Variation in the number of occupied wasp nests (all species pooled).

* =  number of trees with wasp nests;
** =  number of wasp nests along the same perimeter. The data were obtained before and after the 1998–2000 ENSO event in the Saint Elie plantation and forest edges all around (Sinnamary district). Statistical comparisons. (1) Friedman Chi-square test  =  8.3333, df  =  3, P = 0.0396. (2) GLM Poisson model: the intercept corresponds to the effect ‘year Dec 06’ and the ‘trees of the forest edge’. The effect of grapefruit trees and A. sciophilum is significant as well as that of the year Dec 92 at P<0.001. Tukey post-hoc test: the number of wasp nests recorded during the Dec 92 survey is significantly different from that of the three other years (Dec 06, Dec 07, and Dec 10) at P<0.001; the differences between the number of wasp nests recorded during these three latter years were not significant. (3) Under the model: Nt  =  a * Nt-1 * ε, assuming a  =  1, the passage from 121 wasp nests in 1992 to 11 wasp nests in 2006 has a probability of 0.0355. It is therefore unlikely that this passage was due to demographic stochasticity only.