 Reference Base  Progress with developing statistical models to determine ... |
|
|
|
|
|
| |
| Location: |
World |
| Subject: |
Management |
| Species: |
All Rhino Species |
 |
|
It is often said that 'you are what you eat'. Because elements and different isotopes present in food plants, affected by climate, geology and type of plant, can be absorbed into rhino horn through the digestive processes, chemical analysis of rhino horn offers the potential of determining both the source of the horn as well as the species of rhino that produced it.
Results of the WWF-funded AfRSG continental horn fingerprinting project, which aims to develop forensic techniques to determine the source and species of rhino horn based on analysis of its chemistry, have been written up in a detailed confidential final report submitted to WWF.
This report discusses the statistical analysis of rhino-horn chemistry data for samples of horn from populations holding approximately 70% of Africa's rhinos. After dealing with problems of high data dimensionality, multi-colinearity and zero values, successful species and source-identification models were built using discriminant functions - often with 100% post-hoc classification success. Horn chemistry was also related to rainfall and primary productivity, and horn tips were found not to differ consistently chemically from the rest of the horn. The best source determination models used data from all three labs and analysed data for the two species separately at the finer spatial scale of park or area within a park. Graphical presentation of the results (canonical plots, traces and icon plots) also enables them to be understood by laymen.
However, despite these successes, results should be treated as preliminary until they are validated independently using jack-knifing, which requires the acquisition of more advanced statistical software. Further work is also required to improve identification of samples that have come from areas not yet covered by the horn fingerprint atlas.
As part of the process of taking the analyses further, with the ultimate aim of developing a routine forensic source-determination technique, the AfRSG Scientific Officer will inform the peer review of the work done at the Southern African Statistical Association Conference in November 2000. Dr Rajan Amin of the Zoological Society of London has offered to assist with further statistical analysis.
|
|
|
File available
details
Original text on this topic: