*3.3. Equilibrium Phosphorus Concentration (EPC0)*

EPC0 values across the 15 legacy sediment sites ranged from 0–0.24 (mean: 0.044) mg L−<sup>1</sup> (Table 3). Ten of fifteen sites had EPC0 values greater than the baseflow stream water concentration during sediment sampling indicating that the sediment could be a potential P source if deposited into the channel (Table 3). Five of the fifteen sites had EPC0 values that were less than the stream water P concentrations indicating that the sediment would act as a potential sink for P (Table 3). While there was no consistent relationship between EPC0 values and stream water PO4 <sup>3</sup><sup>−</sup> concentrations, two of the highest EPC0 values were associated with Brandywine Zoo and Cooch's Bridge locations, both urban stream locations (Table 3).

**Table 3.** Sediment EPC0 concentrations for the 15 streambank legacy sediment sites (one selected depth) and comparisons against baseflow stream water PO4 <sup>3</sup><sup>−</sup> concentrations to determine if sediments would act as a source or sink for P.


Our EPC0 values were within the range of values reported by other studies (Table 4). A study in Maryland (Kimages creek) that investigated legacy sediments reported an EPC0 value of 0.010 mg L−<sup>1</sup> [47], while that from till bank material near Lake Pepin in Minnesota had an EPC0 value of less than 0.1 mg L−<sup>1</sup> [48]. Studies have shown that there is a strong correlation between the EPC0, soluble reactive phosphorus (SRP), and dissolved reactive phosphorus (DRP) in stream water [43,49]. The EPC0 in sediments may change depending upon the SRP concentration in the sediment and stream waters [49] as, potentially, in the case of our urban, P-rich, Brandywine zoo site (Table 3). One study [49] looked at the influence of sewage treatment works (STWs) on riverbed sediment to determine if sediments could act as a buffer for the increase in stream water P concentrations. They found that regardless of STW influence, the riverbed sediments always acted as sinks in the water column. Despite higher concentrations of SRP in the rivers downstream of STWs, sediment near STWs had a higher capacity to absorb SRP [49]. Similarly, [48] showed that fine bank sediments transported in stream waters sorb elevated P from sewage and industrial waste and then deliver it to Lake Pepin. Given that controlling bank erosion could be expensive, they suggest reducing P inputs to waterways to control this loading to Lake Pepin.


**Table 4.** Comparison of EPC0 values from various sites and sediments types reported in the literature.

For our sites, we did not find any significant correlations between EPC0 concentrations and particle size classes %fine, %sand, %silt, and %clay. In contrast, EPC0 was significantly negatively correlated with M3Al (r = −0.70; p = 0.0033) and positively correlated with M3Fe (r = 0.54; p = 0.03). Contrary to our observations, a significant inverse correlation between M3Fe and EPC0 was reported by [36], indicating that as the amount of M3Fe increased, the potential for sorption increased resulting in lower EPC0 values. In the same study, EPC0 values were positively correlated with sand content but negatively correlated with silt and clay content [31,36], indicating that with finer fractions greater retention of P occurred.

EPC0 values also varied with bank depth for both sites that were evaluated with depth. At BEB, for depths 60, 122, 183, and 260 cm from the top, the EPC0 values were 0.032, 0.028, 0.002, and 0.032, respectively. At SM3, for depths 132, 173, 231, 267 cm from the top, the values were 0.031, 0.024, 0.047, and 0.031, respectively. This variation in EPC0 values with bank depth was likely due to variation in sediment composition and characteristics with depth [22]. When the BEB EPC0 values are compared against seasonally varying stream water PO4 <sup>3</sup><sup>−</sup> concentrations (Figure 4), we note that the source-sink behavior of sediments varies temporally. The sediments behave as a sink when stream water P concentrations are elevated, particularly during stormflows (Figure 4), and as a source during low-P baseflow conditions. For example, for the sediment at 60 cm depth (EPC0 = 0.032; Figure 4), the sediment would serve as a sink for P for 21 out of the 57 sampling points and as a source for the remainder 35 points (assuming stream waters are in contact with sediment at this depth). In comparison, sediments at 183 cm depth behaved as a sink for 49 out of the 57 sampled stream water concentrations. A similar temporal variation in source-sink behavior of sediments was also reported by [36] by valuating sediment EPC0 values against stream water P concentrations through the year. While we did not measure EPC0 at multiple times of the year, other studies suggest that sediment EPC0 could also be temporally variable, driven by sediment conditions and stream water concentrations. For example, EPC0 can increase under reducing conditions due to the loss of crystalline forms of Fe oxides resulting in a release or reduced retention of P [44,50]. EPC0 values could also fluctuate as a result of stream water concentrations with an increase in EPC0 with increasing stream water concentrations, resulting in a reduction of the sediment buffering capacity [51]. On the other hand, hydrologic dilution after storms could result in a release of P from sediments [52]. This dynamic behavior could make determining source-sink behavior of legacy sediments more complicated since such a behavior would now be dictated by stream water concentrations as well as time-variable sediment EPC0 values.

While we were not able to assess the EPC0 values for bed sediments, others have made such assessments and evaluated them against streambank sediment values. EPC0 values for bed sediments at all of the 10 sites in river Wensum in Norfolk, UK [49], were below the stream water SRP concentrations suggesting that bed sediments were always acting as a sink for P. Two studies [36,39] found that bed sediment EPC0 values were greater than those for bank sediments (Table 4). They suggested that, in general, banks had higher proportion of fine grained material, including P-sorbing clay and metal hydro-oxides (and thus lower EPC0), as opposed to bed sediments [36,39]. Additionally, erosion and downstream transport or loss of finer fractions from bed sediments could further increase EPC0 values for bed sediments and thus reduce the sorption potential for bed sediments compared to the original bank sources [39].

**Figure 4.** Comparison of sediment EPC0 values for four depths for stream bank site BEB against stream water PO4 <sup>3</sup><sup>−</sup> concentrations over the period August 2017 to July 2018. Sediment sampling was done only one time in October–November 2017. EPC0 values for four depths 60, 122, 183, and 260 cm from the top were 0.032, 0.028, 0.002, and 0.032, respectively.

#### *3.4. Legacy Sediment Sorption Under Anoxic and Oxic Conditions:*

Our experiment with P-sorbed legacy sediments revealed that the solution PO4 <sup>3</sup><sup>−</sup> concentrations under all sediment treatments under anoxic conditions (mean 1.40 <sup>±</sup> 0.32 mg P L<sup>−</sup>1) were significantly greater (t-test, p < 0.001) than those measured under oxic conditions (mean 0.26 <sup>±</sup> 0.19 mg P L−1) (Figure 5). Similar to studies for other sediments and soils [32,44,53], these results showed that the P-sorbed legacy sediments released more PO4 <sup>3</sup><sup>−</sup> under anoxic conditions than under oxic conditions. Thus, legacy sediments with elevated P could release P in greater amounts under anoxic conditions, effectively acting as a source. Anoxic conditions result in the reduction of Fe (III) to Fe (II) [54,55] releasing the tightly bound P from the iron oxide surfaces [32,44]. Reduced conditions also allow Fe (II) to preferentially bind to sulfide, releasing the PO4 <sup>3</sup><sup>−</sup> ions [29]. Similar work [40] with aerobic and anaerobic treatments however suggested that the amount of inorganic P release could vary with different soils and would depend on the various P fractions associated with oxides. They found that anaerobic conditions released P associated with Fe, but the same did not extend to slowly-cycling P that was associated with Ca, stable P or residual P [40]. Temperature could also be a factor influencing the release of P under anoxic conditions. Soils under warm flooded and anaerobic conditions released more P than similar anoxic soils under cold or unfrozen/frozen conditions [56]. These responses would likely extend to legacy sediments too with some sediments releasing relatively more P than others under varying anoxic conditions.

**Figure 5.** Solution PO4 <sup>3</sup><sup>−</sup> concentrations after a 24-hour incubation of P-sorbed legacy sediments in oxic and anoxic conditions. Site IDs along with their depths (in cm) are listed. Solution PO4 <sup>3</sup><sup>−</sup> concentrations under anoxic conditions (mean = 1.40 <sup>±</sup> 0.32 mg P L<sup>−</sup>1) for sediment treatments were significantly greater (t-test, p < 0.001) than those measured under oxic conditions (mean = 0.26 <sup>±</sup> 0.19 mg P L<sup>−</sup>1). Control treatment did not contain any sediments.
