Individual identification via non-invasive sampling is of prime importance
in conservation genetics and in behavioral ecology. This approach allows fo
r genetics studies of wild animals without having to catch them, or even to
observe them. The material used as a source of DNA is usually faeces, shed
hairs, or shed feathers. It has been recently shown that this material may
lead to genotyping errors, mainly due to allelic dropout. In addition to t
hese technical errors, there are problems with accurately estimating the pr
obability of identity (PI, or the probability of two individuals having ide
ntical genotypes) because of the presence of close relatives in natural pop
ulations. As a consequence, before initiating an extensive study involving
non-invasive sampling, we strongly suggest conducting a pilot stud) to asse
ss both the technical difficulties and the PI for the genetic markers to be
used. This pilot study could be carried out in three steps: (i) estimation
of the PI using preliminary genetic data; (ii) simulations taking into acc
ount the PI and choosing the technical error rate that is sufficiently low
for assessing the scientific question; (iii) polymerase chain reaction (PCR
) experiments to check if it is technically possible to achieve this error
rate. (C) 1999 The Linnean Society of London.