1. Biological Oceanography – 16
Abandoning Sverdrup’s Critical Depth Hypothesis on phytoplankton blooms?
Gitai Yahel School of Marine Science, Ruppin Academic Centre
Partly After M J. Behrenfeld (Ecology 91, 2010) 80 70 60 50 40 30 20 10 70 70 60 60 50 50 40 40 30 30
June Chlorophyll (mg m-3)
Tel.(052) , Skype gitaiyahel, Web

2. Biological Oceanography - 15
Abandoning Sverdrup’s Critical Depth Hypothesis on phytoplankton blooms?
Gitai Yahel School of Marine Science, Ruppin Academic Centre
Partly After M J. Behrenfeld (Ecology 91, 2010)
A long and detailed review of the subject
Behrenfeld, M. J., & Boss, E. S. (2018. Global Change Biology, 24, 55–77.
Tel.(052) , Skype gitaiyahel, Web

3. Net phytoplankton growth occurs when area ABCD < AEC
1953
Scholar citation count: 1624 (2015)
FIGURE 3 “Digital age” reproduction of figure 2 of Sverdrup (1953). Top half of figure shows phytoplankton (green bars) and zooplankton
(red bars) abundances observed between March and May 1949 at Weather Station “M.” Bottom half of figure shows mixed layer depths
(vertical black lines) and the range of Sverdrup’s critical depth estimates (hatched area) for diffuse attenuation coefficients (k) of 0.10 and
0.075 m1. Using these values for k and assuming constant cloudiness values following Sverdrup’s scale, our recalculated critical depth values
are shown by the solid red (k = m1; cloudiness = 7.5) and blue lines (k = 0.10 m1; cloudiness = 7.5) and dashed red (k = m1;
cloudiness = 6.0) and blue lines (k = 0.10 m1; cloudiness = 8.5). For these calculations, we assumed a constant respiration rate of
0.14 g cal cm2 hr1, which is the average of two values given by Sverdrup for the studies of Jenkins (1937; 0.13 g cal cm2 hr1) and
Pettersson, H€oglund, and Landberg (1934; 0.15 g cal cm2 hr1)
Net phytoplankton growth occurs when area ABCD < AEC
Biological Oceanography, (3)
Wednesday, September 04, 2019 3

4. R = phytoplankton respiration + grazing + sinking + all other losses
Sverdrup’s 1953 paper was a formalization of the ‘critical depth’ concept originally proposed by Gran and Braarud in the 1930’* and later developed by Riley, Atkins, and others.
The critical depth hypothesis attempts to predict if a bloom can occure, not what controls the magnitude of a bloom
A bloom is an increase in biomass, not photosynthetic rate, but there is no clear definition of a bloom
The hypothesis states that a bloom begins when the mixed layer shoals to a depth above the critical depth horizon where production (P) > respiration (R)
R = phytoplankton respiration + grazing + sinking + all other losses
R is assumed constant (that is, variation in top-down processes were not really considered)
Inverse of Sverdrup: prior to crossing the critical depth criterion, net growth is negligible or negative R P
*e.g., Gran & Braarud J. Biol. Board Can. 1 (5),
Biological Oceanography, (4)
Wednesday, September 04, 2019

5. Net growth can be independent of gross production under heavy grazing
Reservations
From Sverdrup:
Net growth can be independent of gross production under heavy grazing
the ‘bloom’ observed 2 days after “the depth of the mixed layer was for the first time smaller than the critical depth” likely reflected advection not rapid local growth
The first increase in biomass occurred before stratification
“It is therefore not advisable to place too great emphasis on the agreement between theory and [the Weather Ship ‘M’] observations”
Behrenfeld1 and Boss Ann. Rev. 2014
Biological Oceanography, (5)
Wednesday, September 04, 2019

6. Earlier reservations:“ *

7. Abandoning Sverdrup
Gran and Braarud’s Critical Photosynthesis Hypothesis
Critical Division Rate Hypothesis
Boss & Behrenfeld, 2010
Behrenfeld & Boss 2014
Behrenfeld, et al., 2013
Behrenfeld, 2014
Behrenfeld et al., 2016, 2017
Behrenfeld & Boss 2018 *

8. Abandoning Sverdrup – the underlying data80 70 60 50 40 30 20 10 A 70 70
NA-12 60 60
NA-11
NA-10
NA-9
NA-8
NA-7 50 50
NA-6
NA-5
NA-4
NA-3
NA-2
NA-1 40 40 30 30
0.01
0.1
1.0 10
Chlorophyll (mg m-3)
= 1989 NABE
= 2008 NAB
= focus in later slides
SeaWiFS data 1998 – 2006
8-day resolution
12 central NA bins, minimize advect.
Chlsat = OC4-V4
Cphyto = GSM / Westberry et al 2008
Biological Oceanography, (8)
Wednesday, September 04, 2019

9. Good correlation between estimates of chlorophyll and biomass
NA-5 - Latitude range: 45o – 50oN
North Atlantic seasonal cycles are dominated by changes in biomass
Thus, Cphyto ~ Chlsat
Differences between Cphyto and Chlsat consistent with photoacclimation
All analyses have been completed with both C and Chl
Results to follow are the same irrespective of C or Chl
Remaining slides focus on C
Within-bin standard deviations shown above as gray lines
Biological Oceanography, (9)
Wednesday, September 04, 2019

10. Correlation between shallowing of the MLD and Phyto. bloom
Mixed layer depth
NA-5 - Latitude range: 45o – 50oN
“Biomass”
PAR MLD
Light (PAR)
Peak biomass occurs in spring
Coincident with rising PAR and shoaling MLD
Also associated with rapid rise in primary production
Conclusion: phytoplankton in the North Atlantic exhibit a repeated vernal bloom caused by increased primary production and growth associated with rising light and shallowing mixed layers – aka, Sverdrup
… unfortunately, biomass can be a terribly misleading thing….
& correlation is not causation
Biological Oceanography, (10)
Wednesday, September 04, 2019

11. and since r= -l  A bloom start when >l
Bloom in a Bottle
To understand what causes a bloom, it is necessary to first identify when a bloom started
The start of a bloom can not be defined by biomass - e.g., when biomass X mg m-3 or Y% above annual median
Using biomass can lead to the wrong start date and association of bloom initiation with the wrong environmental forcing
Bloom initiation implies a change in the rate of growth – for Sverdrup it was the beginning of positive net growth
5% > mean
Days
A bloom starts when:
r (d-1)= >0
and since r= -l  A bloom start when >l
 = phytoplankton specific growth rate
l = bulk specific loss rate
r = net population growth rate
Biological Oceanography, (11)
Wednesday, September 04, 2019

12. Bloom in a Bottle
An easy way to get a first-order sense of rate changes is to plot biomass on a logarithmic scale:
5% > mean
5% > mean
Net population growth rate= = slope of log plot
Biological Oceanography, (12)
Wednesday, September 04, 2019

13. Bloom in a Bottle Autumn Winter
Population specific net growth rates (r) can be calculated from changes in phytoplankton concentration (m-3) as long as the mixed layer is either shoaling or not deepening
However, one must consider the influence of dilution when the mixed layer is deepening
A dilution correction should be considered when assessing growth rates during mixed layer deepening
Depth
Biological Oceanography, (13)
Wednesday, September 04, 2019

14. Abandoning Sverdrup Mixed layer depth Light (PAR)
Deep
NA-5 - Latitude range: 45o – 50oN
Shallow
Light (PAR)
Population net specific growth rate (r) for the active water column:
Negative in summer
positive in late-autumn / early winter and remains positive through the spring until nutrients are depleted
Growth-phase maxima in r can occur during MLD deepening, MLD maximum, or MLD shoaling
Overall, r is inversely related to PAR and gross growth rate (µ)
In 100% of available complete annual cycles, r becomes positive before PAR begins to increase
Biological Oceanography, (14)
Wednesday, September 04, 2019

15. Abandoning Sverdrup Mixed layer depth Light (PAR)
Deep
NA-5 - Latitude range: 45o – 50oN
Shallow
Light (PAR)
Population net specific growth rate (r) for the active water column:
Negative in summer
positive in late-autumn / early winter and remains positive through the spring until nutrients are depleted
Growth-phase maxima in r can occur during MLD deepening, MLD maximum, or MLD shoaling
Overall, r is inversely related to PAR and gross growth rate (µ)
In 100% of available complete annual cycles, r becomes positive before PAR begins to increase
Biological Oceanography, (15)
Wednesday, September 04, 2019

16. Abandoning Sverdrup
The North Atlantic bloom does not begin in the spring
Net exponential growth begins mid-winter
Shift from negative to positive biomass changes coincides with the cessation of the deepening of the mixed layer
Net growth rates are, on average, comparable from winter through spring
Net growth rates do not reflect changes in incident light, photosynthetic rate, or gross growth rate (µ)
The critical depth hypothesis can be dismissed
NA-5 - Latitude range: 45o – 50oN
FIG. 2. (A) Nine-year record of phytoplankton biomass (Cphyt) at eight-day resolution for bin NA-5 plotted with a log transformed
y-axis (same data as in Fig. 1B, C). Gray bars indicate within-bin standard deviations. Vertical dashed lines indicate 1
January of each year. (B–F) Mean annual cycles in Cphyt at eight-day resolution (black symbols, log-transformed left axis) and
mixed layer depth (MLD) from the Fleet Numerical Meteorology and Oceanography Center (FNMOC) and the Simple Ocean
Data Assimilation (SODA) models (black line, right axis). Gray symbols (right axis) represent mean mixed layer depth for three
models, showing close agreement in timing of mixed layer deepening and shoaling despite differences in magnitude of midwinter
mixing depth (gray bars indicate standard deviation between models). The three MLD estimates are from the FNMOC/SODA
model (as in Fig. 1), the Miami Isopycnic Coordinate Ocean Model (MICOM; Bleck et al. 1992), and a higher resolution (20–40
km) version of the MICOM (Ha´ tu´ n et al. 2005; see Milutinovic et al [2009] for additional details). Annual cycles begin in January
and end in December (x-axes).
Cphyto
Biological Oceanography, (16)
Wednesday, September 04, 2019

17. The timing of the “spring” bloom Initiation
Mixed layer depth
Shallow
Photic depth)
Specific growth rate
Deep
Starting now in July
The ‘vernal’ bloom appears to be an event initiated in late fall
Triggering of the bloom appears to be associated with mixed layer
deepening (not shoaling)
How is this possible? Why the mid-winter decrease in r?
Biological Oceanography, (17)
Wednesday, September 04, 2019

18. The timing of the “spring” bloom Initiation
Behrenfeld and Boss Ann. Rev. 2014
The ‘vernal’ bloom appears to be an event initiated in late fall
Triggering of the bloom appears to be associated with mixed layer
deepening (not shoaling)
How is this possible?
Biological Oceanography, (18)
Wednesday, September 04, 2019

19. How is this possible? * * µ = NPP / CZ
Specific phytoplankton growth rate calculated from primary production models *
(B) Annual mean cycles from July to following July of r (open symbols, left axis), MLD (heavy black line, right axis), and euphotic depth (Zeu: heavydotted line, right axis) for bin NA-5. Note that in this figure MLD increases downward. Gray bars indicate standard deviation in r for each eight-day period over the nine-year satellite record (panel A). ‘‘Positive net growth phase’’ is indicated at top. Horizontal dashed line indicates r ¼ 0. (C) Annual mean cycles in phytoplankton specific growth rate (l) for NA-5 (left axis) based on net primary production (NPP) estimates from (solid line) the standard Vertically Generalized Production Model (VGPM); (dashed line) the VGPM with an exponential description of chlorophyll-specific light-saturated photosynthesis (Pb
opt); and (dotted line) the VGPM with a regionally tuned description of Pb opt (see the Abandoning Sverdrup section for details). Note again that the x-axis begins in July and ends in July a year later. Also shown is the mean annual cycle in r from panel B (open symbols, right axis, with same scale as in panel B). Gray hatched box shows where the entire annual range in r is found when plotted on the same left axis as l, emphasizing how small r is relative to l throughout the year.
A net specific growth rate of 0.02 implies approximately 1 division per month
Typical winter C = 4 – 8 mg m-3, Typical spring C peak = 25 – 70 mg m-3
NA bloom requires 2 – 4 doublings over months, or average r of to 0.03 d-1
Biological Oceanography, (19)
Wednesday, September 04, 2019

20. Covariation of losses and production
Positive r through winter is allowed because losses co-vary with µ
(Sverdrup assumed this ‘respiration’ to be a constant)
The increase in r during winter implies that the fraction of µ that escapes predation and other losses (i.e., r:µ) must increase in the winter
Biological Oceanography, (20)
Wednesday, September 04, 2019

21. Recall the Dilution Method*
Landry & Hassett 1982 Mar. Biol. 67,
Landry et al Mar. Ecol. Prog. Ser. 120, 53-63 *
Biological Oceanography, (22)
Wednesday, September 04, 2019

22. Behrenfeld’s ‘Disturbance-Recovery Hypothesis’
As a replacement for the Critical Depth Hypothesis, it is proposed that the north Atlantic bloom is a consequence of a massive scale ‘dilution experiment’
Mixed layer deepening causes a slight decoupling between phytoplankton growth and losses (grazing, mostly)
The ‘decoupling’ increases so long as the mixed layer continues to deepen
Mixed layer shoaling drives a ‘re-coupling’ of phytoplankton growth and losses (grazing)
Thus, while spring shoaling and increasing light favor enhanced photosynthesis and growth, they also favor heavier grazing losses
Biological Oceanography, (23)
Wednesday, September 04, 2019

23. F IGURE 2 Annual cycles of phytoplankton (top black line)
maximum division rate for a nutrient-replete euphotic zone and
(bottom black line) biomass accumulation rate, reproduced from
figures 18 and 20 of Riley (1946). Maximum division rates are based
on measured incident sunlight and diffuse attenuation. These values
are then adjusted downward to account for nutrient stress (yellow
shading) and mixing deeper than the euphotic zone (blue shading) to
yield gross production rates (blue line). Net production rates (red
line) are calculated from gross production by subtracting values of
respiration (red shading) calculated as an exponential function of
temperature. Finally, biomass accumulation rates are calculated from
net production by subtracting grazing losses (green shading) based
on observed zooplankton abundances. Biomass accumulation rates
above the horizontal black-dashed line correspond to increasing
phytoplankton concentrations, while below the dashed line biomass
is decreasing
After Riley, G. A. (1946). Factors controlling phytoplankton populations on Georges Bank. Journal of Marine Research, 6, 54–73.

24. Behrenfeld’s & Boss ‘Disturbance-Recovery Hypothesis’
Modeled annual cycles in the concentrations of phytoplankton ( green lines) and herbivores (orange lines)
No carnivores and incremental changes in phytoplankton division rate (μ).
Addition of carnivore predation on herbivores.
μ is calculated from an annual cycle in mixed-layer light levels corresponding to mixed-layer depth (MLD)
Accounting for the physical effects of MLD changes on phytoplankton and herbivore concentrations and interactions.
The bloom is caused by MLD deepening having a greater impact on herbivore populations than on phytoplankton.
Modeled annual cycles in the concentrations of phytoplankton ( green lines) and herbivores (orange lines).
(a) Model results with no carnivores (i.e., no density-dependent herbivore-specific mortality term) and
incremental changes in phytoplankton division rate (μ). (b) Model results after the addition of carnivore
predation on herbivores. Here, the green and orange lines are phytoplankton and herbivore concentrations,
respectively, for incremental changes in μ; the black lines are phytoplankton (top) and herbivore (bottom)
concentrations when μ changes smoothly over the annual cycle. (c) The same as panel b except that μ is
calculated from an annual cycle in mixed-layer light levels (upper subpanel ) corresponding to an annual range
in mixed-layer depth (MLD) that is representative of areas in the subarctic Atlantic. (d ) The same as panel c
but additionally accounting for the physical effects of MLD changes (upper subpanel ) on phytoplankton and
herbivore concentrations and interactions. Abbreviation: PAR, photosynthetically active radiation (mol
photons m−2 h−1).
Biological Oceanography, (28)
Wednesday, September 04, 2019

25. Critical Turbulence Hypothesis
May 2015

26. Behrenfeld’s & Boss ‘Disturbance-Recovery Hypothesis’
Modeled annual cycles in the concentrations of phytoplankton ( green lines) and herbivores (orange lines).
(a) Model results with no carnivores (i.e., no density-dependent herbivore-specific mortality term) and
incremental changes in phytoplankton division rate (μ). (b) Model results after the addition of carnivore
predation on herbivores. Here, the green and orange lines are phytoplankton and herbivore concentrations,
respectively, for incremental changes in μ; the black lines are phytoplankton (top) and herbivore (bottom)
concentrations when μ changes smoothly over the annual cycle. (c) The same as panel b except that μ is
calculated from an annual cycle in mixed-layer light levels (upper subpanel ) corresponding to an annual range
in mixed-layer depth (MLD) that is representative of areas in the subarctic Atlantic. (d ) The same as panel c
but additionally accounting for the physical effects of MLD changes (upper subpanel ) on phytoplankton and
herbivore concentrations and interactions. Abbreviation: PAR, photosynthetically active radiation (mol
photons m−2 h−1).
Our finding is that phytoplankton blooms require a physically forced preconditioning of ecosystems that significantly disrupts phytoplankton-herbivore relationships. In a sense, the ability of phytoplankton to become uncommonly abundant (i.e., bloom) relies on them first being uncommonly rare
Biological Oceanography, (30)
Wednesday, September 04, 2019

27. Modeling the ‘Grand Dilution’
Biological Oceanography, (31)
Wednesday, September 04, 2019

28. Satellite data tells a very different story about the bloom
Traditionally, the bloom is thought to occur here, due to increased sunlight and shallower mixing fueling rapid photosynthesis
Traditionally, low light and deep mixing in winter is thought to prevent a bloom
surface
Mixing Depth
100 m
200 m
Satellite Data show that phytoplankton population growth rates increase in late autumn, reach a peak by early winter, and then remain high until late spring !
In other words, the bloom periods starts earlier than we thought, lasts longer than we thought, and begins when environmental conditions for phytoplankton growth are at their worst (or are they?)
Population Growth Rate
Sunlight
April
July
April
July
January
October
January
October
Month
Biological Oceanography, (32)
Wednesday, September 04, 2019

29. Satellite data tells a very different story about the bloom
Traditionally, the bloom is thought to occur here, due to increased sunlight and shallower mixing fueling rapid photosynthesis
Traditionally, low light and deep mixing in winter is thought to prevent a bloom
surface
Mixing Depth
100 m
200 m
Satellite Data show that phytoplankton population growth rates increase in late autumn, reach a peak by early winter, and then remain high until late spring !
In other words, the bloom periods starts earlier than we thought, lasts longer than we thought, and begins when environmental conditions for phytoplankton growth are at their worst (or are they?)
Population Growth Rate
Sunlight
April
July
April
July
January
October
January
October
Month
Biological Oceanography, (33)
Wednesday, September 04, 2019

30. Final Comments
Temporal coverage of the satellite record provides a unique opportunity to re-evaluate bloom dynamics
The critical depth hypothesis is found wanting
A Disturbance-Recovery Hypothesis is suggested, but is not the only potential explanation (nutrients, aggregation, temperature effects, sinking….?)
Climate change effects on North Atlantic (and other) blooms may be very much different for a ‘Critical Depth’ concept of blooms and a ‘Dilution’ concept of blooms
Alternative explanations where recently suggested
Biological Oceanography, (34)
Wednesday, September 04, 2019

32. “ ” “ * * Winter Chlz Lat trends Mixing velocity Feb C max Lat r µ
Cphyt vs Csat ” * “ *