Spatial variability of phosphorus sorption dynamics in Louisiana salt marshes

Authors


Abstract

Phosphorus (P) biogeochemistry has been studied in multiple wetland ecosystems, though few data exist on P sorption in U.S. Gulf Coast marshes. There also is a limited understanding of how oil spills in coastal zones can influence P dynamics in wetland soils. In this study, we measured P sorption potential, using the P sorption index (PSI), soil properties, and P saturation at increasing distances from the marsh edge in oiled and unoiled marshes in three regions along the southeastern Louisiana coast (Terrebonne Bay, western, and eastern Barataria Bay). Individual PSI values were highly variable, ranging from 19.5 to 175.6 mg P 100 g−1 and varying by at least a factor of five within each of the three regions, and did not significantly differ between regions or between oiled and unoiled marshes. Soil pH, organic matter, total N, N:P ratio, moisture content, cation exchange capacity, and P saturation differed between regions, and all soil parameters showed great variability between and within individual marshes. Extractable iron was the strongest predictor of PSI across all regions, explaining between 51 and 95% of the variability in individual regions. PSI increased with distance from marsh edge in Terrebonne Bay where other soil properties exhibited similar trends. Results suggest mineral composition of marsh soils, influenced by elevation-inundation gradients, are critical in dictating P loading to estuaries and open waters, and overall marsh functioning. Further, within 2 years of the Deepwater Horizon oil spill, oiled marshes are able to sorb phosphorus at comparable levels as unoiled marshes.

1 Introduction

Positioned between terrestrial and aquatic systems, salt marshes are uniquely situated to intercept and serve as a sink for excess nitrogen (N) and phosphorus (P), and therefore potentially reduce the degree of eutrophication and overall water quality deterioration that impact many coastal systems including the Gulf of Mexico [Rozan et al., 2002; Dunne et al., 2006; Alexander et al., 2008; Jordan et al., 2008]. Further, N and P are critical to the overall health and function of marsh processes such as aboveground and belowground production, nutrient transformations, and carbon (C) mineralization. Despite the importance of both nutrients, much less research has focused on P dynamics.

Phosphorus is delivered to marshes through rivers and estuaries with the dominant portion of P in suspended sediments [Sutula et al., 2004; Jordan et al., 2008]. As particulate P moves from fresh to saline waters, a large portion of the iron (Fe)-bound P is released as Fe is reduced and precipitates with sulfides [Sundareshwar and Morris, 1999; Rozan et al., 2002; Jordan et al., 2008]. Once released, dissolved P becomes available for further transport downstream, burial within sediment, or uptake by microbes [Richardson, 1985; Reddy et al., 1999; Jordan et al., 2008]. Once delivered into wetlands, P cycling relies on the interaction between macrophytes and periphyton communities, exchanges between the soil, pore water, and overlying water, and microbial communities via uptake and changes in redox status and pH [Reddy et al., 1998; Reddy et al., 1999; Noe et al., 2001]. These processes dictate the balance between storage and release of P.

Phosphorus sorption has been studied in wetlands in other regions and many of the mechanisms controlling P retention are well understood. However, there is not much known about P sorption in Gulf of Mexico wetlands. Louisiana wetlands are of particular interest because they cover an extensive area but are experiencing rapid loss and were heavily influenced by the 2010 Deepwater Horizon oil spill. Iron (Fe), aluminum (Al), calcium (Ca), magnesium (Mg), and organic matter are all able to sorb available P in wetland soils [Richardson, 1985; Dunne et al., 2006; Bruland and DeMent, 2009], thereby reducing P loads to adjacent waters and minimizing the potential for eutrophication. Further, P sorption has been shown to differ across salinity gradients, often being lower in brackish and salt marshes relative to freshwater wetlands [Sundareshwar and Morris, 1999; Jordan et al., 2008; Bruland and DeMent, 2009]. Greater ionic strength in saline waters also reduces the P sorption potential due to competition for sorption sites [Sundareshwar and Morris, 1999; Bruland and DeMent, 2009]. Mineral composition and pH of wetland soils also can influence P sorption. Negatively charged clay particles and organic anions can bind with metal cations such as Fe and Al, which in turn attract the negatively charged PO43− ion [Edzwald et al., 1976; Axt and Walbridge, 1999; Bruland and DeMent, 2009], and soils with a lower pH have more Fe and Al hydroxides, which carry a positive charge, thereby facilitating P sorption. Therefore, differences in salinity, SO42− availability, soil organic matter, and Fe and Al concentrations between wetlands with different soil types will likely have different capacities for removing available P from surface and groundwater [Edzwald et al., 1976; Axt and Walbridge, 1999; Sundareshwar and Morris, 1999; Jordan et al., 2008]. Complicating our understanding of P sorption dynamics is the overall lack of information regarding small- and large-scale spatial variability in soil properties in coastal Louisiana marshes.

Louisiana marshes are disappearing at an alarming rate due to multiple stressors such as sea-level rise, insufficient sediment supply, and coastal development. Given their importance to regional water quality, it is necessary to understand how coastal marshes will respond to different perturbations, such as oil exposure. The Deepwater Horizon (Macondo 252) oil spill of 2010 released an estimated 4.4 million barrels of oil into the Gulf of Mexico [Crone and Tolstoy, 2010]. Oil exposure has the potential to influence rates of P sorption and release and to interact with other stressors, leading to complex and unanticipated changes in wetland biogeochemical processing [Lin and Mendelssohn, 2012; Mendelssohn et al., 2012; Sillman et al., 2012]. DeLaune et al. [1979] found that crude oil additions into wetland soils had no impact on biogeochemical processes such as Fe(III) and SO42− reduction, both of which can influence P sorption and release. However, they also found that crude oil inhibited oxygen penetration into the soil, thereby slowing oil degradation and preventing surface soils from becoming oxidized. In contrast, Shin et al. [2000] found greater oxygen demand in wetland soils following crude oil addition, which led to twofold increases in sulfate reduction. The increased rates of sulfate reduction, likely from increased soil carbon content following an oil spill, could increase the concentration of H2S, further promoting the reduction of Fe(III) to Fe(II), thereby releasing additional Fe-bound P [Smolders and Roelofs, 1993; Rozan et al., 2002]. Currently, the long-term effects of oil exposure on P sorption dynamics, and the mechanisms controlling P retention and release, in wetlands soils still are poorly understood.

Altering the delivery of nutrients to the Gulf and the mechanisms that remove N and P can have negative impacts on local and regional water quality, and P retention dynamics in Gulf of Mexico coastal wetlands are not well documented. Therefore, increasing our understanding of the ability of brackish and salt marshes to intercept and retain nutrients, especially following large-scale oil spills, will expand our ability to develop management strategies to reduce and mitigate excess P loading to the Gulf of Mexico under increasing anthropogenic stress. To address these knowledge gaps, we measured P sorption and associated soil properties in 13 brackish and salt marshes in three regions along the southeastern Louisiana coast. In each region we paired oiled and unoiled marshes (one additional oiled marsh was added in one of the regions) and the three regions differed based on vegetation community structure and salinity. The primary goal was to generate the first P sorption numbers for the region and assess spatial variability in both within marshes and across the LA coast. Secondarily, we were interested in determining if there were any long-lasting effects from the Deepwater Horizon oil spill. We hypothesized that (1) there would be regional differences in P sorption based on differences in Fe and Al content (P sorption increasing with Fe and Al), and salinity (greater P sorption in sites with lower salinity) and (2) P sorption would be greater in unoiled marshes than in their oiled counterparts.

2 Methods

2.1 Site Description and Soil Sampling

We sampled 13 marshes across three regions of coastal Louisiana in September 2012: Terrebonne Bay, western Barataria Bay, and eastern Barataria Bay (Figure 1). In each of the three regions, we selected two sets of paired oiled and unoiled sites; sites within each pair were within 2 km of each other, pairs were separated by at least 4 km, and regions were at least 20 km apart. In western Barataria Bay we included an additional oiled marsh. Oiled sites were identified using Shoreline Cleanup and Assessment Technique maps (http://www.noaa.gov/deepwaterhorizon/maps/traj_maps.html) and were verified as having detectable quantities of Macondo oil in the top 5 cm (R. E. Turner et al., Distribution and recovery trajectory of Macondo (Mississippi Canyon 252) oil in Louisiana saltmarshes, Marine Pollution Bulletin, in review, 2014). No Macondo oil was detected in the unoiled marshes.

Figure 1.

Site map of the three sampled regions along the southeastern Louisiana coast.

Sites in Terrebonne Bay were dominated by Spartina alterniflora Loisel., though Juncus roemerianus Scheele was present in all marshes. Western Barataria marshes were dominated by S. alterniflora Loisel.; however, Avicennia germinans (L.) L. was present along the marsh edge at all sites. Eastern Barataria marshes were codominated by S. alterniflora Loisel. and Juncus roemerianus Scheele.

At each of the 13 marshes, we collected two 5 cm deep soil cores (6.7 cm diameter) at 5, 10, 15, and 20 m (for a total of 8 cores per marsh) from the marsh edge using an acrylic soil core with beveled edges. Each core was collected from Spartina-dominated soils. Cores were stored in Whirl-Packs, placed on ice, and transported to the laboratory where they were processed within 24 h. One core from each plot was homogenized and used for P sorption experiments and measurements of extractable Fe, Al, PO4-P, and soil pH (n = 4 for each marsh, with 13 marshes, totaling 52 cores analyzed for PSI). The second core from each plot was used for determination of bulk density, soil texture, cation exchange capacity (CEC), soil water content, organic matter, organic C, total N, and total P concentrations. When overlying water was present, we measured the salinity using a refractometer and collected water for analysis of dissolved inorganic nutrient concentrations. At all 13 sites, we measured temperature and salinity, and collected and filtered water from the bay adjacent to the marsh. Elevation gradients were quantified by measuring water depth at each plot on a total of 3 to 10 visits to each site. By setting the relative elevation at the 5 m plot to 0 and measuring changes in water depth at the three remaining plots, we were able to calculate the change in relative elevation with increasing distance from the marsh edge.

2.2 Soil Analyses

2.2.1 Phosphorus Sorption

We quantified phosphorus sorption using the phosphorus sorption index (PSI) [Bache and Williams, 1971; Dunne et al., 2006; Bruland and DeMent, 2009]. Briefly, 5 g of field-moist soil were added to a 50 mL centrifuge tube with 25 mL of a 130 mg P L−1 solution, amended with two drops of toluene to prevent microbial uptake of P, and shaken for 24 h at 250 rpm. The solution was prepared by adjusting full-strength seawater with deionized water to match the salinity measured in bay water adjacent to the marshes at the time of soil collection. We also measured the salinity at each plot when standing water was present, and salinity did not vary by more than two practical salinity units from the bay water. After 24 h, samples were centrifuged and 10 mL of the supernatant were passed through 0.20 µm syringe filters (Corning, #431224, Corning, NY) and stored frozen until analysis. PSI was calculated as X/log(C), where X equals the amount of P sorbed over 24 h relative to the initial concentration in the stock solution and C is the final inorganic P concentration.

2.2.2 Soil and Water Properties

Iron, Al, and PO4-P (hereafter referred to as POX) were extracted from 10 g of field-moist soil using 0.175 M acidified ammonium oxalate (pH 3.0) by shaking in the dark for 2 h at 250 rpm [Bertsch and Bloom, 1996; Loeppert and Inskeep, 1996] and plant available PO4-P was extracted from 20 g of field-moist soil by shaking for 16 h with 0.5 M sodium bicarbonate (hereafter to referred to as PPA). All samples were centrifuged following shaking and the supernatant was filtered through 0.20 µm syringe filters; exposure to light was kept at a minimum. Samples for Al were acidified with two drops of concentrated nitric acid. Extractable Fe was measured colorimetrically following the method of Loeppert and Inskeep [1996], and extractable Al was measured colorimetrically following the method of Bertsch and Bloom [1996] using a Thermo Finnigan UNICAM UV 300 spectrophotometer. Bay and plot water samples were filtered through acid-cleaned (10% HCl), 0.2 µm pore size, membrane filters (Pall Supor® 200) under low vacuum pressure, and stored frozen until analysis. Samples for dissolved inorganic nutrients (NO3- + NO2-, PO43-, and NH4+) were analyzed using a Lachat Instruments QuickChem® FIA + 8000 Series Automated Ion Analyzer with an ASX-400 Series XYZ Autosampler. Samples were analyzed simultaneously for dissolved NO3 + NO2 (by Cu-Cd reduction followed by azo colorimetry) and PO43− (by the automated ascorbic acid reduction method) but were analyzed separately for dissolved NH4+ (by phenate colorimetry) to prevent contamination of the samples by fumes from the NH4Cl buffer used in the analysis for NO3 + NO2 [American Public Health Association, 1992]. Phosphate from extractions and PSI experiments were measured using similar methods after samples were filtered (0.2 µm syringe filters) and diluted to reduce acidity and allow for good color development. Recoveries of the 130 mg P L−1 PSI stock solutions ranged from 98 to 104%. Phosphate also was measured in the oxalate extractions (POX) because the acidic extraction solution removes P bound to metal oxides, whereas the sodium bicarbonate removes loosely bound P and can provide a greater index of exchangeable P. Further, P, Fe, and Al from the oxalate extractions are used in calculating additional P storage metrics (see below). Standard curves for analyses were prepared from standard stock solutions and yielded r2 values of ≥0.99 for Fe and PO4-P and ≥0.98 for Al. Soil pH was measured using a 1:1 soil:water ratio with a Thermo Scientific Orion 3-Star pH meter equipped with an Orion ROSS Ultra Triode pH/ATC probe. Soil water content was calculated by weighing a 5 g subsample into aluminum weight boats and drying to a constant weight. PSI and all extractable and total nutrients were reported in terms of dry soil.

The remaining soil was dried, ground, and passed through a 2 mm sieve. Bulk density (g cm−3) was calculated as the total dry weight of the soil core divided by the volume. Organic matter content (%) was determined by loss on ignition at 500°C for 4 h. Subsamples were placed into 20 mL glass scintillation vials and fumigated in a desiccator for 24 h under concentrated HCl fumes to remove inorganic C [Hedges and Stern, 1984]. Organic C and total N were then measured using a Flash 1200 Elemental Analyzer (CE Elantech, Lakewood, New Jersey). Buffalo River Sediment standards (National Institute of Standards and Technology, Buffalo River Sediment, 2704) were run concurrently and yielded mean (± SE) organic C and total N concentrations of 3.30 ± 0.01 and 0.26 ± 0.005%, respectively, representing a 98.5% recovery for C (nitrogen is not a certified element in the standard). Total P was extracted by adding 0.5 mL of 50% (w/v) Mg(NO3)2 to subsamples and combusting at 550°C for 1.5 h then shaking for 16 h with 10% HCl, and samples were frozen until analysis. Samples were then diluted and analyzed on a Lachat QuickChem® as described above. Estuarine Sediment standards (National Institute of Standards and Technology, Estuarine Sediment, 1646a) were digested and analyzed concurrently with the samples and had analytical recoveries of 96 to 101%. Cation exchange capacity (CEC) was measured using the compulsive exchange BaCl method [Sumner and Miller, 1996]. Subsamples of soil were treated with 30% H2O2 to remove carbonates and organic matter and the sand, silt, and clay content was measured using the hydrometer method [Gee and Or, 2002].

2.2.3 Phosphorus Metrics

In addition to the potential P sorption (measured as PSI), we also characterized P saturation using two different methods. First, we estimated the phosphorus saturation ratio (PSR) as

display math

where Pox, Feox, and Alox are the PO4-P, Fe, and Al concentrations (molar units) in the ammonium oxalate extracts. This ratio estimates the amount of P absorbed on the soil particles relative to the theoretical absorption capacity (based on the Fe and Al content) [Nair et al., 2004; Dunne et al., 2006]. Using the PSR, we then calculated the soil P storage capacity (SPSC), which is an indicator of how much additional P soil can absorb before switching from a sink to a source of P [Nair and Harris, 2004; Dunne et al., 2006]. SPSC is calculated as

display math

where 0.15 is an estimated phosphorus saturation ratio at which PO4-P contributes to eutrophication of nearby waters [Nair et al., 2004], M1Al and M1Fe are Melich 1 extractable Al and Fe. We used an ammonium oxalate extraction instead of Mehlich 1, similar to both Dunne et al. [2006] and Reddy et al. [1998].

2.2.4 Statistical Analysis

Data for PSI and all soil properties were tested for normality using the Shapiro-Wilk test and for homogeneity of variance using Levene's test. When necessary, data were natural-log transformed. Linear relationships between P sorption and soil and water variables were tested using Pearson's correlation. All variables were tested for significant differences using a two-way analysis of variance (ANOVA) with region (Terrebonne Bay, western Barataria Bay, and eastern Barataria Bay) and oil exposure (oiled, unoiled) as main factors. Significant differences between main effects were tested using Tukey's Honestly Significant Difference posthoc test. We then evaluated PSI with analysis of covariance (ANCOVA) with region and oil exposure as main factors and significantly correlated (p < 0.05) soil properties as covariates. We used regression analyses to evaluate the change in relative elevation across sampling transections by setting the relative elevation to 0 at the 5 m plot. We then regressed water depth against distance from the 5 m plot. Regressions were determined for each site based on the 3 to 10 visits, and then site averages for each region were plotted and regressed to determine regional averages. Unless otherwise stated, all analyses were conducted using SPSS Version 20 (IBM SPSS 2011) at α = 0.05.

3 Results

3.1 Phosphorus Sorption

Phosphorus sorption was highly variable within each marsh and each of the three regions ranging from 31.8 to 175, 19.8 to 149, and 19.2 to 98.4 mg P 100 g−1 in Terrebonne, western Barataria, and eastern Barataria marshes, respectively. Individual values spanned nearly an order of magnitude and PSI in each region varied by at least a factor of 5. Because of the high degree of variability PSI was statistically comparable (p = 0.44) in soils from each of the three regions (Figure 2). Further, mean PSI did not significantly differ between oiled and unoiled marshes (p = 0.74). The variability within individual marshes was often as high as the variability between marshes and between regions. Though there were no differences between regions, there were significant differences between sites within the eastern Barataria marshes (Figure 2). There were no differences between sites within Terrebonne or western Barataria marshes.

Figure 2.

Mean (± standard error) PSI in each site in Terrebonne (n = 4), western Barataria (n = 5), and eastern Barataria (n = 4) marshes. White bars represent unoiled marshes and gray bars represent oiled marshes. Different capital letters indicate significant differences between regions and different lowercase letters indicate significant differences between sites within regions. Significant differences were determined using Tukey's Honestly Significant Difference test.

Relative elevation across the 15 m gradients (from the 5 to 20 m plot) from the four Terrebonne marshes increased linearly (r2 = 0.96) by 12.3 cm and PSI significantly increased with distance from the marsh edge (r2 = 0.47, p = 0.0032) (Figure 3). PSI averaged ~39 mg P 100 g−1 at 5 and 10 m from the marsh edge and was approximately 2 and 2.5 times greater at 15 and 20 m, respectively. A 7.8 cm decrease in relative elevation was measured between the 5 and 20 m sampling plots (r2 = 0.99) in western Barataria marshes, and there was a general decrease, though not significant, in PSI from 78.2 mg P 100 g−1 at 5 m to 56.1 mg P 100 g−1 at 20 m. Mean relative elevation did not change over 15 m in our eastern Barataria marshes (r2 < 0.01), and, similarly, no pattern was observed with distance from marsh edge in eastern Barataria marshes. The relative elevation gradient did not vary significantly with oil status in any of the three regions.

Figure 3.

Linear relationship between PSI and distance from marsh edge in (a) Terrebonne, (b) western Barataria, and (c) eastern Barataria marshes.

3.2 Soil and Water Properties

Similar to PSI, soil properties exhibited a high degree of variability between and within regions. Further, many soil properties exhibited comparable variability within individual marshes as compared to the three different regions. As a result, few soil bulk soil properties differed between regions, and with the exception of mean sand content (42.4% and 35.4% in oiled and unoiled marshes, respectively), no differences were detected between oiled and unoiled marshes (Tables 1 and 2 and Figure 4). Mean soil pH was significantly lower (p = 0.011) in eastern Barataria soils (7.67) than in western Barataria (8.02) and Terrebonne soils (7.92). Mean soil water content and CEC were highest in Terrebonne soils, lowest in western Barataria soils, and intermediate in eastern Barataria soils (p < 0.05) (Table 1). Mean bulk density, soil Fe, Al, POX, and PPA concentrations (Figure 4), and mean sand, silt, and clay content did not significantly differ between regions (p > 0.05) (Table 1). However, similar to PSI, high variability within and between marshes and regions likely precluded the detection of significant differences in soil Fe, Al, POX, and PPA concentrations. For example, soil Fe content varied by a factor of 26, 17, and 8 within Terrebonne, western Barataria, and eastern Barataria Bay marshes, respectively, and comparable variability was found within individual marshes in each region.

Table 1. Mean (± Standard Error) Bulk Soil Properties From Terrebonne (TB), Western Barataria (WB), and Eastern Barataria (EB) Marshes (n = 4 Samples Per Site)a
  pHBulk Density (g cm−3)Soil Water Content (%)CEC (Meq 100 g−1)Sand (%)Silt (%)Clay (%)
  1. a

    Means with different uppercase letters are significantly different based on Tukey's Honestly Significant Difference test (α = 0.05). Bold text indicates oiled marshes.

RegionSite       
TB17.81 ± 0.170.33 ± 0.0277.6 ± 1.725.8 ± 2.8637.9 ± 4.237.7 ± 3.524.4 ± 3.4
 27.96 ± 0.050.30 ± 0.0280.6 ± 2.025.7 ± 2.9941.5 ± 3.738.3 ± 2.720.3 ± 1.9
 38.05 ± 0.120.36 ± 0.0576.6 ± 2.324.4 ± 2.0633.2 ± 2.244.3 ± 2.322.4 ± 0.3
 47.86 ± 0.020.36 ± 0.0875.8 ± 4.625.5 ± 3.0042.2 ± 3.735.1 ± 2.322.6 ± 2.3
WB17.80 ± 0.080.42 ± 0.0268.4 ± 2.38.00 ± 1.8533.8 ± 4.542.3 ± 2.523.9 ± 3.0
 28.03 ± 0.020.50 ± 0.0466.2 ± 3.19.06 ± 0.4436.5 ± 2.941.2 ± 1.022.3 ± 2.2
 37.94 ± 0.140.54 ± 0.0563.8 ± 2.37.15 ± 0.6734.6 ± 2.345.3 ± 3.820.1 ± 1.9
 48.06 ± 0.280.66 ± 0.0760.0 ± 3.410.5 ± 1.4044.4 ± 1.029.3 ± 8.426.3 ± 7.7
 58.28 ± 0.080.99 ± 0.1441.1 ± 6.79.39 ± 1.7247.8 ± 9.531.4 ± 5.520.8 ± 4.0
EB17.75 ± 0.080.36 ± 0.0274.2 ± 0.3411.2 ± 0.4744.4 ± 2.532.2 ± 2.923.4 ± 4.1
 27.76 ± 0.020.40 ± 0.0370.9 ± 1.513.0 ± 1.6433.8 ± 2.939.4 ± 2.826.8 ± 4.0
 37.69 ± 0.070.31 ± 0.0372.1 ± 2.211.6 ± 0.6830.2 ± 4.240.1 ± 4.229.7 ± 6.2
 47.46 ± 0.120.47 ± 0.0269.5 ± 2.212.0 ± 1.0749.1 ± 4.337.0 ± 2.813.9 ± 1.5
Means        
TB 7.92 ± 0.05A0.33 ± 0.02A77.6 ± 1.0A25.3 ± 1.24A38.7 ± 1.8A38.8 ± 1.5A22.4 ± 1.1A
WB 8.02 ± 0.08A0.62 ± 0.05A59.9 ± 4.9B8.81 ± 0.59C39.4 ± 2.4A37.9 ± 2.4A22.7 ± 1.8A
EB 7.67 ± 0.07B0.39 ± 0.02A71.7 ± 1.0AB11.9 ± 0.51B39.4 ± 2.5A37.2 ± 1.7A23.5 ± 2.5A
Oiled 7.94 ± 0.06A0.53 ± 0.05A66.9 ± 2.6A14.6 ± 1.4A42.4 ± 1.8A36.2 ± 1.7A21.3 ± 1.5A
Unoiled 7.81 ± 0.04A0.39 ± 0.02A71.4 ± 1.4A15.2 ± 1.8A35.4 ± 1.6B40.0 ± 1.4A24.6 ± 1.5A
Table 2. Mean (± Standard Error) Soil Organic Matter, Organic Carbon, Total Nitrogen and Phosphorus, and Molar Ratios From Terrebonne (TB), Western Barataria (WB), and Eastern Barataria (EB) Marshes (n = 4 Samples Per Site)a
  Organic Matter (%)Organic C (%)Total N (%)Total P (µg g−1)C:N (mol:mol)N:P (mol:mol)
  1. a

    Means with different uppercase letters are significantly different based on Tukey's Honestly Significant Difference test (α = 0.05). Bold text indicates oiled marshes.

RegionSite      
TB118.2 ± 2.910.9 ± 1.50.59 ± 0.05517 ± 1321.2 ± 1.225.3 ± 2.3
 218.7 ± 1.910.2 ± 1.10.58 ± 0.05546 ± 1320.6 ± 0.523.3 ± 1.9
 319.1 ± 2.29.78 ± 1.30.53 ± 0.06538 ± 3421.4 ± 0.721.7 ± 1.4
 430.7 ± 5.513.7 ± 3.10.66 ± 0.12541 ± 9623.8 ± 0.927.4 ± 2.4
WB111.3 ± 1.78.76 ± 1.80.42 ± 0.07744 ± 10224.1 ± 1.612.5 ± 1.3
 29.8 ± 0.75.43 ± 0.60.30 ± 0.03537 ± 920.8 ± 0.612.5 ± 1.2
 39.2 ± 1.25.05 ± 0.70.27 ± 0.03494 ± 821.6 ± 0.912.1 ± 1.3
 47.6 ± 1.04.85 ± 0.70.24 ± 0.03471 ± 1223.3 ± 1.411.2 ± 1.0
 54.6 ± 1.03.49 ± 0.90.16 ± 0.03462 ± 1727.6 ± 4.27.64 ± 1.4
EB114.8 ± 2.08.59 ± 1.20.50 ± 0.04518 ± 2619.9 ± 1.121.3 ± 1.7
 213.0 ± 0.66.64 ± 0.50.42 ± 0.05513 ± 2418.5 ± 0.619.0 ± 2.1
 317.2 ± 1.211.6 ± 1.00.58 ± 0.02516 ± 1023.0 ± 1.325.1 ± 1.0
 415.4 ± 0.912.0 ± 1.00.55 ± 0.05641 ± 4525.3 ± 0.419.7 ± 3.0
Means       
TB 21.7 ± 3.0A11.1 ± 0.9A0.58 ± 0.04A535 ± 23A21.8 ± 0.5A25.3 ± 1.2A
WB 8.5 ± 1.1B5.52 ± 0.6B0.28 ± 0.03B542 ± 30A23.5 ± 1.0A11.2 ± 0.6B
EB 15.1 ± 0.86AB9.92 ± 0.7AB0.52 ± 0.02A545 ± 19A21.9 ± 0.8A21.4 ± 1.1A
Oiled 12.9 ± 1.1A7.70 ± 1.2A0.41 ± 0.06A531 ± 14A22.7 ± 1.1A16.8 ± 2.3A
Unoiled 16.6 ± 1.8A9.44 ± 1.3A0.49 ± 0.06A554 ± 28A22.1 ± 0.8A20.2 ± 2.8A
Figure 4.

Box plots of extractable (a) Fe, (b) Al, (c) POX, and (d) PPA.

Figure 5.

Relationship between soil Fe concentration and PSI in (a) all marshes, (b) oiled marshes, (c) unoiled marshes, (d) Terrebonne Bay marshes, (e) western Barataria Bay marshes, and (f) eastern Barataria Bay marshes (f).

Similar to soil water content, organic matter content was 1.5 to 2.5 times greater in Terrebonne soils than in western and eastern Barataria soils, respectively (Table 2). Soil total N was significantly lower (p = 0.001) in western Barataria soils (0.28%) than in Terrebonne (0.58%) and eastern Barataria soils (0.52%). Soil N:P ratios were significantly greater in Terrebonne and eastern Barataria marshes than in western Barataria marshes (p = 0.001) (Table 2). Soil organic C was greatest (p = 0.016) in Terrebonne soils (11.1%), lowest in western Barataria soils (5.5%), and intermediate in eastern Barataria soils (9.9%). Soil total P and C:N ratios were comparable between all three regions and between oiled and unoiled marshes (p > 0.05).

Differences in soil CEC, molar N:P ratios, and organic matter between the three different regions were significantly related to salinity. Cation exchange capacity exhibited a significant exponential decline with increasing salinity (r2 = 0.68, p < 0.0001). Conversely, N:P ratios and organic matter showed linear declines with salinity, with salinity explaining 65 and 54% of the variance, respectively. Soil organic C and total N exhibited similar trends with salinity as soil organic matter content. However, salinity is temporally dynamic within the region, whereas soil properties such soil organic matter content, soil organic C, and total N and P, tend to be relatively stable throughout the year (J. M. Marton et al., Nitrification potential and ammonia-oxidizer abundances in Louisiana salt marshes two years after the Deepwater Horizon oil spill, Limnology and Oceanography, in review, 2014).

Salinity in bay water adjacent to the marshes in September 2012 was greatest in western Barataria Bay ranging from 16 to 30, whereas Terrebonne Bay and eastern Barataria Bay had salinities of 10 and 16, respectively (Table 3). Terrebonne Bay had the greatest dissolved inorganic N concentrations (NO3-N + NH4-N), whereas PO4-P was greatest in eastern Barataria and lowest in western Barataria. Surface water PO4-P, collected in the marsh at each of the four plots, was greatest in western Barataria marshes and lowest in Terrebonne marshes.

Table 3. Bay and Plot (n = 1–4 Per Site) Water Characteristics of Study Sites in Terrebonne Bay (TB), Western Barataria Bay (WB), and Eastern Barataria Bay (EB)a
   Bay Water 
RegionSiteSalinityNO3 (μM)NH4 (μM)PO4 (μM)Plot PO4 (μM)
  1. a

    Bold text indicates oiled marshes.

TB1100.220.670.530.40
2100.330.931.090.70
3100.6813.40.390.43
4100.703.160.550.45
WB1160.881.840.391.86
2250.160.490.271.98
3250.664.160.717.98
4251.131.530.602.01
5300.241.030.395.60
EB1160.050.340.754.69
2160.190.150.801.17
3160.212.261.60--
4160.890.060.722.48
Means      
TB  0.484.540.64 
WB  0.611.810.50 
EB  0.330.700.97 

3.3 Phosphorus Storage

Mean P saturation ratio (PSR) was significantly highest (ANOVA; p = 0.034) in Terrebonne soils (9.7%), lowest in western Barataria soils (5.9%), and intermediate in eastern Barataria soils (7.2%) (Figure 6a). Within both the Terrebonne and eastern Barataria regions, PSR was similar among sites, whereas significant differences were found between marshes within the western Barataria region. PSR did not significantly vary with distance from marsh edge in any of the three regions.

Figure 6.

Mean (± standard error) (a) PSR and (b) SPSC in each site in Terrebonne (n = 4), western Barataria (n = 5), and eastern Barataria (n = 4) marshes. White bars represent unoiled marshes and gray bars represent oiled marshes. Different capital letters indicate significant differences between regions and different lowercase letters indicate significant differences between sites within regions. Significant differences were determined using Tukey's Honestly Significant Difference test.

Mean soil phosphorus storage capacity (SPSC) ranged from 2.8 μM g−1 in Terrebonne soils to 5.4 μM g−1 in western Barataria soils, with no significant differences being detected between region, oiled, and unoiled marshes, or sites within any of the three regions (Figure 6b). SPSC significantly increased with distance from marsh edge in Terrebonne marshes (r2 = 0.27, p = 0.022) but decreased with distance into the marsh in western Barataria marshes (r2 = 0.21, p = 0.024). No spatial pattern was detected in eastern Barataria marshes.

3.4 Relationships Between Phosphorus Dynamics and Soil Properties

Phosphorus sorption was most strongly related to soil Fe (r2 = 0.71, p < 0.001), but significant relationships were also found with Al (r2 = 0.29, p < 0.001), PPA (r2 = 0.34, p < 0.001), POX (r2 = 0.15, p = 0.005), organic C (r2 = 0.14, p = 0.007), total N (r2 = 0.08, p = 0.038), and total P (r2 = 0.19, p = 0.001) concentrations; therefore, these parameters were included as covariates in the ANCOVA models. Soil aluminum, organic C, total N, and total P did not yield significant interaction terms with either region or oil exposure in the ANCOVA models.

Soil Fe concentration was a significant covariate for both region (p = 0.029) and oil exposure (p = 0.031). Across all 52 plots in the three different regions, soil Fe concentration explained 71% of the variability in PSI, though this relationship was stronger in unoiled marshes (r2 = 0.88) than in oiled marshes (r2 = 0.55) (Figures 5a–5c). When separated by region, soil Fe concentration explained 95, 51, and 89% of the variability in PSI in Terrebonne, western Barataria, and eastern Barataria soils, respectively (Figures 5d–5f). However, when a single outlier from one oiled marsh in eastern Barataria was removed, Fe was no longer a significant covariate for region or oil. Similarly, there was a significant interaction between region and PPA (p = 0.002). This indicates that the response of PSI to soil PPA differs between the three regions, which had regression slopes of 4.1, 2.5, and 2.2 in Terrebonne, western Barataria, and eastern Barataria marshes, respectively. There also was a significant interaction between oil exposure and PPA (p = 0.046). In Terrebonne marshes only, extractable Fe, Al, and PPA significantly increased with distance from marsh edge (Figures 7a–7c), likely explaining the significant increase in PSI with distance from marsh edge. There were no other significant relationships between soil properties and distance to marsh edge in the other two regions, though they followed a comparable pattern to PSI and relative elevation, with Fe, Al, and PPA tending to decrease with distance in western Barataria and showing no pattern in eastern Barataria.

Figure 7.

Relationship between (a) Al, (b) Fe, and (c) PPA with increasing distance from marsh edge in Terrebonne marshes.

Across all sites and plots (n = 52), PSR was significantly correlated to PSI (r = −0.43, p = 0.001), Fe (r = −0.57, p < 0.001), Al (r = −0.62, p < 0.001), and total P (r = −0.39, p = 0.005). Similarly, SPSC was significantly correlated with PSI (r = 0.61, p < 0.001), Fe (r = 0.92, p < 0.001), Al (r = 0.84, p < 0.001), POX (r = 0.40, p = 0.004), and soil total P (r = 0.62, p < 0.001). As P sorption sites became less available, as indicated by an increase in the PSR, PSI decreased exponentially (Figure 8a). Conversely, as the potential SPSC increased, there was a concomitant increase in PSI (Figure 8b). None of the dissolved nutrient concentrations in the bay or marsh surface water were correlated to PSI, soil properties, or P storage.

Figure 8.

Relationship between (a) PSI and PSR and between (b) PSI and SPSC.

5 Discussion

5.1 Phosphorus Sorption

This is one of the first studies to investigate P sorption and associated soil properties across multiple spatial scales in coastal wetlands of the Gulf of Mexico. The large geographic area from which we sampled included marshes with salinity values ranging from 10 to 30 at the time of sampling (Table 3). Even though soil properties are relatively stable throughout the year and we only measured salinity at the time of sample collection, the trends we observed for salinity were consistent with the general exposure at these sites. In 2012, salinity values in Terrebonne Bay ranged from approximately 2 to 25 with an annual median of 13.0 (Terrebonne Bay Monitoring Station, LUMCON: www.lumcon.edu), whereas values in western Barataria ranged from 0.4 to 32 with a 2012 median of 15.8 (USGS Monitoring Station 07380251: http://nwis.waterdata.usgs.gov/). The monitoring station in Terrebonne Bay is more seaward than our study sites (~5 km), resulting in salinity values at our marshes typically being lower than at the monitoring station. Despite the range in salinity, PSI exhibited large variations within each of the three regions, spanning ranges of 143, 129, and 79.2 mg P/ 100 g soil in Terrehonne, western Barataria, and eastern Barataria marshes, respectively. The high variability associated with each region likely precluded our finding any significant differences between regions or any correlation with salinity, though our values and ranges were comparable to other published PSI values. Axt and Walbridge [1999] measured PSI in freshwater forested wetlands, streambanks, and uplands in the Piedmont and Coastal Plain regions in Virginia and reported values ranging from 61 to 175 mg P 100 g−1 and a single value as high as 366 mg P 100 g−1; across all habitats, wetlands had the greatest PSI. Richardson [1985] also found high variability (~10–165 mg P 100 g−1) in PSI between and within bogs, fens, and nontidal swamps from Michigan, Maryland, and North Carolina. Bruland and Dement [2009] found that PSI was significantly greater in Hawaiian freshwater marshes (salinity of < 0.5) relative to the brackish (salinity of 0.5–30), euhaline (salinity of 30 to 40), and hyperhaline (salinity of >40). They reported a mean PSI of 209.7, 133.2, 65.5, and 73.2 mg P 100 g−1 in freshwater, brackish, euhaline, and hyerhaline wetland soils, respectively, whereas we measured mean regional PSI values ranging from 45.2 to 64.3 mg P 100 g−1, lower than the freshwater and brackish, but comparable to the euhaline Hawaiian marsh soils. Sundareshwar and Morris [1999] also found that P sorption was 33 times greater in freshwater than in salt water marsh soils, and Jordan et al. [2008] showed that phosphate release increased as suspended sediments moved from fresh to saline conditions.

Though we did not find a significant relationship between salinity and PSI, we did measure declines in soil properties previously found to influence P sorption. Soil cation exchange capacity, N:P ratios, and soil organic matter content (and similarly, soil organic C and total N) all decreased with increasing salinity, and the patterns with CEC and soil organic matter were strongest in Terrebonne soils. For example, we found a significant correlation between PSI and CEC (r = 0.54, p = 0.030) and soil organic matter (r = 0.66, p = 0.005) in Terrebonne soils, and much weaker relationships in western and eastern Barataria soils. Both CEC and soil organic matter declined as salinity increased, which could have indirectly decreased P sorption and retention. Sundareshwar and Morris [1999] also found decreases in CEC along a salinity gradient in the Cooper River estuary, South Carolina. Others have reported that PSI is strongly dependent upon negatively charged soil organic matter [Axt and Walbridge, 1999; Bruland and Richardson, 2006; Bruland and Dement, 2009], which can form complexes with soil metals, such as Fe, Al, Ca, and Mg. However, we did not find a correlation between soil organic matter and PSI across all plots, suggesting that the exchangeable Fe and Al were more associated with the mineral, rather than the organic, component of the soil.

After Fe, marsh position was the strongest predictor of PSI; however, marsh position also explained 31% of the variance in Fe concentrations in Terrebonne Bay. Several soil properties and PSI exhibited significant increases with increasing distance from the marsh edge in Terrebonne marshes, whereas decreases, though nonsignificant, were detected in western Barataria marshes, and no pattern was observed in eastern Barataria marshes. These patterns are consistent with differences in relative elevation for the three regions. Relatively higher locations within the marshes of the different regions exhibited relatively higher soil organic matter and Fe concentrations and rates of PSI. Further, stronger patterns were observed in the Terrebonne marshes which also exhibited strongest relative elevation gradient. These findings are consistent with patterns in nitrification potential and abundances of ammonia-oxidizing bacteria and archaea (J. M. Marton et al., in review, 2014). It is probable that these spatial patterns are the result of the interaction between tides and marsh surface elevation. With marsh inundation driven more by wind than actual tides, and a tidal range of only between 45 and 60 cm [Orson et al., 1985], environmental gradients are established over relatively short distances into the marsh. These differences in elevation likely structure soil properties within the marsh, which in turn affect processes such as P sorption and nitrification potential.

5.2 Legacy Effects of Oil

In addition to quantifying P sorption over a large geographic range (~80 km) in Louisiana's coastal marshes, this data set is one of the first to investigate how oil spills can affect P dynamics in these systems. Our results show that Louisiana's coastal wetlands are highly variable between and within individual marshes and regions and can serve as significant short-term sinks for P, thereby providing a valuable ecosystem service to the Gulf of Mexico.

The Deepwater Horizon oil spill resulted in heavy oiling of many salt marshes in Terrebonne and Barataria Bays, potentially impacting many important ecosystem services. Immediately following the oil spill, alkanes and aromatics in marsh soils increased on average 481 and 46,631 times above background, respectively (R. E. Turner et al., in review, 2014) Over 2 years later, and 1 month after our study was conducted, alkane and aromatic concentrations in October 2012 had degraded on average by 98 and 90%, respectively, of their initial concentrations, though these concentrations were still 10 and 4,811 times greater than preoiled concentrations. No Macondo oil was detected within the unoiled marshes; however, crude oil (alkanes plus aromatics) were detected in all marshes indicating that the baseline condition across Louisiana's marshes is one of a history of oil exposure. Two years following the Deepwater Horizon oil spill, oiled marshes sorbed comparable amounts of P as their “unoiled” counterparts suggesting that if there were any effects on P sorption, they were relatively short-lived. Further, oiled marshes are able to sorb as much, and in some cases more, P than unoiled marshes in Louisiana and other coastal areas. However, some of the relationships between soil properties and P sorption differed between oiled and unoiled marshes.

Both Fe and Al are often significantly correlated with P sorption in wetland systems. Sundareshwar and Morris [1999] reported that Fe and Al were the dominant controls in P retention in marsh soils along a salinity gradient in the Cooper River estuary, South Carolina. They further found that P retention declined as salinity increased due to decreases in Fe and Al content and increases in ionic strength of saline water, which creates greater competition for sorption sites. Similarly, Jordan et al. [2008] confirmed decreasing extractable Fe concentrations and P retention in suspended sediments along a salinity gradient in the Patuxent River estuary. Their results confirmed that as salinity increases, sulfides in saline sediments reduce Fe(III) to Fe(II), thereby releasing Fe-bound P. P sorption measured in mineral and peat swamps, fens, and pocosins was significantly correlated to extractable Al (r = 0.93) and Fe (0.62), and less so with organic matter [Richardson, 1985]. In soils from freshwater forested wetlands in Virginia, extractable Al and organic matter were the dominant controls on PSI [Axt and Walbridge, 1999]. In our study, extractable Fe was the strongest predictor of PSI across all sites and regions, though the strength of this relationship varied between the three regions and between oiled and unoiled marshes. Our ANCOVA models found a significant difference between the slopes of PSI on extractable Fe between the three regions and between oiled and unoiled marshes. These differences were driven primarily by one of the oiled marshes in eastern Bartaria, which had the highest measured Fe concentration (4.47 mg g−1) but a PSI value of only 59.7 mg P 100 g−1. When this one point was removed from the analyses, slopes were statistically comparable between regions and oiled and unoiled marshes.

Since this study was conducted 2 and a half years following the Deepwater Horizon oil spill, we cannot say for certain that there were no immediate effects on P sorption, but we can say that if effects did occur, they were relatively short-lived as no differences were detected in PSI between oiled and unoiled marshes. Additional work needs to be conducted in which soils are exposed to oil in a controlled laboratory setting to accurately determine immediate effects on P sorption and the associated controlling factors.

5.3 Phosphorus Saturation and Storage

Terrebonne marsh soils were more P saturated than both western and eastern Barataria marshes, due to lower extractable Fe and Al, but greater oxalate-extractable P concentrations, though this pattern was primarily driven by one of the oiled marshes. Further, soils in Terrebonne marshes had greater organic matter content, and the adjacent waters had a lower salinity, both of which can increase P retention [Richardson, 1985; Jordan et al., 2008; Bruland and Dement, 2009]. This means that across the southeast Louisiana coast, despite higher soil organic matter and lower salinity, coastal marshes in Terrebonne Bay are less likely to retain future P additions from surface and ground water over the long-term relative to the marshes of Barataria Bay, which are less P saturated. Nair et al. [2004] identified a PSR value of 0.15 as a critical threshold above which soils are no longer capable of retaining P which can potentially lead to eutrophication of nearby waters. In our study, only three individual plots, all in Terrebonne Bay marshes, exhibited a PSR greater than 0.15. Using this critical threshold value, the SPSC is an indicator of how much additional P could be stored by soil Fe and Al. Though not statistically significant, Terrebonne soils had the lowest overall SPSC value, which means that despite currently having comparable PSI to the Barataria marshes, they are less likely to continue to serve as P sinks as they approach the critical PSR value. Both PSR and SPSC exhibited high variability, similar to that observed in PSI and soil properties. High variability in P sorption dynamics has been reported by others [Dunne et al., 2006; Bruland and Dement, 2009] and has been attributed to various biogeochemical processes occurring along salinity and inundation gradients. Interestingly, even though the relationship between PSI and Fe significantly differed between oiled and unoiled marshes, PSI, PSR, and SPSC did not significantly differ between oiled and unoiled marshes. Whereas oil exposure may impact potential P sorption rates over short temporal scales, there appears to be no effect on overall P storage 2 years following oil exposure. Despite a high capacity to remove P, these marshes have yet to become P saturated and will continue to intercept and retain P, thereby reducing the total load to the Gulf of Mexico.

5.4 Regional Implications

Terrebonne and Barataria Bays have among the highest surface areas of both wetland and open water habitats of any estuaries along the Gulf Coast of the U.S. [Turner and Rabalais, 1999]. Our study sites spanned most (~80 km) of these estuarine systems and most of the region between the Mississippi and Atchafalaya Rivers, which are the main conduits of water, sediments, and nutrients to Louisiana marshes. Across this large region, we observed similar rates and variability in P sorption and storage dynamics.

We hypothesized that PSI would be driven by Fe, Al, and salinity, which we expected to vary by region and thus lead to differences in PSI. Surprisingly, we found that Fe and Al varied nearly as greatly within regions (and even with individual marshes) as between regions, thereby precluding any significant regional differences in PSI. Our PSI values ranged by nearly an order of magnitude across all sites and varied by at least a factor of 5 within each of the three regions. This regional similarity is important for making generalizations about the Louisiana coast and for future modeling efforts. Despite the fact that Fe and Al content were significantly correlated with PSI, soil properties were highly variable as well, suggesting several samples from each marsh are required to adequately characterize a system and to properly identify relationships between soil properties and processes. The fact that the soil properties most strongly influencing PSI (e.g., Fe, Al, and soil organic matter) varied with distance from the marsh edge and that these soil properties followed similar patterns as relative elevation suggests that inundation patterns are one of the dominant structuring factors in how these marshes function. Further, the region with the greatest change in relative elevation across our sampling transect also exhibited the greatest variability in PSI.

Our second hypothesis was that PSI would be greater in unoiled than in oiled marshes, though our findings did not support this. This study was not designed to determine initial impacts of oil exposure on P sorption and the lack of effects 2 years later suggest that any effects on P sorption, if present, were temporary and are not affecting P retention and storage over longer time periods. It is also possible that the lack of detectable oil effects may be in part due to all sites demonstrating a history of oil exposure, suggesting that “unoiled” marshes simply no longer exist in the region.

6 Conclusions

Our results suggest that brackish and salt marshes along the southeastern Louisiana coast are capable of removing P at levels comparable to marshes in other saline systems. Overall, P sorption and retention are driven primarily by soil Fe and Al and did not vary along a salinity gradient. Marsh elevation and inundation patterns also seem important in influencing soil properties, and therefore P sorption, leading to high but somewhat predictable variability within marshes. Over 2 years following the Deepwater Horizon (Macondo 252) oil spill, no legacy effects on P sorption were observed suggesting that P sorption was never affected, this function rapidly recovered, or that the baseline of oil exposure makes it difficult to identify oil effects 2 and a half years postexposure.

Acknowledgments

We would like to thank Anya Hopple and Shauna-Kay Rainford for their help with sample collection and processing, Brad Rosenheim for help with some sample analyses, Maggie Marton for editorial assistance, and the comments from the anonymous reviewers who improved this manuscript. All data are available through the Gulf of Mexico Research Initiative Information and Data Cooperative (https://data.gulfresearchinitiative.org). This research was made possible by a grant from BP/The Gulf of Mexico Research Initiative to the Coastal Waters Consortium.

Ancillary