Climatic control of trophic interaction strength: the effect of lizards on spiders.

We investigated how temporal variation in rainfall influences the impact of lizards on spiders inhabiting small islands in Abaco, Bahamas. Annual censuses of web spiders were conducted on nine lizard islands and on eight no-lizard islands 1994-2003. Repeated-measures ANOVA showed that annual variation in spider density (time) and in the lizard effect on spider density (lizard x time) were both significant. Correlation coefficients between the lizard effect (ln ratio of no-lizard to lizard spider densities) and number of rainfall days were generally negative, and strengthened with length of the time period during which rainfall was measured prior to annual spider censuses. Spider density was also negatively correlated with rainfall days and strengthened with length of the prior time period. Longer time intervals included the hurricane season, suggesting that the strong negative correlations were linked to high rainfall years during which tropical storms impacted the region and reduced spider and lizard densities. Split-plot ANOVA showed that rainfall during the hurricane season had a significant effect on the lizard effect and on spider density. Results in this study are opposite to those found in our previous 10-year study (1981-1990) conducted in the Exuma Cays, a moderately xeric region of the Bahamas, where the relation between rainfall and the lizard effect on spider density was positive. Combined data from the Exuma and Abaco studies produce a unimodal relation between trophic interaction strength and rainfall; we suggest that the negative effect of storms associated with rainfall was paramount in the present study, whereas the positive bottom-up effect of rainfall prevailed in our previous study. We conclude that climatic variability has a major impact on the trophic interaction and suggest that a substantial change in precipitation in either direction may weaken the interaction significantly.