Nitrogen, phosphorus, and eutrophication in streams - Dodds and Smith (2016)

Introduction to Eutrophication in Streams

Anthropogenic activities have significantly impacted rivers and streams, leading to increased nutrient levels.
Concerns about stream eutrophication have grown as countries adopt nutrient control measures.
Stream eutrophication science is less developed than that for lakes, creating a need to understand how knowledge transfers between lentic (lakes) and lotic (streams) ecosystems.

The N vs. P Controversy in Lakes

A major debate exists regarding the roles of nitrogen (N) and phosphorus (P) in controlling lake eutrophication.
The traditional view suggests P primarily limits freshwater systems, while N limits marine waters, with estuaries being transitional.
This view evolved from historical debates and Schindler's (1974) demonstration that inorganic carbon was not growth-limiting.
Researchers argue N also plays a significant role (Howarth and Paerl 2008, Lewis and Wurtsbaugh 2008, Paerl 2009), while others believe P loading control is sufficient (Schindler and Hecky 2008, Schindler et al. 2008, Schindler 2012).

Evidence for N's Role

Over 50 years of bioassays indicate N limitation or N and P co-limitation (Morris and Lewis 1992, Dodds et al. 1989, Elser et al. 1990, 2007).
Analyses show more algal biomass produced per unit total P when total N to total P ratios (TN:TP) are high (Smith 1982, Prairie et al. 1989).
Dissolved inorganic N can be lost as N_2 via microbial denitrification, reducing N availability (Hill 1979, Paerl et al. 2014).

Arguments for P Primacy

N fixation should compensate for N deficiencies.
Bioassays may not reflect whole-lake responses due to unnatural conditions and short durations.
N-loading control could be unnecessarily expensive (Schindler 1974, Schindler and Hecky 2008, Schindler et al. 2008, Schindler 2012).

Applying Lake-Based Approaches to Streams

The aim is to explore N and P as pollutants in streams and assess the applicability of lake-based water quality management arguments.
This involves discussing how lake approaches apply to streams regarding algal biomass, assessing changes in nutrients and nutrient ratios over time, and considering community responses beyond biomass accumulation.
Results from bioassays, nutrient additions, and analyses of food webs are put in context with N and P enrichment.
Also considers the potential for N fixation to compensate for N deficiencies in stream ecosystems.

Correlations Among Nutrients and Algal Biomass in Flowing Waters

A key advance in eutrophication science was the development of statistical relationships between epilimnetic algal biomass and water column concentrations of TP (Sakamoto 1966, Dillon and Rigler 1974, Vollenweider 1976, Cooke et al. 2005).
Quantitative estimates of in-lake TP concentrations are predicted from nutrient inputs, aiding eutrophication control planning.
Lohman et al. (1992) linked benthic algal biomass and water column nutrients for Missouri streams, broadened by Dodds et al. (2002, 2006) for worldwide streams.
TN and TP correlated with benthic algal biomass up to a point, leveling off at high concentrations.
N and P together described benthic algal biomass better than either alone (Fig. 1).

Practical Applications and Caveats

This relationship guided nutrient control in the Clark Fork River, Montana (Dodds et al. 1997), where benthic algal biomass declined as predicted (Suplee et al. 2012).
The evidence suggests both N and P jointly stimulate benthic algal productivity in streams.
Less algal biomass is expected if both N and P are lowered.
High algal biomass can increase N and P levels because cells contain these elements.
Spatial separation of primary productivity could be one reason the chlorophyll/nutrient relationships are much weaker in streams than in lakes.
Other factors include flooding (Biggs 1995), herbivory, and light attenuation by riparian vegetation.

Specific Studies and Conflicting Findings

Lewis and McCutchan (2010) found no correlation of periphyton biomass with P and a weak correlation with dissolved inorganic N in Colorado streams.
Greater concentration increases could be necessary to increase benthic algal biomass.
Invertebrate grazing could not explain these patterns.
Other factors like algal biomass at the start of the growing season, length of the growing season, and water temperature explained more variation.

Planktonic Algal Biomass

Planktonic algal biomass in rivers is also strongly linked to concentrations of TP in the water column.
Van Nieuwenhuyse and Jones (1996) observed a leveling off of chlorophyll at high P, similar to Dodds et al. (2002).
They did not test whether including total N improved their regression relationship for suspended chlorophyll a.
Further analyses might reveal river phytoplankton biomass depends on both TN and TP, but such analyses have not been attempted.

Overall Conclusions

A strong statistical link exists between both N and P and algal biomass in streams (at least benthic algal biomass).
Both nutrients should be considered in eutrophication management efforts for flowing waters.
There is strong evidence that anthropogenic N and P enrichment has influenced eutrophication-related water quality in fluvial ecosystems.

Nutrient Enrichment in US Rivers

A key issue in eutrophication science is determining baseline conditions before anthropogenic modification of landscapes and nutrient losses from modern agriculture and atmospheric deposition.
Several approaches have been suggested for local determination of reference nutrient levels, with the best approach involving comparisons of present conditions to reference or slightly modified systems.
Many areas of the US do not contain reference-quality streams because nutrient exports have remained unaltered since European colonization in only a small number of relatively isolated watersheds.
Widespread atmospheric deposition of both N and P makes finding a watershed that can be used as a reference even more unlikely.

Approaches to Defining Reference Nutrient Concentrations

The initial approach developed to define reference nutrient concentrations delineated areas expected to naturally have higher or lower concentrations of nutrients.
This spatial delineation led to the idea of nutrient ecoregions (Omernik 1987, 1995); however, some US ecoregions (e.g., the Corn Belt) were defined without available true natural reference conditions.
Given these uncertainties, 2 approaches have since been used to estimate baseline pre-European settlement nutrient concentrations across the United States.
Smith et al. (2003) used the Sparrow model to model expected streamwater nutrient concentrations across the country.
Dodds and Oakes (2004) later linked stream nutrient concentrations to land-use/land-cover data (a multivariate approach similar to that used previously by Omernik 1987) and used these relationships to factor out the anthropogenic influence (i.e., by extrapolating to zero anthropogenic effect using multiple regression methods).

Comparing Baseline Nutrients

The conclusions from these 2 independent approaches agreed broadly.
Subsequently, the predicted baseline nutrients were compared to river surveys by the Environmental Protection Agency (using nationwide data filtered to remove bias, mostly from the 1990s) to estimate the degree of enrichment in US streams (Dodds et al. 2009).
These data revealed that median values of TN and TP have increased over time in most of the ecoregions of the United States, in many cases dramatically (Fig. 2).
We thus conclude that the nutrient enrichment of flowing waters in the United States is a common and real problem.

Range of TN and TP in US Rivers

Data from US rivers collected by the United States Geological Survey (Alexander et al. 1996) reveal a wide range of TN and TP in the water column of US rivers (Fig. 3a).
The observed range of TN and TP concentrations predicts N limitation of algal growth in some stream and river ecosystems and P limitation in others.
A Redfield ratio N:P of 16:1 by moles in general indicates a roughly balanced supply of N and P, and algae assemblages tend to mirror this ratio fairly closely when growing under balanced growth conditions (Hillebrand and Sommer 1999).
Many TN:TP ratios observed in US rivers and streams greatly exceed the Redfield ratio; however, many others are far lower than 16:1 (Fig. 3b).

Nutrient Stoichiometry Changes

The median values of the reference and current values for data from Dodds et al. (2009) can also be used to estimate how nutrient stoichiometry changes in response to nutrient enrichment.
The increases in N and P concentrations have not been proportional, and therefore the degree of N or P limitation in streams could have shifted with anthropogenic inputs; however, this change is not consistent among nutrient ecoregions (Fig. 2).
Although an increase in TN:TP is more common across ecoregions, the sampling protocol for these data is not proportional to the actual number of streams in each ecoregions, and ecoregions vary considerably in total area and drainage density within the ecoregion.

Community Structure and Stoichiometry

These changes in nutrient stoichiometry also can influence both community structure in rivers and streams and the proportion of different elements transported downstream (Justić et al. 1995).
By extension, changes in stoichiometry may change trophic state and food web pathways in rivers and streams.

The Concept of Trophic State in Streams

Stream researchers have long recognized that carbon sources for stream organisms can originate both from within (autochthonous) or outside (allochthonous) the system (Vanotte et al. 1980), this important distinction has less recently transferred to the traditional concept of trophic states.
Traditionally, the trophic state of lakes has been based on phytoplankton biomass concentration, which was thought to represent the main food source for consumers in the food web.
Thus, eutrophication science has historically been focused on nutrient stimulation of algal productivity (with biomass serving as a proxy for production).
The role of allochthonous materials in food webs of lakes and oceans, however, is now recognized as substantial, and their contribution can at times be greater than autochthonous production (Dodds and Cole 2007).

A New View of Eutrophication

A new view of eutrophication, based on the Greek root of the word eutrophia for food, is therefore necessary to capture the full ecological consequences of nutrient enrichment in streams because nutrient availability can influence the processing of detritus as well as the quality of algal food sources for consumer organisms.
This view adds more complexity to trophic state assessment approaches based on measured concentrations of nutrients and chlorophyll alone (e.g., Table 1).
Dividing the concept of trophic state into autotrophic and heterotrophic facets makes clear that both internal and external sources of carbon to stream food webs should be considered to best understand energy flows through these ecosystems (Dodds 2007).

Heterotrophic Processes

The classic definition of trophic state based on phytoplankton for lakes was initially transferred directly to benthic algal biomass in streams, and the links between algal biomass and nutrient additions were the only factors considered to respond to nutrient enrichment.
Many heterotrophic processes, such as carbon cycling, however, can potentially be limited by the supply rates of inorganic nutrients because many sources of detritus are carbon-rich and N- and P-poor (e.g., Ferreira et al. 2015).

Bioassays in Streams

Several reviewers have considered bioassays in streams, typically based on the use of nutrient-diffusing substrata such as permeable clay pots (Pringle and Bowers 1984) or plastic containers with permeable tops (Tank and Webster 1998) filled with nutrient-enriched agar and placed in streams.
After an in situ colonization and incubation period of several weeks, these substrata are removed and analyzed for attached chlorophyll density on the solid surfaces.
In general, early chlorophyll-based bioassays demonstrated that responses could be found to N alone, P alone, N+P, or neither nutrient (e.g., Francoeur 2001, Elser et al. 2007).

Bioassay Results

Although no response was fairly common, growth responses to N alone or N+P were also frequently observed.
These data suggest that, at least in the short-term, benthic algal biomass production in streams can be limited by factors other than N and P alone.
Additional twists to enrichment bioassays include adding a diffusing substrate made of thin wood (Tank and Webster 1998) to simulate natural detrital surfaces.
Researchers have also started to measure variables in addition to chlorophyll, including measurements of ergosterol content for fungal biomass response (Tank and Dodds 2003) and analyses of gross primary production and respiration (Johnson et al. 2009).

Biofilm Studies

Johnson et al. (2009) explored nutrient responses using both inert substrata (porous glass) and wood veneers.
They monitored respiration, gross primary production, and chlorophyll as the response variables (Fig. 4) from 62 sites in reference, agricultural, and urban settings across the United States and found several interesting patterns:
- (1) the chlorophyll response varied across substrata, but N limitation and N and P co-limitation were indicated as for prior bioassays;
- (2) the primary production and respiration responses could be stimulated by factors other than P alone, and the 2 responses could vary at the same site (i.e., heterotrophic and autotrophic responses did not always match);
- (3) responses varied on wood and glass substrata, indicating that the carbon in the wood or physical differences in the 2 substrata led to the development of different biofilm communities that in turn responded differently to nutrient enrichment.

Stream Bioassays

Given the observed diversity of responses to nutrient enrichment, stream bioassays unfortunately tend to leave us in much the same position as in lakes.
What if, as Schindler (2012) claims, short-term bioassay results are simply not indicative of long-term responses?
Whole-stream nutrient enrichment experiments for streams are far less common than for lakes; some studies have reported increases in algal production, but not all of these responses are completely attributable to P enrichment alone (e.g., Stockner and Shortreed 1978, Perrin et. al. 1987, Peterson et al. 1993, Borchardt 1996).

Whole-Stream Nutrient Additions

More recently, some interesting results have indicated that both nutrients can be important, not only to whole-ecosystem algal biomass but also to whole-ecosystem carbon processing and transport (e.g., Gulis and Suberkropp 2003, Ferreira et al. 2006).
For example, a series of whole-stream nutrient additions in the US Adirondacks was recently performed at varied N:P ratios, and multiple response variables including rates of litter breakdown were followed.
In that study, both N and P played a role in stimulating leaf decomposition and altering the rate of carbon retention (Rosemond et al. 2015).
Additional information indicates that nutrients can alter biotic properties of stream animal communities, perhaps independent of more traditional influences on algal productivity and litter breakdown rates.

Impacts on Stream Communities

The data of Wang et al. (2007) suggest that increased N levels as well as P levels can decrease the diversity of key groups of fishes as well as invertebrates (Wang et al. 2007).
In a similar study, Evans-White et al. (2009) found that invertebrate primary consumer diversity decreased steeply with increasing P, eventually leveling out to relatively low diversity.
Evans-White et al. (2009) hypothesized that these responses were a result of altered food quality mediated through eutrophication-related changes in the stoichiometry of the consumers’ food resources.

Stoichiometric Shifts

The weaker relationship for predators (who eat other invertebrates with more constrained stoichiometric compositions than basal food sources) than for primary consumers was considered to be correlative evidence for this argument.
Stoichiometric shifts in food quality are documented to have negative impacts on grazers and higher trophic levels in lakes (Hessen 2013), and there is no reason to assume similarly mechanisms related to sto- ichiometric shifts would not apply in streams as well (e.g., Frost et al. 2002).

Importance of N and P

Broadening the view of trophic state in streams beyond considering only stimulatory effects on autotrophs opens more avenues for both N and P to be important system drivers.
The concept of heterotrophic state in addition to autotrophic state (Dodds 2006) is also encapsulated in the terrestrial literature by the recent concept of green and brown food webs (e.g., Wu et al. 2011).
Understanding the importance of the brown food webs has been clear in stream ecology for canopy-covered systems as well as turbid large rivers (Vanotte et al. 1980).
Data considered under this broadened view suggest that N as well as P, or N+P, are important determinants of system activities in streams, and thus eutrophication by both N and P deserves consideration.

Cyanobacteria and N Fixation in Streams

Cyanobacteria are widespread in the world’s flowing waters (Scott and Marcarelli 2012), and slow-moving, nutrient-enriched rivers and streams have been observed to develop planktonic blooms of potentially harmful cyanobacteria (CyanoHABs) analogous to those found in lakes (e.g., Murray River, Australia: Baker and Humpage 1994; Neuse River: Affourtit et al. 2001).
Where water clarity permits the formation of benthic growth, however, stream algal biomass is more frequently dominated by substrate-attached cyanobacteria when these organisms are present.
Nitrogen-fixing, heterocystous cyanobacteria observed in the benthic habitats of streams include Rivularia, Nostoc, Scytonema, Calothrix, and Gleotrichia.
Additionally, Epipthemia containing N-fixing endosymbionts is common in some streams (e.g., Richardson et al 2009).

Heterocystous Genera

Heterocystous genera of benthic algae, however, are seldom found in highly enriched waters, although some dominate in naturally high P–low N waters (e.g., Nostoc; Dodds and Castenholz 1988).
Landcare Research, 1 of 7 Crown Research Institutes in New Zealand, view heterocystous species of cyanobacteria as mostly occurring in high-quality New Zealand streams (http:// www.landcareresearch.co.nz/resources/identification/ algae/identification-guide, accessed 2 July 2015).
A key issue in eutrophication science is the potential for N-fixing cyanobacteria to compensate for any deficiency in biologically available N.

N Fixation

Can heterocystous cyanobacteria dominate N cycling, and will the local ecosystem retain this new N once it is fixed?
Unfortunately, N fixation is much less frequently measured than other N-cycling rates (e.g., nitrate and ammonium uptake and denitrification), and reviews of N budgets for flowing waters suggest that N-fixation rates are rarely measured in rivers and streams (Marcarelli et al. 2008).
In stream studies that measured both N uptake and fixation, the rates of N fixation have rarely been observed to equal or exceed rates of dissolved inorganic N uptake (Marcarelli et al. 2008); however, advective N fluxes and frequent seasonal scouring of biofilms transport much of the newly fixed N out of the local system.

Retention Times

In contrast to lakes, where retention times may be months or years, stream networks often have retention times on the order of days, making it unlikely that the slow process of N fixation can ultimately satisfy total biological demands for inorganic N.
We note in the experimental stream study by Stelzer and Lamberti (2001) in which nutrients were manipulated for a month, low N:P supply ratios did not lead to a shift to cyanobacterial dominance, in direct contrast to the expected response of lake phytoplankton.
Scott and Marcarelli (2012) considered grazing and scouring more important determinants of benthic cyanobacterial dominance in streams than nutrient conditions.

Denitrification

Moreover, because streams are dominated by benthic habitats, the potential for denitrification is high (Mulholland et al. 2008), perhaps leading to greater proportional N losses in the N budgets of streams relative to lakes.
We provisionally conclude that the probability that N fixation can eventually compensate for N-limited conditions induced by excess P loading is much lower in streams than it is in lakes, but this hypothesis should be directly tested.

Conclusions

Although eutrophication science in fluvial ecosystems lags well behind that for lakes, major advances are being made.
We present broad, multiple lines of evidence that note the importance of both N and P in stream trophic state.
This evidence includes statistical correlations, small scale bioassays, and whole-stream enrichment experiments.
Both autotrophs and heterotrophs seem to be influenced by changes of nutrients.
Both N and P pollution are common in the United States (and elsewhere), and the relative increases are spatially distinct.
Any view of nutrient limitation that focuses on P alone in streams will miss much of the nuance of nutrient limitation for primary producers as well as for detritivores and other heterotrophs higher in the food web.
We thus urge further research on the impacts of nutrient enrichment on fluvial ecosystems.

Further Research

We also stress that multiyear nutrient water quality databases exist for large numbers of rivers and streams located worldwide and that comparative analyses of these data are likely to provide important new insights.
In particular, we urge our colleagues to perform studies that parallel Minaudo et al. (2015), who analyzed 30 years of data documenting the recovery of the Loire River (France) from eutrophication.
It will be crucial to examine the speed with which eutrophic flowing waters respond to N and P loading controls and the degree to which hysteresis effects can be expected to occur during the eutrophication recovery process (Jarvie et al. 2013).