Seawater microbes as well as those associated with macrobiota are
increasingly recognized as a key feature affecting nutrient cycling.
Tidepools are ideal natural mesocosms to test macrofauna and microbe
interactions, and we quantified rates of microbial nitrogen processing using
tracer enrichment of ammonium (
Nitrogen cycle processes are carried out by a diversity of taxa, from microbes to macrofauna, that can all reside in the same habitat. Nevertheless, most studies tend to focus on characterizing and/or measuring the rate of only a single transformation at a time (e.g., nitrification or nitrate reduction), despite the co-occurrence of a diversity of nitrogen processes, including those leading to loss or retention. Given an anthropogenic doubling over the past century of the supply rate of biologically available nitrogen to ecosystems (Galloway et al., 2008; Fowler et al., 2013) simultaneous with accelerated harvest of animals that recycle nitrogen (Worm et al., 2006; Maranger et al., 2008), it is essential that we understand how microbes and macrobiota interact to influence nitrogen cycling. Using the experimental tractability of rocky shore tidepools as natural mesocosms, coupled with isotope tracer enrichments and mathematical modeling, we estimate here the rate of simultaneous nitrogen transformations as a function of animal abundance and time of day.
Along upwelling shores, the paradigm of productivity driven by upwelled nitrate has been challenged by studies quantifying the effects of animal excretion and regeneration (Dugdale and Goering, 1967; Aquilino et al., 2009; Pather et al., 2014). It is well known that nitrogen regeneration is quantitatively significant in a variety of ecosystems (Schindler et al., 2001; Vanni, 2002; Allgeier et al., 2014; Subalusky et al., 2014), However, to make a significant contribution to productivity, uptake of animal excreted ammonium by photo- and chemolithotrophs needs to be sufficiently rapid to retain nitrogen locally to avoid dispersion into the larger environment.
Microbial nitrogen transformations are diverse, converting inorganic nitrogen
among different biologically available (NH
There is parallel evidence that marine animals host diverse microbiomes (Pfister et al., 2010, 2014b; Miranda et al., 2013; Moulton et al., 2016) as well as stimulate phototrophs with excreted nitrogen (Taylor and Rees, 1998; Plaganyi and Branch, 2000; Bracken, 2004). Incubating seawater or sediment separate from the natural environment has provided controlled estimates of single nitrogen transformations (e.g., Yool et al., 2007). However, a principle challenge has been quantifying in situ the simultaneous nitrogen transformations that characterize natural communities. Animal species may host nitrogen-metabolizing microbes while phototrophs in the same environmental setting simultaneously compete for the animals' excreted ammonium. Light levels controlling phototroph ammonium uptake may thus mediate nitrogen transformations.
Stable isotope enrichment experiments are an established methodology for
quantifying nitrogen processing in marine environments where the transfer of
a tracer between source and product pools is measured over time (Glibert et
al., 1982; Lipschultz, 2008). Typically, these assays are done on seawater or
sediments (e.g., review by Beman et al., 2011), though there are some examples
where an organism is assayed (e.g., Heisterkamp et al., 2013). One
acknowledged challenge of these experiments is the simultaneous occurrence
of multiple processes that can isotopically dilute the source pool. For
example, in a
A schematic of the nitrogen cycling model used in this study, where
microbial processes include
Here, we quantify the influence of a common coastal marine animal, the California mussel, on the overall magnitude of and the partitioning between simultaneous nitrogen transformations, using tidepools at low tide as experimental mesocosms. We use an experimental approach to test the possible interacting roles of this animal and light on the rates of nitrogen transformations that, in particular, influence net nitrogen retention. We manipulated the presence and absence of mussels and light in combination with stable isotope tracer addition to directly test their effects on nitrogen transformations. Microbial nitrogen transformations estimated from differential equation models were much higher than published rates for which rate estimates are treated singularly. We use the experiment and model together to test whether nitrogen transformations in the tidepools are elevated by mussels, inhibited by light or affected by other environmental variables. We also test for evidence of interactions between phototrophs and nitrogen-utilizing microbes.
A list of observed and modeled parameters used in this study.
The fates of three forms of inorganic nitrogen (ammonium, nitrite, nitrate) in
an isolated tidepool include a variety of processes mediated by microbes and
other intertidal inhabitants, and are illustrated in Fig. 1. For ammonium,
increases in concentration (and dilution of an enriched tracer) can occur
via excretion by animals and is represented by remineralization (
The traditional source–product model generally involves estimating an
average rate from time 0 to time
A recognized shortcoming of Eq. (1) is that multiple simultaneous processes
(e.g., Fig. 1) can change the concentration and isotopic composition of
source and product nitrogen pools (Lipschultz, 2008). Resolving the influence
of multiple, contemporaneous nitrogen transformations requires a new
approach that accounts for their influence over time on the distribution of
In our differential equation model (Fig. 1), three differential equations
describe how the concentrations of ammonium [A], nitrite [Ni] and nitrate
[Na] in nmol L
Three equations model the time-varying concentrations (nmol
We solved Eqs. (2–7) for the six parameters (
All isotope enrichment experiments were done in tidepools at Second Beach, a
rocky north-facing bench 2 km east of Neah Bay, WA, USA (48
Four
In all experiments, a water sample prior to tracer addition was collected to
verify natural abundance isotope levels (
Each tidepool experiment had three time points for nitrogen isotope composition
and concentration, making it possible to fit our model to the data for each
experiment. We solved our differential equations using the ODE function of
R (in the deSolve R package, Version 3.1.0,
Finally, we compared our differential equation model with the source–product
model shown in Eq. (1). Because our tracer experiments had three time points
(
We measured multiple responses in our experimental manipulation. We analyzed
all responses with a linear mixed effects model using tidepool as a random
effect and testing for a statistical interaction between mussel presence and
light (R,
The ending measured concentrations (in
A statistical summary of the role of mussels and day vs. night
on resulting seawater chemistry and temperature immediately prior to
tidepool re-inundation. We used linear mixed effects models with tidepool as
a random effect and log-transformed estimates for nutrient concentration;
After approximately 5–6 h of isolation at low tide, results were
dependent on both the presence of mussels and the availability of sunlight
(Fig. 2, Table 2). Ammonium concentration was overall greater with mussels
and during the day, and oxygen, temperature and pH all tended to be greater
during the day. Tidepool pH was lower at night (
ODE modeled
The advantage of using our tidepool experiments is that they contain the full
range of actual biological components and environmental fluctuations; but as
they vary in the composition of these components they also show individual
differences. We thus fit the model to each tidepool individually, rather than
a mean value, allowing any influences due to environmental differences to be
incorporated into parameter estimates. ODE model predictions were generally
highly concordant with the observed nutrient and isotope data measured for
each tidepool experiment (Fig. 3). Our estimates of
The relationship between the predicted total
A summary of all estimated rates by treatment in the ODE model
(Eqs. 2–7). Means and (se) are shown with
The estimated rates (nmol L
The rates of ammonium remineralization in tidepools that we estimated with
our ODE model were greatest during the day when mussels were present, as was
the uptake of ammonium (Fig. 2). In turn, all nitrogen metabolisms showed
the greatest rates in the presence of mussels (Fig. 5, Table 3, Table 4).
Further, all nitrogen transformations were greatest during the day with the
exception of nitrate reduction. For ammonium and nitrite oxidation (
A statistical summary of the role of mussels, day vs. night and
their interaction on the rates of nitrogen transformations (in nmol L
All rates of nitrogen transformation during the day and with mussels
estimated with our differential equation model (Eqs. 2–7) were greater than
those estimated by the traditional source–product model (Fig. 5, Table 4). The ODE
model always produced an estimate of the ammonium oxidation rate far greater
than that of the source–product model, particularly during the day. The
ammonium oxidation rate estimated with our differential equation model was
uncorrelated with the estimates from the source–product model (Spearman's
Parameter estimates from our model allowed us to assess the potential
interaction among nitrogen processes. We tested how model estimates of
photosynthetic vs. microbial chemolithotrophic nitrogen use were related.
If competition for ammonium occurs, then ammonium oxidation (
Finally, we found few correlations between nitrogen transformation rates and
oxygen, temperature and pH in tidepools at the end of the low tide period.
Only remineralization and nitrogen uptake rates show a positive correlation
with higher temperatures (
The remineralization of ammonium, oxidation and reduction of inorganic
nitrogen, and the uptake of ammonium and nitrate were all greater in
tidepools with mussels vs. those where mussels were removed. Mean nitrate
flux due to microbial processing (the sum of microbial nitrate
transformations in Table 3) ranged from 8 to 61 % of the total nitrate
uptake attributed to both microbes and phototrophs, with the highest values
when mussels were present and it was daylight. Microbial processing
accounted for an average 32 % of the total ammonium flux with mussels and
daylight. Processing of both nitrate and ammonium by microbial
chemolithotrophs was thus significant in this rocky shore environment, and
especially so when mussels were present. Previous analysis of ammonium
uptake in this system indicated that suspended particles (e.g.,
phytoplankton) in tidepool seawater account for a negligible amount of
ammonium uptake (only 1–3 nmol L
Previous genomic analyses showed that inert substrates (e.g., rocks) in
tidepools with mussels host a nearly identical microbial community to those
in tidepools without mussels (Pfister et al., 2014b), while mussel shells
themselves host a rich diversity of nitrogen-metabolizing microbial taxa
(Pfister et al., 2010). Combined with the nitrogen processing rates we
quantified here, these studies suggest that California mussels are loci for
the microbial processing of nitrogen. Marine invertebrates acting as hosts for
significant nitrogen processing is further supported by work with snails and
other bivalves, which are demonstrated sites of nitrogen transformations
including ammonium oxidation (Welsh and Castadelli, 2004; Stief et al., 2013;
Heisterkamp et al., 2013). N
In high-energy coastal environments, animal-regenerated ammonium could be
advected by waves and currents rather than retained. Because the rates we
quantified are rapid, and because tidepool habitats are high flow refugia,
net retention of inorganic nitrogen in nearshore areas can result, and is a
phenomenon that is likely to enhance local primary production. Over a diel
cycle, both ammonium and nitrite oxidation and nitrate and nitrite reduction
occurred, and all are processes that retain dissolved and biologically
available nitrogen. Although we did not follow our tracer into all tidepool
species, previous analyses showed it was readily incorporated into tidepool
algae (Pather et al., 2014). Nitrogen loss processes were not quantified,
though other experiments with gas-tight chambers indicated no loss of
nitrogen via enriched N
Both nitrate and nitrite reduction rates were significant and are evidence
for incomplete denitrification or DNRA processes thought to be occurring
only at low oxygen. Even during daytime periods of high oxygen, nitrate and
nitrite reduction were observed, suggesting that tidepools provide
microsites where these microbial reducing processes can take place. The
oxidation of ammonium and nitrite, though not positively related to final
oxygen level, was greatest during the day and with mussels. Even at night
when oxygen could be very low, there was sufficient ambient oxygen to permit
nitrification. Thus, even though remineralization decreased at night and
oxygen levels dropped, presumably associated with decreased mussel
metabolism, ammonium oxidation remained at an average of 160.6 nmol L
Although competition for ammonium between nitrifiers and phototrophs is
poorly understood, the preference for ammonium uptake may make it a
contested resource. Sediment microalgae have been shown to be competitively
superior to ammonium oxidizing bacteria, likely due to higher specific
uptake rates and faster growth (Risgaard-Petersen et al., 2004). Here, we
found little evidence for competitive interactions for either ammonium or
nitrate between photosynthetic processes and microbial chemolithotrophs.
Microbial transformations in the dark did not increase, suggesting that
microbial nitrogen metabolism is driven more by the stimulation of animal
excretion that occurs in these tidepools during the day, perhaps because of
increased tidepool temperature (Bayne and Scullard, 1977). We also show no
evidence of UV inhibition of nitrification (e.g., Horrigan and Springer,
1990; Guerrero and Jones, 1995). We note also that tidepool ammonium levels
rarely were lower than several
We developed the ODE model to simultaneously estimate multiple microbial
transformation rates which provide a more realistic descriptor of microbial
activity in nature. Our model's focus on the rates of simultaneous nitrogen
transformations assures that it is general and applicable to any system. A
key result here is that rate estimates from the differential equation model
were often much greater than those from the source–product model (Lipschultz,
2008; and Glibert et al., 1982). We suggest two reasons that our ODE estimated
greater rates. First, the rapidity of microbial transformations combined
with the diversity of microsites in nature means that tracer enrichment can
readily cycle through multiple products. Thus,
Tidepools demonstrated a range of prokaryotic and eukaryotic nitrogen metabolisms that varied with animal presence and the time of day, echoing other recent studies that demonstrated that marine animals serve as sites for a diversity of nitrogen metabolisms (Fiore et al., 2010; Heisterkamp et al., 2013). The ubiquity in the coastal environment of the flora and fauna found in tidepools suggests that microbial nitrogen transformations are not unique to tidepools but a general feature associated with macrobiota. The relatively high variability in the estimates of all microbial nitrogen transformations we documented is paralleled by variability in the environmental variables (e.g., oxygen, pH, temperature, species composition) that may also foster a rich mosaic of tidepool microsites for microbial biogeochemical processing and nitrogen regeneration and retention. Scaling up to the entire rocky shore ecosystem suggests a large potential role for animals in ameliorating fluctuations in upwelling and nutrient delivery. Meanwhile, ongoing animal harvest in ocean systems has greatly impacted nitrogen cycling (e.g., Maranger et al., 2008), making it imperative to understand the links between nitrogen in coastal systems and animal harvest.
Catherine A. Pfister, Mark A. Altabet and Santhiska Pather designed the experiments and Catherine A. Pfister, Mark A. Altabet and Santhiska Pather carried them out and did laboratory analyses. Catherine A. Pfister, Greg Dwyer and Mark A. Altabet developed the model. Catherine A. Pfister prepared the manuscript with contributions from all co-authors.
We thank E. Altabet, S. Betcher, B. Colson, M. Kanichy, O. Moulton, P. Zaykowski for making the field experiment a success, including R. Belanger's lab efforts. K. Krogsland provided nutrient analyses and J. Larkum laboratory isotope expertise. We thank the Makah Nation for access. Funding was provided by NSF-OCE 09-28232 (CAP), NSF-OCE 09-28152 (MAA), and a Fulbright Foreign Student Award (SP). Edited by: J. Middelburg