The nightly variation of activity was first analysed by modelling
mean activity in the tubes with a mixed model using light treatment,
earthworm origin,mean air temperature and mean relative
air humidity as fixed explanatory variables.The model included all
interactions with earthworm origin. “Tube” was used as a random
variable in the model(six tubes for each origin).The means for each
night were calculated from the twelve observations recorded at
0.5 h intervals.The effects of temperature and humidity on activity
was further studied by dividing the nights into three temperature
classes(cold(9 C))and
two air humidity classes(dry(90%)).Categorical,
instead of continuous variables were used in the statistical
analyses because weather during many nights was quite similar,
the relationship between weather variables and studied responses
were not linear(N was too small for non-linear modelling) and
weather data was incomplete.The classification of both variables
was done so that N was sufficiently high in each class for the
analyses