Introduction


Deltaic marshes at the mouth of the Williamson River, Oregon, probably were once among the most important habitats for larvae of the endangered Lost River sucker, Deltistes luxatus, and shortnose sucker, Chasmistes brevirostris, because of their location downstream of known productive spawning grounds. In the 1940s, the delta was diked and drained for agriculture, which resulted in loss of organic matter, soil compaction, and subsidence, particularly in the northern part of the delta. In 1996, The Nature Conservancy purchased the property surrounding the mouth of the Williamson River (fig. 1) and began a large-scale restoration project to reconnect the Williamson River with 2,500 ha of former agricultural land (hereafter referred to as “the delta”). In October 2007, the levees around the northern part of the delta, known as Tulana, were breached, flooding approximately 1,500 hectares (fig. 1). In November 2008, levees around the southern part of the delta, known as Goose Bay, were breached, flooding an additional 1,000 ha. A primary restoration goal was to restore the function of the delta and its wetlands as nursery habitat for the endangered suckers.


Lost River and shortnose suckers are long-lived (as many as 57 yr) lake dwellers that typically migrate into tributaries of Upper Klamath Lake to spawn in the early spring (Scoppettone and Vinyard, 1991; Terwilliger and others, 2010). In Upper Klamath Lake, the present-day population stronghold for both species, a large portion of each population spawns in the Williamson and Sprague Rivers, and a small group of Lost River suckers spawns at lakeshore springs on the eastern edge of the lake (Janney and others, 2008). After leaving the gravel at the spawning grounds, larvae drift downstream at night with river flow and, prior to restoration, entered Upper Klamath Lake in as little as a day (Cooperman and Markle, 2003). Larval sucker habitat includes nearshore and open water areas of Upper Klamath Lake, but proximity to shore seems most important (Reiser and others, 2001; Cooperman and Markle, 2004; Crandall and others, 2008; Burdick and Brown, 2010).


Restoration of the delta was expected to create complex pathways connecting the lower 5 km of the Williamson River channel to the lake and to increase travel time from spawning locations to the lake for many larvae (Markle and others, 2009). Over time, the increase in emergent vegetation associated with wetland restoration at the delta is expected to decrease the flow velocity across the wetland and to provide sanctuary from wind turbulence (Cooperman and others, 2010), ample feeding and growing opportunities (Crandall and others, 2008), and predator protection (Markle and Clausen, 2006; Markle and Dunsmoor, 2007). These effects of wetland restoration potentially contribute to lower early-life mortality, less emigration from the lake, stronger year class formation, and aid the recovery of the species.


Field collections are spatially and temporally limited, and a spatially distributed model can provide information at temporal and spatial scales that are unattainable through field sampling. Models also provide iterative visual feedback, with field data providing guidance about model assumptions and the model simulations providing guidance about appropriate field sampling strategies and, when confidence is established, predictions under different scenarios. Models of marine larval dispersal are numerous and have been applied to connectivity between populations (Hare and others, 2002; Nahas and others, 2003; Paris and others, 2009; Ashford and others, 2010; Watson and others, 2010), adaptive sampling (Voss and Hinrichsen, 2003; Pepin and others, 2009), and recruitment prediction (Reyns and others, 2006; Hinckley and others, 2009; Mariani and others, 2010). Examples of models for freshwater larval dispersal are fewer than for marine larval dispersal (Beletsky and others, 2007). For this study, the relevant concepts developed for marine larval dispersal were applied and adapted to a large, shallow lake with a complex spatial geometry in which the water currents are primarily wind-driven, using target species that are endangered and, therefore, rare. Larval dispersal was simulated to gain insight into how the physical configuration of the delta landscape and environmental conditions affect the transport of a larval cohort from spawning to rearing habitat.


Purpose and Scope


The purpose of this study was to develop and to validate with field data a biophysical model of larval sucker dispersal from spawning grounds through the delta and into the lake. Wood and others (2008) developed a hydrodynamic and heat model for Upper Klamath Lake and Agency Lake (a northern embayment of Upper Klamath Lake), and Wood and others (2012) used an advection-diffusion approach to simulate water currents transporting larval cohorts through the system. The advection-diffusion approach is limited to describing transport of larvae by currents in combination with passive dispersal and, therefore, cannot be used to simulate active dispersal resulting from horizontal swimming.


This report documents an individual-based approach used to simulate the ensemble age of larvae at larval catch sites and, using a length-at-age regression, the estimated lengths were compared to the length distribution of fish captured in nets. The individual-based approach cannot provide concentration information at the spatial and temporal resolution of the advection-diffusion approach, but it has the advantage of allowing the age of individual particles to be tracked through the system, and it allows the spreading of the larvae that is superimposed on the advective transport to be described as active dispersal resulting from horizontal swimming. Thus, this approach is particularly suited to addressing how the ability of the larvae to swim, and to swim in response to certain stimuli, might affect the dispersal of the larvae through the delta. We used the model to understand travel times and pathways through the delta, how those travel times and pathways changed as restoration altered the shoreline configuration in two phases between 2007 and 2009, and how those travel times and pathways changed with assumptions about larval swim behavior. We also used the model to explore how the management of lake elevation at higher or lower stage could alter larval transport. We assessed the model assumptions about dispersal by converting particle ages to fish lengths and comparing the resulting simulated lengths to field data.


First posted January 31, 2014