Print

Cianfrani, Christina M.

Civil and Environmental Engineering

2005

PhD

Stream reaches from two geographic regions were studied to determine how aquatic ecosystem health is affected by local-scale and watershed-scale variables. In Part 1 of the study, 54 stream reaches in southeastern Pennsylvania and northern Maryland were used to assess the relationship between stream field-measured channel characteristics and watershed landuse, specifically watershed impervious area. Principal components analysis (PCA) was used to identify the most important variables (vegetation type, bankfull width/depth ratio, and large woody debris) in separating stream reaches among nineteen typically collected geomorphic characteristics. In Part 2, twenty-five independent stream reaches in northwestern Vermont were surveyed to assess the influences of local-scale and watershed-scale variables on stream biota (fish, birds, and macroinvertebrates) diversity and condition. Data was collected during the summers of 2003 and 2004 at multiple spatial scales. Local-scale data included quantitative and qualitative geomorphic and habitat surveys. Watershed-scale data included landuse characterization and modeled average annual flow and sediment loading. Diversity and condition for each taxon were assessed using common measures including: 1) fish species richness, Shannon-Weaver Index, Simpson's Index, Vermont Mixed Waters Index of Biotic Integrity, and biomass; 2) macroinvertebrates - %EPT, %Chironomids, and density; and 3) birds - river corridor and piscivore abundance, and species richness. First, streams were classified by channel type to determine whether fish community diversity was higher in pool-riffle or ecotonal/transitional zones. For all three fish community diversity measures, diversity was higher in the transitional zone stream reaches (p £0.05). These results suggest that transitional zones are important for maintaining fish diversity and, therefore, must be included in stream classification systems, especially those used for ecological applications. Second, PCA (with varimax rotation) was used with the Vermont data to generate sets of factors for combinations of local-scale and watershed-scale characteristics. The factors were then used as the independent variables in multiple regression models using the biological data as the dependent variables. Forty significant models were built using the eleven biological community variables and eight sets of local/watershed variables. Fish community diversity and condition were best predicted when using a combination of local-scale and watershed-scale data. However, based on the factor loadings, the qualitative local data were more useful than the quantitative data. In significant macroinvertebrate models, local-scale data and sediment from modeling were important factors. Bird abundance and richness were best predicted using local geomorphic characteristics and the qualitative geomorphic and habitat assessments. These results reinforce the concept that, while both local-scale and watershed-scale variables impact stream biota to some extent, their relative influence depends upon the individual ecology of each taxon. In order to address these issues, comprehensive watershed management and restoration/protection plans should include assessment at multiple scales from an in-stream geomorphological, watershed, and ecological perspective.