**1. Introduction**

Sediment and sediment-bound nutrients such as phosphorus (P) can be detrimental to the health of aquatic ecosystems and are an important concern for watershed managers and natural resource agencies [1–3]. Fine sediments can decrease light penetration and reduce the primary productivity of aquatic vegetation [4]. Inputs of excess nitrogen (N) and/or P to aquatic systems can cause eutrophication and enhance the production of harmful algal blooms [5]. Subsequently, algal decomposition can reduce dissolved oxygen levels in the water column resulting in fish kills [6,7]. The Chesapeake Bay Program ranks nutrients (N and P) and sediment as the top two polluters for the Bay, and the Chesapeake Bay commission seeks to reduce P inputs by 24% and sediment loads by 20% per year [8]. While billions of dollars are being invested in agricultural management practices (e.g., \$3.6 billion by 2025 in 2010 dollars in the Chesapeake Bay; [9]) such as riparian buffers, cover crops, nutrient management, etc., water quality improvements have not achieved their targets, especially in agricultural watersheds [2,10]. In 2017, the Chesapeake Bay Model indicated that an additional 0.46 million kg of P reductions will be required from agricultural sources per year to meet the watershed management plan goals for the endpoint year 2025 [2]. While much of the current management and regulatory focus is on upland agricultural and urban sources that constitute 56% and 18% of the total bay P loads [2], other potential

sources such as streambanks [11,12] and associated legacy sediments [13,14] have also been gaining increasing attention [15].

Legacy sediments have been defined as deposits of historic sediments that have accumulated in the valley-bottoms of eastern US following European settlement [13,14]. These sediments have resulted from extensive erosion from land clearing and agriculture and the simultaneous construction of mill dams and other structures on streams [13,14,16,17]. Many of the milldams, which were particularly ubiquitous across the mid-Atlantic region (every few kilometers), have now breached or have been removed, leaving incised streams vulnerable to bank erosion [14,17,18] (Figure 1). Not surprisingly, recent studies reveal that streambank erosion of legacy sediments could constitute as much as 50 to 100% of the watershed fine sediment exports [19–21] and need to be accounted for in watershed sediment budgets and management plans. Beyond sediments, however, very few studies have characterized the concentrations of P in legacy sediments and their contributions to sediment-bound and total P budgets [15]. We know of only three studies, including two of our own that have determined P concentrations in legacy sediments and estimated their contributions to watershed P loadings [21–23]. One other study [24] investigated anthropogenic signatures in legacy sediments through the use of elemental ratios. Our work showed that while P concentrations in legacy sediments were on the lower side (25-1293 mg/kg; [22]), streambank legacy sediment contributions to watershed fluxes could be as much as 50% and 32% for sediments and sediment-bound P, respectively [21].

**Figure 1.** Examples of two streambank legacy sediment sites of the 15 sampled in this study which were upstream of now breached mill dams. Left: Nate Sienkiewicz at the streambank at Nature Center Beach (NCB; bank height 2.59 m). Red flags indicate locations for bank sampling. Right: Alyssa Lutgen in front of the eroding/slumping streambank at Scotts Mill Dam (SM2; bank height 2.74 m).

Beyond total amounts, we know even less about the fate of legacy sediment P as it is eroded from streambanks and deposited in the stream channel or transported downstream to receiving water bodies. While there is substantial information on streambank sediments and associated nutrients [11,25], we do not know if legacy sediments, deposited in streams, become sources or sinks of P and how this behavior varies with sediment and stream water P concentrations and redox conditions in stream sediments. A recent report by the Chesapeake Bay Science and Technology Advisory Committee (STAC) [15] highlighted the lack of information on N and P concentrations in legacy sediments and their fate as a major knowledge gap that urgently needs to be addressed. Without consideration of legacy sediments and sediment-bound P, there is concern that we may have an inaccurate assessment of the

watershed loadings and may not be appropriately allocating sediment and nutrient reductions and remedial/management resources [26].

Our interest here was to address these knowledge gaps for legacy sediment P and gain a better understanding of the fate of legacy sediment P in aquatic ecosystems. Key questions we addressed were: (1) What are the concentrations of P in streambank legacy sediments and how do they compare against concentrations for upland soils, stream bed sediments, and water quality P thresholds? (2) What is the fate and net source-sink behavior of sediment-bound P under varying stream water P concentrations and redox conditions? and (3) What are the broader implications of legacy sediment P for water quality and watershed management? To address the first question we collated our recent legacy sediment P concentrations reported by [21,22], compared them to P values reported for other streambank legacy and non-legacy sediment sites, upland soil concentrations, stream bed sediments, and water quality thresholds. Comparisons were performed for total P, bioavailable fraction of P quantified by Mehlich-3 extracts [27], and the % degree of phosphorus saturation (%DPS; [27]). The %DPS has been used to estimate the amount of P sorbed on soils and therefore the potential for P loss via desorption from soils [27].

To address the second question, we analyzed selected legacy sediment samples that were collected [22] using standard analytical methods and indices that have been used to determine the potential for P leaching and its source-sink behavior [28]. These indices included the phosphorus sorption index (PSI, [27,29]) and equilibrium phosphorus concentration (EPC0; [30]). The PSI provides a measure of the maximum phosphorus sorption capacity and can be used to assess the upper limit for P sorption of legacy sediments. In contrast, EPC0 represents the lower end and is the concentration at which there is no net sorption or desorption from the sediment [30,31]. If EPC0 values of sediments are lower than that of the stream water concentrations, the sediments will sorb nutrients from stream water and considered net sinks for P. Inversely, if the sediment EPC0 value is greater than the stream water concentration, sediment will desorb P and thus act as net P source. Thus, the higher the value of EPC0, the greater the vulnerability of P leaching from sediments. In addition to these metrics, we also determined how legacy sediments with known P concentrations responded to oxic and anoxic redox conditions through laboratory incubation assays. The intent here was to investigate if reductive dissolution of P [32] occurred from legacy sediments under anoxic conditions and its extent. These observations were taken together to address the broader implications and water quality challenges for question 3.
