GEOS 26400/36400/BIOS 23255/EVOl 32300: Principles of Paleontology

Diversity, Origination, and Extinction, II (Error and Bias)

I. Error vs. Bias
      -Incompleteness: observed diversity less than actual
      -Error: factors that add noise but don't systematically distort temporal pattern
      -Bias: factors that systematically distort temporal pattern
      *NB: Probability of our knowing about a particular taxon in a particular
stratigraphic interval is a compound probability: 
            *Pr(known)=Pr(preserved)*Pr(collected)*Pr(described)*Pr(published) etc.
            **Therefore, various biases mimic each other.
            **Therefore, various biases can (in principle) be corrected in a similar way.

II. Monographic effects
      -General concept: effects of interests and practices of particular workers
      -Paleontologic interest
            --strata or regions receiving more attention have higher apparent diversity
		*fossil record best known from Europe and North America
            --example: shift in times of apparently high brachiopod origination and extinction (Williams 1957)
            --Does interest dictate diversity or vice versa?
            --Possible solution: Rarefaction (number of observed taxa versus number of
                  recorded observations)  (See section VIII below.)
      -Differences in taxonomic practice
            --often relate to "lumping" vs. "splitting", but also much more complicated
	       --see Williams's (1957) brachiopod study
            --Difficult to correct analytically
      -Boundary effects
            --sometimes different workers study fauna on either side of real extinction boundary
                  --Lack of standardization enhances apparent size of extinction.
                  --E.g. Cambrian-Ordovician trilobites
      -Monographic effects can be recognized but hard to correct for.
		**taxonomic standardization

III. Differential sampling (between groups)
      -Solutions: 
            1. Make comparisons only among taphonomically comparable groups.
            2. Use taphonomic control taxa.
                  -Take absense of taxon at face value only if you find a
                  taphonomically/ecologically similar species.

IV. Lagerstaetten
      -Transient preservation of soft-bodied species ==> orig., ext. & diversity spikes.
      -General approach: omit Lagerstaetten from analysis
      -Problem: no perfectly clear distinction between Lag. and non-Lag. deposits
      -Another possible solution: Omit single-interval taxa from analysis

V. Systematic change in completeness over time
      -On first principles, expect completeness to increase toward Recent.
            -Less time for erosion, burial, and metamorphism of fossiliferous rocks
      -This should tend to inflate apparent diversity as you approach Recent.
      -Possible Example: Taxon-level effect
            --as completeness increases, so does apparent ratio of lower to higher taxa
            --this pattern has been interpreted ecologically, but may be partly artificial

VI. Edge effects
      -Fossil record (overall, or for particular section or region) has beginning and end.
       *Result: ability to infer presence of taxa by range-through approach diminishes
            toward beginning and end of record.
         *Results:
            1. Apparent diversity artificially low toward beginning and end.
            2. Origination rate artificially high toward beginning.
            3. Extinction rate artificially high toward end.
      -Edge effects can be corrected (to some extent) if we have estimate of sampling rate. 
      -Edge effect no longer felt within a couple of average taxon lengths.
            *Avoid edge effect by avoiding edges.

VII. Pull of the Recent
      1. Increase in exposed sedimentary rock toward Recent
      2. Extant taxa have only one end of geologic range truncated.
            -The closer a taxon is to Recent, the more likely it is to be extant.
            -Apparent diversity inflated toward Recent, even if fossil record sparse.
            -Possible solutions: throw out extant taxa or ignore their Recent occurrences
      3. If clade still extant, we may see how fine-scale morphological features aid in
       species discrimination.
            -We may then be able to make finer distinctions among fossil forms.
            -Relative to clades that are extinct, this may lead to higher diversity
            estimates and shorter taxonomic ranges.
     4. Conflicting results on empirical importance of Pull of the Recent
		a. Paleobiology Database (unpublished)
			*Suggests many long-ranging, extant genera with sparse fossil record
	 	b. Jablonski et al. (2003)
			*Suggests nearly all bivalve genera that range through the Plio-Pleistocene are 
			sampled during that time interval
		c. Possible reasons for discrepancy: 
			*"random" sampling in PBDB vs. concerted search for known genera by Jablonski et al.
			*bivalves may be among better preserved and sampled taxa

VIII. Effect of rock area and volume (and extent of sampling more generally)
      -Amount of sedimentary rock correlates with observed taxonomic richness.
      *Possible example: taxonomic diversity during Ordovician
            -Raw data show continued diversification through Late Ordovician.
            -Data adjusted with rarefaction suggest diversity peaked earlier.
		*This general approach of "sampling standardization" now possible on global
		scale for entire Phanerozoic, thanks to growth of Paleobiology Database 
		(see below).
      -Two possible explanations for correlation between observed diversity and amount of
            preserved rock:
            1. amount of rock dictates perceived diversity artificially
            2. rock and diversity both affected by common factor (e.g. habitable shelf area)
      -Bambach's work (and later studies) on diversity within communities 
		suggests species richness may have truly increased since the Paleozoic.
                  -But isn't it easier to extract fossils from younger sediments?
			*evidence from recent studies of lithified vs. unlithified sediments
				--unlithified samples contain not just more specimens, but a broader range
				of species in terms of body size and abundance: more species per unit sample size
      -Sampling-standardization of Phanerozoic, marine data (Paleobiology Database)
		*global scale (cf. Bambach)
		*suggests much more muted post-Paleozoic radiation than seen in raw genus data
      -Use of numerical optimization to estimate true orig. and extinct. rates from first & last appearances
		-suggests true rates much more volatile than face-value rates ("face-value" mean taking
		first and last appearances as accurate proxies for originations and extinctions)
		-extinction captured with higher fidelity than origination (i.e. better match between
		face-value and optimized rates)
			-this may partly reflect temporal coincidence between extinction events and decline
			in quality of the record (common-factor hypothesis)
		-comparison between face-value rates, optimized rates, and "turnover" of stratigraphic
		sections also provides evidence for common-factor hypothesis