Abstract:
Analyses of 2 light trap catches comprising 6 041 and 1 598 Culicoides showed that the reliability of subsampling such catches increased with subsample size, while the subsampling error decreased with an increase in the number of individuals per species present in a subsample. Subsamples comprising
approximately 500 Culicoides are deemed sufficient for comparing population densities between
sites and sampling occasions and also give an acceptable indication of relative species abundance at a
site.
It is recommended that species for which the mean number of specimens in subsamples originating
from 3 catches at a site is less than 7, should not be included in biometric analyses aimed at
comparisons of population densities. For all other species a test level of 1% rather than 5% should be
used for such comparisons. When species ratios obtained from subsampled catches are employed as indicators of abundance, the chi-squared test should be utilized at a 1% level if the ratios originate from
3 catches at a site and at a 0,1% level if only one catch per site is made. Due to poor representativeness
of small catches, it is suggested that species for which fewer than 7 individuals are present in a
single subsample, be excluded from chi-squared tests.
A 5-point procedure for subsampling a large light trap collection of Culicoides is given.