Historical Phosphorus Kinetics and Ambient Orthophosphate Concentrations in the St. Lawrence Great Lakes Erie, Huron, Michigan, St. Clair, and Superior by a Modified Inverse Isotope Dilution Method

Abstract

Historical measurements of phosphate turnover and uptake confirm that bacterioplankton dominate phosphate dynamics at ambient steady state conditions in all but the most eutrophic samples, but phytoplankton exhibits increased control at phosphate additions as low as +10 nM. The results are consistent with the theory that uptake mechanisms of bacterioplankton become saturated as soon as phosphate concentrations are elevated above ambient levels. Uptake dynamics were consistent with multiphasic kinetics by bacterioplankton versus phytoplankton. Temperature dependence of phosphate turnover was demonstrated for Lake Superior but not for other Great Lakes in which temperatures were largely homogeneous. Ambient concentrations of orthophosphate were estimated by an inverse isotope dilution method that indicated concentrations ranged from roughly 1 to 7 nM across all the lakes surveyed.

1. Introduction

A recent report of widespread increases in soluble phosphate concentrations in streams across the Great Lakes region of North America from 2003 to 2019 implicates changing climate and land use practices for increasing frequencies and intensities of harmful algal blooms [1]. The authors suggest that even if total phosphorus concentrations remain constant, changes in phosphorus chemical speciation may exert strong ecological effects. Other reports suggest that the lower food webs of Lakes Michigan and Huron are experiencing oligotrophication and are exhibiting increased similarity to Lake Superior [2,3,4,5,6,7,8,9,10]. Whether or not these different observations and speculations can be reconciled remains to be seen, but they suggest the need for increased attention to historical datasets. I recently archived and published a dataset consisting of physical, chemical, and biological data collected from the North American Great Lakes from 1986 to 1997 [11,12]. Additional historical and more recent data about lower trophic level food web structure in those lakes using N-15 stable isotope analyses have likewise been published [13]. Here, I report experiments designed to measure kinetics and ambient concentrations of orthophosphate in several Great Lakes using carrier-free P-33. P-33 was the radioisotope of choice for long research cruises owing to its longer half-life and lower beta particle emission energy than P-32.

Methodological difficulties with measuring ambient phosphate concentrations in lakes have been known for decades [14,15,16]. Spectrophotometry is the principal method used to monitor phosphorus in lakes, but the chemical speciation of the multiple forms of P is typically operationally defined and is resultingly inexact [17]. As early as 1966, Frank Rigler proposed a radiobioassay method to estimate orthophosphate [18]. There has been decades-long controversy about the relative contributions of bacterioplankton, phytoplankton, and zooplankton to total phosphorus flux in lakes [19]. These relative contributions may differ between ambient steady-state conditions and experimentally perturbed concentrations.

Some authors have tried to reduce the detection limits for orthophosphate concentrations by spectrophotometric means by concentrating the phosphate by magnesium-induced coprecipitation [20,21,22], although others have reported unsatisfactorily high blank values with that method [23].

For the conceptual framework of this study, I adopted a conventional two-compartment model of P uptake and excretion [19]. Particulate cellular P is represented by a hypothetical small metabolic compartment that both acquires and releases phosphate from and to solution, as well as a larger structural compartment that is less labile. For concentrations at or near ambient, I assumed that phosphorus dynamics were at steady state and that first-order kinetics applied. An opportunity to test these assumptions against data arose during my research cruise from July to August 1997 on lakes Erie, St. Clair, Huron, Superior, and Michigan. The cruise had multiple investigative aims, but these results have not previously been published. They may now provide a baseline and reference for future investigations of changing phosphorus dynamics in the Great Lakes.

2. Materials and Methods

Preliminary experiments to test the feasibility of shipboard radiophosphorus-33 (P-33) kinetics measurements during long research cruises were performed at 23 stations in Lakes Erie, Huron, and Michigan from 24 August to 2 September 1993 (

Table 1. Station IDs, decimal Latitude (N), decimal Longitude (W), and station maximum depth z (m), 1993.

Station ID	Latitude	Longitude	z
E1	41.605	82.657	10
E2	41.665	82.268	15
E3	41.722	81.883	20
H1	43.500	82.500	20
H3	43.500	82.233	50
H4	44.167	82.500	64
H6	44.500	82.555	50
H10	45.750	84.252	20
H11	45.750	84.063	40
H12	45.750	83.550	106
M1	43.000	86.267	20
M2	43.000	86.667	100
M3	43.077	87.838	20
M4	43.187	87.668	100
M5	44.142	87.495	20
M7	44.018	87.395	83
M9	45.000	86.200	20
M11	45.000	86.443	142
M16	44.460	87.430	40
M17	44.500	87.300	100
M21	45.500	86.643	20
M22	45.500	86.620	40
M23	45.500	86.482	100

).

In 1997, 32 sampling stations were visited in western Lake Erie, St. Clair, Huron including Saginaw Bay, eastern Lake Superior, and Lake Michigan including outer Green Bay (Figure 1,

Table 2. Station IDs, decimal Latitude (N), decimal Longitude (W), and station maximum depth z (m), 1997.

Station ID	Latitude	Longitude	z
E3	41.722	81.883	20
E2	41.665	82.268	15
E1	41.605	82.657	10
SC	42.500	82.703	4
H1	43.500	82.500	20
H4	44.167	82.500	64
SB1	44.167	83.000	40
SB2	44.083	83.500	11
SB3	43.834	83.667	8
H5	44.500	83.225	21
H6	44.500	82.555	58
H8	45.000	82.750	110
H12	45.750	83.550	106
S1	46.667	85.003	20
S2	46.667	84.867	98
S3	46.803	85.000	25
S4	46.938	85.003	109
S5	46.720	85.500	20
S6	47.001	85.503	155
S8	47.000	86.000	141
S10	47.208	85.930	95
S11	46.915	87.843	20
S12	47.135	87.200	100
S13	47.190	87.742	100
M10	45.322	85.498	101
M9	45.155	85.465	40
M8	45.053	85.500	20
M7	45.500	86.482	100
M5	45.525	87.008	21
M3	44.621	87.186	102
M2	43.000	86.667	101
M1	43.000	86.267	20

).

Isotope dilution experiments in 1993 were conducted with 100 mL samples of lake water from 5, 10, or 15 m depth at ambient water temperature. All treatments involved additions of 100 μL carrier-free P-33 obtained from New England Nuclear, followed by time-series filtration of 10-mL subsamples through 0.2 μm Nucleopore filters. At seven of the sites, plankton samples were diluted by 50% using lake water filtrate from 0.2 μm filtration. In 1997, all samples were collected from 5 m depth except for Lake St. Clair, where the collection depth was 2 m. Treatments consisted of no addition, +10 nM PO₄, and +20 nM PO₄ together with either 50, 80, or 100 μL of carrier-free P-33 obtained from New England Nuclear. The solution for standard additions was prepared so that 1 mL contained 10 nmol P-31 as orthophosphate. At intervals from <1 to 20 min 5-mL, subsamples were removed and filtered through either 0.2 μm or 1.0 μm Nucleopore filters. Then, the filters were placed in glass scintillation vials and stored frozen pending return to my University of Michigan-Ann Arbor laboratory at cruise’s end for liquid scintillation counting. In addition to the filters, 0.5 mL of the 100 mL samples was placed in scintillation vials and likewise stored frozen until counted in parallel with the filters. Doing so provided internal standards for the decay of P-33 stock solution during the long cruise (16 July to 12 August 1997).

In some experiments, 0.1 mL formalin was added to 5-mL subsamples in parallel with uptake time series to determine how much of the sequestered P-33 was labile enough to leak from dead cells, representing a metabolic, non-structural pool. In other cases, 0.1 mL formalin was added to the time-series treatments after ca. 10 min to confirm whether further P-33 uptake ceased, demonstrating that the uptake was biological. In other experiments, time-series uptake measurements were interrupted by addition of ca. 500 nM PO₄ to determine whether cellular P-33 could be chased into solution, representing metabolism and excretion of the structural phosphorus pool.

Upon return to the laboratory, time-series subsamples were analyzed within 36 h by liquid scintillation counting. Total P-33 cpm on each filter were divided by 10 times the P-33 cpm contained in the 0.5 mL control samples for each experiment. Resulting fractions of total P-33 on each filter were plotted versus time, and slopes plus standard errors were calculated by linear regression. A first estimate of ambient [PO₄] was determined by inverse isotope dilution as follows:

P₀ = ambient lake water [PO₄]

(1)

P₁₀ = lake water amended with +10 nM [PO₄]

(2)

k₀ = first order rate constant for carrier-free P-33 uptake in ambient lake water

(3)

k₁₀ = first order rate constant for P-33 uptake in lake water amended with +10 nM [PO₄]

(4)

P₀ = [10]/(k₀/k₁₀ − 1)]

(5)

The coefficients k₀ and k₁₀ were obtained by empirical fit to the time-series data. Improved estimates of P₀ were obtained by assuming that uptake was first-order linear with additions of both 10 nM and 20 nM PO₄, using the Solver™ function of Microsoft Excel™ version 2502 to iterate to a P₀ value for which uptake rates (nM/min) were linear (r² = 1.000) for three points: no add = P₀, P₀ + 10 nM, and P₀ + 20 nM treatments. This effectively permitted calculation of uptake rates of nM/min instead of simply the fraction of P-33 assimilated per minute.

Additional concentration-amended uptake kinetics were conducted using 10-mL lake water samples at ambient lake temperatures. Amendments were +100 and +300 nM P-31 together with 0.1 mL carrier-free P-33 stock solution. After ca. 15 min incubation, the entire samples were filtered through either 0.2 μm or 1.0 μm Nucleopore filters, placed in scintillation vials, and stored frozen until analysis. P-33 kinetics data were analyzed in the context of my archived Great Lakes dataset [12], including water temperature and concentrations of total P (TP), dissolved P (DP), and chlorophyll a (Chl a).

3. Results

3.1. P-33 Turnover Rates Under Ambient (No Add) Conditions in 1993

Turnover rates of P-33 phosphate exhibited statistically significant differences among Lakes Erie, Huron, and Michigan (one-way AOV p << 0.0001). The differences trace almost entirely to Lake Michigan; Lakes Huron and Erie exhibited no statistically significant differences (p > 0.08). These differences could not be ascribed to differences in Chl a (Figure 2), total dissolved P (Figure 3), or temperature.

To ascertain whether the measured turnover rates were reliable and reproducible, rates calculated for undiluted samples were compared with rates calculated for samples that had been diluted by 50%. For the seven samples analyzed from the three lakes, the mean ratio of diluted sample (50%) turnover rate to undiluted samples was 0.462 (SE = 0.031, n = 7), which was not significantly different from the expected mean of 0.500.

3.2. P-33 Turnover Rates Under Ambient (No Add) Conditions in 1997

Turnover rates of carrier-free P-33 phosphate added to lake water at ambient temperature showed almost no relationship to water temperature except for the coldest conditions encountered in Lake Superior (Figure 4).

Likewise, there was no relationship between turnover rates and Chl a (Figure 5) nor with total dissolved P (r² = 0.014, p > 0.1) or with particulate P (r² = 0.033, p > 0.1). Turnover rates for Lake Michigan measured in 1997 were lower than those from 1993 and were more comparable to Lakes Huron and Erie.

The only compelling predictor of turnover rates under ambient conditions was the ratio of particulate P-33 retained on 1.0 μm filters to that retained by 0.2 μm filters (Figure 6).

The pattern suggests that the smallest particles, bacterioplankton, with their high surface-to-volume ratios, play a larger role in phosphate turnover kinetics under ambient, steady-state conditions than do larger phytoplankton particles.

3.3. Kill Experiments

Uptake of P-33 promptly ceased in time-series trials when formalin was added (Figure 7), demonstrating that the uptake was biological in nature.

In time-series trials using 100-mL samples in which one 5 mL subsample was filtered live (CTL) and a second 5-mL subsample simultaneously received 0.1 mL formalin (+Formalin) before being filtered, measurable fractions of P-33 retained by CTL organisms were lost into solution when plankton was killed (Figure 8).

Formalin-treated reduced retention of P-33 by particles differed among Lakes Huron, Superior, and Michigan (

Table 3. Mean and SE of the ratio of P-33 retained on 0.2 μm filters for formalin-killed vs. live subsamples during time-series uptake experiments as illustrated by Figure 8, together with ambient experimental temperatures (C) and Chl a (mg/m³).

Station	Treatment	Mean	SE	n	Temp	Chl a
H6	no add	0.783	0.052	5	16.87	0.31
H6	+10 nM P	0.788	0.082	5	16.87	0.31
H7	no add	0.783	0.037	5	18.16	0.55
H8	no add	0.743	0.027	5	16.87	0.50
H8	+10 nM P	0.802	0.053	5	16.87	0.50
H8	+20 nM P	0.842	0.014	5	16.87	0.50
H12	no add	0.739	0.036	5	18.49	0.78
S1	no add	0.650	0.011	5	16.19	0.29
S2	no add	0.628	0.037	5	15.82	0.28
S3	no add	0.551	0.036	5	14.24	0.29
M8	no add	0.830	0.040	5	19.92	0.56
M9	no add	0.914	0.060	5	19.61	0.51
M10	no add	0.799	0.023	5	18.87	0.60

). With 20 determinations for Lake Huron and 15 each for Lakes Superior and Michigan, one-way analysis of variance detected statistical significance at p < 0.0001. Post-hoc Tukey HSD and Bonferroni pairwise comparisons concurred in finding that Lake Superior differed from both Lake Huron and Lake Michigan (p < 0.01), and that Lakes Michigan and Huron likewise differed (p < 0.05). The differences could not be reliably ascribed to differences in Chl a concentrations (p > 0.06), but they could be ascribed to differences in lake temperature (p < 0.002). Given the working hypothesis that the losses traced to a labile metabolic pool that had not yet been assimilated into structural material by enzymatic processes, the apparent temperature dependence is unsurprising.

3.4. Cold Chase Experiments

In some experiments with unamended lake water, P-33 isotope addition time series were permitted to approach equilibrium, and then 500 nM P-31 phosphate was added to dilute dissolved pools of P-33. Time-series samples after P-31 addition were collected to see if any P-33 compounds in structural components might be metabolized and excreted back into solution (Figure 9). These figures are illustrative of results from the five lakes summarized in

Table 4. Time-series change in P-33 content of particulates retained on 0.2 μm filters after addition of 500 nM P-31. Bold face denotes statistical significance.

Station	Fraction of Total P-33/min	Prob
E3	n.s.	0.42
SC	n.s.	0.41
H1	n.s.	0.37
H2	n.s.	0.29
H8	n.s.	0.38
H12	n.s.	0.91
S2	n.s.	0.16
S5	−0.0001 (0.000022)	0.044
S6	n.s.	0.66
S11	−0.00012 (0.000018)	0.023
S12	n.s.	0.13
M5	n.s.	0.87

.

Time-series subsamples collected post-chase addition were analyzed by linear regression to detect decreases in particulate P-33, which could be indicative of the metabolism and excretion of structurally assimilated P-33. Of 12 such experiments conducted on five lakes, statistically significant decreases in particulate P-33 could be detected for only two sampling sites, both in Lake Superior (Table 4).

In both cases, the implied excretion rate of assimilated P-33 was miniscule, about 0.01 percent of assimilated P per minute.

3.5. Estimation of Ambient P₀ Concentrations

Experiments like the CTL series shown in Figure 8 provided rates of uptake of carrier-free P-33 added to samples unamended with P-31 as well as to samples that were amended by addition of either 10 nM or 20 nM P-31 phosphate. After obtaining initial approximations of P₀ by applying Equations (1)–(5), refined estimates were obtained by iterating putative values for P₀ such that the uptake rate constant k₀ would be first-order linear for P₀, P₀ + 10 nM and P₀ + 20 nM, as described in Section 2. P₀ was determined implicitly by defining the initial slopes of CTL treatments, as shown in Figure 8, as k₀, k₁₀, and k₂₀. Uptake rates in mass units were then defined as

nM P/min in no added P treatment = k₀ × P₀

(6)

nM P/min in +10 nM added P treatment = k₁₀ × (P₀ + 10)

(7)

nM P/min in +20 nM added P treatment = k₂₀ × (P₀ + 20)

(8)

These three equations were solved using the Excel™ Solver™ function to identify a value of P₀ for which the three points defined a straight line (R² = 1.000).

The P₀ concentrations thus deduced are displayed in

Table 5. P₀ calculated from CTL (no add), +10 nM P, and +20 nM P treatments and calculated uptake rates (nM/min).

Station	P0 nM	R2	Uptake
E3	4.21	1.000	0.0084
E2	2.63	1.000	0.0114
E1	5.00	1.000	0.0028
H4	4.22	1.000	0.0058
SB1	3.39	1.000	0.0085
SB2	2.63	1.000	0.0921
SB3	1.95	1.000	0.0026
H5	1.40	1.000	0.0158
H6	1.99	1.000	0.0066
H8	4.29	1.000	0.0044
S4	6.43	1.000	0.0456
S8	2.71	1.000	0.0025
M2	5.46	1.000	0.0062

. All are within the nM range, but the available data (temperature, Chl a, as well as DP, TP, and PP by conventional acid-molybdate spectrophotometry after acid-persulfate oxidation [12]) do not identify a plausible correlate with their values.

Inspection of the data revealed that particles retained on 1.0 µm filters assumed increased proportions of P-33 uptake with P-31 additions (

Table 6. Ratios of P-33 retained on 1.0 μm filters to P-33 retained on 0.2 μm filters for unamended and P-31 amended samples plus paired t-test probabilities comparing unamended ratios to each P-31 amended ratio.

Treatment	Mean	SE	n	Paired t-Test Probability
CTL (no add)	0.27	0.11	30
+10 nM P	0.32	0.12	14	0.0002
+20 nM P	0.37	0.10	12	0.0017
+100 nM P	0.37	0.09	10	0.005
+300 nM P	0.50	0.17	27	<0.0001

).

These data imply that phytoplankton becomes increasingly responsible for P uptake as concentrations are elevated above ambient levels when the initial steady state is perturbed. Given the results shown in Figure 6, P₀ values reported in Table 5 are probably just approximations, because steady state turnover rates decline with increasing ratios of 1.0 μm to 0.2 μm P-33 filter retention. The linear trendline displayed in Figure 6 accounts for only 41 percent of the data variation, and the SE of its slope is 21 percent of its mean value, so use of the trendline slope to correct P-33 turnover rates is unreliable, especially because there is also evidence of temperature dependence of turnover rates in Lake Superior (Figure 4).

The P₀ estimates reported in Table 5 do not display any pattern with respect to P-33 retention ratios in no add treatments (Figure 10), nor do they correlate with temperature (p = 0.8) or Chl a (p = 0.4). The two most eutrophic sites sampled, extreme western Lake Erie (E1) and lower Saginaw Bay (SB3), exhibited P₀ values among the highest and the lowest in the dataset.

4. Discussion

Ambient P₀ concentrations measured by radiotracer in this study (2.7 and 6.4 nM) compare favorably with the most sensitive chemical analytical determinations performed more recently for Lake Superior [24], most notably results obtained by magnesium-induced coprecipitation [21], but they are lower than values reported for outer Green Bay and cited by the U.S. EPA for lakes Michigan, Superior, Huron, and Erie [22] (Table 1). The P₀ value for offshore Lake Michigan (5.5 nM) is about one-half that of a more recent report for the offshore lake [25].

Most studies in mesotrophic and oligotrophic lakes, including the Great Lakes, conclude that bacteria sequester most available phosphate at ambient concentrations [26,27,28,29] owing to their higher surface-to-volume ratios compared to phytoplankton. Increased proportional uptake of phosphate by phytoplankton when ambient concentrations are elevated is a common observation, as demonstrated again in this study. Some authors have ascribed those observations to the fact that bacterial uptake of phosphate becomes saturated at low concentrations, possibly because their metabolism is limited by organic carbon [28]. Along a transect from inner Saginaw Bay to offshore Lake Huron in 2002, bacterioplankton productivity was higher relative to phytoplankton productivity in the oligotrophic offshore, although the highest rates of phosphate turnover were observed within Saginaw Bay [29]. Those results are contrary to the observations reported in this study (Figure 6), in that inner Saginaw Bay showed a lower rate of phosphate turnover than offshore Lake Huron. As in that previous report [29], however, phosphate uptake into phytoplankton (>1.0 μm particles) was highest in inner Saginaw Bay, where Chl a concentrations were elevated with respect to the rest of Lake Huron (Figure 5).

The fact that temperature dependence of P turnover rates was evident in Lake Superior and not in the other lakes may simply reflect the fact that the range of ambient temperatures encountered in Lake Superior was far greater than in any of the other lakes. Evidence that turnover is biological (Figure 7) and thus is enzymatic implies that temperature effects should be expected. Existence of those temperature effects is reinforced by the results in Table 3, which imply slower rates of assimilation of P from a labile pool into a more stable structural pool at the cooler temperatures of Lake Superior. The reason that experiments were conducted at ambient temperature rather than at a fixed temperature has precedence [29], and it avoids the potential problem of temperature shock and acclimation effects.

The range of P₀ concentrations encountered in this study, roughly 1 to 7 nM, is within the range reported recently for Lake Superior [21,24], although the measurements were obtained by different methodologies. It would be instructive, therefore, if future studies could compare radiochemical and standard chemical assays to learn if the methodologies produce comparable results. The increased uptake into phytoplankton particles with even 10 nM P enrichment above ambient (Table 6) admittedly injects an unknown amount of uncertainty into the method used to estimate P₀. The estimated values may thus be more analogous to what Rigler [18] called maximum likely phosphate concentrations.

If changing temperatures and land use are increasing the proportion of total phosphorus that enters the Great Lakes as inorganic phosphate [1], the results reported here reinforce concerns that phytoplankton may increasingly dominate phosphate metabolism at ambient concentrations, with the prospect of increasingly frequent and intense algal blooms in the future, as those authors suggest.

5. Conclusions

The historical data reported in this paper represent a baseline against which current and future phosphate dynamics by bacterioplankton and phytoplankton can be evaluated in the St. Lawrence Great Lakes. The results demonstrate that bacterioplankton generally dominated orthophosphate kinetics at ambient concentrations in the nanomolar range, but that phytoplankton exhibited increasing dominance with additions of as little as 10 nM P.