A two-way ANOVA was used to analyse the influence of the
cultivation site and time of measurement on the daily cycles of ΦPSII
and rETR. A posteriori differences were evaluated with the Unequal N
HSD test (Statistica 6.0, StatSoft). All other effects were analysed with
a one-way ANOVA. Data were tested for variance heterogeneity
(Cochran's Test) and, when the answer was significant, the data were
transformed (Log or ArcSin). When the transformation didn't normalize
the data, having a high number of replicates, we used a more
conservative level of significance Pb0.01 (Underwood, 1997). The
differences found were analysed a posteriori with a Student–Newman
Keuls test.