Meromixis in Hypersaline Mono Lake, California. 2. Nitrogen Fluxes

Meromixis in Hypersaline Mono Lake, California. 2. Nitrogen FluxesAuthor(s): Robert Jellison, Laurence G. Miller, John M. Melack and Gayle L. DanaSource: Limnology and Oceanography, Vol. 38, No. 5 (Jul., 1993), pp. 1020-1039Published by: American Society of Limnology and OceanographyStable URL: http://www.jstor.org/stable/2838090 .

Accessed: 10/06/2014 19:29

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

American Society of Limnology and Oceanography is collaborating with JSTOR to digitize, preserve andextend access to Limnology and Oceanography.

http://www.jstor.org

This content downloaded from 62.122.73.137 on Tue, 10 Jun 2014 19:29:26 PMAll use subject to JSTOR Terms and Conditions

http://www.jstor.org/action/showPublisher?publisherCode=limnoc

http://www.jstor.org/stable/2838090?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp


Limnol. Oceanogr., 38(5), 1993, 1020-1039 ? 1993, by the Amencan Society of Limnology and Oceanography, Inc.

Meromixis in hypersaline Mono Lake, California. 2. Nitrogen fluxes

Robert Jellison' Department of Biological Sciences and Marine Science Institute, UCSB, Santa Barbara, California 93106

Laurence G. Miller U.S. Geological Survey, Water Resources Division, Menlo Park, California 94025

John M. Melack Department of Biological Sciences and Marine Science Institute, UCSB

Gayle L. Dana Marine Science Institute, UCSB

Abstract Vertical fluxes of nitrogen were examined in hypersaline Mono Lake over a 9-yr period which encom-

passed the onset, persistence, and breakdown of meromixis. Under monomictic conditions, ammonia, which accumulates in the hypolimnion, is mixed into the euphotic region during autumn overturn. Following the onset of meromixis in 1983 and elimination of the winter period of holomixis, ammonia was depleted in the mixolimnion and accumulated beneath the chemocline. The mean rate of particulate nitrogen deposition, as measured by sediment traps over a 2-yr period during meromixis, was 2.0 mmol m-2 d-1. Until meromixis weakened in 1988, ammonia concentrations in the euphotic zone remained below 5 uM and increased to - 500 uM beneath the chemocline. Meromixis ended in November 1988 and a large pulse of ammonia was injected into surface waters, resulting in surface ammonia concentrations of -45 uM. Because the pH of Mono Lake is high (9.8) the NH3 : NH4+ ratio is --5, and elevated surface concentrations of ammonia during the 2 yr following breakdown of meromixis resulted in high losses of nitrogen via ammonia volatilization (mean, 10 mmol m-2 d-1). High release rates of ammonia from the sediments were estimated from both the ammonia gradients in pore-water profiles (3-10 mmol m-2 d-1) and the balance of mixolimnetic nitrogen fluxes (4-10 mmol m-2 d-'). The monimolimnetic balance suggested fluxes of ammonia out of the sediments below the chemocline were reduced during meromixis.

Saline lakes are common throughout the world and in arid regions are often the major aquatic habitat. Many are noted for high rates of annual production which provide essential resources for large populations of birds (cf. Hammer 1986). Because saline lakes usually lie in endorheic basins, their sizes, and thus salinities, are responsive to climatic variation and human activities that alter inflows. To be able to predict biological responses to changes in salinity requires an understanding of the fluxes of energy and nutrients in these lakes.

A feature of saline lakes in contrast to other

1 Present address: Sierra Nevada Aquatic Research Lab- oratory, University of California, Star Rt. 1, Box 198, Mammoth Lakes 93546. Acknowledgments

We thank Rebecca Todd, Marjorie Palchack, Lee Dyer, Darla Heil, Chuck Culbertson, and Ron Oremland for their assistance in the laboratory and field.

This research was funded by the Los Angeles Depart- ment of Water and Power.

lentic systems is the lack of outflows; hence, the potential for internal recycling to dominate nutrient fluxes is increased. Here, we assess internal recycling of nitrogen in Mono Lake, California, by examining the vertical fluxes of nitrogen in relation to changes in dissolved and particulate nitrogen over a 9-yr period (1982- 1990), including a rare 6-yr interval (1983- 1988) of meromixis (Jellison and Melack 1993b). We examine nitrogen fluxes between the mixolimnion and monimolimnion and estimate ammonia fluxes out of the sediments and lakewide losses due to ammonia volatilization.

Description of study site Mono Lake, a large, moderately deep, saline

lake, is renowned for its biological and geo- chemical features (Patten et al. 1987). The lake covers 160 km2 and has a mean depth of 17 m at an elevation of 1,943 m. It occupies a tectonic basin on the western edge of the North American Great Basin just east of the Sierra

1020



Nitrogen fluxes in Mono Lake 1021

Nevada, California (38?N, 11 9?W). Sodium is the major cation, and chloride and carbonate are the major anions, although sulfate, borate, and silicate concentrations are also high (Ma- son 1967; Jellison and Melack 1993b). The pH is 10, and salinity ranged from 77 to 98 g liter-' during the study. The lake was monomictic when it was studied in the early 1960s (Mason 1967) and late 1970s through early 1980s (Melack 1983). During 1982 and 1983 large freshwater inflows resulting from heavy winter snowfall during an El Ni-no-Southern Oscillation event led to the establishment of meromixis. Chemical stratification persisted until November 1988 (Jellison and Melack 1993b).

The plankton community of the lake has few species, as is typical of hypersaline waters. The phytoplankton are predominantly coccoid chlorophytes, coccoid cyanobacteria, and several bacillarophytes, mainly Nitzschia spp. (Mason 1967; Lovejoy and Dana 1977; Me- lack unpubl.). High abundances of an endemic brine shrimp, Artemia monica, and an alkali fly, Ephydra hians, provide an important food resource for large populations of migrating birds (Patten et al. 1987). A. monica is the only macrozooplankton species (Lenz 1984; Bowen etal. 1985).

The plankton of Mono Lake have marked seasonal cycles of abundance (Melack 1983). Phytoplankton are abundant throughout the lake in winter and increase in the epilimnion after the development of a thermocline in early spring. The spring increase in algae was much reduced during 1984 and 1985, subsequent to the initiation of meromixis, but gradually re- covered to previously observed levels from 1986 to 1988 (Jellison and Melack 1993a). A. monica hatch from overwintering cysts from January through May. By mid-May, the first adults are present, grazing the phytoplankton and causing a rapid decrease in algal abundance in the upper water column. During summer, the lake is separated into an upper region with sparse phytoplankton and abundant A. monica and a deep region with very few A. monica and a dense suspension of phytoplankton. In autumn, the phytoplankton increase in the surface waters as thermal stratification weakens and the A. monica population de- clines. Marked reductions in algal photosynthesis accompanied decreased nitrogen avail-

ability during a recent 6-yr period of meromixis (Jellison and Melack 1993a).

Methods Sampling and laboratory measurements-

The lake was sampled biweekly from March to August and monthly during the rest of the year from 1982 to 1990 except the autumn and winter months (October-February) were not sampled from 1982 to 1984. Additional surveys were conducted at weekly intervals in October and November 1988. Temperature and conductivity sampling and analysis are described by Jellison and Melack (1993b). Sam- ples for the determination of chlorophyll a and ammonia were collected at an eastern and western pelagic station at monthly intervals (Sta. 6 and 11; see figure 1, Jellison and Melack 1993b). The stations were sampled at 7-17 depths with an opaque Van Dorn water sampler. Water for nutrient analyses was filtered immediately upon collection through 25-mm Gelman A/E glass-fiber filters (pore size, 1 m). Samples for the determination of particulate organic N (PON) were collected from 1983 to 1987 with a section of large-diameter (3 cm) Tygon tubing, which integrated the upper 9 m of the water column, and from one or two deeper depths with a Van Dorn water sampler. PON and Chl a samples were immediately filtered through a 120-,um sieve to remove all stages of Artemia. All samples were kept cold (4-10?C) and dark during transport to the laboratory.

Deposition of particulate material was measured with sediment traps from 27 April 1986 to 2 December 1987. Each trap consisted of four replicate cylinders, each with an inner diameter of 7.3 cm and a height of 23.3 cm, resulting in an aspect ratio (height/inner diameter) of 3: 1. Gardner (1980) found that cylinders yield accurate measurements of vertical flux in nonturbulent waters if the aspect ratio is between 2 and 3. A properly propor- tioned cylinder measures 95-100% of the actual sedimentation rate (Bloesch and Burns 1980).

Traps were deployed in the western, southern, and eastern sectors of the lake at a depth of 24 m, which was below the chemocline in anoxic water and from 3 to 11 m above the bottom. Late in 1986, after cyst production by Artemia had stopped, low numbers of Artemia



1022 Jellison et al.

cysts were collected in sediment traps 3 m above bottom at the western station. We at- tributed these cysts to resuspension and moved this trap to a deeper station where it was deployed 7 m above bottom on 27 February 1987. The other two traps were 6 and 11 m above bottom.

The effects of tilting due to wind or currents were minimized by suspending the trap ver- tically between a buoy and anchor. A messen- ger was used to release caps which then covered the open end of the cylinders before they were raised to the surface. Water at the depth of the trap was collected with a Van Dorn sampler and analyzed for particulate N (PN).

The traps were sampled biweekly unless in- clement weather in autumn and winter forced less frequent sampling. Temperatures at the depth of trap deployment were 4-6?C throughout the study. At these low temperatures the biweekly sampling interval is sufficient to avoid significant losses of the material in the traps by anaerobic bacterial decomposition (Bloesch and Bums 1980).

Sediments were collected with a gravity corer at a western station (depth, 29 m) and sub- sampled within several hours by extruding in 2- or 5-cm-thick intervals (Miller et al. 1993). Pore waters were separated by centrifugation or pressure squeezing and filtered before analysis for dissolved ammonia. Additionally, short cores were collected with a Hangve corer (i.d., 2.9 cm; 30 cm long) from 10 to 21 September 1986 at 44 stations arranged in a grid across the lake and used to assess organic nitrogen content in the upper (0-5 cm) sediments. At each station, duplicate cores were taken and extruded into 1-cm sections. Sediment sections were stored in plastic bags and kept frozen until analysis in 1987.

Ammonia concentrations were measured with the indophenol blue method (Strickland and Parsons 1972). Except for mixed-layer samples during periods of low ammonia concentrations, samples were diluted by 2-10x with distilled-deionized (DI) water. The molar extinction coefficient in Mono Lake water (MLW) is smaller than that for DI water with the indophenol blue method. Therefore, molar extinction coefficients were determined for a series of dilutions of MLW and used to correct the reagent blank. Initially, a single internal standard was used for samples from through-

out the water column. In late 1984, tests indicated weaker color development due to unknown interferences in the anoxic monimolimnion. Beginning in 1985, internal standards were prepared with samples throughout the water column. Subsequent analysis re- vealed a near-linear increase in ammonia beneath the chemocline from 1985 to 1987 (see results). Assuming a linear increase from 1982 to 1985 indicates the bias due to weaker color development in the monimolimnion was neg- ligible in 1982, small in 1983, and large in 1984. A correction factor was applied to monimolimnetic samples for 1983 and 1984 based on an assumed linear increase in ammonia below the chemocline. Dissolved ammonia in pore water was analyzed within several hours of separation (Solorzano 1969) with standards prepared in 0.5 M NaCl or MLW. The pre- cision of the pore-water determinations was 2%.

Water-column profiles of nitrate and nitrite were determined in filtered samples during 1983-1985. Nitrate was reduced to nitrite with Cd-Cu reduction columns and nitrite measured colorimetrically with an azo dye technique (Strickland and Parsons 1972). Dis- solved organic N (DON) was determined during 1983-1985 by persulfate oxidation (Valderrama 1981); nitrate concentrations were determined as above. Internal standards were used throughout.

Water-column profiles of soluble reactive P (SRP) were determined with the molybdenum blue method (Strickland and Parsons 1972) on samples that had been diluted 100 x with DI water. Recent measurements used a method wherein determinations are based on the difference in absorbance of samples in which arsenate is reduced to arsenite (Johnson 1971).

The equilibrium between ammonia and ammonium is shifted predominately to ammonia in the lake. The ratio of NH4+ to NH3 was calculated at pH 9.8:

Ka = [NH3][H ]= 10-9= 3 [NH4+]

corrected for the effect of high ionic strength (I = 1.78, A. Maest pers. comm.) on the activity coefficient of NH4+ with the Davies equation (Stumm and Morgan 1970) and on the activity coefficient of NH3 with the equa-




tion of Randal and Failey (1927): yNH3 = 1.15, -yNH4+ = 0.78. The result was [NH3] = 5 x [NH4+].

Phytoplankton samples were filtered onto Gelman A/E glass-fiber filters (pore size, 1 ,um) and kept frozen at -14?C until pigments were analyzed. Chl a was usually determined by spectrophotometric analysis with correction for pheopigments (Golterman 1969) after a 40-min extraction of the macerated filters in 90% acetone at room temperature in the dark. Low Chl a concentrations (<5 mg m-3) were measured on a fluorometer (Turner 1 1 1) cal- ibrated against spectrophotometric measurements with large-volume lake samples.

Downward PN flux was determined from samples collected in the sediment traps. Du- plicate filters were prepared from each col- lecting tube by filtering 3-10 ml of sample through precombusted Gelman A/E filters and drying overnight at 40-50?C. Nitrogen was determined by combustion in a Perkin-Elmer 240B elemental analyzer standardized with ac- etanilide. Results were corrected by subtract- ing the concentrations found in the water samples collected at the trap depth. The elemental analyzer was also used to analyze PON in samples from the upper water column and the top 5 cm of the sediments. Wet and dry weights of sediment samples were determined before analysis to determine the porosity and salt cor- rections.

Numerical procedures-During meromixis, the upward flux of ammonia through the chemocline was estimated assuming a Fickian diffusion equation during the heating season and a simple entrainment process during the winter period of mixed-layer deepening. If we assume that eddy conductivities corrected for molecular conductivity (1.3 x 10-7 m2 S-1) approximate eddy diffusivities for ammonia, then

ON JN =- Kz

N

where JN is the upward flux of ammonia in mol m-2 S-', Kz the eddy diffusivity for ammonia in m2 s-', and aN/Iz the ammonia gradient in mol m-4. Eddy conductivities were derived with the flux-gradient heat method corrected for solar insolation (Jellison and Me- lack 1993b). Because the chemocline was deep and overlain by strong thermal gradients dur-

ing much of the heating season, the vertical heat flux was low and uncertainty in the heat budget using a three-station sampling program precluded analyzing short time periods. For this reason a single mean diffusivity for the thermally stratified period was derived for each year and the average ammonia gradient through the period used to calculate fluxes.

The flux-gradient heat method is applicable only during periods in which the heat content from the depth of interest to the bottom is increasing. During the cooling season, the upward flux is the sum of entrainment plus diffusive fluxes. To determine diffusive fluxes at the chemocline during the cooling period, the small (0.66 mS cm-' yr-1 from 1983 to 1988; Jellison and Melack 1993b) and roughly con- tinuous decrease in conductivity at 28 m can be used to estimate eddy diffusivity. The eddy difflisivity was estimated by integrating the flux of conductivity across the chemocline and di- viding by the appropriate conductivity gradient and time period. The mean conductivity gradient from October to March at the chemocline for the five winter periods (1983-1988) was 1.8 mS cm-' m-1. An upper-bound estimate of 2.3 x 10-7 m2 S-1 was derived for the winter eddy diffusivities by assuming all the conductivity flux across the chemocline occurred during winter periods. The entrainment fluxes were calculated as the sum of the ammonia content of horizontal volume elements through which the mixed-layer descended. The total quantity of ammonia in the lake was calculated by linearly interpolating between sampled depths and then calculating volume- weighted sums based on lake bathymetry (Pelagos unpubl. rep.).

Pore-water ammonia profiles were modeled by fitting an exponential equation that approached bottom water concentrations at the sediment-water interface (Miller et al. 1993). The concentration gradient was determined by the fitted slope at the sediment-water interface, and the flux of ammonia was calculated from Fick's first law of diffusion. The sediment diffusion coefficient for ammonia was obtained by correcting the molecular diffusion coefficients (DNH3, Himmelblau 1964; DNH4+, Li and Gregory 1974) for the effects of viscosity and tortuosity (Miller et al. 1993). Po- rosity at the sediment-water interface was 0.90.

Ammonia volatilization was determined by



1024 Jellison et aL

multiplying the surface ammonia concentrations (corrected for the speciation of NH4+ and NH3) by a gas transfer coefficient for each date a surface sample was obtained. The gas transfer coefficient was based on a two-film model derived for ammonia volatilization from rice paddies (Jayaweera and Mikkelsen 1990). Vol- atilization predicted by their model was in close agreement with results from both wind tunnel data and field experiments conducted in flooded rice paddies (Jayaweera et al. 1990). A qua- dratic equation:

K0 = 0.5107 + 0.2554v + 0.03944v2 r2 = 0.99

where K0 (cm h-') is the overall mass-transfer coefficient and v is the wind velocity (m s-') at 8 m was fitted to values presented by Jay- aweera and Mikkelsen (1990, table 4). This relationship was used in conjunction with the distribution of windspeeds at Mono Lake collected in 1989-1991 to determine mean monthly mass-transfer coefficients. A mean windspeed distribution for each month was derived from the frequency of windspeeds averaged over 10-min intervals in 0.5 m s-I bins. Windspeed data were not available for 1982- 1988. Jayaweera and Mikkelsen (1990) calculated the relative effect of temperature on the mass-transfer coefficient to be nearly linear and small (? 5%) over the range 10-400C. Mass- transfer coefficients calculated for Mono Lake were corrected for this temperature effect based on surface water temperatures.

The downward particulate flux of nitrogen consists of detrital material, live algal cells, and Artemia fecal pellets. Artemia produce small, dense fecal pellets encased in membranes (Reeve 1963) which sink rapidly (90-160 m d-1; Melack unpubl.). Because sediment traps were deployed for only 2 yr of the 9-yr period, a model of deposition that included algal settling and Artemia fecal pellet production was developed. The model assumes deposition is a linear function of algal biomass and Artemia grazing:

N = sA + eG

where N is deposition, s is a "piston velocity" and can be thought of as a settling rate of the algae (A), and e is the proportion of grazing (G) egested as fecal pellets. Grazing depends on the instar-specific abundance of Artemia,

temperature, and algal biomass. The grazing formulation assumes a constant instar-specific filtering rate up to a saturation level. These assumptions result in the following formulation of grazing:

gl = min{g X [1 -exp(g2 x wt,)] X tcf,

-2a X tcf X } Wtmax galg

where g, is the realized grazing rate, wt, the weight of instar i, Wtmax the maximum adult weight, galg a constant describing filtering rates below saturating levels of algal biomass, alg the ambient algal biomass, tcf a temperature correction factor for temperatures below op- timum, g, the maximum ingestion rate for adults, and g2 a constant determining the shape of instar-specific maximum grazing function. The formulation for instar-specific maximum grazing rates was based on feeding experiments conducted by Abreu-Grobois et al. (1991) with Artemia franciscana. This relation predicts higher grazing rates in early instars relative to older ones than a rate based solely on instar weight.

The formulation agrees with the results of feeding experiments conducted by Reeve (1963) which show a rapid increase in maximum grazing at low weights and early instars followed by constant maximal rates at weights equivalent to or greater than juvenile instars. The weight-instar relationship was based on three experiments that measured A. monica development under conditions representative of spring and summer conditions (Jellison et al. unpubl. rep.): 0.00031, 0.0054, 0.0093, 0.0162, 0.0280, 0.486, 0.0843, 0.3340, and 1.3234 mg dry wt for instars 1, 2, 3, 4, 5, 6, 7, 8-11, and adults, respectively. Abreu-Gro- bois et al. (1991) found Artemia adults to be capable of a maximum daily grazing rate of 70-80% of their body weight while earlier instars ingested much higher amounts relative to their weight. Oppenheimer and Moreira (1980) found that Artemia N: dry wt content ranged from 0.052 to 0.092 among instars.

Assuming a constant N: dry wt conversion of 0.07 for both Artemia and algae, our summer weight-length relationship, and the same relative grazing rates as a function of weight




600

E 5400 -

0)

R . . *.o

CL~~~~~~~~~~~~~~~~~R

0 20 40 60 80 Chlorophyll a (mg m4)

Fig. 1. Linear and multiplicative regressions of PN (1- 120 ,uM) on Chi a (n = 332). Linear: PN = 6.5 x Chi a + 70.5, r2 = 0.64. Multiplicative: PN = 49.53 x Chl a?5098, r2= 0.66.

found by Abreu-Grobois et al. (1991) resulted in grazing coefficients of 0.05 9 mg N d-l for g1 and -107 for g2. A normal curve was used to represent the response of grazing to temperature over the observed range:

[ (tgrZtopt)

where top is the temperature of optimal Arte- mia growth, t the ambient temperature, and tgrz a constant that defines the width of the curve describing the temperature effiect. The optimal temperature was assumed to be 30-C; Hemnandorena (1976) found higher growth rates at 30?C than at 25?C for A. salina, and tgrz was estimated to be 1 3?C through least- squares fitting to data from laboratory experiments (Jellison et al. unpubl. rep.).

The lakewide PN in Artemia was calculated by assuming a N: dry wt ratio of 0.07, the above instar-specific weights, and instar-specific Artemia abundance determined from triplicate vertical net tows at 10 lakewide stations.

PN in the size range 1-120,u.m was measured in a subset of water samples collected from 1982 to 1987 and found to be highly correlated with Chl a concentration. Although a linear and multiplicative regression explained about the same amount of the total variation, the residuals were more uniformly distributed about the multiplicative model. Also, the multiplicative model is more reasonable because samples containing high Chi a were collected from low light regimes during late autumn to

80 a) E

> 60 epth c a)

CoZ 'E E 40

a-

1983 1984 1 985 1986 1987 1988

Fig. 2. Relative depth, area, and volume beneath the chemocline based on 1987 Pelagos bathymetry (Pelagos Corp. unpubl. rep.), LADWP surface elevations (pers. comm.), and maximum conductivity gradients. Secondary chemocline, which formed above the primary chemocline in 1986, is not shown.

spring and algae are known to increase their Chi a content under low light conditions (Fal- kowski and LaRoche 1991). A multiplicative regression of PN on Chl a explained 66% of the total variation in PN (Fig. 1).

PN = 49.53 x Chl a05098 n = 332, r2 = 0.66.

Because the 1-1 20-,um size class is dominated by algal cells and changes in N were correlated with Chl a, PN in this size class was referred to as algal N. The above regression was used in conjunction with the extensive Chl a data to estimate lakewide algal N for the entire period.

Results Meromixis, phosphorus, and nitrogen-In

February 1982, the water column was well- mixed to the deepest sampling depth (30 m). During 1982 above-normal inflows into the lake resulted in a 0.6 m rise in lake level. Fresh- water flowed above the seasonal thermocline in 1983, and a large salinity gradient was es- tablished between 8- and 12-m depth. The chemocline persisted for 6 yr, over which time it gradually deepened and weakened until holomixis occurred in late November 1988. Al- though initially above 12 m, the chemocline successively deepened during autumn mixing of each year except 1986 (Fig. 2). In 1986 high inflows led to establishment of a secondary chemocline higher in the water column. Al-



1026 Jellison et aL.

0

5

10

~15

0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~11

20

25 o

0I 0

0 0~~~~~~~~~~ 30~~~~~~~~~~~~~~~~~~~

1982 1983 1984 1985 1986 1987 1988 1989 1990 Fig. 3. Ammonia (+ ammonium) concentrations (4M). Contours based on monthly samples taken from 7 to 17

depths.

though this secondary chemocline was de- stroyed by mixed-layer deepening in autumn of 1986, there was little downward displace- ment of the original chemocline in 1986. The area of the lake beneath the chemocline declined from 63% on 1 June 1983 to 33% on 18 October 1988; the volume of the lake beneath the persistent chemocline declined from 46 to 11% over the same period (Fig. 2).

SRP determinations, uncorrected for As in- terference, were high (>600 ,uM) throughout the water column during the entire study period. Recent measurements on March and September 1992 samples with a method that simultaneously determines phosphate and arsenate indicated high and nearly uniform arsenate concentrations throughout the water column (mean, 76 ,uM). Arsenate accounted for a roughly constant proportion of the absorbance of the molybdate blue complex, 0.265 (SD = 0.032). SRP was similarly high and

nearly uniform with depth in both March and September 1992 with a mean concentration of 460,uM. Given these findings, it was certainly > 400,uM throughout the period and well above saturating concentrations for algal uptake. Thus, no further consideration was given to P dynamics.

Nitrite and nitrate concentrations were low throughout 1983-1985. The overall mean nitrite concentration was 0.11 ,uM (SE = 0.04, n = 31), and the mean nitrate plus nitrite concentration was 0.54,uM (SE = 0.04, n = 178). DON concentration was always high (mean, 145 ,uM; SE = 5.2, n = 88).

Large changes in ammonia concentrations in the mixed layer accompanied the onset and breakdown of meromixis (Fig. 3). In August 1982, prior to meromixis, epilimnetic concentrations were 5-10 ,M. During meromixis from 1983 to early 1988, mixolimnetic concentrations were always < 5 ,M and often <0.5 ,uM.




Despite continued meromixis, mixolimnetic ammonia concentrations were elevated on one date (3 February) in early 1988 due to entrainment during winter mixing. Ammonia concentrations also increased to between 5 and 10 ,uM in midsummer 1988 as the thermocline deepened, entraining ammonia-rich water. This small ammonia increase was followed by an abrupt increase to >45 ,uM when holomixis occurred in November 1988. In 1989, epilimnetic ammonia concentrations were tempo- rarily reduced (< 10 ,uM) from April through June during a period of high algal production; however, concentrations increased through the summer asArtemia grazing converted algal PN to ammonia via grazing and excretion. Am- monia returned to 25-50 ,uM from August through December. In 1990, the seasonal pattern was similar to that in 1989, but concentrations were lower. From March through No- vember concentrations remained <10 ,uM, whereas during spring and autumn algal blooms values were < 1 and 5 ,uM, respectively.

Monimolimnetic ammonia concentrations increased from the onset of meromixis until late 1987 when concentrations exceeded 500 ,uM (Fig. 3). From mid-1984 through 1988, ammonia concentrations in the monimolimnion increased at a rate of 100 ,uM yr-t. In 1989 following breakdown of meromixis, hypolimnetic concentrations increased from 45- 50 ,uM during winter holomixis to 80-90 ,uM during the course of summer thermal stratification. Concentrations were lower (25-30,uM) during the 1989-1990 winter period of holomixis compared to 1988-1989 immediately after breakdown of meromixis. During the thermally stratified portion of 1990, ammonia increased to 65-80 ,uM before holomixis in November, after which concentrations dropped to 12-15 ,uM throughout the water column.

Total lakewide ammonia was at a minimum of 20 Mmol on 3 August 1982 and increased at -50 Mmol yr-1 to 300 Mmol by late 1987 (Fig. 4A). Total ammonia decreased in early 1988 as the mixed layer deepened and, prior to holomixis in late 1988, dropped from -220 to 110 Mmol. Lakewide ammonia continued to decline after the breakdown of meromixis, reaching 22 Mmol in early 1990.

Total lakewide algal N ranged from 12 to 79 Mmol and showed a strong annual pattern of high winter and low summer values (Fig. 4B).

400 I I I I I I

300 Ammonia A-

200 -

100 -

c) 0 CD 1982 1984 1986 1988 1990

E 200 S 200 * Aigal N

150 - ArtemiaN N

0) 1 00- _ 0

50-

-a 1982 1984 1986 1988 1990

03) -" 400.

300 -

200 -

100 - Ammonia + Algal + Artemia N

1982 1984 1986 1988 1990

Fig. 4. Changes in total lakewide nitrogen in dissolved and particulate pools. Open circles in panels A and C (1983 and 1984) reflect indirect estimate of monimolimnetic ammonia based on assumed linear accumulation (see methods).

In 1982, prior to meromixis, the algal-N pool was about equal to the ammonia pool. Once the lake became meromictic, the ammonia pool rapidly increased, becoming much greater than the algal-N pool. During most of meromixis, annual variation in the algal-N pool was smaller than either before (1982) or after (1989- 1990), reflecting decreased size of the spring and autumn algal blooms. The annual variation increased during 1988, as weakening of meromixis resulted in larger algal blooms, and in 1989 and 1990 after meromixis ended. Dur- ing 1989 and 1990, following meromixis, the algal-N and ammonia pools varied inversely with each other with a seasonal variation of 40-60 Mmol.

The Artemia population was near zero in winter and high (>50,000 ind. m-2) in summer. Before meromixis (1982) and immediately after (1989), midsummer peak abundance was much higher than in other years.




300 - AmmoniaA

200 -

I-_ 100 - U,

G O E 1983 1984 1985 1986 1987 1988 Co

30

- 30

0) 20 -B 0

C o - AlgaliN 0

E . 1983 1984 1985 1986 1987 1988

E300- o Ammonia +Algal N c 2 200 -

100 -

1983 1984 1985 1986 1987 1988 Fig. 5. Changes in dissolved and particulate pools of

nitrogen beneath the chemocline. Open circles in panels A and C as in Fig. 4.

The calculated nitrogen in Artemia tissue varied from 0 to 54 Mmol during meromixis with higher peaks before (1982: 110 Mmol), during the breakdown (1988: 88 Mmol), and after meromixis (1989: 167 Mmol) (Fig. 4B). The annual variation in Artemia N was larger than it was in algal N.

The prominent peak in Artemia N in 1989 results from estimates of Artemia abundance on a single date, 13 September 1989. Because the decline in the algal-N pool from 10 August to 13 September was only 7 Mmol, the rapid increase from 73 to 168 Mmol of Artemia N between the two sampling dates implies a large upward flux of ammonia. The thermocline descended 3 m during this period and entrained

12 Mmol of N. Thus, only a 1 9-Mmol increase in Artemia N could be accounted for by N entrained during deepening of the thermocline and passing through algae to Artemia. However, the Artemia abundance data indicate a 95-Mmol increase. On 13 September,

adult abundance was 92,500 m-2 (SE = 25,700) based on a 10-station mean of triplicate vertical net tows at each station. On this date Artemia was more patchily distributed on a lakewide basis than usual. The C.V. was 88%, which is much higher than the mean C.V. for the entire period (22%). Therefore, a portion of the prominent Artemia peak in 1989 is likely due to sampling error accompanying higher than usual patchiness.

The 6-yr pattern of lakewide PN (algal + Artemia) was opposite that of the ammonia pool during meromixis. Total PN was high just before meromixis in 1982, decreased during meromixis, and increased immediately after. Because the ammonia pool was much larger than the particulate pool, the general pattern of the sum of particulate and ammonia pools was similar to the ammonia pool with the ex- ception of the peak in late 1989 due to the high abundance of Artemia (Fig. 4C).

The decline in lakewide ammonia began during mixed-layer deepening late in 1987 and continued through 1990. The sharpest decline occurred in 1988 as the chemocline first deepened and then mixed completely in Novem- ber, at which time all the remaining 150 Mmol beneath the chemocline were redistributed throughout the water column. Much of this ammonia was incorporated into the particulate pool as a pronounced algal bloom occurred in autumn 1988. Total N (ammonia, algal, and Artemia) decreased - 100 Mmol in 1988. Dur- ing 1989 and 1990, total N declined more slowly (avg annual rate, 50 Mmol yr-').

Nearly all the lakewide increase in ammonia during meromixis was due to gradual accumulation beneath the persistent chemocline. Despite deepening of the chemocline, and thus decreasing monimolimnetic volume, total lakewide ammonia in the monimolimnion increased until 1987 (Fig. 5A). The increase in 1985-1987 was approximately linear (r2 = 0.57) at an annual rate of 43 Mmol yr-'. The x-intercept of the linear regression describing the increase from 1985 to 1987 is 25 Novem- ber 1981. The decrease in algal N beneath the chemocline reflects decreasing volume as the chemocline deepened over time because algal concentrations were almost constant there (Fig. 5B). Artemia is limited to oxic water and thus does not contribute to PN beneath the chemocline. The trend of algal N plus ammonia in




the monimolimnion was dominated by changes in dissolved ammonia (Fig. 5C).

Estimates of lakewide N on any date are uncertain due to the few depths sampled, a pronounced chemocline, and potentially large gradients near the bottom. An estimate of the uncertainty is 20-50% based on examining the variation on consecutive dates (Fig. 4). Here, the focus is on periods in which lakewide N varied over 10-fold.

Nitrogen fluxes-Vertical fluxes of N were calculated for three periods of interest: 1983- 1987 which includes the buildup of lakewide nitrogen, the breakdown of meromixis in 1988, and the postmeromictic period, 1989-1990. Estimates of the upward flux of ammonia and its volatilization during the three periods are examined, and the deposition of PN is extrap- olated to the three periods with a model cali- brated to data collected from sediment traps over a 2-yr period. Sediment release of ammonia was estimated as the balance among upward flux, volatilization, and deposition and observed changes in lakewide nitrogen (ammonia, algal, and Artemia N).

The lakewide mean measurements of PN deposition ranged from 1.0 to 4.7 mmol m-2 d-l with a mean of 2.0 (Fig. 6A). A nested ANOVA in which nesting included date, station, sediment trap tube, and duplicates from each tube was performed. In 1986, date and station contributed equally (38 and 35%) to the variance, while the variance among collection tubes was less important (20%). Vari- ance of duplicates, which included subsam- pling and analytical procedures, contributed the least (7%). In 1987 nearly all of the variance (80.5%) was due to date, while lesser amounts were attributable to station (4.2%), tube (8.8%), and replicate (6.5%). The average C.V. of lakewide estimates was 17% in 1986 and 14% in 1987. Average rates of lakewide deposition were similar in 1986 and 1987; midsummer fluxes were mostly between 2 and 3 mmol m-2 d-l, and late autumn to early spring fluxes were between 1 and 2 mmol m-2 d-1.

A least-squares fit of the model of particulate deposition to the sediment trap data yielded an estimate for fecal pellet production of 7.1% of the grazing rate. Given a 44% net assimi- lation efficiency (Reeve 1963), the remaining 48.9% is presumably excreted, primarily as ammonia. The model estimate of algal settling

4

3

2

E , E 1986 1987

0 ._,

O * FecalN B (D o Algal N

-o

C 2 - < 1 < 0

1986 1987

Fig. 6 A. Actual (- -) lakewide deposition of nitrogen measured in sediment traps vs. rates predicted with model of algal sinking and Artemia fecal pellet production (. ). Error bars indicate ? 1 SE. B. Model estimates of algal sinking vs. fecal pellet production.

was 0.12 m d-1. The model captures a significant portion (32%) of the observed seasonal variation in deposition over the 2-yr period (Fig. 6A) and indicates the two components of deposition-algal settling and fecal pellet sinking-are of about the same magnitude but inversely correlated during the year (Fig. 6B). Although the above model of deposition includes many sources of uncertainty, it is likely to provide a better estimate of deposition than simply extrapolating the observed deposition rates to the entire period because algal biomass and Artemia abundance varied from year to year during the period of interest.

The model of deposition was used with algal and Artemia biomass to estimate the deposition of nitrogen over the entire 9-yr period. Algal biomass in the mixed layer, as measured by Chl a concentration at 2 m, ranged from near zero to 80 mg Chl a m-3 (Fig. 7A). In addition to a strong annual signal of summer




A 60-

40-

:E

020

C 1982 1984 1986 1988 1990

15

C\B 12 B z Z) 9

Z3

C 1982 1984 1986 1988 1990

Fig. 7. A. Chl a concentrations at 2 m. B. Lakewide estimates of Artemia biomass integrated over the water column.

minima and winter-spring maxima, Chl a concentrations were significantly lower through most of the meromictic period (1983-1988) than during nonmeromictic periods (1982, 1989, and 1990). During the last year of meromixis (November 1987-November 1988), chemical stratification had weakened and upward flux of ammonia was significantly higher (see below), and algal biomass increased ac- cordingly. Although peak Artemia biomass re- vealed a similar trend of lower values during meromixis, the annual peak in 1990, 2 yr after meromixis ended, was also low (Fig. 7B).

The model predicted deposition to range from 0.5 to 10.6 mmol N m-2 d-l (Fig. 8A). Temporal changes in algal and fecal pellet deposition (Fig. 8B) reflect changes in both ambient algal biomass (Chl a) (Fig. 7A) and Ar- temia (Fig. 7B). The mean deposition rates were 1.3 and 1.1 mmol N m-2 d-l for algal settling and fecal pellet sinking.

The upward flux of ammonia through the chemocline was estimated for the period 23

12 I I I I I I --

Total A

8 CM

4

E L E C 1982 1984 1986 1988 1990 0

12 a Q * Fecal B

0 ~~~~Algal )8 -

0)

z 4 -

0 1982 1984 1986 1988 1990

Fig. 8. A. PN deposition calculated from a model of fecal pellet productioil and algal settling. B. Model-predicted partitioning of nitrogen deposition between algal material and fecal pellets.

March 1983 to 25 November 1988. Eddy diffusivity at the chemocline ranged from molecular in 1986 to 22.8 m-2 S-1 in 1988 (Table 1). The summer estimate of difflusivity decreased from 1983 to 1986 and then increased to a maximum during 1988, when the vertical density gradient approached zero before the end of meromixis in November. Minimum values in 1986 coincided with the formation of a secondary chemocline higher above the initial chemocline due to above-average run- off. During the heating season in 1986, no change in heat content was measured beneath the persistent chemocline. Thus, in 1986 the estimated upward flux was based on the molecular diffusivity of ammonia (1 x 10-9 m2

s-') for this period. Although the vertical ammonia gradients were greater in 1987 than during the first years of meromixis, the chemocline was deeper in the water column (20-21 m) and eddy diffusivity was lower than that calculated for 1983-1985. Thus, upward ammonia flux was also low in summer 1987.




Table 1. Upward ammonia flux through the chemocline.

Upward Depth of Length of Eddy Area of Ammonia ammonia Winter

chemocline period diffusivity chemocline gradient flux entrainment Period (m) (d) (10-7 m2 S-') (kM2) (mmol m-4) (Mmol) flux (%)*

23 Mar-14 Sep 83 12 176 10.35 104.7 2.8 4.6 15 Sep83-4Apr84 12-15 203 2.3t 102.4 10 5.9 31 5 Apr-20 Oct 84 15 199 8.39 97.7 24.4 34.4

21 Oct 84-21 Mar 85 15-17 152 2.3t 93.7 90 31.9 20 22 Mar-23 Aug 85 17 155 2.32 90.1 67.5 18.6 24 Aug 85-13 Mar 86 17-19 202 2.3t 85.0 55 33.0 43 14Mar-14 Oct86 19 215 0.01 83.0 45.9 0.1 15 Oct 86-5 May 87 19-20.5 203 2.3t 79.6 25 32.5 75 6 May-14 Nov 87 20.5 193 0.10 75.8 113.2 1.4

15 Nov 87-11 Apr 88 20.5-23.5 149 2.3t 68.5 50 49.1 79 12 Apr-22 Oct 88 23.5 194 22.80 59.8 59.0 128.0 23 Oct-25 Nov 88 20.5-23.5 33 2.3t -30 25 139.7 100 * Column indicates percent of total upward flux which is due to entrainment. t Upper estimate of mean winter eddy diffusivity at the chemocline for entire period (see methods).

Fluxes increased further during winter mixed- layer deepening in 1987-1988 and then tripled during 1988 as vertical mixing increased. The breakdown of meromixis in October-Novem- ber 1988 resulted in a 139 Mmol pulse of ammonia into the mixed layer.

During meromixis the loss of nitrogen from the mixolimnion via volatilization of NH3 was estimated to range from 0.2 to 3.8 mmol m-2 d-l (Fig. 9). During the breakdown of meromixis in 1988 and the following 2 yr, volatilization increased and exceeded 30 mmol m-2 d-l in early and late 1989. On a lakewide areal basis the loss of nitrogen due to volatilization was 3.5 Mmol d-l in spring and autumn 1989. Sediments-Shallow stations within the 44-

station grid tended to have sandy sediments, while deeper stations were dominated by silt.

40 I l l

-o E 30 - -5 E E

O 20-

0 > 10-

z 0

1982 1983 1984 1985 1986 1987 1988 1989 1990

Fig. 9. Ammonia lost to atmosphere via volatilization.

The organic nitrogen content of individual core sections ranged from 0.2 to 15.9 g kg-' with a mean of 5.6 (SD = 3.1, n = 273). Organic nitrogen increased both as a function of station depth and depth in the sediment. Mean organic N at shallow stations (Z < 10 m) increased from 3.3 g kg-1 in the top centimeter to 4.8 between 2 and 3 cm (Fig. 10). Mean nitrogen content was higher at deep stations (Z 2 20 m), varying from 5.9 to 10.1 g kg-'. With station depth as a covariate, nitrogen content significantly increased with increasing depth in the sediment over the first three 1-cm core sections (P < 0.001). There was no significant difference among nitrogen contents of the third, fourth, and fifth core sections.

0 4 I 0 1

0

E

a)2

CO3

-z- 0_

Shallow Intermediate Deep 5 I 2 4 6 8 10 12

Organic nitrogen (g kg 1)

Fig. 10. Organic nitrogen content in sediments at shallow (depth < 10 m, n = 13), intermediate (depths, 10-20 m, n = 12), and deep (depth 2 20 m, n = 19) stations based on 44 stations cored in September 1986. Error bars indicate ? 1 SE.




0

20

40

60

O Aug 1986 v) 100 E Nov 1986 ( AAug 1988

.E 120 I l a) 0 1000 2000 3000 4000 5000

c0

2

6

VMayl1986 8 OAug 1986I

El Nov 1986 AAAug 1988

10 - 0 1000 2000 3000 DISSOLVED AMMONIA (j.M)

Fig. 11. Pore-water ammonia profiles in cores collected at a pelagic station in the westem basin (station depth, 29 m). A. Full profiles. B. Detail of upper 10 cm. Uncer- tainty in the pore-water concentration gradient was based on exponential profiles bracketing the observed data (dashed lines).

Dissolved interstitial ammonia concentration increased with depth from 1.1 mM at 1.0- cm to 4.8 at 83.5-cm depth (Fig. 1 lA). Gra- dients in the upper 10 cm were large. The best-fit ammonia concentration gradient at the sediment-water interface was 0.82 mM cm-l, with a range of 0.44-1.32 based on the model

required to bracket the data from three profiles (Fig. 1 1B). The calculated ammonia flux out of the sediments using a corrected diffusion coefficient of 0.83 ? 0.05 cm2 s- was 6.0 mmol m-2 d-1 with a range of 3.1-10.4 based on the error in the diffusion coefficient and the range in the concentration gradient.

Discussion A large climatic perturbation caused the on-

set of meromixis in Mono Lake (Jellison and Melack 1 993b) and provided conditions where internal fluxes of nitrogen were altered. The vertical flux of ammonia was reduced throughout the year due to decreased turbulent mixing and the absence of winter holomixis. As a di- rect consequence, ammonia accumulated in the monimolimnion and became depleted in the euphotic zone resulting in decreased algal photosynthetic activity (Jellison and Melack 1993a). During the breakdown of meromixis, a large pulse of ammonia was injected into the euphotic zone and algal photosynthetic activity was markedly increased. In many respects, nitrogen dynamics over the entire period (1982-1990) resembled the classic seasonal progression (stratification, epilimnetic nutrient depletion, hypolimnetic accumulation, and autumn mixing and algal bloom) observed over the course of a year in many temperate lakes. Vertical fluxes of nitrogen were uncoupled from seasonal regimes of temperature and light; consequently interannual differences in photosynthetic activity could be unambiguously attributable to changes in the nutrient regime.

Because external inputs of nutrients to Mono Lake are low, the high rates of algal photosynthetic activity observed at the lake must de- pend on internal recycling. Although SRP was in excess throughout the water column during the entire period, inorganic N concentrations varied over two orders of magnitude and were often low. Nitrate and nitrite concentrations were always low. DON concentration was high, but several types of evidence suggest its availability to phytoplankton is low. A positive response in Chl a was observed when batch algal cultures were enriched with ammonia (Jellison unpubl.). If high concentrations of DON were readily available for algal growth, a positive response would not be expected because DON was high relative to algal biomass in both treat- ments and controls. Also, the correlation of




annual photosynthetic activity with year-to- year changes in ammonia availability (Jellison and Melack 1993a) indicates nutrient limitation despite high DON concentrations throughout the year. Given these facts, we have focused on fluxes of PN and ammonia.

The onset of meromixis and reduction in nitrogen availability resulted in a decrease in photosynthetic activity from 540 g C m-2 yr- in 1983 to 340 in 1984 (Jellison and Melack 1988). If we use daily estimates of integral photosynthetic activity (Jellison and Melack 1993a), the mean annual photosynthetic production over 4 yr of meromixis (15 November 1983-14 November 1987), in which ammonia accumulated beneath the chemocline, was 358 g C m-2 yr-'. The mean annual photosynthetic production was twice as large (749 g C m-2 yr-1) during the 3 yr of increased vertical fluxes of ammonia beginning with the breakdown of meromixis, clearly indicating the importance of internal fluxes of N to the maintenance of high rates of algal photosynthesis in the lake.

Algal sinking rates are expected to be low in the lake because the dominant phytoplankton are small (1-5-,um diam) and the density (1.07 g cm-3, Jellison and Melack 1993b) and relative viscosity (1.21, Mason 1967) of the water are high. Algal sinking rates for mixed assem- blages of phytoplankton determined in the laboratory were very low (4-13 mm d-l, Melack unpubl.) and similar to bacterial sinking rates (Pedros-Alio et al. 1989). The model of particulate deposition validated against sediment trap data produced an algal sinking rate estimate of 0.12 m d-l which, although 10-fold higher than those determined in the laboratory, is still at the low end of the range reported for other algae (Wetzel 1983). The difference between laboratory and model estimates may be due to the different species composition and physiological health of algae as the model estimate is determined primarily by settling in the winter whereas laboratory estimates were performed with summer populations.

The concept of new and recycled production (Dugdale and Goering 1967) can profitably be used in lake ecosystems (Caraco et al. 1992). Although nitrogen uptake by algae was not directly measured, an estimate can be made from measurements of carbon uptake, assuming balanced growth. Although the C: N uptake ratio varies diurnally, long-term net uptake of

rapidly growing cells should approximate the Redfield ratio. Fisher et al. (1982) found uptake rates averaged over the entire day to be close to the molar Redfield C: N ratio of 6 in natural algal populations in Chesapeake Bay. Applying this ratio to estimates of lakewide photosynthetic activity for the period in which sediment traps were deployed results in PN production of 15.6 mol N m-2. Sedimentation of PN over the same period was 1.0 mol N m-2 indicating that only 6.4% of PN production was sedimented through the chemocline. The estimated N deposition rate for the entire period was 1.7 mmol N m-2 d-l while photosynthetic carbon uptake was 135 mmol C m-2 d-l. If we assume a molar C: N uptake ratio of 6, then 8% of production sank through the water column. The fraction of production lost through deposition in the lake is low and more similar to a mean value for the ocean (10%, Eppley and Peterson 1979) than temperate lakes (14-60%, Takamura and Yasuno 1988). Although a large proportion of primary pro-

duction is supported by rapid recycling of nutrients in the mixed layer, external inputs and upward turbulent fluxes of nutrients must balance losses due to particulate deposition if production is to be maintained. Despite low algal sinking rates in the lake, downward particulate fluxes in winter, during periods of high algal biomass but no Artemia, were about equal to those observed under opposite conditions in summer. During summer, particulate deposition was dominated by rapidly sinking fecal pellets. The predominance of the fecal pellet contributions to particulate deposition has been noted in other heavily grazed systems. In an arm of Puget Sound, Lorenzen and Welsch- meyer (1983) found the particulate flux of pheopigments in fecal pellets to be 20 times that of sedimenting Chl a. During the 6-yr period of meromixis in this study, the downward particulate flux of nitrogen represented a significant nutrient loss from the euphotic zone.

Although information necessary to con- struct a complete nitrogen budget is not available for Mono Lake, a balance based on vertical fluxes is informative. Fluxes are compared over three different periods: the portion of meromixis in which ammonia increased beneath the chemocline (23 March 1983-14 No- vember 1987), the erosion and breakdown of




meromixis (15 November 1987-24 November 1988), and postmeromixis (25 November 1988-31 December 1990). Because the vol- umes and areas associated with the mixolimnion and monimolimnion are continually changing over these periods, we use area- weighted fluxes and volume-weighted changes in ambient concentrations.

During meromixis, nitrogen (ammonia + algal N) gradually increased beneath the chemocline at a rate of 117 x 103 mol N d-1 (Fig. 12A). The estimated upper bound on the upward flux was 96 x 103 mol N d-l while the downward particulate flux was much higher (160 x 103 mol N d-l). The balance of these fluxes and the observed change in total N (ammonia + algal N + Artemia N) below the chemocline implies a net release from beneath the chemocline.

Over time periods of > 1 yr, more than half of particulate deposition is likely to be mineralized. Westrich and Bemer (1984) fitted a model consisting of two first-order decay pro- cesses to observations of oxic decomposition and determined sedimenting organic material to be composed of three fractions: highly reactive (- 50%), reactive (- 16%), and non- reactive (- 34%). Their two first-order decay constants were 24 and 1.4 yr-1 at 20-22?C. They also determined first-order decay constants for bacterial sulfate reduction of fresh freeze-dried phytoplankton to be 8.8 and 0.84 yr-1, respectively. Even after correcting for lower temperatures (assuming a Qlo of 2), using these decay constants results in 50-60% remineralization over a year. Sulfate-reduction rates (R. Oremland pers. comm.) in the anoxic water column (12 mmol m-2 d-1) and profundal sediments (13 mmol m-2 d-l) can easily account for mineralization of all organic deposition (2.4 mmol N m-2 d-1), assuming a C: N ratio of 5-9 for material collected in the traps. Therefore, for purposes of estimating sediment fluxes, remineralization was assumed to be 60 and 50% of particulate deposition onto sediments above and below the chemocline. If these rates of mineralization are assumed, the estimated ammonia flux out of the sediments beneath the chemocline was 133x 103 molNd-1.

In a similar fashion, mixolimnetic fluxes during meromixis include deposition onto sed-

iments above the chemocline (145 x 103 mol N d-l), deposition through the chemocline (160 X 103 mol N d-l), upward fluxes from the monimolimnion (96 x 103 mol N d-l), and volatilization (179 x 103 mol N d-l). Assum- ing the mineralization rates of deposited material as described above, an ammonia flux of 330 x 103 mol N d-l out of the mixolimnetic sediments is required to balance the fluxes and the observed change of 29 x 103 mol N d-l in mixolimnetic N (ammonia + algal N + Artemia N).

The magnitude of various fluxes was markedly different during the breakdown of meromixis (Fig. 12B). Total depositional fluxes were higher and the upward flux of N which had accumulated in the monimolimnion was first increased as meromixis weakened and then uniformly mixed throughout the water column when meromixis ended in November 1988. Over the period, the overall upward flux of ammonia was almost 5 times the deposition. Much of this pulse of N into the mixolimnion was lost to the atmosphere, as estimated volatilization (627 x 103 mol N d-l) was nearly double the mixolimnetic increase (335 x 103 mol N d-l). Estimates of sediment release rates based on the N balance were 404 and 62 x 103 mol N d-l for the mixolimnion and monimolimnion.

During the 2 yr after the breakdown of meromixis (25 November 1988 to 31 December 1990), estimated volatilization increased (1,627 X 103 mol N d-l) owing to high ambient ammonia concentrations, and total lakewide N slowly declined (156 x 103 mol N d-l) (Fig. 12C). Because the decrease in lake nitrogen is much less than the estimated volatilization over the period, high sediment fluxes (1,673 x 103 mol N d-l) are implied.

The overall mass transfer coefficients for ammonia volatilization were not directly measured but based on a model validated against measurements in rice paddies (Jayaweera and Mikkelsen 1990; Jayaweera et al. 1990) mod- ified for wind and temperature conditions at Mono Lake. These estimates, however, are reasonable and similar to the gas exchange constant derived empirically at Mono Lake for an inert gaseous tracer. Wanninkhof et al. (1987) measured a gas exchange constant for SF6 of 2.7 cm h-' over a 2-month period at




A. 23 March 1983- 14 November 1987 A

dA59 160 179

b

t\' * L ~~~~~Chemocline

AS ctT-- 330)

(133)

B. 15 November 1987 - 24 November 1988

293 a +335 627

d 167 e

b Chemocline (

4,,2 -69

f (62)

C. 25 November 1988 - 31 December 1990

1627 506 -156

b

(1,673) Fig. 12. Mean lakewide vertical fluxes of nitrogen (103 mol d-') during three periods: A-meromixis (1983-1987);

B-the breakdown of meromixis (1988); C-postmeromixis (1989-1990). Numbers in boxes indicate changes in total N (ammonia + algal N + Artemia N). Arrows are ammonia volatilization (a), particulate deposition through the thermocline (b), upward ammonia flux at the chemocline (c), particulate deposition onto sediments above the chemocline (d), and ammonia fluxes out of sediments above (e) and below (f) the chemocline. These last two fluxes are derived based on the balance of the others, assuming that 60% of deposition onto mixolimnetic sediments and 50% of deposition onto monimolimnetic sediments are remineralized in each time period.

the lake. Our calculated monthly transfer coefficients ranged from 1.3 to 5.1 cm h-I (mean, 3.14).

The above analysis indicates ammonia volatilization to be a significant loss process for

N. Calculated ammonia volatilization was particularly high after the breakdown of meromixis when surface ammonia concentrations were elevated over those observed during previous monomictic conditions in 1982. During




the 2 yr immediately after breakdown of meromixis, volatilization losses were three times greater than nitrogen deposition due to algal and fecal pellet sinking. Ammonia concentrations in 1991 indicate that the ammonia pulse at the end of meromixis had disappeared; estimated volatilization in 1991 was 172 x 103 mol N d-l, similar to the estimated volatilization in 1982 (152 x 103 mol N d-l).

Other nitrogen pathways that are often important terms in the N budget of lakes include denitrification, external inputs of dissolved inorganic nitrogen, and nitrogen fixation. De- nitrification is often a significant loss of N in temperate lakes (Seitzinger 1988). In Mono Lake denitrification is probably low because nitrate and nitrite concentrations are always low and water-column concentrations of ni- trous oxide are below the detection limit (<0.05 AM; R. Oremland pers. comm.). In the Great Salt Lake, no denitrification was detected in experimental microcosms (Post 1977).

Estimates of external inputs of nitrogen to Mono Lake are low relative to the vertical fluxes examined here. Atmospheric deposition, based on mean annual precipitation onto the lake surface and nitrate and ammonium concentrations in precipitation measured at near- by Spuller Lake, is low (- 5 Amol N m-2 d-l). To estimate nitrogen inputs from stream and subsurface inflows to the lake, we assumed nitrogen concentrations in subsurface inflows were similar to those in inflowing streams. Mean ammonium and nitrate concentrations in inflowing streams were 0.46 (n = 15) and 0.54 AM (n = 20), respectively, which, given total inflows of -0.5 x 106 m3 d- and a surface area of 160 kM2, yields an input of -3 gmol N m-2 d-1.

Planktonic heterocystous cyanobacteria are absent in Mono Lake (Melack unpubl.). In addition to high salinity, low availability of Mo may limit planktonic heterocystous cyanobacteria (Howarth and Cole 1985). Marino et al. (1990) found the ratio of sulfate to Mo to be a strong predictor of the abundance of N-fixing cyanobacteria in 13 saline lakes in Alberta. The ratio in Mono Lake, ̂ .1 x 107 (Winkler 1977), is higher than in any of the 13 lakes in their sample and would predict no N-fixing cyanobacteria.

Although nitrogen fixation is low or absent in the pelagic region, it does occur in the sur-

ficial sediments and floating ball-shaped clus- ters consisting of a matrix of filamentous chlorophytes, nonheterocystous cyanobacteria, and anaerobic bacteria (Oremland 1990) found in the littoral region. Oremland (1990) found benthic N-fixation rates of -6 ,umol m-2 h- and estimated lakewide fixation to be - 0. 1 x 106 mol yr-I or 0.6 mmol m-2 yr-1. Subsequent measurements of nitrogen fixation in benthic algal mats have yielded rates - 10-fold higher (65 Amol m-2 h-', D. Herbst pers. comm.). These algal mats are widely distributed and occur down to at least 10-m depth (D. Herbst pers. comm.). The highest hourly rate multi- plied by 24 h and the total area of sediments above 10-m depth (57 km2) yields a lakewide (160 kM2) input of 0.5 mmol m-2 d-l. Because algal mats are patchily distributed, a major portion of the benthic fixation is light-dependent (Oremland 1990), and rates measured by Herbst were measured at temperatures near the annual maximum in well-developed mats; a realistic mean estimate is much less. Realized rates are likely to be not more than a tenth of this estimate or 0.05 mmol m-2 d-l. Therefore the external input of nitrogen due to inflowing streams, atmospheric deposition, and nitrogen fixation is likely <0.1 mmol N m-2 d-l. Al- though these external sources are potentially significant for the long-term nitrogen budget of the lake, they are small relative to the vertical water column and benthic fluxes estimated in this study (Table 2).

The high nitrogen losses due to volatilization and small external sources of N imply substantial inputs from the sediments. The estimated fluxes of ammonia out of the sediments based on balancing the vertical fluxes and observed changes in total N (ammonia + algal N + Artemia N) varied during the course of meromixis (Table 2). The areal estimate of benthic ammonia flux in the monimolimnion during the period of ammonia accumulation beneath the chemocline was 1.5 mmol m-2 d-1. During breakdown of meromixis the estimate was slightly lower (1.0 mmol m-2 d-l). These estimates are lower than those based on sediment profiles of ammonia (3.1-10.4 mmol m-2 d-1).

The estimated rates of ammonia release from sediments above the chemocline were higher: 3.9 (1983-1987) and 3.6 mmol m-2 d-l (1988). These are closer to rates predicted by sediment




Table 2. Ammonia fluxes on an areal basis (mmol m-2 d-'). All rates are derived from total lakewide changes indicated in Fig. 12 divided by the appropriate area as indicated by the row in which each estimate is listed.

Deposition Source Upward onto needed to

Ammonia flux Deposition mixolim- Change balance Area volatil- through through netic in other (kM2) ization chemocline chemocline sediments total N terms*

Meromictic ammonia buildup (23 Mar 83-14 Nov 87) Lake surface 175.1 1.0 - - - Mixolimnetic sediments 85.6 - - - 1.7 +0.2 3.9 Monimolimnetic sediments 89.5 - 1.1 1.8 - +1.3 1.5

Breakdown of meromixis (15 Nov 87-24 Nov 88) Lake surface 171.0 3.7 - - - Mixolimnetic sediments 110.7 - - - 2.6 +2.0 3.6 Monimolimnetic sediments 60.3 - 14.0 2.8 -11.5 1.0

Postmeromixis (25 Nov 88-31 Dec 90) Lake surface 164.9 9.9 - - 3.1 0.9 10.1

* Estimates assume 60% of deposition onto mixolimnetic sediments and 50% of deposition onto monimolimnetic sediments is mineralized within the time pernod.

pore-water profiles. Although estimated sediment release rates were derived from cores taken in sediments overlain by anoxic water, a core taken below aerobic water (28 August 1988, 15 m) contained similar pore-water concentrations of ammonia (0-2 cm, 1,025 ,uM). Estimates of ammonia flux from the mixolimnetic sediments include uncertainties in volatilization and external inputs (e.g. nitrogen fixation) and include both oxic and anoxic sediments. Ammonia release from oxic sediments may be rapidly incorporated into benthic algae mats existing in littoral regions of the lake. The actual release rates of ammonia from the mixolimnetic sediments may span a wide range. During the postmeromictic period, the lakewide estimate of ammonia flux out of sediments was much higher, 10. 1 mmol m-2 d-l.

Our analysis indicates high ammonia release rates (3.6-10.1 mmol m-2 d-l) throughout the period except for those beneath the chemocline during meromixis. Although it is unknown why rates beneath the chemocline were lower, it could be explained if a layer near the bottom formed during meromixis and included elevated ammonia concentrations. The breakdown of the layer during holomixis could explain the higher ammonia release rates estimated after the breakdown of meromixis. Elevated ammonia concentrations were noted on several occasions when local topography resulted in samples being collected very near the sediment-water interface. Because the deepest sampling depths were usually 1-2 m

above the bottom, we were unable to assess the magnitude of near-bottom ammonia gradients.

Over time scales longer than a year, sediments usually act as a net sink for nutrients as partially mineralized organic matter is buried. The fraction of primary production which is buried represents a long-term average of deposition, remineralization, burial, and flux out of the sediments. Nitrogen burial rates can be estimated from volumetric sediment content and deposition rates. 210Pb dating on two freeze cores (30-40 determinations per core) taken in 1991 at a centrally located deep station (36 m) yielded sediment accumulation rates of 7 and 8 mm yr-1 (B. Anderson pers. comm.). The mean density of the in situ sediment of the two cores over the top 33 cm was 0.135 and 0.109 g cm-3 of salt-free dry sediment. Using these accumulation rates and the mean organic N content in the top 2 cm of sediments collected from >20 m depth (0.6% by wt) yields a nitrogen burial rate of 1.17 mmol N m-2 d-l or 49% of the average downward particulate flux.

Given the estimated low external inputs of nitrogen (<0.1 mmol N m-2 d-1), a large net flux of ammonia out of the sediments is required to balance the mean loss due to volatilization (3.5 mmol N m-2 d-1). Thus, a long- term disequilibrium of nitrogen fluxes at the sediment-water interface is indicated.

For the sediments to act as a net source of nitrogen over an extended period, previous production and burial rates must have been




higher. During the past 50 yr, inflowing streams have been diverted out of the Mono basin, resulting in a decline of 14 m in surface elevation and a doubling in salinity (Patten et al. 1987). Nitrogen fixation in benthic algal mats is nearly double at pre- 1941 salinities than at current salinities (D. Herbst pers. comm.). It is also possible that planktonic N fixation was higher. In much less saline Pyramid Lake, north of Mono Lake, nitrogen fixation provided 81% of the total N input and 99.5% of the algal N requirement (Home and Galat 1985). Given higher rates of nitrogen fixation at lower salinities and the N-limited status of Mono Lake, previous rates of algal production and nitrogen burial were likely higher. This could explain part of the current imbalance of nitrogen fluxes at the sediment-water interface.

At present it appears that nutrient cycling in Mono Lake is more similar to marine envi- ronments than to temperate lakes. Similarities with oceanic systems include N-limited status, high rates of intemal recycling, low external inputs and sediment burial rates, and low rates of nitrogen fixation relative to internal fluxes.

References ABREu-GROBOIs, F. A., R. BRISE1RO-DuEfWA, M. A.

HERRERA, AND M. L. MALAG6N. 1991. A model for growth ofArtemiafranciscana cultures based on food ration-dependent growth efficiencies. Hydrobiologia 212: 27-37.

BLOESCH, J., AND N. M. BURNS. 1980. A critical review of sedimentation trap technique. Schweiz. Z. Hydrol. 42: 15-55.

BOWEN, S. T., AND OTHERS. 1985. Ecological isolation in Artemia: Population differences in tolerance of anion concentrations. J. Crustacean Biol. 5: 106-129.

CARACO, N. F., J. J. COLE, AND G. E. LIKENS. 1992. New and recycled primary production in an oligotrophic lake: Insights for summer phosphorus dynamics. Limnol. Oceanogr. 37: 590-602.

DUGDALE, R. C., AND J. J. GOERING. 1967. Uptake of new and regenerated forms of nitrogen in primary productivity. Limnol. Oceanogr. 12: 196-206.

EPPLEY, R. W., AND B. J. PETERSON. 1979. Particulate organic matter flux and planktonic new production in the deep ocean. Nature 282: 677-680.

FALKowsKI, P. G., AND J. LARocHE. 1991. Acclimation to spectral irradiance in algae. J. Phycol. 27: 8-14.

FIsHER, T. R., P. R. CARLSON, AND R. T. BARBER. 1982. Carbon and nitrogen primary productivity in three North Carolina estuaries. Estuarine Coastal Shelf Sci. 15: 621-644.

GARDNER, W. D. 1980. Field assessment of sediment traps. J. Mar. Res. 38: 41-52.

GOLTERMAN, H. L. [ED.]. 1969. Methods for chemical analysis of fresh waters. IBP Handbook 8. Blackwell.

HAmmER, U. T. 1986. Saline lake ecosystems of the world. Junk.

HERNANDORENA, A. 1976. Effects of temperature on the nutritional requirements of Artemia salina (L.). Biol. Bull. 151: 314-321.

HORNE, A. J., AND D. L. GALAT. 1985. Nitrogen fixation in an oligotrophic, saline desert lake: Pyramid Lake, Nevada. Limnol. Oceanogr. 30: 1229-1239.

HIMMELBLAU, D. M. 1964. Diffision of dissolved gases in liquids. Chem. Rev. 64: 527-550.

HOWARTH, R. W., AND J. J. COLE. 1985. Molybdenum availability, nitrogen limitation, and phytoplankton growth in natural waters. Science 229: 653-655.

JAYAwEERA, G. R., AND D. S. MIKKELSEN. 1990. Am- monia volatilization from flooded soil systems: A computer model. 1. Theoretical aspects. Soil Sci. Soc. Am. J. 54: 1447-1455.

- , , AND K. T. PAW U. 1990. Ammonia volatilization from flooded soil systems: A computer model. 3. Validation of the model. Soil Sci. Soc. Am. J. 54: 1462-1468.

JELLISON, R., AND J. M. MELACK. 1988. Photosynthetic activity of phytoplankton and its relation to environ- mental factors in hypersaline Mono Lake, California. Hydrobiologia 158: 69-88.

- , AND . 1993a. Algal photosynthetic activity and its response to meromixis in hypersaline Mono Lake, California. Limnol. Oceanogr. 38: 818-837.

-- , AND . 1993b. Meromixis in hypersaline Mono Lake, California. 1. Vertical mixing and density stratification during the onset, persistence, and breakdown of meromixis. Limnol. Oceanogr. 38: 1008- 1019.

JOHNSON, D. L. 1971. Simultaneous determination of arsenate and phosphate in natural waters. Environ. Sci. Technol. 5: 411-414.

LENZ, P. H. 1984. Life-history analysis of an Artemia population in a changing environment. J. Plankton Res. 6: 967-983.

Li, Y.-H., AND S. GREGORY. 1974. Diffusion of ions in sea water and deep-sea sediments. Geochim. Cos- mochim. Acta 38: 703-714.

LORENZEN, C. J., AND N. A. WELSCHMEYER. 1983. The in situ sinking rates of herbivore fecal pellets. J. Plank- ton Res. 5: 929-933.

LOvEJoY, C., AND G. DANA. 1977. Primary producer level, p. 42-57. In D. W. Winkler [ed.], An ecological study of Mono Lake, California. Inst. Ecol. 12. Univ. Calif., Davis.

MARINO, R., R. W. HOWARTH, J. SHAMESS, AND E. PREPAS. 1990. Molybdenum and sulfate as controls on the abundance of nitrogen-fixing cyanobacteria in saline lakes in Alberta. Limnol. Oceanogr. 35: 245-259.

MASON, D. T. 1967. Limnology of Mono Lake, Califor- nia. Univ. Calif. Publ. Zool. 83: 1-110.

MELACK, J. M. 1983. Large, deep salt lakes: A compar- ative limnological analysis. Hydrobiologia 105: 223- 230.

MILLER, L. G., R. JELLISON, R. S. OREMLAND, AND C. W. CULBERTSON. 1993. Meromixis in hypersaline Mono Lake, California. 3. Biogeochemical response to stratification and overturn. Limnol. Oceanogr. 38: 1040- 1051.

OPPENHEIMER, C. H., AND G. S. MoREIRA. 1980. Carbon, nitrogen and phosphorus content in the develop-




mental stages of the brine shrimp Artemia, p. 609- 612. In G. Personne et al. [eds.], The brine shrimp Artemia. V. 2. Universa.

OREMLAND, R. S. 1990. Nitrogen fixation dynamics of two diazotrophic communities in Mono Lake, Cali- fornia. Appl. Environ. Microbiol. 56: 614-622 (er- ratum 56: 2590).

PATTEN, D. T., AND OTHERS. 1987. The Mono Basin ecosystem: Effects of changing lake level. Natl. Acad. Sci.

PEDR6s-ALI6, C., J. MAS, J. M. GASOL, AND R. GUERRERO. 1989. Sinking speeds of free-living phototrophic bacteria determined with covered and uncovered traps. J. Plankton Res. 11: 887-905.

POST, F. J. 1977. The microbial ecology of the Great Salt Lake. Microb. Ecol. 3: 143-165.

RANDAL, F. M., AND C. F. FAILEY. 1927. The activity coefficient of gases in aqueous salt solutions. Chem. Rev. 4: 271-285.

REEVE, M. R. 1963. Growth efficiency in Artemia under laboratory conditions. Biol. Bull. 125: 133-145.

SEITZINGER, S. P. 1988. Denitrification in freshwater and coastal marine ecosystems: Ecological and geochem- ical significance. Limnol. Oceanogr 33: 702-724.

SoL6RzANo, L. 1969. Determination of ammonia in natural waters by the phenolhypochlorite method. Lim- nol. Oceanogr. 14: 799-801.

STRICKLAND, J. D. H., AND T. R. PARSONS. 1972. A practical handbook of seawater analysis, 2nd ed. Bull. Fish. Res. Bd. Can. 167.

STUMM, W., AND J. J. MORGAN. 1970. Aquatic chem- istry. Wiley.

TAKAMURA, N., AND M. YASUNO. 1988. Sedimentation of phytoplankton populations dominated by Micro- cystis in a shallow lake. J. Plankton Res. 10: 283-299.

VALDERRAMA, J. C. 1981. The simultaneous analysis of total nitrogen and total phosphorus in natural waters. Mar. Chem. 10: 109-122.

WANNINKHOF, R. H., J. R. LEDWELL, W. S. BROECKER, AND M. HAMILTON. 1987. Gas exchange on Mono Lake and Crowley Lake, California. J. Geophys. Res. 92: 14,567-14,580.

WESTRICH, J. T., AND R. A. BERNER. 1984. The role of sedimentary organic matter in bacterial sulfate reduction: The G model tested. Limnol. Oceanogr. 29: 236-249.

WETZEL, R. G. 1983. Limnology, 2nd ed. Saunders. WINKLER, D. [ED.]. 1977. An ecological study of Mono

Lake, California. Inst. Ecol. 12. Univ. Calif., Davis.

Submitted: 27 March 1991 Accepted: 17 November 1992

Revised: 15 April 1993



Documents

Meromixis in Hypersaline Mono Lake, California. 2. Nitrogen Fluxes