Master of Science / Department of Biology / Craig Paukert / We investigated spatial variation in fish assemblage and food web structure in the Kansas River, USA in relation to habitat changes. Fishes were collected at ten sites throughout the Kansas River for assessing assemblage structure in summer 2007 using fish community metrics and at 3 sites in 2006 for food web structure using stable isotope analysis. Satellite imagery indicated riparian habitat on the Kansas River was dominated by agriculture in the upper reaches (>35%) and tended to increase in urban land use in the lower reaches (>58%). Instream habitat complexity also decreased with increased urban area (<25%) becoming more channelized. Jaccard's similarity and percent similarity indices suggested that large-bodied fishes show changes in species presence and composition longitudinally within the river. Also, reaches directly above Bowersock Dam in Lawrence, Kansas and below the Johnson County Weir, near Kansas City, Kansas had low percent similarity compared to other reaches, suggesting the dam and the weir affect community composition. Canonical correspondence analysis indicated that species that prefer high velocity flows and sandy substrate (blue sucker and shovelnose sturgeon) are associated with the upper river reaches. Also, there was a higher abundance of omnivorous and planktivorous fish species in the lower more channelized river. The lower reaches contain more tolerant, macrohabitat generalist species and the upper river contained more intolerant, fluvial specialist species. Fish, macroinvertebrates, and detritus were collected at three river reaches classified as the heterogeneous instream habitat (>40% grass islands and sand bars) intermediate (22% grass islands and sand bars), and homogeneous (6% grass islands and sand bars) instream habitat reaches in June 2006. Riparian land use (proportion as agricultural and urban) was related to instream habitat with homogeneous areas having more urban riparian area compared to the heterogeneous and intermediate reaches. The heterogeneous habitat reach had higher variability in [Delta][superscript]13C for fish classified as piscivores/invertivores (P=0.029) and macroinvertebrates (P=0.004) suggesting the complex habitat in the heterogeneous habitat reach provided more variable food sources. The [Delta}15N values also indicated that ten of the twelve fish species tended to consume prey at higher trophic levels in the heterogeneous habitat reach suggesting a more complex food web. Land use practices are leading to homogenization of instream habitat and this homogenization of habitats may be related to food web diversity and trophic position of fishes. Conserving intolerant, native species in the Kansas River may require maintaining suitable habitat for these species and restoration of impacted areas of the river.

The Staunton River is a high gradient, second order stream approximately 6 km in length located on the eastern slope of the Blue Ridge Mountains in Shenandoah National Park, VA. In June 1995, a catastrophic flood and debris flow altered the instream habitat and <i>Salvelinus fontinalis</i> population of the Staunton River. The debris flow scoured the streambed, deposited new substrate materials, removed trees from the riparian zone, and eliminated fish from a 1.9km section of the stream. By June 1998, both young-of-year (YOY) and age 1+ <i>S. fontinalis</i> had recolonized the debris flow affected area. The event provided a rare opportunity to examine recovery of the <i>S. fontinalis</i> population and instream habitat in addition to addressing potential effects of the debris flow on movement, activity, and growth of fish in the debris flow affected and unaffected areas of the stream. Post-recolonization movement and activity were monitored using two-way fish traps (weirs), mark-recapture techniques, and radio telemetry. The weirs failed to produce any movement data. Most fish (91%) in the mark-recapture study had range sizes less than 100m, however biases common to mark-recapture study designs (low recapture rate, flawed logic, etc.) hampered interpretation of results. For example, subsequent recapture of individually marked fish indicated that as many as 54% of marked fish confirmed to have been alive at the time of a recapture session were not recaptured. Radio telemetry provided information on <i>S. fontinalis</i> movement and activity at seasonal and diel scales during summer and fall. Differences in movement and activity between the debris flow affected and unaffected areas were minimal when compared to seasonal variations. During summer, range sizes were near 0m and crepuscular activity patterns were observed. During the fall range size increased and diel activity was concentrated in the mid-afternoon with a much higher peak than during summer. Basin-wide visual estimation technique (BVET) fish population surveys performed each spring and fall from 1993 = 1999 provided pre- and post-event fish population abundance and density estimates. Post-event fish growth in the debris flow affected and unaffected areas was monitored using mark-recapture techniques. Abundance and density of both YOY and age 1+ <i>S. fontinalis</i> exceeded pre-event levels within 2-3 years. Growth of YOY and age 1+ fish was significantly greater in the debris flow affected area until spring 1999. Population density appeared to have a strong negative influence on growth. The observed changes in fish growth and differences in fish size associated with population density would be of minimal importance to the typical angler but may suggest a mechanism by which <i>S. fontinalis</i> populations can quickly recover from catastrophic events. BVET habitat surveys provided information on total stream area, number of pools and riffles, pool and riffle surface area and depth, substrate composition, and large woody debris (LWD) before (1993), immediately following (1995), and four years post-event (1999). Immediately following the debris flow, the stream channel was highly disordered which resulted in an increase in the total number of habitat units and a decrease in average habitat unit surface area, total stream area, and average depth when compared with pre-event conditions. In addition, substrate composition had shifted from small to large diameter particles and LWD loading had increased in both debris flow affected and unaffected areas. Four years after the event, the total number of habitat units, average habitat unit surface area, total stream area, and average depth had all returned to near pre-debris flow levels and substrate composition had begun to shift towards smaller particle sizes. Changes in LWD loading from 1995-1999 reflected changes in the riparian zone following the debris flow. In the unaffected area, where riparian trees remained intact, LWD loading increased, whereas in the debris flow affected area, where riparian trees were eliminated, LWD loading decreased. For the most part the effects of the debris flow, although immediately dramatic, were in the long term minimal. The debris flow affected area was recolonized rapidly and abundance and density quickly rebounded past pre-event levels. Differences in fish growth between the affected and unaffected area were short lived. Any effect the debris flow affected area may have had on movement or activity was minimal when compared with seasonal variations. Most habitat characteristics reverted to near pre-event levels just four years after the flood and debris flow. Although a number of factors will influence recovery time from such events, these results indicate that immediate management action, such as stocking or habitat modifications, are not necessary in all cases. / Master of Science