Predicting the effects of climate change on freshwater cyanobacterial blooms requires consideration of the complete cyanobacterial life cycle

Abstract

To date, most research on cyanobacterial blooms in freshwater lakes has focused on the pelagic life stage. However, examining the complete cyanobacterial life cycle—including benthic life stages—may be needed to accurately predict future bloom dynamics. The current expectation, derived from the pelagic life stage, is that blooms will continue to increase due to the warmer temperatures and stronger stratification associated with climate change. However, stratification and mixing have contrasting effects on different life stages: while pelagic cyanobacteria benefit from strong stratification and are adversely affected by mixing, benthic stages can benefit from increased mixing. The net effects of these potentially counteracting processes are not yet known, since most aquatic ecosystem models do not incorporate benthic stages and few empirical studies have tracked the complete life cycle over multiple years. Moreover, for many regions, climate models project both stronger stratification and increased storm-induced mixing in the coming decades; the net effects of those physical processes, even on the pelagic life stage, are not yet understood. We therefore recommend an integrated research agenda to study the dual effects of stratification and mixing on the complete cyanobacterial life cycle—both benthic and pelagic stages—using models, field observations and experiments.

INTRODUCTION

Surface aggregations of cyanobacteria (blooms) are increasing in many freshwater systems worldwide, threatening ecosystem services fundamental to society (Paerl & Huisman, 2008; Taranu et al., 2015; Ho et al., 2019). To date, most studies of freshwater cyanobacteria have focused on this conspicuous stage of their life history. However, the pelagic focus overlooks the fact that in temperate and boreal lakes, cyanobacteria are generally not present in the water column year-round (Fig. 1). Instead, a large part of the life cycle is spent in the benthos, on or near the sediment (Fryxell, 1983; Reynolds, 2006; Poulickova et al., 2008; Kaplan-Levy et al., 2010). Subsequent recruitment from the benthos to the open water pelagic zone inoculates blooms (Reynolds, 1972; Hansson et al., 1994; Perakis et al., 1996; Brunberg & Blomqvist, 2003; Padisak, 2003; Stahl-Delbanco et al., 2003; Kravchuk et al., 2006; Torres & Adámek, 2013). Similar benthic-pelagic coupling is also implicated in marine and estuarine harmful algal blooms, including red tides (e.g. Boero et al., 1996; Steidinger, 2010).

Although previous papers have suggested that knowledge of cyanobacterial life cycles is required for effective bloom prediction and management (e.g. Hellweger et al., 2008; Hense & Beckmann, 2010; Suikkanen et al., 2010), we still do not have a good understanding of the environmental conditions in which recruitment from benthic life stages is critical to pelagic population dynamics. When quantified, cyanobacterial recruitment on average contributes <1 to 2% of the pelagic population (reviewed by Tan, 2012)—yet can sometimes account for up to 60% of the pelagic population (reviewed by Carey et al., 2014). Further, eliminating recruitment could result in 50% smaller blooms (Verspagen et al., 2005). A predictive understanding of when, where, and why benthic life stages matter to bloom formation could lead to novel management strategies to diminish benthic survival or recruitment, as has been explored by some researchers (e.g. Visser et al., 1996; Baker, 1999; Tsujimura, 2004; Verspagen et al., 2006; Tan, 2012; Jia et al., 2014; Chen et al., 2016a; Visser et al., 2016; Wu et al., 2017).

In this Horizons article, we urge researchers to consider the complete cyanobacterial life cycle to meet the pressing challenges of predicting and managing cyanobacterial blooms against a backdrop of ongoing global change. First, we review what is known about cyanobacterial life cycles, including how lake thermal stratification and its converse, mixing, have contrasting impacts on benthic versus pelagic life stages. We then explore how these differential responses might alter the current expectation that climate change will continue to increase cyanobacterial blooms (e.g. Paerl & Huisman, 2008) and conclude by proposing a research agenda to advance our understanding of cyanobacterial life history and when it matters to bloom prediction and forecasting. Although our focus is primarily on temperate and boreal lakes that exhibit summer thermal stratification, we also consider waterbodies of other mixing regimes and climates in the proposed research agenda.

Fig. 1

Schematic for a generalized cyanobacterial life cycle; details vary among taxa.

QUICK PRIMER ON CYANOBACTERIAL BENTHIC-PELAGIC COUPLING

Cyanobacterial life cycles are complex, with key ecological constraints in each habitat impacting the transitions between habitats (Fig. 1). Cyanobacterial population growth within the water column is generally enhanced by warm temperatures, high light and nutrient availability, and thermal stability (e.g. Paerl, 1988; Huisman & Hulot, 2005; Reynolds, 2006; Jöhnk et al., 2008; Wagner & Adrian, 2009).

When pelagic conditions for cyanobacteria deteriorate, e.g. due to cooling water temperatures in the autumn in temperate and boreal environments, many cyanobacterial taxa actively exit the water column (Fig. 1). Their transition to the benthos happens via one of two mechanisms. First, in true dormancy, cyanobacteria produce specialized cells called akinetes while still in the water column; the colonies (with akinetes) then senesce and sink to the sediment, forming a “seed bank” (Nichols & Adams, 1982; Gyllstrom & Hansson, 2004; Suikkanen et al., 2010). Cyanobacterial genera with true dormancy include Aphanizomenon, Cylindrospermopsis, Dolichospermum and Gloeotrichia (Adams & Duggan, 1999; Karlsson-Elfgren & Brunberg, 2004; Reynolds, 2006; Rücker et al., 2009; Wood et al., 2009; Kaplan-Levy et al., 2010; Kovacs et al., 2012). Akinetes are highly resistant to environmental stressors, including desiccation, and can either germinate immediately (e.g. Rother & Fay, 1977; Lynch & Shapiro, 1981; Cmiech et al., 1984) or persist for months to decades (Livingstone & Jaworski, 1980; Rasanen et al., 2006; Wood et al., 2009), providing a “storage effect” against adverse conditions (Warner & Chesson, 1985; Caceres, 1997) and facilitating invasion into new lakes (e.g. Padisak, 2003; Rücker et al., 2009; Ramm et al., 2017).

Alternatively, instead of producing akinetes, taxa such as Microcystis and Planktothrix lower their metabolic activity and sink to overwinter on or near the sediments as (mostly) inactive vegetative cells (Fallon & Brock, 1981; Reynolds et al., 1981; Tsujimura et al., 2000; Brunberg & Blomqvist, 2002; Poulickova et al., 2004; Ihle et al., 2005; Sabart et al., 2015; Wang et al., 2018). In some environments, few of these inactive vegetative cells persist through the winter due to mortality, burial and other loss processes (e.g. Baker, 1999; Wang et al., 2018). In others, cells can persist for years (although not decades) before returning to the water column (Caceres & Reynolds, 1984; Bostrom et al., 1989; Brunberg, 1995; Brunberg & Blomqvist, 2003).

Return of benthic life stages to the water column occurs both passively and via buoyancy regulation with gas vesicles (Fig. 1). Passive recruitment occurs when physical processes or bioturbation resuspend benthic akinetes or inactive vegetative colonies, which then become active in the water column (Stahl-Delbanco & Hansson, 2002; Verspagen et al., 2004; Yamamoto, 2010; Gu, 2012; Karlson et al., 2012; Chen et al., 2016b). By contrast, favorable environmental conditions in the benthos can trigger akinete germination or the resumption of metabolic activity by inactive benthic stages. The newly active cyanobacteria then enter the water column after photosynthetic rates are sufficient to promote gas vesicle production, enabling buoyancy (Preston et al., 1980; Trimbee & Harris, 1984b; Karlsson-Elfgren et al., 2003; Carey et al., 2008). Importantly, the conditions triggering the in situ transition from the inactive benthic stage to the active stage are incompletely understood, in part because most studies examine only a few environmental drivers for limited time periods. Moreover, when the same drivers have been studied using both observational and experimental approaches, responses differ across taxa and lakes (Tables S1 and S2). In addition, our understanding of spatial and temporal heterogeneity in recruitment patterns within and among lakes remains limited.

DIFFERENT LIFE STAGES HAVE CONTRASTING RESPONSES TO STRATIFICATION AND MIXING

Regardless of these gaps in current understanding, work to date strongly suggests that cyanobacterial benthic and pelagic stages can respond very differently to the same environmental drivers due to the inherent differences in their habitat and life strategy. For well-studied taxa such as Aphanizomenon, Dolichospermum and Microcystis, cyanobacterial sensitivity to lake stratification and mixing depends on the life stage and whether the mixing is occurring in the water column or at the sediment–water interface.

In the pelagic life stage, the general expectation is that thermal stratification is “good” and mixing is “bad” for cyanobacterial growth and reproduction, especially in deep lakes (M₁ in Fig. 2). Stratification gives buoyant cyanobacteria a competitive advantage (Walsby, 1994), as their ability to regulate their vertical position also enables them to access nutrients elsewhere in the water column (reviewed by Cottingham et al., 2015) and shade out competitors (Carey et al., 2012). Conversely, mixing events break up surface aggregations of cyanobacteria (reviewed in Zhao et al., 2017; Xiao et al., 2018). Mixing events strong enough to disrupt stratification can cause premature loss of active cyanobacteria from the water column because the greater pressures at depth cause gas vesicles to collapse or they transport cells/colonies below the compensation depth (e.g. Walsby, 1994; Visser et al., 1996; Huisman et al., 2004). After gas vesicle collapse, cells are no longer able to control their buoyancy and sink to the benthos (Oliver & Walsby, 1984; Kinsman et al., 1991). Even if cells do not immediately senesce, light limitation and colder temperatures at depth usually result in losses that exceed reproduction (Huisman & Hulot, 2005). These mechanisms explain why increased thermal stability driven by climatic warming is implicated as a leading cause of increasing cyanobacterial blooms (e.g. Jöhnk et al., 2008; Wagner & Adrian, 2011; Carey et al., 2012)—as well as why some lake managers deploy epilimnetic mixing systems to control cyanobacteria (reviewed by Visser et al., 2016; Xiao et al., 2018; Lofton et al., 2019).

Fig. 2

Graphical depiction of the core hypothesized mechanisms: mixing events that disrupt stratification have an adverse effect on pelagic cyanobacteria (M₁), while mixing in shallow water stimulates recruitment from benthic life stages back into the water column (M₂). Both mechanisms are at work in deep lakes, whereas only M₂ would operate in shallow lakes. The net effect of these opposing processes on pelagic population size is not yet known, but is expected to differ among cyanobacterial taxa as a function of lake type, location, and time of year, as explained in the text. The dotted line represents the thermocline.

However, mixing can have positive impacts on benthic cyanobacterial life stages (M₂ in Fig. 2). For example, mixing at the sediment–water interface triggers the return of benthic cyanobacteria to the water column either passively via physical resuspension by sediment mixing (Reynolds et al., 1981; Thomas & Walsby, 1986; MacIntyre & Melack, 1995; Verspagen et al., 2004; Misson et al., 2011; Sejnohova & Marsalek, 2012) or ebullition (Delwiche et al., 2020), or by stimulating germination and active recruitment (Stahl-Delbanco et al., 2003; Karlsson-Elfgren et al., 2004; Rengefors et al., 2004; Misson & Latour, 2012). We reviewed the literature on the environmental factors determining reactivation, germination and recruitment (Tables S1 and S2) and found reports demonstrating positive effects of mixing on recruitment for multiple genera, including Aphanizomenon (Trimbee & Harris, 1984a; Hansson et al., 1994), Dolichospermum (Reynolds, 1972; Rengefors et al., 2004; Bertos-Fortis et al., 2016), Microcystis (Reynolds et al., 1981; Stahl-Delbanco et al., 2003; Verspagen et al., 2004; Ihle et al., 2005; Verspagen et al., 2005; Misson et al., 2011; Misson & Latour, 2012; Su et al., 2016), Nodularia (Bertos-Fortis et al., 2016) and Gloeotrichia (Forsell & Pettersson, 1995; Karlsson-Elfgren et al., 2003,, 2004,, 2005; Rengefors et al., 2004; Carey et al., 2014). Moreover, macrophyte presence, which decreases mixing in shallow waters (e.g. Gebrehiwot et al., 2017), is associated with lower phytoplankton recruitment (reviewed by Villena & Romo, 2007). If mixing indeed facilitates the transition from benthic to pelagic habitats—thereby stimulating blooms—it may explain the equivocal success of bloom management via epilimnetic mixing systems (Visser et al., 2016; Lofton et al., 2019).

Based on M₁ and M₂ (Fig. 2), we predict that cyanobacterial taxa that depend strongly on both recruitment and gas vesicles for bloom formation will be positively affected by mixing at the sediment–water interface in both shallow and deep lakes, but negatively affected (at least in the short-term) by the disruption of summer stratification by water column mixing. However, the net consequences of mixing events for cyanobacterial populations are difficult to predict, especially given different dormancy strategies, physiological traits that affect success in the water column, lake morphometry, stochasticity in the timing of mixing events relative to plankton phenology, and other factors. As such, a careful consideration of all of these effects, and the time scales over which they manifest (e.g. Wilson et al., 2016), is needed to make accurate predictions of how cyanobacteria may change in the future.

INTERSECTIONS OF CYANOBACTERIAL LIFE HISTORY AND CLIMATE CHANGE

The effects of climate change on stratification and mixing (e.g. MacKay et al., 2009; Woolway et al., 2019) are already having, and will continue to have, profound influences on aquatic biota in temperate and boreal lakes, including cyanobacteria (Stockwell et al., 2020). The timing, duration and strength of thermal stratification are changing in lakes and reservoirs globally due to increasing air temperatures (Lehman, 2002; O'Reilly et al., 2003; O'Reilly et al., 2015; Woolway & Merchant, 2019). For pelagic cyanobacterial populations that are “seeded” by mixing events that stimulate recruitment, changes to the temporal dynamics of stratification could have important consequences for summer blooms, especially in deep lakes. However, this possibility has not yet been fully explored.

Moreover, the predicted changes in stratification are not unidirectional, as many global climate change scenarios also predict increased high-intensity storms in both mid-latitude and tropical regions (Christensen, 2007; Hayhoe et al., 2007; Field et al., 2012; Havens et al., 2016; Prein et al., 2017). To date, however, the potential consequences of increased storms have not been investigated as extensively as changes in water temperature and stratification (but see Stockwell et al., 2020). In deep stratified lakes, more storms will increase episodic water column mixing (see e.g. Jennings et al., 2012; Klug et al., 2012; De Eyto et al., 2016; Woolway et al., 2018) during otherwise thermally-stable summers. While this mixing may disrupt pelagic cyanobacterial populations in the short term via M₁ in Fig. 2, it may also increase recruitment via M₂ or stimulate pelagic cyanobacteria over days to weeks due to increased nutrients from runoff, resuspended sediments or entrained hypolimnetic water. By contrast, in polymictic lakes, the positive effects of storms on recruitment or nutrient availability (e.g. Zhu et al., 2014; Havens et al., 2016) could be offset by increased hydraulic flushing rates that remove pelagic cyanobacteria, particularly in systems with short residence times (Havens et al., 2016; Richardson et al., 2018, 2019).

Thus, the interplay between two key physical aspects of climate change—stronger thermal stratification and increased intense storms—is likely to have both negative and positive effects on cyanobacteria, with effects that differ for benthic and pelagic life stages (e.g. Hense & Beckmann, 2006; Jager & Diehl, 2014) and among lakes. Notably, although the impacts of different aspects of climate change are beginning to be addressed for pelagic cyanobacteria in temperate systems (e.g. Taranu et al., 2012; Rigosi et al., 2014; Richardson et al., 2018, 2019), none of these studies has considered benthic life stages or recruitment. To develop robust predictions of how cyanobacteria respond to the physical effects of climate change, we need to better understand how these drivers interact with cyanobacterial traits (sensu  Litchman & Klausmeier, 2008; Kruk et al., 2010). For example, differences in dormancy strategies (i.e. akinetes vs. inactive vegetative cells, as described above), accessory photosynthetic pigments (e.g. Glibert, 2016), buoyancy (e.g. floating velocity, Xiao et al., 2018) and sensitivity to increased temperature (e.g. Lurling et al., 2013) may modulate responses to different aspects of climate change. Further, lake-specific characteristics such as morphometry, watershed characteristics and water chemistry may amplify or diminish climate change impacts (e.g. Richardson et al., 2018). For example, cooler overnight air temperatures increase convective mixing in the littoral zone, potentially stimulating both nutrient cycling and passive recruitment (MacIntyre & Melack, 1995), but the impact of this mechanism likely varies with the proportion of a lake comprising littoral habitat and degree of seasonal change in air temperature.

RESEARCH AGENDA

A comprehensive research agenda is required to advance our understanding of how stratification and mixing impact the complete cyanobacterial life cycle. Some of this work is already underway, yet we need concerted efforts to integrate across taxa, lakes and geographic regions to identify emergent trends that may not be evident within any one system (Burford et al., 2020). Specific research needs include:

Simulation models. Simulation models that capture the dynamic impacts of physical and chemical processes on both benthic and pelagic life stages are needed to answer two key questions: (1) What are the net effects of stratification and mixing on cyanobacterial populations, especially bloom formation? and (2) Under a suite of realistic climate and management scenarios, will cyanobacterial blooms increase or decrease? Understanding how diverse taxa and lakes might be impacted by different scenarios can only be achieved through extensive studies using models with different levels of complexity to see whether findings are consistent across modeling approaches and lakes (Sommer et al., 2012; Hipsey et al., 2015). Some individual-based cyanobacterial population models have included benthic stages (e.g. Hense & Beckmann, 2006; Hellweger et al., 2008) and the effects of mixing on pelagic phytoplankton have been explored previously (e.g. Huisman et al., 2004; Jöhnk et al., 2008; Blottière et al., 2014; Zhao et al., 2017). However, to our knowledge, no lake ecosystem simulation model incorporates recruitment from benthic life stages as a contributor to pelagic populations. Addition of recruitment to simulation models such as PROTECH (Elliott et al., 1999; Gray et al., 2019) and the General Lake Model coupled with the Aquatic EcoDynamics modules (GLM-AED2, Hipsey et al., 2019) would allow for wide-scale exploration of potential scenarios of the strength of stratification; the type, frequency, intensity and duration of mixing; interactions with other environmental drivers; and lake management. The mixing scenarios should be informed by both climate change predictions, specifically those related to temperature, wind and the frequency and intensity of storm events, as well as managers’ needs with respect to anticipated water uses. All models should be parameterized from observations and experiments conducted across a wide range of lakes, as described below, and specifically include core environmental drivers likely to determine growth and survival during each life stage (e.g. light, temperature and mixing; Tables S1 and S2).

Field observations. Field observations of how mixing events of varying magnitude and duration impact both the pelagic and the benthic life stage of important bloom-forming taxa in a broad array of lakes are necessary to parameterize the models for different taxa and lakes. In particular, observational data on recruitment for more taxa over multiple years, especially in tropical lakes and polymictic lakes, are urgently needed to better understand the effects of mixing on cyanobacterial life histories. Simultaneous collection of data on interacting environmental drivers—including temperature, light, mixing, nutrients and dissolved oxygen—allows for correlative identification of potential drivers of pelagic population dynamics (e.g. Zhu et al., 2014; Yang et al., 2016) and recruitment (e.g. Carey et al., 2014). To date, however, most cyanobacterial recruitment studies have been conducted over just one or two summer stratified seasons in temperate regions (Carey et al., 2014); studies that run year-round or across many years, and in boreal or tropical regions, remain rare. In particular, cyanobacteria in tropical regions pose major management concerns, yet much less is known about the drivers of tropical blooms (reviewed by Mowe et al., 2015). Consequently, it remains unknown how mixing events may affect tropical cyanobacterial life histories. Newer technologies such as FlowCAMs coupled with taxonomic identification via machine learning may expedite the tedious work of identification via microscopy (e.g. Thomas et al., 2018; Hrycik et al., 2019), though to our knowledge such approaches have not yet been used in recruitment studies.

Experiments. Because of the difficulty of identifying causal drivers from observational data, laboratory or field mesocosm experiments that manipulate the intensity and duration of mixing can be used to isolate the effects of mixing on both benthic and pelagic life stages (see, for example, the studies of mixing in Table S2). As with the other research approaches, experiments will need to be conducted for multiple taxa, in lakes with different morphometries and seasonal regimes. Field mesocosm experiments have the benefit of generally providing more realistic conditions than smaller scale lab culture experiments (e.g. Wang et al., 2018, but see Park et al., 2018), though field studies can be logistically challenging and lose realism after longer durations of time (Burford et al., 2020). In some lakes, it may be possible to manipulate stratification and mixing using engineered systems, enabling the testing of model predictions at the whole-lake scale (Jungo et al., 2001; Cantin et al., 2011; Read et al., 2011; Chen et al., 2018; Lofton et al., 2019).

CONCLUSION

Bloom-forming freshwater cyanobacteria sequentially occupy the benthic and pelagic zones in temperate and boreal lakes, but typically only the pelagic life stage is studied. Because stratification and mixing can have opposing effects on the benthic and pelagic life stages of cyanobacteria, a more complete understanding of all stages of the cyanobacterial life cycle will enable plankton researchers to better predict how ongoing climate change will affect the frequency, intensity and duration of cyanobacterial blooms.

AUTHOR CONTRIBUTIONS

The core ideas in this manuscript evolved over many years of collaboration among all co-authors. C.C.C. and K.L.C. outlined the paper, performed the literature review for Tables S1 and S2, and wrote the first draft; all authors edited and approved the submitted version.

DATA ARCHIVING

Not applicable.

ACKNOWLEDGEMENTS

We thank Bea Beisner and Ian Sherman for catalyzing our team to prepare these ideas as a Horizons paper; proposal reviewers and talk attendees who pushed us to better articulate our thinking; Mary Lofton for helpful discussions; and our two anonymous reviewers for additional suggestions for the research agenda. Our approach to studying cyanobacteria is informed by ongoing dialog with managers and lake associations, especially the Lake Sunapee Protective Association, Auburn Water District, Lewiston Water Division and Western Virginia Water Authority.

FUNDING

This work was supported in part by the National Science Foundation [grant numbers DEB-0749022, EF-0842267, EF-0842112, EF-0842125, ICER-1517823, CNS-1737424, DEB-1926050; DBI-1933016, and OIA 1923004]; the Auburn Water District and Lewiston Water Division; and internal Dartmouth College funding, including a Senior Faculty Fellowship, to K.L.C. Initial drafts of the manuscript were prepared while K.L.C. was serving as a rotating program officer at the National Science Foundation.