Factors affecting the survival of Chinook salmon (Oncorhynchus tshawytscha) embryos in upper and middle Columbia River watersheds, Washington State, USA
Data files
Feb 26, 2025 version files 9.97 GB
-
cjfas_2024_0250R_data_and_code.zip
9.97 GB
-
README.md
6.05 KB
Abstract
A greater understanding of environmental effects on salmon survival during incubation can aid in prediction and facilitate population modeling at local and regional scales. Four basin-scale studies of Chinook salmon egg-to-fry survival were conducted between 2009 and 2021in spawning in tributaries of the upper and middle Columbia River; each study population is currently supplemented or have a recent history of supplementation, and most listed under the U.S. Endangered Species Act. Together, these studies provided a unique opportunity to assess factors influential to egg-to-fry survival, such as fine sediment infiltration and substrate scour, in the context of differences in survival among rivers and study years. We detected a 16.8% decrease in the odds of survival for every 1-cm increase in substrate scour, a 1.0% decrease in the survival odds with each one-unit change in the percentage of accumulated fines, distinct differences in survival within and among rivers (from 30.5% to 82.5%), and subtle effects of parentage in the presence of environmental factors. The causal and predictive relationships provided here will inform conservation, restoration, and further research.
https://doi.org/10.5061/dryad.wm37pvmx4
Description of the data and file structure
We measured egg-to-fry survival and collected associated environmental data in the Methow, Entiat, Wenatchee, and Yakima river basins in Washington State, USA from 2009 to 2022. In total, the study area represents 380 river kilometers (rkm) of Chinook salmon spawning habitat in the middle and upper Columbia River system. Our study focused on spring Chinook salmon populations, i.e., those returning in the spring of the year, but spawning in the late summer and early fall. Our overall approach was as follows: We selected study sites representing the spatial range of spawning in each river system, fertilized eggs on-site, placed them in egg boxes, and buried them in areas routinely used by naturally spawning Chinook salmon at each study location. We measured substrate scour, fine sediment infiltration, substrate size and embeddedness, and temperature accumulation through the period of incubation at each site, and integrated parentage (i.e., sire and dam) into the study design. Refer to Johnson et al. (2012) for details on the methodology:
Johnson, Christopher L., Philip Roni, and George R. Pess. "Parental effect as a primary factor limiting egg-to-fry survival of spring Chinook Salmon in the upper Yakima River basin." Transactions of the American Fisheries Society 141, no. 5 (2012): 1295-1309.
Files and variables
File: cjfas_2024_0250_data_and_code.zip
Description: This compressed file contains files in four directories:
figures: code for figures 1-10 within the manuscript in .rmd file format
tables: code for tables: 2, 3, A1, A2, and A3 within the manuscript and Appendix 1 in .rmd file format
models: causal and predictive models in .rmd file format.
rds_files: data files to support the code listed above, and the raw data file: all_river_ef_dat. All are in .rds file format, and all are targeted by relative directories from within the figure, table, and model .rmd files. Data files are in the subdirectories for convenience: causal_desc_rds contains data files to support the causal model code, kfold_rds contains output used for model comparison, and pred_fit contains saved model fits as an alternative to running the code, which in some cases may take a very long time.
as mentioned above the raw data are fully provided in the all_river_ef_dat.rds file.
The dataframe consists of 25 columns and 1,197 rows.
columns are as follows:
include_exclude: a categorical column indicating if the data are considered sound with respect to the target analysis
river: the river from which the data were collected
brood_year: the year of Chinook spawning represented by the row of data
reach: the categorical reach of the river from which the data were collected
site: the categorical site (within reach) from which the data were collected
lat: site latitude WGS84
lon: site longitude WGS84
redd: categorical index of the respective artificial redd at each site
box_number: categorical index of the actual egg box placed within the artificial redd
cross: a categorical index value representing the male and female from which eggs and milt were collected
week: relative week of spawning/box placement
stock_date: the date on which the egg box was placed in the gravel
recovery_date: the date on which the egg box was recovered (or attempted recovery)
days: the number of days at large (recovery_date - stock_date)
atu: the number of accumulated thermal units from stocking to recovery (in degrees C)
n_stocked: the number of eggs within each egg box
n_survivors: the number of surviving fry following recovery of the egg box
scour_cm: centimeters of scour measured over the period of deployment, negative values represent deposition.
percent_fines: the percentage of particles =< 2mm in the egg box following recovery
d16: substrate size (mm) at the redd site for which 16% of the particles were smaller (see Wolman 1954; Kondolf et al. 2008)
d50: median substrate size (mm) at the redd site (see Wolman 1954; Kondolf et al. 2008)
d84: substrate size (mm) at the redd site for which 84% of the particles were smaller (see Wolman 1954; Kondolf et al. 2008)
percent_embedded: the percentage of particles at the redd site considered to be embedded (see Chadd 2007).
scour_loss: a categorical variable indicating if the egg box was lost and if so whether due to flow-related scour or redd superimposition
egg_weight_g: egg weight (g) for the female used in each of the egg boxes
Missingness in the data is indicated by an "NA" value.
Code/Software
Data and code were created using RStuido (Build 764), running R version 4.4.1.
Model scripts were written using CmdStanR or Rstan packages in R (Gabry and Cešnovar 2022; Stan Development Team 2023).
All scripts are saved in RStudio notebooks, *.rmd filetypes, and may be readily read and run using the freely available RStudio program.
Specific packages necessary to execute the code are explicitly noted within each .rmd file.
this compressed file contains files in four directories:
figures: code for figures 1-10 within the manuscript in .rmd file format
tables: code for tables: 2, 3, A1, A2, and A3 within the manuscript and Appendix 1 in .rmd file format
models: causal and predictive models in .rmd file format.
rds_files: data files to support the code listed above, and the raw data file: all_river_ef_dat. All are in .rds file format, and all are targeted by relative directories from within the figure, table, and model .rmd files. Data files are in the subdirectories for convenience: causal_desc_rds contains data files to support the causal model code, kfold_rds contains output used for model comparison, and pred_fit contains saved model fits as an alternative to running the code, which in some cases may take a very long time.