Migration by nematode larvae through the tissues of their mammalian ho
sts can cause considerable pathology, and yet the evolutionary factors
responsible for this migratory behaviour are poorly understood. The b
ehaviour is particularly paradoxical in genera such as Ascaris and Str
ongylus in which larvae undergo extensive migrations which begin and e
nd in the same location. The orthodox explanation for this apparently
pointless behaviour is that a tissue phase is a developmental requirem
ent following the evolutionary loss of skin penetration or intermediat
e hosts. Yet tissue migration is not always necessary for development,
and navigation and survival in an array of different habitats must re
quire costly biochemical and morphological adaptations. Migrating larv
ae also risk becoming lost or killed by the host. Natural selection sh
ould therefore remove such behaviour unless there are compensating ben
efits. Here we propose that migration is a selectively advantageous li
fe-history strategy. We show that taxa exploiting tissue habitats duri
ng development are, on average, bigger than their closest relatives th
at develop wholly in the gastrointestinal tract. Time to reproduction
is the same, indicating that worms with a tissue phase during developm
ent grow faster. This previously unsuspected association between juven
ile habitat and size is independent of any effects of adult habitat, l
ife-cycle, or host size, generation time or diet. Because fecundity is
intimately linked with size in nematodes, this provides an explanatio
n for the maintenance of tissue migration by natural selection, analog
ous to the pre-spawning migrations of salmon.