Cases and solutions
Sedimentation in bottomland hardwoods downstream of an east Texas dam J.D. Phillips
practice of assigning a causal role to dams in the erosion of Gulf Coast beaches. There is an intuitive logic to this argument, based on the known propensity of dams and reservoirs to trap sediment, and an apparent coincidence of accelerated coastal erosion with an era of dam building (Morton 1977; Morton and Paine 1990). A drastic reduction in sediment loads immediately downstream of dams is apparently common and is not disputed. It is also clear that in some cases sediment trapping by impoundments can be directly linked to erosion and other sediment starvation effects well downstream (e.g., Stanley 1996). However, the assumption that the sediment starvation outcome is ubiquitous, and that the effect extends for signi®cant distances downstream in Texas has received little critical analysis and is supported by very little empirical data. Given that the general downstream geomorphic effects of dams have been shown to be highly variable and contingent on local conditions (Williams and Wolman 1984; Friedman and others 1998), it is clear that generalizations cannot be widely and uncritically applied. The purpose of this study is to determine whether an east Texas dam and reservoir are signi®cantly reducing sedimentation in ¯oodplain bottomland hardwood forests Keywords Dam á Impoundment á Fluvial downstream. sediments á Alluvial sedimentation á Fluvial The impoundment-beach erosion connection has been geomorphology accepted, essentially without question, by a number of sources, including a handbook on management of Texas barrier islands (Morton and others 1983) and a number of scienti®c studies (e.g., Giardino and others 1995; Davis 1997). The notion is widely accepted by the public. Introduction Whereas scientists and coastal managers view impoundIt is commonly believed in Texas that dams and reservoirs ments as one of several potential contributions to beach erosion, some public of®cials are apparently willing to put trap ¯uvial sediments and thereby starve downstream areas of sediment. This view is evident in the widespread all the blame on dams. The Texas General Land Of®ce Deputy Commissioner told a group of Galveston Island property owners in February 1999 that the damming of rivers is to blame for long-term erosion problems (Galveston Daily News, 21 February 1999). Received: 14 June 2000 / Accepted: 13 September 2000 Despite the logic in the impoundment-erosion linkage and Published online: 24 February 2001 some supporting data (see below), there are reasons to ã Springer-Verlag 2001 question the extent to which ¯uvial sediment supplies are reduced more than a few kilometers downstream of dams. J.D. Phillips The downstream geomorphic effects of dams vary conTobacco Road Research Team, Department of Geography, siderably in both space and time (Williams and Wolman University of Kentucky, Lexington, KY 40506, USA 1984; Friedman and others 1998). In some cases sediment E-mail:
[email protected] loads recover rapidly downstream due to tributary inputs Tel.: +1-859-2576950 and/or channel scour. In some situations where there is a Fax: +1-859-3231969 Abstract Dams and reservoirs are often ef®cient sediment traps, and conventional wisdom holds that ¯uvial sediment supplies are reduced well downstream. However, there are reasons to question the extent to which ¯uvial and alluvial sediment supplies are reduced more than a few kilometers downstream of dams. Sedimentation in bottomlands of Loco Bayou, east Texas, was investigated at a site less than 16 km downstream of Loco Dam and Lake Nacogdoches, which controls 86% of the 265-km2 drainage area. Turbidity levels are generally as high or higher than those on Loco Bayou upstream of the lake. Sedimentation rates on the lower ¯oodplain since the dam was completed are 11 mm year±1 or more. This rate is high enough to suggest that the dam has no effect on sediment supplies 16 km downstream. The spatial pattern of sedimentation and the vegetation distribution suggest that the elevation and frequency of ¯ooding, not ¯uvial sediment availability, are the critical factors in determining sediment supplies to these ¯oodplains.
860
Environmental Geology (2001) 40:860±868
DOI. 10.1007/s002540100246
Cases and solutions
substantial amount of unregulated drainage area downstream of dams, it has been found that the effects of dams on sediment supply at the river mouth are undetectable (Phillips 1992a, 1992b). More generally, there is often an essential decoupling between upstream and downstream sediment dynamics in larger basins (Phillips 1992b; Beach 1994; Brizga and Finlayson 1994; Olive and others 1994; Phillips 1995; Fryirs and Brierly 1999). Such decoupling suggests that the effects of dams, however dramatic they may be immediately downstream, may be minimal or undetectable far downstream. Because it is known that sediment starvation is common immediately downstream of dams, and that there may be upper/lower basin decoupling in large basins, the downstream extent of reduced sediment supplies is an issue for geomorphology, water resource management, and river and ¯oodplain ecology.
Dam effects in east Texas As of 1994 there were 205 reservoirs in Texas with capacities >6,115,000 m3. The sediment-trapping ef®ciency of impoundments may be quite high (up to 99%; Williams and Wolman 1984), and a reduction in sediment concentrations and loads downstream of several Texas impoundments has been documented (Solis and others 1994). Double-mass curves plotting cumulative sediment loads (y-axis) against cumulative discharge (x-axis) were constructed for downstream gauging stations on nine Texas rivers by Solis and others (1994). A break in slope, indicating a change in sediment regimes toward lower sediment loads, was found for the Trinity, Nueces, and Lavaca Rivers. However, the Lavaca River is not impounded, measurements on the Nueces are from a station immediately downstream of a dam, and the six other stations show no clear evidence of a change in sediment delivery. Hudson and Mossa (1997) examined sediment discharge records for the lowermost stations on four large rivers draining to the Gulf of Mexico, including the Rio Grande and Brazos in central and west Texas. Sediment yield records are generally inadequate to directly determine whether a decline in sediment transport is associated with impoundment, due to a paucity of pre-impoundment data. Dams do not appear to have reduced mean discharges for the Brazos and Rio Grande, but Hudson and Mossa (1997) state that peak discharges, which are inordinately important in sediment transport, have likely been reduced. Annual maximum stream ¯ows for three large east Texas rivers, the Sabine, Neches, and Trinity (Fig. 1), were examined for evidence of changes in peak ¯ow following dam construction. Published US geological survey data for annual maximum discharges were examined for the lowermost gauging station for which a signi®cant pre- and post-dam record was available. These periods of record begin in 1908 to 1924 and go through 1998. The Sabine River gauging station near Ruliff, Texas (USGS no. 08030500) has a drainage area of 24,162 km2. The nearest and largest upstream reservoir is Toledo Bend, which was completed in 1967 and which has a drainage area
Fig. 1 The study area in east Texas, USA. The general location of the Loco Bayou study area is shown, along with the location of three gauging stations discussed in the text. Lighter shading Urbanized areas. Darker shading Large water bodies
of 18,591 km2 (about 77% of the area at Ruliff). There is no evidence of any decline in peak ¯ows following the damming of the river (Fig. 2). The Neches River, by contrast, does show clear evidence of decreased peak ¯ows following impoundment of the Sam Rayburn reservoir in 1965 (Fig. 3). The lowermost Neches station is at Evadale, Texas,
Fig. 2 Annual peak ¯ows on the Sabine River near Ruliff, Texas. Arrow Year of completion of Toledo Bend Reservoir (Source: US Geological Survey data for station 08030500; htttp://tx.water.usgs.gov)
Environmental Geology (2001) 40:860±868
861
Cases and solutions
Fig. 4 Fig. 3 Annual peak ¯ows on the Trinity River near Romayor, Texas. Arrow Annual peak ¯ows on the Neches River near Evadale, Texas. Arrow Year of completion of Lake Livingston (Source: US Geological Survey Year of completion of Sam Rayburn Reservoir (Source: US Geological data for station 08066500; htttp://tx.water.usgs.gov) Survey data for station 08041000; htttp://tx.water.usgs.gov)
(USGS no. 08041000) with a drainage area of 20,593 km2. About 95% of this area lies upstream of the Rayburn dam. Annual peaks at the Trinity River station at Romayor, Texas (USGS no. 08066500) show no evidence of reduction in peak ¯ows following damming (Fig. 4). The lowermost dam creates Lake Livingston, ®nished in 1969, with an upstream drainage area of 42,950 km2, which is almost 95% of the 45,242-km2 drainage area of the Romayor station. Published data are therefore equivocal about the downstream effects of dams, even when only a small proportion of the total drainage area lies downstream of the dams. The data at least question the assumption that there are large declines in downstream sediment delivery following impoundment.
Study area and methods The study area is Bayou Loco and Lake Nacogdoches in Nacogdoches County, Texas (Figs. 1, 5). Bayou Loco is a tributary of the Angelina River in east Texas. The total drainage area upstream of the lowermost study site is about 265 km2, with about 228 km2 lying upstream of Loco Dam. The study area lies within the eastern Pineywoods region of the Texas coastal plain. The climate is humid subtropical, with a mean annual precipitation of about 1,200 mm. Although the precipitation is reasonably well distributed throughout the year, summer droughts are not uncommon and August is the driest month. Stream ¯ow is typically highest from December through February, and lowest in mid to late summer. Lake Nacogdoches provides water for the city of Nacogdoches. It has an area of 896 ha, and a volume of 48,224,087 m3 (29,523 acre-ft) at the normal pool elevation of 85 m as of 1994 (TWDB 1994). Loco Dam was completed in 1976. Three sites were examined. The primary study location, called the lower site, is 15.62 km downstream of the dam 862
Environmental Geology (2001) 40:860±868
Fig. 5 The Loco Bayou drainage basin (see Fig. 1 for regional context). Sites 1 and 2 are just upstream of the lake and downstream of the dam, and were examined for comparison with site 3, the primary study site for this investigation
and 16 km upstream of the con¯uence with the Angelina River. Drainage area of the lower site is 264.7 km2, 37.06 km2 of which lies downstream of Loco Dam. Addi-
Cases and solutions
tional sites were examined a short distance upstream of the lake and downstream of the dam (Fig. 5). The sites were instrumented with OBS (Optical Backscattering Sensor) nephelometer devices to measure turbidity. The sensors consist of a 875-nm infrared source and silicon photodetector. IR radiation scattered at angles of 140± 160° in front of the sensor produces a photocurrent in the detector which is ampli®ed by an electronic circuit. The level of the photocurrent is directly proportional to the mass concentration of scattering particles (D&A Instruments 1991). The sensors were calibrated to a formazin turbidity standard, and data were recorded in formazin turbidity units (FTU; 1 FTU=1 nephelometric turbidity unit). Data were recorded sporadically from June 1999 through February 2000. The irregularity of data collection is attributable to dif®culties in keeping the remote sampling stations in constant operation. The data reported focus on July 1999 and January 2000, when the largest turbidity events during the study period occurred. Soil-stratigraphic and dendrogeomorphic evidence was examined in an effort to determine historic sedimentation patterns. A cross-¯oodplain survey line normal to the channel was surveyed using a level and stadia. Living trees were inventoried along the surveyed transect. If any portion of the base of a tree fell within 1 m of the transect center line (e.g., a 2-m swath), it was identi®ed and the diameter at breast height (DBH) measured with a diameter tape. Vegetation indices for the transects were calculated as follows:
however, that turbidity at the lower site approximately 16 km downstream from Lake Nacogdoches was reduced as a consequence of the impoundment. Data comparing the upper and lower site turbidity for July 1999 are shown in Fig. 6. While the mean turbidity for the upper site is higher (49.2, SD 78.3) due to the sustained rise shown in the latter part of the data, it can be seen that for most of the graphed period the lower site turbidity (mean 25 FTU; SD 33.9) is greater than or equal to that upstream of the lake. In January 2000, the lower site can be seen to have higher turbidity during those periods for which there are data from both sites (Fig. 7). For the period 17±31 January when the data overlap completely, the mean is 166.9 FTU for the lower site (65.5 upper), and the standard deviation is 86.3 (17.1 upper).
1. Basal area
m2 p
DBH=1002 =4, where DBH is the diameter at breast height (cm). 2. Relative dominance of species A = basal area of A/basal area of all trees. 3. Relative density of species A=number of individuals of Fig. 6 Comparison of turbidity records for the upper and lower sites for a A/number of all trees. period in July 1999. x-Axis Number of readings, which 4. Basal area ratio = basal area/area of transect, where the three-week were collected every 15 min transect area excludes the stream channel and pasture areas. 5. Density = number of trees/area of transect.
Results Turbidity Turbidity is in¯uenced by biological factors as well as inorganic suspended solids and sediment loads. However, given the close proximity of the sites, it is reasonable to assume that most of the variation in turbidity is associated with suspended solids. Turbidity readings at the middle site were consistently low, as would be expected immediately downstream of the dam. The exception is when dam releases stirred up loose bottom sediments, resulting in brief turbidity spikes. In general, however, turbidity here was generally
