Do invertebrate communities in floodplains change predictably along a river's length?