Lake toxin concentrations increase when phosphorus is reduced

doi:10.21203/rs.3.rs-4328909/v2

Download PDF

Biological Sciences - Article

Lake toxin concentrations increase when phosphorus is reduced

https://doi.org/10.21203/rs.3.rs-4328909/v2

This work is licensed under a CC BY 4.0 License

Version 2

posted

You are reading this latest preprint version

Cyanobacterial harmful algae blooms (e.g. those of Microcystis) and their toxins (e.g. microcystins - MCs) threaten the safe use of lakes for drinking and recreation and this hazard is anticipated to worsen due to a warming climate. Present management approaches aim at lowering biomass by reducing the input of the growth-limiting nutrient, typically phosphorus. It is assumed that phytoplankton stoichiometry (MC per biomass) remains constant and that cyanotoxins decrease proportionally. However, lower biomass due to phosphorus reduction means an increase in other resources, like nitrogen and light, which stimulate MC production (we show here a comprehensive summary of published laboratory results). High MC cell quotas benefit toxin-producing strains within the community, causing the toxigenic fraction to increase. While reducing phosphorus may increase MC concentrations in lakes, there is currently insufficient evidence from natural field conditions to support this effect. Here, we present two analyses leveraging the National Lakes Assessment (NLA) dataset. The first focuses on changes in MC concentration within individual lakes across time. In the second analysis, we applied Generalized Additive Models (GAMs) across all lakes and time points. Both analyses indicate that lower phosphorus levels reduce biomass but increase available inorganic nitrogen, available light the MC/biomass ratio, which, in turn, can lead to peaking MC concentrations. Hence, field data from hundreds of lakes confirm that a reduction in phosphorus increases MC concentrations. However, when nitrogen is reduced together with phosphorus, both biomass and MC concentrations decrease. As such, there is a surging need to focus on dual reduction management.

Earth and environmental sciences/Limnology

Biological sciences/Ecology/Freshwater ecology

Earth and environmental sciences/Biogeochemistry/Element cycles

Earth and environmental sciences/Natural hazards

cyanobacteria

Microcystis

microcystin

lake management

field data analysis

Cyanobacteria can produce toxins that make lakes unsafe for drinking and recreation. This is an urgent environmental problem, but there is no scientific consensus about the best management approach. Reducing inputs of the growth-limiting nutrient phosphorous can be successful in decreasing cyanobacterial biomass but the effect on toxin concentration is unclear. Evidence from laboratory data and models suggests toxin concentrations may actually increase. What has been missing is field data to support or refute this hypothesis. We analyzed observations from hundreds of lakes (US National Lakes Assessment) and show that when phosphorus only is reduced, microcystin (the toxin produced by the cyanobacteria Microcystis) concentration increases. A dual reduction of nitrogen and phosphor, however, results in a decrease of both, biomass and toxins.

Harmful algal blooms and their management

Cyanobacterial harmful algae blooms (CyanoHABs), most prominently those of Microcystis and the produced hepatotoxin microcystin (MC), threaten the safe use of freshwater resources for drinking and recreation as well as ecosystem health ^1,2. A prominent example is the interruption of the drinking water supply of Toledo in 2014 ³. The National Lakes Assessment (NLA) showed the presence of HABs in 45% and of MCs in 21% of all examined US lakes in 2017 ⁴. The problem is predicted to get worse in a future warmer climate ^5,6.

Effective management of CyanoHABs is a grand societal challenge. The most common strategy is to control cyanobacteria biomass by reducing growth-limiting nutrients ¹. A debate over which nutrient(s) to reduce, nitrogen and/or phosphorus, is ongoing in the limnological literature ^7,8, but a phosphorus-only reduction is the more commonly applied approach in practice. There are many success stories where phosphorus reduction has led to a decrease of phytoplankton biomass (Chlorophyll a concentration - Chla) ⁸. But what about toxin concentrations, presumably one of the most important targets of a management action? The implicit, and entirely reasonable assumption (although faulty as we will show here), is that toxin concentrations will decrease proportionally with biomass. Thus, it is assumed that phytoplankton stoichiometry (here MC/Chla) remains unchanged. This idea has deep roots in limnology and oceanography ever since Redfield´s seminal work illustrating relatively constant nitrogen-to-phosphorus ratio in marine phytoplankton ⁹, which was subsequently extended to include other elements, e.g. C₁₂₄N₁₆P₁S_1.3K_1.7Mg_0.56Ca_0.5Fe_0.0075Zn _0.0008Cu _0.00038Cd _0.00021Co _0.00019^10,11.

SHOULD toxin concentration increase when phosphorus loading is reduced?

Natural phytoplankton communities consist of numerous species and strains including those that can and cannot produce toxins ^1,2,12. Further, the toxigenic strains do not constantly synthesize toxins (i.e. the MC cell quota is variable) ¹². Reducing phosphorus loading will lower phytoplankton biomass, but also increase the availability of other resources, like nitrogen and light, which will in turn change the MC quota and community composition.

For MC production, nitrogen and light are important resources: nitrogen is part of the MC molecule (min. 7 N/MC) and light is needed as an energy source for its synthesis². This dependence is evident in several laboratory experiments, which show increasing nitrogen results in higher MC quota (fig. 1A, for method description, see SI). For example, Van de Waal (V09 in fig. 1A) ¹³ found that the cellular toxin quota in nitrogen-replete medium (containing 12 mM NO₃) was more than double (log2 of the ratio of MC quotas is 1.1) compared to nitrogen-deplete medium (0.02 mM NO₃). A clear outlier is the study by Pimentel and Giani (PG14 in fig. 1A) ¹⁴, which showed increased mcyD expression and MC production under low nitrogen cultivation. The authors suggest that severe nitrogen limitation caused oxidative stress, which reduced MC production (see discussion about the biological role below). However, the majority of studies do not show this effect. Other studies investigated the effect of irradiance on MC quota (fig. 1B). Most results show higher light intensity increases MC quota. For instance, the fold-change of MC quota was 1.2 (e.g. MC quota has more than doubled) when cultivated at 80 compared to 20 µmol photons/m²/s (UG92 in fig. 1B) ¹⁵. However, a few studies showed a decreased MC quota with higher irradiance. For example, Phelan and Downing (PD11 in fig. 1B) ¹⁶ used continuous light intensities of 8 and 37 µmol photons/m²/s and showed a reduction of MC quota to 22% on average (log2 of ratio = - 2.1). They suggest that missing dark phases in their experimental set-up prevent sufficient regeneration of cell damages and respective MC synthesis. Hence, this observation is not relevant to the field. Generally, there is a lot of variability which can be attributed to strain and experimental variability, as well as feedbacks of effects. For example, higher nitrogen in nitrogen-limited cultures will increase biomass and decrease light due to self-shading, which in turn may decrease MC production. Nonetheless, laboratory data suggest that increased availability of nitrogen and light, an expected result of a phosphorus-only reduction, increases MC production and quota. A phosphorus-only reduction means the toxigenic strain will become more toxic.

A change in MC quota (of the toxigenic strain) may also translate to a change in community composition. This relates to the biological role of MC, which is not yet fully understood. Potential functions include protection against oxidative stress ^18,19, carbon limitation ^20,21 or parasite infection ^22,23. Despite this uncertainty, the highly conserved genome region of the enzyme complex for synthesis ²⁴ and wide distribution of cyanotoxins ²⁵ indicate a substantial fitness benefit for the producer. Hence, higher MC quota (due to increased nitrogen and light availability) will benefit the toxin-producing cells in competition against other phytoplankton and result in an increased fraction in the community, which is also evident in field observations. Welker et al. ²⁶ investigated Microcystis biomass and MC dynamics in Lake Müggelsee and concluded that toxin-producing strains dominate at times of high nitrogen and low phosphorus availability. Also, several other studies show that the growth of toxigenic strains was associated with nitrogen availability in lakes ^17,27,28. For instance, in Lake Mikata in 2005 ²⁷, the population becomes progressively more toxigenic as phosphorus decreases and nitrogen increases. Hence, a phosphorus-only reduction means there will be more toxigenic cells.

In summary, the underlying mechanisms, laboratory experiments and limited indirect evidence from field observations, i.e. none of the above analyses include direct measurements of how MC concentration changes with total phosphorus (TP), suggest that higher availability of nitrogen and light increases both, MC cell quota and toxigenic fraction. This may have important ramifications for lake management. For a phosphorus-only reduction, the phytoplankton stoichiometry (here MC/Chla) will change in a direction that will counteract the reduction in biomass. Decreasing phosphorus may actually increase toxin concentration. This was predicted by a model that includes these mechanisms, which estimated that a 40 % phosphorus-only reduction will increase MC concentration by 14 % in Lake Erie ¹⁷.

DO toxin concentrations increase when phosphorus loading is reduced?

Several management case studies of phosphorus reduction in lakes showed reduced biomass ⁸ but whether toxin concentrations decreased proportionally is often not quantified. We explored this question for the lakes summarized by Schindler et al. by performing a strategic literature review (for methodology see SI). Of the 35 lakes, time-series data of MC concentrations are only available for 9 lakes, and they do not cover the time from before to after the phosphorus-only reduction. Therefore, there is a lack of direct field observations that show how toxin concentrations change when total phosphorus is reduced.

Alfred Redfield analyzed large oceanic datasets to quantify stoichiometric relations in phytoplankton. Similarly, we seek insight into stoichiometry, here MC per biomass, and how that may relate to phosphorus by analyzing a large database of lake observations. Specifically, we ask: Do decreasing phosphorus concentrations in lakes increase MC concentration?

We analyzed the NLA dataset, which includes relevant observations, including TP, total nitrogen (TN), Chla, dissolved inorganic nitrogen (DIN, i.e. biologically available nitrogen), turbidity (NTU, i.e. parameter for available light) and MC concentration (for methodology see SI). Several previous studies analyzed NLA data, using different approaches to find relations between MCs and environmental variables across lakes ^29–32. We used two different methods. The first is a simple and targeted approach to basic regression analyses. Here, we limit our analysis to the change of MC concentrations within individual lakes across time, because the absolute concentrations may vary among different lakes due to other factors, and this is more representative of a change due to management. With the second and more complex method, we investigate overall patterns across all lakes, using Generalized Additive Models (GAMs)^33,34.

We first present the basic regression analysis. Samples were collected in 2007, 2012 and 2017 and the database comprises 9821 sampling events including 1999 detections of MC concentrations (fig. 2). This included 453 time pairs, i.e. data points with detectable MC concentrations at different times within a lake. We correlate changes of different parameters (e.g. MC concentrations) with changes in TP concentrations, where change is quantified as a relative change using log2 of the ratio, i.e. fold change. However, note that the observed change in phosphorus may not be due to management, but natural hydrological variations. This means that nitrogen may also change along with it (e.g. dry or wet years affect nutrient nonpoint source input and flushing). Thus, a change in TP may or may not be associated with a concomitant change in TN so it may be a representative of a phosphorus-only or dual nutrient reduction of a management plan. We therefore split the dataset into two groups (see visualization in fig. 2): In the first group TP and TN concentrations change in the same direction, corresponding to a dual nutrient reduction. A representative lake is South Lake, Minnesota, where a TP concentration of 1.06 mg/L in 2007 and 0.52 mg/L in 2012 was measured, corresponding to a fold change of - 1.02. TN concentration also decreased from 14.4 to 4.9 mg/L, a fold change of -1.49. In the second group, TP and TN concentrations change in opposite directions, corresponding to a phosphorus-only reduction. An example lake is Lyman Lake, Arizona, which experienced a fold change of TP concentrations of log2 (0.16/0.23) is -0.53 and TN concentrations of log2(0.67/0.62) is 0.09. This group was further divided into two sub-groups which are either phosphorus-limited or not, depending on changes in biomass in the same or opposite direction as TP concentrations. We first present and discuss the results for the time pairs where phosphorus only was reduced and which were phosphorus limited.

In fig. 3 changes in TP concentrations are shown, correlated with changes in various other parameters. A phosphorus-only management action would move a lake from right to left on the x-axis. The phosphorus-limited subset was defined based on a positive correlation between changes in TP and biomass, for instance in the representative Lyman Lake a fold change of chlorophyll a of -0.4 was measured (fig. 3 A). Changes in TP correlate negatively with changes in DIN concentration (fig. 3 B1, fold change of DIN in Lyman Lake was 0.7) and positively with changes in turbidity (fig. 3 B2, fold change of turbidity in Lyman Lake was -1), showing an increasing availability of nitrogen and light with reduced TP concentrations. A negative correlation of TP concentrations with MC concentration per biomass was identified (fig. 3 C, fold change of MC/biomass in Lyman Lake was 1.55). This analysis illustrates that if TP concentration decreases, the phytoplankton stoichiometry changes. Specifically, MC concentrations per biomass increase, consistent with the above mechanism: a reduction in biomass results in higher availability of nitrogen and light (both observed in the data), which stimulates MC production and increases toxigenic fraction. This counteracts the decrease in biomass (theoretical dashed line in fig. 3 D) and results in an increase in MC concentrations with reduced TP concentrations (solid line in fig. 3 D). This is also evident in the example Lyman Lake, showing a fold change in MC concentration of 1.14, the concentration has more than doubled.

The results for different subsets of the data provide further insights into how MC concentration is expected to change under different management scenarios. In figure 4, we show the correlations of TP and MC concentrations (solid lines) and the corresponding theoretical lines assuming a constant phytoplankton stoichiometry (i.e. changes in biomass, dashed lines) for the different groups defined in fig. 2. In the subset where TP and TN concentrations change concomitantly, corresponding to a dual nutrient reduction management strategy, MC concentration decreases along with biomass (figure 4B). In the South Lake representing this group, the MC concentration decreased to more than a half from 1.3 µg/L to 0.5 µg/L, fold change is -1.26. This is consistent with the mechanisms discussed above: nitrogen availability decreased and less MC was produced (see also fig. S2). In the group corresponding to a phosphorus-only reduction, lower TP increased MC concentration (fig. 4 C). Biomass does not change with reduced TP concentrations (dashed line), because not all lakes were phosphorus-limited. This prompted us to split the group into sub-groups being phosphorus limited and not phosphorus limited. In the group of phosphorus-limited lakes, MC concentrations increase with reduced TP concentrations as discussed above (fig. 4 D). In the group which are not phosphorus-limited, a reduction in TP is associated with an increase in TN and an increase in biomass, and MC concentration changes along with biomass (fig. 4 E and S4).

The time series analysis above is limited because requiring time pairs within individual lakes eliminates most (80% including non-detects) of the observations. We, therefore, applied a second approach where we analyzed the NLA data with a focus on patterns across lakes. We used all observations and developed models for the same variables as in the first approach (i.e. Chla, DIN, turbidity, MC/Chla, MC). Briefly, we applied GAMs with a Tweedie distribution using a log link function. The Tweedie distribution is ideal for positive response variables such as concentrations or their ratios as it has the advantage that it can handle data with zeros and right-skewed data. Based on the estimated power parameter it resembles commonly used distributions such as the Poisson or Gamma distribution³⁵. Predictor variables for the mean included log-transformed TN (log TN) and TP (log TP). GAMs were fitted to each response variable using smoothing splines to allow for non-linear main effects and non-linear interactions between log TN and log TP (for more information on methodology see SI).

In fig. 5, predictions of each model are shown as heatmaps on a grid of log TP and log TN. The model results support the findings of changes in individual lakes discussed above and extend it to more general patterns across lakes with different N/P ratios. Specifically, the heatmap allows us to investigate systems with different nutrient conditions, for instance at high TN and low TP conditions, the system is probably phosphorus-limited and vice versa. Also, the outcome of different management actions can be simulated with the heatmap. A management action of phosphorus-only reduction would move a lake horizontally from the right to the left side, whereas a dual reduction would move it diagonally to the lower left corner.

These model predictions, which are based on observed field data, support the mechanism described previously. Reducing phosphorus only decreases biomass and increases nitrogen and light availability (i.e. turbidity decreases) as well as MC per biomass, especially in medium to lower TP concentration (fig. 5A, B1, B2, C). This may increase the toxigenic fraction in the cyanobacterial community (no observations are available for that). The increase in MC quota and toxigenic fraction counteract the decrease in biomass and increase total MC concentration. The prediction for MC concentration supports this and indicates an even more diverse pattern: MC concentration first increases when phosphorus only is reduced but decreases at very low TP concentrations (fig 5D). Hence, when phosphorus is reduced to a large extent, the MC concentration decreases with the reduced biomass. Nevertheless, the phytoplankton stoichiometry (i.e. MC/Chla) is still higher than before the reduction (fig. 5C). The same pattern was also shown in a simulation of a gradual phosphorus reduction of an agent-based Microcystis model¹⁷. However, the predictions of the GAMs also show that a dual reduction of TP and TN reduces both, biomass and MC concentrations.

Implications for management

The NLA data are consistent with the underlying mechanisms, reported in laboratory observations and circumstantial evidence found in field observations, and models, and strengthen the notion that reducing only phosphorus may reduce biomass, but toxin concentrations will not reduce proportionally. Hence, reducing only phosphorus loading to a lake is risky, and field data show that MC concentrations actually increase (except for extreme reduction in phosphorus). A dual reduction of phosphorus and nitrogen seem more effective in reducing toxin concentration.

The “Redfield ratio” of a constant nitrogen-to-phosphorus ratio has served as a useful reference for data analysis and modeling in limnology and oceanography for almost a century. At the same time, it is clear that phytoplankton stoichiometry does change and that this can be important to consider in some cases ^11,36. Our findings here show that, when it comes to toxin content, specifically MC, it is important to consider variable stoichiometry, and highlight the importance of moving on from the paradigm that “reducing phosphorus equals reducing biomass equals reducing toxicity” ³⁷. This will require more field observations and toxin quantification should be included as a standard measure when evaluating lake status before and after a management action.

Acknowledgement

This project was financed by Deutsche Forschungsgemeinschaft (DFG), GRK 2032: Urban Water Interfaces (UWI) and National Oceanographic and Atmospheric Administration grant NA18NOS478017. Thanks to Falk Eigemann, Jutta Hoffmann, Carsten Lackner and Behnam Zamani for constructive feedback.

Author contributions

Conceptualization: F.L.H. and C.S., Data curation: C.S., J.M., T.N. and J.P., Formal analysis: C.S. and J.M., Funding acquisition: F.L.H., Investigation: C.S., J.M. and T.N., Methodology: F.L.H., C.S. and J.M., Project administration: F.L.H., Resources: F.L.H., Supervision: F.L.H., Validation: C.S. and J.M., Visualization: C.S., J.M. and F.L.H. Writing - original draft: C.S. and F.L.H., Re-writing draft: all authors

Competing interest declaration

The authors declare no competing interest.

1. Huisman, J. et al. Cyanobacterial blooms. Nat Rev Microbiol 16, 471–483 (2018).

2. Chorus, I. & Welker, M. Toxic Cyanobacteria in Water; A Guide to Their Public Health Consequences, Monitoring and Management; Second Edition. (2021).

3. Bullerjahn, G. S. et al. Global solutions to regional problems: Collecting global expertise to address the problem of harmful cyanobacterial blooms. A Lake Erie case study. Harmful Algae 54, 223–238 (2016).

4. Environmental Protection Agency (EPA). 2017 National Lakes Assessment - Infographic. https://www.epa.gov/system/files/documents/2022-09/infographic_9_7_22b.pdf (2022).

5. Pearl, H. W. & Huisman, J. Blooms Like It Hot. Science (1979) 320, 57–58 (2008).

6. Chapra, S. C. et al. Climate Change Impacts on Harmful Algal Blooms in U.S. Freshwaters: A Screening-Level Assessment. Environ Sci Technol 51, 8933–8943 (2017).

7. Paerl, H. W. et al. It Takes Two to Tango: When and Where Dual Nutrient (N & P) Reductions Are Needed to Protect Lakes and Downstream Ecosystems. Environ Sci Technol 50, 10805–10813 (2016).

8. Schindler, D. W., Carpenter, S. R., Chapra, S. C., Hecky, R. E. & Orihel, D. M. Reducing phosphorus to curb lake eutrophication is a success. Environ Sci Technol 50, 8923–8929 (2016).

9. Redfield, A. On the proportions of organic derivatives in sea water and their relation to the composition of plankton. in Daniel RJ (ed) James Johnstone memorial volume 176–192 (University Press of Liverpool, 1934).

10. Quigg, A. et al. The evolutionary inheritance of elemental stoichiometry in marine phytoplankton. Nature 425, 291–294 (2003).

11. Klausmeier, C. A., Litchman, E., Daufresne, T. & Levin, S. A. Phytoplankton stoichiometry. Ecol Res 23, 479–485 (2008).

12. Dick, G. J. et al. The genetic and ecophysiological diversity of Microcystis. Environmental Microbiology vol. 23 7278–7313 Preprint at https://doi.org/10.1111/1462-2920.15615 (2021).

13. Van De Waal, D. B. et al. The ecological stoichiometry of toxins produced by harmful cyanobacteria: An experimental test of the carbon-nutrient balance hypothesis. Ecol Lett 12, 1326–1335 (2009).

14. Pimentel, J. S. M. & Giani, A. Microcystin production and regulation under nutrient stress conditions in toxic Microcystis strains. Appl Environ Microbiol 80, 5836–5843 (2014).

15. Utkilen, H. & Gjolme, N. Toxin Production by Microcystis aeruginosa as a Function of Light in Continuous Cultures and Its Ecological Significance. Appl Environ Microbiol 58, 1321–1325 (1992).

16. Phelan, R. R. & Downing, T. G. A growth advantage for microcystin production by microcystis PCC7806 under high light. J Phycol 47, 1241–1246 (2011).

17. Hellweger, F. L. et al. Models predict planned phosphorus load reduction will make Lake Erie more toxic. Science (1979) 376, 1001–1005 (2022).

18. Zilliges, Y. et al. The cyanobacterial hepatotoxin microcystin binds to proteins and increases the fitness of Microcystis under oxidative stress conditions. PLoS One 6, (2011).

19. Dziallas, C. & Grossart, H. P. Increasing oxygen radicals and water temperature select for toxic microcystis sp. PLoS One 6, (2011).

20. Guljamow, A. et al. Diel variations of extracellular microcystin influence the subcellular dynamics of rubisco in microcystis aeruginosa pcc 7806. Microorganisms 9, (2021).

21. Barchewitz, T. et al. Non-canonical localization of RubisCO under high-light conditions in the toxic cyanobacterium Microcystis aeruginosa PCC7806. Environ Microbiol 21, 4836–4851 (2019).

22. Rohrlack, T., Christiansen, G. & Kurmayer, R. Putative antiparasite defensive system involving ribosomal and nonribosomal oligopeptides in cyanobacteria of the genus planktothrix. Appl Environ Microbiol 79, 2642–2647 (2013).

23. Agha, R. & Quesada, A. Oligopeptides as biomarkers of cyanobacterial subpopulations. Toward an understanding of their biological role. Toxins (Basel) 6, 1929–1950 (2014).

24. Rantala, A. et al. Phylogenetic evidence for the early evolution of microcystin synthesis. PNAS 101, 568–573 (2003).

25. Harke, M. J. et al. A review of the global ecology, genomics, and biogeography of the toxic cyanobacterium, Microcystis spp. Harmful Algae 54, 4–20 (2016).

26. Welker, M., Von Döhren, H., Täuscher, H., Steinberg, C. E. W. & Erhard, M. Toxic Microcystis in shallow lake Müggelsee (Germany) - Dynamics, distribution, diversity. Arch Hydrobiol 157, 227–248 (2003).

27. Yoshida, M., Yoshida, T., Takashima, Y., Hosoda, N. & Hiroishi, S. Dynamics of microcystin-producing and non-microcystin-producing Microcystis populations is correlated with nitrate concentration in a Japanese lake. FEMS Microbiol Lett 266, 49–53 (2007).

28. Li, J., Xian, X., Xiao, X., Li, S. & Yu, X. Dynamic characteristics of total and microcystin-producing Microcystis in a large deep reservoir. Environmental Pollution 335, (2023).

29. Merder, J. et al. Geographic redistribution of microcystin hotspots in response to climate warming. Nature Water 1, 844–854 (2023).

30. Beaver, J. R. et al. Land use patterns, ecoregion, and microcystin relationships in U.S. lakes and reservoirs: A preliminary evaluation. Harmful Algae 36, 57–62 (2014).

31. Yuan, L. L. & Pollard, A. I. Using National-Scale Data to Develop Nutrient-Microcystin Relationships That Guide Management Decisions. Environ Sci Technol 51, 6972–6980 (2017).

32. Buley, R. P., Correia, H. E., Abebe, A., Issa, T. B. & Wilson, A. E. Predicting microcystin occurrence in freshwater lakes and reservoirs: assessing environmental variables. Inland Waters 11, 430–444 (2021).

33. Wood, S. N. Generalized Additive Models: An Introduction with R. (CRC Press, A Chapman & Hall Book, 2017).

34. Rigby, R. A. & Stasinopoulos, D. M. Generalized Additive Models for Location, Scale and Shape. Appl. Statist vol. 54 (2005).

35. Dunn, P. K. & Smyth, G. K. Series Evaluation of Tweedie Exponential Dispersion Model Densities. Statistics and Computing vol. 15 (2005).

36. Droop, M. R. The nutrient status of algal cells in batch culture. Journal of the Marine Biological Association of the United Kingdom 55, 541–555 (1975).

37. Wilhelm, S. W. et al. Response to “Model assumptions limit implications for nitrogen and phosphorus management”: The need to move beyond the phosphorus = biomass = toxin doctrine. J Great Lakes Res 48, 1738–1739 (2022).

The authors declare no competing interests.

2407016SINLA.docx

Download PDF

Version 2

posted

You are reading this latest preprint version

Lake toxin concentrations increase when phosphorus is reduced

Status:

Version 2

Abstract

Figures

Significance Statement

Main text

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 2