Abstract:
Migratory animals often use environmental cues to time their seasonal migrations. Local conditions may, however, differ from distant ones, and current conditions may poorly predict future conditions. This may be particularly true for early wet season conditions in tropical systems, as storms and associated rainfall events are generally not predictable at the scale of weeks or days and are heterogeneously distributed even at the scale of a few kilometres. How migratory animals cope with such challenges, and the consequences they may have, remain poorly known. We used time-to-event models based on GPS data from 19 African elephant herds from Hwange National Park (Zimbabwe) to study the effect of local and distant rainfall events on the elephants’ decision to initiate their wet season migration. Elephants relied more on distant rainfall events occurring along the future migration route than on local events when initiating their migration. Such ability to use distant cues does not, however, ensure an immediate migration success. In over 30% of the cases, the elephants came back to their dry season range, sometimes after having travelled > 80% of the expected migration distance. This happened particularly when there was little additional rain falling during the migration. All elephants successfully migrated later in the season. Our study improves the understanding of the migratory ecology of elephants. More broadly, it raises questions about the reliability of rainfall as a migratory cue in tropical systems, and shed light on one of its potential consequences, the poorly quantified phenomenon of migration false starts.
