Modeling Transport of Larval Suckers Through A Restored Delta in Upper Klamath Lake, Oregon, Using Density-Based and Individual-Based Approaches

Part of a recovery effort for endangered suckers in Upper Klamath Lake has been restoration of deltaic marshes at the mouth of its primary tributary, the Williamson River. In this system, an important function of shoreline nursery habitat is to slow larval transport out of the lake. Marsh restoration should help retain larvae in the lake and reduce this loss to the population. We used a hydrodynamic and larval transport model to generate density simulations and individual-based simulations to compare with larval catches in and around the delta, before and after delta restoration. The hydrodynamic model included physical simulations with passive particles and biophysical simulations with particle behaviors of night-time-only drift, positive or negative rheotaxis or random swimming, and mortality. Prior to restoration, larvae left the river mouth and were usually entrained in a southward flowing eastern boundary current. After restoration, larvae still exited through the river mouth, but also tended to move through a marsh complex around Goose Bay before entrainment in the boundary current. Rank correlation coefficients from density simulations and field data were almost all uniformly positive and often significant, but most were about 0.30 – 0.70, suggesting that the model predicts the general pattern of distribution, but usually explains less than half the variation in site rank densities. Sized-based density simulations correctly replicated an increase in median particle age consistent with north to south transport along the eastern shoreline of Upper Klamath Lake. Our individual-based approach compared the length of passive particles (based on a length-at-age linear regression) to the length in larval catches. Three types of swimming behavior and passive dispersion were tested. The highest correlation (R=0.68) was obtained with night-time-only drift, mortality, and random swimming, and was an improvement over the model of passive particles (R= 0.47). Unlike most biophysical models of larval drift, our validation was not based on an abundant species. Instead, the two endangered species comprised only 6.3% of our catch. Despite this limitation, our field data provided surprisingly good levels of corroboration for both the density and the individual based simulations and showed that currents are an important component of larval sucker dispersal and that shoreline configuration is an important component of larval sucker retention. Further progress on this biophysical model will require understanding cues responsible for volitional swimming and the biological factors responsible for retention at nursery sites.