SUMMARY 1. We examined the relationship between catchment land cover, sediment regime and fish assemblage structure in four small streams in the upper Little Tennessee River basin of North Carolina. Study streams drained similar sized catchments (17–31 km2) with different fractions of non‐forested land cover. Non‐forested land cover was <3% in two ‘reference’ streams, whereas it was 13 and 22% in two ‘disturbed’ streams. Land cover data were compared with sediment transport data (suspended and bedload), benthic habitat data (embeddedness, substratum composition and coverage of fines) and fishes collected in autumn 1997. 2. Suspended sediment concentration was significantly higher in disturbed streams during both baseflow and stormflow. During baseflow disturbed streams nearly always exceeded 10 nephelometric turbidity units (NTU), whereas reference streams never exceeded this threshold. The difference in suspended sediment concentration between reference and disturbed streams was more consistent at baseflow than at stormflow. Therefore, baseflow turbidity may be a useful indicator of potential stream degradation. 3. Disturbed sites had five‐ to nine‐fold more bedload transport than reference sites. Both embeddedness and streambed instability increased with increasing non‐forested land cover. 4. Relative abundance of fishes requiring clean cobble/gravel substratum for spawning was lower in disturbed streams, whereas relative abundance of mound‐building cyprinids, their nest associates and fishes that excavate nests in soft sediments (centrarchids) was higher. Relative abundance of fishes spawning in benthic crevices and gravel (BC + G) declined as the proportion of non‐forested land cover increased. This study supports growing evidence that human‐induced sedimentation alters stream fish assemblages.

Adult aquatic insects emerging from streams may be a significant source of energy for terrestrial predators inhabiting riparian zones. In this study, we use natural abundance delta(13)C and delta(15)N values and an isotopic (15)N tracer addition to quantify the flow of carbon and nitrogen from aquatic to terrestrial food webs via emerging aquatic insects. We continuously dripped labeled (15)N-NH(4) for 6 weeks into Sycamore Creek, a Sonoran desert stream in the Tonto National Forest (central Arizona) and traced the flow of tracer (15)N from the stream into spiders living in the riparian zone. After correcting for natural abundance delta(15)N, we used isotopic mixing models to calculate the proportion of (15)N from emerging aquatic insects incorporated into spider biomass. Natural abundance delta(13)C values indicate that orb-web weaving spiders inhabiting riparian vegetation along the stream channel obtain almost 100% of their carbon from instream sources, whereas ground-dwelling hunting spiders obtain on average 68% of their carbon from instream sources. During the 6-week period of the (15)N tracer addition, orb-web weaving spiders obtained on average 39% of their nitrogen from emerging aquatic insects, whereas spider species hunting on the ground obtained on average 25% of their nitrogen from emerging aquatic insects. To determine if stream subsidies might be influencing the spatial distribution of terrestrial predators, we measured the biomass, abundance and diversity of spiders along a gradient from the active stream channel to a distance of 50 m into the upland using pitfall traps and timed sweep net samples. Spider abundance, biomass and richness were highest within the active stream channel but decreased more than three-fold 25 m from the wetted stream margin. Changes in structural complexity of vegetation, ground cover or terrestrial prey abundance could not account for patterns in spider distributions, however nutrient and energy subsidies from the stream could explain elevated spider numbers and richness within the active stream channel and riparian zone of Sycamore Creek.

A proactive sampling strategy was designed and implemented in 2000 to document changes in streams whose catchment land uses were predicted to change over the next two decades due to increased building density. Diatoms, macroinvertebrates, fishes, suspended sediment, dissolved solids, and bed composition were measured at two reference sites and six sites where a socioeconomic model suggested new building construction would influence stream ecosystems in the future; we label these "hazard sites." The six hazard sites were located in catchments with forested and agricultural land use histories. Diatoms were species-poor at reference sites, where riparian forest cover was significantly higher than all other sites. Cluster analysis, Wishart's distance function, non-metric multidimensional scaling, indicator species analysis, and t-tests show that macroinvertebrate assemblages, fish assemblages, in situ physical measures, and catchment land use and land cover were different between streams whose catchments were mostly forested, relative to those with agricultural land use histories and varying levels of current and predicted development. Comparing initial results with other regional studies, we predict homogenization of fauna with increased nutrient inputs and sediment associated with agricultural sites where more intense building activities are occurring. Based on statistical separability of sampled sites, catchment classes were identified and mapped throughout an 8,600 km(2) region in western North Carolina's Blue Ridge physiographic province. The classification is a generalized representation of two ongoing trajectories of land use change that we suggest will support streams with diverging biota and physical conditions over the next two decades.