2015 GSA Annual Meeting in Baltimore, Maryland, USA (1-4 November 2015)

Paper No. 42-2
Presentation Time: 9:00 AM-5:30 PM


ZHOU, Heather J.1, WANG, Daniel1, WANG, Steve C.1, WANG, Chengying1, GAI, Linda1, MOORE, John L.2, PORTER, Susannah M.3 and MALOOF, Adam C.4, (1)Mathematics and Statistics, Swarthmore College, 500 College Ave, Swarthmore, PA 19081, (2)Earth Research Institute, University of California, Santa Barbara, Santa Barbara, CA 93106, (3)Department of Earth Science, University of California, Santa Barbara, Santa Barbara, CA 93106, (4)Department of Geosciences, Princeton University, Princeton, NJ 08544, jzhou2@swarthmore.edu

The Signor-Lipps effect has typically been discussed in the context of mass extinctions, but it can also arise in their mirror-image equivalent, "mass origination" events. The most famous such example is the Cambrian explosion, a pivotal event in the history of life. Although the broad outline of the Cambrian explosion is now well understood, details such as its duration and pattern of origination events are not well constrained. Maloof et al. (2010) found that a dataset of small shelly fossils through the earliest Cambrian showed three pulses of diversification over approximately 16 million years, but they were unsure whether these features were real or artifacts due to the incompleteness of the fossil record. Here we use novel methods to estimate the duration and number of pulses in the Cambrian explosion while accounting for Signor-Lipps–type effects.

We used a revised a dataset of fossil occurrences of 166 genera of small shelly fossils from Mongolia, Siberia, and China, dating from the earliest part of the Cambrian (Nemakit-Daldynian and Tommotian, or Terreneuvian). To estimate the duration of the origination event, we calculate a confidence interval for the time span between the true first and last originations, both of which are likely to be underestimated in the fossil record. The confidence interval is constructed by inverting a hypothesis test for whether a given duration is consistent with the observed fossil occurrences. To estimate the number of pulses, we take a classification-based approach using a k-Nearest Neighbor (kNN) classifier. This classifier takes as its input a vector of AIC and BIC weights, which are based on the likelihood of each possible number of pulses given the observed fossil occurrences. Both of these methods are based on methods originally developed for mass extinctions, but modified so that uniform recovery potential is not assumed.