*2.3. Equilibrium Phosphorus Concentration (EPC0)*

Since EPC0 analysis required sorption assays with multiple PO4 <sup>3</sup><sup>−</sup> solution concentrations for each sediment sample, to keep things manageable, we limited the sediment analysis to only one randomly-selected depth across the 15 streambank sites and the fine sediment fraction. The intent here was to survey the EPC0 behavior across sites. With the inclusion of two replicates, 30 sediment samples were analyzed in total (15 sites × 1 depth × 2 replicates). In addition, to assess EPC0 variability with streambank depth, we determined the EPC0 values for all depths at two sites—BEB and SM3 (four depths each). These samples were also replicated twice.

Four PO4 <sup>3</sup>−-P solution concentrations, representing the likely range of stream water P concentrations, were used for the assay and included n.d., 0.25, 0.50, and 2.0 mg P L<sup>−</sup>1. All solutions were made by dissolving KH2PO4 in filtered stream water collected from a forested headwater tributary of Big Elk Creek, Maryland, with undetectable P concentration. Stream water was used to simulate natural ionic strength conditions. About one gram of sediment (exact weight was noted) was added to a 50 mL centrifuge tube to which 20 mL of PO4 <sup>3</sup><sup>−</sup> solution was added; this mixture was created for each of the four P solution concentrations. Two drops of chloroform were added to the solution to inhibit microbial activity. Samples were placed on an end-over-end shaker and incubated for 24 h at 25 ± 2 ◦C. Once incubated, samples were centrifuged at 2000 rpm for 30 min. Using Sterlitech Glass Microfiber 0.7-μm filters, the centrifugate was filtered into 40 mL amber vials. The solution PO4 3−

concentrations were measured colorimetrically using EPA-118-A Rev 5 method on the AQ2 Discrete Analyzer (Seal Analytical, Mequon, Wisconsin).

The EPC0 was computed using the procedures described by [36] based on [30]. The P sorbed on the sediment phase (*S*) (μg g<sup>−</sup>1) after a 24 hour period was calculated using the following equation:

$$S = \frac{v}{m} \text{ (C}\_0-\text{C}\_{24}\text{)}$$

where *v* is the total volume of the solution (0.02 L in this case), m is the mass of dry sediment in (g), *C*<sup>0</sup> (μgPL<sup>−</sup>1) is the concentration of solution prior to incubation and *C*<sup>24</sup> (μgPL<sup>−</sup>1) is the concentration of solution after the 24-hour incubation. S was plotted (on the y-axis) against *C*<sup>24</sup> (x-axis) and the data points were fitted with a linear regression. At low solution P concentrations (as in this case) the relationship between C and S can be estimated using linear regression [30,37]. The EPC0 (μgPL−1) value is the coordinate at which the linear fit line intercepts the x-axis and is computed by substituting y as 0 and solving the linear equation for x. EPC0 values were then expressed in mg L<sup>−</sup>1. Additionally, Pearson correlations (r) were determined to investigate the relationships between the EPC0 value, particle size fractions, and M3Fe and M3Al concentrations using JMP software (version 14.0).

#### *2.4. Sorption and Desorption Under Anoxic and Oxic Conditions*

The intent of this experiment was to investigate how legacy sediment P sorption or desorption could vary if streambank legacy sediments were eroded from the bank and deposited in the stream under anoxic and oxic conditions [32]. Legacy sediment samples for four sites, one depth each, and only the fine sediment fraction were selected for this analysis. Three replicates were used for each sediment sample. Since the legacy sediments had low initial inorganic P, the sediments were exposed to an elevated P solution (10 mg L<sup>−</sup>1) prepared using KH2PO4 prior to the experiment. For this, thirty grams of the dry sediment sample was placed in an acid-washed, ethanol cleaned, glass tray and 600 mL of 10 mg L−<sup>1</sup> PO4 <sup>3</sup><sup>−</sup> solution was added to saturate the sediment with P. The sediment was placed on a shaker table for 24 h at 100 rpm, drained, and then placed under a dry hood until any excess liquid was evaporated. The sediment was then removed and placed in a sterile Ziploc bag until the second part of the experiment.

For the second part of the experiment, two grams of P-sorbed legacy sediments were added to 40 mL of filtered stream water in an amber vial and placed on a shaker table for 24 h at 112 rpm. A control treatment was also created where no sediment was added to the stream water. Unlike that for EPC0, stream water PO4 <sup>3</sup><sup>−</sup> values were above detection (0.04 mg P L<sup>−</sup>1) for this experiment. This was likely because of stormflow conditions prior to sampling; this however was not a problem for this experiment since our intent was to investigate the differences between oxic and anoxic conditions. The vials were subject to both oxic and anoxic treatments. To maintain oxic conditions, the cap was left off the amber vial to ensure oxygen would not be depleted in the water. Anoxic conditions were created by adding one gram of sodium sulfite (Na2SO3; an oxygen scavenger) to the solution and the vial was sealed by closing the cap. Anoxic conditions were verified using a dissolved oxygen meter. After 24 h, samples were filtered using Sterlitech glass microfiber 0.7 μm filters. The sample solutions were measured for their solution PO4 <sup>3</sup><sup>−</sup> concentrations (EPA-118-A Rev 5) on an AQ2 Discrete Analyzer (Seal Analytical, Mequon, Wisconsin). A t-test was used to determine the significant differences between the oxic and anoxic treatment groups.

#### **3. Results and Discussion**

*3.1. Concentrations of P in Streambank Legacy Sediments and Comparisons Against Upland Soils, Stream Bed Sediments, and Water Quality Thresholds*

P concentrations measured by [21,22] are presented in Table 1 and are compared against other studies with data on streambanks, bed sediments, and upland soils. It should be noted that while P concentrations for both the coarse and fine fractions of the soils/sediments were measured by [38], only the fine fraction values were reported in [21]. The comparisons here provide important insights. Total P concentrations for streambank legacy sediments were the lowest of all sediment sources (Table 1) with mean P concentrations for the banks nearly half of those reported for cropland and developed soils [21]. The difference in mean concentrations between legacy sediments and other sediment categories was less for M3P, but nonetheless, M3P values for streambanks were the lowest among all sources (Table 1). Mean concentrations for total and M3P for streambanks [21,22] were comparable to streambank values from other studies (Table 1). Percent DPS values for streambanks were also low and particularly lower than values reported for cropland and developed soils (Table 1).


**Table 1.** Comparison of legacy sediment P concentrations against other soil/sediment sources. Comparisons are made for particle size class (<63 μm: fine), coarse, and bulk sediments. Concentrations include a range of values and mean in brackets ( ). Where available, sample numbers are indicated in square brackets [ ]. Table modified from [22].

When P concentrations are assessed for particle size class (Table 1), total P concentrations for fine sediments were twice or more than those for coarse fractions across upland soils, banks and bed sediments. The same level of separation, however, was not observed for M3P and %DPS values. Mean %DPS was generally greater for the coarse versus the fine fractions but a similar consistent trend was not observed for M3P. Similarly, when mean %DPS values were compared for bed sediments and streambanks, bed sediment values were slightly greater than the streambank values, with no consistent pattern for total P and M3P.

While M3P was originally developed for agronomic needs, i.e., determining crop nutrient requirements and associated fertilizer application, it has been used for determining environmental risk associated with P leaching [27]. In Delaware, M3P and %DPS values less than 50 mg kg−<sup>1</sup> and 15%, respectively, are considered "below optimum" and do not pose any risk to water quality [27]. In contrast, M3P and %DPS values in excess of 100 mg kg−1and 58%, respectively, are considered a threat to water quality [27].

In Arkansas, environmental threshold for water quality concern for M3P is higher at 150 mg kg<sup>−</sup>1. When compared against these water quality thresholds, studies listed in Table 1 arrived at the same conclusion that streambank sediments (legacy and non-legacy) likely posed a low risk for P leaching under well oxygenated conditions and served as a net sink for P [36,41–43]. All of the studies, however, did recognize that while P concentrations were low and most of the P was likely bound to metal hydro-oxides, this P could be released into solution due to the reductive dissolution of the oxides under anoxic conditions [44,45]. The low P concentrations in streambank legacy sediments should not be very surprising considering that much of these sediments were likely deposited prior to the 1950s [13,15,17], before the significant increase in use of synthetic N and P fertilizers on agricultural lands [3]. One way these buried legacy sediments could have acquired elevated P concentrations would be through contact with P-rich streamflow (along the banks or during overbank flooding) and/or upland runoff carrying fertilizer nutrients that infiltrated through the soil profile.

Elevated concentrations of total P in bank versus stream bed sediments (Table 1) were attributed to a greater fraction of fine sediments in the banks, which includes P-sorbing iron oxides [39]. The same pattern, however, has not been reported by other studies in Table 1 (e.g., [21,40]). This could be because bed sediments typically represent a mixture of various sediment sources including P-rich upland sediments. Bed sediments could also acquire elevated P concentrations from P-rich stream runoff. However, broadly, most studies do report that bed sediments are more coarse grained than bank sediments [39] and this would likely result in less P sorption capacity for bed sediments as reflected by the higher %DPS values for bed sediments in Table 1.

#### *3.2. Phosphorus Sorption Index (PSI) for Legacy Sediments*

Solution concentrations of PO4 <sup>3</sup><sup>−</sup> after 18 h of legacy sediment incubation were lowest for the fine fraction (<63 <sup>μ</sup>m) (43.7 <sup>±</sup> 8.5 mg P L−1; Figure 3) down from the starting concentration of 75 mg P L<sup>−</sup>1. While these solution concentrations are much greater than what one would typically observe in streams, this experiment indicates that fine legacy sediments have a fairly high capacity for P sorption. In comparison, solution concentrations for the coarse fraction (>63 μm) were higher (61.5 <sup>±</sup> 10.9 mg P L<sup>−</sup>1; Figure 3), indicating lower sorption for this sediment class. Three samples within the coarse fraction had solution concentrations greater than the starting solution of 75 mg P L−<sup>1</sup> (Figure 3) indicating some release of P from sediments. The mean PSI value for coarse and fine legacy sediment fractions taken together was 472.3 <sup>±</sup> 270.4 mg kg−1, while that for the coarse fraction was 292.6 <sup>±</sup> 224.4 mg kg<sup>−</sup>1, and that for the fine fraction was 652.0 <sup>±</sup> 177.3 mg kg−<sup>1</sup> (Table 2). There was a significant difference in PSI values between the coarse and fine size fractions (p < 0.001). There was a strong positive correlation between PSI values for the coarse fraction and M3Al (r = 0.77; p < 0.001) and a weak and insignificant correlation with M3Fe (r = 0.17; p = 0.15). For the fine fraction, the relationship between M3Al was slightly weaker, but still positive (r = 0.69; p < 0.0001) and there was a positive correlation with M3Fe (r = 0.26; p = 0.029).

**Figure 3.** Solution PO4 <sup>3</sup>−P (mg P L<sup>−</sup>1) concentrations after 24-hour incubation for coarse fraction and fine fraction sediments. The dashed line represents the starting solution concentration (75 mg PO4 <sup>3</sup>−P L<sup>−</sup>1). Larger decreases below this value indicate greater sorption by sediments. The labels on the x-axis include site name and bank sampling depth in cm.


**Table 2.** PSI (mg/kg) values for agricultural and streambank soils in the Mid-Atlantic Region.

\* Agricultural soils; \*\* Field Border areas separate crop fields from drainage ditches.

The PSI experiment confirmed that fine legacy sediments have a greater sorption capacity than the coarser fractions. The PSI value for the fine legacy sediments was also greater than most of the agricultural soils (e.g., Table 2, Evesboro loamy-sand (136–263 mg kg−1) and Pocomoke sandy clay loam (95–714 mg kg<sup>−</sup>1), with the exception of the Matawan sandy loam, which had a higher sorption capacity (588–2564 mg kg<sup>−</sup>1) [46]. In another sorption study [39], maximum sorption values for exposed and submerged streambank sediments in an agricultural catchment in central Pennsylvania were 259 mg kg−<sup>1</sup> and 214 mg kg<sup>−</sup>1, respectively (Table 2). These values were much lower than values for our fine legacy sediment fraction, but comparable to the coarse fraction PSI value.
