INTRODUCTION
Eutrophication in rivers, lakes, and coastal areas worldwide has
persisted for decades due to intensive agriculture and urbanization
(Diaz & Rosenberg, 2008; Galloway et al., 2004; Le Moal et al., 2019;
Royer et al., 2006; Van Cappellen & Maavara, 2016; Van Meter et al.,
2017). In the U.S. alone, over 60% of U.S. estuaries and coastal water
bodies have been degraded by excessive nutrient inputs (Howarth et al.,
2002). The concentration levels and export patterns of nitrate, a major
dissolved nitrogen (N) species, are essential to understand and quantify
N export from terrestrial to aquatic ecosystems. Export patterns are
often quantified using concentration - discharge (C-Q) relationships
with a power law of C = a Qb . The C-Q
relationships determine loads of solute export (load = C x Q) and
reflect the response of earth systems to hydrological changes. The value
of b indicates different export patterns: a positive, high value
means a flushing pattern where concentrations increase with streamflow,
whereas a negative b indicates decreasing concentrations with
streamflow (Godsey et al., 2009; Moatar et al., 2017; Musolff et al.,
2015). Chemostatic patterns (with absolute b values close to zero) occur
when concentrations vary little compared to hydrological variations.
These export patterns have important implications on how and how much
nitrate exports across different hydrological regimes. High bvalues with pronounced flushing patterns indicate highly sensitive,
escalated export during extreme hydrological events such as flooding.
Agriculture and urbanization have significantly modified subsurface
physical and chemical structures, including the spatial distribution of
N (Figure 1). Agricultural lands are often characterized by shallow flow
via tile drainage and heavy fertilization that enriches N in shallow
soils (Van Meter et al., 2016; Woo & Kumar, 2019). Urban watersheds
represent a different type of human perturbation, often characterized by
impervious surfaces that facilitate surface runoff during storms and
sewer pipes that enhance rapid shallow subsurface flow (Grimm et al.,
2008). Nitrogen in urban watersheds can come from a variety of surface
and underground sources including atmospheric deposition, lawn
fertilizers, and domestic and industrial wastewater (Baker et al., 2001;
Divers et al., 2014). Buried leaky sewer systems often elevate the N
input to urban groundwater (Groffman et al., 2003; Pickett et al.,
2011). These manipulations do not occur in undeveloped, pristine
watersheds, where N content is often low and tightly cycled (Weitzman &
Kaye, 2018). These differences, in particular their vertical
distribution of N content, can lead to distinct C-Q patterns across
different land use conditions.
Given the tremendous human influence, an open question that is key to
understand nutrient export is:How do subsurface physical
and chemical structures from different land uses in governing nitrate
export patterns? Contrasting observations in literature do not converge
to a unifying framework that answers these questions. In agricultural
lands, both dilution and flushing have been observed (Jiang et al.,
2010; Miller et al., 2017). The common perception is that chemostasis or
biogeochemical stationarity prevails in agriculture lands, primarily due
to the large legacy store of nitrogen (N) that induces transport
limitation and buffers concentration variability (Basu et al., 2010;
Basu et al., 2011; Thompson et al., 2011). In undeveloped, forest
watersheds, nitrate concentrations have been shown to often peak during
spring floods (Creed & Band, 1998; Creed et al., 1996; Pellerin et al.,
2012; Sickman et al., 2003), whereas chemostasis has not been as
commonly observed (Huang et al., 2012; Jacobs et al., 2018). Although
much less examined, existing urban studies have observed both dilution
(Duncan et al., 2017) and flushing patterns (Poor & McDonnell, 2007;
Wollheim et al., 2005).
Growing literature has recognized the importance of subsurface
structure, particularly the hydrological and biogeochemical contrasts in
shallow and deeper zones in governing nitrate levels, denitrification
potentials (Bishop et al., 2004; Creed et al., 1996; Kolbe et al.,
2019), and solute export in general (Musolff et al., 2017; Zhi et al.,
2019). Seibert et al. (2009) developed a flow-concentration integration
model that quantitatively links variations of stream chemistry to flow
rates and vertical profile of shallow soil water chemistry. Zhi et al.
(2019) showed that vertical chemical contrasts between shallow soil
water and deeper groundwater can explain diverse export patterns in
multiple watersheds under a gradient of climate, geology, and land cover
conditions. Here we propose the shallow versus deep hypothesis:
the N concentration contrasts in shallow and deep zones determine
nitrate export patterns under different land use conditions . In
particular, we hypothesize that nitrate concentration contrasts in
shallow water (Csw) and deep water (Cdw)
differ in different land uses, and that the magnitude of concentration
contrasts governs export patterns. Here “shallow water” and “deep
water” are broadly defined as waters from different depths contributing
to streams (Figure 1). The shallow water can come from the shallow
surface / subsurface zone that dominates streamflow at high flow; the
deep water may derive from deeper subsurface zones and predominate at
low flow. These shallow and deep waters are loosely defined. In pristine
sites, the shallow water may be the shallow soil water; in agriculture
sites, it can be the combination of surface runoff, tile drainage, and
soil water; in urban lands, it may be the runoff from impervious
surfaces and shallow subsurface pipes. The deep water is the groundwater
that interacts with streams and carries chemical signatures that are
distinctive from shallow water.
If this hypothesis is true, we expect higher concentrations in shallow
water in agricultural lands, leaning more toward a flushing pattern. In
contrast, in urban watersheds, concentrated nutrients accumulated in
leaky pipes in deeper subsurface are often higher than shallow, rapid
runoff on impervious surfaces, leading more toward a dilution pattern
(Duncan et al., 2017). Nitrogen in forests and pristine sites can come
from decomposition from organic matter in shallow soils and leaching
from N-containing rocks (Mayer et al., 2002; Montross et al., 2013). In
addition, they are often tightly cycled. These characteristics can lead
to varied patterns. We test this hypothesis combining a national-scale
synthesis for 228 sites in the United States data synthesis and
mechanism-based reactive transport modeling. Our goal is to propose a
general, simple conceptual model that can explain and quantify export
patterns under diverse conditions.