Full Paper 11th ISE 2016, Melbourne, Australia

MODELLING OF MOUNTAIN WHITEFISH (PROSOPIUM WILLIAMSONI) EGG

STRANDING IN THE COLUMBIA RIVER IN BRITISH COLUMBIA, CANADA AND
AN ASSESSMENT OF EFFECTS ON RECRUITMENT

DANA C. SCHMIDT, BRAD HILDEBRAND, SIMA USVYATSOV

Golder Associates Ltd., 201 Columbia Ave.
Castlegar, BC, V1N1A2,Canada

WOLF PLOEGER
Golder Associates Ltd., 102 2535 3 Ave SE,
Calgary, AB T2A7W5,Canada

The Lower Columbia River Egg Loss Model was created in the 1990’s to predict Mountain Whitefish
(Prosopium williamsoni) egg stranding as a result of hydropower operations on the lower Columbia River in
British Columbia, Canada. The model was updated between 2010 and 2013 using results from individual
RIVER2D hydraulic models developed for two key spawning areas, and incorporates spawning timing,
incubation time, egg deposition probabilities and dewatering time. The updated model estimated relative annual
stranding rates of eggs deposited during the winter months from the two spawning areas that were believed to
represent the Mountain Whitefish population of the lower Columbia River in Canada. Results from this work
provided updated tools to reduce the uncertainty related to the overall reliability of egg loss estimates and will
help guide future management of the Mountain Whitefish population in the lower Columbia River.

1 INTRODUCTION
Mountain Whitefish (Prosopium williamsoni) are the most abundant sportfish in the Keenleyside Reach of the
lower Columbia River [LCR - defined as the Columbia River from Hugh L. Keenleyside Dam (HLK) to the
Canada-US Border and including the lower Kootenay River below Brilliant Dam (BRD)]. This species uses this
area for all life history functions [1, 2]. Water level fluctuations associated with dam operations on both the
Columbia (HLK) and Kootenay (BRD) rivers can negatively impact whitefish spawning success by exposing
incubating embryos when water levels recede. In addition, armoured substrates found in regulated systems like
the lower Columbia River have been identified as potentially detrimental to whitefish egg survival by decreasing
the egg retention capabilities of incubation habitat. Flow regulation of the lower Columbia River may also affect
whitefish spawning behaviour, hatch periodicity, and hatch success through the modification of flows that may
provide essential spawning and hatching cues. Finally, flow fluctuations may also affect larval and juvenile
Mountain Whitefish, which prefer near-shore rearing habitats with relatively low velocities and gradients [3].
These concerns led to the development and initiation of BC Hydro’s Whitefish Flow Management (WFM)
program in the winter of 1994 and 1995. A series of intensive studies on Mountain Whitefish spawning and life
history characteristics were subsequently conducted annually between 1995 and 1999. These monitoring
programs identified that whitefish eggs are dewatered by flow changes in the lower Columbia River [4]. As part
of the ongoing implementation of a Water Use Plan (WUP) for the operations of the hydroelectric facilities in
this area, a series of Management Questions were developed through an open Stakeholder consultation process.
The primary question to be addressed by the studies is: “How do daily flow changes contribute to cumulative
channel dewatering in key spawning areas over the whitefish reproductive period?” To address this issue, in
2003, we developed the first version of a Mountain Whitefish Egg Loss Model (ELM), a tool that estimates the
risk of egg loss under alternative WFM flow scenarios [5]. Because of imprecision and high uncertainties, we
developed a detailed River2D hydraulic model [6] of two of the primary spawning areas for Mountain Whitefish
in the lower Columbia River with the details previously described [7]. The present study’s site selection,
approach, and design were based on the results of previous study programs on Mountain Whitefish distribution,
movements, spawning behaviour, habitat selection, and early life stage biology in the lower Columbia River,
plus the primary literature reviewed during and subsequent to these studies. This summary provides a brief
description of the approach and resulting knowledge gained by the study team from over 10 years of study.
Detailed descriptions of the modelling methodology, hydraulic modelling, field biological sampling and
assumptions used in development of the model are available in the primary report [5].

2 METHODS

2.1 Key Spawning Area Selection

Sampling effort and analysis for this program were concentrated in the two key Mountain Whitefish spawning
areas: CPR Island on the Columbia River and the lower Kootenay River in British Columbia, Canada (Figure 1
from [7]). Based on data from previous studies, these two areas are used consistently and extensively for
Mountain Whitefish spawning, and exhibit the range of depths, substrate, and velocity characteristics utilized by
the majority of spawning whitefish. Egg catch rates indicate a substantially lesser degree of use of other
spawning areas compared to the key spawning areas at the CPR Island and Kootenay River sites [8]. The
physical habitat characteristics at these spawning areas are similar to those found at other areas in the LCR; the
reasons why Mountain Whitefish spawn in some specific areas but not in other apparently similar areas is
unknown, but could be related to site fidelity or microhabitat conditions in the vicinity of spawning habitats.

COLUMBIA
SECTION

Figure 1. Map of the lower Kootenay and Columbia River’s in southern British Columbia, Canada. The specific
study area for ELM encompasses the “Kootenay Section” and “Columbia Section”, the reach of the Columbia
River below Norn’s Creek to just above the confluence with the Kootenay River (from [7]).
2.2 Mountain Whitefish Egg Loss Model (ELM) Development
The updated ELM includes data from the River2D Models [6], as well as data collected during Mountain
Whitefish spawn monitoring as part of the Whitefish Life History and Egg Mat Monitoring program [8]. The
locations of the cross sections used in the model are depicted in Figure 2.

Figure 2. The confluence areas of the lower Kootenay River and the Columbia River with the GRTS selected
cross sections covering the two major Mountain Whitefish spawning areas that are used for the detailed
hydraulic analysis.

A flow chart of the ELM described the combined processes used to develop the model (Figure 3). The
results of the 99 runs of the River2D models were interpolated to provide a continuous description of changes in
depth and velocity at the two spawning sites as a function of HLK and BRD discharge. The interpolation was
performed separately for each node, which resulted in individual functions of depth changes. The functions used
were selected to best fit each node. The output was a matrix of 6,206 and 3,873 nodes in Columbia and
Kootenay, respectively, with coordinates, water elevation, depth, and water velocity data associated with each
node. For each node, the probability of spawning on each day of the spawning season was estimated using the
depth and time from the beginning of the year. Then, incubation time was estimated using the input temp data,
which was in turn used to evaluate whether the node will dewater prior to egg hatch. The daily estimates of
dewatered egg proportions were then summed to provide a total yearly estimate of egg stranding.
 
Figure 3. Flow chart of the components used in modeling egg loss. Green boxes designate input data required for
every run, brown boxes designate model components developed for this study, and blue boxes designate steps in
the model computation. HLK=Hugh L. Keenleyside Dam on the Columbia River; BRD=Brilliant Dam on the
Kootenay River.

All statistical analyses were performed in R v. 3.0.1 [9]. All non-linear regressions were performed using
‘nls’, a non-linear least squares function in R. The incorporation of uncertainty into the estimates of stranding
was performed using bootstrapping, a resampling technique widely used to assess uncertainty [10]. In this study,
bootstrapping was used to provide confidence intervals around the final estimate of total egg stranding, as well
as for propagating the different sources of error throughout the model. These sources include:
o variability of Accumulated Thermal Units (ATU) required for hatching;
o variability of egg deposition levels with time; and,
o variability of egg deposition levels with depth.
The program estimated 300 stranding values based on the bootstrapping procedure that were then used to
create a 95% confidence interval around the median daily estimate of the proportion of stranded eggs out of eggs
deposited each day.

3 RESULTS
The calibrated model was run using 2007-2012 discharge and temperature data to predict relative egg loss
caused by stranding for at least 8 hours of dewatering (Table 1; Figure 4). The model indicated substantial
uncertainty in the estimates, but indicated operations result in different risks among years, with lower risk in
2007, 2010 and 2011 than in 2008, 2009 and 2012.

Table 1. Summary of ELM simulations for years 2007-2012. Values represent bootstrapped estimates of median
and 95% CI’s of the proportion of total eggs deposited that were stranded at each of the two spawning sites for
each year.

CPR Island Kootenay River

Lower Upper Lower Upper
Year 95% CI Median 95% CI 95% CI Median 95% CI
2007 0.133 0.272 0.461 0.043 0.114 0.258
2008 0.231 0.454 0.675 0.096 0.305 0.592
2009 0.146 0.33 0.553 0.078 0.327 0.738
2010 0.109 0.231 0.44 0.016 0.043 0.12
2011 0.146 0.298 0.521 0.034 0.087 0.187
2012 0.239 0.468 0.701 0.096 0.337 0.665
 1.0
CPR Island Kootenay

0.8

0.6
Total egg loss

0.4

0.2

0.0

2007 2008 2009 2010 2011 2012 2007 2008 2009 2010 2011 2012

Sampling year
Figure 4. Comparison of yearly total stranding estimates in both key Mountain Whitefish spawning areas from
2007 to 2012. Each year represents a spawning/hatching period from November 1 of indicated year to May 1 of
the next year. Egg loss represents the proportion of the total amount of deposited eggs that strands throughout the
period of November 1 to May 1 (median with 95% CI).

The model also provides data that can be mapped visually to determine spatially, what is the location of
most of the stranding, based on predictions of egg deposition rates and subsequent dewatering for any time
interval, based on the hydraulic conditions resulting from the operations of the two dam on the Columbia and
Kootenay Rivers. Figure 5 illustrates spatial deposition and relative egg stranding rates for the Kootenay River
during the 2003-2004 spawning and egg incubation season.
Figure 5. Maps of total egg deposition (upper panel) and proportion of stranded eggs out of those deposited
(lower panel) at Kootenay throughout the entire spawning and incubation periods (November 1 to May 1). The
maps are colour-coded based on deposition and stranding levels, respectively

These maps provide a tool for directing future field sampling efforts to maximize sampling efficiency and
also serve to identify areas of importance for Mountain Whitefish that may have dewatering avoided by future
mitigation.

4 DISCUSSION
The ELM has been developed as a tool to both predict future stranding under a given set of operations of two
hydroelectric dams on the Columbia and Kootenay Rivers in British Columbia, in addition to the uncertainty
associated with the predictions. The uncertainty developed is based on the imprecision of the field data used to
estimate various biological parameters related to the hydraulic model output. The hydraulic model was also
developed using sampling protocols [1] that could be used to determine uncertainty in its predictions of depth
and velocity (only depth was used in the model), but initial examination indicated that uncertainty was very low
when compared to the biological components of the model that were used in the simulations. The model is
currently being used to both guide operations to reduce egg stranding as well as the input into additional analysis
of egg loss rates related to recruitment of adult Mountain Whitefish [11]. The model also can be updated as new
data are acquired to refine parameter estimates and reduce uncertainty in estimated egg loss rates. The model
only estimates uncertainty related to observation error associated with field measurements of parameters
included in the model, but does not address process error. Possible process error include other factors that may
impact egg mortality predictions, such as predation rates on eggs, survival rates of dewatered eggs beyond 8
hours of dewatering, etc. In addition, the real question, “Does egg stranding impact Mountain Whitefish
populations?” requires additional complex modelling and relating other field sampling programs and their
uncertainty to the predictions of the ELM [11]. Because of major costs of operational changes to dam operations
during the high power demand period of the winter, which corresponds to the Mountain Whitefish spawning and
incubations season, improving our understanding of the risks and uncertainties associated with egg stranding
impacts on fish populations is an important endeavor.

5 ACKNOWLEDGMENTS
These studies have been funded by BC Hydro and are part of the ongoing Water Use Plan in British Columbia.
There have been numerous reviewers, but we would like to especially thank Guy Martel and James Baxter of BC
Hydro who have provided numerous suggestions and feedback into the development of these models and
contributions from Robyn Irvine and Joe Thorley of Poisson Consulting Ltd. who have assisted in the analysis of
much of the data used in field programs in developing these models.

REFERENCES

[1] Hildebrand, L., and K. English. “Lower Columbia Development. Lower Columbia River fisheries inventory.
1990 Studies.” Volume I Main Report. Prepared for BC Hydro, Environmental Resources by R.L. & L.
Environmental Services Ltd., in association with L.G.L. Ltd., Sydney, B.C.,(1991) 166 p. + app.
[2] R.L.&L. Environmental Services Ltd.. “Columbia Basin Developments Lower Columbia River. Fisheries
Inventory Program 1990 to 1994”. Prepared for BC Hydro, Environmental Affairs. R.L. & L.,(1995) Report
No. 381 95D: 156 p. + 7 app.
[3] R.L.&L. Environmental Services Ltd. “Lower Columbia River Mountain Whitefish monitoring program:
1994-1996 investigations.” Unpublished Report prepared for BC Hydro, Kootenay PS/PF. R.L. & L.,
(2001), Report No. 514F: 101 p + 8 app.
[4] Golder Associates Ltd. “Estimates of Mountain Whitefish Egg Stranding Mortality for potential Columbia
River Flow Reductions in 2002 - 2003.” Report prepared for BC Hydro, Castlegar, BC, (2003) Golder
Report No. 02-28-057D. 12 p.
[5] Golder Associates Ltd. “Lower Columbia River whitefish spawning ground topography survey: Year 3 data
report.” Report prepared for BC Hydro, Castlegar, (2013),BC Golder Report No. 10-1492-0142F: 68 p. + 3
app. URL=https://www.bchydro.com/content/dam/BCHydro/customerportal/documents/corporate/
environment-sustainability/water-use-planning/southern-interior/clbmon-47-yr3-2014-01-29.pdf
[6] Steffler, P. and J. Blackburn, “River2D: Two-Dimensional Depth Averaged Model of River Hydrodynamics
and Fish Habitat, Introduction to Depth Averaged Modeling and User's Manual”,(2002), University of
Alberta, Calgary, Alberta, CA, 119 pp.
[7] Schmidt, D., B. Hildebrand, D. Ciobotaru, W. Ploeger, D. DeRosa, and T. Oussoren. “Generalized random
tessellation stratified (GRTS) survey designs for hydraulic modeling of Mountain Whitefish egg deposition
and stranding on the Columbia River in British Columbia, Canada.” In: Proceedings of the 9th
International Symposium on Ecohydraulics, Vienna, (2012). 10 p.
[8] Golder Associates Ltd. “Lower Columbia River whitefish life history and egg mat monitoring program:
Year 5 interpretive report.“, (2014) Report prepared for BC Hydro, Castlegar, BC Golder Report No. 11-
1492-0111: 86 p. + 5 app.
[9] R Core Team.” R: A language and environment for statistical computing. R Foundation for Statistical
Computing, Vienna, Austria.” (2013). ISBN 3-900051-07-0, URL http://www.R-project.org/.
[10] Efron, B., and R. J. Tibshirani. “An introduction to the bootstrap.” (1993).Chapman & Hall, New York.
[11] DeHaan,M. ,J. Korman, G. Martel, J. Baxter, P. Bradshaw, and J. Thorley.” Whitefish Flows Management
Proposal, Mountain Whitefish Flows Technical Review Report”,(2015).Unpublished Report prepared by BC
Hydro, Burnaby,BC.. 46 p.