Limnol. Oceanogr., 54(6, part 2), 2009, 2520–2528
2009, by the American Society of Limnology and Oceanography, Inc.
E
Nitrogen deposition, catchment productivity, and climate as determinants of
lake stoichiometry
Dag O. Hessen,a,* Tom Andersen,a Søren Larsen,a Brit Lisa Skjelkvåle,b and Heleen A. de Witb
a Department
b Norwegian
of Biology, University of Oslo, Oslo, Norway
Institute for Water Research (NIVA), Oslo, Norway
Abstract
Nearly 1000 Norwegian lakes in catchments with low human activity were surveyed. By covering a wide range
of nitrogen (N) deposition (0.1–2 g m22 yr21) along a latitudinal and climatic gradient, we clearly demonstrate
how nitrogen (N) deposition, climate, and a few key catchment properties, notably the terrestrial vegetation
density and the fraction of bogs, together serve as major predictors of concentrations and ratios of carbon,
nitrogen (N), phosphorus (P), and silicate (Si) in downstream lakes. Inorganic N in lakes was positively correlated
with N deposition, while organic N was closely associated with allochthonous dissolved organic carbon. The ratio
of NO3 to total N as well as NO3 to total P and NO3 to SiO2 were highly variable, and most of this variability was
explained by N deposition; terrestrial vegetation density, as inferred from the Normalized Difference Vegetation
Index; temperature; runoff; and the fraction of bogs in the catchment. Climate-induced changes in element
concentrations and elemental ratios could profoundly affect the lake metabolism and community composition. By
linking these data with downscaled climate change predictions we may also predict future shifts in element export
and element ratios in various lakes with reasonable accuracy.
Human alterations of biogeochemical cycles have
become a major concern both on the local and the global
scale. Climatic changes, via changes in temperature,
hydrology, vegetation, and other catchment properties,
strongly affect fluxes of matter and elements in the
watershed and, thus, recipient lakes and marine areas.
While changes in the global carbon (C) cycle are an issue
of major concern, the relative changes in the global
nitrogen (N) cycle are even more pronounced. Transformation of atmospheric N2 to more reactive reduced or
oxidized forms of inorganic N by the fertilizer industry and
various combustion processes has accelerated dramatically,
and recent analyses of the global N cycle (Vitousek et al.
1997; Galloway and Cowling 2002; Galloway et al. 2008)
indicate that various human activities currently fix more N2
than do natural ecosystems.
Increased N deposition can affect surface waters in many
different ways. Increased deposition of inorganic N
promotes soil and water acidification through increased
NO3 in surface waters (Stoddard 1994; Henriksen et al.
1997). Furthermore, N deposition affects the nature of
elemental limitation for both autotrophs and heterotrophs
in lakes and rivers, with consequent effects on community
composition and ecosystem processes. One effect of such
increased inputs of N over phosphorus (P) would be an
intensified P limitation in surface waters (Hessen et al.
1997; Interlandi and Kilham 1998; Bergström et al. 2005),
or even large-scale shifts from N limitation to P limitation
(Bergström and Jansson 2006).
Precipitation affects concentrations of various compounds and elements differently. In areas affected by
anthropogenic N emissions, precipitation is enriched in
inorganic N and is an important source of N to surface
* Corresponding author: [email protected]
waters. Other elements, such as C, P, and silicate (Si), that
are of primarily terrestrial origin are generally diluted by
precipitation. Thus, in areas with elevated N deposition,
one might expect ratios of inorganic N to C, P, and Si to
increase.
Organic N, on the other hand, is primarily associated
with dissolved organic matter (DOM), mostly humic
matter, and should be diluted with increased precipitation
in parallel with dissolved organic carbon (DOC). Hence,
the ratios of total N to other elements will depend on the
relative contribution of organic vs. inorganic N, which
again will depend on catchment properties. It is important
to keep in mind that concentrations and flux rates may
relate differently to precipitation, such that high runoff
may dilute the concentration but still increase export flux.
Climate-induced changes on mineralization and mobilization of elements may thus be easier to detect in downstream
lakes than in the terrestrial part of the catchment. Hence,
lakes may serve as sentinels of ongoing catchment changes;
such changes may also affect physical, chemical, and
biological properties in the recipient lakes as well.
In this study, we utilize a large data set on lake
chemistry, catchment properties, temperature, precipitation, and N deposition for almost 1000 lakes in Norway
and assess how these factors affect export fluxes and, thus,
ratios of C, N, P, and Si in lakes. Annual N deposition
rates vary by more than a 10-fold measure between
southern areas that receive high anthropogenic inputs and
the comparatively pristine northern areas. There is also a
strong gradient in temperature, precipitation, vegetation
density (as indicated by the Normalized Difference
Vegetation Index, NDVI; Tucker 1979), and other catchment-specific properties that allow for testing for the effects
of climatic variables, N deposition, and catchment attributes on elemental concentrations and ratios in lakes. The
2520
N-deposition, climate, and stoichiometry
fate and flux of DOC is clearly important also for the flux
of organic fractions of N and P (Dillon and Molot 2005),
but a full study on drivers for concentrations and export
fluxes of DOC (or total organic carbon, TOC) will be
reported elsewhere. In this study we focus on the drivers for
elemental ratios in lakes with fairly pristine catchments.
The comparison of catchments and lakes under different
climatic regions represents a type of ‘‘space for time’’
approach to climate effects. Given the long response time
of landscape elements such as forest cover, bogs, and lakes,
we believe this to be a valid approach. The use of remote
sensing data for obtaining this range of key catchment
properties should be generally applicable, and we believe
both the applied methods and the results of this study to be
of relevance for a wide range of boreal lakes and
catchments.
Methods
The study is based on a nationwide database from a
regional lake survey in Norway in 1995 (Henriksen et al.
1998). The lakes span high-altitude to lowland locations,
with forested and nonforested watersheds and with highly
variable land cover. Most lakes have fairly pristine
catchments, with .95% of the catchments having ,5%
arable land. A single sample was taken from each lake
close to the lake outlet at the time of autumn overturn.
Most lakes were sampled in October 1995, while a few
were sampled the month before or after. As a result of the
water residence time-dependent averaging of concentrations in lake water, the samples are assumed to represent
an average of summer and early-fall conditions. Surveyed
lakes were selected using random sampling stratified by
county, with sampling fractions given by a national lake
register maintained by the Norwegian Water Resources
and Energy Directorate (NVE). All samples were analyzed
by standard methods in accredited laboratories within 2–
3 d of collection. Total N and P were measured as NO3
or PO4, respectively, after wet oxidation with alkaline
(acid) peroxodisulfate. NO3 and SiO2 were analyzed
colorimetrically in a segmented flow autoanalyzer, while
PO4 was analyzed by manual spectrophotometry. TOC
was converted to CO2 by catalytic high-temperature
combustion and detected by infrared gas analysis. NH4
concentrations were negligible in most lakes and thus are
not included.
Catchment boundaries from a NVE digital polygon layer
were combined manually so that each lake sample was
associated with a single polygon covering the entire
watershed. Digital line feature maps from NVE representing rivers and streams were used to guide the manual
watershed assembly process. Catchment areas were calculated directly from this polygon layer. The resulting
catchment polygon layer was then used to extract
catchment-averaged statistics from other Geographic Information System layers. Spatial analysis was performed
with Hawth’s Analysis Tools 3.27 (www.spatialecology.
com/htools), which is an extension for ESRI’s ArcMap 9.2.
Fractional land-cover statistics were extracted from
digital 1 : 50,000 land-use maps from the Norwegian
2521
Mapping and Cadastre Authority (Statkart). Land-use
classes included open water, bog (peatland), arable land,
and forest; the remainder represented largely unforested
moor or grassland. Most of the surveyed lakes have pristine
catchments, with cultivated land constituting only 0.5% of
total area, and bogs covered 5% of the area. Forested areas
represented 28% of total area, the rest being nonforested,
either as a result of high elevation or latitude and/or coastal
areas without forest cover. Lake area, on average, was 12%
of the total watershed area.
Runoff (mm yr21) was averaged for each catchment
polygon using 1960–1990 averages of area-specific runoff
on a Norwegian 1 3 1–km NVE raster map. Catchmentaveraged terrain slopes were calculated as the rate of
change between adjacent cells in a 1 3 1–km digital
elevation model of Norway from the Norwegian Mapping
and Cadastre Authority. Atmospheric N deposition (including dry deposition) was averaged for each catchment
polygon from a digital map of yearly, accumulated total
atmospheric N deposition for 1995. The N-deposition map
was constructed by spatial interpolation (kriging with a
spherical semivariogram model) on 1u 3 1u gridded output
data from the Unified EMEP MSC-W modeling system
(http://www.emep.int/). NDVI was acquired as monthly
composites over a year obtained from the U.S. Geological
Survey Eurasia Land Cover Characteristics database
(http://edc2.usgs.gov/glcc/). Annual mean temperatures
were averaged from the 1 3 1–km WorldClim raster
(Hijmans et al. 2005; http://www.worldclim.org/).
Multiple regression analysis was used for exploring
factors that are influencing both lake concentrations of
major elements as well as their ratios. In an initial screening
phase, regression models were constructed by stepwise
backward elimination using the Bayesian Information
Criterion for model selection (Johnson and Omland
2004). Based on this set of minimal adequate models, we
selected a common set of independent variables that we
used in all models to simplify interpretation and comparison of effects. Independent variables in the final set of
regression models were annual mean temperature, specific
runoff, total N deposition, NDVI, and terrain slope as well
as the catchment properties as area fractions of the landuse classes open water, bog, and arable land (Table 1).
Land-use fractions were arcsine square root–transformed
to stabilize variances and to make distributions more
symmetrical. Strongly skewed, positive variables like slope,
runoff, and N deposition were log transformed, while
NDVI and mean temperature were not transformed. A
scatterplot matrix of the eight independent variables
(transformed as indicated in Table 1) is shown in Fig. 1.
All dependent variables (total N, total P, NO3, and SiO2, as
well as their ratios) were log transformed in all models.
Since the independent variables have widely different scales
we have chosen to visualize effect sizes as regression
coefficient 3 the interquartile range of each variable. This
should represent the contrast between ‘‘typical’’ low and
high values of the independent variable in their effect on the
dependent variable. All statistical analyses were performed
with the R statistical programming environment version
2.6.1 (www.R-project.org).
2522
Hessen et al.
Table 1.
Variable
temp
runoff
NDVI
N-dep
slope
bog
lake
arable
Abbreviations, descriptions, units, transformations, and sources applied for catchment key variables.
Description
Unit
Transformation
Source
uC
mm yr21
No transformation
log10
WorldClim, Hijmans (2005)
Norwegian Water Resources and
Energy Directorate (NVE)
Index (0 : 200)
No transformation
Deposition of reactive nitrogen (N)
species from the atmosphere to the
biosphere
Maximum rate of change in z-axis
between a cell and its neighboring cells;
Calculated with the ArcGis 9.2 tool:
‘Slope’
Fraction of watershed covered with bogs;
nonforested area with peat vegetation
Total lake area to watershed area ratio
mg N m22 yr21
log10
Monthly NDVI composites from the
USGS Eurasia Land Cover
Characteristics database
EMEP MSC-W modeling system
u
log10
Norwegian Mapping and Cadastre
Authority
%
Arcsine square root
%
Arcsine square root
Fraction of watershed suitable to be
farmed or cultivated
%
Arcsine square root
Norwegian Mapping and Cadastre
Authority
Norwegian Mapping and Cadastre
Authority
Norwegian Mapping and Cadastre
Authority
Annual mean temperature
The precipitation or snowmelt that runs
off the land into streams or other
surface water
An indicator of terrestrial vegetation
density; Based on remote sensing
Results
Atmospheric N deposition (N-dep) ranged from 96 mg
to 1934 mg m22 yr21, with a strong latitudinal gradient
(Fig. 2a). This gradient partly reflects the precipitation
pattern and partly the fact that southern Norway is
strongly influenced by N deposition from central Europe.
Precipitation is the major source of inorganic N for most
catchments. The annual runoff displays primarily a
longitudinal pattern, with more than 50-fold variation,
ranging from 127 to 9644 mm m22 yr21, reflecting topography, with typically high orographic precipitation in the
west and dry on the eastward leeside (Fig. 2b). The other
two major predictor variables, the fraction of bogs in the
catchment (Fig. 2c) and NDVI (Fig. 2d), also show a
distinct geographical pattern, with NDVI reflecting lowland, productive habitats, while the most boggy areas are
typically located in the central regions.
The different roles of catchment properties, runoff, and
N-dep on lake concentrations of organic N and NO3 are
clearly shown in Fig. 3. While organic N reflects the effects
of bogs and vegetation as well as the diluting effects of
precipitation, NO3 generally follows N-dep. Deposition of
oxidized and reduced N was highly correlated (r2 5 0.96)
and close to a 1 : 1 ratio over the entire gradient of total
deposition. Lake concentrations of reduced N (NH4) are in
general low or negligible. This is probably due to preferential
uptake by roots and microorganisms of reduced N,
oxidation of NH4 to NO3 in soils and surface waters, and
to lower mobility of NH4 causing higher soil retention.
NDVI was the major predictor of lake concentrations of
organic N, total P, NO3, and total SiO2 (Fig. 4). DOM was
the most important determinant of concentrations of total
N and P, both being dominated by organic fractions; this is
reflected by the positive contribution of the fraction of bogs
in the catchment, which was not observed for SiO2. N
deposition not only contributed positively to total N, but it
also had a significant negative effect on total P and SiO2.
Mean annual temperature had contrasting positive (NO3)
and negative (P and SiO2) effects. Including all significant
predictor variables explained 71% of the variability in total
N but only 44% and 56% for total P and SiO2, respectively.
Concentrations of organic N followed total N closely,
reflecting that organic N was a major pool of total N, or
rather that total N reflect the dynamics of organic N, where
71% of observed variability could be explained by the
included parameters. This is in contrast to NO3, in which N
deposition was by far the major predictor, in addition to
positive contributions from temperature and % lake area of
total catchment (Fig. 4). Interestingly, the fraction of bogs
in the catchment was a major (negative) predictor of NO3.
In total, 58% of the NO3 concentration variance was
explained by the regression model. It should be noted that
there is a significant effect of arable land on most
parameters, in spite of the visual impression in Fig. 4. This
is due to the extreme skewness of this variable, with most
catchments having no agricultural land at all (less than
0.5% of the total area is arable), so that its interquartile
range becomes practically zero. When present in a
catchment, however, arable land made a notable contribution because of inputs of manure and fertilizers.
The processes affecting elemental concentrations in lakes
also had important effects on elemental ratios (Fig. 5). The
ratio between TOC and total organic nitrogen (TON) had
low variability as a result of their close correlation
(regression slope C : N 5 28.1 6 0.5, r2 5 0.76, p ,
0.0001). From a biological perspective, however, the
relative concentration of inorganic N is more relevant than
that of organic N. The ratio of NO3 to organic N was
below unity in 70.5% of the lakes, and the variability in this
ratio is striking, spanning more than four orders of
magnitude. For the NO3 : TP ratio, a 2700-fold range was
observed, and likewise, the NO3 : SiO2 ratio showed a
strong variability.
N-deposition, climate, and stoichiometry
Fig. 1.
2523
Scatterplot matrix of transformed independent variables (see Table 1) used in the regression models (n 5 987–992).
A fair share of this biologically important variability was
accounted for by climatic and catchment-specific variables.
The full analysis of determinants of DOC flux and
concentrations is beyond the scope of this analysis, but we
have included TOC in the analysis of elemental ratios. Sixtysix percent of the variance in TOC : total N was explained by
these variables, with NDVI as the major positive predictor
and N deposition as the major negative predictor (Fig. 6). As
a result of the close correlation between total N and total P,
only a modest part (36%) of the variation in the ratios
between these fractions was explained by the model, with
temperature and N deposition as major positive contributors. Also, for TOC : TON, only a modest part of the
variability was explained (43%), again partly owing to the
low variability in the ratio between TOC and organic N,
both being closely associated with DOM.
For NO3 : total P, NO3 : SiO2, and NO3 : organic N, N
deposition was, not surprisingly, the key variable, and was
also the key variable in all cases with NDVI as the major
negative predictor, annual mean temperature as the major
positive predictor, and with a fairly high predictive power
(53%, 66%, and 55%, respectively).
Discussion
By covering a wide range of N deposition along a
latitudinal gradient and by sampling pristine catchments
with negligible effects from agricultural runoff or sewage,
this study clearly demonstrates that terrestrial productivity,
N deposition, climate, and a few key catchment properties,
notably the fraction of bogs, together serve as major
predictors of element export to surface waters. This finding
2524
Hessen et al.
Fig. 2. Polygon maps showing regional patterns in some of the independent variables used
in the regression models: (a) nitrogen deposition (mg N m22 yr21, for the year 1995), (b) runoff
(mm yr21; average 1961–1990), (c) area fraction of bog and peatland (dimensionless), (d) NDVI
(index; Modis average for 1992): 22,500 watershed polygons; sampled lakes are represented as
dots in map (a).
is in accord with previous findings from boreal lakes
(Dillon and Molot 2005; Kortelainen et al. 2006). Since the
different predictor variables act differently on the various
elements, they also profoundly affect the stoichiometry of
key elements in lakes. In general, ratios with a narrow
range of variability (e.g., TOC : TON) were predicted with
lower accuracy than ratios with high variability (e.g.
NO3 : SiO2).
The co-variation of N deposition and precipitation and
the additional diluting effect of precipitation on non–N
elements acted together to increase all NO3 : element ratios.
It should be noted that this pattern reflects long-term
averages across different ecosystems, while the diluting
effect of precipitation will obviously not be equally clear in
single catchments on shorter time scales. Long-lasting,
elevated N deposition may also lead to N saturation
(Stoddard 1994; Kaste et al. 1997). Organic N, on the other
hand, was closely associated with TOC, both originating
primarily from allochthonous DOM (cf. Kortelainen et al.
2006). The ratio of total N to other elements was also
N-deposition, climate, and stoichiometry
2525
Fig. 3. Polygon maps of regression model predictions for concentrations of (a) organic N
and (b) NO3 (both as mg L21) in all 22,500 Norwegian watersheds.
influenced by N deposition, but the relatively large fraction
of organic N associated with allochthonous DOM dampens
the stoichiometric effect of N deposition on total N relative
to other elements. The effect of temperature on these ratios
is not clear-cut, but both NO3 : total P and NO3 : SiO2 were
apparently positively affected by high temperature. This
could be a confounding effect, since the southern, coastal
areas with high N deposition also typically have higher
annual temperatures, but it could also indicate increased
mineralization of organic N and mobilization of inorganic
N in soils, relative to P and Si, under elevated temperature
(Rustad et al. 2001). Intriguingly, N deposition apparently
yielded lower lake concentrations of P and SiO2, perhaps as
a result of stimulated terrestrial uptake of these nutrients
due to this N fertilization. This effect could add to the
elevated NO3 : total P and NO3 : SiO2 ratios in areas with
high N deposition.
NDVI presumably serves as a good proxy of terrestrial
net primary production in these catchments, where
precipitation normally is plentiful (Pettorelli et al. 2005),
Fig. 4. Effect size plots for the independent variables in regression models of lake-water concentrations of (a) total N, (b) total P, (c)
NO3, and (d) SiO2. Effect sizes are computed as regression coefficients 3 interquartile ranges of the different independent variables.
2526
Hessen et al.
Fig. 5. Frequency distribution histograms for elemental ratios (mol : mol) in the lakes: (a) TOC : TON, (b) NO3-N : TON, (c) NO3N : total P, and (d) NO3-N : SiO2.
and, hence, it also correlates with inorganic N uptake in
roots and high N retention (Kaste et al. 1997). NDVI
also turned out to be a major predictor of lake
concentrations of DOC, DON, NO3, total P, and SiO2.
The causality is not straightforward, however. Productive
catchments with high vegetation density will provide
more litterfall and a more active root mineralization, but
also a higher uptake of nutrients. Hence, the net effect of
NDVI on elemental ratios requires stronger mechanistic
investigation.
Fig. 6. Effect size plots for the independent variables in regression models of selected elemental ratios (a) TOC : total N, (b) total
N : total P, (c) NO3 : total P, (d) NO3 : SiO2, (e) NO3 : organic N, and (f) TOC : organic N. Effect sizes are computed as regression
coefficients 3 interquartile ranges of the different independent variables.
N-deposition, climate, and stoichiometry
For most all elements or elemental ratios, the catchment
fraction of bogs was a major predictor, yet with contrasting
effects. It is quite noteworthy that while bogs were
positively correlated with total N, this variable did not
explain NO3 or the ratio of NO3 relative to other elements
and organic N. Peats and wetlands are typical sinks of N as
a result of their high capacity for denitrification (Jansson et
al. 1994); nevertheless, the export of organic N also
indicates that a considerable fraction of inorganic N must
be converted to organic forms within these waterlogged
systems. Bogs were also positively related to total P,
reflecting that DOM also is a major carrier of organic P
(Dillon and Molot 2005; Kortelainen et al. 2006). Even if
only a modest fraction of the large pools of organic N and
organic P are converted to bioavailable forms by photooxidation or microbial mineralization, this would be a
major source of N for autotrophs (Humborg et al. 2004).
The association between SiO2 and TOC is subtle, and there
was no significant relationship between the fraction of bogs
and concentrations of SiO2, which primarily is determined
by chemical weathering induced by CO2 and organic acids
leached from roots and soils (Conley et al. 2000; Humborg
et al. 2004).
Both the absolute concentrations of NO3 as well as the
elemental ratios may have a suite of biological effects, and
N deposition and shifts in climate and vegetation cover
could profoundly affect these parameters. Increased NDVI
and fraction of bogs would clearly shift lakes toward N
limitation, while N deposition would work in the opposite
direction. Bergström et al. (2005) demonstrated that lakes
receiving high inputs of atmospheric N are pushed toward
extreme P limitation, reflected in increased phytoplankton
biomass yield with increased N deposition. Changes in both
NO3 : total P and NO3 : SiO2 may induce shifts in elemental
limitation and production of autotrophs in lakes and
downstream coastal areas (Conley et al. 1993; Hessen 1999;
Humborg et al. 2004).
Short- or long-term climatic changes will affect nutrient
export to lakes and coastal areas, as well as total loading
and nutrient stoichiometry (Meybeck 1993). Climate also
has effects on concentrations and runoff of DOC—and
thus also organic N and organic P. For most temperate
areas, DOC concentrations are increasing in rivers and
lakes (Skjelkvåle et al. 2005), and while decreased acid
deposition could contribute to this (Monteith et al. 2007),
there is no doubt that temperature and hydrology could
also be key players influencing the export of organic matter
to lakes (Schindler et al. 1997; Erlandsson et al. 2008).
Shifts in both the absolute and relative loadings of DOC to
nutrients, for example, will affect the whole lake metabolism. Typically, a high ratio of DOC : P is characteristic of
high ecosystem respiration to production ratio (Hanson et
al. 2004), and this ratio is likely to be strongly affected by
climate change (Benoy et al. 2007). This again would
promote feedbacks from lakes to the atmosphere.
A major outcome of this analysis of temperate, boreal
lakes and catchments is thus the important role of
terrestrial vegetation productivity on lake-water chemistry.
Increased terrestrial productivity is likely to change
concentrations of most elements in lakes, yet not at the
2527
same rate or in the same direction. In particular, the ratios
of NO3 to P, Si, or organic N are expected to decrease as a
result of increased retention of inorganic N at increased
temperatures, while enhanced N deposition will work in the
opposite direction. For all relationships there is a large,
unexplained variance. Presumably, part of the unexplained
variance in observed concentrations and ratios is due to
variations in retention between lakes with different water
residence times: low renewal rates imply high retention and
vice versa, but this will not be equal for all elements.
Typically areas receiving high inputs of atmospheric N may
have low catchment retention of N and virtually zero lake
retention of N (Kaste et al. 1997), while P retention would
be substantial as a result of a strong P limitation. Climateinduced changes in precipitation will thus be a major
determinant not only of concentrations of elements in
lakes, but also of elemental ratios.
Climate change scenarios point toward a future with
more forests and increased terrestrial production in areas
receiving sufficient precipitation, while the opposite will
happen under drier conditions. On a long-term scale
(decades to centuries), changes in catchment properties,
such as fractions of bogs and forest cover, and weathering
rates will regulate the flux and fate of C, N, P, and Si in
catchments, while on a shorter time scale (years to
decades), anthropogenic effects on the N cycle, changed
temperature, precipitation patterns, and soil or root
processes will be dominant. In any case, the entangled
responses and interactions of these drivers may be hard to
detect directly in the terrestrial catchment, while downstream lakes will pick up and integrate these stoichiometric
parameters. Hence, lakes are valuable sentinels for detecting global changes, which in turn will have a major direct
and indirect effect on the biota of boreal lakes.
Acknowledgments
We are indebted to our colleagues in the project ‘‘Biogeochemistry in Northern Watersheds, a Reactor in Global Change’’
for stimulating cooperation and discussions, and we are grateful
to two anonymous reviewers for most-constructive comments and
suggestions.
The study is based on a regional survey financed by the
Norwegian Pollution Control Authorities conducted by the
Norwegian Institute for Water Research (NIVA) and forms part
of a Norwegian Research Council grant to D.O.H. (165139/S30).
References
BENOY, G., K. CASH, E. MCCAULEY, AND F. WRONA. 2007. Carbon
dynamics in lakes of the boreal forest under a changing
climate. Environ. Rev. 15: 175–189.
BERGSTRÖM, A.-K., P. BLOMQVIST, AND M. JANSSON. 2005. Effects
of atmospheric nitrogen deposition on nutrient limitation and
phytoplankton biomass in unproductive Swedish lakes.
Limnol. Oceanogr. 50: 987–994.
———, AND M. JANSSON. 2006. Atmospheric nitrogen deposition
has caused nitrogen enrichment and eutrophication of lakes in
the northern hemisphere. Glob. Change Biol. 12: 1–9.
CONLEY, D. J., C. L. SCHELSKE, AND E. F. STOERMER. 1993.
Modification of the biogeochemical cycle of silica with
eutrophication. Mar. Ecol. Prog. Ser. 101: 179–192.
2528
Hessen et al.
———, P. STÅLNACKE, H. PIKKANEN, AND A. WILANDER. 2000.
The transport and retention of dissolved silicate by rivers in
Sweden and Finland. Limnol. Oceanogr. 45: 1850–1853.
DILLON, P. J., AND L. A. MOLOT. 2005. Long-term trends in
catchment export and lake retention of dissolved organic
carbon, dissolved organic nitrogen, total iron and total
phosphorus: The Dorset, Ontario, study, 1978–1998. J.
Geophys. Res. 110: 1–7.
ERLANDSSON, M., I. BUFFAM, J. FÖLSTER, H. LAUDON, J.
TEMNERUD, G. A. WEYHENMEYER, AND K. BISHOP. 2008.
Thirty-five years of synchrony in the organic matter
concentrations of Swedish rivers explained by variation in
flow and sulfate. Global Change Biology 14: 1191–1198.
GALLOWAY, J. N., AND E. B. COWLING. 2002. Reactive nitrogen
and the world: 200 Years of change. Ambio 31: 64–71.
———, AND oTHERS. 2008. Transformation of the nitrogen cycle:
Recent trends, questions, and potential solutions. Science 320:
889–892.
HANSON, P. C., A. I. POLLARD, D. L. BADE, K. PREDICK, S. R.
CARPENTER, AND J. A. FOLEY. 2004. A model of carbon
evasion and sedimentation in temperate lakes. Glob. Change
Biol. 10: 1285–1298.
HENRIKSEN, A., A. HINDAR, D. O. HESSEN, AND Ø. KASTE. 1997.
Contribution of nitrogen to acidity in the Bjerkreim River in
Southwestern Norway. Ambio 26: 304–311.
———, AND oTHERS. 1998. Northern European lake survey—
1995. Finland, Norway, Sweden, Denmark, Russian Kola,
Russian Karelia, Scotland and Wales. Ambio 27: 80–91.
HESSEN, D. O. 1999. Catchment properties and the transport of
major elements to estuaries. Adv. Ecol. Res. 29: 1–41.
———, A. HINDAR, AND G. HOLTAN. 1997. The significance of
nitrogen runoff for eutrophication of freshwater and marine
recipients. Ambio 26: 321–325.
HIJMANS, R. J., S. E. CAMERON, J. L. PARRA, P. G. JONES, AND A.
JARVIS. 2005. Very high resolution interpolated climate
surfaces for global land areas. Int. J. Climatol. 25: 1965–1978.
HUMBORG, C., AND oTHERS. 2004. Nutrient variations in boreal
and subarctic Swedish rivers: Landscape control of land–sea
fluxes. Limnol. Oceanogr. 49: 1871–1883.
INTERLANDI, S. J., AND S. S. KILHAM. 1998. Assessing the effects of
nitrogen deposition on mountain waters: A study of
phytoplankton community dynamics. Water Sci. Technol.
38: 139–146.
JANSSON, M., H. ANDERSSON, H. BERGGREN, AND L. LEONARDSON.
1994. Wetlands and lakes and nitrogen traps. Ambio 23:
320–325.
JOHNSON, J. B., AND K. S. OMLAND. 2004. Model selection in
ecology and evolution. Trends Ecol. Evol. 19: 101–108.
KASTE, Ø., A. HENRIKSEN, AND A. HINDAR. 1997. Retention of
atmospherically-derived nitrogen in subcatchments of River
Bjerkreim in southwestern Norway. Ambio 26: 296–303.
KORTELAINEN, P., T. MATTSSON, L. FINÉR, M. AHTIAINEN, AND T.
SALLANTUS. 2006. Controls on the export of C, N, P and Fe
from undisturbed boreal catchments, Finland. Aquat. Sci. 68:
453–468.
MEYBECK, M. 1993. C, N, P and S in rivers: From sources to
global inputs, p. 163–193. In R. Wollast, F. T. Mackenzie and
L. Chou [eds.], Interactions of C, N, P and S biogeochemical
cycles and global change. Springer-Verlag.
MONTEITH, D. T., AND oTHERS. 2007. Dissolved organic carbon
trends resulting from changes in atmospheric deposition
chemistry. Nature 450: 537–539.
PETTORELLI, N., J. O. VIK, A. MYSTERUD, J.-M. GAILLARD, C. J.
TUCKER, AND N. C. STENSETH. 2005. Using the satellite-derived
NDVI to assess ecological responses to environmental
change. Trends Ecol. Evol. 20: 503–510.
RUSTAD, L. E., AND oTHERS. 2001. A meta-analysis of the response
of soil respiration, net nitrogen mineralization, and aboveground plant growth to experimental ecosystem warming.
Oecologia 126: 543–562.
SCHINDLER, D. W., P. J. CURTIS, S. E. BAYLEY, B. PARKER, K. G.
BEATY, AND M. P. STAINTON. 1997. Climate-induced changes
in the dissolved organic carbon budgets of boreal lakes.
Biogeochemistry 36: 9–28.
SKJELKVÅLE, B. L., AND oTHERS. 2005. Regional scale evidence for
improvements in surface water chemistry, 1990–2001. Environ. Pollut. 137: 165–176.
STODDARD, J. L. 1994. Long term changes in watershed retention
of nitrogen; Its causes and aquatic consequences, p. 223–284.
In L. A. Baker [ed.], Environmental chemistry of lakes and
reservoirs. Advances in chemistry, series no. 237. American
Chemical Society.
TUCKER, C. J. 1979. Red and photographic infrared linear
combinations for monitoring vegetation. Remote Sens. Env.
8: 127–150.
VITOUSEK, P. M., AND oTHERS. 1997. Human alterations of the
global nitrogen cycle: Sources and consequences. Ecol. Appl.
7: 737–750.
Associate editors: John P. Smol and Warwick F. Vincent
Received: 15 September 2008
Accepted: 19 February 2009
Amended: 09 March 2009