Thread by @Gangryong

Congratulations to Sally Reynolds and René Bobe, for this amazing new book! I’ll take this opportunity to talk a bit about the chapter I contributed to, led by @bobe_rene and co-authored by me, @carvalhoprimate and Meave Leakey.

The Koobi Fora Formation in the Lake Turkana Basin is one of the most important rock units in the world, and continues to provide one of the richest records of our evolution in Africa. #paleoanthropology

There are nearly 250 hominins documented from East Turkana. It includes periods of astonishing hominin diversity with Homo habilis, H. rudolfensis, H. erectus, and Paranthropus boisei occupying the region between 2—1.4 million years ago.
#fossilfriday

In this contribution we present a synthesis of previous paleoecological work at East Turkana and add new paleontological analyses relevant to understanding the ecology of early hominins. #paleoecology

My main role in this collaborative endeavor, was the introduction of a new method for analysis of faunal associations, using the APRIORI algorithm for #datamining of association-rules.

#rstats #arulesViz #arules

#APRIORI is used mostly for recommending products in e-commerce platforms by discovering item-sets of related products, e.g. if costumers are interested in book A the site recommends also book B, etc. based on a set of rules defined and learned by the algorithm.

We can also find rules in a database of extinct animals, e.g. “If Taxon-A (is present) then Taxon-B (is likely present)”. For our paleofaunal association analysis we focused on the period from 2 to 1.4 Ma. The results are provided in an interactive online platform, #FARUBO

FARUBO landpage ‘Data Exploration’ includes an ordered table with the most important faunal associations found by APRIORI, as scored by the parameters explained in the ‘About’ tab.

You can see it for yourself, here: osteomics.com/FARUBO

The ‘Paleofaunal Network’ tab provides a graph #dataviz of the results. Rule number (1 being the highest) = lift, circle size = support, circle shading = confidence. Remember that the results are dynamic and interactive, i.e., the model’s parameters can be changed by the users.

Using this method we found #Theropithecus to be a key taxon in predicting the occurrence of both Homo and Paranthropus.

When multiple taxa were allowed, the most important association found by APRIORI was (Metridiochoerus, Theropithecus) ⇒ (Paranthropus). In the chapter we discuss the potential ecological and taphonomic reasons, and the importance of these genera in East Turkana.

Theropithecus and Metridiochoerus combined potential as paleontological predictors for hominins in the region can be inferred from our results. Both are more abundant than #hominins, and thus are easier to detect during paleontological surveys.

As a last note, we present a new map of the Paleontological Collection Areas that revises previously published maps and an appendix with notes describing all the areas. These were drafted by Frank Brown (1943– 2017) and this chapter is dedicated to his memory.

cambridge.org/core/books/abs…