Reconstructing Fork Length at Ocean Entrance for Successfully Returned Cowichan River Chinook Salmon: Phase 2 (DRAFT)

Abstract

The impact of climate change on salmon stocks along the Pacific Northwest is a crucial issue, given that the muting or complete removal of specific life-history characteristics in salmon stocks have altered run timings and depressed adult returns. Climate change-related impacts and ecosystem alterations are uniquely present worldwide, and a recent report published by the Cowichan Valley Regional District (CVRD) has outlined these potential impacts on the Cowichan Valley and, specifically related to this study, the Cowichan River. Understanding the impacts of different hydrological and temperature regimes on the expression and success of certain traits in salmon species is essential in understanding climate change impacts. Fish otoliths have been used extensively for providing the most accurate fish age determinations due to their continued growth throughout life. Juvenile Chinook salmon life histories can be determined using minor elements, and the ability of managers to utilize data that demonstrates the impacts different hydrological regimes have on the success of certain traits can be of great use in regulated systems such as the Cowichan River.

Introduction

The diversity of life-history characteristics (genetics) has been cited as one of the most significant drivers in population dynamics and resilience (Hanski 1998; Sturrock et al. 2015; Wittmann et al. 2018). The muting and complete removal of specific life-history characteristics in salmon stocks along the Pacific Northwest have resulted in altered run timings and depressed salmon stocks (Heath et al. 2003; Smith et al. 2014; Jones et al. 2018). With climate change altering rivers’ hydrological dynamics (discharge and temperature), understanding these alterations’ impact on pacific salmon stocks is crucial. Climate change-related impacts and ecosystem alterations are uniquely present worldwide (Schindler et al. 2008). A recent report published by the Cowichan Valley Regional District (CVRD) has outlined these potential impacts on the Cowichan Valley and, specifically related to this study, the Cowichan River.

The impacts of climate change on the Cowichan Valley are projected to result in a doubling of summer days above 25°C, and the “1-in-20 hottest temperature is projected to increase from 33°C to 37°C by the 2050s” (CVRD 2017). Furthermore, this projected increase in temperature will likely cause an increase in dry spells by about 20%; this equates to a rise in dry spell days from 22 to 26 days annually (CVRD 2017). The Cowichan Valley will see an increase in precipitation by 5% (CVRD 2017) delivered during extreme precipitation events, which will also increase by 65% (99th percentile wettest days indicator), and snowpack is projected to decrease 85% by the 2050s (CVRD 2017). Furthermore, the increase in precipitation will not occur during critical times, and summer rainfall is expected to decrease by 17% (CVRD 2017). The projected reductions in snowpack and summer precipitation create challenges for water management. Sustaining critical river flow and temperature conditions during the Chinook salmon spring outmigration period – and through the summer months for stream-type salmon and trout species – is a top priority. Understanding the impacts different hydrological and temperature regimes have on the expression and success of certain traits in salmon species are essential in understanding climate change impacts; this is especially important for river systems impacted by decreasing snowpack and increased drought conditions (Schindler et al. 2008).

Fish otoliths (ear stones) are calcified structures in the saccule and utricle of the inner ear that have been used to decipher juvenile salmon’s life histories for many years (Campana 1999). Unlike scales, the chemical elements contained in otoliths do not reabsorb during food deprivation and starvation (Campana 1999). Consequently, otoliths have been used extensively for providing the most accurate fish age determinations due to their continued growth throughout life. Recently, microchemistry has been used as a method to develop life history maps for juvenile salmon during their freshwater-rearing phase (Kennedy et al. 2002). The elemental composition of otoliths is dominated by calcium, oxygen, and carbon, which comprise the two calcium carbonate (CaCO3) matrices (Campana 1999). Although these are the major components, 31 elements have been detected in fish otoliths (Campana 1999). The minor elements that can be utilized for microchemistry life-history purposes are Na, Sr, K, S, N, Cl and P; these elements are found in concentrations of >100 mg/l (Campana 1999).

Juvenile Chinook salmon life histories can be determined using minor elements (Na, Sr, K, S, N, Cl, and P) (Campana 1999). Chinook salmon, one of the most economically and ecologically important, have recently declined within the Salish Sea and, more broadly the Pacific Ocean (DFO 2011). Specifically, the Cowichan River Chinook are an indicator stock for the Salish Sea (DFO 2015). Juvenile salmon exhibit significant variation in size, timing and age at which they outmigrate from their natal rivers. These variations and the subsequent expression and success of specific phenotypic traits can be driven by hydroclimatic conditions experienced during critical developmental periods (Beechie et al. 2010; Sturrock et al. 2015).

In a system such as the Cowichan River, where flow is regulated, the ability of managers to utilize data that demonstrates the impacts different hydrological regimes have on the success of certain traits can be of great use. Moreover, it has been suggested that the Cowichan Watershed has been impacted by climate change more than the majority of rivers on Vancouver Island (C. Wightman, British Columbia Conservation Foundation, personal communication. 2018). Cowichan River Chinook salmon are mostly comprised of the ocean-type life history phenotype, with juveniles migrating to the ocean within three months of emergence from the gravel. However, this life history phenotype’s size- and timing-specific characteristics and its contribution to the adult spawning population are relatively unknown (K. Pellett, Department of Fisheries and Oceans, personal communication. 2018). Historically the Cowichan River supported a spring-run Chinook population (stream-type) that resided in the river for over one year before outmigration (T. Kulchyski, Cowichan Tribes, personal communication. 2020).

The diversity of Cowichan River Chinook life history phenotypes and the population’s overall abundance has declined over the last few decades (Slaney et al. 1996). However, recent returns to the system have been historically high (K. Pellett, personal communication), and the widespread exploitation of this stock is far below what it was in the 1900s. Diversification of life-history strategies has been linked to the overall resilience of salmon populations; this is based on empirical evidence that multiple life-history strategies reduce long-term population variance and extinction risk (Reisenbichler and Rubin 1999).

This study aims to provide information on fork length size at marine emergence for Cowichan River Chinook during different hydrological regimes and their subsequent contribution to the adult spawning population. This report documents the study’s second phase, which examined the 2014, 2015 and 2016 outmigration years. Additional Chinook adult samples from 2013 are included in the initial analysis. Peak outmigration for fry Cowichan Chinook occurs between mid-March and early April (Nagtegaal 1996; 1998-2002), while peak fingerling outmigration happens from mid-May to mid-June. There are few data to suggest when the S1 outmigration occurs. Results of this study, once completed and with the inclusion of a number of different years with differing hydrological conditions, Cowichan Tribes, regional biologists and fisheries managers may be able to understand better the juvenile Chinook population dynamics of the Cowichan River. As a result, managers may be able to forecast subsequent adult returns better and assess the resilience of this stock in the face of climate change.

Study Area

The Cowichan River is located in the southeastern portion of Vancouver Island, B.C. and flows southeast for 47 km before draining into Cowichan Bay and the Salish Sea (Fig. 1 and Fig. 2). With a watershed area of approximately 939 km2 and a mean annual discharge of 53 m³/s, the Cowichan River ranks fourth in size on Vancouver Island. The Cowichan River flow is regulated by a dam located at the outlet of Cowichan Lake. Historically, summer base flows have been regulated at 7.5 m3/s and have been reduced to 4.5 m3/s to conserve water for mill operations and sustain river flows throughout the summer months.

Figure 1. Map of Cowichan River. Map highlights the Cowichan River mainstem, its inlet at Cowichan Lake and its outlet in Cowichan Bay. 2023.

Materials and Methods

Hydrological Regime

Hydrological data were compiled from the Water Survey of Canada’s Cowichan Lake Outlet station (08HA002). Discharge data have gone through quality assurance and have been published to three decimal places due to the Water Survey of Canada’s standard practice. The temperature was recorded at the same station (08HA002); however, this data as it is not an official Water Survey of Canada product, has not undergone a quality assurance program.

Adult Chinook Sampling

To develop an understanding of outmigration phenotype contributions to adult returns, otolith samples were collected from returned adult Chinook salmon throughout the length of the river during fall spawn migrations. Samples were collected through multiple different sampling types (i.e. dead pitch and Broodstock). Sample collection data including date and location of sample, sample crew and Poh length were recorded where possible.

Samples collected from 2009, 2010, 2011, and 2012 outmigration years were included to show additional comparisons of mean size at outmigration and mean size at adult return. However, these samples were collected for a study assessing the 2013 adult return year and thus are comprised of four outmigration years which only comprise of a single adult return age class. These data were not included in phenotype contribution analyses.

All of the 205 Chinook salmon samples analyzed were from natural origin fish (i.e. wild cohorts). A summary of sample sizes for each ocean emergence year for the key outmigration years of this study (2014, 2015, and 2016) is provided in Table 1. The percent contribution of each age class (21, 31, and 41) to their escapement year are shown in Table 1.

Table 1. Adult Cowichan Chinook age structure, sample sizes, sex ratios and return year summary.

Otolith Elemental Analyses

Adult Chinook otoliths were embedded, polished, and analyzed at the Bob Wright Centre for Oceans, Earth and Atmospheric Sciences at the University of Victoria, Victoria, British Columbia, Canada. Cleaned and dried otoliths were placed sulcus-side down on a labelled plastic mould and covered in epoxy (Buehler Epoxy-Cure Resin). Otoliths were inspected before the epoxy cured to ensure no air bubbles were trapped against the otolith. After 12 hours of cure time, the hardened epoxy samples were removed from the moulds. Labelled otoliths were polished using waterproof, adhesive-backed silicon carbide lapping paper with 320, 600, and 1200 grit sizes (Allied High Tech). Using a smooth circular hand motion, the 320 grit size was used to grind close to the core, then 600 grit paper was used to expose the core and 1200 grit for smoothing. To prevent debris transfer and contamination, samples were rinsed with deionized water between each grit size. A final polish for 10 minutes using 0.25 µm diamond suspension spray (Buehler, Metadi Supreme) on 2500 grit polishing pads (Allied High Tech, Pan-B) resulted in a highly polished surface. Polished samples were rinsed and sonicated in deionized water for 1 minute and allowed to dry before ablation.

Microchemistry data were collected from polished otolith sections using Laser Ablation Inductively Coupled Plasma Mass Spectrometer (LA-ICP-MS) at the School of Earth and Ocean Sciences, University of Victoria. The LA-ICP-MS is a UP-213 Laser Ablation System (New Wave Research) coupled to an X Series II ICP-MS (Thermo Electron Corporation). The analyses were performed at a frequency of 20 Hz, using a spot size of 30 μm. The laser travelled 5.0 μm/second, and the mass spectrometer collected element concentrations for 25Mg, 43Ca, 55Mn, 66Zn, 86Sr, and 138Ba. The quantified NIST standard glasses 615, 613, and 611 were analyzed at the beginning and end of each run (after eight samples) to generate calibration curves for each element. Background intensity data were collected for 20 seconds before running the laser for each sample. Calcium was used as the internal standard to produce quantitative data assuming the aragonite in otoliths is composed of 40% Ca. PlasmaLab (ver.2.6.1.335, Thermo Electron 2007) software was used to complete data collection and reduction. The concentration data (mg/L) was exported to an Excel spreadsheet, where it was then converted to element:Ca molar mass ratios.

During data collection, the laser-ablated an otolith transect starting at the ventral edge, through the primordium, and finished at the dorsal edge (some samples were run in the reverse order). Otolith natal regions were determined by visually locating the exogenous feeding check using a compound microscope and measuring the distance from the primordium. Samples were removed from analyses if the core or exogenous feeding check were not visible using 40x magnification. Microchemistry data were collected from a 50 µm otolith transect which commenced immediately after the exogenous feeding check, representing the natal tributary chemistry signatures (Olley et al. 2011).

Reconstructed Size at Outmigration

The equation used for the back-calculation of fork lengths at marine entry is based on a fork length/otolith radius baseline prepared by Lance Campbell (Washington Department of Fish and Wildlife) developed from multiple Puget Sound Chinook populations (L. Campbell, personal communication). Fork length/otolith radius relationships vary significantly between ESUs (Evolutionary Significant Units). Thus, the Puget Sound baseline (Fig. 2) used to date should provide a close representation of the Cowichan Chinook population, but should be verified with a Cowichan fork length/otolith radius baseline to determine suitability.

We grouped fish into 5 mm fork length bins and categorized them into fry and fingerling, S1 (Sturrock et al. 2015). Sturrock et al. (2015), conducted a similar study on the Stanislaus River in the California Central Valley, USA and defined three migratory phenotypes (fry, parr and smolt) by size bins <55 mm, >55 to <75 mm and >75 mm, this was done based on previous work (Miller et al. 2010). For this study and based on discussions by local experts three life history phenotypes were decided fry (<55 mm), fingerling (>55 mm - <90 mm) and S1 (> 90 mm). While it is recognized that S1 Chinook are rare and there is overlap between these groups, the author feels this is a reasonable conclusion based on literature and experience with Cowichan Chinook.

knitr::include_graphics("Puget_Sound_Baseline.png")

Figure 2. Fork Length to otolith radius baseline, produced from Puget Sound Chinook salmon stocks

Phenotype Contributions to Adult Returns

The primary objective of this study was to determine outmigration phenotypic contributions to adult returns for each outmigration year analyzed in the study. To do this we attempted to mirror our sample sizes to outmigration year age class adult return contributions, provided by DFO (Table 2).

Factors Influencing Age (size) at Adult Return

Variation in age (size) at adult return is not well understood. In order to understand if environmental factors (hydrological regime) or biological factors (size at outmigration) influence the age or size at return for adult Cowichan Chinook Salmon we utilized a generalized linear mixed regression model (GLMM).

The mixed model analysis in this study was performed using the “glm” function in the R programming language. The response variable was age, which was scaled for analysis, and the predictor variables included life history stage, weighted average discharge, and sex. The model also included a random intercept for ocean entry year to account for potential clustering of individuals within years.

  Age at Return = β_0+ β_1 (life history type)+ β_2 (sex)+ β_3 (weighted average discharge)

The model was fit using maximum likelihood estimation, and model assumptions were checked by examining the distribution of residuals. The model’s goodness of fit was assessed using the null deviance, residual deviance, and AIC values. The null deviance represents the deviance of the null model, while the residual deviance represents the deviance of the fitted model. A lower residual deviance indicates a better fit of the model to the data. The AIC value provides a measure of the model’s goodness of fit, with lower values indicating a better fit.

Statistical significance was determined using p-values, with a significance level of 0.05 used as the threshold for statistical significance. Singularities in the model were also noted and reported.

Data Wrangling and Statistical Analysis

All data wrangling and statistical analysis were conducted using the statistical programming software R version 1.2.5033, “Orange Blossom” (R Development Core Team 2019). Data wrangling was completed with package dplyr (Wickham et al. 2020) All figures and graphs were produced by ggplot2 (Wickham 2009).The final report was created using R Markdown and knitr package (Xie et al. 2021; Xie, Y. 2021).

Results

Hydrological Regime

Upload river discharge and temperature data and calculate weighted averages for days between March 1 and July 1 of each year.This data will be used to derive relationships between river conditions and size at outmigration as well as assess their influences on size (age) of returning adults.

Mean discharge and temperature for the six month outmigration period for each of the key outmigration years 2014, 2015, and 2016 were 45.2, 26.8 and 36.9 m³/s (Fig 4). Mean weighted temperature for these years were xxx, 14.7 and 14.3 degrees Celsius (Fig 4).

Figure 3. Cowichan River discharge and temperature for all study years (2014 - 2016), from January 1 to July 1. Discharge is marked by years 2014 (Blue), 2015 (Red), and 2016 (Green). Temperature is marked by 2014-T (Blue), 2015-T (Red), and 2016-T (Green). Temperature values lie on the secondary y-axis. The value of 15°C, which is known to impair smoltification in Chinook salmon, is shown by the horizontal red dashed line.

Backcalculation of Size at Ocean Entrance

The fork length to otolith radius used for this study was produced from a compilation of Puget Sound Chinook stocks (Figure 2; Campbell L. WDFW. 2015). This relationship shows a positive linear trend across the full range of fork lengths (r²= 0.911)and was used to backcalculate fork length at marine entrance using the linear regression equation derived from the baseline, similar to Volk et al. (2010).

Backcalculated Size at Ocean Entrance All Years

The backcalculated size at ocean entrance for all years ranged from 22.16 to 165.60 mm (Figure 3). The mean fork length for each of the three phenotypes (fry, fingerling, and s1) across all years were 44.39, 73.06, and 101.30 mm, respectively (Figure 3). Between year variation in mean fork lengths were evident however they not outside of the calculated standard deviation and were not significantly different.

Figure 3. Cowichan River Chinook fork lengths at ocean entrance for all years separated by life history phenotype fry (red) fingerling (blue) and S1 (green). Mean fork length of each phenotype is marked by the black vertical line.

Fork Length at Adult Return By Ocean Entry Year

Fork length at adult return was estimated from POH length using an equation developed and provided by DFO stock assessment (y = 1.1985x + 21.212) (Kevin Pellett. DFO Stock Assessment, Personal. Communication. 2023). Fork length at adult return to be variable between outmigration years (Figure 4). However, while there were differences between fork lengths of each age class between years, there was still relative consistency between size at age of return for each year. Of note is the relatively larger age 2 fish that are returning from S1 life history phenotypes.

Figure 4. Cowichan River Chinook fork length and age at return.

Key Outmigration Years

Due to the nature of the dataset we have three years which have samples from the primary contributing age classes for adult returns in the Cowichan River. Ocean entrance years 2009 to 2012 were samples analyzed in 2015, for contributions from different life history strategies for the return year. However, assessing life history phenotype contributions based on outmigration year is preferred as we can then assess environmental variables that may contribute to their specific size at outmgiration and which then influences size at return. Due to this years 2009 - 2012 are not included in these additional analyses.

Adult Age Structure

For the key migration years, most returning Chinook matured at age 3 (~43%) and comprised 123 males and 82 females (Table 1). The next most numerous age class was 2-year-olds (Jacks) for all years (Table 1). The age-2 class was comprised primarily of male contribution (96%; Table 1). Sample sizes were chosen to reflect the percent contribution of age classes to each return year (Table 1 and Table 2). The abundance of adult returns to the Cowichan River are known to vary greatly between years and returns for the key migration years showed this trend (Table 2). Further, the contributions of age class to a given years returns are equally as variable (Table 2).

Table 2. Summary of Cowichan River Chinook Adult Return Estimates for Key Migration Years

Backcalculated Size at Ocean Entrance

The backcalculated size at ocean entrance for Phase 2 of this study ranged from 22.2 mm to 165.6 mm (Figure 5). The mean fork length for each of the three phenotypes (fry, fingerling and s1) across all years were 44.23, 74.65, and 101.65 mm, respectively (Figure 5). Between year varation in mean fork lengths were evident however these varitians were not outside of the calculated standard deviation and were not significantly different.

Figure 5. Fork Length at ocean entrance for the three study years and separated by life history phenotypes Fry (red), fingerling (blue) and S1 (green).

Contributions of Life History Phenotypes

Fingerlings were the most commonly observed phenotype in the survival adult population (58 - 68%), with fry and S1 phenotype contributions being similar (Figure 6). Moreover, fingerling contributed the most to the three-year-old class when assessing phenotype contributions to specific age classes and had similar, but variable, contributions to two- and four-year-olds (Figure 8). Life history phenotype did not show to have a strong influence on size or age at return (Figure 9).

Figure 6. Life history phenotype percent contributions to adult return age classes for Cowichan River Chinook

Figure 7. Life history phenotype percent contributions to adult return age classes for Cowichan River Chinook

Figure 8. Life history phenotype contributions to adult return age classes and associated fork lengths for Cowichan River Chinook

Linear Mixed Regression Model

In this study, a mixed model analysis was performed to investigate the relationship between age and several predictor variables in a fish population. The response variable, age, was scaled for analysis, and the predictor variables included life history phenotype, weighted average discharge, and sex. The model also included a random intercept for ocean entry year.

The results of the mixed model analysis showed that the intercept, representing age, was significantly related to the predictor variables weighted average discharge, and sex. Specifically, the lack of significance for phenotype was an interesting result. The estimate for weighted average discharge was 0.126, indicating a positive relationship with age and it was determined to be a significant preditor for age at return (P = 0.045). The estimate for sex was -6.20, suggesting that males tend to be younger than females, and it was also found to be a significant predictor on age at return ( P > 0.05).

However, the model reported singularities for ocean entry year, indicating that the model could not estimate a coefficient for this variable. The null deviance for the model was 11417.6 on 204 degrees of freedom, while the residual deviance was 8923.3 on 200 degrees of freedom.

Figure 9. Model-averaged, standardized coefficients from generalized linear regression models of age at return. Red vertical lines represent the extent of the 95% confidence intervals for each explanatory variable. Blue vertical lines represent the extent of the adjusted standard error for each explanatory variable. Coefficients are standardized so that the effect sizes are comparable among variables.

Discussion

This study aimed to investigate the relationship between age at return and several predictor variables in a fish population. A mixed model analysis was performed, and the results showed that weighted average discharge and sex were significant predictors for age at return; the lack of significance for life history phenotype was an interesting finding; however, the fry phenotype were shown to have a negative influence on age at return. The results suggest that life history phenotype, river discharge, and sex all play a role in age at return, but further analysis is required to fully understand the factors that influence the rate of jacking in the Cowichan population.

Data from previous PIT tagging projects in the Cowichan River and Bay have indicated that freshwater growth rates for juvenile Chinook salmon were on average, 0.6 mm/day, while marine growth rates were 1.0 mm/day (Pellett 2014; Pellet 2017). These increased growth rates in the marine environment may influence return rates and fry contributions to the age 2 population. As fry will have outmigrated months earlier than their fingerling counterparts and in some years (low temperature), this may provide increased growth that results in fish becoming large enough to return to spawn later in the same year. Fingerlings were the most commonly observed phenotype in the survival adult population, contributing to 58-68% of the population, while fry and S1 phenotype contributions were similar. The life history phenotype did not strongly influence size or age at return, suggesting that other factors, such as environmental conditions, may be more important determinants of age at return.

In order to fully understand what factors influence the rate of jacking in the Cowichan population, marine emergence time of year is required. As marine growth rates are almost double those found in freshwater, fry and their modestly higher contribution to the jack population may be influenced by their earlier marine emergence and increased growth rate (Nagtegaal et al. 2002).

Nagtegaal et al. (1996; 1998-2002) conducted previous Cowichan River fry abundance estimates using rotary screw traps and recapture estimates from bismark brown subsamples. Fry abundance estimates during these studies ranged from 173,225 - 3,964,347, showing that abundance is highly variable. However, it should be noted that adult returns during these years were low compared to present-day returns (K. Pellett, DFO Stock Assessment, personal communication. 2020). Nagtegaal et al. (2002), assumed that the fry population was the primary driver of the Cowichan Chinook population. However, it has been suggested that the fry population is a minor contributor, and this study provides further evidence of this (Figure 6; Figure 7). The 51 age class was not assessed as their numbers to the returning population were very low (K. Pellett, personal communication).

The back-calculated fork length data aligns with the PIT tag-calculated marine entrance data developed by the British Columbia Conservation Foundation’s Pacific Salmon Commission PIT tag study of Cowichan Chinook (Atkinson 2019; Appendix A Figure 1). Tagging restrictions limited the PIT tag study’s sampling to fish ≥ 65 mm. Comparing these fork length distributions at outmigration provides extra confidence in both data sets.

Fork lengths were based on a fork length/otolith radius baseline developed from multiple Puget Sound Chinook populations in Washington State. Caution should be taken when assessing the exact size at outmigration using this non-Cowichan baseline. However, trends in size and differences between size at ocean emergence and returning age class are unlikely to be meaningfully altered once a Cowichan-specific fork length/otolith radius baseline is developed.

Overall, this study provides important insights into the factors that influence age at return in the Cowichan River Chinook population. The results suggest that river discharge, sex, and potentially other environmental factors play a significant role in determining age at return. However, further research is required to fully understand the complexities of the factors that influence the rate of jacking in this population. This study also highlights the importance of using multiple data sources to confirm and validate results and the importance of considering the potential implications of environmental factors on population dynamics.

Next Steps

This analysis is based on an otolith-to-fork length baseline developed from a conglomerate of Chinook from Puget Sound. This is being used as a proxy for a Cowichan River-specific baseline. In 2022, the BCCF collected, mounted and constructed an otolith to fork length baseline from Cowichan Chinook captured in the marine environment in May. These samples, once analyzed showed to be not constructive for developing a baseline. Differences in daily growth rates were apparent in the data. These drastic differences in growth rates, likely attributed to time spent in the estuary and nearshore marine environment resulted in a poorly fitted baseline (y = 0.0248x + 37.733; r^2 = 0.5652).

Moving forward, the BCCF will collect 10 samples per 10 mm fork length bin from 30 mm to 130 (a total of 100 samples) from the Cowichan River mainstem to develop a Cowichan River-specific otolith width to fork length baseline.