Você está na página 1de 5

Click

Here

GEOPHYSICAL RESEARCH LETTERS, VOL. 37, L11603, doi:10.1029/2010GL043617, 2010

for

Full
Article

Untangling the uncertainties about combined effects


of temperature and concentration on nutrient uptake
rates in the ocean
S. Lan Smith1
Received 15 April 2010; accepted 10 May 2010; published 9 June 2010.

[1] I show that assumptions about how uptake rates depend


on concentration strongly impact the interpretation of field
observations, specifically with respect to the combined
effects of temperature, T, and nitrate concentration, [NO3],
on nitrate uptake. The standard assumption that maximum
uptake rate, V max , is independent of ambient nutrient
concentration, S a , contrasts with the prediction of the
recently developed Optimal Uptake kinetics that Vmax should
increase hyperbolically with S a . Assuming Arrhenius T
dependence, I fit the respective equations to field
observations of chlorophyllspecific V max, T and [NO 3].
The inferred sensitivity to T differs by a factor of two.
Considerable uncertainty therefore remains about the T
dependence of uptake rates, and therefore about biological
production and biogeochemical cycles. Given that both
climate change and anthropogenic nutrient inputs are
altering the relationship between T and nutrients in the
ocean, these uncertainties limit our understanding of the
direct effects and associated feedbacks. Citation: Smith, S. L.
(2010), Untangling the uncertainties about combined effects of
temperature and concentration on nutrient uptake rates in the ocean,
Geophys. Res. Lett., 37, L11603, doi:10.1029/2010GL043617.

1. Introduction
[2] Smith et al. [2009] found that the observed pattern in
MichaelisMenten (MM) halfsaturation constants, KNO3,
as determined by shipboard experiments for uptake of the
important nutrient nitrate, agrees with the prediction of the
recently developed Optimal Uptake (OU) kinetics, based on
physiological acclimation by phytoplankton to optimize
uptake rate. Maximum uptake rates, Vmax, as fit to the MM
equation, vary more widely with environmental conditions
[Kanda et al., 1985; Harrison et al., 1996; Collos et al., 2005]
and are more difficult to interpret. Therefore, both Collos et
al. [2005] and Smith et al. [2009] chose to examine only the
pattern for KNO3. Yet understanding and quantitatively
modeling the largescale pattern of nutrient uptake in the
ocean requires consistently addressing the dependencies of
both Vmax and halfsaturation constants (or more generally,
affinities).
[3] Field experiments observe the combined effects of
temperature (T), ambient nutrient concentration, Sa, and
potentially light. In the surface ocean, T and Sa are strongly
negatively correlated [e.g., SilioCalzada et al., 2008]; cold,
1
Environmental Biogeochemical Cycles Research Program, RIGC,
JAMSTEC, Yokohama, Japan.

Copyright 2010 by the American Geophysical Union.


00948276/10/2010GL043617

nutrientrich water is upwelled, then warmed as nutrients are


depleted. Therefore, interpretation of the data (i.e., untangling
the effects) depends on assumptions about how uptake rates
depend on each factor, respectively.
[4] Arrheniustype T dependence of chemical reaction rates
is well established, as is a similar exponential T dependence
for phytoplankton growth rates [Eppley, 1972; Bissinger et al.,
2008]. However, Moisan et al. [2002] present an alternative
T dependence, based on simulations of the dynamic response
of a diverse assemblage of phytoplankton. Two competing
assumptions exist about how uptake rates depend on concentration. The most common assumption for decades has
been that Vmax (as fit to the MM equation) is independent of
Sa. However, the recently developed OU kinetics makes the
rather different prediction [Smith et al., 2009] that typical
shortterm incubation experiments should observe a saturating
increase in Vmax as a function of Sa, to which phytoplankton
were preacclimated upon sampling. Given that compared to
MM, OU kinetics has been found to better describe uptake
in chemostat experiments under limitation by N, P and
vitamin B12 [Smith and Yamanaka, 2007], the dependence of
KNO3 on ambient nitrate concentration ([NO3]a) in seawater
[Smith et al., 2009], and changes in Si and N uptake rates
during an oceanic ironfertilization experiment [Smith et al.,
2010], its prediction about Vmax should be considered as an
alternative.
[5] Here I examine the consequences of these different
assumptions about the nutrient and T dependence of Vmax for
interpreting field observations of uptake rates, specifically
with respect to the combined effects of T and [NO3] on
nitrate uptake. Furthermore, I seek to evaluate whether the
predictions of OU kinetics are consistent with field observations of both Vmax and Ks.

2. Methods
2.1. Data
[6] I analyzed data for chlorophyll (chl) specific Vmax for
nitrate from the shortterm (13 h), shipboard incubation
experiments (at ambient T) of Kanda et al. [1985] (North
Pacific, n = 17) and Harrison et al. [1996] (North Atlantic,
n = 60). G. W. Harrison (personal communication) provided
the latter data set as a digital file, from which I matched the
observed uptake rates to the ambient T and [NO3], by cross
referencing the cruise numbers, dates and T from that file
with the uptake rates and [NO3] as digitized from Harrison
et al. [1996]. Both data sets provide values of Vmax and
KNO3, with corresponding ambient T and [NO3]. However,
for Kanda et al. [1985], KNO3 must be estimated by subtracting the observed [NO3] from the reported estimate of

L11603

1 of 5

L11603

L11603

SMITH: UNTANGLING UNCERTAINTIES ABOUT UPTAKE

ature in K, S is the nutrient concentration, and Ks is the


MM halfsaturation constant for nutrient S. This is a multiplicative combination of Arrheniustype T dependence and
the Monod (or MM) kinetics for growth (or uptake) as
applied by Dugdale [1967]. Thus the MMbased the expression for dependence of Vmax on T is:
Vmax; MM Vmax;r eEa =RTr eEa =RT

In contrast, OU kinetics [Smith et al., 2009] predicts that


apparent values of both Vmax and Ks, as obtained by fitting
the MM equation to data from shortterm experiments, should
depend on Sa, the ambient nutrient concentration (in seawater) to which the phytoplankton were preacclimated
before the experiments:
app
Vmax

Ksapp

V0
q
V0
A0 Sa

r
V0 Sa
A0

where V0 and A0 are, respectively, the potential maximum


values of Vmax and affinity [Smith and Yamanaka, 2007].
Analogous to the equation of Goldman and Carpenter
[1974], combining Arrheniustype T dependence with OU
kinetics yields for uptake rate:
vOU

V0; r eEa =RTr eEa =RT


S
q
q
V0
V0 S a
1 A0 Sa
A0 S

where S is the concentration in the incubation flasks. This


gives the OUbased equation for Vmax as a function of T:
Vmax; OU

Figure 1. Schematic of the implications of combining


multiplicatively an exponential dependence of maximum
uptake rate, Vmax, on T with either MichaelisMenten or
Optimal Uptake kinetics, respectively, to describe the
dependence on ambient nutrient concentration Sa. Because
of the negative relationship between T and Sa in the near
surface ocean, the dependencies counteract one another in
the case of Optimal Uptake kinetics.
(KNO3 + [NO3]), whereas Harrison et al. [1996] report
values of KNO3 directly from nonlinear fits of the MM
equation to their data, which includes more precise measurements down to nanomolar [NO3].
2.2. Equations
[7] The equation most commonly applied to describe the
combined effects of nutrient concentration and T is that of
Goldman and Carpenter [1974], originally applied to growth,
but here written (as often applied) for uptake rate:
vMM Vmax; r eEa =RTr eEa =RT

S
Ks S

where Vmax, r is the Vmax at reference temperature Tr, Ea is


the activation energy, R is the gas constant, T is temper-

V0; r eEa =RTr eEa =RT


q
1 AV0 S0 a

I similarly tested MM and OUbased equations using the


T dependence of Moisan et al. [2002, equation 9] for the
effective population response in a fluctuating environment.
2.3. Fitting
[8] Log transformations of equations (2) and (6) were fit,
respectively, to values of logVmax, using observed ambient
T and [NO3] to calculate the fitted values of Vmax for each
data set, respectively. For the OU case, the ratio V0/A0,
which determines how the apparent values of MM constants depend on nutrient concentration (equations (3) and
(4)), was determined from fits of logKNO3 versus log[NO3]
as by Smith et al. [2009] for each data set, respectively.
This left two adjustable parameters to be fit in either case:
for MM (equation (2)), Vmax, r and Ea, and for OU
(equation (6)), V0, r and Ea. The nls algorithm in Splus
software (v. 8) was used for nonlinear fitting.

3. Results
[9] Qualitatively, the most striking difference between the
MM and OUbased assumptions is that the latter implies
that the greatest value of Vmax should occur at intermediate
values of both T and Sa (Figure 1), because the latter two are

2 of 5

L11603

SMITH: UNTANGLING UNCERTAINTIES ABOUT UPTAKE

L11603

[10] With Arrhenius T dependence, the fits (Table 1) of


the OUbased equations to the data were nominally better
than those of the MMbased equations. However, neither
model reproduces the wide variability in the data, and these
observed chlspecific rates are not ideal for testing the OU
predictions for biomassspecific rates (see section 4). Therefore, these results alone are not sufficient to draw conclusions
about which model is better. Still, with OU kinetics the
nutrientdependence of Vmax was modeled according to
equation (3), using values of the ratio V0/A0 obtained from
separate fits of logKNO3 vs. log[NO3], based on equation (4).
No inconsistency was found in these inextricably linked
predictions of OU kinetics about the respective patterns of
variation in Vmax and KNO3 as a function of Sa. Compared to
the exponential T dependence, fits to the data of Harrison
et al. [1996] using the T dependence of Moisan et al.
[2002], which adds one extra parameter, were worse for
MM and only slightly better for OU kinetics (Figure 3).

4. Discussion

Figure 2. Observed (circles) Vmax of nitrate from the short


term (12 h) incubation experiments of Harrison et al.
[1996] in the North Atlantic Ocean plotted versus (a, b)
ambient ocean T and (c, d) ambient nitrate concentration,
[NO3]a at time of sampling. Fits (crosses and lines) of two
different models, based respectively on MIchaelisMenten
(Figures 2a and 2c) and Optimal Uptake (Figures 2b and 2d)
kinetics for uptake as a function of concentration. Crosses
are bestfits of each model, respectively, using the observed
T and [NO3]a. Lines are model fits using bestfit parameters,
but assuming that T and [NO3]a follow exactly the loglog
correlation from this data set.

strongly and negatively correlated in the nearsurface ocean.


For the data set of Harrison et al. [1996] the inferred T
dependence, Q10 (the factor by which rates increase for a
10C increase in T), was approximately twice as strong
assuming OU versus MM kinetics (Table 1). This is because
of the counteracting effects of the two dependencies, as can be
seen by comparing the effects of each separately (Figures 2a
and 2b). Results were similar for the data of Kanda et al.
[1985] (not shown) with less difference between the MM
and OUbased estimates of Q10 (Table 1) because of the
lower estimate of V0/A0 for this data set. However, given
that the values of both KNO3 and [NO3]a were less precise
for this data set, which contains fewer data, I have more
confidence in the results based on the data of Harrison et al.
[1996]. Using that value, V0/A0 = 0.0807, to fit the OU model
to the data of Kanda et al. [1985] yields a Q10 of 3.8, with very
little difference in quality of fit (results not shown). Even for
the data set of Harrison et al. [1996], there
is considerable

p
uncertainty in the estimate of V0/A0: log( V0 =A0 ) = 0.546
with 90% confidence intervals of 0.118, equivalent to a
90% confidence range from 0.047 to 0.14 for V0/A0, which
results in a range of inferred Q10 from 3.17 to 3.74 (with little
difference in the quality of fit, results not shown).

4.1. Comparison With Growth Kinetics


[11] The finding that phytoplankton maximum growth
rates, mmax, increase exponentially with T, with Q10 = 1.8
[Eppley, 1972; Bissinger et al., 2008], is an idealization of
the interspecies variation in mmax based on fitting through
the top of the data (maximum value of mmax at each ambient
temperature, Ta). In controlled laboratory experiments growth
rates of individual species decrease above some species
specific optimal temperature, Topt. Application of exponential T dependence to model largescale growth (or uptake)
rests on the assumption of a continuum of species, such that
the dominant species at any time and place is the one having
Topt = Ta. Speciesspecific Topt are also observed for nutrient
uptake [Dauta, 1982; Tilman et al., 1981], and an analogous
idealization is possible for Vmax. However, that approach
cannot examine the combined effects of T and nutrients, nor
do I expect that it would yield meaningful results for the data
sets considered here, which contain far fewer values than
those analyzed for mmax.
[12] The Q10 estimated herein based on MM kinetics is
essentially the same as that reported for growth, but that
based on OU kinetics is twice as large. Furthermore, Lomas
and Gilbert [1999] have hypothesized that nitrate uptake
rates may decrease with T if coolwater diatoms are reducing
nitrate in order to dissipate excess energy under highlight
conditions. Their hypothesis implies a distinctly different
pattern of variation for MM constants for nitrate than for
other nutrients and would impact the interpretation of field
observations, because diatoms dominate phytoplankton
biomass in many nitraterich areas of the ocean.
[13] Moisan et al. [2002] present an alternative theoretical
argument that the greatest values of mmax should occur at
intermediate T, which if applied to Vmax could potentially
explain the wide variability observed. In their simulations of
assemblages in a fluctuating environment, the dominant
species at any given time did not have Topt = Ta, because of
the dynamics of competition. However, they made no
quantitative comparison to observations, and the specific
shape of their population response depended both on the
assumed speciesspecific shape of T dependence and on the
fluctuations (e.g., on seasonality, which varies with latitude).

3 of 5

L11603

L11603

SMITH: UNTANGLING UNCERTAINTIES ABOUT UPTAKE

Table 1. BestFit Parameter Values and Statistics for Fits of the Respective Equations for Vmax to the Data Sets From Field Experimentsa
Fit to Data Set
Kanda et al. [1985]

Harrison et al. [1996]

MM

OU

MM

OU

0.31 (0.030)

3570 (1540)
1.5

0.585 (0.049)
8110 (1300)
2.7

10.1 (1.08)

4410 (1090)
1.7

40.3 (4.92)
10250 (1240)
3.4

0.82
15

0.34
15

9.3
58

8.5
58

0.035

1.6 105

1.6 104

2.5 1011

Moisan et al. [2002] T Dependence


Parameter (units)
Vmax, r (nmol h1 (mg Chl)1
V0, r (nmol h1 (mg Chl)1
T opt (K)
Tscale (K)

5.30 (0.72)

285 (0.52)
1402 (272)

43.7 (7.14)
299 (12.8)
9326 (14700)

Residual Square Error (RSE)


degrees of freedom (df)

15.8
57

7.9
57

Arrhenius T Dependence
Parameter (units)
Vmax, r (nmol h1 (mg Chl)1
V0, r (nmol h1 (mg Chl)1
Ea/R (K1)
T Sensitivity (Q10)
Residual Square Error (RSE)
degrees of freedom (df)
Significance Level of fit for Ea/R
p<

a
SE, standard errors, in parentheses. Values of rate coefficients are for a reference T of 293 K. For fitting the OUbased equation, a value of V0/A0 = 7.07
103 was applied for the data set of Kanda et al. [1985], and 0.0807 for that of Harrison et al. [1996], based on separate fits of logKNO3 vs. logNO3 as by
Smith et al. [2009], for each data set, respectively. Fits of the equation of Moisan et al. [2002] to the data of Kanda et al. [1985] did not converge, and for
Harrison et al. [1996] in OU case the value of Tscale is less than its SE.

Therefore, their relatively more complex equation is not a


compelling alternative for quantitative largescale modeling.
4.2. Caveat
[14] The data analyzed here were for chlspecific Vmax,
whereas the predictions of OU kinetics are strictly valid on a
cellspecific basis and should ideally be compared to cell
specific, or at least biomassspecific, observations. Chl:C
ratio varies by a factor of from 5 to 10 depending on light
and nutrient environment [Flynn, 2003], and chl:N by at
least a factor of two to three [Geider et al., 1998]. Chl:N
ratios tend to be higher in nutrientrich areas, which tend to
have deeper mixed layers and therefore darker conditions,
compared to nutrientpoor areas with shallow mixed layers.
Therefore, relative to the data analyzed herein, Nspecific
rates would tend to be faster at high [NO3] (low T) and
slower at low [NO3] (high T), which would yield lower
estimates of Q10 based on either uptake kinetics.
[15] Still, given the lack of field observations of biomass
specific Vmax, the data sets analyzed herein are the best that I
know. Furthermore, OU kinetics will consistently result in
greater values of Q10 than MM kinetics, even for biomass
specific Vmax, because the nutrientdependence in OU kinetics
counteracts the T dependence. This is an inescapable result of
the negative correlation between T and Sa in the nearsurface
ocean. Although the data and quality of fits in this study alone
are insufficient to determine which model is superior, these
results together with those of Smith et al. [2009] do show that
OU kinetics provides a consistent interpretation for the
observed patterns of both Vmax and KNO3.

5. Conclusions
[16] Significant uncertainty remains about the T dependence of nutrient uptake rates, and therefore also in our

understanding of largescale marine biogeochemistry, most


acutely for our ability to quantitatively model nutrient cycles.
Although three recent studies [Smith and Yamanaka, 2007;
Smith et al., 2009, 2010] have shown OU kinetics to be a

Figure 3. Same as Figure 2, but with fits using the equation


of Moisan et al. [2002] for T dependence, with the assumption of symmetry about the effective optimal temperature for
the assemblage, T opt, implying a single value of Tscale.

4 of 5

L11603

SMITH: UNTANGLING UNCERTAINTIES ABOUT UPTAKE

superior alternative to MM kinetics, the predictions of the


former specifically for Vmax have yet to be tested against
controlled experiments. This means that in addition to the
quantitative uncertainty about Q10, there is also structural
uncertainty about the correct form of the equations for
describing uptake rates. This study has shown that structure
to be important for untangling combined effects in field
observations. If we are to understand the direct effects and
associated feedbacks, given that both climate change and
anthropogenic nutrient inputs [Duce et al., 2008] are altering
the relationship between nutrients and T in the ocean, these
uncertainties need to be reduced. This will require controlled
laboratory experiments examining the combined effects of
T and preconditioning to Sa, field observations of biomass
specific uptake rates, and data assimilation studies.
[17] Acknowledgments. I thank G. W. Harrison for providing the
data set, J. D. Annan, J. C. Hargreaves, J. Kanda, M. Kishi, A. Oschlies,
M. Pahlow, Y Yamanaka and N. Yoshie for helpful discussions, and the
anonymous reviewers.

References
Bissinger, J. E., D. J. S. Montagnes, J. Sharples, and D. Atkinson (2008),
Predicting marine phytoplankton maximum growth rates from temperature: Improving on the eppley curve using quantile regression, Limnol.
Oceanogr., 53, 487493.
Collos, Y., A. Vaquer, and P. Souchu (2005), Acclimation of nitrate uptake
by phytoplankton to high substrate levels, J. Phycol., 41, 466478.
Dauta, A. (1982), Conditions for phytoplankton development, comparative
study of the behaviour of eight species in culture. II. Role of nutrients:
Assimilation and intracellular storage, Ann. Limnol., 18, 263292.
Duce, R. A., et al. (2008), Impacts of atmospheric anthropogenic nitrogen
on the open ocean, Science, 320, 893897, doi:10.1126/science.1150369.
Dugdale, R. C. (1967), Nutrient limitation in the sea: Dynamics, identification, and significance, Limnol. Oceanogr., 12, 685695.

L11603

Eppley, R. W. (1972), Temperature and phytoplankton growth in the sea,


Fish. Bull., 70, 10631085.
Flynn, K. J. (2003), Modeling multinutrient interactions in phytoplankton:
Balancing simplicity and realism, Prog. Oceanogr., 56, 249279.
Geider, R. J., H. L. MacIntyre, and T. M. Kana (1998), A dynamic regulatory
model of phytoplankton acclimation to light, nutrients, and temperature,
Limnol. Oceanogr., 43, 679694.
Goldman, J. C., and E. J. Carpenter (1974), A kinetic approach to the effect
of temperature on algal growth, Limnol. Oceanogr., 19, 756766.
Harrison, W. G., L. R. Harris, and D. B. Irwin (1996), The kinetics of nitrogen utilization in the oceanic mixed layer: Nitrate and ammonium interactions at nanomolar concentrations, Limnol. Oceanogr., 41, 1632.
Kanda, J., T. Saino, and A. Hattori (1985), Nitrogen uptake by natural populations of phytoplankton and primary production in the Pacific Ocean:
Regional variability of uptake capacity, Limnol. Oceanogr., 30, 987999.
Lomas, M. W., and P. M. Gilbert (1999), Temperature regulation of nitrate
uptake: A novel hypothesis about nitrate uptake and reduction in cool
water diatoms, Limnol. Oceanogr., 44, 556572.
Moisan, J. R., T. A. Moisan, and M. R. Abbott (2002), Modelling the effect
of temperature on the maximum growth rates of phytoplankton populations, Ecol. Modell., 153, 197215.
SilioCalzada, A., A. Bricaud, and B. Gentili (2008), Estimates of sea surface nitrate concentrations from sea surface temperature and chlorophyll
concentration in upwelling areas: A case study for the benguela system,
Remote Sens. Environ., 112, 31733180.
Smith, S. L., and Y. Yamanaka (2007), Optimizationbased model of multinutrient uptake kinetics, Limnol. Oceanogr., 52, 15451558.
Smith, S. L., Y. Yamanaka, M. Pahlow, and A. Oschlies (2009), Optimal
uptake kinetics: Physiological acclimation explains the pattern of nitrate
uptake by phytoplankton in the ocean, Mar. Ecol. Prog. Ser., 384, 112.
Smith, S. L., N. Yoshie, and Y. Yamanaka (2010), Physiological acclimation by phytoplankton explains observed changes in Si and N uptake
rates during the SERIES ironenrichment experiment, Deep Sea Res.,
Part I, 57, 394408, doi:10.1016/j.dsr.2009.09.009.
Tilman, D., M. Mattson, and S. Langer (1981), Competition and nutrient
kinetics along a temperature gradient: An experimental text of a mechanistic approach to niche theory, Limnol. Oceanogr., 26, 10201033.
S. L. Smith, Environmental Biogeochemical Cycles Research Program,
RIGC, JAMSTEC, 317325 Showamachi, Kanazawaku, Yokohamashi,
Kanagawaken 2360001, Japan. (lanimal@jamstec.go.jp)

5 of 5

Você também pode gostar