FEMS Microbiology Ecology, 91, 2015, fiv041
doi: 10.1093/femsec/fiv041
Advance Access Publication Date: 6 April 2015
Research Article
RESEARCH ARTICLE
Iron supply constrains producer communities in
stream ecosystems
Chad A. Larson1 , Hongsheng Liu and Sophia I. Passy∗
Department of Biology, University of Texas at Arlington, Arlington, TX 76019, USA
∗ Corresponding author: Department of Biology, University of Texas at Arlington, Arlington, TX 76019, USA. Tel:+817-272-2415;
E-mail: [email protected]
Present address: Department of Ecology State of Washington, Olympia, Washington.
One sentence summary: This is the first continental and experimental investigation to demonstrate that iron limitation is potentially widespread in US
streams and can have negative impacts on producer biodiversity and biomass accumulation.
Editor: Riks Laanbroek
1
ABSTRACT
The current paradigm that stream producers are under exclusive macronutrient control was recently challenged by
continental studies, demonstrating that iron supply constrained diatom biodiversity and energy flows. Using algal
abundance and water chemistry data from the National Water-Quality Assessment Program, we determined for the first
time community thresholds along iron gradients in non-acidic running waters, i.e. 30–79.5 μg L−1 and 70–120 μg L−1 in
oligotrophic and eutrophic streams, respectively. Given that Fe concentrations fell below both thresholds in 50% of US
streams, and below the eutrophic threshold in 75% of US streams, we suggest that Fe limitation is potentially widespread
and attribute it to the restricted distribution of wetlands. We also report results from the first laboratory experiments on
algal-iron interactions in streams, revealing that iron supplementation leads to significant biovolume and biodiversity
increase in both nitrogen fixing and non-nitrogen fixing algae. Therefore, the progressive brownification of freshwaters due
to rising dissolved organic carbon and iron levels can have a stimulating influence on microbial producers with cascading
effects along the trophic hierarchy. Future research in running waters should focus on the role of iron in algal physiology
and biofilm functions, including accumulation of biomass, fixing atmospheric nitrogen and improving water quality.
Keywords: algae; biodiversity; biomass accumulation; brownification; eutrophication; iron; nitrogen fixers; nutrient
colimitation; streams
INTRODUCTION
The ‘iron age in oceanography’ began in the late 1980s with the
discovery that iron limited the growth of oceanic phytoplankton, which has since been confirmed in 40% of the world’s ocean
(de Baar et al. 2005). Numerous investigations over the following decades concluded that Fe limitation had a strong impact
on oceanic algal carbon uptake, composition and productivity
(Martin and Fitzwater 1988; Coale et al. 1996; Boyd et al. 2000). The
paradigm in freshwater research, on the other hand, is that photosynthetic communities are products of exclusive macronutrient control (Borchardt 1996; Elser et al. 2007; Sterner 2008).
Recent continental studies have challenged this paradigm by
showing that iron supply constrains the energy flows in diatom
communities (Passy 2012), as well as their biodiversity at scales
ranging from individual stream reaches to entire watersheds
(Passy 2008; Passy 2009; Passy 2010). However, unlike the ocean,
experimental research on the algal-iron relationships in streams
is lacking, despite evidence of a significant increase in periphyton biomass in response to micronutrient addition (Pringle et al.
1986). Although iron is one of the most abundant elements on
Earth, it is often limiting to producers across aquatic ecosystems, including the open ocean (Martin and Fitzwater 1988;
Received: 3 February 2015; Accepted: 1 April 2015
C FEMS 2015. All rights reserved. For permissions, please e-mail: [email protected]
1
2
FEMS Microbiology Ecology, 2015, Vol. 91, No. 5
Falkowski 1997), lakes (North et al. 2007; Downs, Schallenberg
and Burns 2008) and streams due to low terrestrial inputs and/or
poor solubility. As running waters are often non-humic, nonacidic and oxygen-rich, dissolved Fe quickly precipitates as
amorphous non-bioavailable forms (Vuori 1995).
Iron supply also plays a crucial role in the nitrogen and carbon cycles through its influence on N2 fixation and primary productivity (Falkowski 1997). Nitrogen fixation is carried out by
diazotrophs (nitrogen fixers) by means of an iron-rich multimeric enzyme complex, nitrogenase (Berman-Frank, Lundgren
and Falkowski 2003). Oceanic investigations have shown that
iron availability controls N2 fixation by constraining nitrogenase
synthesis and activity as well as growth and photosynthesis of
diazotrophs (Paerl, Prufertbebout and Guo 1994; Mills et al. 2004;
Moore et al. 2009). Although the taxonomic, morphological and
physiological diversity of photosynthetic diazotrophs is much
greater in freshwaters than in the open ocean (Berman-Frank,
Lundgren and Falkowski 2003), the influence of Fe on stream diazotroph composition is completely unknown. This is a substantial gap in our knowledge of the nitrogen cycle, considering that
streams are often N-limited (Vitousek and Howarth 1991; Elser
et al. 2007) and N2 fixation is an important autogenic source of
new nitrogen in these systems (Grimm 1994; Grimm and Petrone
1997).
Eutrophication and micronutrient enrichment represent major and generally independent gradients in freshwaters, i.e. eutrophic streams in the USA can be Fe-poor or Fe-rich (Passy
2012) and eutrophic lakes across the globe can be micronutrientsufficient or deficient (Downs, Schallenberg and Burns 2008).
However, unlike eutrophication, which has been broadly studied (Smith and Schindler 2009; Evans-White, Haggard and Scott
2013), the extent of Fe limitation for stream producers is unknown. Considering that Fe supply is controlled by wetlands
(Dillon and Molot 1997; Passy 2010) and large wetlands have
a very restricted geographic distribution (Mitsch and Gosselink
2007), we expect most streams to be Fe-deficient. Here, we confirmed this prediction by analyzing iron and land cover data
from continental stream and watershed surveys, respectively,
carried out by the US Geological Survey as part of the National
Water-Quality Assessment (NAWQA) Program.
Biofilm algae are important contributors of organic carbon
in many small streams and large rivers (Stevenson 1996; Vis
et al. 2007), supporting higher trophic levels and maintaining water quality. Nevertheless, our understanding of how their communities function under the array of nutrient limitations and
colimitations apparently present in natural running waters is
inadequate. To address this issue, we performed a continentalscale analysis of microbial compositional responses along iron
gradients in streams of varying macronutrient supply (nitrate
and phosphate). We expected that the community Fe threshold
would be greater under eutrophic than under oligotrophic conditions because, by definition, eutrophic species have a much
higher macronutrient demand than oligotrophic producers, and
may show a similarly higher Fe-dependence.
We also carried out the first experimental investigation in
the stream benthos, testing the independent and interactive
effects of Fe and macronutrients on biodiversity and biomass
accumulation of both diazotrophs and non-diazotrophs. We
hypothesized that Fe would have a strong independent effect on
biodiversity, because each limiting resource provides new opportunities for tradeoff and coexistence (Harpole and Tilman 2007;
Passy 2008). Conversely, biomass accumulation, which relies on
macronutrients for building proteins, nucleic acids and phospholipids would show only a weak response to enrichment with
Fe only. However, maximum producer richness and biomass
would be observed when all resources were replete, including
N, P and Fe, consistent with the benthic model of coexistence,
which predicts that adding nutrients at high supply increases
the niche dimensionality of the algal habitat and produces thick
and speciose biofilms (Passy 2008). Given the substantial energy and Fe requirements of N2 fixation, we also expected that
high levels of both P and Fe were necessary for establishing a
diverse and abundant diazotroph community under N-limiting
conditions.
MATERIALS AND METHODS
The NAWQA Program data
Proportional area of all major land cover types in 2946 US watersheds and total dissolved iron in 2437 distinct stream localities were measured by the USGS as part of the NAWQA
Program (http://water.usgs.gov/nawqa). Both land cover and
iron data were available for 1670 streams. Algal communities
were examined in a subset of 392 stream localities, spanning
37.42 latitudinal and 78.55 longitudinal degrees. Of these localities, 162 were classified as oligotrophic (nitrate ≤ 245 μg L−1
and phosphate ≤ 22 μg L−1 ) and 230, as eutrophic (NO3 − >
245 μg L−1 and PO4 3− > 22 μg L−1 ), following Hill and Fanta
(2008) and Passy (2008). The 392 streams were sampled one
to six times between 1993 and 2006 and 600 quantitative algal samples were collected from the richest-targeted habitats.
These habitats harbor the most diverse periphytic assemblages
within the reach, including epilithon, epiphyton and epidendron. Algal collection, processing and enumeration followed
established protocols (http://pubs.usgs.gov/of/2002/ofr-02-150/).
Streams sampled for algae also had data for the month of
algal collection on total Fe, NO3 − -N and PO4 3− -P (all water filtered), generated by the NAWQA Program according to
Fishman and Friedman (1989). If multiple measurements of the
studied nutrients were taken during this month, they were
averaged.
Experimental data
Microcosm study
In February–March 2011 and September–October 2011, we performed two nutrient manipulation experiments in a facility with
24 stream microcosms (see Fig. S1, Supporting Information).
These experiments were conducted in different seasons to test
the generality of the observed patterns. The microcosms are
round glass dishes, measuring 30.5 cm in diameter × 10.2 cm
in height and holding 4.5 L of water. In each microcosm, an 8.9
cm propeller mounted on an IKA RW-20 Digital Overhead StirR
rer (IKA
Works, Inc., Wilmington, North Carolina, USA) was
placed just above the substrate in the center of the dish and
set at 600 rpm, creating a constant current velocity of 8 cm
R
sec−1 . The stainless steel propellers were coated with Plasti Dip
R
Primer, then multiple times with Plasti Dip , after which they
were placed in water for 2 weeks. Chemical analysis of this water as well as in control treatments (without Fe) found no traces
of Fe. The microcosms were illuminated by 250 W metal halide
lamps for 14 h daily at levels sufficient for photosynthesis, i.e.
∼200 μmol m−2 sec−1 (Hill and Fanta 2008).
We sampled streams with broad nutrient ranges, i.e. 8.4–
260 μg L−1 NO3 − , <1–25 μg L−1 PO4 3− and <1–50.4 μg L−1 Fe, in the
Dallas–Fort Worth area to obtain a maximally diverse species
pool for our experiments. The source algae were scraped from
Larson et al.
hard substrates and homogenized to break up any large algal
clumps and to release trapped particulates, which could have altered the nutrient levels. To eliminate all insect grazers, the seed
algae were treated for 24 h with 0.32 mg L−1 Malathion, which
is the manufacturer recommended maximum application dose.
After the Malathion treatment, seed algae were washed, placed
in a microcosm with COMBO medium (Kilham et al. 1998), modified by excluding all tested nutrients, i.e. nitrate, phosphate and
iron, and illuminated for 14 h daily. The medium was replaced
daily for 7 days to remove traces of all tested nutrients from the
seed source. Inoculation with 100 mL of algal suspension took
place once on day 1 of each experiment.
The microcosms were filled with modified COMBO medium,
prepared with carbon-filtered water and supplemented with
NaNO3 (14.0 mg L−1 N), K2 HPO4 (1.55 mg L−1 P), EDTA + FeCl3 ·H2 O
(0.21 mg L−1 Fe) and all their combinations or left as control
(none of the aforementioned nutrients added). Other than the
manipulated nutrients, modified COMBO medium included all
constituents (major stocks and algal trace elements) in their normal concentrations (Kilham et al. 1998). Employing a fully factorial design resulted in eight different nutrient treatments with
three replicates each. Every 3 days, one-third of the medium was
replaced with fresh medium of the respective nutrient combination and the nutrient levels were checked for consistency with
the COMBO levels with an AutoAnalyzer III (SEAL Analytical Inc.,
Mequon, Wisconsin, USA).
The bottom of the microcosms was lined with 550 g of small
tumbled glass pieces (size 2 jelly bean glass, The Garden of Glass,
https://thegardenofglass.com), providing a substrate for 36 circularly arranged natural stone tiles (3.1 cm × 1.5 cm × 0.7 cm)
and additional surface area for algal colonization. Every 10 days,
three randomly chosen tiles from the same relative position in
each microcosm were thoroughly scraped with a toothbrush until visibly clean, then returned to the microcosm but never sampled again. The scraped area at each sampling was estimated
at 0.5–0.8% of the total area available for colonization (tiles and
jelly bean glass), indicating that sampling generated minimal
biomass loss. Experiments continued until natural sloughing,
which occurred after day 40 and 60 in the two runs, respectively.
Algal processing and identification
The collected algae were fixed with 4% formaldehyde. Soft algae
units (unit = a cell for unicellular algae, a colony, or 25 μm of a
filament) were counted in 30 random fields in a Palmer-Maloney
R
cell. Samples were acid digested and mounted with Naphrax
for diatom identification. At least 400 diatom frustules were enumerated per sample. For most species, we measured at least
20 individuals and calculated their average biovolume, as previously described (Hillebrand et al. 1999). For rare species (<20
individuals), the cell dimensions were taken from the literature
and used to calculate biovolume as above. Densities of all algae
were converted to biovolume (μm3 cm−2 ).
Statistical analyses
Using NAWQA data, Threshold Indicator Taxa ANalysis (TITAN)
(Baker and King 2010) was implemented to estimate community
shifts (change points) across a gradient of iron concentrations,
as determined by the synchronous changes in the abundance of
their constituent taxa. Acidic streams (pH ≤ 6.9, about 13% of
all streams) were excluded because they have distinct species
composition, low richness and greater Fe concentrations than
non-acidic streams; therefore, they are not representative of the
majority of streams in this study. To control for the differences in
3
species composition between oligotrophic and eutrophic sites,
TITAN was performed in the two groups separately. Concentrations of iron ranged from 3 to 950 μg L−1 in the oligotrophic
stream samples (n = 245) and from 3 to 1485 μg L−1 in the eutrophic stream samples (n = 355), but these differences were not
significant as shown in the section ‘Results’.
TITAN was run in R.3.0.1 (R Development Core Team, Vienna,
Austria). Species with occurrences in at least five sites were used
in the analysis, resulting in 391 and 454 species in the oligotrophic and eutrophic groups, respectively. For each nutrient
group, 250 random permutations of the taxa data were carried
out to compute standardized z scores for estimating community
change points. Indicator value (IndVal), measuring species association with a specific group (z+ or z−), and z score (IndVal standardized by subtracting the mean and dividing by the standard
deviation of the permutations) were calculated for all species.
The indicator quality of each species was assessed by two diagnostic metrics—purity and reliability, calculated from the bootstrap resampling. Purity represents the proportion of bootstrap
replicates matching the observed group assignment (either z+
or z−). Reliability is the proportion of bootstrap replicates with
observed IndVal scores equal to or larger than expected from
random data (probability values ≤ 0.05). Species of high purity
show a bootstrap response that is consistent with the observed
response in ≥95% of the replicates. Species of high reliability
have indicator values that are significant in ≥95% of the bootstrap replicates. Additionally, bootstrap metrics of uncertainty
for sum(z) and individual taxa z scores were calculated using
500 unique iterations.
In the continental data, equality of mean Fe concentrations
and equality of Fe variance between oligotrophic and eutrophic
streams were tested by a t-test and a Bartlett test, respectively. Proportional area of land cover types (arcsine square-root
transformed) and iron concentrations (ln-transformed) were
analyzed by correlation. In the experimental dataset, algal
richness (the numbers of all identified species, all nondiazotrophs and all diazotrophs) and ln-transformed biovolume
for the same three groups of species were compared across treatments using repeated measures ANOVA (rANOVA), followed up
by a Tukey post hoc test.
RESULTS
TITAN of continental survey data revealed strong species responses and pronounced community shifts along Fe gradients
in both oligotrophic and eutrophic streams. From the species
with significant responses to Fe (P < 0.05) and purity and reliability ≥0.95, a higher number, 122, was identified as positive
(z+) indicators (increasing with Fe concentrations), while fewer,
63 taxa, emerged as negative (z−) indicators (decreasing with Fe
concentrations) (see Table S1, Supporting Information). A community change point occurs when the sum of species’ z scores
is maximized. This was observed for z+ species at 79.5 μg L−1 Fe
in oligotrophic streams and 120 μg L−1 Fe in eutrophic streams
(Fig. 1). However, the sum of z scores for z+ taxa exhibited a
broader peak in oligotrophic streams, i.e. between about 30 and
79.5 μg L−1 Fe and a second, smaller peak in eutrophic streams at
70 μg L−1 . These results indicate that the community thresholds
lie within the ranges 30–79.5 μg L−1 Fe in oligotrophic streams
and 70–120 μg L−1 Fe in eutrophic streams. As expected, the community threshold was higher in eutrophic sites. This was not
due to differential Fe concentrations because mean Fe concentrations and equality of variance across these groups were not
4
FEMS Microbiology Ecology, 2015, Vol. 91, No. 5
Figure 1. TITAN of periphyton community responses along iron gradients in oligotrophic streams (n = 245) (a) and eutrophic streams (n = 355) (b) across the USA.
Sum(z−) and sum(z+) values indicate all candidate change points, while vertical lines denote the maximum sum(z+) values.
significantly different, as determined by a t-test and a Bartlett
test, respectively (P > 0.05). In both stream types, the community
change points for the z− indicator taxa were similar and close
to the end of the Fe gradient (3.0–4.0 μg L−1 ), but they were not
distinct (Fig. 1). Thus, determination of community thresholds
was based on the distribution of the positive indicators only.
We then assessed the variability in Fe concentrations in 2437
US streams, sampled over an 18 year period, and tested what
major land cover types contributed the most to this variability. In
50% of the studied streams, minimum total iron was ≤15 μg L−1
(Fig. 2a), i.e. substantially below the community thresholds in
both oligotrophic and eutrophic streams (Fig. 1) and below the
individual change points for the majority of z+ species (see Table S1, Supporting Information). In 75% of the streams, minimum total Fe was ≤67 μg L−1 (Fig. 2a) and thus lower than the
Fe threshold in eutrophic streams. This means that stream algal
communities may experience extensive, yet still unappreciated
Fe limitation. The origin of this deficiency was traced back to
the proportion of wetlands in the watershed, which was most
strongly correlated (Pearson r = 0.50, P < 0.000 0001, n = 1670)
with Fe concentrations (averaged for each locality). All other
land cover types were comparatively weak or non-significant
(NS) correlates of Fe, including shrubland (−0.16, P < 0.000 0001),
surface water (0.11, P < 0.000 005), agriculture (−0.07, P = 0.005),
forest (0.06, P < 0.02), urban development (0.06, P < 0.02), barren land (NS), grassland (NS) and perennial ice and snow (NS).
Therefore, streams in watersheds with large wetlands had high
Fe concentrations (Fig. 2a and b).
To test our hypotheses on the influence of iron on community structure and function, we measured the independent iron
effect, excluding all confounding macronutrient effects, as well
as the iron-macronutrient interactive effects with two 40–60 day
experiments in laboratory microcosms (Fig. S1, Supporting Information). Despite compositional differences between seed algae,
the trends in biomass and species accumulation were consistent in the two experiments. For example, Encyonema silesiacum,
Cymbella helvetica, Achnanthidium minutissimum and E. minutum
were the most abundant species in the algae source for the first
Larson et al.
5
Figure 2. (a) Map of 2437 distinct stream localities, where the USGS collected 27 555 water samples and measured Fe concentrations from January 1993 to June 2011.
Localities were sampled between 1 and 211 times and the minimum recorded Fe concentration is plotted here. (b) Map of percent wetland cover in 2946 watersheds,
determined from GIS coverages by the USGS.
experiment, while Brachysira vitrea, Leptolyngbya vandenberghenii,
A. minutissimum and Nitzschia amphibia were the dominant seed
algae for the second experiment. Consequently, we conducted
most of our analyses with a pooled dataset (n = 144) from the
mature biofilm (after day 20 of colonization).
Treatment had a significant influence on biomass accrual
(non-diazotroph biovolume, diazotroph biovolume and total biovolume) and algal species richness (non-diazotroph richness,
diazotroph richness and total richness) over the experimental course. Since the duration of the two experiments differed,
rANOVA was performed for days 30 and 40 with data from both
experiments as well as for days 30, 40, 50 and 60 with the respective data from experiment 2 only (n = 96 samples in both
analyses). The between-subjects-treatment effect was highly
significant (P < 0.000 0001) in all analyses with F-ratios ranging between 17.3 and 76.0 in the first set of rANOVAs (degrees
of freedom = 7 and 40) and between 75.3 and 327.8 in the second set of rANOVAs (df = 7 and 16). To achieve maximum power,
Tukey post hoc tests were carried out using all samples from both
experiments (n = 144).
Biovolume of non-diazotrophs was the highest in the NPFe
treatment, second highest in the NP treatment, followed by the
PFe and P treatments (Fig. 3). Total biovolume followed the exact same pattern (data not shown), indicating that iron supplementation stimulated significantly biomass production, but
only when both N and P were present, as expected. Iron by itself as well as in combination with N did not generate significant biovolume increases compared to control. The PFe and P
6
FEMS Microbiology Ecology, 2015, Vol. 91, No. 5
A notable exception was the PFe treatment, where total richness
was significantly greater than in the NP treatment (Fig. 4b). As
hypothesized, Fe had a significant independent effect on richness (non-diazotroph and total richness). Iron also displayed a
significant synergistic effect with P, whereby richness was significantly greater in the PFe treatment than in both the P and Fe
treatments. Supplementation with both N and Fe, on the other
hand, did not produce higher richness than this observed with
either N or Fe.
Nitrogen fixers reached the highest biovolume (Fig. 3,
Table S2, Supporting Information) and richness (Fig. 4a) in the
PFe treatment, where they comprised about 9% of the total biovolume on average. Diazotroph biovolume and diversity were
significantly greater in the PFe treatment than in both the P and
Fe treatments (Figs 3 and 4a), suggesting that PFe colimitation
may have an impact on the nitrogen cycle in stream ecosystems.
DISCUSSION
Figure 3. Mean biovolume (ln-transformed) with 95% CI of non-diazotroph and
diazotroph algae across treatments in two experimental runs. Different letters
indicate significant differences between treatments at P = 0.000 002 for nondiazotrophs and 0.000 002 < P ≤ 0.002 for diazotrophs following Tukey post hoc
tests, n = 144.
treatments were not significantly different from one another either. However, the addition of Fe when N and P were replete resulted in six times greater total biovolume on average than in the
NP treatment, revealing a strong synergistic Fe-macronutrient
effect on biovolume production.
Compared to biomass accumulation, both the independent
and interactive effects of Fe on species richness were pronounced. Species richness of non-N2 fixers was the highest
in the NPFe treatment and the second highest in the NP and
PFe treatments, significantly exceeding all other treatments
(Fig. 4a). The variability of total richness across treatments was
similar to this of non-diazotrophs, including an over 50% increase in the NPFe treatment compared to the NP treatment.
The present investigation reports community thresholds along
Fe gradients in streams. Previous research in freshwaters,
conducted exclusively with phytoplankton, has shown single
metrics, such as biomass accumulation, cyanobacterial colony
formation and carbon and nitrogen fixation, to be inhibited
by Fe limitation (Schelske 1962; Wurtsbaugh and Horne 1983;
Hyenstrand, Rydin and Gunnerhed 2000). Here, we demonstrate
continentally by a multivariate approach that the effects of iron
in the benthos can be detected from individual species to entire communities, undergoing distinct shifts as Fe concentrations change. The thresholds of these changes were estimated
at Fe concentrations exceeding the minimum Fe levels observed
in 50–75% of the US streams. These results reveal potentially
wide-spread Fe deficiency for the majority of benthic species,
most likely originating from the limited distribution of wetlands,
as previously suggested (Passy 2010). A comparative analysis
of watershed land cover types and stream iron concentrations,
which, we believe, is the most comprehensive to date, lent support to our hypothesis that wetlands were a major Fe source
for streams. As expansive wetlands are confined primarily to
Figure 4. Mean species richness with 95% CI of non-diazotroph and diazotroph algae (a) and total algae (b) across treatments in two experimental runs. Different letters indicate significant differences between treatments at 0.000 002 ≤ P ≤ 0.01 for non-diazotrophs, 0.000 002 ≤ P ≤ 0.03 for diazotrophs and 0.000 002
≤ P ≤ 0.01 for total algae following Tukey post hoc tests, n = 144. The inset shows some of the dominant N2 fixers in the PFe treatment, Anabaena variabilis and
Rhopalodia gibba.
Larson et al.
coastal areas in the east and southeast USA, the Great Lakes
region and the Mississippi valley (Fig. 2b), the majority of US
streams remains comparatively Fe-poor (Fig. 2a). This can be a
problem for many benthic producers, given their high Fe requirements (Fig. 1, Table S1, Supporting Information).
We also carried out laboratory experiments in stream
biofilms, which revealed that biodiversity, responding significantly to the independent Fe effect, was more sensitive to
Fe supplementation than biomass production, as hypothesized. The synergistic effects of Fe and macronutrients, on the
other hand, were important for both producer biodiversity and
biomass. Specifically, the NPFe treatment had on average 16
species more than the second richest treatment, PFe, and significantly greater biovolume than the second most abundant
treatment, NP. These results are in general agreement with the
benthic model (Passy 2008), supporting an evolved understanding of natural communities as being strongly colimited by a
higher number of resources and because of this more vulnerable to loss of biodiversity and changes in biomass when the
resource amounts are altered (Harpole and Tilman 2007; Passy
2008). However, we further showed stream biofilms to exhibit
more complex patterns of resource colimitation than previously
thought (Harpole et al. 2011), involving both macronutrients and
micronutrients. In view of the broad scale iron deficiency in
streams and the strong dependence of stream producers on
iron supply, confirmed here continentally and experimentally,
we conclude that iron limitation is a major driver of spatial
algal dynamics in running waters. Our experimental observations that the PFe treatment supported greater overall biodiversity than the NP treatment, and richness in the N but not in
the Fe treatment was statistically equivalent to this in the control treatment, raise the question of whether iron may have a
greater impact on stream algal biodiversity than nitrogen. This
is plausible considering that in the N-deplete conditions of the
PFe treatment there was a well-developed assemblage of N2 fixers, which may have supplied the rest of the community with usable forms of nitrogen. These results suggest that coexistence of
diazotrophs and non-diazotrophs in N-limited natural streams
may be controlled by PFe colimitation.
Our finding of Fe-macronutrient colimitation in the biofilm
has far-reaching implications for management of stream water quality. Streams and rivers contain only 0.0001% of the
water on Earth but supply nearly two-thirds of the water resources used by humans (Perry and Vanderklein 1996). The uptake of nutrient pollutants (nitrate) increases with algal richness
(Cardinale 2011) and here we detected the highest richness in
the NPFe treatment, exceeding this in the NP treatment by 22
species on average (Fig. 4b). However, as discussed, eutrophication does not correlate with Fe supply (Passy 2012) and our
oligotrophic and eutrophic streams did not differ in Fe concentrations. This means that Fe-deficient streams, receiving
excessive nitrate and phosphate loads from agriculture and urbanization will have diminished biodiversity and consequently,
weaker ability to reduce pollution. Eutrophication is a global
problem in aquatic ecosystems, causing water quality deterioration (Smith and Schindler 2009); therefore, it is important to
understand how it can be controlled and even, reversed. We recommend that future research in stream remediation explores
whether addition of bioavailable iron would stimulate the selfpurification process in macronutrient-impacted watersheds by
increasing the biodiversity of microbial producers and improving their resource utilization.
7
Our survey and experimental observations provide valuable
insight into the consequences of surface water ‘brownification’,
or the increase of dissolved organic carbon (DOC), which has occurred over the past 30 years as a result of global warming, declining anthropogenic sulfur emissions and reduced soil acidification (Freeman et al. 2001; Monteith et al. 2007; Evans et al.
2012). More recently, we have come to realize that brownification
is also associated with rising iron levels (Kritzberg and Ekstrom
2012; Sarkkola et al. 2013), which, given the findings of this study,
may lead to greater producer biodiversity, biomass production
and N2 fixation, all with favorable effects on consumers. Strong
positive correlations between organic matter and diatom biodiversity have already been reported across local to continental
scales (Passy, Ciugulea and Lawrence 2006; Passy 2010; Pound,
Lawrence and Passy 2013) and it is quite possible that these
trends are driven to a large extent by Fe levels, covarying with
the spatial patterns of DOC concentrations. The information on
species and community thresholds along Fe gradients generated
by this study can be very useful in developing predictive models
for the outcome of future changes in stream Fe concentrations,
which will improve our ability to project how benthic communities will respond to brownification.
This is also the first experimental demonstration of PFe
control of diazotroph composition in the stream biofilm. Previous research on N2 fixation in running waters has generated conflicting results on the influence of P enrichment,
shown to be a positive factor in some systems (Marcarelli and
Wurtsbaugh 2006, 2007) but non-significant in others (Scott
et al. 2009). Although, we did not measure N2 fixation directly,
our microscopic analyses revealed that P addition stimulated
the diazotroph community but was insufficient. This may explain the discrepancy in earlier field studies, i.e. differential Fe
amounts may generate diverging responses to P enrichment.
Here, the P treatment was not significantly different than the
control in terms of diazotroph richness and failed to produce
the N2 fixer diversity and abundance observed when both P
and Fe were plentiful. Species from the cyanobacterial genera
Anabaena, Calothrix, Nostoc, Rivularia and Pseudanabaena and the
diatom genus Rhopalodia established comparatively large populations in the PFe treatment (Fig. 4a) but were absent or in
low abundance in the P treatment (Table S2, Supporting Information), implying that Fe is a limiting nutrient for some of the
most common freshwater N2 fixers. The diazotroph flora in the
freshwater biofilm, encompassing mostly heterocystous and
non-heterocystous filamentous cyanobacteria and some biraphid diatoms, is vastly dissimilar from this in the open ocean,
dominated by the non-heterocystous Trichodesmium and unicellular cyanobacteria (Berman-Frank, Lundgren and Falkowski
2003), yet both floras are constrained by P and Fe, as shown here
and in oceanic reports (Mills et al. 2004). We propose that similarly to the ocean, where iron supply has governed N2 fixation
and primary productivity on geological timescales (Falkowski
1997; Canfield, Glazer and Falkowski 2010), the nitrogen and carbon cycles in streams may too be strongly affected by iron availability. Therefore, despite differences in geologic and natural
history, freshwater and marine producers may be driven by similar mechanisms.
SUPPLEMENTARY DATA
Supplementary data is available at FEMSEC online.
8
FEMS Microbiology Ecology, 2015, Vol. 91, No. 5
FUNDING
This research was supported by the Norman Hackerman Advanced Research Program under grant no. 003656-0054-2009 to
SIP.
Conflict of interest. None declared.
REFERENCES
Baker ME, King RS. A new method for detecting and interpreting
biodiversity and ecological community thresholds. Methods
Ecol Evol 2010;1:25–37.
Berman-Frank I, Lundgren P, Falkowski P. Nitrogen fixation and
photosynthetic oxygen evolution in cyanobacteria. Res Microbiol 2003;154:157–64.
Borchardt MA. Nutrients. In: Stevenson RJ, Bothwell ML, Lowe RL
(eds). Algal Ecology: Freshwater Benthic Ecosystems. San Diego,
CA: Academic Press, 1996, 183–227.
Boyd PW, Watson AJ, Law CS, et al. A mesoscale phytoplankton
bloom in the polar Southern Ocean stimulated by iron fertilization. Nature 2000;407:695–702.
Canfield DE, Glazer AN, Falkowski PG. The evolution and future
of earth’s nitrogen cycle. Science 2010;330:192–6.
Cardinale BJ. Biodiversity improves water quality through niche
partitioning. Nature 2011;472:86–9.
Coale KH, Johnson KS, Fitzwater SE, et al. A massive phytoplankton bloom induced by an ecosystem-scale iron fertilization experiment in the equatorial Pacific Ocean. Nature
1996;383:495–501.
de Baar HJW, Boyd PW, Coale KH, et al. Synthesis of iron fertilization experiments: From the iron age in the age of enlightenment. J Geophys Res 2005;110:C09S16.
Dillon PJ, Molot LA. Effect of landscape form on export
of dissolved organic carbon, iron, and phosphorus from
forested stream catchments. Water Resour Res 1997;33:
2591–600.
Downs T, Schallenberg M, Burns C. Responses of lake phytoplankton to micronutrient enrichment: a study in two New
Zealand lakes and an analysis of published data. Aquat Sci
2008;70:347–60.
Elser JJ, Bracken MES, Cleland EE, et al. Global analysis of nitrogen
and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecol Lett 2007;10:
1135–42.
Evans CD, Jones TG, Burden A, et al. Acidity controls on dissolved organic carbon mobility in organic soils. Glob Change
Biol 2012;18:3317–31.
Evans-White MA, Haggard BE, Scott JT. A Review of stream nutrient criteria development in the United States. J Environ Qual
2013;42:1002–14.
Falkowski PG. Evolution of the nitrogen cycle and its influence
on the biological sequestration of CO2 in the ocean. Nature
1997;387:272–5.
Fishman MJ, Friedman LC (eds). Methods for determination of
inorganic substances in water and fluvial sediments. Techniques of Water-Resources Investigations of the United States
Geological Survey, 3rd edn, 1989, 545.
Freeman C, Evans CD, Monteith DT, et al. Export of organic carbon from peat soils. Nature 2001;412:785.
Grimm NB. Disturbance, succession and ecosystem processes in
streams: a case study from the desert. In: Giller PS, Hildrew
AG, Raffaelli DG (eds). Aquatic Ecology: Scale, Pattern and Process. London: Blackwell Science Ltd, 1994, 93–112.
Grimm NB, Petrone KC. Nitrogen fixation in a desert stream
ecosystem. Biogeochemistry 1997;37:33–61.
Harpole WS, Ngai JT, Cleland EE, et al. Nutrient colimitation of primary producer communities. Ecol Lett 2011;14:
852–62.
Harpole WS, Tilman D. Grassland species loss resulting from reduced niche dimension. Nature 2007;446:791–3.
Hill WR, Fanta SE. Phosphorus and light colimit periphyton growth at subsaturating irradiances. Freshwater Biol
2008;53:215–25.
Hillebrand H, Durselen C, Kirschtel D, et al. Biovolume calculation for pelagic and benthic microalgae. J Phycol 1999;35:
403–24.
Hyenstrand P, Rydin E, Gunnerhed M. Response of pelagic
cyanobacteria to iron additions—enclosure experiments
from Lake Erken. J Plankton Res 2000;22:1113–26.
Kilham SS, Kreeger DA, Lynn SG, et al. COMBO: a defined freshwater culture medium for algae and zooplankton. Hydrobiologia 1998;377:147–59.
Kritzberg ES, Ekstrom SM. Increasing iron concentrations in surface waters - a factor behind brownification? Biogeosciences
2012;9:1465–78.
Marcarelli AM, Wurtsbaugh WA. Temperature and nutrient
supply interact to control nitrogen fixation in oligotrophic
streams: an experimental examination. Limnol Oceanogr
2006;51:2278–89.
Marcarelli AM, Wurtsbaugh WA. Effects of upstream lakes and
nutrient limitation on periphytic biomass and nitrogen fixation in oligotrophic, subalpine streams. Freshwater Biol
2007;52:2211–25.
Martin JH, Fitzwater SE. Iron deficiency limits phytoplankton growth in the north-east Pacific subarctic. Nature
1988;331:341–3.
Mills MM, Ridame C, Davey M, et al. Iron and phosphorus colimit nitrogen fixation in the eastern tropical North Atlantic.
Nature 2004;429:292–4.
Mitsch WJ, Gosselink JG. Wetlands, 4th edn. Hoboken, NJ: John
Wiley & Sons, 2007.
Monteith DT, Stoddard JL, Evans CD, et al. Dissolved organic carbon trends resulting from changes in atmospheric deposition chemistry. Nature 2007;450:537–40.
Moore CM, Mills MM, Achterberg EP, et al. Large-scale distribution of Atlantic nitrogen fixation controlled by iron availability. Nat Geosci 2009;2:867–71.
North RL, Guildford SJ, Smith REH, et al. Evidence for phosphorus, nitrogen, and iron colimitation of phytoplankton communities in Lake Erie. Limnol Oceanogr 2007;52:315–28.
Paerl HW, Prufertbebout LE, Guo CZ. Iron-stimulated N2 fixation
and growth in natural and cultured populations of the planktonic marine cyanobacteria Trichodesmium spp. Appl Environ
Microb 1994;60:1044–7.
Passy SI. A distinct latitudinal gradient of diatom diversity is
linked to resource supply. Ecology 2010;91:36–41.
Passy SI. A hierarchical theory of macroecology. Ecol Lett
2012;15:923–34.
Passy SI. Continental diatom biodiversity in stream benthos declines as more nutrients become limiting. P Natl Acad Sci USA
2008;105:9663–7.
Passy SI. The relationship between local and regional diatom
richness is mediated by the local and regional environment.
Global Ecol Biogeogr 2009;18:383–91.
Passy SI, Ciugulea I, Lawrence GB. Diatom diversity in chronically versus episodically acidified Adirondack streams. Int Rev
Hydrobiol 2006;91:594–608.
Larson et al.
Perry J, Vanderklein E. Water Quality: Management of a Natural Resource. Cambridge, MA: Blackwell Science, Inc., 1996.
Pound KL, Lawrence GB, Passy SI. Wetlands serve as natural sources for improvement of stream ecosystem health
in regions affected by acid deposition. Glob Change Biol
2013;19:2720–8.
Pringle CM, Paabyhansen P, Vaux PD, et al. In situ nutrient assays
of periphyton growth in a lowland Costa Rican stream. Hydrobiologia 1986;134:207–13.
Sarkkola S, Nieminen M, Koivusalo H, et al. Iron concentrations
are increasing in surface waters from forested headwater
catchments in eastern Finland. Sci Total Environ 2013;463:
683–9.
Schelske CL. Iron, organic matter, and other factors limiting primary productivity in a marl lake. Science 1962;136:45–6.
Scott JT, Lang DA, King RS, et al. Nitrogen fixation and phosphatase activity in periphyton growing on nutrient diffusing substrata: evidence for differential nutrient limitation in
stream periphyton. J N Am Benthol Soc 2009;28:57–68.
9
Smith VH, Schindler DW. Eutrophication science: where do we
go from here? Trends Ecol Evol 2009;24:201–7.
Sterner RW. On the phosphorus limitation paradigm for lakes.
Int Rev Hydrobiol 2008;93:433–45.
Stevenson RJ. An introduction to algal ecology in freshwater benthic habitats. In: Stevenson RJ, Bothwell ML, Lowe RL (eds).
Algal Ecology: Freshwater Benthic Ecosystems. San Diego, CA:
Academic Press, 1996, 3–30.
Vis C, Hudon C, Carignan R, et al. Spatial analysis of production by macrophytes, phytoplankton and epiphyton in a large
river system under different water-level conditions. Ecosystems 2007;10:293–310.
Vitousek PM, Howarth RW. Nitrogen limitation on land and in
the sea: How can it occur? Biogeochemistry 1991;13:87–115.
Vuori KM. Direct and indirect effects of iron on river ecosystems.
Ann Zool Fenn 1995;32:317–29.
Wurtsbaugh WA, Horne AJ. Iron in eutrophic Clear Lake,
California: its importance for algal nitrogen fixation and
growth. Can J Fish Aquat Sci 1983;40:1419–29.