Influence of timing of sea ice retreat on phytoplankton size during marginal ice zone bloom period on the Chukchi and Bering shelves

Amane Fujiwara, Toru Hirawake, Koji Suzuki, Lisa Eisner, Ichiro Imai, Shigeto Nishino, Takashi Kikuchi and Sei-Ichi Saitoh

Kew words•  Long time series data analysis•  Spatial analysis•  Phytoplankton size •  Bering & Chukchi Shelves

JAMSTEC, Hokkaido University, NOAA

IntroductionRemote sensing of phytoplankton functional types (PFTs)

Hirata et al. (2011)•  Different PFTs (size classes/groups) play different roles in the marine ecosystem

(e.g., carbon, nitrogen, sulfur cycles, energy transport, biological pump)•  Monitoring of PFTs from satellite can provide valuable information•  Regional model development/tuning are also required for the regional study•  We focused on the phytoplankton size composition in the western Arctic Ocean

size

groups %chla

Introduction

•  Further evaluation of the impacts on the ecosystem is required especiallyconsideringbo0om-upcontrolofthefoodwebthatbeginsfromphytoplanktonproduc9on

•  Ocean color Remote sensing can contribute to understand phytoplankton responses to the sea ice decrease

Changes in ecosystem have been reported•  Continuous decline of sea ice•  Northward shift in the species and

biomass distribution (Grebmeier et al.,2006, 2012 and their references)

•  Little is known for the phytoplankton response

Study Area: Bering & Chukchi shelves

Introduction

Phytoplankton bloom(biomass & size composition)

mixing

nutrients

Apr

May

Jun Jul

Aug

Sep

Organic matter

Ice-algae

Seasonality of Arctic Marine Ecosystem

benthos

fishes

•  Spring Phytoplankton bloom supports annual marine biological production•  Monitoring of inter-annual variability is important as well as primary production

zooplankton

(Modified from Wassmann et al., 2011)

Objectives

1.  Assess the responses of phytoplankton size structure during bloom period to the timing of sea ice

2.  Assess the controlling factors of annual primary production

Prior to this study, size derivation algorithm and PP algorithm were proposed optimally for the study region

(Fujiwara et al., 2011, Hirawake et al., 2012)

Long-time monitoring of phytoplankton size structure and primary production can contribute to understand the recent marine ecosystem changings in the Bering and Chukchi Seas

%chla>5µm (fractional chla biomass of >5µm phytoplankton)= fraction of large phytoplankton

Materials & Methods

Length of open-water

period

Rrs(λ)

Timing of sea-ice retreat

Daily SST

Daily Heat Flux

NCEP/NCAR

Sea IceConcentration

PAR Daily PPeu

Daily %Chla>5µm

Annual Primary

production

SSMI

SeaWiFS & MODIS

∫Heat flux

14-day average after sea-ice melt for each pixel

AVHRR & MODIS

SeaWiFS & MODIS

Overview of the data processing

Daily, 9-km, L-3

Bloom time & annual median

Bloom time

Bloom time & annual median

SeaWiFS (1998–2007) and MODIS (2003–2013) Rrs(λ) were merged à bias correction

SST

%Chla>5µm

Timing of sea ice retreat

Materials & MethodsBias correction between SeaWiFS and MODIS Rrs(λ)

bias and RMSE were simulated changing conversion factor, and optimum values were determined (bias = 0) using data of common observation term (2003–2007)

Materials & MethodsEffect of the bias correction

201510

50

-8 -6 -4 -2 0 2 4 6 8bias of FL

default SeaWiFS converted SeaWiFS

40302010

0

Num

ber o

f obs

erva

tions

-0.4 -0.2 0.0 0.2 0.4bias of chla

40302010

0-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

bias of aph(443)

(a)

(b)

(c)

Mean = -1.43, -0.19Median = -1.40, -0.10Stdev = 1.66, 1.55Min = -7.30, -6.28Max = 3.70, 5.54

Mean = -0.09, -0.00Median = -0.09, 0.00Stdev = 0.05, 0.05Min = -0.38, -0.29Max = 0.05, 0.15

Mean = -0.07, -0.02Median = -0.06, -0.01Stdev = 0.05, 0.05Min = -0.67, -0.68Max = 0.03, 0.08

Histograms of biases between MODIS and SeaWiFS products

201510

50

-8 -6 -4 -2 0 2 4 6 8bias of FL

default SeaWiFS converted SeaWiFS

40302010

0

Num

ber o

f obs

erva

tions

-0.4 -0.2 0.0 0.2 0.4bias of chla

40302010

0-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

bias of aph(443)

(a)

(b)

(c)

Mean = -1.43, -0.19Median = -1.40, -0.10Stdev = 1.66, 1.55Min = -7.30, -6.28Max = 3.70, 5.54

Mean = -0.09, -0.00Median = -0.09, 0.00Stdev = 0.05, 0.05Min = -0.38, -0.29Max = 0.05, 0.15

Mean = -0.07, -0.02Median = -0.06, -0.01Stdev = 0.05, 0.05Min = -0.67, -0.68Max = 0.03, 0.08

bias of %chla>5µm was successfully removed(also for chla and PP)

bias of %chla>5µm

Rrs bias correction

Materials & Methods

Sea-ice Retreat timing bloom time %Chla>5µm

1998

1999

2000

2013

・・・

Spatial correlation analysis→Assess how yearly phytoplankton size composition during bloom period changes corresponding to yearly change of sea-ice retreat timing and other environmental variables

•  %chla>5µm•  SST•  dOHC•  PAR

Correlation analysis

Timing of sea ice Retreat

Bloom9mevalue

One-by-one pixel

Results & Discussion

Distribution of negative correlation coefficient (ρ) dominates (~70% of the area) in the shelf region (~16% was significant (p<0.05))

Earlier ice-retreat generally causes increase of larger phytoplankton (i.e., diatoms) during bloom periodDistribution of correlation coeff.

ρ(timing retreat & %Chla>5µm)

•  Mechanism of Nutrient supply ??•  Under water light ??

positive (p<0.05)

positive

negative

negative (p<0.05)

Timing of sea ice retreat

ρ(retreat timing & %Chla>5µm)

ρ(retreat timing & SST) ρ(retreat timing & ∫heat fluxbloom)

Results & DiscussionWhy earlier sea-ice retreat causes increase of larger phytoplankton??

∫Heat fluxbloom & SST: proxy of development of surface mixed layer

•  ~80% of the area showed positive ρ between SST and retreat timing and negative ρ between ∫heat flux and retreat timing⇒Earlier ice retreat causes colder SST and we can infer slow development of thermal stratification because of colder air temperature and weaker solar radiation in early season

positive (p<0.05)

positive

negative

negative (p<0.05)

Distributions of correlation coeff. (ρ)

Continuous nutrient supply from below can be expected in the early retreat years after sea ice meltà Large %Chla>5µm likely to maintain

Results & DiscussionWhy earlier sea-ice retreat causes increase of larger phytoplankton??

•  Sea ice disappears no later than summer solstice•  Under-ice bloom is likely to occur in the late retreat

years -  Sufficient light penetration into under ice can

be expected (strong radiation around summer solstice and thinner & fragile ice)

Under ice bloom(Arrigo et al., 2012)

Nutrients can be utilized by under-ice bloom before sea ice melt in late retreat years

Distribution of Frequent bloom type from Lowry et al. (2014)

AlaskaSiberia

ρ(retreat timing & %Chla>5µm)

Alaska

SiberiaSpatially matched!

Frequent bloom type is under ice

bloom

Results & DiscussionWhat contributes to annual primary production (APP) in the shelf area??

èStandardized multiple regression analysis

(length of open-water period)(%Chla>5µm) (SST)

Comparison of partial regression coefficients

(APP = α1 %Chla>5µm + α2 SST + α3 LOP)

partial regression coefficient

Selected variables•  Ann. Med. %Chla>5µm à phytoplankton size composition •  Ann. Med. SST à phytoplankton activity•  Length of open-water period à length of growing season

•  %Chla>5µm was the most important variable for APP in the southern Chukchi & Bering Seas•  óLOP was important in the northern Chukchi shelf (r.g., Arrigo et al., 2008, 2011)

è S-MRA allows us to compare the magnitudes of the contribution to APP

Conclusions•  Earlier sea ice retreat triggers increase of larger phytoplankton in the open-

water area–  Response of under-ice phytoplankton community should be taken into

account–  Energy use for ice-favored and open-water zooplankton species can

change•  Phytoplankton size is important factor for Annual Primary Production

especially in the southern Chukchi and Bering shelf–  Nutrients & productive groups are important to determine APP in the

southern longer ice free area–  Length of growth season is main factor to determine APP in the northern

shorter ice-free area•  Continuous monitoring of phytoplankton size and production is needed to

comprehend ongoing ecosystem changings •  Fujiwara et al. Biogeosciences Discuss., vol. 12, p. 12611-12651, 2015 (revised)

Thank you for the attention!

Acknowledgements•  the GRENE Arctic Climate Change Research Project,

NIPR•  GCOM-C RA4, JAXA