Get access

Detecting the impact of oceano-climatic changes on marine ecosystems using a multivariate index: The case of the Bay of Biscay (North Atlantic-European Ocean)


Georges Hemery, e-mail:


Large-scale univariate climate indices (such as NAO) are thought to outperform local weather variables in the explanation of trends in animal numbers but are not always suitable to describe regional scale patterns. We advocate the use of a Multivariate Oceanic and Climatic index (MOCI), derived from ‘synthetic’ and independent variables from a linear combination of the total initial variables objectively obtained from Principal Component Analysis. We test the efficacy of the index using long-term data from marine animal populations. The study area is the southern half of the Bay of Biscay (43°–47°N; western Europe). Between 1974 and 2000 we monitored cetaceans and seabirds along 131000 standardized line transects from ships. Fish abundance was derived from commercial fishery landings. We used 44 initial variables describing the oceanic and atmospheric conditions and characterizing the four annual seasons in the Bay of Biscay. The first principal component of our MOCI is called the South Biscay Climate (SBC) index. The winter NAO index was correlated to this SBC index. Inter-annual fluctuations for most seabird, cetacean and fish populations were significant. Boreal species (e.g. gadiformes fish species, European storm petrel and Razorbill …) with affinities to cold temperate waters declined significantly over time while two (Puffin and Killer Whale) totally disappeared from the area during the study period. Meridional species with affinities to hotter waters increased in population size. Those medium-term demographic trends may reveal a regime shift for this part of the Atlantic Ocean. Most of the specific observed trends were highly correlated to the SBC index and not to the NAO. Between 40% and 60% of temporal variations in species abundance were explained by the multivariate SBC index suggesting that the whole marine ecosystem is strongly affected by a limited number of physical parameters revealed by the multivariate SBC index. Aside the statistical error of the field measurements, the remaining variation unexplained by the physical characteristics of the environment correspond to the impact of anthropogenic activities such overfishing and oil-spills.