Transcript Global warming impact on phytoplankton seasonal cycles

Global warming impact on
phytoplankton seasonal cycles
Stephanie Henson
Harriet Cole, Claudie Beaulieu,
Andrew Yool
Motivation
• Seasonal cycle of phytoplankton is relevant to
higher trophic levels and carbon export
• How will phytoplankton seasonality change with
global warming and why?
• A previous study suggested it takes ~ 30 years
to detect a global warming trend in primary
production
• Could seasonality be a ‘shortcut’ to detecting
effects of climate change?
How will global warming alter seasonality?
Reduced mixing
+ nutrient
limitation ->
weaker seasonal
cycle
Reduced mixing +
light limitation ->
seasonal cycle
remains & earlier
blooms
The canonical view (Doney, 2006)
How will phytoplankton seasonality change
with global warming?
Take coupled climate model simulations using IPCC CMIP5 models run
with the RCP8.5 scenario 2006-2100:
• Canadian Centre for Climate Modelling and Analysis CanESM2
• NOAA Geophysical Fluid Dynamics Laboratory GFDL-ESM2M
• Met Office Hadley Centre HadGEM2-CC
• Institut Pierre Simon Laplace
IPSL-CM5A-MR
• Max Planck Institute
MPI-ESM-LR
• National Oceanography Centre
NEMO-MEDUSA
Phytoplankton seasonal cycle metrics
Timing of peak
Seasonal
amplitude
(max-min)
North Atlantic
seasonal cycle
of primary
production
(GFDL model –
monthly
output)
Trends in phytoplankton seasonality
Primary production
Seasonal amplitude
Average % change per
year, 2006-2090
Timing of peak
Difference in days, 20062026 vs 2071-2090
Trends in phytoplankton seasonality
Decrease in PP, except Arctic
Decrease in seasonality, especially in
North Atlantic
Peak PP ~ advances, particularly
Arctic
Trends in drivers of seasonality
Surface nitrate seasonal
amplitude decreases
almost everywhere
Average % change/year
MLD seasonal amplitude
decreases everywhere
except the Arctic
ΔSST/year
SST amplitude increases
(highs get hotter quicker
than the lows)
SST
MLD
NO3
How much data is needed to detect a global
warming trend?
Signal (i.e. trend) has to exceed noise (i.e. natural
variability)
2/ 3
3.3 1  
*
N
n  

1  
 
n* : number of years required to detect trend
N : standard deviation of the noise (residuals after trend
removed)
 : estimated trend
 : auto-correlation of the noise (AR(1))
Weatherhead et al. (1998)
Detecting a trend in phytoplankton seasonality
Mean annual PP
n* - Number of years to detect a trend above natural variability
Mean PP – 34 years
Detecting a trend in phytoplankton seasonality
Mean annual PP
Seasonal amplitude of PP
n* - Number of years to detect a trend above natural variability
Mean PP – 34 years; seasonal amplitude – 37 years
Effect of model temporal resolution
• Used monthly mean model output here
• But phenological changes may only be observable
at higher temporal resolution
• How does changing the model temporal resolution
alter n* (number of years to detect trend)?
Ongoing work (Harriet Cole)
Effect on n* of calculating trends
in bloom initiation with different
model temporal resolution
Conclusions
• Seasonal amplitude of PP decreases; timing of
peak advances  transformation of bloom regions
to non-bloom regions
• Due to decreased mixing and nutrient supply
• Arctic is an exception: increased seasonality and
earlier peak, but reduced mixing  effect of ice
melt?
• Seasonality metrics are not necessarily a shortcut
to detecting a trend
• For some regions > monthly resolution data
required to detect phenological change
Henson et al. (2010); Beaulieu et al. (2013); Henson et al. (in press) – all
Biogeosciences