Seasonal life strategy of Prorocentrum minimum in Chesapeake Bay, USA: Validation of the role of physical transport using a coupled physical–biogeochemical–harmful algal bloom model

Abstract

Over 40 years ago, it was suggested that Prorocentrum minimum in Chesapeake Bay has a seasonal life strategy that depends on the physical transport by estuarine circulation, bringing cells from lower bay to the mid-bay in spring when they bloom. In this study, a validated hydrodynamic-biogeochemical model is used to simulate the annual cycles of P. minimum in Chesapeake Bay and track its life history over multiple years. The model reproduces the observed seasonal progression of P. minimum without a seed population. Four life stages of P. minimum are faithfully produced in the model: (1) in winter, overwintering populations from the previous bloom are distributed throughout the water column in the lower bay; (2) in late winter/early spring, cells are transported upstream by the landward bottom flows; (3) in May, P. minimum develops a bloom in the mid-bay; (4) in late summer/fall, decaying P. minimum populations are transported downstream by the seaward surface flows. Particle tracking shows that it takes about 3–4 months for the overwintering cells to travel from the lower bay to the mid-bay, but about 6 months for the decaying cells to travel from the mid-bay to the lower bay, as the estuarine circulation is far stronger during the high runoff months of January–May than during the low runoff months of June–December. With the peak growth rate around 20°C, May provides an optimal window of growth opportunity for P. minimum as phytoplankton assemblage transitions from winter–spring diatoms to summer dinoflagellates in a seasonal succession.

In a study published over 40 years ago, Tyler and Seliger (1978, TS78 hereafter) proposed a novel mechanism for the life cycle strategy of the red tide dinoflagellate Prorocentrum minimum in Chesapeake Bay. During January and February, they observed significant concentrations of P. minimum cells below the photic zone in the lower bay, at a distance of 200 km downstream of the region where blooms typically appear later in the year. Based on several bay-wide ship surveys (January, February, March, May, August, November–December) over a span of 2 years (Fig. 1), they hypothesized that P. minimum in surface outflowing waters at the mouth of the bay was recruited in late winter into more dense inflowing coastal waters and then transported northward to reach its bloom area in the upper part of the mid-bay by late spring. They further hypothesized that P. minimum was not introduced into the bay from surface waters of the Atlantic Ocean, but rather resulted from the decaying P. minimum bloom that was transported seaward by the estuarine outflow from the previous year's bloom. This hypothesis of TS78 is remarkable because it suggests a self-sustaining life strategy of an estuarine Harmful Algal Bloom (HAB) species that does not require a remote source of seed populations nor does it involve a life stage as cysts buried under the sea bed. It also begs for a reexamination in light of recent laboratory, field and modeling studies of Prorocentrum blooms and the recognition that planktonic Prorocentrum species have proliferated in estuarine and coastal waters worldwide over the past few decades, especially in relation to eutrophication (Heil et al. 2005; Glibert et al. 2008, 2012, 2015, 2020; Zhang et al. 2021).

Fig 1

(a) Map of Chesapeake Bay in which the dashed black line marks the along-channel section in the deep center channel, the filled squares mark the monitoring sites regularly sampled by EPA Chesapeake Bay Program and the black star represents the location used in Fig. 4. Color contours indicate depth. (b–f) Along-channel distributions of Prorocentrum minimum cell density from the field surveys reported in TS78. Note that contours in (b) and (f) have different units (10⁻² cells mL⁻¹ and cells mL⁻¹) from others (10⁻³ cells mL⁻¹). Figure reproduced from Limnol. Oceanog. with permission of the publisher.

Many HAB species have relatively modest growth rates when compared with other phytoplankton species. It is intriguing how slow-growing HAB species out-compete other species to develop blooms, and life cycle strategies have long been suspected of giving some HAB species a competitive advantage (Stolte and Garces 2006; Hense 2010; Azanza et al. 2018). For example, transitions between resting and vegetative phases in the dinoflagellates Alexandrium fundyense and Pyrodinium bahamense were found to be responsible for initiating or terminating blooms (e.g., Anderson 1998; Garcés et al. 2002; Anderson and Rengefors 2006). Resting cells from previous blooms settle on the sea bed, where they accumulate and form a so-called seed bank. Seed banks and blooms are not necessarily in the same geographic location due to transport of the different life cycle stages by ocean currents; offshore germinating cells may be advected onshore initiating a coastal bloom (e.g., McGillicuddy et al. 2003), while an offshore HAB may be generated by germinating cells originating at a coastal seed bank (Donaghay and Osborn 1997). Other HAB life cycle strategies include Pseudo-nitzschia diatom species undergoing sexual reproduction (Lelong et al. 2012; Montresor et al. 2016) and cyanobacteria forming akinetes as a resting stage (Huber 1984; Suikkanen et al. 2010). All these life cycle strategies involve biological processes, which stand in contrast to the above P. minimum life strategy that relies entirely on physical processes.

The TS78-reported bimonthly surveys of P. minimum and currents in Chesapeake Bay made a strong case for the physical transport mechanism as a life strategy, but a direct test of this mechanism has been lacking. Such a test requires tracking P. minimum cells over a complete life cycle as they move between the bloom region in the upper part of the mid-bay and overwintering site in the lower bay. Although chemical and genomic markers have been used to track marine species such as oyster larvae (Gancel et al. 2019; Houston et al. 2020), it has been impractical to carry out the detailed surveys of P. minimum over several years that would be necessary for such an empirical test. On the other hand, a well-validated numerical model can track movements of HAB cells in space and over time. In this paper, we apply a validated coupled hydrodynamic-biogeochemical model, Regional Ocean Modeling System (ROMS)–Row Column Aesop (RCA)–Prorocentrum (Zhang et al. 2021), to study the life cycle strategy of P. minimum in Chesapeake Bay. Specifically, we run the model in a “perpetual” mode by repeating annual forcing over a number of years thereby testing if the P. minimum life cycle is repeated and sustained every year. This approach has been widely used in climate modeling studies to examine the stability of a seasonal/annual cycle in a climate system under different climatic forcing conditions (e.g., Zwiers and Boer 1987; Brossier et al. 2011).

Methods

ROMS–RCA–Prorocentrum model

The ROMS–RCA–Prorocentrum model has three components (Zhang et al. 2021). The three-dimensional (3D) hydrodynamic model is based on the ROMS (Shchepetkin and McWilliams 2005; Haidvogel et al. 2008), and the biogeochemical model is based on the RCA structure (DiToro 2001; Isleib et al. 2007). The Prorocentrum model uses the rhomboid approach in which the individual HAB taxa are characterized against a background of other functional groups (Zhang et al. 2021).

The ROMS hydrodynamic model domain covers the Chesapeake Bay and its adjacent shelf and has been validated against a wide variety of observational data (M. Li et al. 2005, 2006; Zhong and Li 2006; Xie and Li 2018; Xie et al. 2018). The model has 80 × 120 grid points in the horizontal directions and 20 vertical levels. ROMS is forced by freshwater discharge at river heads, water levels at the open boundary, and heat and momentum flux across the sea surface. The freshwater input is prescribed for the eight major tributaries of Chesapeake Bay, based on measurements at U.S. Geological Survey gaging stations. At the offshore boundary, the tidal component is provided by TPXO7 (Egbert and Erofeeva 2002), and nontidal component is extracted from daily sea level measured at Duck, North Carolina, by National Oceanic and Atmospheric Administration. The air–sea heat flux and momentum flux are calculated using the North America Regional Reanalysis data.

The RCA biogeochemical model is coupled to the ROMS hydrodynamic model in an offline mode. Hourly averages of temperature, salinity, and transport terms from ROMS are used to drive the biogeochemical variables in RCA. The RCA has a water-column component (Isleib et al. 2007; Zhang and Li 2010) and a two-layer sediment diagenesis model (DiToro et al. 2001; Brady et al. 2013). The water-column model includes state variables representing dissolved inorganic nitrogen (DIN), dissolved inorganic phosphorus (DIP), and silicate (Si), particulate and dissolved organic N and P, and dissolved O₂. The phytoplankton module includes two generic algal assemblages and one species: one winter group (optimum temperature ~ 10°C), one summer group (optimum temperature ~ 25°C), and P. minimum. For the P. minimum model, the growth rate depends on temperature, light and nutrient concentrations, while the mortality terms include both grazing and respiration, and the model parameters have been calibrated according to published physiological experiments on P. minimum and numerical sensitivity analysis experiments (Zhang et al. 2021). At the heads of the tributaries, the nutrient concentration and two phytoplankton assemblages are prescribed based on Chesapeake Bay Program data and P. minimum concentration is set to zero as there is no evidence suggesting that external input is a significant source of P. minimum in the Bay. At the offshore boundary, nutrient concentrations on the shelf are acquired from the World Ocean Atlas and Filippino et al. (2011). ROMS–RCA was previously validated in several modeling studies (Testa et al. 2014; Li et al. 2016; Testa et al. 2017; Ni et al. 2020).

A detailed validation of ROMS–RCA–Prorocentrum in Chesapeake Bay was presented in Zhang et al. (2021), showing good skills in predicting DIN as nitrate plus nitrite and ammonium DIP as phosphate chlorophyll a, and P. minimum cell density at a number of monitoring stations. The validated model not only predicted the seasonal timing and location of P. minimum blooms but also captured the observed interannual variations in the magnitude and distributions of P. minimum blooms.

In the ROMS–RCA–Prorocentrum model (Zhang et al. 2021), the growth rate of P. minimum is written as

(1)

The specific growth rate is given by

(2)

where G_p is the maximum growth rate, T_opt is the optimal temperature for the maximum growth, and and are the shape factors characterizing the window of optimal growth. The effect of light (as photosynthetically active radiation, PAR) availability on P. minimum growth (G_par) is parameterized by

(3)

in which is the slope of the P–I curve (in unit of ly⁻¹). The effect of nitrogen limitation on P. minimum growth is parameterized by

(4)

and the effect of phosphorous limitation is parameterized by

(5)

where K_mn and K_mp are the half saturation constants for DIN and DIP, respectively. The net effect of nutrient limitation on P. minimum growth is given by

(6)

These terms are herein analyzed to discern mechanisms regulating the timing and location of P. minimum blooms.

Perpetual run configuration

To configure the perpetual model runs, we selected year 2007 for the annual forcing as the river discharge in that year was close to the long-term average. The ROMS model was initialized with climatological temperature and salinity and was run for a spin-up period of 3 years. The outputs of this spin-up run were used to set the initial conditions for the hydrodynamic model in the perpetual run. The nutrients and phytoplankton assemblages in RCA model were initialized on 01 January 2007 with observational data from Chesapeake Bay Program. Observations of P. minimum were limited during winter with only four data records in the main stem of the Bay. As such, the initial condition for P. minimum was constructed based on both the observation in January 2007 and the bay wide survey from TS78. The ROMS–RCA–Prorocentrum model was forced with the same annual forcing and ran for several years until repeating annual life cycles were established.

Particle transport model

To further clarify the physical transport mechanism, the Larval TRANSport Lagrangian model (LTRANS) was used to track the trajectories of neutrally buoyant particles (North et al. 2008). LTRANS simulates particle advection by the velocity fields and incorporates a random displacement model to simulate particle random walk due to turbulent diffusion. To test the first half of P. minimum life history, 12 particles were released in the model at a lower-bay location on 01 January and tracked until the end of June. To test the second half of P. minimum life history, 20 particles were released at a mid-bay location on 01 July and tracked until the end of December. Both sets of particles were initially placed at 1-m intervals from the surface to the bottom at their respective locations.

Results

Repeating annual cycles

The time series of surface water P. minimum cell concentrations at eight stations along the center axis of Chesapeake Bay show repeating annual cycles (Fig. 2). There are differences in cell density between the first 2 years as the model adjusts to the annual forcing and the influence of the initial condition fades. However, the P. minimum time series show repeating annual cycles at all the stations from years 3. The annual cycles show a pattern of low cell concentration between January and April, an initiation of a bloom in the beginning of May, reaching a peak in mid-May to late May, and then termination by mid-June to late June. After that, the cell concentration drops to low levels during the rest of the year. A late-spring P. minimum bloom is sustained every year, completely independent of the initial seed concentration.

Fig 2

(a–h) Time series of the model-predicted surface Prorocentrum minimum cell density at eight stations arrayed in the along-channel section of Chesapeake Bay (their locations marked in Fig. 1a).

Differences in the bloom magnitude were predicted by the model among the stations (Fig. 2). The predicted peak cell concentration is about 1 × 10⁶ cell L⁻¹ at the two upper bay stations (CB 2.2 and CB 3.1) as well as at the two lower bay stations (CB 5.2 and CB 6.1). In comparison, the predicted peak cell concentration reaches nearly 2 × 10⁶ cell L⁻¹ at the four mid-bay stations (CB 3.3C, CB 4.1C, CB 4.2C, CB 4.3C), indicating that the most intense P. minimum blooms develop in this mid-bay region.

Transport pathways

Next, the along-channel distributions of monthly mean P. minimum cell concentration were compared with the monthly mean residual circulation (Fig. 3). During January, most P. minimum cells are located on the shallow lower bay and are present at all depths, as strong turbulent mixing disperses cells throughout the weakly stratified water column (Fig. 3a). The estuarine return flows (~ 0.05–0.15 m s⁻¹) in the bottom layer transport the cells landward such that a plume of low cell concentration water spills into the deep mid-bay, in agreement with the cell distribution reported in TS78's January survey (compare Figs. 1b and 3a). In February, the cell density in the lower bay decreases but the plume of low cell concentration water penetrates further upstream (compare Fig. 3b with Fig. 1c). The P. minimum cell distribution in March covers a similar latitudinal extent as in February, but reaches higher in the water column (up to 10 m depth) as mixing diffuses the cells upwards (Li et al. 2005, 2016; Li and Zhong 2009; Li and Li 2011). By April, low cell concentrations appear throughout the water column.

Fig 3

Monthly-mean Prorocentrum minimum cell density in the along-channel section. The vectors are monthly-mean subtidal flows. The color bar is in logarithmic scale and the color scale for (g–l) is different from that for (a–f) in order to better show the location of P. minimum when the cell density is low.

A model-predicted P. minimum bloom develops in May, with the highest concentration in the surface waters of the mid-bay, at a distance of 150–250 km from the estuary's mouth (Fig. 3e). This was also shown clearly in TS78's field survey in May (Fig. 1d). This bloom weakens considerably in June and July (Figs. 3f,g). By August, the cells are mostly confined to a region about 250 km from the estuary's mouth (Fig. 3h), as shown in TS78's survey in August (Fig. 1e). Over the next few months, a plume of low P. minimum cell density is advected seaward by the surface outflow, with its front edge reaching 170–150 km from the mouth in September and October (Figs. 3i,j), and at the 120 km mark in November (Fig. 3k), and at the 70 km mark in the lower bay in December (Fig. 3l).

It is of note that the transport of P. minimum cells from the lower bay to the mid-bay (200–250 km) takes about 3–4 months (from January to May), but the transport of P. minimum cells from the upper part of the mid-bay to the lower bay takes about 6 months (from July to December). This difference can be explained by the seasonal difference in the estuarine circulation strength. The river flow is much higher (up to 6000 m³ s⁻¹) between January and May, but much lower during the summer and fall (down to 300–400 m³ s⁻¹) (Fig. 4a). Consequently, the residual estuarine circulation is much stronger during the winter–spring months, with the landward flow in the lower layer averaging ~ 0.1 m s⁻¹ at the mid-bay location that is 150 km from the bay's mouth (Fig. 4b). In comparison, the estuarine circulation is much weaker between July and December, with the seaward flow in the upper layer averaging to ~ 0.05 m s⁻¹.

Fig 4

(a) Susquehanna river discharge. (b) Surface-layer (1 m depth, blue) and bottom-layer (15 m depth, red) currents at 150 km from bay's mouth. Positive value indicates landward flow and negative value indicates seaward flow. Dashed red and blue lines are 10-day low passed velocity, and solid lines are monthly averaged velocity.

The particle tracking model, LTRANS, further demonstrated how the estuarine return flow transports overwintering populations to the mid-bay to fuel a bloom during winter–spring and how the estuarine outflow transports decaying cells to the lower bay during summer and fall (Fig. 5). For the test representing the first half of P. minimum life history, particles were released at a lower bay location on 01 January and tracked until the beginning of June (Figs. 5a–f). Some particles in the surface layer are exported out to the shelf while other particles are advected landward by the estuarine return flow, reaching 38°N by 01 March and 38.5°N by 01 May. The later location is the mid-bay region where P. minimum bloom typically occurs. For the test representing the second half of P. minimum life history, particles were also released at a mid-bay location on 01 July and tracked until the beginning of December (Figs. 5g-l). A few particles move upstream, other particles move progressively seaward. By November and December, some particles reach the lower bay and are located in the lower layer. The two particle tracking calculations confirm the landward transport pathway between January and May and the seaward transport pathway over the summer and fall.

Fig 5

(a–f) Locations of particles released in the bottom water of the lower bay on 01 January, showing how the estuarine return flow transports the particles to the mid-bay to fuel a bloom in May. (g–l) Locations of particles released in the surface water of the mid-bay on 01 July, showing how the estuarine outflow transports the particles to the lower bay during the summer and fall.

A window of growth opportunity for blooms

To understand why favorable conditions for P. minimum occur in the mid-bay in May, model-predicted surface distributions of DIN and DIP, PAR (at 2 m depth), P. minimum cell concentration, DIN limitation, DIP limitation, light limitation, and P. minimum growth rate were examined. High river flows during January and April deliver high concentrations of DIN and DIP to the upper and middle parts of the estuary (Figs. 6a,b) whereas at this time of year PAR is low in the upper bay due to incoming riverine sediment (Fig. 6c). The region between 38.5° and 39.2°N represents the overlap area where both nutrient concentrations and the light field are favorable for P. minimum growth.

Fig 6

Surface distributions of DIN (a), DIP (b), PAR at 2 m depth (c), Prorocentrum minimum cell concentration (d), DIN limitation (e), DIP limitation (f), light limitation (g), and P. minimum growth rate (h) in May.

DIN is not limiting P. minimum growth in most parts of Chesapeake Bay at this time of year, as approaches 1 (Fig. 6e). On the other hand, is in the range of 0.8–1 in the upper Bay but drops dramatically south of 38.5°N (Fig. 6f). Hence nutrient limitation on P. minimum growth is mainly determined by P limitation. Values of G_par are low north of 39°N but approach one south of this latitude (Fig. 6g). The actual growth rate of P. minimum during May is highest in the region between 37.8° and 39°N (Fig. 6h), which corresponds reasonably well to the region with highest cell density (Fig. 6d), although a precise correspondence is not expected as the biomass also depends on grazing and respiration and is affected by physical transport. Therefore, the P. minimum bloom develops in the mid-bay region due to the optimal light and nutrient conditions there.

The month of May also provides an optimal window of growth opportunity for P. minimum as water temperature in May matches the optimal temperature for P. minimum growth at ~ 20°C (Fig. 7a). In comparison, the winter–spring diatom group is parameterized in the model with optimum temperature for growth of ~ 10°C and the summer dinoflagellate group is parameterized in the model with an optimum temperature for growth of ~ 25°C. A comparison of the specific growth rates for the three phytoplankton groups shows the diatom's domination between December to mid-April and the summer assemblage's domination between June and October. The two windows of opportunity for P. minimum growth are (1) a late spring period from mid-April to end of May and (2) a late fall period from late October to end of November (Fig. 7b). The actual growth rates of these phytoplankton species also depend on nutrient concentration and light availability. At the mid-bay Sta. 4.1C, the diatom group reaches a peak growth rate of over 1 in March and April, P. minimum reaches a peak growth rate of about 1 in May, and the summer assemblage reaches a peak growth rate of about 2 in June to August (Fig. 7c). The phytoplankton biomass shows a diatom maximum of 1 during March and April, a P. minimum maximum of 0.5 in May, and a summer assemblage maximum of over 1 in summer and early fall (Fig. 7d). Therefore, due to its window of growth opportunity around ~ 20°C, P. minimum manages to develop a bloom in late spring as the phytoplankton seasonal succession transitions from the winter–spring diatom group to the summer dinoflagellate group.

Fig 7

Time series of (a) temperature, (b) specific growth rates, (c) growth rate, and (d) biomass of winter–spring diatom group (red), summer dinoflagellate group (green) and Prorocentrum minimum (blue) in the surface water of a mid-bay station CB 4.1C.

Sensitivity to model parameters

The above results were obtained from the model run (control run) using a set of parameter values determined according to published physiological experiments on P. minimum (summarized in Zhang et al. 2021). There are uncertainties in estimating some parameters such as the maximum growth rate G_p and the shape factors and characterizing the window of optimal growth. We have conducted four additional sensitivity analysis model runs with G_p, and increased or decreased by 20% (Fig. 8). Reducing G_p by 20% substantially suppresses the bloom (Fig. 8b). In contrast, increasing it by 20% enhances the peak bloom size by ~ 30% and also lengthens the bloom duration (Fig. 8c). When and decrease by 20%, P. minimum has a longer window of opportunity to grow and its bloom lasts longer (Fig. 8d). When and increase by 20%, the window of growth opportunity is shortened, resulting in a shorter bloom duration and a smaller bloom size (Fig. 8e). Overall, the seasonal progression of the P. minimum bloom in the sensitivity analysis model runs is similar to that shown in Fig. 3, despite that the bloom size and duration are sensitive to changes in these parameters.

Fig 8

Time series of Prorocentrum minimum cell density at CB4.1C obtained from the control model run (a) and the sensitivity analysis model runs in which the maximum growth rate G_p of P. minimum and the shape parameters and characterizing the window of optimal growth increase or decrease by 20% (b–e).

Discussion

The perpetual model simulations using a 3D-coupled hydrodynamic-biogeochemical model has confirmed the life strategy of P. minimum proposed by TS78. In summary, P. minimum has four life stages: (1) in winter, overwintering populations from the previous bloom are mixed throughout the water column in the lower bay due to strong turbulent mixing; (2) in late winter/early spring, cells are transported upstream by the landward bottom flows with a travel time of about 3–4 months since the estuarine circulation is strong during the high runoff months of January to April; (3) in May, P. minimum develops a bloom in the upper part of the mid-bay due to optimal growth conditions there; and (4) in late summer/fall, decaying P. minimum populations are transported downstream by the seaward surface flows from the mid-bay to the lower bay by the estuarine outflow in the surface layer, taking about 6 months since the estuarine circulation is much weaker during the low runoff seasons of summer and fall (Fig. 9). During this annual cycle P. minimum exhibits two distinct phases in terms of growth: a “rapid growth phase” between late April and early July when cells are actively growing and a “slow growth phase” between August and April in the follow year when post-bloom cells first move towards the mouth of the Bay and then return in the bottom layer (Figs. 3, 7). During the latter period, both the growth rate and mortality rate are very low. This study has demonstrated the feasibility of a unique self-sustaining life strategy of a HAB species that relies entirely on the physical transport mechanism. It does not require a remote source of seed populations nor involves a life stage as cysts buried under the sea bed.

Fig 9

Schematic diagram of the life strategy of Prorocentrum minimum.

It is also interesting to note that P. minimum takes advantage of an optimal growth window in May as the phytoplankton assemblage in Chesapeake Bay makes a transition from the March–April spring bloom of diatoms to the fast-growing summer assemblage in a seasonal succession. P. minimum typically only constitutes 20%–30% of the total phytoplankton biomass in Chesapeake Bay (Adolf et al. 2006). Moreover, its specific growth rate is lower than that of both the winter–spring diatoms and the summer dinoflagellates (Testa et al. 2014). It appears that P. minimum squeezes a bloom in between the blooms of the two dominant phytoplankton groups because its optimal temperature for growth (~ 20°C) is greater than that of the winter–spring diatoms, but less than that of the summer assemblage. The location of the bloom in the upper part of the mid-bay points to P. minimum's exploitation of optimal conditions of nutrient concentration and light field.

In a subsequent paper, Tyler and Seliger (1981) found that the growth rate of P. minimum depends on both temperature and salinity. In particular, the growth rate increased with salinity in low temperature waters, thus restricting the over-wintering populations to the high-salinity lower bay. This salinity-enhanced growth rate appears to be at odds with the laboratory experiments of Grzebyk and Berland (1996) which showed a moderately higher growth rate in an intermediate range of salinities. Tyler and Seliger (1981) also suggested an optimal temperature growth of P. minimum around ~ 25°C, but recent field observations of P. minimum in Chesapeake Bay clearly showed highest bloom density at a temperature range between 13 and 25°C (Tango et al. 2005). Zhang et al. (2021) compared the model simulations with or without salinity dependence in the specific growth of P. minimum but found little differences in the model results. We also conducted a perpetual model run with the salinity dependence and found a similar result. Therefore, the physical transport mechanism does not require a salinity-enhanced growth rate to sustain the overwintering populations in the high-salinity lower bay.

Sensitivity analysis model showed that the four seasonal life stages of P. minimum remain the same, although the bloom size and bloom duration are sensitive to changes in the maximum growth rate and the shape parameters characterizing the optimal window of growth. Zhang et al. (2021) reported additional sensitivity analysis model and found that the model-predicted bloom size was particularly sensitive to the half saturation constant for phosphorous uptake. Despite the uncertainty in determining these physiological parameters from laboratory experiments and the model's sensitivity to these parameters, it should be noted that the model was able to capture the large interannual variability of the P. minimum blooms in a decadal hindcast simulation using the same set of parameter values (Zhang et al. 2021).

In all, the insight from over 40 years ago, based on multiyear bay-wide field surveys (TS78), has been confirmed using a contemporary 3D-coupled hydrodynamic-biogeochemical model parameterized for P. minimum. The model was run using average flow conditions and captured the peak in bloom development in May in the mid-to-upper parts of the estuary. The model also showed the potential window of opportunity for a fall bloom, which did not develop under average conditions, but with shifts in flows, storms, and temperature, such blooms may occur and are, in fact, observed in some years (Tango et al. 2005; Li et al. 2015, 2020). For example, a fall bloom developed in late fall in 2006, as storms injected nutrients and cells into the surface euphotic layer, leading to a second peak in the growth rate of P. minimum (Zhang et al. 2021).

Globally, P. minimum blooms are expanding in coastal and estuarine waters, and their association with increasing eutrophication has been documented (Glibert et al. 2008, 2012 and references therein). As climate changes, and associated increasing temperatures, altered stratification and density gradients, as well as altered propensity and intensity of precipitation events, the window of opportunity for P. minimum may change in Chesapeake Bay as well as elsewhere (Li et al. 2020). Springs in Chesapeake Bay are expected to become wetter, and it is projected that this will increase N loads, even in the absence of increases in land-based applications; an increase in N flux down the Susquehanna River (the major tributary of Chesapeake Bay) of 17% by 2030 and 65% by 2095 is expected from flow changes alone (Howarth 2008). Based on climate downscaling models for the Chesapeake Bay region, projected for the years 2041–2070, DIN loads will not only increase in spring, but DIN : DIP will also increase substantially, increasing the potential habitat for species such as P. minimum (Glibert 2020; Li et al. 2020). Understanding the role of physical transport in conjunction with habitat changes through modeling gives new opportunities to explore the vulnerability for blooms in the future, not only in the bay, but in coastal and estuarine regions worldwide where physical models are available.

Acknowledgment

We thank the two reviewers for their helpful comments. This work is funded by the National Oceanic and Atmospheric Administration National Centers for Coastal Ocean Science Competitive Research Program under award NA17NOS4780180 to UMCES. This is ECOHAB contribution number ECO996 and contribution number 6032 from the University of Maryland Center for Environmental Science. Model output is available at https://doi.org/10.5281/zenodo.3525203.

Conflict of Interest

None declared.