: The processes driving temporal distribution and abundance patterns of whale sharks
Rhincodon typus
remain largely unexplained. We present an analysis of whale shark occurrence
in the western Indian Ocean, incorporating both spatial and temporal elements. We tested the
hypothesis that the average sighting probability of sharks has not changed over nearly 2 decades,
and evaluated whether variance in sightings can be partially explained by climate signals. We
used a 17 yr dataset (1991 to 2007, autumn only) of whale shark observations recorded in the log-
books of tuna purse-seiners. We randomly generated pseudo-absences and applied sequential
generalized linear mixed-effects models within a multi-model information-theoretic framework,
accounting for sampling effort and random annual variation, to evaluate the relative importance
of temporal and climatic predictors to sighting probability. After accounting for seasonal patterns
in distribution, we found evidence that sighting probability increased slightly in the first half of the
sampling interval (1991−2000) and decreased thereafter (2000−2007). The model including a spa-
tial predictor of occurrence, fishing effort, time
2
and a random spatial effect explained ~60% of the
deviance in sighting probability. After including climatic predictors, we found that sighting prob-
ability increased slightly with rising temperature in the central Pacific Ocean and reduced temper-
atures in the Indian Ocean. The declining phase of the peak, concurrent with recent accounts of
declines in population size at near-shore aggregations and with the most pronounced global
warming, deserves continued investigation. Teasing apart the legacy effects of past exploitation
and those arising from on-going climate changes will be a major challenge for the successful long-
term management of the species.