Data, Walker, 1862

Data View in CoL analyses

The pattern of air-exposure behavior was consistent throughout the populations for the three species from Costa Rica: it was always present in Cl. albidigitus and Ca. obscurus , and always absent in Paguristes sp. ( Tab.1); thus, data were pooled by species to obtain a sample size adequate for robust statistical analyses. No individuals of Paguristes sp. was observed exhibiting air-exposure behavior, thus after testing the statistical significance of this absence by using a chi-squared test, this species was excluded from subsequent statistical analysis.For the remaining three species ( Cl. albidigitus , Cl. erythropus , and Ca. obscurus ), we applied chi-squared tests to assess possible differences in the number of crabs exhibiting air-exposure behavior depending on sex categories and shell types, as well as the number of survivors of physical stress experiments with Cl. erythropus . For each demographic category (male, non-ovigerous female, ovigerous female) and species, size differences between hermit crabs, which performed air-exposure behavior on the rocks and those that were clustered under stones were tested by a one-way ANOVA (factor: position on rocks). Similarly, we analysed the size differences between live and dead hermit crabs used in physical stress experiments of Cl. erythropus by a three-way ANOVA (factors: hermit crab demographic category, shell type and temperature). Data were log-transformed to homogenize variance prior to analysis, and the Student–Newman–Keuls test was used as a post hoc test. A p> 0.05 value was considered as the threshold for statistical non-significance for all statistical tests carried out with these data.