© plosi trends in ocean colour and chlorophyll ... · the record of chlorophyll changes over time...

OPEN 3 ACCESS Freely available online © PLOSI o -

Trends in Ocean Colour and Chlorophyll Concentration from 1889 to 2000, WorldwideMarcei R. Wernand1*, Hendrik J. van der Woerd2, Winfried W. C. Gieskes31 Royal Netherlands Institute for Sea Research, Physical Oceanography, Marine Optics & Remote Sensing, Texel, The Netherlands, 2 Institute for Environmental Studies (IVM), VU University Amsterdam, Amsterdam, The Netherlands, 3 University of Groningen, Department of Ocean Ecosystems, ESRIG, Groningen, The Netherlands

AbstractMarine primary productivity is an important agent in the global cycling of carbon dioxide, a major 'g reenhouse gas', and variations in the concentration of the ocean's phytoplankton biomass can therefore explain trends in the global carbon budget. Since the launch of satellite-mounted sensors globe-wide monitoring of chlorophyll, a phytoplankton biomass proxy, became feasible. Just as satellites, the Forel-Ule {FU) scale record (a hardly explored database of ocean colour) has covered all seas and oceans - but already since 1889. We provide evidence that changes of ocean surface chlorophyll can be reconstructed with confidence from this record. The EcoLight radiative transfer numerical model indicates that the F U index Is closely related to chlorophyll concentrations in open ocean regions. The most complete FU record is that of the North Atlantic in terms of coverage over space and in time; this dataset has been used to test the validity of colour changes that can be translated to chlorophyll. The FU and F(J-derived chlorophyll data were analysed for monotonously increasing or decreasing trends with the non-parametric Mann-Kendall test, a method to establish the presence of a consistent trend. Our analysis has not revealed a globe-wide trend of increase or decrease in chlorophyll concentration during the past century; ocean regions have apparently responded differentially to changes in meteorological, hydrological and biological conditions at the surface, including potential long-term trends related to global warming. Since 1889, chlorophyll concentrations have decreased in the Indian Ocean and in the Pacific; increased in the Atlantic Ocean, the Mediterranean, the Chinese Sea, and In the seas west and north-west of Japan. This suggests that explanations of chlorophyll changes over long periods should focus on hydrographical and biological characteristics typical of single ocean regions, not on those of ' the' ocean.

Citation: Wernand MR, van der Woerd HJ, Gieskes WWC (2013) Trends in Ocean Colour and Chlorophyll Concentration from 1889 to 2000, Worldwide. PLoSONE 8(6): e63766. doi:10.1371/journal.pone.0063766

Editor: Fabiano Thompson, Universidade Federal do Rio de Janeiro, Brazil

Received January 7, 2013; Accepted April 5, 2013; Published June 12, 2013

Copyright: © 2013 Wernand e t al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: The authors have no support or funding to report.

Com peting Interests: The authors have declared that no competing interests exist.

* E-mail: [email protected]

IntroductionO ceanographic da ta archaeology (salvaging historical ocean

data) is, according to Parker [1], a critical requirem ent for clim ate change research w hen the data (e.g., tem peratu re and salinity) cover long tim e periods. C hanging clim ate will affect m arine life, and this, in turn , plays a crucial role in clim ate control [2]. As a m atter o f fact, phytoplankton accounts for nearly ha lf o f E arth ’s total p rim ary productivity [3]. Phytoplankton biom ass is nowadays readily derived from satellite da ta [4]; indeed, chlorophyll provides the best index o f phytoplankton biom ass for p rim ary productivity studies, as was recently confirm ed by H u o t et al. [5]. Long-term changes in ocean prim ary production m ay well have im portant consequences for the ocean food chain, as well as for the global carbon cycle [6 ],

O n e of the so-called ‘Essential C lim ate V ariables’ listed by the W orld M eteorological O rganization (W M O) to detect biological activity in the ocean’s surface layer is ocean colour [7]. M ore recently an im portan t aspect o f clim ate research has been the detection o f ocean colour from space with derived products such as w ater transparency that is influenced by coloured dissolved organic m atter absorption and chlorophyll concentration. W ater transparency can be used to predict the dep th o f the upper ocean

m ixed layer [8 ]. This layer plays a critical role in the flux o f energy betw een a tm osphere an d ocean.

C hanges in w ater colour are caused by a change in the com position o f optically active substances [9]: suspended particulate m atter, pigm ents (chlorophyll-a, b and c and carotenoids) in algae [10], an d dissolved organic m atter [11]. T he blue colour o f oligotrophic oceans is caused by dom ination of selective absorption an d scattering o f the w ater molecules. Chlorophyll colours the w ater green (or even red by the carotenoids perid in in in dinoflagellates), and the presence o f coloured dissolved organic m atter (CDO M ) will, in high concentrations, account for the absorption o f m ost o f the blue pa rt o f the incom ing sunlight, resulting in brow nish-coloured water.

T h e colour o f seas and oceans is, next to w ater transparency, salinity and tem perature, one of the few oceanographic param eters that have been recorded for over a century [12]. O ne outstanding long-term record (since 1931) has been p roduced by the C ontinuous P lankton R ecorder Surveys in the N orth Sea and the N o rth Atlantic [13], [14]; o ther records are the H aw aii- and B erm uda Atlantic O cean Tim e-Series (H O T S an d BATS [15], [16]) and the California C ooperative O ceanic Fisheries Investigations (CalCO FI) [17]. All these series have from the launch o f the Coastal Zone C olor Scanner been supplem ented with satellite-

PLOS ONE I www.plosone.org 1 June 2013 | Volume 8 | Issue 6 | e63766

mailto:[email protected]

http://www.plosone.org

Long-Term Trends in Ocean Colour and Chlorophyll

[O recro ilch . ,10 6 a t l L&nge Ztl1 V esti. L.

°*t*ec^ûa,biatu

D u n k c !

in P ro te s te n CelbmchTorels

W «.«/er/.

M a ß s ta b 1:31157 000*7. — 7* Srmcain

•n ■. .t _ . - t —. ■— . » y xiiojM U r■ -*- —- <#«• Ajju-.f.Oiv;

O.Kr-̂ !»in»J AI

Plankton* Expcdition . C. Tafel II.

A ^ r la t j v o n L ip s in s & T is c h c r - K ie l iL L c ip z iç .C*ogr»fiii-Aasltyi -̂on’YTtpicriD«íc»¡L*ipxí¿

t> «*> liy • iU ». i » ckf £**t>*c>ilui){'e»i.

llKUKSCilK.NDKW A S S E R F A R B E ,

Tiefe» KcboltMnuvon (prtfltcrTnMiijarcnx.

K obxilbtau.

G rünlich P .lau .

AuA¿<pr.tct Crünbtav

Il «jA í i -«¿.Osisocf ru n , m r l i i » n u k £eti\>>'t.

ÜU au£tr\<iW der EaJirCiinU- d er E xpédition veneuhnUen- t'çrbenxablrr, j ù td van Dr.Gerhard. Schott- /& / tend. Ä? beah- cchlet.,Y er/i-rb t * i>e:eiefu\ec g renu , ader Im ane Trübungen, dee

I

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21

Figure 1. Krümmel's contoured North Atlantic FU-map (1889). The sailing tra ck o f th e s te a m s h ip 'N ational ' is sh o w n in black. T he co lo u r (W asserfarbe) w as in d ica ted as a p e rc e n ta g e o f a yellow p o ta ss iu m c h ro m a te so lu tio n a d d e d to a b lu e c o p p e r -s u lp h a te so lu tio n . T he le g en d ind ica tes th e R /-sca le co lou rs 1 to 21. do i:10 .1371/jo u rn a l.pone.0063766 .g001

derived oceanographic products for tem peratu re change and, as to biological characteristics, for ocean colour and its derivative, chlorophyll concentration. From the late 1990s m ore sophisticated im aging ocean colour satellite sensors such as the Sea viewing

W ide angle Field o f view Sensor SeaW iFS, the M E dium Resolution Im aging Spectrom eter M E R IS and the M oderate Resolution Im aging Spectroradiom eter M O D IS sam ple the ocean’s colour. Since the launch o f SeaW iFS, ocean colour from




space w ith rigid m easurem ent protocols [18] has been archived as a standard oceanographic param eter. W ith revisit times of 1 to 3 days, a swath w idth o f a round 1,000 to 2,000 kilom etres an d a spatial resolution o f 250 to 1,000 m eters, the ocean has never before been sam pled optically on such a scale [19].

T h e record o f chlorophyll changes over tim e is lim ited; only a few authors have presented analyses o f trends in chlorophyll over lim ited num bers o f years. V enrick et al. [20] found, by m eans o f tren d analysis o f standing-stock chlorophyll, a doubling of phytoplankton chlorophyll in the central N orth Pacific gyre betw een 1964 and 1985. W e m ust b ear in m ind th a t the analysis o f chlorophyll has becom e m ore accurate with the introduction of H PL C in the late seventies o f the past century [21]. Falkowski and W ilson [22] found areas (10 by 10 degrees boxes) with significant positive and negative average annual changes in Secchi depth since 1900, bu t overall their estim ated annual change averaged over the whole North-Pacific O cean ’s chlorophyll concentration is small and no t significant. A similar transform ation was recently used by Boyce et al. [23] in o rder to derive world-wide changes in ocean phytoplankton biomass since 1899.

Since 1889 a colour com parator m ethod has been used to establish the colour o f the sea, th rough Forel-Ule scale observations [24], [25]. Besides Forel and Ule, the inventors o f the scale, K rüm m el (1893), Luksch (1901) and V on Drygalski (1897) [26], [27], [28] were am ong the first to use this m ethod in the open sea. K rüm m el m ade his observations during the p lankton expedition of the H um bold t Society in 1889. At th a t tim e only the blue to blue- green p a rt o f the scale (scale num bers 1 to 1 1 ) was used to classify the colour o f open w ater. An exam ple o f K rüm m eFs con toured sea colour m ap is shown in Figure 1. This m ap shows surprisingly well the oligotrophic Sargasso Sea gyre (‘deep cobalt blue, FU-V), as well as the outflow o f the C ongo R iver (‘greenish-blue, -FU2-3’), and the C anary and G reen land C urrents (‘dark green, FU4-7’). T h e da ta were collected in the A tlantic during the Plankton Expedition of 1889 and the colour o f the sea was expressed as a percentage o f a yellow potassium chrom ate solution added to a blue copper-sulphate solution. Luksch collected observations over

eight years during the ‘Pola’ expedition (1890-1898) in the M editerranean , Aegean- and R ed Sea and V on Drygalski perform ed observations during the G reen land Expedition o f the G eographical Society o f 1891-1893. Since 1890 the Forel-Ule scale becam e the m ost com m only used and m ost simple scale to determ ine, through com parison, the colour o f seas, lakes and rivers.

O ver the years only a lim ited num ber o f geographical m aps, based on interpolation o f sets o f Forel-Ule data, becam e available. T h e U .S. Navy H ydrographic Office published in the 1950s three atlases on m arine geography, including Forel-Ule contours o f the Seas o f Jap a n , K orea an d Indoch ina [29], [30], [31]. A m ore com prehensive effort was m ade by Frederick [32], who presented contoured Forel-Efle m aps. She had a round 24,000 Forel-Efle observations at her disposal to create con tour m aps o f averaged sea colours, bu t she did no t analyse trends. W ernand and V an der W oerd [33] have recently found, in the N orth Pacific O cean, a greening tren d from 1930 until 1955, an d since th en a bluing, based on an analysis o f 17,171 KU-observations that a re available for that ocean.

T h e aim o f the work presented here was to visualise changes in ocean colour, notably its chlorophyll com ponent, over the longest possible period o f tim e by an analysis o f the ‘forgotten’ globe-wide ocean colour obtained by the Forel-Efle m ethod. T h e tem poral variation o f ocean colour has been analysed from 1889 till the year 2000, based on subsets in a total o f 221,110 F U observations collected globe-wide. First the da ta selection and quality control procedures are described an d a m odel is in troduced that provides simple relations for the conversion from iT7-scale num ber to Chlorophyll concentration. Subsequently, the da ta are grouped in 28 seas and oceans. All this should eventually allow speculation on m echanism s beh ind basin-scale external forcing o f the plankton abundance near the ocean surface.

Materials and MethodsW ith the Forel-Efle m ethod the colour o f natural waters is

com pared to a scale, that consists o f twenty-one tubes filled with

100000

<siao+3>SicuC/3

-O01

b í

10000

100

1889-1899

1900-1909

1910-1919

1920-1929

1930-1939

1940-1949

1950-1959

1960-1969

1970-1979

1980-1989

1990-1999

Total 670 1 39 15589 4798 3664 42327 55254 60036 38732□ Winter 119 1 3022 531 480 7926 11364 12149 9023■ Spring 92 3539 660 749 10227 13977 14998 9236□ Summer 183 1 11 5512 2138 1657 13504 17010 18056 10557H Autumn 276 27 3516 1469 778 10670 12903 14833 9916

Figure 2. The tem poral distribution of the total num ber of 221110 m -observations contained within dataset-0 for the period 1889 to 1999. D ata is b in n e d p e r sea so n an d p e r d e c a d e e x c e p t fo r th e first p e rio d o f 11 years from 18 8 9 -9 9 . N otice: th e p e rio d 1900 to 1929 co n ta in zero o r a lim ited n u m b e r o f o b se rv a tio n s . do i:10 .1371/jo u rn a l.p o n e .0063766 .g002




Table 1. The distribution of FL/-observatlons contained within dataset-1.

Sea Area Abr. Total W W -% Sp Sp-% S S-% A A- %

Antarctica AA 51 48 94.1 3 5.9 0 0 0 0

Atlantic Ocean A T 8161 1318 16.2 1823 22.3 3090 37.9 1930 23.7

Arctic Ocean ARO 1 0 0 0 0 1 100 0 0

Baltic BAL 122 0 0 47 38.5 58 47.5 17 13.9

Barents Sea BAR 2651 25 0.9 477 18.0 1741 65.7 408 15.4

Baffin Bay BFB 5 0 0 0 0 4 80 1 20

Black Sea BL 110 1 0.9 31 28.2 28 25.5 50 45.5

Bay of Biscay BOB 17 2 11.8 3 17.6 7 41.2 5 29.4

Bering Sea BS 1750 14 0.8 56 3.2 1645 94.0 35 2

Caribbean Sea CB 700 179 25.6 158 22.6 210 30.0 153 21.9

Chukchi Sea e s 50 0 0 0 50 100 0 0

East China Sea ECS 3432 638 18.6 859 25.0 1122 32.7 813 23.7

Gulf of Alaska GAK 279 1 0.4 75 26.9 179 64.2 24 9

Gulf of Mexico GM 110 3 2.7 28 25.5 27 24.5 52 47.3

Greenland Sea GRS 21 0 0 4 19.0 13 61.9 4 19.0

Indian Ocean INO 3446 971 28.2 711 20.6 552 16.0 1212 35.2

Kara Sea KRA 97 0 0 0 0 53 54.6 44 45.4

M editerranean MED 1080 163 15.1 155 14.4 366 33.9 396 36.7

North Sea NS 56 4 7.1 9 16.1 22 39.3 21 37.5

Norwegian Sea NWS 1524 27 1.8 277 18.2 1121 73.6 99 6.5

Pacific Coast PC 1842 303 16.4 461 25.0 522 28.3 556 30.2

Pacific Ocean PAC 34611 6281 18.1 7417 21.4 11077 32.0 9836 28.4

Philippine Sea PHS 106557 24931 23.4 27115 25.4 27873 26.2 26638 25.0

Red Sea RS 166 51 30.7 61 36.7 0 0 54 32.5

South China Sea s e s 3055 760 24.9 889 29.1 813 26.6 593 19.4

Sea o f Japan SJP 39198 6432 16.4 10399 26.5 13371 34.1 8996 23

Sea o f O khotsk SOK 1393 18 1.3 261 18.7 897 64.4 217 15.6

Yellow Sea YS 9354 2244 24 1952 20.9 3323 35.5 1835 19.6

Total 219839 44414 20.2 53271 24.2 68165 31 53989 24.6

The table shows from left to right: the seas for which FU data was collected, Its abbreviation, the total number of observations, the observations per season (Winter (W), Spring (Sp), Summer (S) and Autumn (A)) with the percentage of the total number of observations. Seas discussed In this article are In bold Italic. Observations of the Inland Caspian Sea (465 obs.) are filtered out using BM1. dol:10.1371 /journal.pone.0063766.t001

m ixtures o f coloured chem ical solutions. T he m ethod has recently been described by W ernand an d van der W oerd [34], M ost F U observations were retrieved from oceanographic and m eteorological databases archived by N O A A -N O D C [35]. T his dataset contains, besides date, geographical position an d Secchi depth (.SB), the F U index (coded 1 to 21; see introduction). Tw o new attributes were added for each entry, indicating the m eteorological season an d the sea area. T h e geographical nam ing follows the conventions o f the U .S. Geological Survey (USGS) that is based upon the latitude an d longitude a t w hich the observation took place. This dataset contains 220,440 observations collected from 1907 to 1999, inclusive.

From three historic expeditions, betw een 1889 and 1899, 670 observations were digitized and added to the N O A A -N O D C dataset. T h e added iTT-observations from before 1900 consisted of 1) 89 observations from K rüm m eFs plankton expedition in the N orth A tlantic in 1889, 2) 367 observations from Luksch ‘Pola’ expeditions collected from 1890 to 1898 in the M editerranean, Aegean- an d R ed Sea and 3) 214 observations from Schott’s G erm an deep sea expedition on the steam er ‘ Valdivia’ collected

from 1898 to 1899 in the N orth Sea, A tlantic, Ind ian O cean, R ed Sea and M editerranean . D uring V on D rygalsk ii G reen land Expedition o f the G eographical Society o f 1891-1893 a round 70 F U observations were collected in the N orth Sea, N orth Atlantic, Davis Strait and Baffin Bay; how ever the b ad quality o f the p rin ted m ap m ade it impossible to digitize these data. For m ost o f the F U d a ta a Secchi disc dep th observation was also available. For the first quality check we looked at the com bination o f F U and SD values an d om itted da ta w hich seemed incorrect, the criterion being that low F U values should have high SD values and vice versa.

T h e re-form atted dataset is referred to as dataset-0 and contains, after the first quality check, 221,110 F U globally collected observations o f open water, lakes and rivers. Figure 2 shows a ba r chart o f the total num ber o f F U observations collected w ithin a decade and per season. T h e period from 1889 to 1899 period covers eleven years in o rder to gain as m uch available inform ation o f the earliest period o f ocean colour observation.




5 0 °N

4 0 °N

3 0 °N

Figure 3. Example (around Japan) of three defined masks to extract FTAobservations; in white the BM1 mask starting at the land/ sea boundary spreading over the whole sea with on top of this layer, in dark grey, BM2 starting at a distances of > 1 0 0 km o ff land and on top of this layer BM3, a layer in black, starting at a distance of 500 km off land.doi:10.1371/journal, pone.0063766.g003

Data filteringSubsequently, dataset-0 was differentiated into three geograph

ical areas by applying Basin Masks (BM). T h e Basin Masks are used to filter out observations w ithin land boundaries (lakes and rivers), and observations close (100 km or 500 km) to the coast.T h e Basin Masks were created and executed in A rcG IS 9.3 software [36] by m eans o f the A rcG IS clip function.

First, BM1 was created to extract data from dataset-0 free of observations w ithin the land boundaries ju st m entioned. This dataset is referred to as d a tase t-1 and contains 219,839 sea observations. T able 1 shows the distribution o f the F U observations per sea area w ith details on the total and seasonal num bers of F U observations, also expressed as a percentage o f the total. T he definition o f seasons follows the m eteorological convention. W e highlight some results o f the m ajor oceans and seas. T h e selected oceans and seas are the Barents Sea, Bering Sea, East C hina Sea,Ind ian O cean , M editerranean , N orth- and South Atlantic, N orth- and South Pacific, N orw egian Sea, Pacific Coast, Philippine Sea,South C hina Sea, Sea o f Ja p a n , Sea o f O khotsk and the Yellow Sea.

Second, BM 2 was chosen pragm atically to extract open sea observations from dataset-0 , a t a distance o f m ore th an 1 0 0 km off-coast, to avoid effects on w ater colouration o f anthropogenic pressure in the coastal zones such as locally increased nu trien t

Tab le 2. The number of Fl/-observations per dataset unmasked and obtained by the use of 3 data-extraction masks.

M ask - B M I BM2 BM3

Rt-Dataset Dataset-0 Dataset-1 Dataset-2 Dataset-3

Off-Coast + lakes/rivers 0 km >100 km >500km

No. o fO b s . 221110 219839 61434 21971

Off-coast indicates the areas where FL/-data were collected. dol:10.1371/journal.pone.0063766.t002

loading (eutrophication) th a t tends to enhance phytoplankton biomass or high sedim ent loading caused by changes in land use or erosion. These phenom ena result in short-term , drastic colour changes, no doubt influencing the open sea F U values. This dataset is referred to as dataset-2 and contains 61.434 open sea observations. T he last mask, BM 3, was again chosen pragm atically to extract observations from dataset-0 a t a distance o f over 500 km from the coast to include the oceans bu t at the same tim e avoid the influences o f m ixing w ith the differently coloured w ater o f nearby seas. T his dataset will be referred to as dataset-3 and contains 21.971 open ocean observations.

A n exam ple o f the shapes o f the three data extraction masks is given in Figure 3. O ne m ust bear in m ind that by clipping the data to the dim ensions o f the m entioned masks the num ber of observations per sea area dim inish by roughly a factor o f 4 in case o f BM 2 and a factor 10 in case o f BM 3. H ow ever, to establish a realistic colour o f the bulk w ater o f a sea or ocean, no t influenced by its nearby coastal or surrounding seawater (with m uch higher F U num bers) this action is unavoidable. T he data extraction masks, datasets nam e and the num ber o f observations included in the dataset are sum m arized in T able 2. F rom d a tase t-1, a global m ap o f the F77-colour distribution was established th rough an Inverse D istance W eighted interpolation (IDW) [37], [38]. ID W interpolation explicitly im plem ents the assum ption that observations that are close to one ano ther are m ore alike th an those that are fu rther apart. T h e ID W determ ines cell values by calculating a linearly weighted com bination o f nearby F77-observations. T he w eight is calculated by taking the inverse o f the distance betw een the cell and F77-observation, raised to a pow er (usually equal to two). This pow er option controls the significance o f known points on the in terpolated values, based on their distance from the output point. Datasets-2 and -3 are statistically analysed per sea area and per year. D ata were b inned per year to elim inate any influence of a seasonal cycle in the colour o f the sea. T he arithm etic m ean F U values F U w ith the confidence interval o f the m ean (95%) and of the observations (95%) [39] o f each sea were calculated. L inear trends over tim e were calculated by a, to the num ber of observations, weighted- linear regression o f the (FU) values. T he

1 2 0 °E 1 30°E 1 4 0 °E 1 5 0 °E 1 6 0 °E




T able 3. Specifications of the four components used in the Ecolight model for concentration ranges, inherent optical properties and atmosphere.

Ecolight IOP modelspecification Pure w ater Chlorophyll (mg m 3) CDOM (m ’ ) M inerals (g m 3)

Case 1 - ocean chlorophyll dom inated

Pope & Fry 0.1 to 40 Co-varying with chlorophyll -

Case 2 - ocean CDOM dom inated

Pope & Fry Constant with depth 0.1 mg m ~3 Constant with depth varying from 0 to 1 mT1

Constant with depth 0 or 0.2 g rrT3

Specific absorption - £Jp*(z, X) QCDOM*(z, k) ap*(X) for calcareous sand

Specific scattering - b(z, X) - bp*(X) for calcareous sand

Phase function - Morel et al. [56] - Morel et al. [56]

Sky model Normalised radiance by RADTRANX [57] (Harrison and Coombes, 1988)

Diffuse and direct sky by RADTRANX [58] (Kasten and Czeplak, 1980)

Parameters Cloud Fraction = 0, Solar Zenith = S0°, Press = 29.9 in. mere, Day of Year = mean earth-sun distance used, Air Mass Type = 1, Rel.Hum. =80%, Wind Speed = 5m s-1 , Visibly = 15km, Total Ozone =300 Dobson units, Aerosol Opt. Thickness =0.261

doi:10.1371/journal.pone.0063766.t003

w eighted linear regression takes into account the nu m b er o f observations as weights [40],

T o check the presence of a m onotonously increasing or decreasing tren d in the observations the nonparam etric M ann- K endall trend test p e r sea and ocean has been applied as proposed by M ann [41] and further studied by K endall [42] and Sen [43], T h e test was found to be a good tool for trend detection and has been utilized by m any o ther investigators in the analysis o f various types o f environm ental da ta [44], [45], [46], For the statistical analyses including the M ann-K endall test we applied the Excel add-in X LSTA T-softw are (h ttp ://w w w .x lsta t.com ). In the trend analysis all results are presented under the assum ption that the FU- m easurem ents have been m ade independently. T he potential influence o f tem poral and spatial au tocorrelation in the dataset on the simple linear regression analysis has no t been taken into account, m ainly because this is a very com plex and certainly not well defined exercise [47], T herefore the provided tem poral trends

should be treated w ith caution; we are aware of the fact that they provide a simplified description of reality.

Bio-optical ModellingA n F U colour change from a low er to a h igher scale index

indicates a greening o f seaw ater caused by the presence o f light- absorbing substances, notably chlorophyll and C D O M . In order to simulate the relation of chlorophyll or C D O M and the F U index the H ydroLight-EcoLight [48] m odelling software was adopted. This m odel solves the radiation—transfer equations of fight in natu ral waters and com putes radiance or irradiance distributions and derived quantities, like rem ote sensing reflectance i?Rs and F U [49], By varying the m odel’s input param eters, like chlorophyll, C D O M or m ineral concentrations, a relation can be established betw een concentrations and 7?rS or the F U index. H ydroL ight-EcoLight employs m athem atically sophisticated in v arian t im bedding techniques to solve the radiative transfer equation. Details o f this solution m ethod can be found in M obley’s

180” 160” W 140” W 120” W 100” W 80” W 60” W 40" W 20” W 0" 20” E 40” E 60" E 80" E 100” E 120” E 140" E 160” E

Figure 4. The positions o f /^/-observations extracted at a distance > 5 0 0 km off-coast and at a distance o f > 1 0 0 km off-coast (resp. 21971 obs.A, 61434 obs.A). Oceans and seas are Indicated by abbreviations as m entioned In Table 2. Red lines Indicate th e division of north and equatorial regions. A lthough th e M editerranean has a limited num ber of 237 observations, It was found to be Im portant to analyse this sea for which data w ere already collected at th e end of th e 19th century, dolii 0.1371/journal.pone.0063766.g004


http://www.xlstat.com



10000

90 0 0

8000

3000

u

INONA

>2 ON

Eq.A >2 OS <2 ON

NP >2 ON

Eq.P >2 OS <2 ON

Sum

Selected 1314 1304 851 8681 9152 21302

D ri-A utum n 506 212 160 2769 2082 5729

■ n -Sum m er 2 3 7 666 521 2910 2572 6906D n-Spring 221 3 2 7 129 1889 2047 4613

□ n- W in ter 35 0 99 41 1113 2451 4054

Figure 5. The selected 21302 ///-observations extracted under the ocean mask BM3, binned per sea area and per season.dol:10.1 371/journal.pone.0063766.g005

book on Light and Water [50] and in his ‘C om parison o f num erical models for the com putation of underw ater light fields' [51]. Both m odels, H ydroL ight and EcoLight, generate the same output; notice the difference, nam ely th a t H ydroL ight com putes the full 3D -radiance distribution, whereas EcoLight com putes only irradiances.

Tw o sim ulation runs were m ade to establish the influence of either chlorophyll o r C D O M on the colour o f seaw ater i.e. on the rem ote sensing reflectance R r § and on the F U num ber. For

m odelling the oceanic waters, the new IO P m odel “N E W ” case 1 was used. This m odel is based on a com bination o f i) the results of m odelled particle absorption as given by B ricaud et al. [52] and ii) the results o f m odelled particle absorption described in m ore recent publications by M orrison and Nelson [53] and Vasilkov et al. [54], T he “N E W ” case 1 tw o-com ponent IO P m odel consists o f 1) pure w ater [55] and 2) chlorophyll-bearing particles with co- varying C D O M and detritus. O ne o f the m odel's output param eters is the rem ote sensing reflectance R r s(k) w hich was autom atically convoluted with C IE 1931 curves into a chrom aticity coordinate set .v, v and accordingly into a F U scale n u m b er (see [34] for m ore details). In this way a relation betw een chlorophyll and F U could be established for oceanic waters alone, and is used in the trend analysis to convert F U back into a chlorophyll concentration.

For seas we used a generic four-com ponent case 2 w ater specifying concentration profiles and IO P models for each o f four com ponents: 1) pure w ater [55], 2) chlorophyll bearing particles (Op*(K) and b(K) = 0.4070a 7 9 5 0 * (660A ) 1), 3) C D O M given as absorption in m 1 a t 440 nm no t co-varying w ith chlorophyll bearing particles (acDOM*iz , = 1 exp [—.0140 (A,—440)]), and 4)m ineral particles {ap*(k) and bp*(k) o f calcareous sand). For the phase function we used M orel et al. small particle Case 1 phase function [56].

T o simulate the chlorophyll and m ineral concentration in open sea w ater the values o f these properties were set respectively at 0.1 m g nr 3 and 0 or 0.2 g m 3. T he C D O M absorption (a4 4 0) was varied betw een 0.01 and 1.0 m l . Again, the calculated rem ote sensing reflectance Rns(k) was convoluted w ith C IE 1931 curves into a chrom aticity coordinate set .v, v and into a F U scale num ber [34], In T able 3 the EcoLight m odel param eters, including the scattering and absorption properties o f the constituents are tabu la ted for both models.

ResultsFigure 4 shows the geographical positions o f the observations

extracted under respectively BM 3 (> 5 0 0 km, black triangles) and

71Co

>CU71

-QO

20000

cq 15000

10000

5000

I

B H

U

BAR BS ECS INO MED NA >2 ON

Eq.A20S-20N

NP >2 ON

Eq.P20S-20N

NWS PC PHS SCS SJP SOK YS Sum

Selected 709 1251 3129 2597 237 1952 1507 22963 11871 1084 926 7742 927 6740 595 3351 67581Un-W inter 6 6 586 712 33 159 135 3386 3270 24 159 1948 202 891 3 821 12341Mn-Spring 174 37 776 457 13 428 239 5375 2480 184 234 2002 261 1552 48 681 14941□ n-Summer 438 1196 1026 389 73 926 783 8160 3459 808 240 2315 254 2847 433 1222 24569□ n-Autumn 91 12 741 1039 118 439 350 6042 2662 68 293 1477 210 1450 U l 627 15730

Figure 6. The selected 67581 ///-observations extracted under the basin mask BM2, binned per sea area and per season.dol:10.1371/journal.pone.0063766.g006




0.012 -,

0.010

0.008=3

0.006LO

Cc0.004

0.002

0.000

a

0.004

0.003

d0.002

LOo f

0.001

0.000

-0 .1 mg ■■•■■0.5 mg -*-2 mg -1 0 mg — 40 mg

0.Q.0'O*U

Chi = 0 m g n r 3 M inerals = 0 gnv:

— 0.01 m-1 0 .0 2 5 m -l — 0.05 m-1— 0 .2 5 m -l — 1.0 m-1

0.012

0.010

0.008

0.006

^ 0.004 -

0.002

0.000380 420 460 500 540 580 620 660 700 b 3 8 0 4 2 0 4 6 0 5 0 0 5 4 0 5 8 0 6 2 0 6 6 0 7 0 0

Chi = 0.1 m grrr3 M inerals = 0 grrr3

— 0.01 m-1 — 0.05 m-1 — 0 .2 m -l — 0.5 m-1 — 1.0 m-1

0.004 -

0.003 -

> -• 0.002

0.001

Chi = O .lm grrr3 M inerals = 0 .2 g n r 3

— 0.01 m-1 —0.05 m-1 — 0 .2 5 m -l — 0.5 m-1 — 1.0 m-1

0.000380 420 460 500 540 580 620 660 700

Wavelength (nm)380 420 460 500 540 580 620 660 700

Wavelength (nm)

Figure 7. Examples of the Ecolight modelled Rrs (sr 1 ) for case 1 waters (panel a) and case 2 waters (panels b, c and d) w ith variable composition (see also Table 4). In panel (a) only the input chlorophyll concentration is varied betw een 0.1 and 40 mg itT 3. For case 2 w aters the input CDOM4 4 0 absorption (8 4 4 0 ) is varied betw een 0 . 0 1 and 1 . 0 0 m _ 1 with chlorophyll and mineral concentration of bo th 0 (panel b), fixed chlorophyll concentration o f 0.1 mg itT 3 and a mineral concentration o f 0 (panel c) and fixed chlorophyll concentration of 0.1 mg itT 3 and a mineral concentration of 0.2 g itT 3 (panel d). From these Rrs spectral signatures the chrom aticity coordinate set and FU num ber w ere calculated. doi:10.1371/journal.pone.0063766.g007

BM2 (> 100 km, grey-white triangles). Parts o f the globe th a t are under-sam pled are m ainly found in the Southern H em isphere. For further analysis the Adantic and Pacific were split into a nortiiern (NA, NP) and an equatorial (EA, EP) p art, indicated in botii figures by red lines. T h e southern parts o f both oceans were not analysed; for die Soutii A dantic no da ta were available for the period 1900 to 1959 and die rem aining periods were highly undersampled; and as to tile Soutii Pacific observations are lim ited to tile

period 1950 to 1999. H ow ever, during this period, only the last two decades have been sam pled properly. T h e Ind ian O cean has been analysed as one single entity. T h e selected and analysed seas are tile Barents Sea, B ering Sea, E ast C hina Sea, M editerranean, N orw egian Sea, Pacific Coast, Philippine Sea, Soutii C hina Sea, Sea o f Ja p a n , Sea o f Okhotsk, and Yellow Sea. Figure 5 shows tile num ber o f available F U observations o f the selected oceans extracted under BM 3 (> 500 km off-coast). Because the statistical

100▲ Chl=0.0 M inerals= 0 o Chl=0.1 M inerals= 0 • Chl=0.1 M inera ls= 0 .2

y = O.OOOlx4 '4 3

R2 = 0.96

DO

0.1u

0.01“ I--------------1

9 10 1 2 3 4 5 6 7 81 2 3 4 5 6 7 8

A FU B FU

Figure 8. Chlorophyll concentration and CDOM attenuation at 440 nm as a function of FiAnumber. The FU-numbers w ere derived from the Ecolight Rrs spectra. In the left panel (A), chlorophyll w as varied from 0.1 to 40 mg m ~ 3 with a resulting span o f FW\ to FL/10. For case 2 w aters (panel B), the relationship betw een CDOM and FU was calculated for th ree com binations o f chlorophyll concentration (0 or 0.1 mg itT 3) and mineral concentration ( 0 or 0 . 2 g m ~3). dol:10.1371/journal.pone.0063766.g008

PLOS ONE I www.plosone.org June 2013 | Volume 8 | Issue 6 | e63766



Table 4. Results of the weighted linear regression modelling and the Mann-Kendall trend test for the oceans.

O ceans Trend a delta Kendall's p-value Sen 's Period Chi

a tau (Two-tailed) slope start end

NA > 20 BM2 G 0.00692 0.0010 -0 .8 8 <0.0001 -0.0008 1889-1991 0.04 0.75

NA > 20 BM3 G 0.00731 0.0010 -0 .05 0.022 0 1889-1991 0.07 0.82

EQA B M2 G 0.00060 0.0004 -0 .1 9 <0.0001 0 1889-1991 0.33 0.39

EQA B M3 G 0.00031 0.0010 -0 .83 <0.0001 -0.0006 1889-1991 0.35 0.38

NP > 20 B M2 B -0.00163 0.0001 -0 .13 <0.0001 0.0001 1933-1999 0.51 0.40

NP > 20 B M3 B -0.00202 0.0002 -0 .8 7 <0.0001 -0.0001 1934-1999 0.42 0.29

EQP B M2 B -0.00062 0.0001 -0 .0 8 <0.0001 0.0002 1934-1999 0.21 0.17

EQP B M3 B -0.00085 0.0001 -0 .7 6 <0.0001 0 1935-1999 0.22 0.17

INO BM2 B -0.00246 0.0001 -0 .25 <0.0001 0 1898-1993 0.40 0.17

INO BM3 B -0.00165 0.0002 -0.21 <0.0001 0 1898-1993 0.34 0.18

Trends are Indicated by a B (bluing) or a G (greening). Preceding columns show the regression coefficient a with delta a, Kendall's tau, the two-tailed p-value (with a significance level of cz = 0.05) The columns with the heading Chi show the calculated start and end chlorophyll value In mg m~3 over the span of the trend as Indicated In the column named Period (see Figures 13 and 14). dol:10.1371 /journal.pone.0063766.t004

analysis was also done for each separate season, the num ber o f seasonally b inned observations is given too. In Figure 6 these num bers are provided for all seas that were treated under BM2 ( > 1 0 0 km off-coast).

In o rder to understand the oceanic trend analysis o f chlorophyll, we first present the Ecolight m odelling results o f the conversion betw een F U and chlorophyll concentration (chapter 3.1).

H ereafter, a global view o f averaged ocean colour o f the oceans and seas utilizing all observations from 1889 to 2000 is presented in chap ter 3.2. T h e chap ter is followed by a regional trend analysis, first pe r ocean an d then pe r sea.

F U ModellingV arying chlorophyll concentrations were used as input for the

tw o-com ponent case 1 w ater m odel, for the four-com ponent case 2

w ater m odel C D O M as input was varied (a4 4 0 betw een 0 and 1 m *) w ith fixed values for com ponents three an d four m im icking open sea bio-optical values. For each run the EcoLight m odelling software ou tpu t param eters are the spectral rem ote sensing

M onthly m eans (1889-1999) - N orth Atlantic >20°N6.0

5.5

5.0

4.5

g 4.0

3.5

3.0

2.5

2.0

reflectance R-g^s(X), as shown in Figure 7, and the F U num ber shown in Figure 8 .

Figure 7a shows f?RS spectra generated by the case 1 m odel with varying chlorophyll, Figure 7b shows f?RS spectra generated by the case 2 m odel with varying C D O M absorption and chlorophyll and m ineral concentrations set to 0. Figures 7c and 7d show f?Rs spectra generated by the case 2 m odel w ith varying C D O M and chlorophyll set to 0 . 1 m g m an d m inerals set to 0 and to 0 . 2 g m 3 respectively.

T h e results presented in Figure 8 A show a change in chlorophyll concentration from 0.1 to 1 m g m resulting in a shift in F U index from 1 to 4. An exponential fit to the m odel results provides the following relation:

C hl = 0.061 * e 0'666* ^ ' (1)

M onthly m eans (1979-2000) - N orth A tlantic >20°N

la n Feb M ar Apr M ay lu n Jul Aug Sep Oct Nov Dec

-0-NA-BM2 HD-NA-BM3 ^S-Chl-BM3

l.b 1 h

1 4 ___ 1 4

1.2 EM 1.2

1.0 b. 1.0 -

00o

CJ 0.8T!

D.b <U O.bCl)

114 X! 114O

0.2 0.2

0.0 0.0la n Feb M ar Apr M ay lu n lui Aug Sep Oct Nov Dec

O F U (1980-1991) CZCS (1979-1986) SeaWiFS (1997-2000)

Figure 9. Representation of North Atlantic m onthly F U and chlorophyll for the periods 1889-1999 (a) and 1980-1990 (b, only chlorophyll), a - The BM2 (circles) and BM3 (squares) extracted FU da ta with error bars show the sam e pattern and reveal the North Atlantic spring bloom . On the secondary axes, indicated with green line/circles, the m odelled chlorophyll (see Eq. 1) with error bars is show n, b - The m odelled chlorophyll for th e period 1980-1991, com pared to CZCS- (1979-1988) and SeaWiFS (1997-2000) derived chlorophyll, b lended with In-situ data, as p resen ted by Gregg and Conkrlght, 2002). dol:10.1371/journal.pone.0063766.g009



Long-Term Trends ¡n Ocean Colour and Chlorophyll

160°W 140°W 120°W 100°W 80° W 60° W 40° W 20°W 100°E 120°E 140°E 160°E«■■□■■■■■■■■10 11 12 13 14 15 16 17 18 19 20 21

160°W 140°W 120°W 100°W 80° W 60° W 40°W 20°W 20°E 40°E 60°E 80°E I00°E I20°E 140°E 160°E 180°

70°N

50°N

30°N

10°N

10°S

30°S

50°S

70°S

Figure IO. A global IDW interpolation of BM1 clipped FU observations. The w eight is se t to a pow er o f 2. The search radius is set to 4degrees with an o u tp u t grid size of 0.5 degrees. Data collection period covers 1889 to 1999. doi:10.1371/journal.pone.0063766.g010

This relation takes into account the fact th a t in the open ocean chlorophyll bearing particles are co-varying w ith C D O M and detritus, whereas detritus contributes only 5% to 20% to the n o n w ater absorption coefficient [59]. E quation 1 is used to transform the acquired F U into an average chlorophyll concentration per year for oceanic regions. After having consulted a chlorophyll composite o f an entire M O D IS mission, with m axim um chlorophyll concentration o f a few m g m 3, we decided to set the upper lim it o f observed F U to 7 ( ~ 6 m g m 3 chlorophyll) to avoid extrem e outliers w hich could greatly influence the results; rem em ber the exponential relation betw een F U and chlorophyll

(Equation 1). In this way, observations taken above extremely dense phytoplankton blooms (e.g., coccolithophorids) could be avoided. As an exam ple (BM2), for the Ind ian O cean no data were found > F U 1, for the N orth A tlantic 25 observations were om itted and for the N orth Pacific (total o f a round 23,000 obs.) we om itted 240 observations for our chlorophyll calculations.

For the seas, results will be presented as F U only, because they generally have m ore optically active constituents and no single relation betw een F U and chlorophyll can be established. How ever, to dem onstrate the influence o f C D O M , on the FLI scale index we present case 2 m odel results for 3 m odelled p a ram ete r settings.

Equatorial Pacific-EP Indian Ocean-INO

Pacific Coast-PC North Pacific-NP

PhilippineSea-PHS Equatorial Atlantic-EA

M edlterranean-M E D South C hinese Sea-SCS

Sea of Japan-SJP North Atlantic-NA

YellowSea-YS Sea of Okhotsk-SOK

E astC hlnese Sea-ECS Bering Sea-BS

N orw egian Sea-NWS B arents Sea-BAR

FU index

Figure 11. Representation of the oceans and seas identified and arranged in terms of their F U colour as calculated from all available observations extracted under BM2 or BM3 (see Table 4). The Barents Sea is th e m ost greenish sea (bottom ) and th e Equatorial Pacific is th e m ost bluish ocean (top). doi:10.1371/journal.pone.0063766.g011




C hrom ophorlc D issolved O rganic M atter Index C hlorophyll a co n cen tra tio n ( mg / m3 )

Figure 12. An entire MODIS mission composite map of the CDOM Index and chlorophyll-a concentration (4 July 2002-30 Jun 2010).The m ap show s a CDOM index < 0 .5 with the lowest oceanic chlorophyll concentrations of = 0.02 mg m ~ 3 in the South Pacific (white box left). The small white box a t the to p indicates th e open w ater of the Barents Sea with a CDOM index = 3 .5 and chlorophyll concentrations = 5 mg m ~3. Source: level 3 browse, http://oceancolor.gsfc.nasa.gov/. doi:10.1371/journal.pone.0063766.g012

T h e param eter settings are identical to the settings used to generate the results in Figures 7b, 7c and 7d. In Figure 8 B we see that a change in C D O M absorption from 0.01 to 0.1 m _ 1 results in a shift over the first 4 F U scale colours in the case that C D O M , next to w ater itself, is the only absorber o f light. T h rough an exponential fit C D O M 4 4 0 absorption can be calculated w ith a coefficient o f determ ination R 2 = 0.999 according to

C D O M 4 4 0 = 0 .0 0 5 1 * e0J65*FU (2)

Setting bo th the background chlorophyll concentration to 0.1 m g m 3 and the m ineral concentration to zero the F U num bers change from F U = 3 to FU —8 (open circles in Figure 8 B), w hich implies that the low chlorophyll concentration (set a t 0 . 1 m g

m 3) already colours the w ater to a F U value o f 3. This outcom e is som ewhat different from the case 1 m odelling results, shown in Figure 8 A, w here the same chlorophyll concentration corresponds to F U = 1. W e m ust b ear in m ind that in the case 1 m odel, different from the case 2 m odel, C D O M is co-varying with chlorophyll. T h rough a pow er law fit C D O M 4 4 0 absorption can be calculated with a coefficient o f determ ination R 2 = 0.96 according to

C D O M 440 = 0.0001 * F U 4A3 (3)

In the last set o f results o f the case 2 Ecolight m odel, the background chlorophyll concentration was set to 0 . 1 m g m and the m ineral concentration was set to 0.2 g m 3. For F U num bers


http://oceancolor.gsfc.nasa.gov/



Table 5. Results of the weighted linear regression modelling and the Mann-Kendall trend test for the seas.

Seas Trend a delta Kendall's p-value Sen's Period FU

a tau (Two-tailed) slope start end

BAR B -0.0589 0.0100 -0 .056 0.036 0 1925-1992 9.0 5.1

BS B -0.0211 0.0044 -0 .138 <0.0001 0 1935-1998 5.2 4.6

NWS - 0.0016 0.0040 0.023 0.302 0 1889-1990 5.0 5.1

MED G 0.0159 0.0025 0.419 <0.0001 0.0070 1890-1991 1.3 2.9

s e s G 0.0189 0.0020 0.231 <0.0001 0 1933-1994 2.1 3.2

ECS G 0.0293 0.0020 0.103 <0.0001 0.0010 1934-1999 3.1 5.0

PHS G 0.0032 0.0003 0.055 <0.0001 0.0003 1933-1999 2.1 2.3

PC B -0.0909 0.0020 -0 .458 <0.0001 -0.0021 1962-1998 4.6 1.3

YS - -0.0004 0.0010 -0 .027 0.032 0.0006 1933-1996 4.2 4.2

SJP G 0.0098 0.0004 0.171 <0.0001 0.0004 1933-1999 3.1 3.8

SOK G 0.0064 0.0050 0.026 0.404 0 1935-1995 4.2 4.6

Trends are indicated by a B (bluing) or a G (greening). Preceding columns show the regression coefficient a, with delta a, Kendall's tau, the two-tailed p-value (with a significance level of a = 0.05). The columns with the heading FU show the calculated start and end FU value over the span of the trend as indicated in the column named Period (see Figures 15 and 16). doi:10.1371 /journal.pone.0063766.t005

betw een 4 an d 8 the C D O M 4 4 0 absorption can be calculated with a coefficient o f determ ination i ? 2 = 0.99 according to

C D O M 440 = 0 . 0 0 0 0 1 * F U 5'50 (4)

T h e CDOM-TY7 relations above are p resented to help w ith the in terpreta tion and com parison o f a derived F U in seas for which m ore background knowledge of the area-specific inherent optical properties are available.

In the following we show the significance i.e. accuracy and applicability o f the F U scale observations p resented here by visualizing i) an overall (1889-1999) m onthly F U and by ii) visualizing an overall m onthly chlorophyll concentration applying E quation 1. A representation of N orth Atlantic m onthly m eans of F U and chlorophyll, including a com parison o f satellite derived chlorophyll, are shown in Figures 9a and b. N orth A tlantic data, collected betw een 1899 and 1999, w ere extracted under the BM 3 and BM 2 mask to reveal possible divergence betw een bo th datasets, as the latter could be affected by coastal phenom ena influencing the colour o f the sea. T h e num ber o f observations extracted under BM2 accounts for twice the num ber extracted under BM 3. T h e da ta were converted to chlorophyll by E quation 1 and the m onthly F U and chlorophyll were calculated. Sim ilar results were found for b o th the analysed datasets, w ith an

Table 6. The Mediterranean weighted linear regression with modelled chlorophyll, assuming an oligotrophic (case 1) classification.

Period 1890-1991 start end

FUmed = -28.6+0.0159 * Year 1.3 2.9

ChlMED = -6.52+0.00352 * Year 0.13 mg m ~3 0.49 m g m " 3

Chlorophyll could be slightly overestimated due to the presence of CDOM. Chlorophyll values for the Mediterranean increased from the first to last decade by a factor of nearly 4. doi:10.1371 /journal.pone.0063766.t006

identified F U m axim um (spring bloom) in M ay (Figure 9a). C alculated chlorophyll (BM3) shows a concentra tion o f a round 1.4 m g m - 3 in M ay. Also a com parison was m ade betw een the satellite-based and fT7-based chlorophyll seasonal cycle for the years 1980-1989. Figure 9b shows seasonal variations o f the m odelled chlorophyll concentration for the period 1980-1989 and the variation based on C Z C S- and SeaW iFS-derived chlorophyll concentrations, b lended w ith in situ data. T h e agreem ent, also (Figure 9a) with N orth A tlantic b loom values [60], is striking.

From the findings presented above we conclude that F U is a good proxy for chlorophyll in the open ocean. W e have already argued that besides w ater itself chlorophyll (mostly inside the cells o f phytoplankton) gives the ocean its colour [9] ; bu t we m ust bear in m ind that next to chlorophyll the usually non-covarying C D O M can, th rough absorption, have a similar influence on the colour o f the sea an d therefore on FU.

Century averaged ocean colourT o establish a global view o f the F U da ta collected betw een

1889 an d 1999 all observations contained w ithin da tase t-1 were in terpolated according to the ID W technique as described in m aterial and m ethods. T h e inverse distance weight function was set to a pow er law o f index 2. T h e search radius, that limits the num ber o f F U observations used for calculating each interpolated F U value, was set to respectively 2.5 an d 4 degrees with an output grid size o f 0.5 degrees. T he difference betw een results o f interpolation using a search radius o f 2.5 and 4 degrees tu rned out to be negligible, a lthough for under-sam pled seas the applied interpolation technique results in F U values for areas w here no observation were perform ed. A m ap o f ID W interpolated F U values, w ith a search radius o f 4 degrees, is presented in Figure 10. T h e southern oceans, highly under-sam pled, are shown in white.

A first analysis o f Figure 10 shows that the lowest F U num bers are found in the E quatorial Pacific and the Ind ian O cean (FU< 2) and in the E quatorial A dantic (FU<3). N ote the extrem ely high F U values (16 to 21) east o f the O b and Yenisei estuaries in the K ara Sea (75"N, 70"E), m ost likely caused by very high C D O M values [61], Also in the Baltic, w here C D O M is the m ajor light absorber [62], F U num bers o f a round 14 are encountered . T he A tlantic shows increasing F U values (up to F U = 5) tow ards higher



Chl

orop

hyll

(mg

rrr3

)


2.5 2.5North Atlantic >20°N (BM3)Chi = -13.7+7.3E-03*Year

North Atlantic >20°N (BM2)Chi = -13.0+6.9E-03»Year

1.5 1.5

0.5 0.5

1900 1940 .980 p 0 1900 192,0 1940 1960 2000-0.5 -0.5

«-----------------Equatorial Atlantic 20°S-20°N (BM3)Chi = -0.230+3.0E-04‘Year

1

Equatorial Atlantic 20°S-20°N (BM2)Chi = -0.806+6.0E-04*Year1.5 1.5

1

0.50.5

•i ?

0

-0.5-0.51880 1900 1920 1940 1960 1980 20001880 1900 1920 1940 1960 1980 2000

North Pacific >20°N (BM3) f (chl,17)Chi = 4.33-0.00202*year *1

0.6

0.4

0.220 994

1880 1900 1920 1940 1960 1980 2000

1North Pacific >20°N (BM2)Chi = 3.67-0.00163*year

.8

0.6 -»¿O463© © 35? 25i5̂ ,28.2 315,22r

0.4

0.2

0

1880 1900 1920 1940 1960 1980 2000

Equatorial Pacific 20°Sto 20°N (BM2) 13 Equatorial Pacific 20°S to 20°N (BM3)0iiChi = 1.41-0.000620*years 0.4 Chi = 1.88-0.00086*year

82

115Q

6 80 o 33 © 0.3 - ^ 11 14

--- ©15 29

...... - T 31 &39 ...... •...50 7i

20. " " '„ O J ,17̂

- i l _______! ... , > * « > £ ,50 17 «. 1527° i n II P r~|. .... .81 13©.........M.......mi 1 'Hè __

0 . 2 1 1116 °15

‘ 28ci 46 lt>2b©®Ï884..=8 8 5 9 . ^ S . l&S—«--------------- 47 31 ©30 ®i3 13®©,, 1339

Ï 8 ------- ----- \ 26 2

0 . 1 -1880 1900 1920 1940 1960 1980 2000 1880 1900 1920 1940 1960 1980 2000

Year




Figure 13. Ocean trends: The arithm etic mean derived chlorophyll concentration per year (with the no. of obs.) w ith superposed lines: weighted (no. o f obs.) least-squares regression lines (blue or green line) indicate a bluing trend or greening trend of a sea, the 95% confidence interval o f the mean (dotted line) and of the observations (solid line). A sea with no significant trend is indicated by a black line. Regression coefficients are indicated at the to p of each graph. doi:10.1371/journal.pone.0063766.g013

latitudes, m ost likely caused by the influence of m ore extensive phytoplankton blooms.

O f all da ta extracted under either BM 3 or BM 2, the m ean F U 1 8 8 9 1 9 9 9 pe r sea area was calculated, with 95% confidence intervals for the m ean and for the observations. T h e results o f the statistical analysis show that the E quatorial Pacific and Indian O cean are the bluest oceans o f o u r globe with a m ean F U o f 1.6 and 1.7, respectively. W ith the results o f our statistical analysis one can classify an d rank the oceans and greater seas in term s o f their m ean F U colour, presented in Figure 11 (similar to the ID W interpolation o f Figure 10, however, less visible in this coloured graph). This figure shows that in the 20th century the N orth A tlantic O cean has been the greenest ocean of our p lanet while the Barents Sea was the greenest open sea o f all.

In o rder to support these classifications w ith recently collected global data (after 1999), we show in Figure 12 an entire M O D IS mission com posite (4 Ju ly 2002 to 30 J u n 2010) for the C D O M index [63] and for the chlorophyll-o concentrations. N ote that in case 1 waters a m ean relationship exists betw een the C D O M content an d the chlorophyll concentration, anom alies in this relation, for bo th case 1 an d case 2 waters are given by this so- called C D O M index <D [63]. In the hyper-oligotrophic E quatorial Pacific gyre [64], located w ithin the greater white box, we see a <1 X 1 and very low chlorophyll concentrations, that are expected in blue oligotrophic oceans (around 0.02 m g m -3 ). W ith in the smaller white box, indicating the Barents Sea, we see a C D O M index a round 3.5 an d chlorophyll values a round 5 m g m colouring the sea green. Areas with an increasing C D O M index and or increasing chlorophyll concentration (Figure 12) are generally com patible with an increasing .F i/index (Figure 11).

Yearly averaged regional chlorophyll and ocean colourT o establish the influence o f the choice o f Basin M ask on the

tren d analysis for open ocean w ater, we com pared the results o f

dataset-2 (BM2) and dataset-3 (BM3). Because the conversion from F U index to chlorophyll concentration depends on the C D O M absorption, only oceanic waters w ith a C D O M index <1X2.5 were converted with E quation 1; see also the global C D O M composite in Figure 12. For the rem aining seas, as indicated in bold in T able 1, we applied the tren d analyses on BM2 extracted data. For all da ta presented here, per sea o r per ocean, the overall trend covers the whole period o f da ta collection. T h e least-squares regression lines, indicated by a full blue o r green line, indicate either a b luing or greening o f the o cean /sea under investigation. For the oceans b lu ing /g reen ing m eans decrease/ increase in chlorophyll, for the rest o f the seas this can m ean a decrease/increase in F U w hich can either m ean a decrease/ increase in chlorophyll o r C D O M or a com bination o f both.

In the next paragraphs the results o f our statistical analysis are presented per year pe r sea a rea for illustrational purposes only. For the M ann-K endall test all single observations were used (no binning) In some cases, the BM 2 extracted data is enclosed within the boundary of a neighbouring sea or ocean, hence a second data selection was needed, done with the A rcG IS da ta selection tool, to exclude overlapping data. T he total num ber o f extracted F U observations per sea or ocean can therefore be slightly different from the num ber o f observations given in T able 2 and Figures 5 and 6 . D a ta extracted under the BM 3 m ask concerns only the oceans, da ta extracted under the BM 2 m ask concerns bo th oceans and w orld seas.

O cean s. In Figure 13 the results o f the statistical analysis are presented for the A tlantic and Pacific O cean an d in Figure 14 for the Ind ian O cean, per year betw een 1889 and 1999. T he F U has been converted to chlorophyll according to E quation 1. For all oceans analysed, the yearly F U values, including the num ber o f observations, regression coefficients and tren d lines w ith the 95% confidence interval curves for the m ean an d for the observations, are presented in the graphs. At the left-hand side the graphs are

1.2

£ 0 . Í

ÜÛ

0.6

Q . 0.4 Oo

Z 0-2 u

0.5

0.4

0.3

0.2

0.1434430

1880 1900 1920

Indian Ocean(BM2)Chi = 5.08-0.00246*Year

1940 1960 1980

Indian Ocean(BM3)Chi = 3.48-0.00165*Year

2 0 0 0 1880 1900 1920 1940 1960 1980 2000

Year

Figure 14. Ocean trends: the arithm etic mean derived chlorophyll concentration per year (with the no. o f obs.) w ith superposed lines; weighted (no. o f obs.) least-squares regression lines (blue or green line) indicate a bluing trend or greening trend of a sea, the 95% confidence interval o f the mean (dotted line) and of the observations (solid line). A sea with no significant trend is indicated by a black line. Regression coefficients are indicated at the to p of each graph. doi:10.1371/journal.pone.0063766.g014



For

el-U

le


Barents SeaFU = 122-0.059*Year11

9

7

5

3

11880 1900 1920 1940 1960 1980 20002000

8Bering SeaFU = 46.4-0.021‘Year

7

6

5*54'

4

3

2

11900 1920 1940 1960 1980 2000

MediterraneanFU = -28.6+0.016*Year4

25 • • 90

3

2

1

0 ,2000 1880 1900 1920 1940 1960 1980 20001880 1900 1920 1940 1960 1980

5/î'*l-1941 (FU9)South Chinese Sea

FU = -34.5+0.019*Year

4

3

2

11880 1900 1920 1940 1960 1980 2000

East Chinese SeaFU - -53.6+0.029*Year

U S 12889 “

11 «21 ai-RYViiq

1900 1920 1940 1960 1980 20003.5

Philippine SeaFU = -4.14+0.0032*Year

3

--------- . >1°! 29<>12Í120

» J * . 13

—____ .4 ,^ 241«

2.5

2

1.5

1

Pacific CoastFU = 183-0.0914*Year

1880 1900 1920 1940 1960 1980 2000 1880

Year

1900 1920 1940 1960 1980 2000




Figure 15. Sea trends: the arithm etic mean FU values (FU) per year (with the no. o f obs.) w ith superposed lines; weighted (no. of obs.) least-squares regression lines (blue or green line) indicate a bluing trend or greening trend of a sea, the 95% confidence interval o f the mean (dotted line) and of the observations (solid line). A sea with no significant trend Is Indicated by a black line. Regression coefficients are Indicated a t th e to p o f each graph. dol:10.1371 /journal.pone.0063766.g015

based on d a ta extracted u nder the BM 2, graphs on the right are based on d a ta extracted u nder BM 3. T h e num ber o f observations is indicated near the yearly m eans. T h e trend line w ith regression coefficients, w here y is the chlorophyll concentration in m g m and X is the year is indicated in each graph. For the oceans T able 4

shows the regression coefficient a w ith standard deviation delta a, K endall’s tau coefficient (represents the degree o f concordance betw een two colum ns o f ranked data) and the M ann-K endall derived p-value. T h e last two colum ns o f the table shows the period over which observations were collected and the start and end value o f the calculated chlorophyll concentration (mg m ' ).

T h e N orthern and E quatorial Atlantic show a similar greening tren d for bo th BM2 and BM 3. In the late 19th century the m ean chlorophyll concentration a t the surface o f the N orth Atlantic am ounts to 0.06 m g m ' , w hereas the last year o f observations shows a value o f 0.79 m g m 3. T h e overall tren d shows an average grow th (BM2, BM3) o f 0.0071 m g m ' p e r year. T he E quatorial A tlantic shows a sm aller averaged chlorophyll growth o f 0.0005 m g m ' p e r year.

For the N orth Pacific we found a bluing tren d for both BM2 and BM3 with an average decrease o f —0.0018 m g m ' p e r year. T h e chlorophyll concentration o f a round 0.47 m g m ' in 1933 declined to an average o f 0.35 m g m ' in 1999. For the Equatorial Pacific for bo th BM 2 and BM 3, a bluing tren d (on average —.00074 m g m ' per year) was established betw een 1934 and 1999.

T h e results for the Ind ian O cean, presented in Figure 14, show a bluing ocean for bo th BM2 and BM3 d ata w ith an average decay o f —0.0021 m g m ' p e r year. Between 1898 and 1993 the Indian O cean showed significant decline in chlorophyll, from 0.37 to 0.18 m g in ' , at the end o f the observational period.

Seas. For all seas analysed, the yearly F U values, including the nu m b er o f observations, regression coefficients and trend lines with the 95% confidence interval curves for the m ean and for the observations, are presented in Figure 15 and Figure 16. T able 5 shows the general tren d o f a sea, bluing or greening, the weighted (no. o f observations) linear regression coefficient a w ith standard deviation delta a, K endall’s tau and the two-tailed p-value for the

Yellow SeaFU = 5.01-3.9E-04*Year

5

4

• " i 'l r r .........• 76-&ZQ___

Väb• 154

.30

• 1 ' .*.70

6 2* • 47

• 81#9? \• 75« 89

» 100 • 89

2

CUZ) ]_ L 1880 1900 1920 1940CUo

1960 1980 2000

Sea of JapanFU = -15.8+9.8E-03*Year6

5

4

3

2

11900 1920 1940 1960 1980 2000

9Sea of OkhotskFU = -8.17+6.4E-03*Year8

7

6

5

4

3

2

11880 1900 1920 1940 1960 1980 2000

Year

Figure 16. Sea trends: the arithm etic mean FU values (FU) per year (with the no. o f obs.) w ith superposed lines; weighted (no. of obs.) least-squares regression lines (blue or green line) indicate a bluing trend or greening trend of a sea, the 95% confidence interval o f the mean (dotted line) and of the observations (solid line). A sea with no significant trend Is Indicated by a black line. Regression coefficients are Indicated a t th e to p o f each graph. dol:10.1371 /journal.pone.0063766.g016




analysed seas. T h e last two colum ns of the table shows the period over w hich observations w ere collected an d the start an d end value o f the F U num ber. BM2 was used as da ta extraction mask. T he F U values are no t converted into chlorophyll concentrations in waters w ith a C D O M index C » 2 .5 (Figure 12), w hich occurs in m ost o f the seas. In these seas a change in C D O M as well as chlorophyll can cause shifts in sea colour.

Figure 15 shows for bo th the n o rth ern Barents and Bering Sea a b luing trend. T h e Barents Sea, the greenest open sea o f our planet, shows a d ram atic b luing (p = 0.036) with a factor o f —0.059 F U pe r year from — 9.0 (1925) to —5.1 (1992). Between 1935 and 1998 the Bering Sea is b luing (pCO.OOOl) w ith a factor o f —0.0211 F U per year from —5.2 to —4.6. T h e N orw egian Sea, geographically situated below the Barents Sea, has an average F U value o f 5.05 an d shows no significant colour change (p = 0.302) over the period 1889-1990.

T h e M ed iterranean was geographically split into a w estern and eastern part, based on the know n existence o f an east-west oligotrophy gradient a t 15° East [65], [66], [67]. In total 824 F U observations (BM1) betw een 1889 and 1999 were collected. From these only 237 observations (BM2) were analysed. O u r analyses shows th a t the w estern p a r t is slighdy greener com pared to the eastern part; FUMED.West- FUuED.East+O-^FU. BM1 an d BM2 related F U bo th show a significant greening betw een 1890 and 1991. T h e results for the BM2 dataset are plo tted in Figure 15: the whole M ed iterranean has been greening significantly over the past century, with an increase o f 0.016-FI7 per year since 1890. (p<0.0001). Interestingly, this greening ra te is alm ost similar to the results found for the N o rth A tlantic (BM3) with an increase o f 0.02F U per year. As an indication o f the potentially related changes in biomass, the F U da ta o f the M ed iterranean were converted into chlorophyll, assum ing its w ater being C ase l (see T able 6). W ith a F U = 1 .3 during 1890 and a F U = 2.9 in the last year o f observation 1991 M ed iterranean chlorophyll increased from 0.13 to 0.5 m g m - 3 .

Both the South and East Chinese Sea show greening trends (pCO.OOOl for both) betw een 1933 an d 1999; the East Chinese Sea, being greener, even shows an increase o f alm ost two F U scale num bers. T he Philippine Sea shows m inim al greening for the period 1933 (FU= 2.1) - 9 9 {FU= 2.3), b luing w ith 0 .0 0 3 2 T I/p e r year. Anom alies are identified in the periods 1940M 9 and 1950— 59; however, F U values observed during these years are based upon a lim ited num ber o f observations com pared to the rest o f the periods.

D a ta o f the Pacific C oast are da ta o f open sea although the nam e o f the a rea implies som ething else. No o ther sea investigated by us showed such a m ajor change in colour, from F U = 4.6 to 1.3 over the relatively short period 1962-98. This sea is bluing, w ith a ra te o f —0.09F U per year (p = <0.0001). T h e results o f the rem aining seas (Yellow Sea, Sea o f ja p a n and the Sea o f Okhotsk) are shown in Figure 16 and can be found a t the end o f T ab le 5. T h e central Yellow Sea (p = 0.0316) is no t as yellow as its nam e w ould suggest. No significant colour change could be detected; the first observations were collected betw een 1933 an d 1939 an d have a F U = 4.2, the same value still holds, after 60 years, for the last observational year 1996.

For the Sea o f ja p a n an overall greening is found betw een 1933 and 1999 (pCO.OOOl). T h e Sea o f Okhotsk, a bluish-green sea, connected with the Sea o f Jap an , shows an overall greening betw een 1935 and 1995 w ith an undulating pa tte rn in F U not seen so pronounced in o ther seas (m inor trend, p = 0.4).

W e emphasise th a t the trend analysis presented here is only a simplified description of the data. It was already shown in a

previous p ap er on F U m easurem ents m ade in the N orth Pacific [33] th a t oceans can show a m uch m ore com plex tem poral behaviour.

Summarising the resultsLowest open ocean (> 500 km off-coast) chlorophyll values o f

—0.1 m g m 3 were found in the N orth A dantic betw een 1889 and 1899. T h e N orth A dantic chlorophyll shows on average a grow th o f 0.071 m g m 3 pe r year. T h e Equatorial A dantic greens less dram atically w ith an am oun t o f 0.0006 m g m 3 pe r year. Both the N orth- and E quatorial Pacific show a decrease in chlorophyll betw een 1933 and 1999, w ith the E quatorial Pacific b luing a t abou t ha lf the rate o f the N orth Pacific. T h e Indian O cean shows a continuing decline in chlorophyll over the period 1898 and 1993. Bluing seas are the Barents Sea and the Bering Sea. T h e Pacific C oast shows an extrem e an d fast b luing over a relative short period betw een 1962 and 1998. Like the A dantic, the M ed iterranean shows a greening tren d over the analysed period. A t the o ther side o f the globe similar greening trends are found for the East- and South Chinese Sea, and, albeit less extrem e, the Philippine Sea and the Sea o f Okhotsk. T h e Sea of J a p a n shows a distinct an d continued greening over the period 1933 to 1999. T h e N orw egian Sea and the Yellow Sea are, in term s of long-term colour changes, less affected seas.

Discussion and Conclusions

W e conclude th a t the ocean colour dataset o f Forel-U le scale observations th a t has been established over the period 1889 to 1999 offers a unique opportun ity to investigate worldwide tem poral changes o f colour a t the sea and ocean surface in the 20th century, from 1889 to present; long-term colour changes that we show can safely be related to concentrations o f chlorophyll, the m ost used phytoplankton biomass proxy. Bio-optical m odelling, based on the state-of-the-art radiation transfer code called ‘EcoLight’ th a t is based on a tw o-com ponent and four-com ponent m odel has dem onstra ted how the concentration o f chlorophyll and C D O M relates to ocean colour (FU scale index). H ow ever, we m ust keep in m ind th a t the m odelled chlorophyll behaves exponentially. This m eans th a t in case a F U observation was done over an oceanic p lankton b loom (FU5-FU7, personally observed), the m odel generates for a FU5 observation 1.7 m g m -3 o f chlorophyll and for a F U I observation 6.5 m g m -3 of chlorophyll. Indeed, for the clearest oceanic waters (FU = 1 to 4), w here algal pigm ents a re the dom inant factor o f light absorption, a highly significant exponential relation was found betw een the F U index and chlorophyll concentration (Figure 8A). T hus, significant changes in the average chlorophyll concentration pe r year could be reconstructed (Figures 13 and 14). A lthough, for w ater closer to land (< 200 km off-coast) the FU-Chlorophyll relation can be im proved by a better estim ate o f the C D O M index an d specific inheren t absorption by phytoplankton in each ocean, the observed variation in colour is real an d indicates a change in phytoplankton biomass a t the surface.

T h e Forel-Ule classification o f na tu ra l waters, th a t was started long ago, in the late n ineteenth century, can, as we have shown, be regarded as an oceanographic record as reliable for interpretation as the long-term records o f salinity, tem perature and w ater transparency th a t are often used to relate ocean w ater properties to long-term clim ate changes th a t receive so m uch a ttention since the IPC C reporting on this phenom enon . T h e ocean colour record o f the FU-scale observations can hardly be replaced by o ther data if only because it goes back all the way to 1889. W e have shown here th a t observations m ade with this colour scale are no t simply




‘subjective’, as earlier has been stated w hen Secchi disc visibility records w ere evaluated [68].

T h e Forel-U le classification accuracy is highlighted by our analysis o f the N orth A dantic dataset; the annual cycle o f Forel- U le ocean colour and derived chlorophyll results (Figure 9b) are fully in line with the annual cycles established by satellite ocean colour sensors [60]. Also for the N orthern Pacific the TU-based seasonal changes o f calculated chlorophyll concentration com pare well w ith the values o f chlorophyll changes reported for this region [33] (W ernand and van der W oerd, 2010b).

W e are well aw are o f the fact th a t in some seas the num ber o f observations available an d used for our analyses is ra th e r low, bu t they are a t the same tim e the only da ta related to phytoplankton biomass change in these ocean regions. T he M ann-K endall analysis was perform ed to evaluate the ocean colour o r chlorophyll tren d of an ocean or sea. T h e outcom e of the analysis indicates th a t the N orth A dantic and E quatorial A dantic have been greening over the whole past century. T h e increase in N orth A dantic chlorophyll after the 1950’s is consistent w ith the trend of increase in the Phytoplankton C olour Index (PCI) obtained during the C ontinuous Plankton R ecorder (CPR) survey over the past 50 years [14] [69], [70], [71], [72]. Fu rther evidence of a greening A dantic is given by Edw ards et al. [73] in a ‘SA H FO S ecological status report’ o f 2 007 /2008 (‘there has been a large increase in phytoplankton since the late 1980s in most regional area o f the North Atlantic*). For the five oceans under investigation a significant trend was found (pCO.OOOl, significance level o f 0.05). O u t o f eleven investigated seas seven show trends with a pCO.OOOl.

O u r findings do no t indicate a global tren d in the ocean’s colour as established on the basis o f the F U record, an d therew ith in its chlorophyll contents. W e canno t confirm the recent report o f an ocean-wide phytoplankton decrease [23]; Boyce et al. found a global decline o f phytoplankton over the past century th a t they related to increasing sea surface tem perature. Differences in data filtering techniques m ay have been the cause o f the considerable discrepancy betw een the results o f Boyce et al. [23] and the results we present here. Boyce et al. based their conclusions on datasets from w hich only da ta obtained in waters < 2 5 m deep or < 1 km from the coast w ere excluded. This m eans th a t their results close to coasts m ay have been biased by chlorophyll-rich eutrophic coastal zones. This will have an im pact on trends in oligotrophic regions, w here chlorophyll varies m uch less. In contrast to Boyce et al. [23], we have om itted data < 1 0 0 km (seas and oceans) o r < 5 0 0 km (oceans) from the coast.

H enson et al. [74] recenüy concluded th a t detection o f clim ate change-driven trends in the satellite data record is confounded by the relatively short tim e series and large inter-annual and decadal variability in productivity. A ccording to H enson et al., recenüy observed changes in chlorophyll, p rim ary production and the size o f the oligotrophic gyres canno t be expliciüy ascribed to the im pact o f global clim ate change. T hey suggest th a t tim e series o f at least 40 years are needed to distinguish trends from natural variability.

G regg an d C onkright [60] re-analysed the C Z C S global ocean chlorophyll p roduct, using SeaW iFS com patible atm ospheric correction m ethods, and m ade a first quantitative com parison of the decadal trends in global ocean chlorophyll over the periods 1979-1986 (CZCS) and 1997-2000 (SeaWiFS). T hey reported a decrease in global chlorophyll concentrations from the C Z C S records to p resent o f abou t 6%; they found larger reductions in the n o rth ern high latitudes and an increase in chlorophyll in the low latitudes. T h eir conclusions are no t in line with ours b u t we m ust b ear in m ind th a t results are based upon the results o f two relatively short periods - m uch shorter than the shortest period

presented by us. A ntoine et al. [19] also used C Z C S an d SeaW iFS data; they reported an overall increase o f the w orld ocean average chlorophyll concentration by abou t 22% .

RecommendationsT h e dataset presented here contains 220,000 Forel-Ule

observations; 160,000 rem ain unexploited for the tim e being. It w ould be of great interest to investigate this latter dataset, collected in coastal and shelf seas (< 1 0 0 km off-shore) globe-wide, to establish the possible influence o f coloured n ear shore waters on open-ocean colour; m oreover, w ith a refinem ent o f the m odel results presented here it should be possible to classify w ater quality no t only in the open ocean b u t also in seas w here coloured dissolved organic m atter (CDO M ) is no t co-varying with chlorophyll, usually sea regions w here hu m an influence plays a role, notably eutrophication by nu trien t inpu t o f rivers, an d by sedim ent an d C D O M load changes. How ever, we m ust b ear in m ind th a t such a refinem ent is no t an easy task as chlorophyll and C D O M absorption can have a similar influence on sea an d ocean colour an d therefore on F U values. Future work should concentrate on explanations o f driving forces beh ind the decadal trends in open-ocean colour presented here, and anom alies in the ocean colour record over long term s o f up to a century. H indcasting is the only way to understand and pred ic t changes in the ocean’s ecosystem in its relation to clim ate change.

Long-term series are lim ited; a literature search indicates that ‘long-term ’ generally m eans: betw een 5 and a t m ost 30 years. It is therefore no t only challenging b u t also necessary to com bine archived long-term data series o f basic w ater quality param eters such as Secchi disc visibility depth, chlorophyll, o r phytoplankton abundance (e.g., the colour index recorded since 1948 during the C ontinuous P lankton R ecorder Survey of SA HFO S) to establish in ter-annual o r inter-decadal m eans from w hich results can be com pared to our results. An attem pt to com bine historic w ater quality datasets has been presented recenüy by Boyce et al. [75] (2012). D a ta o f o ther long-term tim e series, nam ely the Hawaii- and B erm uda A dantic O cean Tim e-Series (H O T S and BATS, b o th since 1988) and the California C ooperative O ceanic Fisheries Investigations (C alC O FI since 1949) should help to achieve this goal.

Finally, we like to m ake a plea, substantiated in the present paper, for the réin troduction of the TtZ-scale to classify the colour o f seas an d oceans. O cean colour classification by m eans o f a single num erical value instead o f a hyper-spectral classification will facilitate the in terpreta tion of long-term ocean colour data series and a t the same tim e facilitate a connection betw een the present and the past. M E R IS satellite Forel-Ule m apping can play a role here, th rough validation and coupling o f historic observations to the p resent e ra o f satellite-based m onitoring.

AcknowledgmentsThe authors wish to thank Tim Boyer (NOAA, NODC) for providing the bulk of the Forel-Ule data; the reviewers provided highly valuable comments that have been essential for improving the text, tables and figures.

Author ContributionsConceived and designed the experiments: M RW HJvdW W W CG. Performed the experiments: MRW . Analyzed the data: M R W FfJvdW. Contributed reagents/m aterials/analysis tools: M R W HJvdW. W rote the paper: M R W HJvdW W W CG.




References1. P a rk e r BB (1992) O c e a n o g rap h ic d a ta a rchaeo logy : F in d in g h isto rica l d a ta for

c lim ate a n d g lobal c h a n g e research . O c e a n o g ra p h y 5 (2): 124—125.2. M ille r C B (2003) O c e a n Ecology a n d g lobal c lim ate ch an g e. B iological

O c e a n o g rap h y , W iley-B lackw ell. 416 p.3. N R C (2001) N a tio n a l R e se a rc h C o u n c il, C o m m itte e o n E a r th S tudies, Space

S tudies B o ard , Issues in th e In te g ra tio n o f R e se a rc h a n d O p e ra tio n a l Satellite System s for C lim a te R esearch : P a r t I. S cience a n d D esign . N a tio n a l A cadem ies Press, 153 p .

4. S a ich u n T , G u an g y u S (2006) S a tellite-derived p r im a ry p ro d u ctiv ity a n d its sp a tia l a n d tem p o ra l variab ility in th e C h in a seas. J o u rn a l o f G eo g rap h ica l Sciences, Science C h in a P ress 16 (4): 4 4 7 -4 5 7 .

5. H u o t Y , B ab in M , B ru y a n t M F , G ro b C , T w ard o w sk i M S , e t al. (2007) D oes ch lo ro p h y ll a p ro v id e th e b e s t in d ex o f p h y to p la n k to n b iom ass for p r im a ry p ro d u ctiv ity studies? B iogeosciences 4: 707—745.

6. G reg g W W , C o n k rig h t M E , G in o u x P, O ’R eilly J E , C asey N W (2003) O c e a n p r im a ry p ro d u c tio n a n d clim ate: G lo b a l d e c ad a l changes. G eophysica l R esea rch L ette rs 30 (15): 1809. D O I: 1 0 .1 0 2 9 /2 0 0 3 G L 0 1 6 8 8 9 .

7. W M O (2003) T h e seco n d r e p o r t o n th e a d eq u acy o f the g lobal observ ing system s for c lim ate in su p p o rt o f th e U N F C C C . W o rld M eteo ro lo g ica l O rg a n iz a tio n . E x ecu tiv e su m m ary , W M O /T D N o . 1143. 85 p.

8. Y a n X H , S chübel J R , P r itc h a rd D W (1990) O c e a n ic u p p e r m ix ed lay e r d e p th d e te rm in a tio n b y the use o f satellite d a ta . R e m o te Sensing o f E n v iro n m e n t 32 (1): 5 5 -7 4 .

9. M o re l A, P r ie u r L (1977) A nalysis o f v a ria tio n s in o c e an color. L im nology a n d O c e a n o g ra p h y 22 (4): 7 0 9 -7 2 2 .

10. S h en g g u an g Y , D ay in g X , X ia o lo n g Y ,J u n Z (1991) E ffect o f a lgae a n d w a te r o n w a te r co lo r shift. C h in ese J o u rn a l o f O cean o lo g y a n d L im nology 9 (1): 49—56.

11. P lass G N , H u m p h rey s T J , K a tta w a r G W (1978) C o lo r o f th e o cean . A pp lied O p tics 17: 1 4 3 2 -1 4 4 6 .

12. W e rn a n d M R , G ieskes W W C {2011) O c e a n O p tic s fro m 1600 (H udson) to 1930 (R am an ). Sh iftin g in te rp re ta t io n o f n a tu ra l w a te r c o lo u rin g . U n io n des O c é a n o g rap h e s d e F ran ce , Paris , F ran ce . 87 p . ISB N : 978-2 -9510625-3-5 .

13. B a tten S D , B urkill P H (2010) T h e C o n tin u o u s P la n k to n R eco rd er: to w ard s a g lobal persp ec tiv e J . P la n k to n R es. 32 (12): 1 6 1 9 -1 6 2 1 . D O L IO .1 0 9 3 /p la n k t/ f b q l4 0 .

14. H in d e r SL, H ay s G C , E d w ard s M , R o b e r ts E C , W a ln e A W , e t al. (2012) C h a n g e s in m arin e d inoflagellate a n d d ia to m a b u n d a n c e u n d e r c lim ate change. N a tu re C lin i. C h a n g e L ette rs 2: 2 7 1 -2 7 5 . D O I: 1 0 .1 0 3 8 /n c lim a te l3 8 8 .

15. K a r l D M , M ichae ls A F (1996) O c e a n T im e Series: R esu lts f ro m th e H a w a ii a n d B e rm u d a R e se a rc h P ro g ram s, D e e p Sea R e se a rc h P a r t II: T o p ic a l S tudies in O c e a n o g ra p h y 43: Issue 2—3.

16. B id igare R R , C h a i F , L a n d ry M R , L ukas R , H a n n id e s C C S , e t al. (2009) S u b tro p ica l o cean ecosystem s tru c tu re ch an g es fo rced b y N o r th Pacific c lim ate v aria tio n s . J . P la n k to n R es. 31 (10): 1 1 3 1 -1 1 3 9 . D O I: 1 0 .1 0 9 3 /p la n k t/fb p 0 6 4 .

17. B o g rad SJ, C h eck ley J r D A , W o o ste r W S (2003) C a lC O F I: a h a lf c e n tu ry o f physical, ch em ical, a n d b io log ical re se a rch in th e C alifo rn ia C u r re n t System . D e e p S ea R e se a rc h P a r t II: T o p ic a l S tudies in O c e a n o g rap h y 50 , Issues 14—16, A ugust: 2 3 4 9 -2 3 5 3 .

18. M u e lle r J L , D avis C , A rn o n e R , F ro u in R , C a rd e r K , e t al. (2003) A b o v e-W ater R a d ia n c e a n d R e m o te S en sin g R e f le c tan c e M e a s u re m e n t a n d A nalysis P ro tocols. In R a d io m e tric M easu rem en ts a n d D a ta A nalysis, O c e a n O p tics P ro toco ls F o r Satellite O c e a n C o lo r Sensor, V a lid a tio n , R ev ision 4, P ro tocols N A S A /T M -2 0 0 3 - 1 6 2 1 /R e v , J . L. M u elle r, G . S. F a rg io n , a n d C . R . M cC la in , eds., 3, G reen b e lt: 21—30.

19. A n to in e D , M o re l A , G o rd o n H R , B an zo n V , E vans R H (2005) B ridg ing o cean c o lo u r o b serva tions o f th e 1980s a n d 2000s in se arc h o f lo n g -te rm trends.

J o u rn a l o f G eophysica l R e se a rc h 110: 1 -2 2 . D O I: 1 0 .1 0 2 9 /2 0 0 4 JC 0 0 2 6 2 0 .20. V en rick E L , M cg o w an J A , C a y a n D R , H a y w a rd T L (1987) C lim a te a n d

ch lo ro p h y ll a: L o n g -te rm tren d s in the C e n tra l N o r th P acific O c e a n . S cience 238: 7 0 -7 2 .

21. G ieskes W W C , K ra a y G W (1983) U n k n o w n ch lo ro p h y ll-a derivatives in the N o r th -S e a a n d th e tro p ic a l A d a n tic -O c e a n rev e a led b y H P L C analysis. L im nology a n d O c e a n o g ra p h y 28: 757—766.

22. Falkow ski P , W ilson C (1992). P h y to p lan k to n p ro d u ctiv ity in th e N o r th Pacific O c e a n since 1900 a n d im plica tions fo r ab so rp tio n o f an th ro p o g e n ic C 02 . N a tu re 358: 7 4 1 -7 4 3 .

23. Boyce D G , Lew is M R , W o rm B (2010) G lo b a l p h y to p la n k to n dec line o v er the p a s t cen tu ry . N a tu re 466: 5 9 1 -5 9 6 .

24. F o re l FA (1890). U n e nouvelle fo rm e d e la g am m e de co u leu r p o u r l’é tu d e de l ’e a u des lacs. A rchives des Sciences P hysiques e t N a tu re lle s /S o c ié té d e P hysique e t d ’H is to ire N a tu re lle de G en èv e , V I: 25.

25. U le W (1892) D ie B estim m u n g d e r W asse rfa rb e in d en Seen . K le in e re M itth e ilu n g en . D r. A. P e te rm an n s M itth e ilu n g en aus Ju s tu s P e rth es g eo g ra p h isch e r A nsta lt, G o th a : J u s tu s P erthes: 70—71.

26. K rü m m e l O (1893) D ie F a rb e d e r M eere . G eophysikalische B e o b a c h tu n g e n d e r P la n k to n -E x p e d itio n . E rg eb n isse d e r P la n k to n -E x p e d itio n d e r H u m b o ld t- Stiftung , Bd. I -C . K ie l u n d Leipzig: V e r la g v o n L ipsius & T isch er. T a fel II: 8 9 -1 1 0 .

27. L u k sch J (1901) U n te rs u c h u n g e n ü b e r d ie T ra n s p a re n z u n d F a rb e de Seew assers. W issen sch afd ich e E rgebn isse X IX . E x p ed itio n en S .M . Schiff “ P o la ” im M itte llän d isch en , Ä gäischen u n d R o th e n M eere in d e n ja h r e n 1890—1898.

B erich te d e r C o m m iss io n fü r O c e a n o g ra p h isc h e F o e sch u n g en . C o llectiv - A u sg ab e aus d e m L X IX B an d e d e r D en k sch riften K a iserlich en A k ad em ie d e r W issenschafte . A . F o rsch u n g e n im R o th e n M eere . B. F o rsch u n g en im O sd ic h en M itte lm eere . W ien : H of- u n d S taatsd ru ck erei. W ien . 4 0 0 -4 8 5 .

28. V o n D rygalsk i, E D (1897) G ro n lan d -E x p e d id o n d e r G esellschaft fü r E rd k u n d e zu B erlin , 1891—1893 u n te r L e itu n g v o n E rich v o n D rygalsky. Z w eiter B a n d m it 31 A b b ild u n g en im T e x t, 9 T a fe ln u n d 1 K a r te Berlin: W . H . K ü h l. K a r te 10: 3 1 8 -3 2 0 .

29. U S N av v H y d ro g ra p h ie O ffice (1951a) M a rin e g e o g rap h y o f th e Sea o f lap a n . P u b lica tio n N o . 757: 4 8 -5 0 .

30. U S N av y H y d ro g ra p h ic O ffice (1951b) M a rin e g eo g rap h y o f K o re a n w aters. P u b lica tio n N o . 752: 8 -9 .

31. U S N av y H y d ro g ra p h ic O ffice (1951c) M a rin e g e o g rap h y o f In d o ch in ese w aters. P u b lica tio n N o . 754: 36.

32. F red erick M A (1970) A n A das o f Secchi D isc T ra n sp a re n c y M easu rem en ts a n d F ore l-U le C o lo r C o d es fo r th e O c e a n s o f th e W o rld . M a s te r’s thesis. N av a l P o s tg ra d u a te School: U n ited States. 188 p.

33. W e rn a n d M R , V a n d e r W o e rd H J (2010b) O c e a n co lo u r ch an g es in the N o rth P acific since 1930. J o u rn a l E u ro p e an O p d c a l Society R a p id P u b lic a d o ns 10015s.

34. W e rn a n d M R , V a n d e r W o e rd H J (2010a) S p ec tra l analyses o f th e F orel-U le O c e a n co lo u r c o m p a ra to r scale. J o u rn a l E u ro p e a n O p tic a l S ociety R a p id P u b licad o n s 10014s.

35. B oyer T P , A n to n o v J I , G a rc ia H E , J o h n s o n D R , L o c a rn in i R A , e t al. (2006) W o rld O c e a n D a ta b ase 2005. E d. S. L evitus, N O A A A d as N E S D IS 60 , U .S . G o v e rn m e n t P rin tin g O ffice, W a sh in g to n , D .C ., 190 p ., D V D s.

36. E S R I (2008) W h a t is A rc G IS 9.3? A rcG IS 9 ” . R e d la n d s (USA): E S R I, 119 p.37. Philip G M , W a tso n D F , (1982) A Precise M e th o d fo r D e te rm in in g C o n to u re d

Surfaces. A u s tra lian P e tro le u m E x p lo ra d o n A ssoc ia tion J o u rn a l 22: 205—212.38. W a tso n D F , P hilip G M (1985) A R e fin e m e n t o f In v erse D is tan ce W eig h ted

In te rp o la d o n . G eo p ro cessin g 2: 315—327.39. Sokal R R , R o h lf FJ (2012) D esc rip d v e statisdes. B iom etry , th e p rin c ip le s a n d

p rac tic e o f statisdes in b io log ical resea rch 4 th ed itio n , F re em a n a n d C o , N ew Y ork. 937 p.

40. B ev in g to n P R (1969) D a ta red u c tio n a n d e rro r analysis fo r the p hysica l sciences. M c G raw -H ill, N ew Y ork. 336 p.

41. M a n n H B (1945) N o n p a ra m e tric tests ag a in st tren d . E c o n o m etric a 13: 2 4 5 -2 5 9 .42. K e n d a ll M G (1975) M u ltiv a ria te A nalysis. C h a r le s G riffin & C o , L o n d o n . 210 p.43. Sen P K (1968) E stim ates o f th e R egression C oeffic ien t b a sed o n K e n d a ll’s T a u .

J . A m er. S tatist. Assoc. 63: 1379—1389.44. H ip e l K W , M cL eo d A I, Fosu P (1983) E m p irica l p o w er c o m p ariso n s o f som e

tests for tre n d , in statistical A spects o f W a te r Q u a lity M o n ito rin g , D evelopm ents in W a te r S c ien ce 27: 3 4 7 -3 6 2 , E d ite d b y A .H .E l-S h a a ra w i a n d R .E . K w iatkow ski.

45. M cL eo d A I, H ip e l K W , B odo BA (1990) T re n d analysis m e th o d o lo g y fo r w ater q u a lity tim e series. E n v iro n m etrics 2: 169—200.

46. S alm i T , M a a ta A , A ntilla P , R u o h o -A iro la T , A m n ell T (2002) D e te c tin g tren d s o f a n n u a l values o f a tm o sp h e ric p o llu tan ts b y the M a n n —K e n d a ll tes t a n d S e n ’s slope e stim ates — T h e E xcel T e m p la te A p p lica tio n M A K E S E N S . F inn ish M eteo ro lo g ical In stitu te , H elsinki, F in lan d , 35 p.

47. B eale C M , L e n n o n J J , Y e a rs le y JM , B rew er M J, E ls to n D A (2010) R egress ion analysis o f spa tia l d a ta . Ecology L ette rs 13: 246—264.

48. M o b ley C D , S u n d m a n L K (2008) H y d ro L ig h t 5 -E co L ig h t 5 U sers G uide. S eq u o ia S cientific Inc ., U S . 99 p.

49. M o b ley C D (2010) C o m p u ta t io n o f th e F o re l-U le scale co lo r in d ex in H y d ro L ig h t. H y d ro L ig h t T e c h n ic a l N o te 8. 3 p .

50. M o b ley C D (1994) L ig h t a n d W ate r: R a d ia tiv e T ra n sfe r in N a tu ra l W aters. A cad em ic Press, S an D iego. 59 2 p.

51. M o b ley C D , G en tili B, G o rd o n H R , J in Z , K a tta w a r G W , e t al. (1993) C o m p a riso n o f n u m erica l m odels fo r th e c o m p u ta tio n o f u n d e rw a te r l ig h t fields. A p p lied O p tics 32 (36): 7 4 8 4 -7 5 0 4 .

52. B ricau d A, M o re l A , B ab in M , A llali K , C lau s tre H (1998) V a ria tio n s o f l ig h t a b so rp tio n b y su sp en d ed partic les w ith th e ch lo ro p h y ll a c o n c en tra tio n in o cean ic (C ase 1) w aters: analysis a n d im plica tions fo r b io -o p tica l m odels. J o u rn a l o f G eo p h y sica l R e se a rc h 103: 31 ,033-31 ,044 .

53. M o rriso n J R , N e lso n N B (2004) S easonal cycle o f p h y to p la n k to n U V ab so rp tio n a t th e B e rm u d a A d a n tic T im e-series S tudy (BATS) site. L im nology a n d O c e a n o g ra p h y 49 (1): 2 1 5 -2 2 4 .

54. V asilkov A P , H e rm a n J R , A h m a d Z , K a h ru M , M itch e ll B G (2005) A ssessm ent o f th e u ltrav io le t rad ia tio n field in o c e an w aters fro m space-b ased m easu rem en ts a n d full rad ia tiv e -tran sfe r calcu lations. A p p lied O p tics 44: 2863—2869.

55. P o p e R M , F ry E S (1997) A b so rp tio n sp ec tru m (3 8 0 -7 0 0 nm ) o f p u re w ater. II. In te g ra tin g cavity m easu rem en ts . A p p lied O p tics 36: 871 0 —8723.

56. M o re l A, A n to in e D , G en tili B (2002) B id irec tio n a l R eflec tan ce o f O c e a n ic W aters: A cco u n tin g fo r R a m a n E m ission a n d V a ry in g P artic le S ca tte rin g Phase F u n c tio n . A p p lied O p tics 41: 6 2 8 9 -6 3 0 6 .

57. H a rr iso n A , C o o m b es C (1988) A n o p a q u e c lo u d cover m o d el o f sky sh o rt w av elen g th rad ian ce . Sol. E n e rg y 41 (4): 3 8 7 -3 9 2 .

58. K a s ten F, C zep lak G (1980) S o lar a n d te rres tria l rad ia tio n d e p e n d e n t o n the a m o u n t a n d type o f cloud. Sol. E n e rg y 24 (2): 1 7 7 -1 8 9 .




59. N e lso n N B , Siegel D A , M ich ae ls A F (1998) S easonal d y n am ics o f co lo red disso lved m a te ria l in th e Sargasso Sea. D eep Sea Res. I 45 (6): 931—957.

60. G reg g W W , C o n k rig h t M E (2002) D e c a d a l C h a n g e s in G lo b a l O c e a n ch lo ro p h y ll. G eo p h y s ica l R e se a rc h L e tte rs 29 (15): 20-1 - 20-4 . D O I: 1 0 .1 0 2 9 /2 0 0 2 G L 0 14689.

61. K o h le r H , M e o n B, G o rd eev W , Spitzy A , A m o n R M W (2003) D issolved o rg an ic m a tte r (D O M ) in th e estuaries o f O b a n d Y en isei a n d th e a d ja c e n t K a ra -S ea , R ussia. P ro ceed in g s in M a rin e Science 6: 281—309.

62. K o w alczu k P , S to n -E g ier J , C o o p e r W J, W h ite h e ad R F , D u ra k o M J (2005) C h a ra c te riz a tio n o f c h ro m o p h o ric dissolved o rg an ic m a tte r (C D O M ) in the B altic S ea b y ex c ita tio n em ission m a tr ix f luo rescence spectroscopy . M arin e C h e m is try 96 ( 3 ^ ) : 2 7 3 -2 9 2 .

63. M o re l A , G en tili B (2009) A sim ple b a n d ratio tec h n iq u e to q u an tify th e co lo red disso lved a n d d e tr ita l o rg an ic m ate ria l fro m o c e an co lo r rem o te ly sen sed d a ta . R e m o te S ensing o f E n v iro n m e n t 113: 99 8 —1011.

64. M o re l A , G en tili B, C lau s tre H , B ab in M , B ricau d A, e t al. (2007) O p tica l p ro p ertie s o f th e “ c lea res t” n a tu ra l w aters. L im n o lo g y a n d O c e a n o g ra p h y 52 (1),: 2 1 7 -2 2 9 .

65. M cG ill D A (1961) A p re lim in a ry study o f th e oxygen a n d p h o sp h a te d istrib u tio n in th e M e d ite rra n e a n Sea. D eep -S ea R es. 8: 259—268.

66. D o la n J R , V idussi F, C la u s tre FI (1999) P lan k to n ic ciliates in th e M e d ite rra n e a n Sea: lo n g itu d in a l tren d s . D eep -S ea R es. I 46: 2 0 2 5 -2 0 3 9 .

67. S io k o u -F ran g o u I, C h ristak i U , M azzo cch i M G , M o n tre so r M , R ib e ra d ’A lcalá M , e t al. (2009) P la n k to n in th e o p e n M e d ite r r a n e a n Sea: a rev iew . B iogeosciences 6: 1 1 1 8 7 -1 1 2 9 3 .

68. P re isen d o rfe r R W (1986) S ecchi disc science: V isual optics o f n a tu ra l w aters. L im nology a n d O c e a n o g ra p h y 31 (5): 90 9 —926.

69. L e term e SG, E d w ard s M , S e u ro n t L , A ttrill M J, R e id P C , e t al. (2005) D ec a d a l basin -sca le ch an g es in d ia to m s, dinoflagellates, a n d p h y to p la n k to n co lo u r across th e N o r th A d an tic . L im nology a n d O c e a n o g ra p h y 50: 1244—1253.

70. B a r to n A D , G re e n e CEI, M o n g e r B G , P e rsh in g A J (2003) T h e C o n tin u o u s P la n k to n R e c o rd e r survey a n d th e N o r th A d a n tic O scilla tion: In te ra n n u a l- to M u ltid ecad a l-scale p a tte rn s o f p h y to p la n k to n variab ility in th e N o r th A d an d c O c e a n . P rogress in O c e a n o g ra p h y 58: 337—358.

71. H e a d E JH , P ep in P (2010) M o n ito r in g ch an g es in p h y to p la n k to n a b u n d a n ce a n d co m p o s id o n in the N o rth w e s t A d an d c : a co m p ariso n o f results o b ta in e d by c o n tin u o u s p lan k to n rec o rd e r sa m p lin g a n d c o lo u r sa tellite im agery . J . P lan k to n R e se a rc h 32 (12): 1 6 4 9 -1 6 6 0 . D O I: 1 0 .1 0 9 3 /p la n k t/fb q l2 0 .

72. E d w ard s M , R e id P , P lan q u e B (2001) L o n g -te rm a n d reg io n a l variab ility o f p h y to p la n k to n b iom ass in th e N o rth e a s t A d a n tic (1960—1995). J o u rn a l o f M a rin e Science 58 (1): 3 9 -4 9 .

73. E d w ard s M , B e a u g ran d G , J o h n A G , J o h n s D G , L ican d ro P , e t al. (2009) E co log ical S ta tu s R e p o rt: results fro m th e G P R survey 2 0 0 7 /2 0 0 8 . S A H F O S T e c h n ic a l R e p o rt, 6: 1 -1 2 . P ly m o u th , U .K . IS S N 1744-0750.

74. H e n so n SA, S a rm ien to J L , D u n n e J P , B o p p L , L im a I, e t al. (2010). D e te cd o n o f a n th ro p o g e n ic c lim ate ch a n g e in satellite reco rd s o f o c e an ch lo ro p h y ll a n d p ro d u ctiv ity , B iogeosciences 7: 621—640.

75. Boyce D G , Lewis M , W o rm B (2012) In te g ra d n g g lobal ch lo ro p h y ll d a ta fro m 1890 to 2010. L im nol. O c ean o g r.: M e th o d s 10: 8 4 0 -8 5 2 . D O I 1 0 .4 3 1 9 / lo m .2 0 1 2 .10.840.



© plosi trends in ocean colour and chlorophyll ... · the record of chlorophyll changes over time...

Documents