Can Monitoring Data of Forensically Important Blow Flies Be Transferred Between Cities?

• Published in: Journal of applied entomology (1986) 2024-10
• Main Authors: Lutz, Lena, Amendt, Jens, Reckel, Frank, Krebs, Oliver, Klimpel, Sven, Cunze, Sarah
• Format: Article
• Language: English
• ISSN: 0931-2048; 1439-0418
• DOI: 10.1111/jen.13354

The use of monitoring data to build prediction models for the abundance and activity of necrophagous blow flies is common practice in forensic entomology, but its advantages and disadvantages are still debated. A frequently asked question is the transferability of such species prediction models. So far no study has examined whether the assumption of low transferability of these data between cities and seasons holds true. In the present study, we evaluate whether models calibrated with a specific training data set from a specific place and time can be transferred to other data sets for different time periods and locations. We developed models using five different algorithms to predict the activity and abundance of four forensically relevant blow fly species ( Calliphora vicina Robineau‐Desvoidy, Lucilia ampullacea Villeneuve, Lucilia caesar (Linnaeus), Lucilia sericata (Meigen)). The training data set was obtained from a single city, and the transferability of the models was evaluated using monitoring data from this and three other cities. The geographic transferability of the models was confirmed for all algorithms, but only for two species, C. vicina and L. sericata , and for two of the four cities. Lucilia caesar and L. ampullacea were rare in the test data set, and their species‐specific adaptation to environmental parameters was not captured by the models. Cities that did not work differed from the training data set in terms of climate and habitat features. To build generalised predictive models of blow fly abundance and activity, we need training data sets based on monitoring data from different regions, seasons and years to cover a wide range of environmental conditions. This is essential for describing and predicting natural variability.