Factors affecting quality of temperature models for the pre-appearance interval of forensically useful insects

In the case of many forensically important insects an interval preceding appearance of an insect stage on a corpse (called the pre-appearance interval or PAI) is strongly temperature-dependent. Accordingly, it was proposed to estimate PAI from temperature by using temperature models for PAI of particular insect species and temperature data specific for a given case. The quality of temperature models for PAI depends on the protocols for PAI field studies. In this article we analyze effects of sampling frequency and techniques, temperature data, as well as the size of a sample on the quality of PAI models. Models were created by using data from a largely replicated PAI field study, and their performance in estimation was tested with external body of PAI data. It was found that low frequency of insect sampling distinctly deteriorated temperature models for PAI. The effect of sampling techniques was clearly smaller. Temperature data from local weather station gave models of poor quality, however their retrospective correction clearly improved the models. Most importantly, current results demonstrate that sample size in PAI field studies may be substantially reduced, with no model deterioration. Samples consisting of 11-14 carcasses gave models of high quality, as long as the whole range of relevant temperatures was studied. Moreover, it was found that carcasses exposed in forests and carcasses exposed in early spring are particularly important, as they ensure that PAI data is collected at low temperatures. A preliminary best practice model for PAI field studies is given.