A geographical framework for assessing longitudinal patterns in stream habitat and fish distribution

Graduation date: 2002New approaches are needed to quantify and understand spatial patterns of stream fishes and their environment. Concepts in riverine ecology emphasize the importance of thermal zones and gradual longitudinal changes in physical habitat and biota, but little is known about spatial variability within the river continuum. I present a conceptual framework for assessing patterns in streams that are difficult to detect using standard site-based approaches to sampling. The ability to detect pattern is defined as the scope, or the ratio of extent (the distance, area, or volume encompassing all sample points) to grain size (the size of an individual sample unit). By increasing the scope and continuity of sampling, I illustrate how greater sampling effort can\ud reveal new patterns and unexpected relationships between stream fishes and their\ud environment. Merging geography and stream ecology, I describe new approaches including remote sensing, multiscale sampling, and extensive surveys for assessing longitudinal patterns in stream habitat and fish distribution. Airborne thermal infrared\ud remote sensing was effective for quantifying spatially continuous patterns of water\ud temperature over a range of scales from channel units (10–50 m) to entire river\ud sections (30–70 km). To examine factors influencing the spatial distribution of larval Pacific lamprey, a benthic fish species, I applied a nested sampling design and determined that stream habitat variables predicted patterns in larval abundance but\ud played different roles at different spatial scales. Increases in the scope of data\ud collection required adaptations in statistical analysis in order to accommodate larger and more complex ecological datasets. I evaluated multivariate ordination techniques with respect to their ability to describe fish community structure and found that nonparametric multivariate smoothing of presence–absence data was highly effective for detecting patterns in heterogeneous fish assemblage data. Spatially continuous\ud analysis presented challenges in extracting patterns from noisy ecological data but\ud provided the opportunity to evaluate distributional patterns over a range of spatial scales. I examined spatial variability of stream fish assemblages and observed that the relative influences of temperature and channel morphology on fish assemblage structure were dependent on the thermal context and the spatial scale of analysis