Feb 25, 1980 - Amezaga, A. Sands, G. Redfield, G. Ma- lyj, C. Odelius ..... F-nitrate assimilation (same date as E). (-hours). 14C incubations. ...... ISAAC, 1974.
Limnol.
Oceanogr.,
Inorganic Richard
27(l),
1982, 53-65
nitrogen
P. Axler,
assimilation
Richard
in a subalpine
lake1
and Charles
R. Goldman
M. Gersberg,
Institute of Ecology and Division of Environmental University of California, Davis 95616
Studies,
Ahs tract Isotope tracer techniques were used to determine the seasonal patterns of inorganic nitrogen and carbon assimilation in the epilimnion of meso-oligotrophic Castle Lake, California. ammonium and nitrate indicated that the nitrogen Kinetics experiments with lSN-labeled assimilation rate was generally limited by low ambient concentrations of inorganic nitrogen (DIN). The half-saturation constants for DIN assimilation were low (0.7-Q-3 pg N*liter-l) and comparable to ambient levels during the nutrient-depleted, ice-free growing season. Ammonium assimilation was always higher than nitrate assimilation, even when nitrate concentration greatly exceeded ammonium during early summer. This observation was also supported by the results of 13N03-- uptake experiments. The rnidsummer proliferation of nonheterocystous blue-green algae was accompanied by increased carbon and nitrogen assimilation, and higher rates of particulate C and N turnover. Carbon to nitrogen and carbon to maximum nitrogen assimilation ratios were generally high and were not (34.5 +- 13.3 and 15.6 & 4.5 g C:g N), d i d not show obvious seasonal differcnccs, correlated significantly with seston compositional ratios. These results integrated with other Castle Lake studies suggest that phytoplankton growth is maintained by efficient use of internally regenerated ammonium, in agreement with many such studies in nitrogen-deficient marine waters.
daily rate of uptake of nitrogen by the phytoplankton. These results suggest that a quasi-steady state exists, whereby inorganic nitrogen concentrations remain extremely low due to a dynamic balance between ammonium sources and sinks. We wanted to determine the pattern of inorganic nitrogen assimilation relative to inorganic carbon assimilation in a lake with minimal external nitrogen loading and to find out which factors are most important in regulating nitrate and ammonium assimilation, and in turn, the rate of primary production. The few studies of this nature on freshwater ecosystems involved much more productive lakes (or reservoirs) (Brczonik 1968, 1972; Alexander 1970; Liao and Lean 1978a,b) characterized by tither high allochthonous inputs of nitrogen or significant amounts of limnetic nitrogen fixation. The conceptual model of Dugdale and Goering (1967) for the relative importance of the various sources of inorganic nitrogen for algal growth in oligotrophic ocean waters seems equally applicable to Castle Lake. Their hypothesis is that nitrogen regenerated in the euphotic zone is the primary source of inroganic N for
Castle Lake is a small, meso-oligotrophic lake in a granitic cirque basin in the Klamath Mountains of northern California. It has a euphotic zone deep (O-25 m) relative to its mixed layer depth (Z, = 5 m) and low levels of inorganic nitrogen and phosphorus (~10 pug [NH,+ + NO,+ NO,-1-N . liter-l; < 1 pg P043--P. liter-l) during most of the growing season. Allochthonous nutrient inputs are minimal during this period and nitrogen availability is most important in regulating phytoplankton growth during midsummer (Goldman 1978; Gersberg et al. 1981). Kimmel and Goldman (1977) reported high mineralization efficiency in the epilimnion and upper hypolimnion and found that ~20% of the areal particulate carbon and nitrogen production in the euphotic zone reached the sediment surface. Axler et al. (1981) showed that regeneration of NH4+-N via ammonification and zooplankton excretion in the epilimnion occurs at a rate similar to the
’ Castle Lake research support was provided by NSF grants BM74-02246 and DEB76-19524 to C. R. Goldman,
53
54
Axler et ul.
phytoplankton growth (“regenerated” production), but that to maintain these rates of primary production, external N loading must compensate for the continual losses of cellular nitrogen due to sedimentation and zooplankton grazing. In such systems annual production is ultimately limited by the extent to which “new” nitrogen inputs exceed these losses and this difference represents “new” production. This approach has been supported by marine studies (MacIsaac and Dugdale 1969; Goering et al. 1970; Eppley et al. 1973, 1979u,b), and we find a great deal of similarity in the patterns of N assimilation in an oligotrophic freshwater lake. We thank G. C. Anderson, J. R. Postel, N. J. Parks, N. Peek, K. Krohn, E. deAmezaga, A. Sands, G. Redfield, G. Malyj, C. Odelius, and M. Smith for their help. The impetus for this study was generated by discussions with B. L. Kimmel and L. J. Paulson. Methods Morphometric characteristics of Castle Lake are given by Kimmel and Goldman (1977). Water samples were collected from 3 m (midepilimnion) in the deep (35 m) central basin of the lake with a 3- or g-liter opaque Van Dorn bottle. Nitrate N was determined by a modification of the hydrazine reduction method (Mullin and Riley 1955; Kamphake et al. 1967). Ammonium N was determined by the method of Solorzano (1969) immediately after sampling, whereas nitrate samples were sometimes frozen at -20°C for l-2 weeks before analysis. Periodic tests for the recovery of both nutrients from enriched lake water samples were made. The precision of both methods was about * 1 pg N *liter-l (SD) for triplicates in the range of O-20 pg N. liter- l. Particulate C and N were determined (after 80-p filtration onto precombusted GF/C filters) with a Carlo-Erba elemental analyzer. Particulate N was also determined for each 15N filter by measuring the gas pressure of combusted samples at constant pressure and temperature before mass spectrometric analyses. These data were directly
calibrated against known weights of glutamic acid and urea and with parallel determinations using the CIIN analyzer. Experiments in summer 1980 indicated that 64-102-p prefiltration did not significantly affect either chlorophyll fluorescence or photosynthetic 14C uptake. Phytoplankton productivity was measured in situ by the 14C method of Goldman (1963) using 0.45-p Millipore filters. The amounts of incorporated 14C retained on Millipore filters ranging from 0.22- to 5-p pore size and on GF/C and ReeveAngel 984H glass-fiber filters (used for lfjN experiments) were not statistically different, which suggests that the rates of 14C and 15N uptake were determined for similar microbial communities. Dissolved inorganic carbon was measured immediately after sampling using an inwere from frared analyzer. Incubations 1000-1400 hours, and the rates of C uptake were then scaled to daily rates (PC) by multiplying them by the ratio of daily insolation : incubation period insolation. We generally did our N assimilation experiments either on the same day as or within 1 day of the primary production determinations. We avoided cloudy days whenever possible; when necessary, we interpolated between “14C days” and scaled the results according to differences in daily solar insolation (continuously recorded with a Belfort pyrheliometer). The accuracy of this procedure for estimating net daytime primary production has been supported by diel studies with epilimnetic water samples from Castle Lake (C. R. Goldman unpubl.). For nitrogen uptake rates we used the stable isotope r5N as a tracer. Replicate lake water samples (in l- or 2-liter bottles) were enriched with 99 atom-% 15NH4C1 or Na15N0, (after 80-p filtration to remove zooplankton) and incubated in situ for 24 h, after which the particulate material was filtered at low vacuum onto glass-fiber filters (precombusted GF/C or Reeve-Angel 984H), air-dried over silica gel, and stored at -20°C. Samples were later redried at 60°C for 24 h and prepared for mass spectrometry according to the micro-Dumas procedure of Gunther
Inorganic
nitrogen
et al. (1966). The fraction of 15N in the assay gas and rates of uptake were calculated according to Pavlou et al. (1974) and Neess et al. (1962). The rates of NH4+ uptake were then multiplied by a factor of 1.25 according to the results of a single diel experiment using 4-h incubations throughout the day (fLll1 details: Axler 1979). Nitrate uptake was generally below detection in 4-h experiments, and so these data are uncorrected. Final seston enrichments ranged from about 0.1 to 2.5 atom-% excess 15N. Results with final en!richments co.1 atom-% 15N above ambient controls were not used. Ambient rates of NH4+ assimilation (pNH,+) were calculated by assuming that the rate of assimilation was related to the ambient NH4+-N concentration according to the Michaelis-Menten (M/M) equation, s P = PmaxKt + s
where p is rate of uptake, S is substrate concentration (of the particular nitrogenous nutrient), pmax is rate of uptake at saturating levels of S, and Kt is half-saturation constant (at which p = pm,,/2). Regressions of S/p on S for experiments involving serial additions of NI14+ were then used to calculate the parameters Kt and pmax* Half-saturation constants for other dates were estimated by linear interpolation with respect to time for the intervals defined by the dates of kinetics experiments. Additional pmax values were available from other experiments done in 1978, and so rates determined for enrichments ~50 Fg 15N. liter-l were assumed equal to pmax. We also found a strong correlation (T 2 = 0.99, P < 0.01) between pmax and the rate for 9 yg 15N*litcr-1 enrichments, p(S + 9), and so we used the regression equation pmax = 1.26 p(S -k 9)-0.03 to estimate pmax when only p(S + 9) was determined. This relationship is a consequence of a low Kt, low ammonium levels, and the fact that the slope of the p vs. S curve is small for S B &. Although the precision of the calculated uptake rates was good (C.V. for 57 sets of triplicate rate determinations in 1977
assimilation
55
was 17 + 14%), their absolute accuracy is harder to estimate and no doubt considerably worse. Eppley et al. (1977) and McCarthy et al. (1977) emphasized that uncertainties in measuring low NH,+ concentrations limit the precision and accuracy of 15N-derived uptake rates for oceanic samples. For systems such as Castle Lake, where ambient NH4+ levels are comparable to &, a relatively small absolute error in NH,+-N is further amplified by use of the Michaelis-Menten equation. For example, if pmax = 1 pg N-liter-l *d-l, Kt = 2.3 pg N-liter-l, and S = 3 pg N *liter-l (typical values for 1978), an error of only 2 Fg N *liter-l, near the detection limit of the ammonium method, will result in an uncertainty of about 50% in the calculated p. Consequently, even without methodological errors, small diel differences in ambient NH4+-N (and therefore, the time of sampling) have critical importance. Full details of, the methodology involved with the 13N03- uptake experiments arc given elsewhere (Gersbcrg et al. 1978, 1981; Axler 1979). These experiments were not done in situ, but incubations were at ambient water temperatures, under artificial lights approximating epilimnetic irradiance levels. Results Kinetics experiments were done with epilimnetic (3 m) water samples in July 1977 and throughout the 1978 field season (Figs. 1, 2, Table 1). Half-saturation constants for ammonium uptake were uniformly low and comparable to ambient concentrations of ammonium in the Castle Lake epilimnion. Although the final seston 15N enrichments in the 15N03kinetics experiments were generally too low to yield meaningful estimates of &, there were enough data from 3 September 1978 to provide a half-saturation constant of 9.3 pg N03--N *liter-l (Fig. 2F), about three times the & of 3 pg NH4+-N. liter-l estimated for the same date. Midepilimnetic levels of dissolved inorganic nitrogen (DIN), particulate nitrogen (PN) and carbon (PC), and assimila-
Axler et al.
27 JULY
0
IO
20
1977
I
I
I
30
40
50
0
NHf CONCENTRATION ( pg Nd-‘1 Fig. 1. Ammonium assimilation rate (p) as a function of ammonium concentration (S) for 1977 epilimnetic water samples. Also plotted is linear regression equation for the S/p vs. S transform, which was used to estimate Michaelis-Mcnten parameters.
tion rates for inorganic carbon (PC), ammonium N (pNH4+), and nitrate N (pN03-, using Kt = 9.3 pug Neliter-l) for the ice-free portion of 1978 are presented in Fig. 3. An initial increase in primary productivity was followed by a midseason period of relatively constant PC and then a late season decrease coincident with lower water temperature and incident light as a result of storm activity. Ammonium concentrations were low (usually ~5 /-Lg N *liter-l) during the entire period, but NO,- was not depleted below the 5 pg N-liter-l level until late July. The relatively high levels of nitrate (>20 pug N *liter-l) during the early portion of the growing season were the result of a severe winter, producing high spring inputs of snowmelt-nitrate and a late thaw of ice. Nitrate derived from ammonium (by winter nitrification under the ice) released from anoxic sediments during the previous summer and fall can also contribute significantly to this “initial” pool (Axler 1979; Axler et al. 1980). Nitrate assimilation in the epilimnion
was low relative to ammonium assimilation even when nitrate levels were much higher than ammonium levels in June 1978. The maximum rate of nitrate assimilation was generally LO.1 pg N-liter-l * d-’ until mid-August. This result is consistent with our repeated failure to detect significant nitrate depletion in nitrate-enriched water samples and with the low rates in 13N03- and 15N03- uptake experiments in 1977 (Table 2). The 13N tracer technique has the advantage of much higher sensitivity than 15N, but the lomin half-life precludes in situ rate determinations and necessitates short incubations; minutes, in contrast to hours or days for jsN (Gersberg et al. 1978, 1981). In situ estimates from the two methods could not be directly compared because the 15N determinations for low enrichments were below detection all summer, except for the 16 July 1977 rcsult (which agreed well with the I9 July value for 13N). Estimates for pmax however, showed moderately good agreement except for the high 15N value on 6 September, for which we have no explanation. If Kt is assumed to be 9.3 pg N* liter-l (from the 3 September 1978 kinetics run), then we can calculate in situ NO,--uptake rates for the 13N run on 21 July (S = 2.5 pg N-liter-r, pmax = 0.246 pg N *liter-l* d-l) and the 15N run on 16 July (S = 1.0 pg N *liter-l, pmax 5 0.105 pg N *liter-l * d-l) using the MichaelisMenten equation. This yields values of 0.052 and 0.010 pg N *liter-l *d-l, in excellent agreement with the directly determined rates of 0.063 and 0.013, suggesting that the rates of nitrate uptake and assimilation are similar and that the low values we measured arc real. Specific rates of carbon and nitrogen (NH,+ + NOs-) assimilation were calculated by normalizing in situ rates to values of particulate C and particulate N, and these data are given together with maximum rates of DIN assimilation, PC, pN, seston C:N, and C:N assimilation ratios in Table 3. We have several cautionary notes regarding these data. The first qualification involves estimation of daily primary production (PC) from short term
Inorganic
nitrogen
assimilation
57
c
0
,D
0.80 -
T G -0 T v
- 20 07 AUG 1978
Z
IO
20
30
40
50
O
0
03 SEP 1978
- 20 30 JUN 1978
E
0
0
=) 0.30 -
03 SEP 1978
N Fig, 2.
CONCENTRATION
As Fig. 1 but for 1978. A-E-Ammonium
(-hours) 14C incubations. Unlike the light-dark 0, method, the 14C method does not permit respiration losses to be estimated; discussions of the exact meaning of the PC estimate have frequently appeared (Peterson 1978). Devol and Packard (1978) measured seasonal changes in Lake Washington in rcspiration directly by determining electron transport system (ETS) activity, and in PC by the same methods we used in Castle Lake. They found that the measured “daily primary production” was a good estimator of net daily photosynthetic production and that respiration losses at night represented some 3040% of this carbon fixation and presented similar results for a variety of aquatic systems. Therefore, the true net daily primary pro-
(pg
assimilation;
N I-’
F-nitrate
) assimilation
(same date as E).
duction for Castle Lake is probably better represented by -0.6 PC. In the context of our study, this distinction is most relevant with regard to C:N assimilation raTable 1. Kinetic NI&+ assimilation in m). & as pg Nelitcr-’ (*SD), 9 determined
parameters determined for epilimnetic water samples (3 (ISD), pmax as Fg Nsliter-‘*d-l for the S/p vs. S plots.
rq 14 27 14 30 31 7 3 3
Jul 77 Jul 77 Jun 78 Jun 78 Jul 78 Aug 78 scp 78 Sep 78”
* NO,-
assimilntion.
0.7k2.4 6.041.1 2.1k1.7 1.7+1.0 1.6+- 1.9 3.0-+ 1.4 3.0-+ 1.6 9.324.0
Pmox
1.37kO.02 7.71k0.26 0.3OkO.02 0.84 kO.03 0.9010.17 0.85kO.05 1.51-+0.11 0.36kO.04
r-2
0.99 0.99 0.96 0.98 0.84 0.97 0.95 0.89
n
9 9 11 9 5 10 10 10
58
Axler et al.
O-w_ 0’ /’
1fY \ \ IJ \ /’ %go PC ‘u’ P---O I
I .oo
0.75 0.50 0.25 0.00 Fig. 3. Seasonal patterns of concentration production (PC), maximum (P,,.,& and in situ kidepilimnetic water samples.
of particulate (pN03-,
tios, since we determined pN in 24-h incubations, C: N assimilation ratios were high, 34.5 and 15.6 for in situ and maximum N assimilation; but if a 40% rcspiration correction (discussed above) is applied to PC they are reduced to 20.7 and 9.4. A second point to be emphasized involves the use of VC (=pc:PC) and VN ( =~N:PN) as indicators of relative growth with respect to carbon and nitrogen. These specific uptake rates will underestimate true uptake to the extent that detrital C or N contributes to the total pool of PC or PN. There is no general agreement on how to estimate this error for nitrogen, although ATP carbon has been used as an estimate of living carbon, as-
pNI-L,+)
C, particulate N, IDIN, and rates of primary rates of nitrate aud ammonium assimilation for
suming that most of this is due to phytoplankton. Eppley et al. (1973) used the ordinal intercept of the linear regression of PN as a function of ATP carbon as a first approximation of nonliving nitrogen for North Pacific Ocean waters. Previous analyses of seston ATP-C in Castle Lake samples have shown that living C was a small fraction ((10%) of total PC (Jassby 1973; Goldman and Kimmel 1978). Neither PC nor PN correlated significantly with ATP-C. Linear regression analyses of epilimnetic PN concentration and chlorophyll Q concentration during 1978 were statistically significant however, and the ordinal intercept represented 68% of the average PN concentration (P 4 0.001). This suggests that, at most,
Inorganic
nitrogen
59
assi~milution
water, 1977. Incubation times Table 2. Estimates of nitrate uptake using 13N and 15N in epilimnetic were 24 h for 15N assays and 2-10 min for 13N assays, which were converted to daily rates by assuming NO:,- uptake occurs during a 14-h claylight period. Ammonium levels were low,