7853 - Laserfiche WebLink

Home Browse Search

546 GREGOR T. AUBLE ET AL. Ecological Applications Vol. 4, No.3 Middle bar river right Middle bar river left Upstream bar river right ----- . l\t\.i'iel G\ll\l\\sO ~ Upstream bar river left Hydraulic cross section 9 150m I ~ FIG. 2. Location of sampled areas within study reach. gradient of inundation duration. Changes in vegetation distribution were then estimated by determining how a new flow regime would alter inundation durations. Franz and Bazzaz (1977) used a similar approach to assess the hydrologically simpler case of impoundment impacts upstream of a dam. We censused randomly located plots, assigned each plot to one of three vegetative cover types, and deter- mined the inundation duration for each plot based on the discharge required to inundate the plot and the historical flow duration curve. The inundation-dura- tion gradient was broken into 12 regularly spaced class- es. This number minimized the range of individual classes while ensuring that each class contained at least one plot. The 12 classes and the number of plots per class were: 0-0.01, 18 plots; 0.01-0.1, 42 plots; 0.1- 0.2, 11 plots; 0.2-0.3, 3 plots; 0.3-0.4, 6 plots; 0.4- 0.5, 10 plots; 0.5-0.6, 16 plots; 0.6-0.7, 16 plots; 0.7- 0.8, 5 plots; 0.8-0.9, 5 plots; 0.9-0.99, 1 plot; 0.99- 1.0,0 plots. The 12th class in this list was not repre- sented in the set of sampled plots, but was added in order to support prediction about plots that might be- come permanently inundated under an alternative hy- drologic regime. For each class we calculated the pro- portion of plots in each cover type. We then used these proportions as probabilities to estimate the future cov- er type of a plot from its future inundation duration. Flow duration curves describing different hydrologic alternatives were used to produce plot inundation du- rations associated with each alternative. Applying the cover-type probabilities to plots with new inundation durations produced an estimate of the expected value of the new number of plots in each cover type: nij = Pij. ~ where nij = number of plots in cover type i and in- undation duration class j, Pi,} = probability that a plot in inundation duration class j is in cover type i, ~ = total number of plots in inundation duration class j. Application of the model to the plots sampled for calibration thus produces the expected number of plots in each cover type for each alternative hydrologic re- gime. We interpret changes in the proportions of ran- domly located plots as changes in the cover type com- position of the study area. Ancillary model output includes fractions of time inundated and cover-type probabilities for each plot as well as transition matrices depicting aggregate and plot-by-plot change in inun- dation durations and cover-type probabilities. We implemented the model as a series of individual modules, including the HEC-2 step-backwater model for water surface elevations (Hydrologic Engineering Center 1990), the TWINSP AN clustering program for identifying cover types (Hill 1979a), and a series of short programs that we wrote in FORTRAN and SAS (SAS Institute 1987) to interpolate water surface ele- vations, calculate flow duration curves, calibrate and apply cover-type probabilities, and summarize output across plots. Plot sampling We selected and sampled the plots on 18-31 July 1990. Vegetation sampling was restricted to the five relatively flat areas of alluvial sediment, or bars, within the study reach (Fig. 2). Adjacent to the stream, bars were delimited by the water's edge. This line also marked the streamward limit of emergent vegetation. Away from the stream, bars were bounded by the base ofa cliff or the toe ofa talus slope. The narrow, uneven areas of alluvial and colluvial sediment between bars were not sampled. We randomly located 133 rectangular I x 2 m plots on the bars; the long side of each rectangle was placed parallel to the direction of flow. This plot size was big enough to represent the largely herbaceous vegetation and small enough that inundation duration was uni- form throughout the plot. The center of each plot was located by reading x and y coordinates from a table of random numbers and pacing the coordinates in the field. The National Park Service prepared a topograph- ic map of the study reach and determined the elevation and location of all the plots using a total station sur- veying instrument. The density of plots was intended to be the same for all bars; however, the total-station survey, carried out after the plots were established, indicated that plot densities on the five bars varied