Stream ecosystems are holistic systems that incorporate disturbances and abiotic influences at many spatial and temporal scales. This view supports a three-tiered model of variables that determine biotic integrity in streams, with causes and effects flowing from large-scale to fine-scale processes. Tier One characteristics include variables important at the scale of geomorphological processes and land use over entire watersheds. These variables largely determine Tier Two factors, abiotic conditions in a stream reach. Tier Two variables, in turn, largely structure the Tier Three variables, the stream's biotic communities. Through field studies and GIS analysis, relationships among these three tiers of variables were examined in this research to explore the question of how agriculture exerts its influence on stream fishes. This study investigated 27 streams, in two ecoregions and the transition area, or ecotone, between them, in south-central Ohio. The study design allowed questions to be asked concerning the relative influence of geomorphology and land use in varied landscapes, as well as relative impacts of watershed versus riparian land use. The region also contained relatively equal proportions of three types of agriculture (hay, row crops, and pasture) allowing the study to address the question of which land use might be most harmful to stream fish. This study supported the importance of row crop agriculture, finding it to be the most degrading type of agriculture for stream fish, but also found pasture to be an important causal factor in stream community degradation. This study also supported the importance of riparian buffers, finding riparian agriculture to be more degrading than agriculture over the entire watershed. A more interesting finding is the suggestion that a minor amount of nutrient enrichment from agricultural land use may benefit streams that are naturally oligotrophic. A possible mechanism could be increased primary production, which increases macroinvertebrate density, and provides a larger food base for fishes. This study also reports the possible existence of a biodiversity "hotspot" in the transitional region between the two ecoregions. Some evidence exists that greater habitat heterogeneity increases species richness, suggesting a possible cause for higher biodiversity in this ecotonal region. Since habitat heterogeneity over whole streams was not measured in this study, both the existence and mechanism of such a hotspot needs more study. A final conclusion is that geomorphology and agricultural land use may be equally important in structuring stream conditions, and thus, biological stream communities. This study illustrates the difficulties associated with overlapping causes and effects in complex systems such as streams and their catchments. Several variables in the study reported here required examination at multiple scales and with multiple statistical techniques in order to understand relationships that varied across different regions. The effects of a particular agricultural variable were not always equal in the diverse landscapes of southern Ohio. Lotic ecologists must examine a variety of ecoregions, and incorporate a variety of scales with a variety of analytic tools, if predictive stream ecology is to become a reality.