Bias? What bias? The evolution of downstream larval‐feeding in animals

The homology of larval‐feeding modes has been a major element in recent discussions about animal phylogeny. ‘Downstream‐feeding’ is one of the two main larval‐feeding modes, and is usually equated to an opposed‐band system involving ciliary bands called the prototroch and metatroch. Larval‐feeding in Spiralia is reviewed here and the homology hypothesis of downstream larval‐feeding is expanded, encompassing any feeding involving the prototroch. It is often argued that the presence of planktotrophic larvae using downstream‐feeding is plesiomorphic among spiralian animals, and that there is a bias in transformations, such that feeding larvae tend to be lost rather than gained. These hypotheses are assessed in relation to the Spiralia, Trochozoa and particularly polychaete annelids. Cladistic parsimony analyses are performed based on datasets of Rouse & Fauchald (1997) and Rouse (1999), with an additional character based on downstream larval‐feeding. Methods adopted to assess the possibility of a bias in transformations towards loss of downstream larval‐feeding include: expanded primary homology arguments, character transformations favouring reversals and polymorphic terminals coded as having downstream larval‐feeding. These measures all tend to favour the possibility that downstream larval‐feeding will be plesiomorphic, and tends to be lost rather than gained. Nevertheless, all analyses show that downstream larval‐feeding is not plesiomorphic, and appears to have evolved multiple times. The results support a conclusion that the prototroch has become associated with feeding a number of times. Hypotheses of metazoan phylogeny that are predicated on the assumption that downstream‐feeding larvae are plesiomorphic (e.g. Peterson et al . 1997) should be re‐assessed.