Interplay of water quality and vegetation in restored wetland plant assemblages from an agricultural landscape

Ecological Engineering 108 (2017) 255–262

Contents lists available at ScienceDirect

Ecological Engineering journal homepage: www.elsevier.com/locate/ecoleng

Interplay of water quality and vegetation in restored wetland plant assemblages from an agricultural landscape

MARK

⁎

Cory M. Shoemaker , Gary N. Ervin, Evelyn W. DiOrio Department of Biological Sciences, Mississippi State University, PO Box GY, Mississippi State, MS 39762 USA

A R T I C L E I N F O

A B S T R A C T

Keywords: Agricultural runoff Eutrophication Nitrogen Sediment Wetland restoration Plant assemblage

Water quality degradation from excessive fertilizer use and runoff is a worldwide problem. While this degradation impacts wetlands, these systems can also be a vehicle for water quality improvement. Restoration of wetlands in agricultural landscapes has recently increased, but little work has evaluated the relationship of plant assemblages and water quality parameters in restored, non-treatment wetlands. This study examines the impact of self-designed wetland plant assemblages on nitrogen and sediment dynamics. Thirty mesocosms were seeded with soil from restored wetlands and allowed to develop from the seed bank to emergent assemblages. During the 2015 growing season (seven to nine months after establishment), these assemblages were exposed to treatment loads of nitrogen and sediment, common stressors to wetlands in agricultural landscapes. Water samples were taken up to five days post-treatment in July and September to quantify interactions between the stressors and plant assemblages. Analyses showed plant assemblage identify was not structured by treatment, but by the site of soil origin. Treatment removal rates were influenced by total amount of the stressor present, with nitrogen removal rates being higher, in relative terms, in low nitrogen amended treatments. Additionally, plant quality, not quantity, was linked to nitrogen and sediment loss rates, and over time, elevated nitrogen and sediment loads were associated with decreased plant assemblage quality. This study demonstrates the ability of plants from restored wetlands to affect nutrient and sediment dynamics, with three significantly differing plant assemblages all exhibiting substantial nutrient and sediment reduction capacity. Nevertheless, we also found that in a relatively short time (seven to nine months) common stressors in agricultural settings can significantly impact wetland plant assemblage quality, and that this may be linked to a reduced capacity for nutrient and sediment removal.

1. Introduction Worldwide, aquatic environments have been degraded by land use change associated with agricultural production. In particular, excessive fertilizer use and subsequent runoff lead to eutrophication in freshwater and marine systems (Gilbert et al., 2014). Nutrient pollution is a primary cause of water quality degradation, with agricultural inputs acting as a major source of these nutrients (Ribaudo et al., 2001). Synthetic nitrogen (N) fertilizer entering waterways from agricultural field runoff is the largest input of N to the Mississippi River Basin (Howarth 2008) and is estimated to contribute approximately 50% of N entering the Gulf of Mexico (Ribaudo et al., 2001). This N input results in the formation of a hypoxic zone varying in size annually from about 7500 km2 to 17,500 km2 (Rabalais et al., 2001; http://www.gulfhypoxia.net). Water quality degradation is expedited and perpetuated through land conversion, with agricultural landscapes producing increased levels of surface runoff containing nutrients and sediments than undisturbed ⁎

soils (Woltemade, 2000). As primary aquatic systems, wetlands are disproportionately affected by these water quality degradations. Nutrient levels can affect plant assemblage dynamics on a species-by-species basis (Mahaney et al., 2004), but high levels of N seem to promote invasive establishment and/or decreased diversity (Beas et al., 2013). Extended periods of N loading, as well as periodic N pulses, contribute to wetland eutrophication and have the potential to alter plant community structure within aquatic systems (Thiébaut and Muller, 1999). While sedimentation rates vary between wetlands and geographic regions, anthropogenic impacts on sediment runoff are of major concern in agriculturally impacted wetlands (Richards et al., 1996) and affect seed bank and subsequent vegetative development (Peterson and Baldwin, 2004). While water quality degradation impacts wetlands, these systems are also seen as a vehicle for water quality improvement. Increased hydrologic retention time allows for alteration of wetland processes to

Corresponding author. E-mail addresses: [email protected] (C.M. Shoemaker), [email protected] (G.N. Ervin), [email protected] (E.W. DiOrio).

http://dx.doi.org/10.1016/j.ecoleng.2017.08.034 Received 2 May 2017; Received in revised form 15 August 2017; Accepted 30 August 2017 0925-8574/ © 2017 Published by Elsevier B.V.

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

occur, as wetlands develop unique suites of nutrient fixing microbes (Bourgues and Hart, 2007). Within wetlands, hydrophytic vegetation has significant and positive effects on water quality mitigation, with pollutant removal efficiencies varying among species (Brisson and Chazarenc, 2009). Although some uncertainty exists in causal mechanisms behind differences in species’ abilities to alter water quality, certain traits have been suggested to contribute to greater removal efficiency. Tanner (1996) found a linear correlation between plant biomass and removal of total N, while differences in growth form affect the location of nutrient uptake. Increased plant surface area allows for greater microbial attachment (Brisson and Chazarenc, 2009), however Pettit et al. (2016) suggest plant structural complexity was more influential on microbial biomass than surface area alone. Development of oxidized rhizospheres impacts soil biogeochemical environment, and subsequently, microbial communities. Aldridge and Ganf (2003) found differences in species’ abilities to alter oxidation-reduction (redox) potential in soils, with Typha domingensis increasing levels 218 mV above bare sediment compared to 41 mV for Bolboschoenus caldwellii, while Potamogeton crispus had no influence on redox potential in flooded soils. Even in heavily reduced conditions, the presence of plants increases soil microhabitat heterogeneity, engendering a diversity of processes conducive for nutrient removal (Brix, 1994; Sorrell and Armstrong, 1994; Jesperson et al., 1998; Shoemaker and Kröger, 2017). Effects of wetland vegetation on pollutant loads have been extensively noted and studied in wastewater treatment wetlands. In a metaanalysis conducted on vegetated treatment wetlands dosed with livestock effluent, median removal rates of 57% for total suspended sediments (TSS) were observed (Knight et al., 2000) while NO3− and PO43− removal rates of 86% and 56%, respectively, were observed in stormwater retention ponds planted with duckweed (Lemna minor) (Sims et al., 2013). Traditionally, studies on wetland plants’ ability to ameliorate water quality occur in wetlands planted with monotypic stands of vegetation or in microcosms (Reddy et al., 1983; Cronk and Fennessy, 2001; El-Sheikh et al., 2010). The applicability of such results to diverse plant assemblages is less well known, especially the impact of assemblages not directly managed for water quality improvement. The current study is unique in that it addresses effects of plant assemblages on water quality parameters in a replicated experimental study. In particular, a dearth of research exists on water quality improvement in non-flow through wetlands in an agricultural matrix, an area of critical need for mitigation of nutrient and sediment runoff from some 1.5 billion ha of crop-producing land worldwide (FAO, 2015). By determining whether and how self-designed plant assemblages, such as those resulting from passive restoration efforts, affect nutrient and sediment retention in non-treatment wetlands, best management techniques can be developed to ameliorate water quality conditions in areas other than urban and industrial settings. Our specific aim in this study was to measure the relative importance of plant assemblage composition on nutrient and sediment abatement in non-flow-through wetlands, such as those typically enrolled in wetland conservation programs in the United States. We hypothesized that plant species identities, rather than total cover, would be more strongly associated with water quality improvement, and that, as nutrient and sediment exposures increased, differences among plant species assemblage composition would become more evident

Bayou, Moon Bayou) enrolled in the Wetland Reserve Program (WRP) located in the Mississippi River Alluvial Valley (the Delta). Wetlands from which the soil was taken were greater than 10 years post-restoration, with clay soils of the Alligator-Sharkey-Dundee-Forestdale series. One soil sample from each wetland was placed in its own container, for a total of 10 soil samples per site, across three soil seed bank source wetlands (n = 30 total mesocosms). In addition to the 30, three more mesocosms were constructed with sand only and three constructed with sand and soil (but weeded regularly), to estimate the effect of soil and plants, respectively. Mesocosms were placed on leveled gravel and oriented in the same direction, with approximately 0.5 m of buffer space between neighboring mesocosms. Following their construction in December 2014, all mesocosms were flooded for the duration of the winter. Mesocosm hydroperiods were managed following a slow, midseason draw-down, common in WRP wetlands (LMJV, 2007), which occurred from 19 April 2015 to 5 May 2015. Water was removed by siphoning. For the time period covered in this study (July–September), water levels fluctuated between moist soil and inundated to a depth of 5 cm, with water levels at a depth of approximately 5 cm before treatment application (see following paragraph). Mesocosm plant assemblages were allowed to grow from seed banks existing in the soil from the three Delta wetlands. Shortly before treatment application (within a week, see below), percent cover was recorded for each plant species present, to quantify plant assemblage composition. Additionally, similar plant species cover data were collected from the three soil source wetlands, these data were collected four times, once each in May and August of 2014 and 2015, to allow for comparison of plant assemblages between mesocosms and the wetlands from which the seed banks were sourced. In the wetlands, plant assemblages were characterized using 50, 0.8m2 circular quadrats systematically spaced among 10 parallel transects 20 m apart from each other, with quadrat spacing of 20 m along the transect. Species identification followed Godfrey and Wooten (1981), and naming was consistent with the USDA-PLANTS database (plants. usda.gov). Nitrogen and sediment were selected as stressors for this study because they are common contaminants in wetlands on agricultural landscapes (Baker et al., 2016). Four treatments and a control were included in a randomized complete block design, with the three wetland soil sources serving as a blocking factor (Table 1). In addition to controls, which received no experimental nutrient or sediment inputs, we used two levels of N in the form of NO3− and two levels of sediment, in a factorial design. Levels of each contaminant were selected based on observed patterns in Delta wetlands. High and low NO3− loads of 20 g and 1.25 g per year were based on observations from agricultural runoff in Delta drainage ditches (Baker et al., 2016), while high and low sediment accumulation rates of 2 cm and 0.5 cm per year were calculated from rates observed in a Delta oxbow lake wetland over the last 200 years (Wren et al., 2008). Sediment accumulation rates were converted to loads by calculating dried sediment density and converting weight to Table 1 Experimental design of study. Thirty total mesocosms. N = Nitrogen, Sed. = Sediment Treatment

2. Methods This study took place at the Aquatic Plant Research facility located on the R.R. Foil Plant Science Research Center, at Mississippi State University (33° 28′N, 88° 46′W). Mesocosms for this study (30) were 568 L Rubbermaid stock tanks. Each mesocosm was constructed using 57 L of filler sand and 19 L of wetland soil, placed and spread out evenly over top of the sand. The wetland soil was collected on 1 November 2014 from three restored wetlands (Burrell South, Muddy 256

Block (wetland soil site)

(1) High N, High Sed.

(2) High N, Low Sed.

(3) Low N, High Sed.

(4) Low N, Low Sed.

Control

(1) Burrell S (n = 10) (2) Muddy Bayou (n = 10) (3) Moon Bayou (n = 10)

Y11 (n = 2)

Y12 (n = 2) Y22 (n = 2)

Y13 (n = 2) Y23 (n = 2)

Y14 (n = 2) Y24 (n = 2)

Y1c (n = 2)

Y32 (n = 2)

Y33 (n = 2)

Y34 (n = 2)

Y3c (n = 2)

Y21(n = 2)

Y31 (n = 2)

Y2c (n = 2)

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

of presence/absence values. When analyzing nutrient data, assessment of homogeneity of variance assumption was accomplished via the Fligner-Killen Test, a nonparametric test that is robust against departures from normality (Fligner and Killeen, 1976). When found to violate normality assumptions, data were transformed (logarithm, reciprocal). Repeated measures ANOVAs were performed through a multivariate approach using the lm() function to setup the repeated, random effects, and Anova() to conduct the tests. This test determines whether differences in nutrient and sediment reduction were observed between sampling periods (July and September) and assesses whether mesocosms returned to baseline (pre-treatment) levels for nutrient and sediment variables by the end of the observation period. Additionally, a repeated measures ANOVA was conducted to determine if baseline (pre-treatment) nutrient and sediment levels were similar between the start of the July and September sampling periods. In these tests, sampling periods were levels of the repeated measures factor, with individual mesocosms as the subjects. Effects of wetland soil site and treatment on reduction rates of nutrient and sediment were assessed via decay rates of these variables in the mesocosms. For each mesocosm, a linear model was fitted to the initial and final reading of nutrient and sediment concentrations, and the coefficient of this equation recorded as the decay rate with these values used to describe reduction rates (mg/day) under different conditions. A Multivariate Analysis of Variance (MANOVA) was performed for each study period, with dependent variables of nutrient and sediment decay rates and independent variables of treatment, wetland soil site, and their interaction. If no correlations between treatment and wetland soil site were present, then separate ANOVAs were conducted in conjunction with pairwise t-tests and least square means analysis. To assess the relationship between decay rates and plant quantity and quality, total cover and Floristic Quality Assessment Index (FQAI) (Lopez and Fennessy, 2002; Andreas et al., 2004) scores were calculated in each mesocosm for the two sampling periods. Mesocosm FQAI scores were calculated using coefficient of conservatism scores for north Mississippi wetland plants reported in Herman et al. (2006). If an observed species was not included in the north Mississippi list, its score for the southeastern United States, as reported in Gianopulos (2014), was used instead. Normality was assessed for FQAI scores and a two-way ANOVA performed to determine if treatment or wetland soil site affected plant community quality in July and September. The effect of plant quantity on decay rates and total cover and FQAI were calculated using a correlation model, as these variables are continuous. Average coefficient of conservatism (a component of FQAI which indicates the quality of a plant in a region) scores were calculated for all mesocosms over both sampling periods and for the three Delta WRP wetlands. Mean coefficient of conservatism scores were used for comparison instead of FQAI scores, as these control for the marked difference in size between mesocosms (∼1 m2) and wetlands (∼66,000 m2). Assumptions of normality were verified, and differences between mesocosm and wetland mean coefficients of conservatism were assessed using Welch’s t-test. Because the amounts of N and sediment varied considerably between high and low amendments, we also calculated a relative loss rate of nutrients or sediment within each of the two sampling periods. This was meant to standardize decay, as a function of total concentrations presents, and was calculated as = (ln(C1) − ln(C0)) , where C represents the t 1 − t0 concentration, and t represents time (days) which yields a relative change in concentration per day. Analyses of variance were then conducted on these data, and if significant differences found, Tukey’s HSD was used as a post hoc test.

Table 2 Amount of nitrogen and sediment applied per mesocosm per treatment and sampling period. N = NO3−, Sed = sediment. Amendment Sampling Period

High N (g)

Low N (g)

High Sed. (g)

Low Sed. (g)

July September

3.50 4.00

0.22 0.25

1417 945

708.75 236.25

volume. The treatments were applied throughout the year starting in May 2015, with the total divided proportionally to match seasonality of NO3− and sediment loading rates observed by Baker et al. (2016). Current treatments were applied on 14 July and 22 September 2015. Nitrogen was supplied in the form of potassium nitrate (KNO3−; Multi-K GG, Haifa Group, Matam-Haifa, Israel), and soil for sediment treatments was collected from Delta WRP sites in May 2015. Upon collection, the soil was autoclaved for eight hours to kill any seeds or propagules and ground (Royer Model 112, Royer Industries, Oshkosh WI, USA). The soil was then dried and weighed. The July amendments consisted of 3.50 and 0.219 g NO3− for high and low nitrogen treatments and 1410 and 708 g of soil for high and low sediment treatments. As the treatments were applied to reflect seasonality in nutrient and sediment rates, September amendments consisted of 4.00 and 0.250 g NO3− for high and low nitrogen treatments and 945 and 236 g of soil for high and low sediment treatments (Table 2). For application, N and sediment amendments were combined, blended together with water (∼2 L), and the resulting slurry spread evenly over the mesocosms’ surfaces via modified watering cans. Thus, four total treatment combinations and a control were possible. During treatment applications, control treatments received the same volume of water as was used in the slurries. To assess nutrient and sediment dynamics in the water column, 45 mL samples were taken throughout both study periods. Before treatment application, ambient water samples were collected with additional samples collected immediately after treatment application and 8, 32, and 56 h post-application for the July study period. In addition to water samples being collected at the same time post-treatment application as July, samples collected in September also included 80 and 104 h post-treatment samples. After collection, all water samples were acid preserved and stored at 1.1 °C in a refrigerator until processing. Total suspended solids (TSS) was determined by filtering 25 mL of water according to methods outlined in APHA (1998). Nutrient analysis samples were filtered through 0.45 μm cellulose membrane filters for NH4+ (ammonium), NO3−, and PO43− (collectively “nutrients”) using Lachat Flow Injection Analysis (FIA) (Lachat Instruments, Loveland CO, USA). 2.1. Statistical analysis All data analysis was conducted in Program R version 3.0.0 (R Core Team, 2013). To determine if unique plant communities were present during July and September sampling periods, Canonical Correspondence Analysis (CCA) was conducted using plant survey results from the two sampling periods separately. Canonical Correspondence Analysis was used to determine whether community structure was related to either nutrient treatment, wetland soil site, or the interaction of the two. Analysis followed guidelines outlined in McCune and Grace (2002) and was conducted in package vegan (Oksanen et al., 2015), with scaling optimized to sites, weighted average (WA) scores assessed, and convex hulls used to aid in visual interpretation. If distinct communities were present, an Indicator Species Analysis (ISA), a method for determining “the strength and statistical significance of the relationship between species occurrence/ abundance and groups of sites” (De Cáceres, 2013), was performed using package indicspecies (De Cáceres, 2013) with 999 permutations

3. Results During the July sampling period, a rain event (10.15 mm) occurred approximately 30 h post application. This precipitation deposited 257

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

1.13 mg/L NH4+, 0.01 mg/L PO43−, and 0.30 mg/L NO3−in addition to the amendment, calculated from the weeded sand and soil mesocosms. The spike associated with the rain event returned to baseline levels by the end of the sampling period (56 h post treatment). Since subsequent analysis used decay and relative loss rates, and was based on initial and final readings, this spike was not directly addressed in the following results, and it is assumed to have had equivalent effects across all mesocosms, regardless of treatment or soil source. Average air temperatures for July and September sampling periods were 29.5 °C and 22.4 °C, respectively. Beginning in June 2015, insects were observed in increasing numbers in the mesocosms. In particular, they were associated with Ammania coccinea, causing substantial damage to plants. As a result, beginning in July 2015, cyfluthrin (Bayer Advanced Powerforce Multi-Insect Killer, SBM Life Science Corporation, Cary NC, USA) was applied periodically via backpack sprayer, in equal amounts to all mesocosms, for the remainder of the growing season (five applications made overall). During the July sampling, water levels in a high nitrogen, high sediment tank decreased to a level where a water sample could not be obtained for the 56 h post treatment sampling period. To ensure a balanced design, missing data were generated using Bayesian linear imputation (Donders et al., 2006) in package mice (van Buuren and Groothuis-Oudshoorn, 2011). This method creates multiple imputations based off observed values (via Gibbs sampling), decreasing bias associated with other techniques (Donders et al., 2006), and assumes data are missing at random.

3.2. Baseline measurements No differences in nutrient reduction were found between sampling periods, but sediment removal rates were significantly higher during the July sampling period than September (F = 28.30, df = 29, P < 0.001; Fig. 2). Except for PO43−in September (F = 2.73, df = 29, P = 0.11), concentrations did not return to their pre-amendment levels by the end of either sampling period. Baseline concentrations between July and September sampling periods were similar for NH4+ and NO3−, but not for PO43− (F = 6.93, df = 29, P = 0.013) or sediment (F = 4.56, df = 29, P = 0.041). Baseline levels of PO43− were higher in September, and baseline sediment levels were higher in July. 3.3. Treatment and wetland soil site decay functions The MANOVA of decay rate coefficients showed no significant effects of wetland soil site nor treatment on water quality decay rates for July. However, during September strong effects were observed for wetland soil site on NO3− decay (F = 4.32, df = 2, P = 0.033) and treatment on PO43− decay (F = 7.17, df = 4, P = 0.002). No significant interactions were found between treatment and wetland soil site for July (F = 1.42, df = 32, P = 0.122) nor September (F = 1.07, df = 32, P = 0.401), which indicates the use of two-way ANOVAs is appropriate. When analyzing the Least Square Means, a general pattern emerged, with control treatments having the lowest loss rates, which is consistent for all independent variables except NO3−; this is explored further in the Discussion. Additionally, while not significant, mesocosms seeded with soil from the Burrell South site consistently had the lowest values for loss rates across both sampling periods and all independent variables, except for PO43− (data not shown).

3.1. Assemblage composition Plant species assemblage structure was strongly correlated with wetland soil site in July and September (Fig. 1). Both CCAs have 23 unconstrained eigenvalues, with 37.8% and 38.6% of variation explained in the July and September ordination, respectively. The first three axes accounted for 85.7% of the constrained variation in the July plant surveys, while the first three axes accounted for 82.3% of the constrained variation in the September plant surveys. Indicator Species Analysis associated 17 species to three unique groups by wetland soil site and five species to two groups in July and 15 to unique groups and five species to two groups in September (Table 3).

3.4. Treatment and relative loss rates Because control treatments had inherently lower concentrations of each water quality variable of interest (as a result of there being no additions in the control treatment), we examined relative loss rates to control for this potentially confounding factor. No significant difference in relative loss rate was attributed to wetland soil site identity. Fig. 1. Canonical Correspondence Analysis (CCA) showed that plant assemblages in the mesocosms during both July and September were strongly correlated with the wetland from which soil seed banks originally were collected.

258

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

Table 3 Mesocosm indicator species analysis. Specificity indicates percent of a given species’ occurrences that occur in mesocosms from a given site. Fidelity indicates percent of mesocosms from that site which contain a given species. July

Sept

Site

Species

Specificity

Fidelity

P-value

Species

Specificity

Fidelity

P-value

Burrell S

Ludwigia peploides Juncus acuminatus Echinochloa sp. Juncus diffusissimus Ipomoea wrightii

0.940 0.890 0.850 0.760 0.920

0.680 0.400 0.360 0.320 0.200

0.001 0.001 0.001 0.006 0.033

Rotala ramosior Juncus diffusissimus Juncus acuminatus Ipomoea wrightii Rhycospora corniculata

0.920 0.910 1.000 0.890 1.000

0.800 0.400 0.200 0.200 0.160

0.001 0.001 0.009 0.034 0.037

Moon

Echinochloa crus-galli Sagittaria lancifolia Lindernia dubia Echindorus cordifolius Typha latifolia

0.690 1.000 0.730 1.000 0.700

0.960 0.520 0.520 0.240 0.320

0.001 0.001 0.003 0.005 0.021

Lindernia dubia Panicum hians Sagittaria lancifolia Typha latifolia Echinodorus cordifolius

0.820 0.750 0.950 0.710 1.000

0.560 0.600 0.360 0.400 0.200

0.001 0.001 0.001 0.004 0.010

Muddy

Sesbania herbacea Echinochloa muricata Lindernia anagallidea Digitaria sanguinalis Polygonum hydropiperoides Cyperus pseudovegetus Iva annua

0.700 0.820 0.750 1.000 0.780 0.920 0.850

0.920 0.760 0.680 0.480 0.480 0.320 0.320

0.001 0.001 0.001 0.001 0.001 0.003 0.004

Leptochloa fusca Digitaria sanguinalis Sesbania herbaceae Polygonum hydropiperoides Iva annua

0.770 1.000 0.640 0.890 0.800

0.920 0.640 0.960 0.640 0.360

0.001 0.001 0.001 0.001 0.002

Fig. 2. Comparison of decay rates for nutrient and sediment amendments during the July and September sampling periods showed that only sediment decay differed significantly between the two time periods. Note that the nitrate y-axis has a different scale.

Treatment, however, had a significant effect on relative loss rate of all nutrient and sediment variables except for PO43− in the July sampling and significantly affected all water quality variables in September (Fig. 3). Fig. 3. Examination of relative loss rates of nutrients and sediment during the two time periods, among treatments, indicated that generally similar patterns were observed during both months. In July, treatment had a statistically significant effect on relative loss rate overall, but the Tukey HSD test was unable to distinguish among the different treatment groups. Otherwise, similar lowercase letters group statistically indistinguishable treatments (Tukey HSD test) within each category of nutrient or sediment being measured.

3.5. Plant species quality vs. quantity In July, mesocosms seeded with Burrell South soil had a significantly lower FQAI score than those seeded with either Muddy or Moon Bayou soil (F = 5.36, df = 2, P = 0.012), but this trend did not continue during the September sampling period (F = 1.06, df = 2, P = 0.363; Fig. 4). In September, treatment rather than wetland soil site, significantly affected FQAI (F = 3.93, df = 4, P = 0.0142), with tanks receiving the combined high N and high sediment treatment having lower FQAI scores than the other treatments and control (Fig. 4). No significant difference in mean coefficients of conservatism (CC) was observed between mesocosms and their respective Delta wetland sites (t = 0.128, df = 3.04, P = 0.906), with an average CC of 2.77 for the mesocosms and 2.78 for the Delta wetland sites (Fig. 5). Differences in trends between FQAI and mean CC can be attributed to index calculation. The FQAI does not account for nonnative species in its total number of species while mean coefficient of conservatism does, and difference in area sampled can affect FQAI (Ervin et al., 2006). A positive correlation was found between total cover and sediment

removal rates (P = 0.030), but no significant relationship was found between total plant cover and nutrient reduction rates. Total cover was significantly affected by wetland soil site in July (F = 5.20, df = 2, P = 0.014), with Burrell South mesocosms having the lowest values for total cover. In September, however, treatment, not site, affected total cover (F = 3.13, df = 4, P = 0.034). High N, high sediment treatment mesocosms had significantly greater total cover than control mesocosms, with the other treatments not significantly different from high N, high sediment or control treatments.

259

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

Fig. 5. Mean coefficients of conservatism in mesocosms (July and September), with the exception of July Burrell South mesocosms assemblages, were similar to values observed in the soil seed bank source wetlands. Values for the wetland sites are mean values calculated from on-site plant surveys during May and July–August 2015.

understanding assembly processes in these wetlands. While decay rates for nutrient and sediment were not significant among treatments or wetland soil sites (except for NO3− and PO43− among sites and treatments, respectively in September), trends linked to treatment and wetland soil site emerged. In general, control mesocosms had lower decay rates than mesocosms amended with the other four treatments. This trend shows that as amendments were added to these wetlands, the response dynamics were altered from those of the controls. This trend of increasing decay rates in non-control mesocosms does not hold for NO3− dynamics, perhaps suggesting these decay rates may be more assemblage, rather than treatment driven. Assemblages in mesocosms seeded with Burrell South soil had consistently lower decay rates than those from Muddy or Moon Bayou for NH4+, NO3−, and sediment in July and September. While not significant, plant assemblages from these sites altered removal dynamics, suggesting plants have the ability to change nutrient and sediment removal patterns. This is not surprising as Yao et al. (2011) found wetland plant species differed in their ability to remove NO3−, PO43− and NH4+ in an experimental setting conducted in beakers. Functional traits, such as the ability to fix N (Trowbridge, 2007) or morphological traits may also affect nutrient uptake rates/dynamics and subsequently plant assemblage structure (Fujita et al., 2010). Indicator species may have an influence on decay rates, as assemblages from Muddy and Moon Bayou had higher rates of decay for most nutrient and sediments compared to Burrell South. Of note, Sesbania spp. (Yao et al., 2011), Typha spp. (Peterson and Teal, 1996), and Sagittaria spp. (Moore et al., 2013) have high rates of N uptake and are indicators of Muddy and Moon Bayou. Ammania coccinea, an annual broadleaf, was strongly linked to Muddy and Moon Bayou seeded mesocosms in July, but not in September. This relationship may have broken down as a result of heavy insect grazing on A. coccinea, which may have influenced overall assemblage removal rates. Insect herbivory has previously been found to alter N dynamics in old-field systems (Uriarte, 2000) and thus, targeted grazing of dominant plants may have affected nutrient decay rate dynamics in the current study. Additionally, the presence of nitrogen-fixing Sesbania herbacea may have

Fig. 4. Mean FQAI values for July were influenced by soil source (top panel), whereas FQAI in September was influenced by nutrient and sediment additions, specifically the combination of high sediment and high nitrogen addition (bottom panel).

4. Discussion Plant assemblages were clearly associated with strong fidelity to wetland soil source. These assemblages retained their unique character throughout the study period, with the same patterns holding 12 months after the conclusion of this study (Shoemaker, unpublished data). The lack of assemblage convergence based off treatments could be attributed to the influence of priority effects or stochastic, rather than deterministic assembly (Young et al., 2001). The rather limited timeframe of the present study (three months) could also have affected results, since treatment impacts may not have manifested themselves within this time period. Management of many WRP wetlands, from which the soil for this study was obtained and hydrology mimicked, recommends regular disturbance at periods of two to three years to insure assemblages are kept in an early successional sere (LMVJV, 2007). Therefore, short-term dynamics are important both to the landowner and for 260

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

performance in wetlands (Spayd et al., 2013.), this work is usually conducted either in monotypic, planted tanks (e.g., Iamchaturapatr et al., 2007) or at the landscape level (e.g., El-Sheikh et al., 2010). The current study suggests that in agricultural, non-treatment wetlands, plant quality is more influential on decay rate coefficients than total plant quantity. In July, Burrell South had the lowest rate FQAI scores and lowest decay rates. This relationship potentially indicates quality of plants may drive nutrient reductions. Conversely, in September, treatment type significantly affected FQAI scores, with the high N, high sediment treatment linked to a decrease in FQAI, suggesting that over time, nutrient load may drive quality of plants within wetlands. This dynamic has been observed in restored wetlands, with decreases in water quality resulting in the establishment of invasive species and overall degradation of plant assemblages (Brooks et al., 2005). Findings from the present study indicate that vegetation quality initially had a positive influence on water quality parameters, but continued high loading resulted in assemblage quality degradation. The similarity of mean coefficient of conservatism between mesocosms and wetland soil sites allows for tentative extrapolation of these results to field-level settings, but further research in non-wastewater wetlands is needed. Results from this study demonstrate the ability of wetland plants to affect nutrient uptake dynamics, but total removal rates seem to be influenced by the total amount of nutrients/sediment present, especially in the case of NO3−. Consideration should be given not only to the plant cover, but also to the quality of these plants. Continual rates of high nutrient and sediment inputs lead to degradation of plant communities over a relatively short period of time. Plant assemblages in non-treatment wetlands can improve water quality, but long-term loading will lead to alterations of assemblages. This may affect the attainment of the primary management goals. This research is limited in that these are non-flow through wetlands, which may lead to altered nutrient and sediment patterns. Further work is needed to determine the impact of specific assemblages on water quality parameters in fieldscale settings.

modified N dynamics (Trowbridge, 2007). A closely related species, Sesbania cannabina, negatively affected PO43− uptake rates when compared to eight other wetland species (Yao et al., 2011). If S. herbacea has the same effect, this may explain why Burrell South has elevated rates of PO43− removal compared to the other wetland soil sites. Increasing surface area/structural complexity may increase the ability of sediments to precipitate out of the water column (Brisson and Chazarenc, 2009; Pettit et al., 2016). Greater sediment decay rates were found in Muddy and Moon Bayou mesocosms, which are characterized by several broadleaf species, compared to Burrell South, which had a greater portion of gramminoid indicator species. The ability of plants to reduce suspended sediment is well known in lotic systems (Madsen et al., 2001), but results from this study indicate vegetation may also affect suspended sediment reduction in lentic systems as well. When examining relationships between relative loss rates of water quality parameters and independent variables of wetland soil site and treatment, wetland soil site did not influence rates, while treatment load influenced rates across a variety of variables. Sediment removal rates were always lowest in controls compared to other treatments. This is not altogether unexpected, as an increase in sediment input from control levels can be accounted for via precipitation processes alone. While NO3− loss rates were influenced by treatment, plant response differed between treatment and sampling period. In particular, plant communities seemed to reach a saturation point in the amount of NO3− they are able to remove, with high N amended mesocosms having similar uptake rates to control mesocosms and lower rates of uptake than low N amended mesocosms (Tanner, 1996). Ammonium followed a similar, though not significant, pattern to NO3−, suggesting that N species may act similarly in these systems. Nitrate uptake reaches asymptotic levels in wastewater treatment wetlands, with higher nutrient removal rates usually associated with lower loading rates, particularly in young treatment wetlands (Gottschall et al., 2007). Both Gottschall et al. (2007) and the present study indicate that in agricultural, non-treatment wetlands, upper limits of N uptake occur at potentially lower concentrations than in high-load treatment systems, with these trends increasing in strength as wetlands mature. Choices in wetland management actions have the potential to alter plant assemblage dynamics, nutrient cycling, and their interaction. The timing, frequency, intensity, and duration of disturbance may shift assemblages between those structured more strongly by competition to more dispersal-tolerant species (Kross et al., 2008). Draw-down timing may also affect water quality parameters. While this study used a slow, midseason draw-down, shifting draw-down timing to earlier or later in the season will affect abiotic factors such as soil and water temperature and evapotranspiration, while increasing or decreasing the draw-down rate may affect germination, root development, and seed production (Fredrickson, 1991). In addition, water level management over the growing season affects survivability of plants and alters nutrient cycling. Wetter conditions, such as those in this study, may yield assemblages with higher frequencies of facultative wetland or obligate wetland plants compared to drier sites. Additionally, the timing and duration of soil drying and wetting cycles affect N and P dynamics, especially N mineralization and nitrification and P sorption/desorption to soil particles (Reddy and DeLaune, 2008). Uncertainty exists in the role of plant quality and quantity in wetland water quality dynamics. The presence of plants increases nutrient uptake rates over bare soil (Cronk and Fennessy, 2001), with widespread differences in uptake efficiencies between macrophyte species (Brisson and Chazarenc, 2009). Mixed-species stands of vegetation may be most effective for N and P mitigation, as species differ in N and P mitigation rates (Moore et al., 2013). Plant cover/biomass influences water quality, but specific dynamics seem to be system and species specific (Brix 1994; Thullen et al., 2002). While types and composition of plant assemblages have been shown to impact nutrient mitigation

Acknowledgements The authors thank Dr. Charles Bryson and John MacDonald for their assistance in identifying unknown plants. We also thank Gray Turnage and the Mississippi State University (MSU) Geosystems Research Institute for use of their facilities and Dr. Gary Lawrence and the Department of Plant and Soil Science for use of their soil grinder and autoclave. The authors greatly appreciate Dr. Matthew Moore providing a very helpful critique of an early version of this manuscript. Additionally the technical expertise and assistance provided by the MSU Water Quality Lab was invaluable. This research was supported by the Mississippi Water Resource Research Institute (grant number G11AP20088), as well as the MSU Department of Biological Sciences. References APHA, 1998. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Washington D.C., USA. Aldridge, K.T., Ganf, G.G., 2003. Modification of sediment redox potential by three contrasting macrophytes: implications for phosphorus adsorption/desorption. Mar. Freshwater Res. 54, 87–94. Andreas, B.K., Mack, J.J., McCormac, J.S., 2004. Floristic Quality Assessment Index (FQAI) for Vascular Plants and Mosses for the State of Ohio. Ohio Environmental Protection Agency, Division of Surface Water, Wetland Ecology Group, Columbus, OH, USA (219 pp). Baker, B.H., Kröger, R., Prevost, J.D., Pierce, T., Ramirez-Avila, J.J., Prince-Czarnecki, J.M., Faust, D., Flora, C., 2016. A field-scale investigation of nutrient and sediment reduction efficiencies of a low-technology best management practice: low-grade weirs. Ecol. Eng. 91, 240–248. Beas, B.J., Smith, L.M., Hickman, K.R., LaGrange, T.G., Stutheit, R., 2013. Seed bank responses to wetland restoration: do restored wetlands resemble reference conditions following sediment removal? Aquat. Bot. 108, 7–15. Bourgues, S., Hart, B.T.H., 2007. Nitrogen removal capacity of wetlands: sediment versus epiphytic biofilms. Wat. Sci. Tech. 55, 175–182. Brisson, J., Chazarenc, F., 2009. Maximizing pollutant removal in constructed wetlands:

261

Ecological Engineering 108 (2017) 255–262

C.M. Shoemaker et al.

Mahaney, W.M., Wardrop, D.H., Brooks, R.P., 2004. Impacts of sedimentation and nitrogen enrichment on wetland plant community development. Plant Ecol. 175, 227–243. McCune, B., Grace, J.B., 2002. Analysis of Ecological Communities. MjM Software Design, Gleneden Beach, OR, USA. Moore, M.T., Kröger, R., Locke, M.A., Tyler, H.L., Cooper, C.M., 2013. Seasonal and interspecific nutrient mitigation comparisons of three emergent aquatic macrophytes. Bioremediat. J. 17, 148–158. Oksanen, J., Guillaume-Blanchet, F., Kindt, R., Legendre, P., Minchin, P.R., O'Hara, R.B., Simpson, G.L., Solymos, P., Stevens, M.H.H., Wagner, H., 2015. Vegan: Community Ecology Package R Package Version 2.2-1. http://CRAN.R-project.org/package= vegan. Peterson, J.E., Baldwin, A.H., 2004. Seedling emergence from seed banks of tidal freshwater wetlands: response to inundation and sedimentation. Aquat. Bot. 73, 243–254. Peterson, S.B., Teal, J.M., 1996. The role of plants in ecologically engineered wastewater treatment systems. Ecol. Eng. 6, 137–148. Pettit, N.E., Ward, D.P., Adame, M.F., Valdez, D., Bunn, S.E., 2016. Influence of aquatic plant architecture on epiphyte biomass on a tropical river floodplain. Aquat. Bot. 129, 35–43. Rabalais, N.N., Turner, R.E., Wiseman, W., 2001. Hypoxia in the gulf of Mexico. J. Environ. Qual. 30, 320–329. Reddy, K.R., Sacco, P.D., Graetz, D.A., Campbell, K.L., Porter, K.M., 1983. Effect of aquatic macrophytes on physico-chemical parameters of agricultural drainage water. J. Aquat. Plant Manag. 21, 1–7. Reddy, K.R., DeLaune, R.D., 2008. Biogeochemistry of Wetlands: Science and Applications. CRC Press, Boca Raton, FL USA (800 pp). Ribaudo, M.O., Heimlich, R., Claassen, R., Peters, M., 2001. Least-cost management of nonpoint source pollution: source reduction versus interception strategies for controlling nitrogen loss in the Mississippi Basin. Ecol. Econom. 37, 183–197. Richards, C., Johnson, L.B., Host, G.E., 1996. Landscape-scale influences on stream habitats and biota. Can. J. Fish. Aquat. Sci. 53, 295–311. Shoemaker, C., Kröger, R., 2017. Frequentist and Bayesian approaches to understanding changes in redox potential due to hydrology and vegetation in agricultural drainage ditches. Wetlands. http://dx.doi.org/10.1007/s13157-017-0901-9. Sims, A., Gajaraj, S., Hu, Z., 2013. Nutrient removal and greenhouse gas emissions in duckweed treatment ponds. Wat. Res. 47, 1390–1398. Sorrell, B.K., Armstrong, W., 1994. On the difficulties of measuring oxygen release by root systems of wetland plants. J. Ecol. 82, 177–183. Spayd III, R.E., Clark, S.E., Baker, K.H., 2013. Impact of plant assemblages on nutrient removal in constructed wetlands. Int. J. Environ. Waste Manag. 11, 315–322. Tanner, C.C., 1996. Plants for constructed wetland treatment systems- a comparison of the growth and nutrient uptake of eight emergent species. Ecol. Eng. 7, 59–83. Thiébaut, G., Muller, S., 1999. A macrophyte communities sequence as an indicator of eutrophication and acidification levels in weakly mineralized streams in northeastern France. Hydrobiologia 410, 17–24. Thullen, J.S., Sartoris, J.J., Walton, W.E., 2002. Effects of vegetation management in constructed wetland treatment cells on water quality and mosquito production. Ecol. Eng. 18, 441–457. Trowbridge, W.B., 2007. The role of stochasticity and priority effects in floodplain restoration. Ecol. Appl. 17, 1312–1324. Uriarte, M., 2000. Interactions between goldenrod (Solidago altissima L.) and its insect herbivore (Trirhabda virgate) over the course of succession. Oecologia 122, 521–528. van Buuren, S., Groothuis-Oudshoorn, K., 2011. mice: multivariate imputation by chained equations in R. J. Stat. Softw. 45, 1–67. Woltemade, C.J., 2000. Ability of restored wetlands to reduce nitrogen and phosphorus concentrations in agricultural drainage water. J. Soil Water Conserv. 55, 303–309. Wren, D.G., Davidson, G.R., Walker, W.G., Galicki, S.J., 2008. The evolution of an oxbow lake in the Mississippi alluvial floodplain. J. Soil Water Conserv. 63, 129–135. Yao, F., Jianbing, S., Chaoqun, T., Wuzhong, N., 2011. Kinetics of ammonium, nitrate and phosphate uptake by candidate plants used in constructed wetlands. Procedia Environ. Sci. 10, 1854–1861. Young, T.P., Chase, J.M., Huddleston, R.T., 2001. Community succession and assembly: comparing paradigms in the context of ecological restoration. Ecol. Restor. 91, 5–18.

should we pay more attention to macrophyte species selection? Sci. Total Environ. 407, 3923–3930. Brix, H., 1994. Functions of macrophytes in constructed wetlands. Wat. Sci. Tech. 29, 71–78. Brooks, R.P., Wardrop, D.H., Cole, C.A., Campbell, D.A., 2005. Are we purveyors of wetland homogeneity? A model of degradation and restoration to improve wetland mitigation performance. Ecol. Eng. 24, 331–340. Cronk, J.K., Fennessy, M.S., 2001. Wetland Plants, Biology and Ecology. CRC Press, Boca Raton, FL, USA (462 pp). De Cáceres, M., 2013. How to Use the Indicspecies Package (ver. 1.7.1). http://cran.rproject.org/web/packages/indicspecies/vignettes/indicspeciesTutorial.pdf. Donders, A.R.T., van der Heijden, G.J.M.G., Stijnen, T., Moons, K.G.M., 2006. Review: a gentle introduction to imputation of missing values. J. Clin. Epidemiol. 59, 1087–1091. El-Sheikh, M.A., Saleh, H.I., El-Quosy, D.E., Mahmoud, A.A., 2010. Improving water quality in polluated drains with free water surface constructed wetlands. Ecol. Eng. 36, 1478–1484. Ervin, G.E., Herman, B.D., Bried, J.T., Holly, D.C., 2006. Evaluating non-native species and wetland indicator status as components of wetlands floristic assessment. Wetlands 26, 1114–1129. Food and Agriculture Organization of the United Nations (FAO), 2015. FAO Statistical Pocketbook: World Food and Agriculture. Food and Agriculture Organization of the United Nations, Rome, Italy. Fligner, M.A., Killeen, T.J., 1976. Distribution-free two-sample tests for scale. J. Am. Statist. Assoc. 71, 210–213. Fredrickson, L.H., 1991. Strategies for Water Level Manipulations in Moist-soil Systems Waterfowl Management Handbook 13.4.6, USFWS. Fujita, Y., Robroek, B.J.M., de Ruiter, P.C., Heil, G.W., Wassen, M.J., 2010. Increased N affects P uptake of eight grassland species: the role of root surface phosphatase activity. Oikos 119, 1665–1673. Gianopulos, K., 2014. Coefficient of Conservatism Database Development for Wetland Plants Occurring in the Southeast United States. North Carolina Dept. of Environment & Natural Resources, Division of Water Resources: Wetlands Branch (Report to the EPA, Region 4). Gilbert, P.M., Maranger, R., Sobota, D.J., Bouwman, L., 2014. The haber bosch-harmful algal bloom (HB-HAB) link. Environ. Res. Lett. 9. Godfrey, R.K., Wooten, J.W., 1981. Aquatic and Wetland Plants of the Southeastern United States: Monocotyledons and Dicotyledons. The University of Georgia Press, Athens, GA USA. Gottschall, N., Boutin, C., Crolla, A., Kinsley, C., Chapagne, P., 2007. The role of plants in the removal of nutrients at a constructed wetland treating agricultural (dairy) wastewater Ontario, Canada. Ecol. Eng. 29, 154–163. Herman, B.D., Madsen, J.D., Ervin, G.N., 2006. Development of Coefficients of Conservatism for wetland vascular flora of north and central Mississippi. Geo. Resour. Inst. Rep. 4001. Howarth, R.W., 2008. Coastal nitrogen pollution: a review of sources and trends globally and regionally. Harmful Algae 8, 14–20. Iamchaturapatr, J., Yi, S.W., Rhee, J.S., 2007. Nutrient removals by 21 aquatic plants for vertical free surface-flow (VFS) constructed wetlands. Ecol. Eng. 29, 287–293. Jesperson, D.N., Sorrell, B.K., Brix, H., 1998. Growth and root oxygen release by Typha latifolia and its effects on sediment methanogenesis. Aquat. Bot. 61, 165–180. Knight, R.L., Payne Jr., V.W.E., Borer, R.E., Clarke Jr., R.A., Pries, J.H., 2000. Constructed wetlands for livestock wastewater management. Ecol. Eng. 15, 41–55. Kross, J., Kaminski, R.M., Reinecke, K.J., Penny, E.J., Pearse, A.T., 2008. Moist-soil seed abundance in the mississippi alluvial valley. J. Wildl. Manage. 72 (3), 707–714. LMVJV Forest Resource Conservation Working Group, et al., 2007. In: Wilson, R. (Ed.), Restoration, Management, and Monitoring of Forest Resources in the Mississippi Alluvial Valley: Recommendations for Enhancing Wildlife Habitat. Lower Mississippi Valley Joint Venture, Vicksburg, MS. Lopez, R.D., Fennessy, M.S., 2002. Testing the Floristic Quality Assessment Index as an indicator of wetland condition. Ecol. Appl. 12, 487–497. Madsen, J.D., Chambers, P.A., James, W.F., Koch, E.W., Westlake, D.F., 2001. The interaction between water movement, sediment dynamics and submersed macrophytes. Hydrobiologia 444, 71–84.

262