P. Sirabella et al., Breaking down the climate effects on cod recruitment by principal component analysis and canonical correlation, MAR ECOL-PR, 216, 2001, pp. 213-222
The pattern of temporal correlations between cod recruitment and sea temper
ature, in conjuction with the climate variability of atmospheric pressure a
nomalies (NAO index) was investigated by means of a combined use of princip
al component analysis (PCA) and canonical correlation analysis (CCA), using
time series collected in the area surrounding the Kola peninsula (Barents
Sea) and in the North Sea. The proposed data analysis strategy, namely to c
arry out a PCA of the temperature, cod recruitment and NAO time series foll
owed by a CCA between the component spaces of all the possible data sets co
uples (recruitment vs temperature, recruitment vs NAO and NAO vs temperatur
e), allowed us to sketch a general model of correlation between climate and
cod recruitment dynamics. Two independent effects of temperature variabili
ty on cod recruitment emerged for the Kola region, pointing to the existenc
e of at least 2 different mechanisms of comparable importance by which temp
erature may affect cod recruitment. In the North Sea the situation is somew
hat simpler, and the data are compatible with only 1 major interaction mech
anism. Moreover, the general effect of temperature on cod recruitment was o
pposite in the 2 regions: direct correlation for the Barents Sea, inverse c
orrelation for the North Sea. This is probably due to the existence of an o
ptimal temperature regime for cod recruitment lying in between the 'cold' B
arents Sea and the 'warm' North Sea.