Abstract:
We analyzed the suitability of a combined sampling approach ? consisting of visual search
and branch-beating ? for quantifying tropical caterpillar communities. Surveys were conducted in the
Ecuadorian montane forest zone, with two shrub species from the genus Piper serving as focal targets.
We sampled 160 shrubs in the course of four experiments following a standardized sampling protocol.
Subsequently each shrub was completely defoliated accompanied by an intensive leaf-by-leaf search, in
an effort to extract as close to 100% of all present caterpillars as possible. We analyzed the resulting
dataset with regard to completeness, taxonomical bias, and influences of daytime, complexity of shrub
structure, or experience of the researcher. The standardized sampling protocol extracted between
50.6% and 71.6% of the caterpillars present on a shrub. A minor taxonomic bias of the sampling
protocol was observed, but appears to be of a simple and predictable nature, and is therefore easy to
account for. We did not find any significant influences of daytime. Structure and size of shrubs had
a strong influence on sampling results with small and simply structured shrubs being sampled most
completely, large and complex shrubs most incompletely in our dataset. Researcher experience did
not appear to have an influence on the sampling efficiency or taxonomic composition of samples
obtained when we compared caterpillars obtained by standardized sampling with those collected by
exhaustive leaf-by-leaf search. Comparison of caterpillar sizes revealed however, that inexperienced
field assistants tended to overlook large fractions of the smallest caterpillars entirely. We conclude
that our standardized combined sampling approach is fairly suitable for studies concerning caterpillar
communities, especially when resampling of the same shrub individuals is desired.