Page 17 - Lloreta_alii_2001

th
catch data that only cover the 20 century (from 1920 or 1950 up to now). When
considering the 1920-1960 period, BFT trap catches also appeared correlated to the

LOD. However, the correlation vanished over the 1850-1890 period. Thus, when the

long-term is considered, we do not find a relationship between BFT fluctuations and the

LOD/ACI.

A potential long-term forcing of the temperature was finally investigated through

six different time series of temperature. The Jones’s proxy and the instrumental series of

temperature from Cadix and Dar-El-Beida were reliable and validated. D’Arrigo’s

proxy was reconstructed from tree-rings records, so that other meteorological (e.g. rain)

or climate-independent factors could blur this index (Schmutz et al., 2000), which is

calculated on inland information. This could explain why results of regressions and

correlations are less clear with D’Arrigo’s time series than with other temperature time

series. The two Italian time series (i.e. Palermo and Cagliari) could present some

problems of homogeneity between the early and the later period (Maugeri, Insituto di

Fisica Generale Applicata, Italy, pers.com., see also Brunetti et al., 2000). Therefore,

analyses were re-computed using a regional series for Centre-Italy, which was

homogeneous and validated (Brunetti et al., 2000). The results (not shown) led to the

same conclusions: BFT trap catches appeared closely and negatively related to

temperature. The analyses both in time and frequency domain between BFT and

temperature time series lead thus to consistent and trustworthy results. Significant

regression or correlation analyses do not imply causal relationship, but neither we can

deny the possibility that BFT long-term fluctuations were induced by changes in

temperature. To check for it, further modelling and experimental studies are needed.

However, we may discuss, on the basis of our current knowledge, the most probable

underlying processes of such a relationship.

18