Are phytoplankton blooms occurring earlier in the Arctic? M. KAHRU *, V. BROTAS w , M. MANZANO-SARABIA z and B. G. MITCHELL * *Scripps Institution of Oceanography, University of California San Diego, La Jolla, California, USA, wCentre of Oceanography, Faculdade de Cie ˆncias da Universidade de Lisboa, Lisboa, Portugal, zFacultad de Ciencias del Mar, Universidad Auto ´noma de Sinaloa, Mazatla ´n, Sinaloa, Me ´xico Abstract Time series of satellite-derived surface chlorophyll-a concentration (Chl) in 1997–2009 were used to examine for trends in the timing of the annual phytoplankton bloom maximum. Significant trends towards earlier phytoplankton blooms were detected in about 11% of the area of the Arctic Ocean with valid Chl data, e.g. in the Hudson Bay, Foxe Basin, Baffin Sea, off the coasts of Greenland, in the Kara Sea and around Novaya Zemlya. These areas roughly coincide with areas where ice concentration has decreased in early summer (June), thus making the earlier blooms possible. In the selected areas, the annual phytoplankton bloom maximum has advanced by up to 50 days which may have consequences for the Arctic food chain and carbon cycling. Outside the Arctic, the annual Chl maximum has become earlier in boreal North Pacific but later in the North Atlantic. Keywords: arctic, climate change, ocean color, phenology, phytoplankton blooms, remote sensing Received 26 April 2010; revised version received 30 June 2010 and accepted 9 August 2010 Introduction Both satellite and in situ studies on land have shown that in many temperate ecosystems the spring onset of vegetation greenness has advanced, the growing season has lengthened, and both these changes are correlated with rising temperatures (Myneni et al., 1997; Cleland et al., 2007). A limited number of corresponding obser- vations using phytoplankton species in the ocean (Ed- wards & Richardson, 2004; Winder & Schindler, 2004; Thackeray et al., 2008) have also been able to show changes in the timing of annual phenomena but others (Winder & Cloern, 2009) question the ability to detect climate driven trends in phytoplankton communities due to their high spatio-temporal variability. Satellite- derived time series of chlorophyll-a concentration (Chl, mg m 3 ), used as a proxy for phytoplankton biomass, have shown trends of increased phytoplankton bloom magnitude in coastal and shelf areas (Kahru & Mitchell, 2008; Kahru et al., 2009) as well as multidecadal and interannual oscillations in oceanic patterns of Chl and sea surface temperature (Behrenfeld et al., 2006; Marti- nez et al., 2009). The detection of changes in the timing of blooms is complex due to the various shapes of the annual cycle in phytoplankton biomass and the high variability in the oceanic time series (Vargas et al., 2009). Phytoplankton blooms stimulate the production of zooplankton which, in turn, provides forage for larval fish (Platt et al., 2003). The height and timing of the annual Chl maximum are a result of many interacting processes (Siegel et al., 2002). It appears that secondary producers, such as shrimp, may have adapted their egg hatching times to the phytoplankton spring bloom (Koeller et al., 2009). The match or mismatch between the reproductive cycles of marine organisms and their food may partly determine the year class strength (Cushing, 1990). Here, we use satellite-derived Chl to objectively map changes in the timing of the annual phytoplankton maxima, with an emphasis on the Arctic Ocean and the adjacent boreal oceans. We acknowledge that the annual cycle can have a multitude of shapes (cf. Vargas et al., 2009; Platt et al., 2010); however, as a first approximation, we search for trends in the timing of the annual maximum without considering the shape of the annual cycle. Data and methods Satellite-derived Level-3 (i.e. binned and mapped) data sets of chlorophyll-a concentration (Chl, mg m 3 ) were obtained from NASA’s Ocean Color website (http:// oceancolor.gsfc.nasa.gov/) and the European Space Agency’s GlobColour project (http://www.globcolour.- info). Daily composite concentrations of Chl using stan- dard Case 1 water algorithms (O’Reilly et al., 1998; Morel & Maritorena, 2001) were obtained. Any single ocean color sensor has a limited daily coverage result- ing from gaps between the swaths, sun glint and cloud cover. Merging data from multiple sensors – if data Correspondence: M. Kahru, Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA, USA, tel. 1 1 858 534 8947, fax 1 1 858 822 0562, e-mail: mkahru@ucsd.edu Global Change Biology (2011) 17, 1733–1739, doi: 10.1111/j.1365-2486.2010.02312.x r 2010 Blackwell Publishing Ltd 1733