Premium
Reconciling landscape and local views of aquatic communities: lessons from Michigan trout streams
Author(s) -
WILEY MICHAEL,
KOHLER STEVEN,
SEELBACH PAUL
Publication year - 1997
Publication title -
freshwater biology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.297
H-Index - 156
eISSN - 1365-2427
pISSN - 0046-5070
DOI - 10.1046/j.1365-2427.1997.00152.x
Subject(s) - streams , trout , variance (accounting) , sampling (signal processing) , ecology , temporal scales , population , scale (ratio) , spatial ecology , sample (material) , sampling design , geography , environmental resource management , environmental science , computer science , cartography , fish <actinopterygii> , fishery , biology , computer network , chemistry , demography , accounting , filter (signal processing) , chromatography , sociology , business , computer vision
1. Rapidly advancing geographical information systems (GIS) technologies are forcing a careful evaluation of the roles and biases of landscape and traditional site‐based perspectives on assessments of aquatic communities. Viewing the world at very different scales can lead to seeming contradictions about the nature of specific ecological systems. In the case of Michigan trout streams, landscape analyses suggest a predictable community shaped by large‐scale patterns in hydrology and geology. Most site‐based studies, however, suggest these communities are highly variable in structure over time, and are strongly shaped by site‐specific physical and biological dynamics. As the real world is comprised of processes operating both at local and landscape scales, an analytical framework for integrating these paradigms is desirable. 2. Decomposition of variances by factorial ANOVA into time, space and time–space interaction terms can provide a conceptual and analytical model for integrating processes operating at landscape and local scales. Using this approach, long‐term data sets were examined for three insects and two fishes common in Michigan trout streams. Each taxon had a unique variance structure, and the observed variance structure was highly dependent upon sample size. 3. Both spatially extensive designs with little sampling over time (typical of many GIS studies) and temporally extensive designs with little or no spatial sampling (typical of population and community studies), are biased in terms of their view of the relative importance of local and landscape factors. The necessary, but in many cases costly, solution is to develop and analyse data sets that are both spatially and temporally extensive.