Factors influencing movement probabilities of Florida tree snails Liguus fasciatus (Muller) in big cypress national preserve following Hurricane Andrew

Beginning in 1993, approximately one year after Hurricane Andrew, we marked 2,547 Florida tree snails (Liguus fasciatus) during six sampling (mark-res ighting) occasions twice per year over a three-year period. During each sam pling occasion, we conducted a search for all tree snails within 16 sample plots on eight hammocks (tree islands) and for an extended radius of approx imately 20 m around each plot. We individually marked all trees > 5 cm dbh within each plot and all trees within the extended radius in which a marked snail was found. Because the host tree for each marked animal was known, i nter-tree movements could be directly measured as the proportion of animals found on different host trees at times t and t + 1. We observed 533 marked snails during two or more consecutive sampling occasions. Of these, 414 (7 7.7%) snails had moved from one host tree to another between sampling occas ions. Based on a conditional logistic regression model, movement probabilit ies were influenced by whether or not the snail's host tree had been Lysilo ma at time t, the diameter of the host tree at time t, and interactions amo ng these and with year. Overall, snails had a lower probability of moving f rom one tree to another ii their host tree had been Lysiloma and ii they we re on larger host trees. We found no evidence that movement probabilities w ere influenced by the snail's subspecies, age, hammock size class, or by th e season. An effect of latitudinal gradient or proportion of host-trees dam aged was not retained in our final model. However, a preliminary univariate analysis did indicate a difference in movement probabilities between our n orthernmost hammocks, which had relatively little hurricane damage, and the southern hammocks, which had extensive damage.