An examination of the impact of Olson's extinction on tetrapods from Texas

Abstract

It has been suggested that a transition between a pelycosaurian-grade synapsid dominated fauna of the Cisuralian (early Permian) and the therapsid dominated fauna of the Guadalupian (middle Permian) was accompanied by, and possibly driven by, a mass extinction dubbed Olson's Extinction. However, this interpretation of the record has recently been criticised as being a result of inappropriate time-binning strategies: calculating species richness within international stages or substages combines extinctions occurring throughout the late Kungurian stage into a single event. To address this criticism, I examine the best record available for the time of the extinction, the tetrapod-bearing formations of Texas, at a finer stratigraphic scale than those previously employed. Species richness is calculated using four different time-binning schemes: the traditional Land Vertebrate Faunachrons (LVFs); a re-definition of the LVFs using constrained cluster analysis; individual formations treated as time bins; and a stochastic approach assigning specimens to half-million-year bins. Diversity is calculated at the genus and species level, both with and without subsampling, and extinction rates are also inferred. Under all time-binning schemes, both at the genus and species level, a substantial drop in diversity occurs during the Redtankian LVF. Extinction rates are raised above background rates throughout this time, but the biggest peak occurs in the Choza Formation (uppermost Redtankian), coinciding with the disappearance from the fossil record of several of amphibian clades. This study, carried out at a finer stratigraphic scale than previous examinations, indicates that Olson's Extinction is not an artefact of the method used to bin data by time in previous analyses.

Keywords: Olson’s extinction; Permian; Redtankian; Tetrapods; Texas; Time bins.

Figure 1. The time bins used in the diversity analysis.

(A) The cluster dendrogram indicating the grouping of the formations by CONISS; (B) the tetrapod bearing formations in Texas; (C) the Land Vertebrate Faunachrons (LVFs) redefined by CONISS; (D) the original LVFs.

Figure 2. Taxic diversity estimates.

Diversity estimates without correcting for sampling, using four different methods of time-binning the data. (A) Species level diversity estimate; (B) Genus level diversity estimate.

Figure 3. Subsampled diversity estimates (Original Land Vertebrate Faunachrons).

Numbers of species (A) and genera (B) in each land vertebrate faunachron (original definitions), corrected for sampling heterogeneity using shareholder quorum subsampling. Legend indicates quorum level.

Figure 4. Subsampled diversity estimates (redefined Land Vertebrate Faunachrons).

Numbers of species (A) and genera (B) in each land vertebrate faunachron (definitions based on CONISS), corrected for sampling heterogeneity using shareholder quorum subsampling. Legend indicates quorum level.

Figure 5. Subsampled diversity estimates (formations).

Numbers of species (A) and genera (B) in each formation, corrected for sampling heterogeneity using shareholder quorum subsampling. Legend indicates quorum level.

Figure 6. Subsampled diversity estimates (half-million-year time bins).

Means of the numbers of species (A) and genera (B) found in each half-million-year time bin in each of the 100 stochastic distributions of specimens, corrected for sampling heterogeneity using shareholder quorum subsampling. Legend indicates quorum level.

Figure 7. Extinction rates.

Median (thick black lines) of the extinction rates calculated for each half-million-year time bin in each of the 100 stochastic distributions of specimens at the genus (A) and species (B) levels. Dashed lines indicate standard error around the median.

Figure 8. Origination rates.

Median (thick black lines) of the origination rates calculated for each half-million-year time bin in each of the 100 stochastic distributions of specimens at the genus (A) and species (B) levels. Dashed lines indicate standard error around the median.