To understand the origins of cultivated taro (Colocasia esculenta), we must study wild populations and closely-related wild species. A broad outline of the possible history of this crop has begun to emerge. Taro may have diverged from an ancestor adapted to cool and humid montane environments in mainland Southeast Asia. If the genus has existed since at least the Miocene geological period, then Colocasia species must have dispersed and evolved through multiple cycles of glacial and interglacial climatic conditions. Over these cycles, species may have differentiated along three main environmental gradients: wet to dry, cool to warm, and open to shaded. Two evolutionary lineages of C. esculenta have been domesticated and both tolerate open cultivated habitats: var. antiquorum and var. esculenta. These are associated with (i) montane cool and dry, and (ii) lowland warm and wet environments respectively. Perhaps the greatest mystery is how the montane lineage adapted to open habitats, given that the most obvious natural open habitats in Southeast Asia are permanent springs and the banks of meandering rivers in lowland plains. Around the world, farmers have adopted these two lineages and developed cultivation systems that must to some extent mirror the natural ecology of wild source populations.