Aboveground Net Primary Productivity in a Riparian Wetland Following Restoration of Hydrology

This research presents the initial results of the effects of hydrological restoration on forested wetlands in the Mississippi alluvial plain near Memphis, Tennessee. Measurements were carried out in a secondary channel, the Loosahatchie Chute, in which rock dikes were constructed in the 1960s to keep most flow in the main navigation channel. In 2008–2009, the dikes were notched to allow more flow into the secondary channel. Study sites were established based on relative distance downstream of the notched dikes. Additionally, a reference site was established north of the Loosahatchie Chute where the dikes remained unnotched. We compared various components of vegetation composition and productivity at sites in the riparian wetlands for two years. Salix nigra had the highest Importance Value at every site. Species with minor Importance Values were Celtis laevigata, Acer rubrum, and Plantanus occidentalis. Productivity increased more following the introduction of river water in affected sites compared to the reference. Aboveground net primary productivity was highest at the reference site (2926 ± 458.1 g·m−2·year−1), the intact site; however, there were greater increase at the sites in the Loosahatchie Chute, where measurements ranged from 1197.7 ± 160.0 g m−2·year−1·to 2874.2 ± 794.0 g·m−2·year−1. The site furthest from the notching was the most affected. Pulsed inputs into these wetlands may enhance forested wetland productivity. Continued monitoring will quantify impacts of restored channel hydrology along the Mississippi River.


Introduction
The Mississippi River is the fourth longest and tenth largest river in the world. The Mississippi River Basin (MRB) is the fourth largest watershed in the world, draining more than 3,220,000 km 2 covering about 40% of the landmass in the continental United States. The Lower Mississippi River sub-basin is characterized by low, flat topography, alluvial soils in a wide flood plain, relatively high rainfall, and high water tables [1]. The Mississippi Alluvial Valley (MAV) once contained nearly 10 million ha of bottomland hardwood forests, representing the largest area of bottomland hardwood wetlands in North America [2]. Historically, black willow stands in the rive floodplain occurred in low moist spots in parts of the alluvial valley near river channels [3]. Settlement along the Mississippi River led to intensive harvesting of forests, isolation of much of the original floodplain with levees, the development of large-scale agricultural practices, urbanization, and the draining of wetlands [4]. The Mississippi River and Tributaries Project included the levee system and many stone dikes in order to divert most river flow from secondary channel complexes into the main navigation channel [1]. Dams and reservoirs, especially in the Missouri basin, resulted in the retention of sediment stored in pools behind the dams and in the fields of the winged dams, which enhances the flow magnitude in the of the Loosahatchie Chute and determine if the reintroduction of flood pulses would impact species composition and the changes in plant production. This study was part of a larger study that involved measurement of nutrient dynamics, wetland productivity, and greenhouse gas emissions at three sites in the MRB [36]. The specific objective was to determine spatial and temporal patterns of productivity and structure of the forest in response to restored hydrology. We hypothesized that aboveground net primary productivity would increase at the restored sites. Increased productivity would indicate that the restored hydrology was positively impacting wetland function.
Biology 2016, 5, 10 3 of 14 in plant production. This study was part of a larger study that involved measurement of nutrient dynamics, wetland productivity, and greenhouse gas emissions at three sites in the MRB [36]. The specific objective was to determine spatial and temporal patterns of productivity and structure of the forest in response to restored hydrology. We hypothesized that aboveground net primary productivity would increase at the restored sites. Increased productivity would indicate that the restored hydrology was positively impacting wetland function.

Study Area
Loosahatchie Bar is located on the west bank of the Mississippi River, opposite Memphis, Tennessee, between river miles 736.5 and 742.8 on the border of Shelby County, Tennessee, and Crittenden County, Arkansas ( Figure 1, note: river miles are the official designation of distance along the Mississippi and are used here so that the location of our study can easily be located in the context of the larger system). Stone dikes were constructed during the 1960s by the United States Army Corps of Engineers (USACE) to keep water from the Mississippi River from flowing through the Redman Point Loosahatchie Bar secondary channel complex (herein referred to as the Loosahatchie Chute). In an attempt to restore the area, twelve notches were excavated by the USACE from nine of the existing dikes at Loosahatchie Chute during October 2008-February 2009 ( Figure 1). Dikes measure between 100 and 600 m in length, 7.5 to 60 m wide at top, and 20 to 64 m wide at the bottom. Each notch was 0.9

Study Area
Loosahatchie Bar is located on the west bank of the Mississippi River, opposite Memphis, Tennessee, between river miles 736.5 and 742.8 on the border of Shelby County, Tennessee, and Crittenden County, Arkansas ( Figure 1, note: river miles are the official designation of distance along the Mississippi and are used here so that the location of our study can easily be located in the context of the larger system). Stone dikes were constructed during the 1960s by the United States Army Corps of Engineers (USACE) to keep water from the Mississippi River from flowing through the Redman Point Loosahatchie Bar secondary channel complex (herein referred to as the Loosahatchie Chute). In an attempt to restore the area, twelve notches were excavated by the USACE from nine of the existing dikes at Loosahatchie Chute during October 2008-February 2009 ( Figure 1). Dikes measure between 100 and 600 m in length, 7.5 to 60 m wide at top, and 20 to 64 m wide at the bottom. Each notch was 0.9 to 3.3 m deep. These notches restored flow in more than 17 km of secondary channel in the Loosahatchie Chute ( Figure 2). This effort was funded by the US Fish and Wildlife Service Fish Passage Program, the Audubon Society, and non-governmental conservation organizations in order to restore flow to habitats of two federally listed species (S. albus and S. antillarum athalassos), as well as other wildlife.
Biology 2016, 5, 10 4 of 14 to 3.3 m deep. These notches restored flow in more than 17 km of secondary channel in the Loosahatchie Chute ( Figure 2). This effort was funded by the US Fish and Wildlife Service Fish Passage Program, the Audubon Society, and non-governmental conservation organizations in order to restore flow to habitats of two federally listed species (S. albus and S. antillarum athalassos), as well as other wildlife.

Site Description
Study site selection and data collection began February 2009. Sites were selected in relation to the Redman Point-Loosahatchie Bar Environmental Restoration Project ( Figure 1; Table 1). Three study sites (Near, Mid and Far) were established along the Loosahatchie Chute, and a reference site (Ref) was positioned north of the Redman point dikes. All sites were riparian forest composed primarily of black willow (Salix nigra Marsh.; Figure 3). Three 25 × 25 m plots were established at each site for measurement of net primary productivity. The number and size of the plots were established to quantify biomass change as a representation of the area being measured. It is important to note that logistics and access constrained the selection of sites. During high discharge of the river, high river stages and strong current limit access to many areas due to safety concerns. By contrast, during low flow, boat access is limited due to low water and it was practically impossible to go to many areas on foot. For these reasons, the reference site was less directly impacted by river flow than the study sites.
The river stage for the Mississippi River at Memphis, Tennessee, was measured at the Weather Bureau Gage, where the National Weather Stage Gage height of 0 m is measured at an elevation of 56.1 m; this is based on older USGS topographic maps and NGVD29 benchmarks.

Site Description
Study site selection and data collection began February 2009. Sites were selected in relation to the Redman Point-Loosahatchie Bar Environmental Restoration Project ( Figure 1; Table 1). Three study sites (Near, Mid and Far) were established along the Loosahatchie Chute, and a reference site (Ref) was positioned north of the Redman point dikes. All sites were riparian forest composed primarily of black willow (Salix nigra Marsh.; Figure 3). Three 25ˆ25 m plots were established at each site for measurement of net primary productivity. The number and size of the plots were established to quantify biomass change as a representation of the area being measured. It is important to note that logistics and access constrained the selection of sites. During high discharge of the river, high river stages and strong current limit access to many areas due to safety concerns. By contrast, during low flow, boat access is limited due to low water and it was practically impossible to go to many areas on foot. For these reasons, the reference site was less directly impacted by river flow than the study sites.
The river stage for the Mississippi River at Memphis, Tennessee, was measured at the Weather Bureau Gage, where the National Weather Stage Gage height of 0 m is measured at an elevation of 56.1 m; this is based on older USGS topographic maps and NGVD29 benchmarks.

Soil
Soil was classified according to the United States Department of Agriculture (USDA) National Resources Conservation Service Web Soil Survey. The Ref site soil data were collected 17 November 2008. The soil was classified as Bowdre silty clay and was frequently flooded. At a depth from 0 to 12.7 cm, the cation-exchange capacity (CEC) was 25 to 45 meq 100 g´1 with a pH of 5.6 to 7.3; at a depth from 12.7 to 43.2 cm, the CEC was 22-40 meq 100 g´1 with a pH of 5.6 to 7.3; at a depth from 43.2 to 106.7 cm, the CEC was 7.0 to 18 meq 100 g´1 with a pH of 6.1 to 8.4; and at a depth from 106.7 to 152.4 cm, the CEC is 5.0 to 18 meq 100 g´1 with a pH of 6.1 to 8.4. The soil at the Mid site was classified as Crevasse fine sand from data collected 23 September 2008. At a depth from 0 to 20.3 cm, the CEC was 1.8 to 6.8 meq 100 g´1 with a pH of 5.6 to 8.4; and at a depth from 20.3 to 152.4 cm, the CEC was 1.4 to 6.1 meq 100 g´1 with a pH of 5. 6-8.4. Soil at the Near and Far sites have not been surveyed.

Aboveground Productivity
Total aboveground NPP was calculated as the sum of stem growth and litterfall. Woody stem growth productivity was calculated using diameter at breast height (dbh) measurements of all trees with dbh greater than 10.0 cm. Trees were tagged with an aluminum identification tag positioned~1.3 m from the base of the tree ( Figure 5, left; [37][38][39]). Forest composition including species composition, percent cover, basal area and importance values were quantified at each study site as described below. Dbh was measured above and below (~5 cm) the nail holding the identification tag [40]. Diameter tape (d-tape) designed to read dbh from circumference was used to take measurements. Measurements of dbh began in the winter of 2009 and were taken yearly in the winters of 2010 and 2011. Woody biomass was calculated using species-specific allometric equations based on stem dbh (average of top and bottom measurement) as the independent variable (Table 2; [40,41]. Biomass was summed at each of the three plots at each study site and divided by the area of the plot (625 m 2 ) resulting in the biomass per unit area (kg¨m´2) for each plots for both years. Litterfall productivity was measured using five leaf litter collection traps placed randomly within each plot, totaling 15 traps per site (5 per subplot). The collection traps were constructed from 1.91 cm diameter polyvinyl chloride (PVC) pipe, measuring 0.25ˆ0.25 m, with a screened net bottom. Each trap was elevated (~3 m) above the ground to prevent inundation during high water periods ( Figure 5, right). The traps were installed in June 2009. During the high litterfall period from October to December, litter was collected approximately every two weeks until all leaves had dropped from the trees. Litter was not collected during the rest of the year due to spring floods washing away most of the traps and inaccessibility of the sites. Leaf litter was separated from woody litter and dried to constant mass at 65˝C. Final dry weights were weighed to the nearest 0.01 g.
Tree species composition analysis used equations 1-3 modified from Barbour et al. (1987) [42]. Basal area was defined as the trunk cross-sectional area of a given species (i.e., cm 2¨m´2 ). Importance Values equal the sum of relative density and relative dominance for a maximum value of 2.
Relative density " pindividuals of a speciesq{ptotal individuals of all speciesq (1) Relative dominance " ptotal basal area of a speciesq{ptotal basal area of all speciesq Importance " Relative density`Relative dominance

Data Analyses
To determine if restoring hydrology had increased productivity, NPP among sites within each sampling year was tested with general linear model for analysis of variance (ANOVA; [43]). Significant results were followed by Tukey's post-hoc analysis [44]. NPP for both 2010 and 2011 were compared at each site using paired sample t-test. Differences were considered significant at α < 0.05.

Results
Trees at the Near site were more exposed to wind compared to the other sites, which were protected by surrounding forests, contributing to stem breaks and canopy debris on the forest floor. Tree density at all sites was dominated by S. nigra, which also had the highest Importance Value for every site (Table 3). Species with minor Importance Values identified at some sites included sugarberry (Celtis laevigata Willd.), red maple (Acer rubrum L.), American elm tree (Ulmus americana L.), and sycamore (Platanus occidentalis L.) The Ref site had the lowest tree density of 296 trees ha −1 , however, these were the largest trees increasing in basal area from 3.26 cm 2 ·m −2 in 2009 to 3.47 cm 2 ·m −2 in 2010 and 3.67 cm 2 ·m −2 in 2011 ( Table 3). The relative density and relative dominance of S. nigra was 100% for 2009-2011, with an importance value of 2 for each year (Table 3).

Data Analyses
To determine if restoring hydrology had increased productivity, NPP among sites within each sampling year was tested with general linear model for analysis of variance (ANOVA; [43]). Significant results were followed by Tukey's post-hoc analysis [44]. NPP for both 2010 and 2011 were compared at each site using paired sample t-test. Differences were considered significant at α < 0.05.

Results
Trees at the Near site were more exposed to wind compared to the other sites, which were protected by surrounding forests, contributing to stem breaks and canopy debris on the forest floor. Tree density at all sites was dominated by S. nigra, which also had the highest Importance Value for every site (Table 3). Species with minor Importance Values identified at some sites included sugarberry (Celtis laevigata Willd.), red maple (Acer rubrum L.), American elm tree (Ulmus americana L.), and sycamore (Platanus occidentalis L.) The Ref site had the lowest tree density of 296 trees ha´1, however, these were the largest trees increasing in basal area from 3.26 cm 2¨m´2 in 2009 to 3.47 cm 2¨m´2 in 2010 and 3.67 cm 2¨m´2 in 2011 ( Table 3). The relative density and relative dominance of S. nigra was 100% for 2009-2011, with an importance value of 2 for each year (Table 3).  The Near site had nearly twice the density of trees than the Ref site and approximately a quarter of the basal area of trees growing at the other sites (Table 3). There was a decline in tree density at the Near site from 543 trees ha´1 in 2009 to 506 trees ha´1 in 2011. This particular site experiences direct winds from the west, which are strong enough to snap tree stems. Saplings were observed in the understory, which will grow to occupy the spaces opened from the damaged trees, however, the DBH measurements of these trees had not yet reached 10 cm. The basal area of S. nigra at the Near site increased from 0.63 cm 2¨m´2 in 2009 to 0.76 cm 2¨m´2 in 2010 and 0.86 cm 2¨m´2 in 2011 ( Table 3). The basal area of C. laevigata in 2009 was 0.01 cm 2¨m´2 and increased slightly up to 0.02 cm 2¨m´2 in 2010 and 2011. This decrease in density and basal area at the Near site was reflected in the stem growth measurements. The relative density of S. nigra was 98% and C. laevigata was 2% for 2009-2011 ( Table 3). The relative dominance of S. nigra was approximately 98% and C. laevigata was approximately 2% for all three years (Table 3). At the Near site S. nigra was the most important species (1.96) with C. laevigata being of minor importance (0.04) for 2009-2011 ( Table 3).
The Mid site maintained a density of 319 trees ha´1 during all three years ( Table 3). The basal area of S. nigra at the Mid site increased from 2.84 cm 2¨m´2 in 2009 to 3.02 cm 2¨m´2 in 2010 and 3.22 cm 2¨m´2 in 2011 ( Table 3). The basal area of P. occidentalis in 2009 was 0.01 cm 2¨m´2 for each year. The relative density of S. nigra was 98.3% and P. occidentalis 1.7% for 2009-2011 ( Table 3). The relative dominance of S. nigra was approximately 99.7% and P. occidentalis was approximately 0.3% for all three years (Table 3). At the Mid site S. nigra was the most important species (1.98) with P. occidentalis being of minor importance (0.02) for 2009-2011 ( Table 3).
Stem growth in 2010 ranged from 595.2˘237.6 g m´2¨year´1 at the Near site to 2105.1˘267.1 g m´2¨year´1 at the Ref site, however, differences were not significant between sites (F 3,8 = 2.714, p = 0.115; Table 4; Figure 6a). In 2011 stem growth ranged from 525.5˘195.8 g m´2¨year´1 at the Near site to 2383.4˘439.6 g m´2¨year´1 at the Ref site, and again there was not a significant difference between sites (F 3,8 = 2.667, p = 0.119). For all sites combined, there was no significant difference in stem growth increase in 2011 compared to 2010 (t 11 =´1.446, p = 0.176).  Table 4). Table 4. Summary of average woody stem growth, litterfall, and total aboveground net primary productivity (NPP) per plot for restored forested wetlands along the Mississippi River. Each value is the mean for the measurements (˘se). Sites with significant differences are identified by lowercase letters, a or b. This is calculated according to Tukey's post-hoc analysis. Differences were considered significant at α < 0.05.

Site
Plot  Figure 6c). The Near site had a 45% increase (239.2 g m´2¨year´1) in litterfall, contributing 56% of total NPP, compared to 42% contribution the previous year. Total NPP increased at the Near site by 169.5 g m´2¨year´1 despite the loss of 69.70 g m´2¨year´1 in stem productivity. The Mid site had increased production in both litterfall (51.5 g m´2¨year´1) and stem growth (197.7 g m´2¨year´1), and thus in total NPP (249.2 g m´2¨year´1). The Far site had the greatest changes in both litterfall (354.6 g m´2¨year´1) and stem growth (611.9 g m´2¨year´1), which is reflected in the 50% increase in the total NPP (966.5 g m´2¨year´1). The first year the Far site had 766.7 g m´2¨year´1 less total NPP than the Ref site, and following hydrologic restoration it was within 51.9 g m´2¨year´1 of total NPP compared to the Ref site. Total NPP at the Ref site increased the least (11.4%, 251.70 g m´2¨year´1), having an increase in stem growth of 278.3 g m´2¨year´1 and a decrease in litterfall of 26.6 g m´2¨year´1.

Discussion
The impact of the reintroduction of flood pulses to the functioning of the bottomland hardwood forests measured by species composition and the changes in plant production rates was detectable. The dominance of S. nigra at all sites in the present study reflects past plant community composition of similar locations at the same elevation and river stage. Historically, the Mississippi River floodplain has had black willow stands in low moist spots in parts of the alluvial valley near river channels, where water levels are 4.6 to 9.1 m above mean low water [3].
The frequency and duration of water availability to the Loosahatchie Chute increased due to notching and this was reflected by the response of tree productivity. Initially, measures of NPP indicated there were differences between the sites, with the greatest difference being between the Ref and Near sites. Following hydrologic restoration, the NPP was not significantly different between sites due to the increased productivity of all the restored sites. The restored study sites generally had a greater percentage increase compared to the reference site. This finding of increased productivity is important. Even though the Ref site had the greatest overall NPP for both years, it had the least amount of change. A previous study of Mississippi delta freshwater forested wetlands, which included S. nigra, receiving municipal effluent discharge had a measured reduction of an average of 79% of total Kjeldahl nitrogen and an average of 88.5% total phosphorus, where sites nearest the secondarily treated municipal effluent had increased productivity of vegetation compared to control sites [45]. In that study, the site closest to receiving the discharge had greater litterfall (717 g¨m´2¨year´1) and total NPP (1467 g m´2¨year´1) than a control site (412 and 714 g m´2¨year´1, respectively), and stem growth at all sites ranged between (302-776 g m´2¨year´1; [45]). In another previous study of seasonally flooded forests in Louisiana, bottomland hardwood woody productivity averaged 574 g¨m´2¨year´1 litterfall, 800 g¨m´2¨year´1 stem growth, and 1374 g¨m´2¨year´1 total NPP [46]. The present study had similar measurements in productivity as the forested wetlands in the delta region (Table 4). In general, it has been found that the most productive forested wetland sites are those with intermediate flooding [6,46]. In general, the values for total NPP from this study are in the upper range of those reported in the literature [6,46]. This is likely related to the seasonal flooding as well as nutrient supply from the river.
Measurements of the soil structure at the sites revealed that the restored sites had a higher proportion of silt than sand and clay particles [47]. The restored sites, beginning with the Near site, had greater reductions of total carbon, nitrogen, and phosphorus with decreased reduction down the channel [47]. The Ref site had a higher percentage of clay than other soil particles, and higher concentrations of all the nutrients measured [47]. Floodplain fertility depends on quality of deposited sediment and dissolved inorganic compounds, including plant nutrients, being replenished and deposited by inflowing river water [23].
The effectiveness of functions of restored hydrology that maintains restored wetland flora and soil carbon resources may take centuries to return to a healthy state [48]. As ecosystems age, they tend to grow in increasing orders of complexity. The pulsing of water and nutrients during seasonal flooding, followed by the released nutrients and dry period, promote the enhanced productivity of seasonally flooded wetlands [6]. Swamps with flowing water and seasonal flooding regimes have higher forest productivity [46]. Continued study of net primary productivity and tree communities at these sites will help quantify impacts of the restored flood pulse over time following restored channel hydrology. It is possible that a long-term evaluation of tree growth at these sites would indicate differences among sites. Restored hydrology to secondary channels can improve habitat for federally endangered and commercially important species, provide public benefits include boating, fishing, and birding, and improve water quality through wetland filtration. Flood pulses are an important, integral part of the Mississippi River floodplain, having a hydrological effect on the biota, improving system functioning [23]. These initial results indicate that restoration via notching of weirs is promising.
The functioning of the LC and the effectiveness of the restoration can be understood within the context of the River Continuum Concept (RCC; [33]) and the Flood Pulse Concept (FPC; [23,34,35]. In the RCC, longitudinal pattern changes from upstream to downstream are emphasized. Construction of rock weirs led to a reduction in upstream-downstream connectivity. This has impacted the productivity of the floodplain forest ecosystem as well as leading to negative impacts on the Pallid Sturgeon and the Interior Least Tern. Our results suggest that restoring the connectivity of the river by notching rock weirs led to greater increases in productivity as a result of enhanced exchange between the river and its floodplain as described by the FPC. Future measurements of the area will allow the determination of long-terms benefits of this river reconnection project.

Couclusions
The initial results of the effects of hydrological restoration on forested wetlands in the Mississippi alluvial plain near Memphis, Tennessee showed that enhancing the connectivity of the river by notching rock weirs led to greater increases in productivity of the study sites as a result of improved exchange between the river and its floodplain. The site furthest from the notching was the most affected. Pulsed inputs into these wetlands appeared to have enhanced forested wetland net primary productivity. However, continued monitoring is needed to quantify the long-term impacts of restored channel hydrology on forest productivity.