1. Lowland temperate rain forest was sampled in Fiordland, New Zealand
. An assembly rule was sought in terms of relative constancy in the pr
oportion of species from seven synusial (stratum) guilds, compared to
a null model in which species associate at random. Two forest types we
re examined (coastal broadleaved and Nothofagus broadleaved), separate
ly and together. 2. The results show more significances than expected
on a random basis. 3. In two forest-type/synusia analyses, guild propo
rtions were more variable than expected under the null model. For lian
es this is suggested to be the result of old canopy gaps and for shrub
s because of topographic variation. 4. There was significant constancy
in the proportion of species from the herb synusia. This effect was s
een separately in both forest types. A similar effect was seen for the
ground synusia, significant when the two forest types were combined.
These effects suggest that there is a limitation on the assembly of th
ese forest communities, a constraint on species coexistence that ensur
es relatively constant representation from the ground and herb guilds.
This represents one of the first demonstrations of an assembly rule i
n plant communities.