Quantitative model for geolocating pollen samples

March 19, 2019
ingehmvhaaften

Quantitative model for geolocating pollen samples

By Inge van Haaften (currently studying for MSc Biological Sciences, Ecology & Evolution track at the University of Amsterdam)

For this “Amsterdam Palaeoecology Club” (APC) meeting we did not discuss a paper, instead I presented my progress on my second research internship of my master’s on the geolocation of pollen samples under the supervision of dr. C. N. H. McMichael. The other students were asked to read the paper ‘Forensic palynology: Why do it and how it works’ by Mildenhall et al. (2006). This paper gives a review of the use of palynological analysis for criminal investigation, which ties in with my research.

Criminal investigators, customs, scientists and regulatory bodies have an interest in knowing the origins of materials, for example when dealing with poached goods. Because pollen can only originate from a locale that features the plants that produce that pollen, items that contain or are covered by specific pollen types can be geolocated. There are many items that come from regions of high biodiversity, using expert knowledge to geolocate samples becomes more time and energy intensive when you are dealing with a larger range of probable origin and many genera. For that reason, this study will be using surface samples from 705 sites in the neotropics containing 473 genera.

I have been working on a computational method that uses freely available presence databases from the Global Biodiversity Information Facility (GBIF) and climatic data (worldclim 2.0) to compute presence maps for all genera included in the sample using ensemble models (Araújo & New, 2007). By combining the computations of several models though the ensemble model technique, a more robust presence map can be achieved. A site can then be geolocated using the presence models of all genera from that site. This method can be very effective, but there is a tradeoff between the surface area of the predicted region (specificity) and whether those regions indeed include the sites (correctness).

I hope my presentation gave the other APC members a view into some applications of palynology on a modern time scale. It certainly sparked a fun discussion on pollen- and vegetation samples coming from places that are relatively easy to reach and how illegal operations would prefer hard to reach places to avoid detection and how that could create bias in the reference database.

References

Araújo, M. B., & New, M. (2007). Ensemble forecasting of species distributions. Trends in Ecology & Evolution, 22(1), 42-47. DOI: 10.1016/j.tree.2006.09.010

Mildenhall, D., Wiltshire, P. E., & Bryant, V. M. (2006). Forensic palynology: Why do it and how it works. Forensic Science International, 163(3), 163-172 DOI: 10.1016/j.forsciint.2006.07.012 

3 Comments

  1. Sounds very cool. May I ask, what programming language you are writing your model in?

Leave a Reply to Simon Cancel reply

Required fields are marked *.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Blog at WordPress.com.
%d bloggers like this: