*3.5. Conceptual Model for Source*/*Sink Behavior of Legacy Sediment P and Broader Implications for Water Quality and Watershed Management*

Evaluation of legacy sediments through a variety of P indices provided important insights into the potential source-sink behavior of legacy sediments under contrasting redox conditions and sediment and water P concentrations. Overall, our results suggest that legacy sediments, which are particularly rich in silts and clays [15,17,22], have a large capacity for P sorption and can act as a sink for P under well-oxygenated conditions and in stream waters with moderate to high levels of dissolved P (e.g., as in the case of agricultural and urban streams). Floodplains could be important reservoirs of such P (e.g., conceptual model in Figure 6). In contrast, if legacy sediments are introduced into stream waters with low dissolved P concentrations (e.g., in forested streams with P concentrations below sediment EPC0), or are deposited in anoxic conditions, e.g., sediment interfaces in lowland streams or at the bottom of ponds and lakes (Figure 6), these locations could become potential "hotspots" [57] for release P to overlying waters. Because of the greater amount of fine fractions, streambank legacy sediments are likely greater sinks of P than the stream bed sediments.

At the full drainage network scale, legacy sediments in headwater streams, with steeper slope gradients, shallower and faster flows, and likely more oxygenated conditions, could potentially serve as net sinks/repositories of P as opposed to lower positions in the drainage network where deeper and slower flows with anoxic conditions may encourage P-release. Increasing stream water P concentrations lower in the drainage network due to urban/agricultural land use and/or inputs from wastewater and other sources could also increase P sorption on legacy sediment surfaces and thus reduce the buffering potential [48]. Another factor that could affect legacy sediment P dynamics lower in the drainage basin, particularly in estuaries, is the effect of salinity on sediment P. Studies report that increasing salinity could result in release of P from sediments, but to varying extents [58,59]. It should also be noted that beyond abiotic sorption-desorption processes, P release or sequestration in sediments could also be influenced by biotic decomposition of P associated with sediment organic matter [44,60]. Typically, however, organic matter or organic carbon (OC) content of legacy sediments are low (OC < 1%; [22]), hence, such decomposition-related P release/source is likely to be small.

**Figure 6.** Conceptual model illustrating how stream bank legacy sediments could act as either a source or sink for P at various fluvial positions. Sorption (P sink) could occur on oxygenated floodplains if sediment EPC0 values are below stream water P concentrations. In contrast, sediments could become a source of P due to reductive dissolution of iron oxides under anoxic conditions or when EPC0 values are greater than stream water P concentrations. Deposition of legacy sediments in the channel could occur due to fluvial and/or subaerial bank erosion or mass wasting of the bank material.

We recognize here that we investigated legacy sediment behavior through controlled laboratory experiments. Field or in-situ conditions could be much more complicated with multiple factors and conditions (e.g., pH, temperature, salinity, oxygen, water flow and diffusion gradients, dry-wet cycles, etc.) simultaneously affecting P processes and dynamics with one factor reinforcing or cancelling out the effects of the other [11,25,61,62]. Because of this, measuring and understanding legacy sediment responses in-situ could be both, important and challenging [63]. Our observations were also based on only 15 sites and additional legacy sediment sites across the mid-Atlantic and elsewhere, particularly across a gradient of land uses and stream water P concentrations extending from the headwaters to the estuaries, could further help generalize and validate our conceptual model.

Given this conceptual model and the potential source-sink effects of legacy sediments on stream water P, one wonders how the presence of large, valley-bottom stores of P-poor legacy sediments have impacted stream water quality over the past century. Use of synthetic fertilizers started increasing around the 1950s and since then sources such as wastewater, sewage, septic, and fertilizer and manure applications on agricultural lands have been contributing elevated P to our surface waters [2,3,64]. Simultaneously, over the same period or even earlier, streambank legacy sediments have been eroding and contributing sediments to the fluvial system as a result of milldam breaches and removals [17]. Key questions then are: how have the interactions of P-poor legacy sediments and P-rich upland waters affected the trajectory of water quality of our surface waters over the past few decades? Have P-poor legacy sediments buffered the full impacts of upland sources on downstream aquatic ecosystems? On the other hand, have fine legacy sediments enriched with upland-P (among other sediment sources), which have been deposited in downstream aquatic systems and bays, become a long-term, internal, source of P (via reductive dissolution)?

Furthermore, many milldams still exist in the mid-Atlantic with stores of legacy sediments upstream of these dams [17,65]. Backing up and pooling of stream waters in mill ponds encourage anoxic conditions in sediments. This raises the question of how do the anoxic conditions influence the fate and release of P associated with millpond and streambank sediments? Moreover, dams are increasingly being removed across the mid-Atlantic and elsewhere [66–69]. An important question that is on the mind of many watershed managers and environmental agencies is how does the removal of the dams influence particulate and dissolved fractions of P stored upstream of the dams in the stream channel and along the banks? Dam removals could reduce anoxic conditions and the associated reductive dissolution of P, but increase erosion of particulate P associated with in-stream and streambank sediments. Addressing these questions and determining the net balance of these fluxes and processes is important if we want to fully understand the complex interactions of human land use legacies on water quality and the health of our ecosystems.

Our observations on P source-sink mechanisms also have important implications for contemporary watershed management and restoration and watershed P budget and regulatory assessments. Many of the streams with legacy sediments, particularly in the mid-Atlantic, are incised and hydrologically disconnected from the streambanks or legacy sediment terraces [14]. Stream restoration efforts are currently underway that enhance the hydrologic connectivity (or exchange) of the stream with the streambanks or floodplains via bank grading, floodplain creation, and/or stream uplift [70–72]. Such restoration efforts should particularly try to leverage the untapped P sorption capacity of legacy sediments by increasing the surface area of floodplain sediments that are exposed to and interact with stream waters, maintaining the floodplains in oxic conditions (so that P is retained on iron oxides), and reducing the exports of fine legacy sediments to downstream flows. If streambank legacy sediments do not contain other contaminants (e.g., metals or organics), these sediments could serve as valuable P sinks and should be used on-site in floodplain creation and other restoration activities and not moved offsite to landfills for disposal (as happens in some restoration projects (personal observations of the first author)). If legacy sediments have to be removed from valley bottoms [73] and if they are not contaminated, one potential opportunity could be to spread them back on croplands. After all, legacy sediments are the silt and clay-rich topsoil of precolonial America. Their use would particularly be beneficial on cropland soils saturated with decades of fertilizer applications of P, also referred to as legacy P [62,74]. Legacy sediments may be able to mitigate the pollution potential of these P-rich soils via sorption and P sequestration.

Given the various source-sink P mechanisms highlighted in this study, watershed P budgets, models, and regulatory assessments will have to account for the inputs of sediment-bound P with bank erosion of legacy sediments, the sorption capacity of these sediments in the fluvial system under oxic conditions (thus removal of solution P via by sediments), stream water P concentrations and the P sorption-desorption potential, and the release of P from sediments into solution under anoxic conditions. The magnitude of each of these processes would likely vary spatially along the length of the drainage network, from the headwaters to the bays. Assessing the *net* effect of legacy sediments on aquatic P would thus require a spatially-distributed, integrated, quantitative assessment or model of each of these processes over the drainage network. Characterizing the role of legacy sediments for fluvial P budgets will become more important and urgent given that climate intensification of storm events and other processes could potentially affect the erosion and inputs of these sediments into aquatic ecosystems [75]. Understanding and quantifying this variability and targeting P hotspots with appropriate best management practices (BMPs) will be key to achieving the P reduction goals established by the Chesapeake Bay and other watershed programs.

#### **4. Conclusions**

Using measured P concentrations and a variety of P indices, we investigated the potential for legacy sediments to act as a source or a sink of P in aquatic ecosystems. This study shows that streambank legacy sediments have low P concentrations and large capacity for P sorption. Overall, under oxic conditions and in streams with moderate to elevated P (in excess of sediment EPC0 values, e.g., agricultural and urban streams), legacy sediments will likely serve as a net sink for P. However, legacy sediments could release P and become a net source under low stream water P conditions and/or anoxic sediment conditions due to reductive dissolution of iron oxides. Thus, the net source-sink effect for P at the catchment or drainage network scale will have to be assessed by accounting for the spatial variability in source-sink behavior with stream water P and redox conditions. Understanding these processes and the balance of these fluxes is critical to understanding the impact of legacy sediments on aquatic ecosystems. Whether they serve as P sources or sink, legacy sediments and the processes described herein need to be considered in nutrient budgets and watershed models that are being implemented to assign BMPs and meet regulatory load reductions for water quality.

**Author Contributions:** Conceptualization, S.I. and N.S.; methodology, N.S., A.L., and G.J.; formal analysis, N.S.; writing—original draft preparation, S.I. and N.S.; writing—review and editing, S.I., N.S., A.L., G.J., and J.K.; funding acquisition, S.I. and J.K. All authors have read and agreed to the published version of the manuscript.

**Funding:** Funding for this project was provided by the US Department of Agriculture via grant # USDA-NIFA 2017-67019-26330.

**Acknowledgments:** We thank the Fair Hill Natural Resources Management Area staff and the University of Delaware Soils Testing Lab for their analytical support. We also recognize the guidance of Maria Pautler and Karen Gartley on the P test indices.

**Conflicts of Interest:** The authors declare no conflict of interest.
