The Effects of Urbanization on Local First-Order Streams: One Year Later

Kristen K. Cecala, Steven J. Price, William Ringle, and Michael E. Dorcas

Davidson College

To determine the effects of urbanization on first order streams within one year of disturbance, as part of a larger study, we sampled 10 streams directly before and one year following the initiation of urbanization of the stream watershed. Each site was located within either Mecklenburg (Sites 1-2) or Cabarrus Counties (Sites 3-10) of the western Piedmont of North Carolina (Fig. 1). Our study sites were selected from properties in which we knew urbanization would occur within one year of our initial sampling in March and April of 2005. Sampling points were haphazardly chosen along the stream such that they would test points downstream of watershed disturbance.
Literature Cited
Anthropology Department
Davidson College Herpetology Laboratory

Yellow stars indicate sampling locations in this inset map.


Sampling was conducted in March and April of 2005 before disturbance and in March and April of 2006 following watershed disturbance. Each stream was sampled twice per year, once in March and a second time in April. As part of a larger study, sampling events included sampling two separate points of a stream approximately 10 m apart before returning to one point 3 subsequent times within a week. At each sampling event, we recorded dissolved oxygen and conductivity corrected for temperature using a YSI Model 85 Probe (YSI Inc., Yellow Springs, OH) as well as measuring maximal width and depth at three controlled sites within a 10 m reach. Values for dissolved oxygen and conductivity were averaged for both sampling events per year. All three measures of maximal width and depth were averaged for all sampling events per year. Secondly, we also characterized the stream substrate once per sampling event in which we approximated the proportion of sand, gravel, pebble, cobble, boulder, bedrock, silt, detritus, clay, and other such as artificial substrate (e.g. riprap or culverts) within the same 10 m transect as used above. For analyses, we chose to examine differences in the proportion of sand, detritus, and the combined category of fine sediments (clay, silt, muck)..

To determine the proportion of the watershed urbanized within one year of sampling, we used the program ArcGIS 9.2 (ESRI, Redlands, CA) for all analyses. First, we used elevation data (NC State DOT) for Mecklenburg and Cabarrus Counties to develop catchments for each first-order stream sampled. Catchments were derived by using the extension ArcHydro (ESRI, Redlands, CA). To develop these catchments, we proceeded to first find the flow direction and flow accumulation based on the elevation. Next, stream segmentation and definition was defined using information described by flow direction and flow accumulation. From this information, ArcHydro calculated the catchment grid and delineated the catchment polygon. Lastly, by calculating the drainage line, we were able to develop the stream layer for our sampled first-order stream.


See ArcGIS methods for more detail on this process.

Completed catchment and development delineation of site # 4. The pale yellow indicates the watershed for site # 4.

Site # 4
Flow Direction of Site # 4
Flow Accumulation of Site # 4
Stream Link of Site # 4
Catchments for Site # 4

To develop the area of these watersheds developed before our 2006 sampling, we used either development plans confirmed by groundtruthing (Sites 4-7, 10) as well as actual groundtruthing and manual collection of disturbed area for the remaining sites (Sites 1-3, 8, 9). Once delineations of the stream’s watershed and the area of urbanization were calculated, we were able to determine the proportion of the watershed disturbed within one year. Furthermore, we were able to measure the minimal and maximal distance of disturbance to the stream to better determine the impact of stream buffers on first-order streams.

We analyzed our data using several linear regressions to determine the relationship between the proportions of urbanized watershed to the change in our various stream characteristic values (Minitab v.12.1). Proportional data was transformed using an arcsine transformation to prevent our data from being baised by the mean of the data (Sokal and Rohlf, 1981). Because this is an exploratory analysis and landscapes are highly variable, significant results were evaluated at an α =0.10.

Questions? Email: