures of glaciers and rock glaciers: Defining glaciers- 769 type signatures. J. Geophys. Res. Biogeosciences 121(3): 919–932. 770 doi:10.1002/2015JG003236. 771 Finney, B.P. 2000. Impacts of climatic change and fishing on Pacific Salmon abundance over the 772 past 300 years. Science 290(5492): 795–799. doi:10.1126/science.290.5492.795. 773 Finney, B.P., Gregory-Eaves, I., Douglas, M.S.V., and Smol, J.P. 2002. Fisheries productivity in 774 the northeastern Pacific Ocean overDraft the past 2,200 years. Nature 416(6882): 729–733. 775 doi:10.1038/416729a. 776 Gallegos, C.L., Davies-Colley, R.J., and Gall, M. 2008. Optical closure in lakes with contrasting 777 extremes of reflectance. Limnol. Oceanogr. 53(5): 2021–2034. 778 doi:10.4319/lo.2008.53.5.2021. 779 Geen, G.H., and Andrews, F.J. 1961. Limnological changes in Seton Lake resulting from 780 hydroelectric diversions. Progress Report, International Pacific Salmon Fisheries 781 Commission. 782 Glew, J.R., Smol, J.P., and Last, W.M. 2002. Sediment Core Collection and Extrusion. In 783 Tracking Environmental Change Using Lake Sediments. Edited by W.M. Last and J.P. 784 Smol. Kluwer Academic Publishers, Dordrecht. pp. 73–105. doi:10.1007/0-306-47669- 785 X_5. 786 Grant, S.C.H., and Pestal, G. 2012. Integrated biological status assessments under the Wild 787 Salmon Policy using standardized metrics and expert judgement: Fraser River Sockeye 788 Salmon (Oncorhynchus nerka) Case Studies. Canadian Science Advisory Secretariat 789 Research Document 2012/016. 790 Gregory, R.S. 1993. Effect of Turbidity on the Predator Avoidance Behaviour of Juvenile 791 Chinook Salmon (Oncorhynchus tshawytscha). Can. J. Fish. Aquat. Sci. 50(2): 241–246. 792 doi:10.1139/f93-027. 793 Gregory-Eaves, I., Selbie, D.T., Sweetman, J.N., Finney, B.P., and Smol, J.P. 2009. Tracking 794 sockeye salmon population dynamics from lake sediment cores: A review and synthesis. 795 In Challenges for diadromous fishes in a dynamic global environment: proceedings of the 796 international symposium “Challenges for Diadromous Fishes in a Dynamic Global 797 Environment” held in Halifax, Nova Scotia, Canada, June 18-21, 2007. Edited by A. 798 Haro and American Fisheries Society. American Fisheries Society, Bethesda, Md.

34

https://mc06.manuscriptcentral.com/cjfas-pubs Page 35 of 45 Canadian Journal of Fisheries and Aquatic Sciences

799 Hansen, A.G., and Beauchamp, D.A. 2015. Latitudinal and photic effects on diel foraging and 800 predation risk in freshwater pelagic ecosystems. J. Anim. Ecol. 84(2): 532–544. 801 doi:10.1111/1365-2656.12295. 802 Hart, R.C. 1988. Zooplankton feeding rates in relation to suspended sediment content: potential 803 influences on community structure in a turbid reservoir. Freshw. Biol. 19(1): 123–139. 804 doi:10.1111/j.1365-2427.1988.tb00334.x. 805 Healey, M. 2011. The cumulative impacts of climate change on Fraser River sockeye salmon 806 (Oncorhynchus nerka) and implications for management. Can. J. Fish. Aquat. Sci. 68(4): 807 718–737. doi:10.1139/f2011-010. 808 Hodson, A., Anesio, A.M., Tranter, M., Fountain, A., Osborn, M., Priscu, J., Laybourn-Parry, J., 809 and Sattler, B. 2008. Glacial ecosystems. Ecol. Monogr. 78(1): 41–67. doi:10.1890/07- 810 0187.1. 811 Hodson, A., Mumford, P., and Lister, D. 2004. Suspended sediment and phosphorus in proglacial 812 rivers: bioavailability and potential impacts upon the P status of ice-marginal receiving 813 waters. Hydrol. Process. 18(13): 2409–2422. doi:10.1002/hyp.1471. 814 Horppila, J., and Liljendahl-Nurminen, A. 2005. Clay-Turbid interactions may not cascade-A 815 Reminder for Lake Managers. Restor. Ecol. 13(2): 242–246. doi:10.1111/j.1526- 816 100X.2005.00031.x. 817 Hume, J.M., Shortreed, K.S., and Morton, K.F. 1996. Juvenile sockeye rearing capacity of three 818 lakes in the Fraser River system. Can. J. Fish. Aquat. Sci. 53(4): 719–733. 819 doi:10.1139/f95-237. Draft 820 Hyatt, K.D., McQueen, D.J., Shortreed, K.S., and Rankin, D.P. 2004. Sockeye salmon 821 (Oncorhynchus nerka) nursery lake fertilization: Review and summary of results. 822 Environ. Rev. 12(3): 133–162. doi:10.1139/a04-008. 823 Hylander, S., Jephson, T., Lebret, K., Von Einem, J., Fagerberg, T., Balseiro, E., Modenutti, B., 824 Souza, M.S., Laspoumaderes, C., Jonsson, M., Ljungberg, P., Nicolle, A., Nilsson, P.A., 825 Ranaker, L., and Hansson, L.-A. 2011. Climate-induced input of turbid glacial meltwater 826 affects vertical distribution and community composition of phyto- and zooplankton. J. 827 Plankton Res. 33(8): 1239–1248. doi:10.1093/plankt/fbr025. 828 Ichimura, S., Parsons, T.R., Takahashi, M., and Seki, H. 1980. A comparison of four methods for 829 integrating 14C-primary productivity measurements per unit area. J. Oceanogr. Soc. Jpn. 830 36(5): 259–262. doi:10.1007/BF02072127. 831 IPCC. 2014. Climate Change 2014: Synthesis Report. Contribution of Working Groups I, II and 832 III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. 833 IPCC. 834 Irwin, J. 1974. Water clarity records from twenty- two New Zealand lakes. N. Z. J. Mar. Freshw. 835 Res. 8(1): 223–227. doi:10.1080/00288330.1974.9515498. 836 Jarvis A., H.I. Reuter, A. Nelson, E. Guevara, 2008, Hole-filled seamless SRTM data V4, 837 International Centre for Tropical Agriculture (CIAT), available from 838 http://srtm.csi.cgiar.org. 839 Jeppesen, E., Leavitt, P., De Meester, L., and Jensen, J.P. 2001. Functional ecology and 840 palaeolimnology: using cladoceran remains to reconstruct anthropogenic impact. Trends 841 Ecol. Evol. 16(4): 191–198. doi:10.1016/S0169-5347(01)02100-0. 842 Jiilavenkatesa, A., Dapkuna, S.J., and Lum, L.H. 2001. NIST recommended practice guide: 843 particle size characterization. National Institute of Standards and Technology, 844 Gaithersburg, MD. doi:10.6028/NBS.SP.960-1.

35

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 36 of 45

845 Joint, I.R., and Pomroy, A.J. 1981. Primary production in a turbid estuary. Estuar. Coast. Shelf 846 Sci. 13(3): 303–316. doi:10.1016/S0302-3524(81)80028-X. 847 Kirk, K.L. 1991. Inorganic particles alter competition in grazing plankton: The role of selective 848 feeding. Ecology 72(3): 915–923. doi:10.2307/1940593. 849 Koenings, J.P., Burkett, G.B., and Kyle, G.B. 1985. Limnological and fisheries evidence for 850 rearing limitation of sockeye production in crescent Lake, southcentral Alaska (1979- 851 1982). FRED Division Report, Alaska Department of Fish and Game. 852 Koenings, J.P., Burkett, R.D., and Edmundson, J.M. 1990. The exclusion of limnetic Cladocera 853 from turbid glacier-meltwater lakes. Ecology 71(1): 57–67. doi:10.2307/1940247. 854 Korhola, A., and Rautio, M. 2001. Cladocera and Other Branchiopod Crustaceans. In Tracking 855 Environmental Change Using Lake Sediments. Edited by J.P. Smol, H.J.B. Birks, and 856 W.M. Last. Springer Netherlands, Dordrecht. pp. 5–41. doi:10.1007/0-306-47671-1_2. 857 Korosi, J.B., Kurek, J., and Smol, J.P. 2013. A review on utilizing Bosmina size structure 858 archived in lake sediments to infer historic shifts in predation regimes. J. Plankton Res. 859 35(2): 444–460. doi:10.1093/plankt/fbt007. 860 Korosi, J.B., and Smol, J.P. 2012a. An illustrated guide to the identification of cladoceran 861 subfossils from lake sediments in northeastern North America: part 1—the Daphniidae, 862 Leptodoridae, Bosminidae, Polyphemidae, Holopedidae, Sididae, and Macrothricidae. J. 863 Paleolimnol. 48(3): 571–586. doi:10.1007/s10933-012-9632-3. 864 Korosi, J.B., and Smol, J.P. 2012b. An illustrated guide to the identification of cladoceran 865 subfossils from lake sediments inDraft northeastern North America: part 2—the Chydoridae. J. 866 Paleolimnol. 48(3): 587–622. doi:10.1007/s10933-012-9636-z. 867 Kurek, J., Korosi, J.B., Jeziorski, A., and Smol, J.P. 2010. Establishing reliable minimum count 868 sizes for cladoceran subfossils sampled from lake sediments. J. Paleolimnol. 44(2): 603– 869 612. doi:10.1007/s10933-010-9440-6. 870 Laspoumaderes, C., Modenutti, B., Souza, M.S., Bastidas Navarro, M., Cuassolo, F., and 871 Balseiro, E. 2013. Glacier melting and stoichiometric implications for lake community 872 structure: zooplankton species distributions across a natural light gradient. Glob. Change 873 Biol. 19(1): 316–326. doi:10.1111/gcb.12040. 874 Liljendahl-Nurminen, A., Horppila, J., and Lampert, W. 2008. Physiological and visual refuges 875 in a metalimnion: an experimental study of effects of clay turbidity and an oxygen 876 minimum on fish predation. Freshw. Biol. 53(5): 945–951. doi:10.1111/j.1365- 877 2427.2008.01952.x. 878 MacCauley, E. 1984. The estimation of the abundance and biomass of zooplankton in samples. 879 In A Manual on Methods for the Assessment of Secondary Productivity in Fresh Waters. 880 Edited by J.A. Downing and F.H. Riglers. Oxford, U.K.: Blackwell Scientific 881 Publications. pp. 228–265. 882 Mazumder, A., and Edmundson, J.A. 2002. Impact of fertilization and stocking on trophic 883 interactions and growth of juvenile sockeye salmon (Oncorhynchus nerka). Can. J. Fish. 884 Aquat. Sci. 59(8): 1361–1373. doi:10.1139/f02-111. 885 Moreira, A.L., and Taylor, E.B. 2015. The origin and genetic divergence of “black” kokanee, a 886 novel reproductive ecotype of Oncorhynchus nerka. Can. J. Fish. Aquat. Sci. 72: 1584– 887 1595. doi:10.1139/cjfas-2015-0145. 888 Moyer, A.N., Moore, R.D., and Koppes, M.N. 2016. Streamflow response to the rapid retreat of 889 a lake-calving glacier: Streamflow Response to Lake-Calving Glacier Retreat. Hydrol. 890 Process. 30(20): 3650–3665. doi:10.1002/hyp.10890.

36

https://mc06.manuscriptcentral.com/cjfas-pubs Page 37 of 45 Canadian Journal of Fisheries and Aquatic Sciences

891 Nehlsen, W., Williams, J.E., and Lichatowich, J.A. 1991. Pacific Salmon at the Crossroads: 892 Stocks at Risk from California, Oregon, Idaho, and Washington. Fisheries 16(2): 4–21. 893 doi:10.1577/1548-8446(1991)016<0004:PSATCS>2.0.CO;2. 894 Nykänen, M., Vakkilainen, K., Liukkonen, M., and Kairesalo, T. 2009. Cladoceran remains in 895 lake sediments: a comparison between plankton counts and sediment records. J. 896 Paleolimnol. 42(4): 551–570. doi:10.1007/s10933-008-9304-5. 897 Obertegger, U., Flaim, G., Braioni, M.G., Sommaruga, R., Corradini, F., and Borsato, A. 2007. 898 Water residence time as a driving force of zooplankton structure and succession. Aquat. 899 Sci. 69(4): 575–583. doi:10.1007/s00027-007-0924-z. 900 Pangle, K.L., Malinich, T.D., Bunnell, D.B., DeVries, D.R., and Ludsin, S.A. 2012. Context- 901 dependent planktivory: interacting effects of turbidity and predation risk on adaptive 902 foraging. Ecosphere 3(12): art114. doi:10.1890/ES12-00224.1. 903 Parsons, T.R., Maita, Y., and Lalli, C.M. 1984. A manual of chemical and biological methods for 904 seawater analysis. In 1st ed. Pergamon Press, Oxford [Oxfordshire] ; New York. 905 Porder, S., and Ramachandran, S. 2013. The phosphorus concentration of common rocks—a 906 potential driver of ecosystem P status. Plant Soil 367(1–2): 41–55. doi:10.1007/s11104- 907 012-1490-2. 908 Rose, K.C., Hamilton, D.P., Williamson, C.E., McBride, C.G., Fischer, J.M., Olson, M.H., 909 Saros, J.E., Allan, M.G., and Cabrol, N. 2014. Light attenuation characteristics of 910 glacially-fed lakes: transparency of glacially-fed lakes. J. Geophys. Res. Biogeosciences 911 119(7): 1446–1457. doi:10.1002/2014JG002674.Draft 912 Saros, J.E., Rose, K.C., Clow, D.W., Stephens, V.C., Nurse, A.B., Arnett, H.A., Stone, J.R., 913 Williamson, C.E., and Wolfe, A.P. 2010. Melting alpine glaciers enrich high-elevation 914 lakes with reactive nitrogen. Environ. Sci. Technol. 44(13): 4891–4896. 915 doi:10.1021/es100147j. 916 Schelske, C.L., Peplow, A., Brenner, M., and Spencer, C.N. 1994. Low-background gamma 917 counting: applications for 210-Pb dating of sediments. J. Paleolimnol. 10(2): 115–128. 918 doi:10.1007/BF00682508. 919 Scheuerell, J.M., Schindler, D.E., Scheuerell, M.D., Fresh, K.L., Sibley, T.H., Litt, A.H., and 920 Shepherd, J.H. 2005. Temporal dynamics in foraging behavior of a pelagic predator. Can. 921 J. Fish. Aquat. Sci. 62(11): 2494–2501. doi:10.1139/f05-164. 922 Schindler, D.E., Leavitt, P.R., Brock, C.S., Johnson, S.P., and Quay, P.D. 2005a. Marine-derived 923 nutrients, commercial fisheries, and production of salmon and lake algae in Alaska. 924 Ecology 86(12): 3225–3231. doi:10.1890/04-1730. 925 Schindler, D.E., Rogers, D.E., Scheuerell, M.D., and Abrey, C.A. 2005b. Effects of changing 926 climate on zooplankton and juvenile sockeye salmon growth in southwestern alaska. 927 Ecology 86(1): 198–209. doi:10.1890/03-0408. 928 Schoen, E.R., Wipfli, M.S., Trammell, E.J., Rinella, D.J., Floyd, A.L., Grunblatt, J., McCarthy, 929 M.D., Meyer, B.E., Morton, J.M., Powell, J.E., Prakash, A., Reimer, M.N., Stuefer, S.L., 930 Toniolo, H., Wells, B.M., and Witmer, F.D.W. 2017. Future of Pacific Salmon in the face 931 of environmental change: Lessons from one of the world’s remaining productive salmon 932 regions. Fisheries 42(10): 538–553. doi:10.1080/03632415.2017.1374251. 933 Selbie, D.T., Lewis, B.A., Smol, J.P., and Finney, B.P. 2007. Long-Term population dynamics of 934 the endangered Snake River Sockeye Salmon: Evidence of past influences on stock 935 decline and impediments to recovery. Trans. Am. Fish. Soc. 136(3): 800–821. 936 doi:10.1577/T06-100.1.

37

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 38 of 45

937 Selbie, D.T., Sweetman, J.N., Etherton, P., Hyatt, K.D., Rankin, D.P., Finney, B.P., and Smol, 938 J.P. 2011. Climate change modulates structural and functional lake ecosystem responses 939 to introduced anadromous salmon. Can. J. Fish. Aquat. Sci. 68(4): 675–692. 940 doi:10.1139/f11-006. 941 Shi, H., Chen, G., Lu, H., Wang, J., Huang, L., Wang, L., Zhao, S., and Liu, X. 2016. Regional 942 pattern of Bosmina responses to fish introduction and eutrophication in four large lakes 943 from Southwest China. J. Plankton Res. 38(3): 443–455. doi:10.1093/plankt/fbv118. 944 Shortreed, K.S., Hume, J.M.B., and Stockner, J.G. 2000. Using photosynthetic rates to estimate 945 the juvenile Sockeye Salmon rearing capacity of British Columbia lakes. In Sustainable 946 Fisheries Management: Pacific Salmon. Edited by E.E. Knudsen, D.D. Steward, J.E. 947 MacDonald, J.E. Williams, and D.W. Reiser. CRC Press, USA. pp. 505–521. 948 Shortreed, K.S., Morton, K.F., Malange, K., and Hume, J.M.B. 2001. Factors limiting juvenile 949 sockeye production and enhancement potential for selected BC nursery lakes. CSAS, 950 Canadian Science Advisory Secretariat. 951 Slemmons, K.E.H., and Saros, J.E. 2012. Implications of nitrogen-rich glacial meltwater for 952 phytoplankton diversity and productivity in alpine lakes. Limnol. Oceanogr. 57(6): 1651– 953 1663. doi:10.4319/lo.2012.57.6.1651. 954 Smol, J.P. 1992. Paleolimnology: an important tool for effective ecosystem management. J. 955 Aquat. Ecosyst. Health 1(1): 49–58. doi:10.1007/BF00044408. 956 Sommaruga, R. 2015. When glaciers and ice sheets melt: consequences for planktonic 957 organisms. J. Plankton Res. 37(3):Draft 509–518. doi:10.1093/plankt/fbv027. 958 Steeman Nielsen, E. 1952. The use of radioactive carbon (14C) for measuring organic production 959 in the sea. J. Cons. Int. Explor. Mer. 18: 117–140. 960 Stewart-Oaten, A., Murdoch, W.W., and Parker, K.R. 1986. Environmental Impact Assessment: 961 “Pseudoreplication” in Time? Ecology 67(4): 929–940. doi:10.2307/1939815. 962 Stockner, J.G., and Macisaac, E.A. 1996. British Columbia Lake Enrichment Program: two 963 decades of habitat enhancement for Sockeye Salmon. Regul. Rivers Res. Manag. 12(4– 964 5): 547–561. doi:10.1002/(SICI)1099-1646(199607)12:4/5<547::AID- 965 RRR407>3.0.CO;2-M. 966 Sweetman, J.N., and Finney, B.P. 2003. Differential responses of zooplankton populations 967 (Bosmina longirostris) to fish predation and nutrient-loading in an introduced and a 968 natural sockeye salmon nursery lake on Kodiak Island, Alaska, USA. J. Paleolimnol. 969 30(2): 183–193. doi:10.1023/A:1025543421436. 970 Szeroczyñska, K., and Sarmaja-Korjonen, K. 2007. Atlas of subfossil Cladocera from Central 971 and Northern Europe. Friends of the Lower Vistula Society, Świecie. 972 Vinebrooke, R.D., Thompson, P.L., Hobbs, W., Luckman, B.H., Graham, M.D., and Wolfe, A.P. 973 2010. Glacially mediated impacts of climate warming on alpine lakes of the Canadian 974 Rocky Mountains. SIL Proc. 1922-2010 30(9): 1449–1452. 975 doi:10.1080/03680770.2009.11902351. 976 Weyhenmeyer, G.A. 1996. The influence of stratification on the amount and distribution of 977 different settling particles in Lake Erken. Can. J. Fish. Aquat. Sci. 53(6): 1254–1262. 978 doi:10.1139/f96-057. 979

980

38

https://mc06.manuscriptcentral.com/cjfas-pubs Page 39 of 45 Canadian Journal of Fisheries and Aquatic Sciences

981 Acknowledgements

982 This study was funded by BC Hydro, with administration provided by Teri Neighbour,

983 Ahmed Gelchu, Darin Nishi, and Jeff Walker. We thank Bonnie Adolf, Gilda Davis, and Jude

984 Manahan of St’át’imc Eco-Resources Ltd who managed the study as part of a larger water use

985 planning project. Access to BC Hydro field facilities was provided by Dorian Turner. Field staff

986 included Garett Lidin for the cores collection; and Allison Hebert, Dani Ramos, Annika Putt,

987 Marc Laynes, Tyler Creasey, L.J Wilson, Caroline Melville, John Goes, Jessica Hopkins, Petra

988 Wykpis and Frank Richings for the limnological sampling. Limnological data compilations were

989 completed by Mike Chung. Lab support was provided by Shannon Harris, Danusia Dolecki, 990 Lech Dolecki, Dr. Scott Lamoureux, KeithDraft Harper, Ron Kerr, Shauna Bennett and staff of 991 Chemical Engineering Department, Queen’s University, Kingston, Ontario, Canada; Fisheries

992 and Oceans lab at Cultus Lake, British Columbia, Canada; ALS Environmental, Burnaby,

993 Canada.

994

39

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 40 of 45

995 Figure Captions

996 Figure 1: Location of Seton Lake and Anderson Lake on the Fraser River watershed in British 997 Columbia, Canada. Location of the five sediment cores used in this study (A1 & A2 in Anderson 998 Lake; S1, S2 & S3 in Seton Lake), and limnological monitoring stations. Catchment area slope 999 rasters were generated from the NASA Shuttle Topography Radar Digital Elevation Model 1000 (Jarvis et al. 2008) using the slope tool in ESRI's Spatial Analyst extension. The water ways 1001 and water bodies shapefiles were retrieved from BCData Catalogue. 1002 Figure 2: Variation of the monthly average (over the three sampling years 2014, 2015 and, 2016) 1003 of the depth of the Euphotic Zone (estimated from the PAR measurements), turbidity, total 1004 phosphorus (TP), total nitrogen (TN), gross primary production (GPP) and total zooplankton 1005 density in Seton Lake (grey line), and Anderson Lake (blue, dotted line). Error bars represent the 1006 standard deviation. 1007 Figure 3: Monthly average of the cladoceran (left panel) and copepod (right panel) densities (top 1008 graphs) and biomasses (bottom graphs) in Seton Lake (grey circle) and Anderson Lake (blue 1009 triangle). Error bars represent the standard deviation. 1010 Figure 4: Total cladoceran flux (#individuals·cm-2·year-1 ×102) and total diatom flux (#valves·cm- 1011 2·year-1 ×108) for the Anderson Lake cores (black line and triangular symbols, A1 and A2) and 1012 the Seton Lake cores (grey line and squareDraft symbols, S1, S2 and S3) over the last 200 years. The 1013 blue shaded area illustrates the turbidity gradient of the two lakes (dark blue corresponding to 1014 clear water, and teal corresponding to turbid water rich in glacial flour) and indicates the timing 1015 of the construction of the diversion. 1016 Figure 5: Relative abundance (%, bar graph) and flux (#individuals·cm-2·year-1 ×102, grey shaded 1017 area) of the dominant cladoceran groups found in the cores from Seton Lake (cores S1, S2, S3) 1018 and Anderson Lake (cores A1 and A2) over the last 200 years. The dots along the Seton Lake 1019 cores correspond to the intervals where the minimum count of 20 individuals could not be 1020 reached. The red arrow corresponds to the time of establishment of the diversion.

40

https://mc06.manuscriptcentral.com/cjfas-pubs Page 41 of 45 Canadian Journal of Fisheries and Aquatic Sciences

r oi C rv a e rp es enter R Ü

Alberta Seton Lake S3 British Columbia British ?!

S2 S1 &< Lillooet ?! &

A2 ?!

Legend Washington D'Arcy &< Monitoring Stations ! ?! Coring Sites 0 100 200 400 0 2.5 5 10 Bridge River Diversion Km Km

Draft

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 42 of 45

May June July Aug Sept Oct 0 (m) 20 Euphotic ZoneEuphotic 6 Turbidity 40 (NTU) 3 0 6 ) 1 - L · 3 TP g μ ( 0 100 ( μ g TN · L - 1 60 ) ) 1 - 20 400 day day · 3 - m GPP · 200 mgC ( Zooplankton Zooplankton density 0 (

Draft# ind · 1000 m - 3 X 10 500 3 ) 0

https://mc06.manuscriptcentral.com/cjfas-pubs Page 43 of 45 Canadian Journal of Fisheries and Aquatic Sciences

Cladocera Copepod ) 3 300 Seton Lake 1000

10 Anderson Lake x 3 - 150 500 m ⋅ Density 0 0 (#ind 600 150 ) 4 10 x 3 75 - 300 m ⋅ g Biomass μ 0 0 ( May June July Aug Sept Oct May June July Aug Sept Oct

Draft

https://mc06.manuscriptcentral.com/cjfas-pubs Canadian Journal of Fisheries and Aquatic Sciences Page 44 of 45

40 20 A1 0 40 ) 2 20 A2 10

x 0 1 - flux 10 5 S3 year year

⋅ 0 2 - 40 cm ⋅ Cladocera 20 S2 ind 0 (# 50 S1 25 0 1800 1840 1880 1920 1960 2000

100 Draft A1 0 ) 8 100

10 A2 x 0 1 - 20 year year

⋅ 10 S3 2 - 0 cm

⋅ 50 Diatom flux Diatom 25 S2

valves 0 (# 50 S1 25 0 1800 1840 1880 1920 1960 2000 Year (C.E.)

https://mc06.manuscriptcentral.com/cjfas-pubs Page 45 of 45 Canadian Journal of Fisheries and Aquatic Sciences

S1 S2 S3 Bosmina spp. % Daphnia longispina Bosmina spp. % Daphnia longispina Bosmina spp. % Daphnia longispina complex % complex % complex % 0 50 100 0 50 100 0 50 100 0 50 100 0 50 100 0 50 100 ° ° ° 2000 ° ° ° ° ° ° ° ° ° ° 1960 ° ° °

) ° °

C.E 1920 Year ( Year

SetonLake 1880

1840

1800 0 25 50 0 5 10 0 10 20 0 5 10 0 5 10 0 5 A1 A2 0 50 100 0 50 100 0 50 100 0 50 100

2000

1960

1920

Year (C.E.) Year 1880 Anderson LakeAnderson 1840

1800 0 10 20 0 10 20 0 10 20 0 10 20 Flux (#ind⋅cm-2⋅year -1 x102) Draft

https://mc06.manuscriptcentral.com/cjfas-pubs