Data from a long-term capture-recapture program were used to assess the status and dynamics of populations of two long-lived, federally endangered catostomids in Upper Klamath Lake, Oregon. Lost River suckers (Deltistes luxatus) and shortnose suckers (Chasmistes brevirostris) have been captured and tagged with passive integrated transponder (PIT) tags during their spawning migrations in each year since 1995. In addition, beginning in 2005, individuals that had been previously PIT-tagged were re-encountered on remote underwater antennas deployed throughout sucker spawning areas. Captures and remote encounters during spring 2012 were used to describe the spawning migrations in that year and also were incorporated into capture-recapture analyses of population dynamics.

Cormack-Jolly-Seber (CJS) open population capture-recapture models were used to estimate annual survival probabilities, and a reverse-time analog of the CJS model was used to estimate recruitment of new individuals into the spawning populations. In addition, data on the size composition of captured fish were examined to provide corroborating evidence of recruitment. Model estimates of survival and recruitment were used to derive estimates of changes in population size over time and to determine the status of the populations in 2011. Separate analyses were conducted for each species and also for each subpopulation of Lost River suckers (LRS). Shortnose suckers (SNS) and one subpopulation of LRS migrate into tributary rivers to spawn, whereas the other LRS subpopulation spawns at groundwater upwelling areas along the eastern shoreline of the lake.

In 2012, we captured, tagged, and released 749 LRS at four lakeshore spawning areas and recaptured an additional 969 individuals that had been tagged in previous years. Across all four areas, the remote antennas detected 6,578 individual LRS during the spawning season. Spawning activity peaked in April and most individuals were encountered at Cinder Flats and Sucker Springs. In the Williamson River, we captured, tagged, and released 3,376 LRS and 299 SNS, and recaptured 551 LRS and 125 SNS that had been tagged in previous years. Remote PIT tag antennas in the traps at the weir on the Williamson River and remote antenna systems that spanned the river at four different locations on the Williamson and Sprague Rivers detected a total of 19,321 LRS and 6,124 SNS. Most LRS passed upstream between late April and mid-May when water temperatures were increasing and greater than 10 °C. In contrast, most upstream passage for SNS occurred in early and mid-May when water temperatures were increasing and near or greater than 12 °C. Finally, an additional 1,188 LRS and 1,665 SNS were captured in trammel net sampling at pre-spawn staging areas in the northeastern part of the lake. Of these, 291 of the LRS and 653 of the SNS had been PIT-tagged in previous years. For LRS captured at the staging areas that had encounter histories that were informative about their spawning location, over 90 percent of the fish were members of the subpopulation that spawns in the rivers.

Capture-recapture analyses for the LRS subpopulation that spawns at the shoreline areas included encounter histories for more than 12,150 individuals, and analyses for the subpopulation that spawns in the rivers included more than 29,500 encounter histories. With a few exceptions, the survival of males and females in both subpopulations was high (greater than 0.9) between 1999 and 2010. Notably lower survival occurred for both sexes from the rivers in 2000, for both sexes from the shoreline areas in 2002, and for males from the rivers in 2006. Between 2001 and 2011, the abundance of males in the lakeshore spawning subpopulation decreased by 53–65 percent and the abundance of females decreased by 36–48 percent. Capture-recapture models suggested that the abundance of both sexes in the river spawning subpopulation of LRS had increased substantially since 2006; increases were due to large estimated recruitment events in 2006 and 2008. We know that the estimates in 2006 are substantially biased in favor of recruitment because of a sampling issue. We are skeptical of the magnitude of recruitment indicated by the 2008 estimates as well because (1) few small individuals that would indicate the presence of new recruits were captured in that year, and (2) recapture probabilities in recruitment models based on just physical recaptures were lower than desired for robust inferences from capture-recapture models. If we assume that little or no recruitment occurred in 2006 or 2008, the abundance of both sexes in the river spawning subpopulation likely has decreased at rates similar to the rates for the lakeshore spawning subpopulation between 2002 and 2011.

Capture-recapture analyses for SNS included encounter histories for more than 17,700 individuals. Most annual survival estimates between 2001 and 2010 were high (greater than 0.8), but SNS experienced more years of low survival than either LRS subpopulation. Annual survival of both sexes was particularly low in 2001, 2004, and 2010. In addition, male survival was somewhat low in 2002. Capture-recapture models and size composition data indicate that recruitment of new individuals into the SNS spawning population was trivial between 2001 and 2005. Models indicate substantial recruitment of new individuals into the SNS spawning population in 2006, 2008, and 2009. As a result, capture-recapture modeling suggests that the abundance of adult spawning SNS was relatively stable between 2006 and 2010. We are skeptical of the estimated recruitment in 2006, 2008, and 2009 because few small individuals that would indicate the presence of new recruits were captured in any of those years, and recapture probabilities in recruitment models were low. The best-case scenario for SNS, based on capture-recapture recruitment modeling, indicates that the abundance of males in the spawning population decreased by 71 percent and the abundance of females decreased by 69 percent between 2001 and 2011. The worst-case scenario, which assumes no recruitment and seems more likely, suggests an 86 percent decrease for males and an 81 percent decrease for females.

Despite relatively high survival in most years, we conclude that both species have experienced substantial declines in the abundance of spawning fish because losses from mortality have not been balanced by recruitment of new individuals. Although capture-recapture data indicate substantial recruitment of new individuals into the adult spawning populations for SNS and river spawning LRS in some years, size data do not corroborate these estimates. In fact, fork length data indicate that all populations are largely comprised of fish that were present in the late 1990s and early 2000s. As a result, the status of the endangered sucker populations in Upper Klamath Lake remains worrisome, and the situation is especially dire for shortnose suckers. Future investigations should explore the connections between sucker recruitment and survival and various environmental factors, such as water quality and disease. Our monitoring program provides a robust platform for estimating vital population parameters, evaluating the status of the populations, and assessing the effectiveness of conservation and recovery efforts.