Data from 2026 Mark-Recapture Analysis on Water and Endangered Fish in the Klamath Basin: Does Old Age or Upper Klamath Surface Elevation and Water Quality Affect Adult Lost River and Shortnose Sucker Survival? A Response to Kann and Walker (ver. 2.0, Sep

The data provided were used in a mark-recapture analysis conducted in 2026. Environmental covariates included water quality and Upper Klamath Lake surface elevation. Raw data were downloaded from the USGS Oregon Water Science Center for multiple sampling sites (422622122004000_MDNL_2026-4-13.csv, 422622122004003_MDNU_2026-4-10.csv, 422042121513100_RPT_2026-4-10.csv, 422719121571400_WMR_2026-4-13.csv, 422305121553800_MDTL_2026-4-13.csv, 422305121553803_2026-4-9.csv, MDNL_WQ2_Ammonia_2026-4-13.csv, WMR_WQ2_Ammonia_2026-4-13.csv, Elevation_2026-4-15.csv) and were formatted for use in mark-recapture models. A Program R script (Covariate_Cleaning2025Update.R) formats data and then writes multiple csv files with annual environmental covariates for use in subsequent mark-recapture models. Annual survival was estimated using a Cormack-Jolly-Seber (CJS) model for three adult sucker spawning populations: Lost River sucker lake (lrss) and river spawners (lrsr) and Shortnose suckers (sns). The recapture data for each population from 1999-2025 are found in the files lrsrboth.inp, lrssboth.inp, and snsboth.inp. R script ran CJS models using a maximum likelihood approach (LRSS_CJS_ProgramMARK.R, LRSR_CJS_ProgramMARK.R, and SNS_CJS_ProgramMARK.R) that compared multiple competing environmental survival models. Using a Bayesian framework, the top covariate survival models from the maximum likelihood approach for each population were run by calling JAGS (Just Another Gibbs Samplers) through R using the following scripts: LRSS_BayesianCJS_TopCovariateModel.R, LRSR_BayesianCJS_TopCovariateModel.R, SNS_BayesianCJS_TopCovariateModel.R. The previous version is available through ScienceBase, all the records in previous versions can be found in version 2.0 which is the most up to date and accurate. There may be uncorrected errors in the previous version and significant changes are noted in the "Version 2.0" text file: First release: 2022