Summary 


Endangered suckers endemic to Upper Klamath Lake spawn in the Williamson River and its tributary, the Sprague River, and the larvae drift downstream to Upper Klamath Lake. A major recovery effort for endangered suckers in Upper Klamath Lake has been predicated on the assumption that recovery of deltaic marshes at the mouth of the Williamson River would help retain larvae in the lake and provide important nursery habitat. We used the UnTRIM two-dimensional finite difference hydrodynamic model running on an unstructured grid to investigate the effects of deltaic modifications that occurred between 2007 and 2009 on the distribution of larval suckers. We compared model simulation results to larval catches for years before and after old levees were breached in the Williamson River Delta to validate the model assumptions. We constructed advection-diffusion based density simulations to describe the effects of deltaic modifications on larval transport, under the assumption that the larvae drift passively in the currents. These simulations were compared to larval catches at the mouth of the Williamson River and along the shorelines in Upper Klamath and Agency Lakes and the Williamson River Delta.


Model simulation results showed that the density of passively transported larvae through the Williamson River Delta would be expected to decrease along shorelines in Upper Klamath Lake after the flooding of restored wetlands. Larvae that previously entered the lake at one location—the mouth of the Williamson River—will now enter Agency and Upper Klamath Lakes at multiple locations through breaches in the levees and over submerged levees that surround the Williamson River Delta. The model simulation results indicated that prevailing winds would cause larval aggregations to be advected to a larger extent through Goose Bay than through Tulana.


Results from density simulations were compared to field data using rank correlation. Correlation coefficients were almost uniformly positive and often significant, but most were about 0.30–0.60, suggesting that although the model predicts the general pattern of distribution, usually less than one-half of the variation in site rank densities was explained. When the density simulations were correlated with fish catches grouped into different size classes, the correlation coefficients were higher for small larvae near the larval source in the river and for larger larvae farthest from the source. The correlations were generally higher in years and with gear types in which shortnose/Klamath largescale larvae dominated, indicating that the assumptions used in the model better described the transport of shortnose/Klamath largescale than Lost River suckers. 


We also used modeling to understand the implications of nighttime-only drift behavior on larval sucker dispersal. We incorporated nighttime-only drift by “freezing” the density in the river channel during the day. Variation in the duration of nighttime drift (defined as the distance down the river channel the behavior persists) had no noticeable effect on correlations between the simulated densities and the larval catches. 


Our field data provided moderate levels of corroboration for the model and showed that current patterns through the modified landscape at the Williamson River Delta are an important component of larval sucker dispersal. Better field data for model validation will be difficult to obtain, but at a minimum require information on gear size efficiency, more sampling, and different sampling strategies.


First posted April 2, 2012