WETLANDS, Vol. 29, No. 2, June 2009, pp. 451–464 ’ 2009, The Society of Wetland Scientists COMPLEX EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION ON WESTERN TENNESSEE FLOODPLAIN FOREST FUNCTION Scott B. Franklin1,2, John A. Kupfer2,3, S. Reza Pezeshki4, Randy Gentry5, and R. Daniel Smith6 1 School of Biological Sciences, University of Northern Colorado Greeley, Colorado, USA 80639 E-mail: [email protected] 2 Edward J. Meeman Biological Field Station Millington, Tennessee, USA 38053 3 Department of Geography & University of South Carolina Columbia, South Carolina, USA 29208 4 Department of Biology, University of Memphis Memphis, Tennessee, USA 38152 5 Southeastern Water Resources Institute, The University of Tennessee Knoxville, Tennessee, USA 37996 6 US Army Corps of Engineers Waterways Experiment Station Vicksburg, Mississippi, USA 39180 Abstract: Data on vegetation composition and structure, soil and leaf nutrient pools, soil redox potential, and surface water hydrologic connectivity were collected from floodplains along six river reaches in western Tennessee to examine the effects of channel modifications on associated riparian systems. Comparisons among channelization treatments (non-channelized reaches, channelized and leveed reaches, channelized but non-leveed reaches) and floodplain geomorphology (depression and nondepressional sites) showed that hydrologic connectivity was affected by channelization treatments, particularly leveeing. The disconnected floodplains were drier, maintained higher nutrient pools, and had greater herbaceous biomass than floodplains still connected to channel hydrology. Runoff onto floodplains from the agriculturally dominated landscape of channelized and leveed tributaries, and flooding stress in the form of scour on floodplains along streams without levees may explain the observed pattern. Channel and floodplain hydrologic processes were most strongly connected for unchannelized streams. Unchannelized streams were varied in soil redox potential, water table, and nutrient pools. Vegetation composition reflects both historical hydrologic regimes and disturbances, and thus complex relationships to channel modifications. Results suggest both the subsidy (i.e., nutrient inputs) and the stress of flood events have been altered by anthropogenic activities, but these alterations were greatest in channelized systems compared to unchannelized systems. Key Words: ecology, hydrology, nutrient pools, productivity, soil redox INTRODUCTION water flow and alter the flood regime. With greater pressure on the utilization of floodplain systems, there exists a global effort to understand floodplain functions so land stewards may assess effects of anthropogenic manipulations (Van Looy et al. 2003, Oswalt and King 2005) and success of restoration efforts (Palmer et al. 2005, Orr et al. 2007). River channels in many areas of the world have been straightened, deepened, widened, and leveed to accelerate storm water drainage, decrease overbank Floodplains have the ability to maintain or improve water quality (e.g., Tabacchi et al. 2000), but this function relies on the preservation of intact riparian ecosystems that in turn limits the conversion of potentially productive floodplains to agriculture. In the southeastern U.S., changes in land use and land cover in riparian zones have been made possible by the implementation of stream management practices designed to expedite or regulate 451 Wetlands wetl-29-02-04.3d 7/4/09 11:13:19 451 Cust # 08-59 452 WETLANDS, Volume 29, No. 2, 2009 flood events, lower water tables of bottomland forests, and protect agricultural land from flooding (Brookes 1989). Channelization alters stream power and sediment transport capacity of a river, which in turn triggers changes in erosion and deposition that result in morphological adjustments of the river channel (Landwehr and Rhoads 2003). Channelization may thus alter hydrologic regimes and fluvial processes in complex manners that are manifested at a range of spatial scales and are related not only to local conditions but also to landscape-level patterns and processes (e.g., modifications to upstream river systems). Levee construction is another ubiquitous alteration, often accompanying channelization in a further effort to constrain floodwaters within the active channel. Levees are barriers to flood pulses (Junk et al. 1989, Bayley 1995), eliminating the connection between channels and floodplains important for fish spawning (Bayley 1995), macroinvertebrate dispersal (Jenkins and Boulton 2003), nutrient cycling (Valett et al. 2005), and other landscape-level functional processes (Ward et al. 2002). Alteration of flood regimes has repercussions on floodplain ecosystem structure as well, albeit not well understood, simply due to the decreased heterogeneity of disturbance and environmental conditions throughout the floodplain leading to lower overall diversity (Simon and Hupp 1992, Sparks 1995, Shankman 1996). Effects of channel modifications are likely complicated by a range of confounding factors, including microtopography (e.g., depressions vs. nondepressions), differences in geomorphic responses of the streams (e.g., incision vs. aggradation), nonequilibrium conditions and responses of the floodplain ecosystem (e.g., the ‘‘storage effect’’: Chesson 1990), and the extent and effectiveness of the channel modifications themselves (which may change with time). Such inherent variability among modified streams has hampered generalizations of their modification effects (Steiger et al. 2005). The overall objective of this study was to clarify the effects of channelization and levee construction on floodplain forest composition, diversity, structure, productivity, surface hydrologic connectivity, and leaf and soil nutrient pools. Hydrologically, we expected that channelization could result in either: 1) lowered water table levels, as water is moved more quickly out of channelized reaches, or 2) higher water table levels, in areas where upstream channelization is exporting water at a faster rate than can be accommodated due to nearby impoundments such as valley plugs, beaver dams, or other channel Wetlands wetl-29-02-04.3d 7/4/09 11:13:53 452 Figure 1. Location of river floodplain study areas and selected USGS gage stations in western Tennessee, USA. Unchannelized reference reaches: Hatchie and Upper Wolf Rivers; Channelized stream reaches: Stokes Creek and Obion Rivers; Channelized and leveed reaches: Lower Wolf and North Fork-Forked Deer Rivers. Channelized reaches are shown in grey; unchannelized reaches are shown in black. blockages. In leveed systems, we anticipated that levees had cut off river inputs and converted these sites to rain-fed systems (although sub-surface hydrology may still be connected), making leveed floodplains drier than non-leveed systems still connected to river hydrologic processes. This conversion could also lead to lower soil nutrient levels because levees would remove inputs associated with periodic flood events, or increased nutrients deposited by sheet flow from neighboring agricultural fields. Finally, we expected that changes in floodplain hydrologic connectivity and nutrient availability would result in structural and compositional differences of the floodplain forest communities, but that such differences would not be consistent among rivers with comparable modifications due to stream-to-stream variability. METHODS Study Areas We selected six river sections ranging in length from 3–8 km for this study, two from unchannelized reaches (the upper Wolf River and the Hatchie River in the Hatchie National Wildlife Refuge), two from Cust # 08-59 Franklin et al., EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION channelized reaches (Stokes Creek, a tributary to the North Fork of the Forked Deer River, and the Rutherford Fork of the Obion River in the Milan Arsenal), and two from channelized and leveed reaches (the North Fork of the Forked Deer River in Tigrett Wildlife Management Area and the lower Wolf River in the Lucius Burch Natural Area) (Figure 1). Although the main channels of the unchannelized stream reaches have not been altered, many of their tributaries have been channelized, affecting their hydrologic processes and geomorphology. Our reference rivers are thus not truly natural, but they represent the best remaining examples of large, free flowing unchannelized rivers in the region and should be classified as the least disturbed condition based on Stoddard et al. (2006). Further, they have been used to develop reference standards for the hydrogeomorphic classification system for low gradient riverine systems in western Tennessee (Franklin et al. 2009). Vegetation in the reference systems is typical of the northern portion of the southern floodplain forest (Sharitz and Mitsch 1993, Hodges 1998), with prominent species including Taxodium distichum (L.) Rich., Salix nigra Marsh., and Quercus nigra L. in depressions and Liquidambar styraciflua L., Platanus occidentalis L., and Acer saccharinum L. on floodplain nondepressions (Franklin et al. 2001a,b). However, timber cutting has been pervasive throughout West Tennessee and has likely affected forest structure and composition in all of the reaches to some degree. Regionally, clearing for agriculture began in the early 1800s as cotton farming developed. Additional waves of forest loss and conversion coincided with railroad expansion during the turn of the 20th century and agricultural expansion in the first third of the century, so that all areas of this study were likely logged at least once (Table 1). More recently, bottomland hardwood area in the southeastern United States, and western Tennessee specifically, decreased again by 25% between 1950 and 1971 (Steed et al. 2002). Concurrent with increases in agriculture, modifications to many rivers in western Tennessee, including snag removal, dredging, widening, straightening, and leveeing, began in the early 1900’s, and most major stream channels in western Tennessee had been dredged and straightened by 1926 (Simon and Hupp 1992). Subsequent periods of debris accumulation and channel filling from sediment deposition necessitated further stream clearing, channel snagging, and occasional widening and straightening of stream reaches (Robbins and Simon 1983, Johnson 2007). Even though these early attempts at channelization failed, a re–invigorated Wetlands wetl-29-02-04.3d 7/4/09 11:13:54 453 Cust # 08-59 453 effort was begun in 1961 and led by the Corps of Engineers as part of the West Tennessee Tributaries Project. These additional efforts also failed and were stopped by landowners who petitioned not to have their land channelized (Johnson 2007). Of the reaches examined in this study, none have experienced channel modifications (excluding clearing and snagging) since the mid-1970’s. ‘‘Natural’’ channel widening in the modified areas, an indicator of channel–bed degradation and ongoing stream adjustment, was occurring at rates in the late-1980’s that were comparable to those due to widening from bank caving on the unchannelized reaches (Simon and Hupp 1992). As might be expected, the channelized systems examined in this study had less forest cover in the surrounding watersheds (Table 1) and higher sediment loads. For example, shortly after the most recent channel adjustments, sediment output in the Forked Deer River averaged 9178 kg ha21 versus only 2608 kg ha21 in the Hatchie (Johnson 2007). Beyond the effects of overbank flow events, bottomland forests can be affected by burial from sediments eroding from uplands and tributary channels and increased inundation due to levees that impeded drainage of floodplains and valley plugs. Climatically, the annual average temperature in the region is 16uC, with mean January and July temperatures of 4uC and 26.6uC, respectively. Average growing season is 230 days, and average annual precipitation is 132 cm with a majority occurring during the winter and spring months (USDA 1978). Soils along all six reaches are of the Falaya-Waverly-Swamp associations (coarse–silty, mixed, active, acid, thermic Fluvaquentic Endoaquepts and thermic Aeric Fluvaquents), which consist of level, poorly drained silty soils on low, broad first bottoms (USDA 1965, 1978). In the absence of channel modifications, these areas would have been flooded in most years during the winter and spring and occasionally during the summer, with periods of inundation ranging from weeks to months. These soils formed in acidic loess washed from uplands and are thus highly susceptible to erosion. Data Collection We established four to nine transects in each of the study reaches, with river study sections ranging from 3 to 8 km. Transects were aligned perpendicular to the stream channel, positioned at least 100 m apart on forested first bottoms, and spatially stratified along the entire river reach. To avoid Wetlands wetl-29-02-04.3d 7/4/09 11:13:54 454 Cust # 08-59 Unchannelized Unchannelized Upper Wolf River Hatchie River Channelized Stokes Creek Channelized & Leveed Lucius Burch NA Channelized Channelized & Leveed Tigrett WMA Milan Arsenal Treatment River Loess, and Claiborne & Wilcox Formations Loess, and Claiborne & Wilcox Formations Loess and Holocene Alluvium Loess, and Claiborne & Wilcox Formations Loess and Holocene Alluvium Loess and Holocene Alluvium Geology 3,112 79,303 259,204 12.9 50.1 60.2 31,384 192,257 86.6 36.8 120,525 64.7 Length Drainage (km) Area (ha) 42 50 2 29 25 12 % Basin Wetlands 32 49 98 68 30 87 % Basin Ag 5 8 4 5 3 5 Study Section Length (km) Logged in early 1900s, ponding death None in recent history Logging History Logged in early 1900s, ponded death, planted cypress Selectively Middle of thinned study area 1947– in main 1948, channel ponded death Privately Above and owned below study patchy area in main cutting, channel ponded death Above study 1960s, but not on area in study sites tributaries Tributaries Not seen, but floodplain leveed for hunting Not seen Channel Sediment Plugs 1964 Enlarging and Straightening none none None 1970s None Clearing and snagging 1976–1978 Clearing and Snagging, below study area 1977 Enlarging and Straightening Recent Date Modifications Completed Table 1. Modification history of the six West Tennessee river reaches used to examine the effects of channel alteration on floodplain forests. Nearly all original channelization projects were begun in the early 1900s. Geology was from Simon and Hupp (1992). Channel sediment plugs were from personal observation and Diehl (2000). Dates for recent modifications are from Hupp (1992) and Johnson (2007). Recent modifications on the Upper Wolf and Hatchie Rivers are only on tributaries and not given. Logging history was developed from Johnson 2007 and interviews with local residents. 454 WETLANDS, Volume 29, No. 2, 2009 Franklin et al., EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION autocorrelation of samples, sample sites (1 to 4) were located at least 50 m apart along each transect but within 1 km of the stream channel and stratified evenly according to their locations in depressions (areas of concave surface microtopography) or nondepressions (areas of straight or convex microtopography). Thus, half of the sites were in depressions for each sample area, and depressions and nondepressions alternated along transects. Data on water table level and variability were collected from January 1999–June 2000 using surface wells installed at each sampling site in 1998. A total of 588 measurements were recorded, ranging from 29 to 146 samples per site, spaced over three to eight time periods. While such limited data collection does not permit a refined analysis of floodplain hydrologic processes, it does allow for a limited comparison of mean soil water table levels and variability among research areas. In the summer of 1999, we installed redox potential electrodes at 24 sites, with two each randomly selected from depressions and nondepressions at each study reach. Redox potential measurements at three depths (15, 30, 60 cm) were taken every two weeks from June 1 through October 30, 1999. Historical data from USGS river gages were collected and compared to floodplain soil water levels to examine the strength of connection between channel and floodplain hydrologic processes. Data on forest structure and composition were recorded in summer 1998 using 20 3 10 m quadrats established at 89 sites, with 36 in unchannelized reaches, 28 in channelized reaches, and 25 in channelized and leveed reaches. Diameter and species of each live tree $ 10 cm in diameter at breast height (dbh; 1.37 m) were recorded to provide information on overstory trees while individuals 0.1– 10 cm dbh were identified by species and counted to determine density in the midcanopy (3.0– 9.9 cm dbh) and sapling (0.1–2.9 cm dbh) strata. At the 24 sites where redox potential was measured, we established nine 1 m2 herbaceous biomass subplots in an area of uniform herbaceous flora composition. In June 1999, we removed and separately bagged all above ground herbaceous biomass and dead litter biomass from three randomly selected plots. All bagged material was dried at 60uC for 48 hours and weighed to the nearest 0.01 g, and averages of the three plots were determined. We also placed 1 m2 litter fall traps to collect leaf fall on these same 24 sites from October 1–December 31 in 1998 and 1999. Nets were emptied every other week, and the downed material was oven–dried and weighed. Ten surface soil samples were collected from the same 24 sites in spring 1999 Wetlands wetl-29-02-04.3d 7/4/09 11:13:55 455 Cust # 08-59 455 and 2000 using a 7.62 cm diameter soil corer to a depth of 20 cm. Soil samples from each site were mixed thoroughly, dried, and sifted through a 2 3 2 mm sieve. Both leaf litter tissue and soil samples were sent to A & L Agricultural Laboratories (Memphis, TN) for nutrient analyses. Data Analysis We examined the effects of channelization and leveeing on a wide range of biotic and abiotic characteristics, including: Site hydrologic processes: mean water table depth (cm) below ground surface, water table variability (reach: among plots on a given date; local: within surface well during year 2000), and soil redox potential. Forest structure: overstory basal area (m2 ha21), overstory density (# ha21), mid–canopy density (# ha21), and sapling density (# ha21); Forest composition and diversity: species composition, species richness (# of species per plot), and species evenness (based on equitability of Shannon-Wiener Index values; McCune and Mefford 1997), in both the overstory and mid–canopy; Nutrient pools: leaf litter nutrient pools: Ca (ppm), K (ppm), N (ppm), P (ppm), and C/N (carbon/ nitrate–N ratio); soil nutrients and characteristics: N (ppm), P (ppm), organic matter (%), cation exchange capacity (meq/100 ml), and C/N (carbon/nitrate–N ratio); In most cases, a nested ANOVA (reaches were nested in TRT, SAS PROC: GLM) was used to examine the effect of channelization treatments (TRT; unchannelized vs. channelized vs. channelized and leveed), topography (DND; depressions vs. nondepressions), and their interaction. Treatment was also included as a subplot of reach to test for differences among stream sections. Upon a significant REACH effect, factors were subsequently blocked by reach to test for significant TRT, DND, and interaction effects. Significant treatment effects were examined with a Tukey post hoc analysis. All samples were considered replicates because of their spacing and alternation of depression and nondepression sites, decreasing autocorrelation. Floristic differences among the sites were investigated using multi–response permutation procedures (MRPP; McCune and Mefford 1997), a technique that tests for differences among groups defined a priori based on the ratio of dissimilarity among groups compared to within groups in Euclidean space. Hydrologic relationships were explored with 456 WETLANDS, Volume 29, No. 2, 2009 correlation analyses of channel discharge and floodplain water table readings from a particular day. To examine the concept that water table fluctuations yield greater productivity (Robertson et al. 2001), leaf and soil nutrients were correlated with mean water table and water table variability (i.e., standard deviation). All inferential analyses were performed in SAS (2003). RESULTS Floodplain Hydrologic Processes Daily water table readings were correlated with stream discharge for both unchannelized reaches and one channelized reach. The strongest correlation was for the Hatchie River (r 5 0.48), while the Upper Wolf and Rutherford Fork of the Obion could only be modeled about half as well (r 5 0.22 and 0.20, respectively). Both the Upper Wolf and Hatchie scattergrams had a consistent relationship between floodplain water–table level and stream gage discharge, while channelized and channelized and leveed stream reaches show the entire range of floodplain water–table levels under low flow conditions, suggesting a disconnect between channel and floodplain hydrologic processes at low flows. Mean depth to water table differed as a function of topographic setting (F 5 7.3; p.F 5 0.02) and reach (F 5 20.6; p.F 5 0.0001). Sites along stream reaches without levees had much shallower water tables that differed substantially between depressions and nondepressions (Figure 2a). Those along leveed reaches, on the other hand, had a greater depth to water table and little difference between topographic settings. The floodplain hydrologic regime also differed among reaches (Figure 2b), with sites at Lucius Burch (channelized and leveed) being the driest and those along the Upper Wolf (unchannelized) the wettest. Water table variability (standard deviation of individual surface wells) had a treatment by topography interaction (F 5 4.9; p.F 5 0.01), with differences between depressions and nondepressions in channelized and unchannelized reaches but not in channelized and leveed reaches. Variability was higher in depressions for unchannelized reaches and higher in nondepressions for channelized reaches (Figure 3a). Water table variability also differed among reaches (F 5 7.8; p.F 5 0.0002), with less variability in channelized and leveed reaches than channelized or unchannelized reaches (Figure 3b). River water table reach variability (variance in water tables across wells for a day in March) was extremely low at the Lucius Burch reach, and thus differed from all other reaches (Figure 3c). Wetlands wetl-29-02-04.3d 7/4/09 11:13:55 456 Figure 2. Mean depth to water table and soil redox potential in floodplain sites along six river reaches in western Tennessee: (a) and (c) treatments include topography, depression versus nondepression sites, and river alterations, including channelized (C), channelized & leveed (C&L), and unchannelized (U); (b) and (d) rivers Cust # 08-59 Franklin et al., EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION 457 Patterns of soil redox potential mirrored but were even stronger than those for water table depth; a significant treatment by topography interaction (F 5 8.8; p.F 5 0.001) and reach effect (F 5 24.8; p.F 5 0.0001; Figure 2c). Redox potential was greatest in channelized and leveed reaches (irrespective of topographic position), with soils remaining in the oxidation zone (Eh . 350 mV) throughout the sampling season. Sites along both channelized and unchannelized reaches, where the river and floodplain hydrologic processes were still connected, had lower redox potentials in depressions compared to non–depressions (Figure 2c), and the channelized reaches and one unchannelized reach (Upper Wolf River) began the growing season with reducing soils (Figure 2d). The reach effect occurred because while soil redox potential was similar between channelized reaches and between channelized and leveed reaches, there was a greater disparity between the two unchannelized locations. Forest Structure and Composition Overstory basal area did not differ among treatments. There was a reach effect on mid–canopy density (F 5 4.8; p.F 5 0.04), indicating variability among river systems unrelated to their treatment. Otherwise, forest structure showed little response to treatment, topography or location. Overstory and mid–canopy composition, however, both showed an interaction between treatment and topography (overstory: MRPP delta 5 0.86, p , 0.0001; mid–canopy: MRPP delta 5 0.82; p , 0.00001). Of particular note was the absence or reduced prominence of characteristic depressional species Taxodium and Nyssa in the modified stream reaches, the variability of specific early successional species among samples (e.g., Salix nigra, Betula nigra, Quercus michauxii, Liquidambar styraciflua), and the prominence of Acer rubrum in the modified stream system overstories and in nearly all mid– canopies (Table 2). Also notable is the strong r include two channelized and leveed stream sections, the North Fork of the Forked Deer Tigrett Wildlife Management Area (NFFD) and Lucius Burch (LB, lower Wolf River), two channelized stream sections, the Rutherford Fork of the Obion River in the Milan Arsenal (RF Obion) and Stokes Creek (Stokes), and two unchannelized stream sections, the Hatchie River Wildlife Refuge (Hatchie) and the Upper Wolf River near Moscow (Up Wolf). Bars are standard error. Different letters represent significant differences among treatments. Wetlands wetl-29-02-04.3d 7/4/09 11:14:05 457 Cust # 08-59 Figure 3. Mean of standard deviations for individual surface wells from six river reaches in western Tennessee: a) Treatments include topography, depressions versus nondepressions, and river alterations, including channelized, channelized & leveed, and unchannelized; b) rivers include two channelized and leveed stream reaches, the North Fork of the Forked Deer in Tigrett Wildlife Management Area (NFFD) and Lucius Burch (LB, lower Wolf River), two channelized stream reaches, the Rutherford Fork of the Obion River in the Milan Arsenal (RF Obion) and Stokes Creek (Stokes), and two unchannelized stream reaches, the Hatchie River Wildlife Refuge (Hatchie) and the Upper Wolf River (Up Wolf); c) standard deviation of surface wells for one day in March 2000 for each stream section. Bars for a and b are standard error. Different letters represent significant differences among treatments. 458 WETLANDS, Volume 29, No. 2, 2009 Table 2. Mean density (stems ha21) for midcanopy (3–10 cm dbh) from floodplain sites along six West Tennessee river reaches. Data are presented by treatment (channelized, C; channelized & leveed, CL; unchannelized, U), and topography (depressions, D; nondepressions, N). Species n5 CL, N CL, D 15 Taxodium distichum 0.0 Nyssa aquatica 0.0 Planera aquatica 0.0 Salix nigra 0.0 Platanus occidentalis 3 6 13 Carya spp.1 20 6 75 Quercus spp.2 7 6 18 Fraxinus pennsylvanica 123 6 230 Ilex decidua var. decidua 3 6 13 Walt. Acer rubrum 253 6 437 Carpinus caroliniana Walt. 60 6 112 Celtis laevigata Willd. 17 6 36 Acer saccharinum 3 6 13 Acer negundo 57 6 111 Betula nigra 0.0 Liquidambar styraciflua 0.0 Ulmus americana 30 6 72 Other species 57 6 123 Total Density 683 6 410 Richness 2.5 6 1.8 Evenness 0.49 6 0.4 C, N C, D 13 14 19 16 3 6 13 27 6 107 23 6 70 0.0 20 6 47 53 6 173 67 6 267 7 6 27 29 6 81 0.0 0.0 79 6 216 5 6 23 29 6 96 5 6 17 11 6 37 61 6 147 0.0 7 6 17 0.0 71 6 161 7 6 17 18 6 75 0.0 10 20 6 53 0.0 10 6 32 40 6 68 30 6 63 0.0 15 6 41 90 6 211 90 265 60 30 40 15 5 85 75 970 2.4 0.66 6 198 6 610 6 87 6 79 0.0 6 67 6 48 6 16 6 171 6 180 6 853 6 1.3 6 0.5 8 6 36 0.0 238 6 400 8 6 40 8 6 32 0.0 8 6 33 92 6 157 88 588 19 12 12 77 27 15 50 1400 3.9 0.65 6 121 6 1335 6 38 6 42 6 42 6 177 6 48 0.0 6 42 6 100 6 813 6 2.4 6 0.3 73 120 17 17 7 10 90 37 770 2.9 0.70 0.0 6 280 6 211 6 53 6 53 6 26 0.0 6 27 6 281 6 120 6 703 6 1.2 6 0.3 U, N 0.0 529 6 1141 50 6 111 3 6 12 0.0 8 6 26 58 6 151 8 6 34 42 6 98 24 6 77 1129 6 1430 4.7 6 1.8 0.84 6 0.1 U, D 14 325 68 21 7 43 7 96 57 68 1171 2.9 0.52 6 6 6 6 6 6 6 6 6 6 6 6 6 39 770 151 75 17 126 17 341 141 153 1910 2.1 0.4 1 Carya glabra (Mill.) Sweet, C. illinoensis (Wangenh.) K. Koch, C. laciniosa (Michx.) Loudon, C. ovata var. ovata (Mill.) K. Koch, C. tomentosa (Poir.) Nutt. Quercus lyrata, Q. michauxii, Q. nigra, Q. texana Buckl., Q. phellos L., Q. pagoda Raf. 2 difference in mid-canopy density of A. rubrum and F. pennsylvanica between nondepressions and depressions on channelized and unchannelized sites, but not on channelized and leveed sites (Table 2). Despite the mid-canopy differences, sapling composition was not different based on either treatment or topography. Overstory richness and evenness were comparable among treatments and between topographic settings, but an interaction effect for richness (F 5 6.2; p.F 5 0.003) suggested the importance of stream modifications on floodplain forest diversity. While nondepressions in the unchannelized reaches had the highest richness values (R 5 4.9), depressions in the same reaches had the lowest values (R 5 3.1). In contrast, richness values were intermediate in the modified stream reaches, and depressions were either comparable to or richer in species than nondepressions, suggesting a treatment effect but one that is altered by topographic setting. These patterns were reinforced by comparisons of mid– canopy diversity, with treatment–level differences again obscured or altered by topographic setting (nondepression sites had higher richness; F 5 7.0; p.F 5 0.01) or a treatment by topography interaction (same evenness relationship as overstory; F 5 Wetlands wetl-29-02-04.3d 7/4/09 11:14:15 458 4.9; p.F 5 0.01). Unchannelized reaches had mid– canopy diversity values similar to those for the overstory, but richness was distinctly lower (and relatively equal between depressions and nondepressions) in the channelized and leveed reaches (Table 2). Maximum litter biomass was different based on treatment (F 5 92.34; p.F 5 0.0004), with sites along channelized and leveed reaches having higher biomass than those along channelized or unchannelized reaches. No treatment effects were found for live shoot or root biomass. Soil Characteristics and Leaf Nutrients Analyses of soil characteristics showed almost no direct treatment effects but were complicated by interaction effects, year-to-year variation, and reachto-reach variability (Figure 4). A treatment by topography interaction was found for all measures in at least one year, indicating that channel modifications may influence floodplain nutrient dynamics but that any changes are likely dependent on the topographic setting (Table 3). For example, C:N ratios were greater in depressions compared to nondepressions in channelized and unchannelized Cust # 08-59 Franklin et al., EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION 459 Figure 4. Percent organic matter, cation exchange capacity, and carbon/nitrate-nitrogen ratios from floodplain soils along six western Tennessee river reaches during two spring sampling periods (1998 and 1999). Treatments include topography, depression versus nondepression sites, and river alterations, including channelized (C), channelized & leveed (C&L), and unchannelized (U). Bars are standard error. Different letters represent significant differences among treatments. reaches in 2000 (Figure 4f). On the other hand, in 1999, C:N was higher only in nondepression sites of channelized and leveed reaches. In 1999, organic matter was greatest in depressions (Figure 4a), resulting in lower cation exchange capacity as well (Figure 4c), but only for sites in channelized reaches. Further, a reach effect was noted for all measures except those involving nitrogen (which was most strongly affected by topographic setting), meaning that the different stream reaches had different soil nutrient pools and structure. A topographic effect (F 5 7.2; p.F 5 0.01) for leaf calcium in 1998 was the only significant treatment effect; depression sites had lower leaf calcium pools than nondepression sites. While relationships between leaf and soil nutrients might be expected, soil and leaf nutrient data were Wetlands wetl-29-02-04.3d 7/4/09 11:14:16 459 Cust # 08-59 uncorrelated (p . 0.10) for all treatments, and there were no relationships between mean water table level and either soil or leaf nitrogen and phosphorus (largest r 5 0.27; lowest p 5 0.26). Negative correlations were found, however, between the variability of the water table at a particular well (i.e., standard deviation) and leaf nitrogen (r 5 20.58; p 5 0.01) and soil phosphorus (r 5 20.44; p 5 0.06). DISCUSSION We hypothesized that channelization and leveeing of western Tennessee rivers would directly affect floodplain hydrologic processes and thereby alter floodplain forest structure, composition and function. In particular, we thought that levees should 460 WETLANDS, Volume 29, No. 2, 2009 Table 3. Nested ANOVA results for comparisons of soil characteristics from floodplains along six westernTennessee river reaches. TRT 5 treatment (channelized, channelized & leveed, and unchannelized); DND 5 topography (depression or nondepression); RIVER 5 the individual stream. Bold indicates significance ( 5 0.05< 0.2). Variable dferror Soil Organic Matter 1999 2000 15 Cation Exchange Capacity 1999 2000 15 Carbon/Nitrate–N ratio 1999 2000 15 Nitrogen 1999 2000 15 Phosphorus 1999 2000 15 Potassium 1999 2000 15 Calcium 1999 2000 15 TRT DND TRT*DND River (TRT) df 5 2 df 5 1 df 5 2 df 5 3 F p.F F p.F F p.F F 1.85 0.35 0.30 0.71 0.93 0.03 0.34 0.88 3.00 3.78 0.05 0.05 7.83 4.2 0.0003 0.03 0.66 0.16 0.58 0.86 0.12 0.35 0.73 0.57 5.57 0.00 0.007 0.99 8.69 7.37 0.0001 0.004 1.38 1.82 0.38 0.30 3.80 1.32 0.06 0.27 3.14 5.47 0.05 0.02 1.79 1.05 0.17 0.41 3.71 1.48 0.16 0.35 14.10 3.13 0.001 0.10 6.30 3.12 0.004 0.08 1.36 0.78 0.27 0.53 0.99 7.29 0.47 0.07 5.87 0.39 0.02 0.55 3.77 2.17 0.03 0.16 2.50 0.85 0.07 0.49 0.82 0.25 0.52 0.79 1.68 0.12 0.20 0.73 7.10 0.47 0.002 0.63 1.16 0.54 0.42 0.63 0.35 1.38 0.56 0.26 6.35 0.49 0.004 0.62 have the greatest impact, and indeed, the clearest distinction between altered and unaltered channels occurred when levees were constructed. Levees detached the floodplain from the channel, decreased overbank flood events and rendered lower water tables and higher soil redox potential compared to non–leveed stream systems (Gergel et al. 2002, Giller 2005, Kang and Stanley 2005). While treatment effects were evident for some aspects of floodplain form and function, the effects of stream channel modifications were often manifested through interactions of river modifications with other factors influencing site hydrologic processes, including topographic setting and reach– specific conditions. Direct treatment effects were thus absent from many comparisons of forest structure, composition, and soil nutrient pools, even though stream modifications (as altered by interaction effects) were clearly crucial. The interactions underscore the complexity of floodplain ecosystems and the importance of topographic heterogeneity (Richter and Richter 2000, Ward et al. 2002, Oswalt and King 2005, Steiger et al. 2005). Wetlands wetl-29-02-04.3d 7/4/09 11:14:21 460 18.0 27.4 4.95 5.45 p.F ,0.001 ,0.001 0.005 0.01 If a stream is channelized and leveed and the levees are not breeched, then the floodplain is completely cut off from the channel with no surficial exchanges of nutrients, moisture, or organic matter, except potentially during extreme flood events. An incising (i.e., degrading) channel, caused by an increased gradient and stream power, will result in a drier floodplain with greater soil aeration (Hupp 1992). The lower Wolf River reach (Lucius Burch) is the best example from this study. If the levees are breeched or no levees were created, then flooding remains a potentially important function of channelized reaches, albeit to a varying degree. A primary goal of channelization is to export water from the modified area more quickly and thereby prevent flooding and lower the water table. High water tables have nonetheless been shown to occur in channelized stream reaches as a result of upstream channelization, a flashier basin hydrology (with land use conversion to agriculture), and the development of stream blockages (e.g., Shankman and Pugh 1992, Diehl 2000) that locally maintain higher water tables during low flow periods (Piégay 1997, Oswalt Cust # 08-59 Franklin et al., EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION and King 2005). Such blockages occurred on both unchannelized and channelized reaches in our study area and contributed to the high water table levels observed in the latter. We found only one significant correlation between channel stage and floodplain water table of the four channelized stream reaches, while both unchannelized reaches had a significant correlation. This suggests that while channelized streams may still exhibit hydrological variability, floodplain water table was not as strongly connected to channel flow as unchannelized streams, especially during low flow periods (Valett et al. 2005). Aggradation of river channels may be due to several factors, including bank collapse, headcutting, and sediment inputs from surrounding agricultural land (Brown 1988, Shankman and Samson 1991, Nakamura et al. 1997), so morphological stream adjustments and the amount of agriculture in the watershed are directly related to stream sediment input (Coleman and Kupfer 1996, Craft and Casey 2000). Uplands in western Tennessee are capped with highly erodible loess soils, with annual erosion estimates averaging 15 MT ha21 (Troy Taylor, NRCS, pers. comm., September 2001). Deposition in litter bags from a decomposition study at our sites (Molavi unpubl. thesis) showed high values for both the channelized (688–748 g m22 yr21) and unchannelized (757–1628 g m22 yr21) sites, while channelized and leveed streams averaged an annual deposition rate of only 22.7 g m22 yr21. Hupp and Bazemore (1993) also suggested unchannelized streams would have high sedimentation rates. The aggradation or scouring of sediment on floodplains has repercussions on other functions. The nutrient subsidy (Odum 1979) from flood events in channelized systems is likely greater than for unchannelized systems due to nonpoint source fertilizer movement from the surrounding watershed. Channelized reaches averaged 78% agricultural land in their watersheds while unchannelized reaches averaged only 40% agricultural land. Thus, nitrogen, phosphorus, and potassium load were expected to be higher in channelized reaches than in unchannelized reaches, as was evidenced with soil nutrient pools (this study, Franklin et al. 2001b). Grubaugh and Maier (unpublished data) found higher water nutrient concentrations in Stokes Creek, surrounded by agriculture, compared to the Upper Wolf River surrounded by mainly forest, corroborating our results. Obviously, this is an additional subsidy, but these streams are also receiving an additional stress due to increased sediment loads. We found negative relationships between water table variability, and leaf nitrogen and soil phosphorus. While periodic flooding Wetlands wetl-29-02-04.3d 7/4/09 11:14:22 461 Cust # 08-59 461 generally results in higher nutrient availability (Mitsch et al. 1979, Baldwin and Mitchell 2000, Brunet and Astin 2000), higher productivity (Robertson et al. 2001), and increased decomposition (Brinson 1981, Lockaby et al. 1996), it is not the case when flows carry heavy sediment loads (Brookes 1986); sedimentation stress effects outweigh subsidy effects. The importance of floodplain microtopography in shaping patterns of hydrologic connectivity, species composition, and some soil nutrients was not surprising. Previous studies suggest that continuous flooding of depressions, resulting in anoxic soil conditions, increases the mobilization of some minerals for plant uptake (e.g., phosphorus) while decreasing nutrient uptake capacity of roots (Pezeshki 1994) and decomposition rates (Mitsch and Gosselink 1993, Brunet and Astin 2000). As expected for sites along unchannelized and channelized reaches, mean depth to water table was less in depressions than nondepressions, and soil redox potential was lower, representing moderately reducing soil conditions. Contrarily, floodplain water table and soil redox potential of sites along channelized and leveed sites did not differ between depressions and nondepressions. Thus, while we hypothesized that depressions would have lower soil and leaf nutrient pools because more of the nutrients would be tied up in slowly decaying biomass, microtopography and nutrient levels differed only when considered within the context of channelization effects. By decoupling the floodplain and stream systems, levees minimize the differences between depressions and nondepressions and potentially reduce the amount of floodplain environmental heterogeneity (Hughes and Cass 1997, Ward et al. 2002). The two key factors determining the responses of the floodplain ecosystems to stream modifications are thus the effects of the modifications on site hydrologic connectivity itself and the manner in which the energy regime and sediment transport capacity of rivers is altered, particularly as it affects the movement and accumulation of sediment, the same two factors suggested by Steiger et al. (2005). Consequently, flood regimes and site hydrologic processes are considered the key factors constraining many aspects of floodplain forest structure, composition, and function due to the varying tolerances of species to water table dynamics (Hosner 1960, Hook and Brown 1973, Sharitz and Mitsch 1993, Pezeshki and Anderson 1997). For example, both A. rubrum and F. pennsylvanica had greater density on nondepressions than depressions of channelized and unchannelized sites, but no difference on channel- 462 WETLANDS, Volume 29, No. 2, 2009 ized and leveed stream reaches, again suggesting loss of topographic heterogeneity following levee construction. While we found differences in composition, the cause was confounded by disturbances that occurred on all rivers (e.g., logging and ponding) and the storage effect of long-lived forest species (Warner and Chesson 1985). For example, floodplains completely disconnected from the channel (Tigrett WMA and Lucius Burch NA) supported some of the best examples of older Quercus forest but were typified by greatly impoverished understory communitites. The older forest is a remnant of the hydrologic processes that existed prior to channel modifications and survives now only due to the long life span of the trees. The unchannelized systems, while still affected by ponding and cutting, did not have as much of their floodplain area affected by these disturbances and also had natural degradation and aggradation occurring along their meandering channel, leading to structural and functional heterogeneity (Kupfer and Malanson 1993). Thus, they tended to be more diverse overall when considering overstory and understory (this study) and floodplain (beta) diversity (Franklin et al. 2009). Oswalt and King (2005) similarly concluded that channelization impacts on floodplain forests along the Middle Fork of the Forked Deer River (TN) were more temporally and spatially complex than previously thought. Due to such confounding factors, direct effects between vegetation composition and/or structure could not be established, and were less helpful in understanding channelization effects. Another factor leading to greater interaction effects than direct effects was the amount of variability among the six reaches. Basin characteristics, site hydrologic connectivity and subsequently stream modification effects can vary greatly even within a relatively small region such as western Tennessee. Indeed, Ward and Tockner (2001) suggest hydrologic diversity is the unifying theme for river ecology. Our soil water table and soil redox potential data supported the obvious: channelized and leveed sites were less variable. The unchannelized river reaches displayed the greatest amount of variation in several characteristics, likely due to the large variations in hydrologic regime typically found on unaltered floodplains (Robertson et al. 1978, Richter and Richter 2000, Steiger et al. 2005). This variation was evident, for example, in considering soil redox potential, where the two unchannelized reaches had the highest and lowest values of all six rivers in the July sample, while soil redox potentials were quite similar within (but not between) the channelized and the channelized and leveed stream Wetlands wetl-29-02-04.3d 7/4/09 11:14:23 462 replicates. One important indirect effect of channelization (leveed or not) may thus be a decrease in floodplain variability, both topographical and hydrological (Ward et al. 2002, Valett et al. 2005). In conclusion, levees had the greatest impact on floodplain forest functions, eliminating exchanges of nutrients and organic matter and rendering a drier floodplain environment, but many responses to stream modifications were predicated on topographic setting. Channel hydrologic regime was tightly linked to floodplain hydrologic regime for unchannelized reaches, while channel and floodplain hydrologic regimes were decoupled in channelized stream reaches during low flow events. Thus, channelization (accompanied by levee construction or not) led to decreased hydrologic variability and subsequently decreased floodplain habitat heterogeneity. Both channelized and unchannelized stream systems suffer from sediment movement and accumulation, decreasing decomposition rates, nutrient cycling, and productivity. ACKNOWLEDGMENTS This research was supported by the Tennessee Department of Environment and Conservation and the U.S. Army Corps of Engineers. We especially wish to thank Ellen Williams (TDEC) for her help and Steve Stephenson for allowing data collection from the Milan Arsenal. We also thank the many people who assisted in the data collection and analysis, especially Natasja van Gestel, Tanya Scheff, Ryan Hanson, Stacy Anderson, Karla Gage, Kit Brown, Melissa Lee, Mitch Elcan, and Jason Farmer. Facilities provided by the Edward J. Meeman Biological Field Station were essential for the project’s completion. LITERATURE CITED Baldwin, D. S. and A. M. Mitchell. 2000. The effects of drying and re-flooding on the sediment and soil nutrient dynamics of lowland river-floodplain systems: a synthesis. Regulated Rivers: Research & Management 16:457–67. Bayley, P. B. 1995. Understanding large river-floodplain ecosystems. Bioscience 45:153–58. Brinson, M. 1981. Primary productivity, decomposition, and consumer activity in freshwater wetlands. Annual Review of Ecology and Systematics 12:123–61. Brookes, A. 1986. Response of aquatic vegetation to sedimentation downstream from river channelization works in England and Wales. Biological Conservation 38:351–67. Brookes, A. 1989. Channelized Rivers: Perspectives for Environmental Management. John Wiley, New York, NY, USA. Brown, R. G. 1988. Effects of wetland channelization on runoff and loading. Wetlands 8:123–33. Brunet, R.-C. and K. B. Astin. 2000. A 12-month sediment and nutrient budget in a floodplain reach of the River Adour, southwest France. Regulated Rivers: Research & Management 16:267–77. Cust # 08-59 Franklin et al., EFFECTS OF CHANNELIZATION AND LEVEE CONSTRUCTION Chesson, P. 1990. Geometry, heterogeneity, and competition in variable environments. Philosophical Transactions: Biological Sciences 330:165–73. Coleman, D. S. and J. A. Kupfer. 1996. Riparian water quality buffers: estimates of effectiveness and minimum width in an agricultural landscape, western Tennessee. Southeastern Geographer 36:113–27. Craft, C. B. and W. P. Casey. 2000. Sediment and nutrient accumulation in floodplain and depressional freshwater wetlands of Georgia, USA. Wetlands 20:323–32. Diehl, T. H. 2000. Shoals and valley plugs in the Hatchie River watershed. US Geological Survey Water-Resources Investigations Report 00-4279, Denver, CO, USA. Franklin, S. B., J. A. Kupfer, S. R. Pezeshki, T. Scheff, R. Hanson, and R. Gentry. 2001a. A comparison of hydrology and vegetation between a channelized stream and a nonchannelized stream in western Tennessee. Physical Geography 22:254–74. Franklin, S. B., J. A. Kupfer, S. R. Pezeshki, N. van Gestel, and R. Gentry. 2001b. Channelization effects on floodplain functions in western Tennessee. p. 189–201. In R. A. Falconer and W. R. Blain (eds.) River Basin Management. WIT Press, Southhampton, Boston, MA, USA. Franklin, S. B., J. A. Kupfer, S. R. Pezeshki, R. Gentry, and R. Dan Smith. 2009. Efficacy of the Hydrogeomorphic Model (HGM): a case study from western Tennessee. Ecological Indicators 9:267–83. Gergel, S. E., M. D. Dixon, and M. G. Turner. 2002. Effects of altered disturbance regimes: levees, floods, and floodplain forests on the Wisconsin River. Ecological Applications 12:1755–1770. Giller, P. S. 2005. River restoration: seeking ecological standards. Editor’s introduction. Journal of Applied Ecology 42:201–07. Hodges, J. D. 1998. Minor alluvial floodplains. p. 325–43. In M. G. Messina and W. H. Conner (eds.) Southern Forested Wetlands: Ecology and Management. Lewis Publishers, Boca Raton, FL, USA. Hook, D. D. and C. L. Brown. 1973. Root adaptations and relative flood tolerance of five hardwood species. Forest Science 19:225–29. Hosner, J. F. 1960. Relative tolerance of complete inundation of fourteen bottomland tree species. Forest Science 6:246–51. Hughes, J. W. and W. B. Cass. 1997. Pattern and process of a floodplain forest, Vermont, USA: predicted responses of vegetation to perturbation. Journal of Applied Ecology 34:594–612. Hupp, C. R. 1992. Riparian vegetation recovery patterns following stream channelization: a geomorphic perspective. Ecology 73:1209–26. Hupp, C. R. and D. E. Bazemore. 1993. Temporal and spatial patterns of wetland sedimentation, west Tennessee. Journal of Hydrology 141:179–96. Jenkins, K. M. and A. J. Boulton. 2003. Connectivity in a dryland river: short-term aquatic macroinvertebrate recruitment following floodplain inundation. Ecology 84:2708–23. Johnson, J. W. 2007. Rivers Under Siege: The Troubled Saga of West Tennessee Wetlands. University of Tennessee Press, Knoxville, TN, USA. Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river-floodplain systems. Canadian Special Publication of Fisheries and Aquatic Science 106:110–27. Kang, H. and E. H. Stanley. 2005. Effects of levees on soil microbial activity in a large river floodplain. River Research and Applications 21:19–25. Kupfer, J. A. and G. P. Malanson. 1993. Observed and modeled directional change in riparian forest composition at a cutbank edge. Landscape Ecology 8:185–99. Landwehr, K. and B. L. Rhoads. 2003. Depositional response of a headwater stream to channelization, east central Illinois, USA. River Research and Applications 19:77–100. Lockaby, B. G., A. L. Murphy, and G. L. Somers. 1996. Hydroperiod influences on nutrient dynamics in decomposing litter of a floodplain forest. Soil Science Society of America Journal 60:1267–72. Wetlands wetl-29-02-04.3d 7/4/09 11:14:23 463 Cust # 08-59 463 McCune, B. and M. J. Mefford. 1997. PC-ORD, Multivariate Analysis of Ecological Data, Version 3.0. MjM Software Design, Gleneden Beach, OR, USA. Mitsch, W. J., C. L. Dorge, and J. R. Wiemhoff. 1979. Ecosystem dynamics and a phosphorus budget of an alluvial cypress swamp in southern Illinois. Ecology 60:1116–24. Mitsch, W. J. and J. G. Gosselink. 1993. Wetlands. Van Nostrand Reinold, New York, NY, USA. Nakamura, F., T. Sudo, S. Kameyama, and M. Jitsu. 1997. Influences of channelization on discharge of suspended sediment and wetland vegetation in Kushiro Marsh, northern Japan. Geomorphology 18:279–89. Odum, E. P. 1979. Ecological importance of the riparian zone. p. 2–4. In B. R. Johnson and J. F. McCormick (eds.) Strategies for Protection of Floodplain Wetlands and Other Riparian Ecosystems. U.S. Forest Service General Technical Report WO-12, USA. Orr, C. H., E. H. Stanley, K. A. Wilson, and J. C. Finlay. 2007. Changes in denitrification following reintroduction of flooding in a leveed Midwestern floodplain. Ecological Applications 17:2365–76. Oswalt, S. N. and S. L. King. 2005. Channelization and floodplain forests: Impacts of accelerated sedimentation and valley plug formation on floodplain forests of the Middle Fork Forked Deer River, Tennessee, USA. Forest Ecology & Management 215:69–83. Palmer, M. A., E. S. Bernhardt, J. D. Allan, P. S. Lake, G. Alexander, S. Brooks, J. Carr, S. Clayton, C. N. Dahm, J. Follstad Shah, D. L. Galat, S. G. Loss, P. Goodwin, D. D. Hart, B. Hassett, R. Jenkinson, G. M. Kondolf, R. Lave, J. L. Meyer, T. K. O’Donnell, L. Pagano, and E. Sudduth. 2005. Standards for ecologically successful river restoration. Journal of Applied Ecology 42:208–17. Pezeshki, S. R. 1994. Plant response to flooding. p. 280–321. In R. E. Wilkinson (ed.) Plant Environment Interactions. Marcel Dekker, Inc., New York, NY, USA. Pezeshki, S. R. and P. H. Anderson. 1997. Responses of three bottomland hardwood species with different flood tolerance capabilities to various flooding regimes. Wetlands Ecology and Management 4:245–56. Piégay, H. 1997. Interactions between floodplain forests and overbank flows: data from three piedmont rivers of southeastern France. Global Ecology and Biogeography Letters 6:187–96. Richter, B. D. and H. E. Richter. 2000. Prescribing flood regimes to sustain riparian ecosystems along meandering rivers. Conservation Biology 14:1467–78. Robertson, A. I., P. Bacon, and G. Heagney. 2001. The responses of floodplain primary production to flood frequency and timing. Journal of Applied Ecology 38:126–36. Robertson, P. A., G. T. Weaver, and J. A. Cavanaugh. 1978. Vegetation and tree species patterns near the northern terminus of the southern floodplain forest. Ecological Monographs 48:249–67. Robbins, C. H. and A. Simon. 1983. Man-induced channel adjustment in Tennessee streams. U.S. Geological Survey Water-Resources Investigations Report Vol. 82-4098, USA. SAS. 2003. SAS 9.1. SAS Institute Inc., Cary, NC, USA. Shankman, D. 1996. Stream channelization and changing vegetation patterns in the U.S. coastal plain. Geographical Review 86:216–32. Shankman, D. and T. B. Pugh. 1992. Discharge response to channelization of a coastal plain stream. Wetlands 12:157–62. Shankman, D. and S. A. Samson. 1991. Channelization effects on Obion River flooding, western Tennessee. Water Resources Bulletin 27:247–54. Sharitz, R. R. and W. J. Mitsch. 1993. Southern floodplain forests. p. 311–72. In W. H. Martin, S. G. Boyce, and A. C. Echternacht (eds.) Biodiversity of the Southeastern United States: Lowland Terrestrial Communities. John Wiley & Sons, Inc., New York, NY, USA. 464 WETLANDS, Volume 29, No. 2, 2009 Simon, A. S. and C. R. Hupp. 1992. Geomorphic and vegetative recovery processes along modified stream channels of West Tennessee. U.S. Geological Survey Report 91-502, USA. Sparks, R. E. 1995. Need for ecosystem management of large rivers and their floodplains. Bioscience 45:168–82. Steed, R., J. Plyler, and E. Buckner. 2002. The bottomland hardwoods of the Hatchie River, the only unchannelized Mississippi River tributary. p. 543–47. In K. W. Oucalt (ed.) Proceedings of the eleventh biennial southern silvicultural research conference. US Department of Agriculture, Forest Service, Southern Forest Experiment Station, Asheville, NC, USA. Gen. Tech Report SRS-48. Steiger, J., E. Tabacchi, S. Dufour, D. Corenblit, and J. -L. Peiry. 2005. Hydrogeomorphic processes affecting riparian habitat within alluvial channel-floodplain river systems: a review for the temperate zone. River Research and Applications 21:719–37. Stoddard, J. L., D. P. Larsen, C. P. Hawkins, R. K. Johnson, and R. H. Norris. 2006. Setting expectations for the ecological condition of streams: the concept of reference condition. Ecological Applications 16:1267–76. Tabacchi, E., L. Lambs, H. Guilloy, A. -M. Planty-Tabacchi, E. Muller, and H. Décamps. 2000. Impacts of riparian vegetation on hydrological processes. Hydrological Processes 14:2959–76. Wetlands wetl-29-02-04.3d 7/4/09 11:14:23 464 USDA. 1965. Soil survey of Dyer County, Tennessee. USDA Soil Conservation Service, Series 1962 No. 5, Washington, DC, USA. USDA. 1978. Soil survey of Madison County, Tennessee. USDA Soil Conservation Service, US Department of Agriculture, Washington, DC, USA. Valett, H. M., M. A. Baker, J. A. Morrice, C. S. Crawford, M. C. Molles, Jr., C. N. Dahm, D. L. Moyer, J. R. Thibault, and L. M. Ellis. 2005. Biogeochemical and metabolic responses to the flood pulse in a semiarid floodplain. Ecology 86:220–34. Van Looy, K., O. Honnay, B. Bossuyt, and M. Hermy. 2003. The effects of river embankment and forest fragmentation on the plant species richness and composition of floodplain forests in the Meuse Valley, Belgium. Belgium Journal of Botany 136:97–108. Ward, J. V. and K. Tockner. 1983. Biodiversity: toward a unifying theme for river ecology. Freshwater Biology 46:807–19. Ward, J. V., K. Tockner, D. B. Arscott, and C. Claret. 2002. Riverine landscape diversity. Freshwater Biology 47:517–39. Warner, R. and P. L. Chesson. 1985. Coexistence mediated by recruitment fluctuations: a field guide to the storage effect. The American Naturalist 125:769–87. Manuscript received 6 March 2008; accepted 21 January 2009. Cust # 08-59
© Copyright 2026 Paperzz