Analysis of Pine Marten Distribution in England and Wales

The Vincent Wildlife Trust had developed a structured scoring system for ad hoc sightings of pine martens (Martes martes) reported from England and Wales by casual observers. In this contract BioEcoSS Ltd. analysed the geographical distribution of sightings of this rare mustelid using computer-intensive techniques such as genetic algorithms and permutation tests. The contract comprised two stages:

a) Firstly, to develop a single-user database in MS Access to hold details over 800 sightings records. Each record included a large number of variables that were assessed objectively to generate a score from 1 - 10 that the sighting was "good". These records could then exported in spreadsheet format or filtered and tabulated to give a large combination of outputs, in both report and graphical formats. Furthermore, the geographical locations could be collated and output in a format suitable for use by Dmap (www.dmap.co.uk).
b) Secondly, these data were used to parameterise a predictive model with 58 binary variables. The weightings for the parameters were generated using a genetic algorithm which produced an objective scoring system. These scores were then analysed using permutation tests to predict the spatial density of sightings, taking account of the effort expended by observers. The results indicated that an objectively weighted scoring system could be used successfully to score sightings from members of the public and were presented at the 2004 Martes Symposium in Lisbon (Poulton et al; 2006).