Understanding the relationship between nutrient availability and freshwater cyanobacterial growth and abundance

ABSTRACT

Anthropogenic eutrophication is a well-established cause of cyanobacterial blooms in freshwaters. Early studies proposed eutrophication control focused on phosphorus (P) as the key limiting nutrient. The rationale was that nitrogen (N) limitation is alleviated by diazotrophic cyanobacteria that fix atmospheric N. However, more recently, studies suggest that the importance of N fixation may have been overstated, and some call for anthropogenic N control as well as P. Reducing nutrient concentrations below critical levels clearly reduces phytoplankton biomass, but increasing evidence shows that cyanobacterial species can adapt to low and variable nutrient conditions, outcompeting eukaryotic phytoplankton. Strategies supporting cyanobacterial dominance include (1) high affinity P uptake, (2) preferencing cellular storage of P over immediate metabolic utilization for growth, (3) alternation between N fixation and inorganic N utilization as needed, and (4) use of organic nutrients when inorganic nutrient supplies are limited. Predicting the ecological responses of cyanobacteria to nutrients therefore requires more complex models than the simple resource models using inorganic nutrients widely used for contemporary studies. Additionally, many ecological studies are not conducted on timescales relevant to understanding how nutrient fluxes drive physiological responses. We advocate for greater use of high-frequency automated measurements of nutrients and cyanobacteria to address the timescale issue. Additionally, field-based experimental studies focused on nutrient cycling, fluxes, and pools are needed to validate laboratory findings and extrapolate them to in situ conditions. The challenge is to undertake these studies at sufficient spatial and temporal scales to predict large-scale responses of cyanobacteria to nutrients.

KEYWORDS:

• cyanobacteria
• fluctuating
• limiting
• nitrogen
• phosphorus
• physiology

Introduction

The effect of nutrient loading in stimulating algal blooms has been a major area of research for many decades (Conley et al. 2009). Despite this knowledge and the related mitigation actions involving nutrient reductions, the frequency and scale of blooms continues to increase globally (O’Neil et al. 2012, Huisman et al. 2018). Examples of nutrient load reduction exist, particularly from point sources such as upgrades to sewage treatment plants, and although these reductions have helped reduce algal biomass (e.g., Edmondson 1972), tackling load reduction from nonpoint sources has been less effective (e.g., Osgood 2017, Le Moal et al. 2019). Additionally, robust studies investigating the scale of nutrient load reductions needed, which nutrients should be reduced, and the timeframes for corresponding reductions in algal biomass remain limited, partly due to our overly simplistic understanding of the complexities of nutrient-algal interactions, both in terms of generalizations across species and the monitoring and experimental approaches used to test responses of algae to nutrients. Ultimately, these factors impact our ability to predict blooms and the capacity to improve management responses to blooms.

Freshwater algal blooms are often dominated by cyanobacteria (O’Neil et al. 2012), an ancient group of species that has survived for eons and often outcompeted eukaryotic algae in freshwater (Mays et al. 2021). The availability of nutrients for cyanobacterial growth is controlled by a complex suite of biogeochemical and physical processes. Nutrient processes are directly related to cyanobacterial utilization (Fig. 1). Nutrients enter waterways in bioavailable and refractory forms from the land and air and are transformed between organic and inorganic forms via a suite of physical, chemical, and biological processes. Microbes play a key role in the generation of dissolved inorganic nutrients, fuelled by organic matter from the terrestrial environment and generated within freshwater systems themselves (e.g., decaying algal biomass). These inorganic nutrients are utilized by algae, aquatic plants, and microbes and then transformed back into organic forms to be recycled, buried, or exported from the system (Essington and Carpenter 2000, Moss 2012)

Figure 1. Utilization of N and P by cyanobacteria in lentic waterbodies. PON = particulate organic nitrogen, DON = dissolved organic nitrogen, NH₄⁺ = ammonium, NO₃⁻ = nitrate, N₂ = nitrogen gas, POP = particulate organic phosphorus, DOP = dissolved organic phosphorus, DIP = dissolved inorganic phosphorus.

Nitrogen (N) biogeochemistry is more complex than that of phosphorus (P) in aquatic systems because it is driven by a wider suite of microbial processes. Accordingly, turnover of N, particularly in the water column, is regulated predominantly by algal and microbial processes and is extremely rapid compared with cycling of P (Reynolds 2006). Sediment remineralization processes may be a more important source of bioavailable P for cyanobacterial growth (e.g., Sakamoto and Okino 2000). Measurements of nutrient concentrations at frequencies typically used in most monitoring programs and reported in the scientific literature (e.g., weekly or monthly) have limited value because of rapid turnover. Our ability to resolve nutrient transformations and fluxes, particularly in the trophogenic zones where cyanobacteria grow and nutrients are often strongly depleted, is therefore strongly compromised.

Studies of N vs. P effects on eutrophication

The notion that P may be a limiting factor for algal growth goes back more than a century (Correll 1999). Early studies of the importance of P for algae, and particularly cyanobacteria in lakes, focused on the impact of large-scale P addition to oligotrophic lakes in the Experimental Lakes Area in northwestern Ontario, Canada (Schindler 1974). Artificial P inputs resulted in cyanobacterial blooms in these systems, leading to the concept that P is the key limiting nutrient for algal growth in freshwater systems. By implication, the nutrient load was related to the nutrient concentration in the waterbody and the resulting biological response. Later, Vollenweider (1968) formulated quantitative loading criteria for P and N and the expected trophic conditions in waterbodies. The correlation between P and chlorophyll a (Chl-a) as a measure of phytoplankton was substantiated using a wide array of data for lakes across the globe (e.g., Oglesley and Schaffner 1978, Canfield and Bachmann 1980, Vollenweider and Kerekes 1980). Conversely, a reduction in algal biomass has also been demonstrated by decreasing P loading to lakes (e.g., Edmondson 1972, Schindler et al. 2016).

Many subsequent studies have focused on the positive correlation between P and cyanobacterial biomass. However, Lewis et al. (2011) noted that more attention has been given to P than to N because (1) P is more easily removed from anthropogenic sources than N, (2) N fixation by cyanobacteria has been assumed to make N control ineffective, and (3) the correlation between Chl-a and total P among lakes is stronger than the correlation between Chl-a and total N. They note that the stronger correlation is unsurprising, given that P is a more conserved element than N, an element strongly regulated by fluxes between water–atmosphere phases. Studies of riverine inputs to lakes suggest excess N inputs, over P, relative to requirements for homeostasis of primary producers (Howarth et al. 2021). Lewis et al. (2011) has also noted that the correlation between P and mean or peak Chl-a concentrations among lakes has been erroneously interpreted as showing cause and effect. While the correlation is useful for determining the maximum biomass of cyanobacteria, and algae more generally, the correlation reveals little about how different species respond to nutrient limitation because P is an essential constituent of algal biomass, as is Chl-a. Chl-a and P will always be present together, whether P is limiting or not, and nutrient limitation cannot necessarily be inferred from such correlations.

A number of recent studies demonstrated the need to control N as well as P, despite the ability for some cyanobacterial species to fix N (Lewis et al. 2011). Many cyanobacterial species are not diazotrophic and form blooms where dissolved inorganic N (DIN) concentrations are high (e.g., Lake Taihu, dominated by the non-N fixing species Microcystis spp.; Xu et al. 2010). Additionally, cyanobacteria have a higher N:P requirement than some other algal groups (e.g., diatoms), and therefore reducing P, but not N, will favor species with elevated N:P ratios (Glibert and Burford 2017). Other studies have also shown that N fixation contributes less N to lakes than previously thought and may not compensate for N deficiencies at a system scale (Burford et al. 2006, Scott and McCarthy 2010).

Establishing an unequivocal answer to the relative importance of N versus P control to reduce cyanobacterial blooms has also been confounded by a number of issues, identified by Post (2005) as (1) chemical measurements that identify pools (relevant to maximum algal biomass) rather than fluxes of nutrients (relevant to growth); (2) nutrient addition bioassays that produce artificial limitation of either nutrients or trace elements, often associated with the effects of sample enclosure; and (3) utilization of a wide range of N and P sources (e.g., organic nutrients) frequently not measured.

P strategies used by cyanobacteria

Ecological studies on the role of P in stimulating cyanobacteria have drawn a direct, and rather simplistic, line between P availability and cyanobacterial biomass. While biomass accumulation may increase with P availability, physiological studies have demonstrated a succession of cyanobacterial species according to P availability, with utilization strategies specific among species and even within species (i.e., at strain level; see review by Xiao et al. 2022). One review showed, for example, that blooms can also form in oligotrophic lakes because cyanobacterial species have adapted to low nutrient availability (Reinl et al. 2021).

Australian strains of the toxic cyanobacterial species Raphidiopsis raciborskii (formerly Cylindrospermopsis raciborskii) are an example of a species highly adapted to low DIP availability (Burford et al. 2016). Studies in subtropical Australian reservoirs have shown that R. raciborskii typically formed blooms in summer months when dissolved inorganic P (DIP) concentrations were near the chemical detection limit (Burford and O’Donohue 2006, Burford et al. 2014). The response of R. raciborskii to low DIP availability was to increase DIP uptake rate via a physiological shift from passive to high-affinity uptake (Prentice et al. 2015, Willis et al. 2017). In the field, a threshold concentration of 4.7 µg L⁻¹ DIP was identified as the switch from passive to high-affinity uptake during an R. raciborskii bloom (Prentice et al. 2015). The high-affinity uptake was substantiated in studies showing upregulation of the active P uptake gene, Pst (Willis et al. 2019). Below DIP availability thresholds, other studies have shown that high-affinity uptake genes, pstS and sphX, are upregulated, and the passive uptake gene, pit, is down-regulated (Orchard et al. 2009). Rapid uptake of DIP follows, the rate of which varies widely among species and strains (Xiao et al. 2022).

High-affinity uptake relies on biogeochemically generated P. Studies have shown that turnover of the DIP pool (i.e., cycling of P in the water column) can occur within a few minutes during summer stratified conditions, when DIP concentrations are close to or below detection limits (Nowlin et al. 2007, Prentice et al. 2015). Regeneration rates at the community level may cycle particulate P pools within the eutrophic zone every 0.7–2.1 days (Nowlin et al. 2007). This regeneration is the result of cell membrane leakage, cell death, and viral lysis as well as zooplankton excretion (Nowlin et al. 2007, Hong et al. 2013).

High-affinity uptake therefore provides a mechanism for P acquisition to maintain growth under low DIP conditions, which may occur when the water column is stratified. The DIP store in the surface mixed layer has largely been assimilated by algae. Additionally, high-affinity uptake allows depletion of this P store during periods between replenishment from high-flow events. In this way blooms can be perpetuated as external DIP is depleted. Galvanese et al. (2019) also showed that elevated temperature enhanced the ability of R. raciborskii to grow at low DIP concentrations. Reynolds (2006) contends that P availability does not limit cyanobacterial (algal) metabolism or growth until external DIP is below conventional analytical detection limits.

Another strategy to ensure cyanobacterial growth when DIP is depleted in surface waters is high P storage capacity, such that P reserves persist through multiple generations. Greater relative storage of P, over N, can result in low cellular N:P ratios. Studies of the toxic cyanobacterium R. raciborskii show that this species does not increase its growth rate with increasing DIP concentrations, but preferentially stores P (Xiao et al. 2020). Over short time scales (hours to days), growth rates have been shown to be unaffected by P availability until starvation and senescence occur (Willis et al. 2017, Xiao et al. 2020). Storage allows sustained growth for multiple generations after P becomes limiting, providing flexibility for the species to grow under low and/or variable DIP concentrations. Reynolds (2006) has suggested that P storage in cyanobacteria can theoretically support 3–4 doublings without any DIP uptake; therefore, accumulation of polyphosphate can provide a mechanism for storage of P to support continued growth during P-depauperate conditions. Considerable variability in the ability to accumulate polyphosphate has been measured among different species of cyanobacteria (Li and Dittrich 2019). Threshold external P levels are also likely, at which preferential polyphosphate degradation occurs. Conversely, the relatively high concentrations of P typically used in growth media can result in detrimental effects on R. raciborskii growth. For example, Willis et al. (2015) found that the highest growth rates in R. raciborskii occurred when concentrations of P in growth media were half or lower than those typically used.

Another strategy for accessing P when external DIP reserves are low is the use of organic forms of P (dissolved organic P [DOP]; Tiwari et al. 2015). DOP research has focused on the capacity to utilize P-esters via activity of the alkaline phosphatase enzyme (APA). For example, a threshold for APA based on cellular P stores in cyanobacterial blooms has been determined in a water supply reservoir (Prentice et al. 2019). The advent of molecular methods has provided the mechanism to study a range of other DOP-related genes associated with increased DOP utilization (Harke et al. 2012, Lin et al. 2018). In some species, genes involved in DOP utilization (i.e., phoA, phoD, and phoX) were upregulated at low DIP concentrations, but not for all species, and genes for DOP utilization may be upregulated even at relatively high DIP concentrations (e.g., Harke et al. 2012, Willis et al. 2018). A study of the P metabolism genes in genomes of multiple Australian strains of R. raciborskii found that the genes were identical (Sinha et al. 2014, Willis et al. 2018), so differential transcription likely affects how the strains use P.

Marine cyanobacteria seem to have additional strategies to continue to grow under low DIP concentrations that to date have not been observed in freshwater cyanobacteria, such as production of non-P membrane lipids (Van Mooy et al. 2009) and use of the organic compound phosphite (Martínez et al. 2012, Polyviou et al. 2015). However, considerably less research is available on alternative strategies for utilization of DOP by freshwater cyanobacteria than marine.

The importance of different P utilization strategies varies among cyanobacterial species (Dignum et al. 2005). Species can be categorized as having greater emphasis on one or more of the following strategies:

Growth: a response to transient DIP inputs by modifying growth rates.

High-affinity uptake: efficient use of external DIP.

Storage: storing P by transforming external supplies into polyphosphate.

Scavenging: use of enzymes to access less bioavailable P.

N strategies used by cyanobacteria

Much of our understanding of N utilization by cyanobacteria has focused on N fixation as a strategy for growth and dominance. However, many species are incapable of N fixation (e.g., Microcystis), although recent studies have shown that microbes associated with colonies of the cyanobacterial genus Microcystis may also fix N, providing an N supply external to Microcystis cells, but within the colonial sheath (Cook et al. 2020). Additionally, Microcystis may capitalize on short-term N fertilization of waterbodies as a result of N fixation by other cyanobacteria species and their subsequent senescence (Beversdorf et al. 2013).

N fixation in capable species will only occur when dissolved N is insufficient for growth (e.g., Burford et al. 2006). N fixation is an energetically expensive process, so use of dissolved forms of N, especially in lower-light environments, is a more efficient strategy. The relative inefficiency of N fixation is exemplified by a study showing a 3–5 day development of heterocysts (required for N fixation) for R. raciborskii strains once dissolved N becomes limiting (Willis et al. 2016). This species produces only terminal heterocysts but can compensate for separation of central cells from heterocystous cells within filaments using a highly efficient N transfer mechanism (Plominsky et al. 2013, 2015). Once heterocysts are produced, N fixation begins immediately, but subsequent addition of higher concentrations of DIN (i.e., to saturate uptake requirements) caused heterocyst loss after about 5 days, halting N fixation (Willis et al. 2016). A nitrate-N threshold of ∼30 μg L⁻¹ has been determined for heterocyst differentiation in Cyanothece sp. (Agawin et al. 2007). Moisander et al. (2012) showed that R. raciborskii maintained similar net growth rates under diazotrophic and nondiazotrophic conditions, whereas Dolicospermum (previously Anabaena) sp. growth was significantly reduced under DIN enrichment. However, studies of other strains of R. raciborskii showed slower growth rates under diazotrophic versus DIN replete conditions (Willis et al. 2016). Therefore, generalization about cyanobacterial N fixation responses to DIN availability is difficult and requires an improved understanding of N quotas and upregulation of N-fixation genes as a trigger for N fixation.

Studies have shown that the importance of N fixation as a significant source of N into lakes may be overstated (e.g., Scott and McCarthy 2010, Scott et al. 2019). As a result, dissolved forms may provide the primary N source for much of the year, as shown for R. raciborskii-dominated populations (Burford et al. 2006). Studies have also shown that ammonium may be a preferred DIN source over N fixation or nitrate (Ferber et al. 2004, Burford et al. 2006). Glibert (2017) highlights studies showing that cyanobacteria prefer ammonium compared with diatoms, which may prefer nitrate. Most cyanobacteria assimilate N through the GS-GOGAT biochemical pathway, but some also have glutamate dehydrogenase (GDH), which is advantageous for those species because ammonium assimilation through GDH does not require ATP (Muro-Pastor et al. 2005). Additionally, relatively low ammonium concentrations can suppress nitrate uptake. The presence of ammonium itself may decrease any further ammonium uptake. For example, in many cyanobacteria, the transcriptional activator of N assimilation genes, NtcA, is negatively controlled by ammonium and the metabolite 2-OG (Coruzzi and Bush 2001, Lindell and Post 2001, Muro-Pastor et al. 2005). This metabolite accumulation represses further N assimilation of ammonium (Post 2005). Glibert et al. (2016) showed that nitrate uptake was directly related to the fraction of diatoms in the community while the proportion of ammonium uptake was related to the fraction of cyanobacteria in the community. Differing abilities to take up and assimilate nitrate and ammonium have also been reported among cyanobacterial species that fix N compared with those that do not (e.g., Flores and Herroro 1994). Studies by Harris et al. (2016) in reservoirs also suggest that lower nitrate:ammonium ratios and higher ammonium concentrations favor production of cyanobacterial secondary metabolites.

DIN may cycle rapidly through the water column of lakes and reservoirs, making it difficult to assess its availability based on an analysis of N pools and to determine if N fixation is likely to dominate. For example, a 1 yr study in a subtropical reservoir found that ammonium cycled through the water column every 0.3–5.8 h while nitrate cycled at a slower rate, 3.5–218 h (Burford et al. 2006). Most studies do not measure DIN concentrations at these time scales and more often monitor pools of N (i.e., concentrations) at time scales of weeks to months. Our rudimentary understanding of DIN dynamics and its uptake as DIN or atmospheric (di-nitrogen) sources could be greatly enhanced by aligning time scales of sampling frequency to N cycle rates among its constituent pools.

In summary, the role of N fixation in promoting dominance of particular cyanobacterial species should not be oversimplified. DIN is clearly an important source of N for N fixing and non-N fixing species, and a dynamic interplay exists between pools and fluxes of N within and between its organic and inorganic constituents within the water column and at air–water and sediment–water boundaries.

N and P availability and impacts on cyanobacterial physiology

Few studies have examined how fluctuations in N and P availability impact physiological processes affecting cyanobacterial growth and survival. Of the studies examining the causal links, most have inferred the physiological status of cells from N:P stoichiometric ratios. Klausmeier et al. (2004) showed that at low growth rates, algal stoichiometry matched the supply ratio while at high growth rates the ratio approached an optimal ratio that maximized growth. Alternating between DIN utilization and N fixation in N-fixing species also affects P requirements. For example, diazotrophic cyanobacteria require more P to fix atmospheric N under DIN limitation than under DIN replete conditions (Moisander et al. 2003, 2007). Burford et al. (2014) showed that for R. raciborskii, production of heterocysts (necessary for N fixation in this freshwater species) increased with additions of DIP and DIN + DIP but decreased with DIN addition alone. Thompson et al. (1994) showed that Anabaena flos-aquae (Anabaena subsequently renamed to Dolichospermum) stored P among different cellular P fractions depending on the available N source. When fixing N, P was stored as photosynthate (sugar with phosphate groups, as well as RNA and DNA), whereas when nitrate was supplied, acid-soluble P accounted for 15% of cellular P, and no sugar P was detected. These studies suggest P availability limiting N fixation. High N:P ratios may also increase the proportion of toxic versus non- or less-toxic blooms for some species (e.g., Microcystis; e.g. Vézie et al. 2002, Van De Waal et al. 2009). Conversely, a mesocosm study during a bloom of R. raciborskii showed that the proportion of toxic strains increased at lower N:P ratios (Burford et al. 2014).

The findings of a significant body of research on utilization of N and P by R. raciborskii can be summarized schematically (Fig. 2). Highest growth rates occur under N and P replete conditions, but removal of P does not necessarily reduce growth rates until limitation becomes extreme. Growth rates are also slower during periods of N fixation, when DIN is limiting, than when N and P conditions are replete. Relatively few other cyanobacterial species have been studied at this level of detail, but comparing and contrasting the range of strategies undertaken by different species would be interesting as well as provide model inputs for predicting blooms of individual species. Strains within populations can vary in terms of rates of P uptake, N fixation, and P storage capacity as well as a range of other processes (e.g., Willis et al. 2015, 2017, Xiao et al. 2020).

Figure 2. Transitions of physiological processes under different nutrient ratios and availability for Raphidiopsis raciborskii. DIN = dissolved inorganic nitrogen; DIP = dissolved inorganic phosphorus; DOP = dissolved organic phosphorus.

The rapid cycling of N and P in freshwater systems (and resulting fluctuations in N and P availability for cyanobacterial growth) is not currently captured in most culture experiments conducted in laboratories. Studies have overemphasized equilibrium states rarely seen in the natural environment. The complexity of responses observed under natural conditions is exemplified in lakes dominated by Planktothrix agardii, where natural fluctuations in DIP concentration regulated the P supply for the algal community, and hence the kinetic and energetic properties of the community (Aubriot et al. 2011). Both the history and pattern of DIP fluctuations resulted in rapid and varying physiological adaptations. Subsequent studies with blooms of Planktothrix and R. raciborskii showed that the growth response may be more dependent on the ability to optimize the uptake of DIP during the time span of nutrient fluctuations than on the amount of nutrient taken up per se (Aubriot and Bonilla 2012). Culture studies of R. raciborskii showed that the frequency of DIP pulses affected the growth rate of cultures (Amaral et al. 2014). Despite important insights from these studies, few studies of other lakes and species are available to broaden our understanding of the fluctuations between P availability and physiological responses of cyanobacteria.

These studies also highlight the need to undertake measurements of nutrient concentrations in the field with sufficient precision and frequency to determine N availability and form, matching the time scales of adjustment in the physiological responses of cyanobacteria. Håkanson et al. (2007) supported this assertion, stating it is generally pointless to base predictions of Chl-a and cyanobacteria on different “bioavailable” forms of the nutrients (DIN, DIP) because they do not predict cyanobacteria and chlorophyll well, and the coefficient of variation for dissolved inorganic nutrient concentrations is high. Therefore, higher frequency sampling is needed to obtain reliable empirical data necessary in useful models with high predictive power (e.g., Defew et al. 2013). Additionally, as discussed earlier, cyanobacterial strategies for utilization of DIN and DIP are complex and not directly linked to concentrations. For example, storage of P by cyanobacteria allowed growth to continue for a number of generations despite undetectable DIP concentrations is (Xiao et al. 2022).

Automated water samplers and in situ nutrient sensors are gradually being included in field sampling to measure ammonium, nitrate, and DIP concentrations (e.g., Wade et al. 2012, Mahmud et al. 2020), but methodological challenges remain, including insufficient sensitivity at low concentrations (Marcé et al. 2016), maintenance of sensors over longer-term deployments (Chen and Crossman 2021), and the capacity to measure a broader suite of bioavailable nutrients (e.g., urea). However, success of these approaches for measuring parameters such as chlorophyll indicate that sensors for nutrients are becoming more routine in the medium term (Brentrup et al. 2016).

Conclusions

Our current paradigms of nutrient availability for cyanobacteria need to better reflect the continuum of availability, sources, and fluxes as well as the range of physiological processes that allow continued nutrient acquisition and growth under nutrient limited conditions. Further work is required to better integrate cell quotas and stoichiometry into understanding nutrient acquisition, storage, and growth responses. The thresholds for factors such as nutrient sufficiency, as well as the rate and dominance of key processes, varies among species and strains within species; however, the physiological state of cyanobacteria can change rapidly in response to changing DIN and DIP concentrations. A simplified version of the linkages between nutrient availability, stoichiometry, and physiological responses (Fig. 3) shows that at a practical level, studies typically do not effectively encapsulate all these levels of responses, meaning that scaling up the findings spatially and temporally to be useful for predictive models remains problematic.

Figure 3. Linkages of low nutrient availability with cyanobacterial responses based on measures of nutrients, stoichiometry, and physiological processes. DIP = dissolved inorganic phosphorus; DIN = dissolved inorganic nitrogen; DOP = dissolved organic phosphorus.

As we develop new tools and approaches to studying cyanobacteria we learn more about their physiological responses to low and variable nutrient availability. One of the challenges is that generalizing across species and strains is not possible; therefore, better integration of field and laboratory studies is needed as well as contemporary automated methods of sampling and sensing in the field. This information is fundamental to improving prediction and management of cyanobacterial blooms.

Acknowledgements

MAB thanks the SIL Baldi award committee for the opportunity to present this paper. We also thank the large number of researchers who contributed to the research presented in this paper. Finally, we thank Jack Jones and Ingrid Chorus for helpful comments on the paper.

Disclosure statement

MAB is an Associate Editor and DPH Editor in Chief of Inland Waters. The review of this paper was conducted by an independent editor to avoid conflict of interest with the authors.

Additional information

Funding

This work was supported by Australian Research Council: [Grant Number DP190101848].