testing the quality of the fossil record through geological time alexander m. dunhill...

17
TESTING THE QUALITY OF THE FOSSIL RECORD THROUGH GEOLOGICAL TIME ALEXANDER M. DUNHILL [[email protected]] SCHOOL OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, U.K. 1

Upload: thomas-clement-maxwell

Post on 18-Dec-2015

216 views

Category:

Documents


3 download

TRANSCRIPT

TESTING THE QUALITY OF THE FOSSIL RECORD

THROUGH GEOLOGICAL TIME

ALEXANDER M. DUNHILL[[email protected]]

SCHOOL OF EARTH SCIENCES, UNIVERSITY OF BRISTOL, U.K.

1

THE INCOMPLETENESS OF THE FOSSIL RECORD

Sepkoski (1984)

Is our knowledge of the fossil record good enough to tackle macroevolutionary

questions?2

A BIASED FOSSIL RECORD...

Raup (1972)

Raup (1976)

Biodiversity in the fossil record is significantly influenced by sampling bias.

Most research focusing on geologically driven biases and subsequent collecting biases.

3

...OR COMMON CAUSE & REDUNDANCY

COMMON-CAUSE: geological and fossil records covary as both driven simultaneously by a common environmental agent.

4

REDUNDANCY: sampling and paleodiversity correlate as they are not independent signals.

Benton et al. (2011) from Fröbisch (2008)

Hannisdal (2011) Peters & Heim (2011)

SAMPLING PROXIES

“...metric that represents collecting effort in some way... should represent some or all of the geological and human factors that can introduce error into interpretations of data from the fossil record.” Benton et al. (2011).

Outcrop areae.g. Smith & McGowan (2005, 2007), Wall et al. (2009, 2011), Marx (2009), Uhen & Pyenson (2007) etc.

Formation Countse.g. Peters & Foote (2001, 2002), Butler et al. (2009), Barrett et al. (2009), Benson et al. (2010, 2011) etc.

Wall et al. (2009) Peters & Foote (2001)

5

SAMPLING PROXY PROBLEMS

Wall et al. (2009)

Sampling proxies are largely untested.

Global or continental scale studies.

Arguably vague sampling proxies (e.g. global geological maps) and crude estimations of paleodiversity.

6

GIS & REMOTE SENSINGPRECISE SMALL-SCALE CASE STUDIES

7

8

POSTER

T163. Geologic Timescale (Posters)Hall B, Poster booth no. 280

Wednesday 7th November, 2-4pm, 4.30-6pm

OUTCROP AREA = rock area that is displayed on a geological map. EXPOSURE AREA = rock area that is visible at the Earth’s surface.9

TESTING SAMPLING PROXIESOUTCROP vs EXPOSURE

Dunhill (2011, 2012)

TESTING SAMPLING PROXIESOUTCROP vs EXPOSURE

Dunhill (2011, 2012)10

CALIFORNIA NEW YORK

AUSTRALIA UK

TESTING SAMPLING PROXIESDIFFERENT ASPECTS OF SAMPLING DO NOT CORRELATE

Triassic-Jurassic UK; Dunhill et al. in review

Proxies for different aspects of sampling DO NOT consistently correlate.

Sampling proxy precision deteriorates when scaling up data.

Singular sampling proxies should NOT be used to correct diversity curves.

11

SAMPLING PROXIES & PALEODIVERSITYINCONSISTENT CORRELATIONS

Lower Jurassic SW UK; Dunhill et al. (2012)

Sampling proxies and paleodiversity show limited to no correlation in small-scale studies.

12

13

Triassic-Jurassic UK; Dunhill et al. in review

Sampling proxies and paleodiversity correlate in marine systems, but not in terrestrial systems.

SAMPLING PROXIES & PALEODIVERSITYINCONSISTENT CORRELATIONS

ACCURACY OF SAMPLING PROXIESFACIES DEPENDENCE

Facies and lithology effects are more pronounced than rock volume effects

(Triassic-Jurassic UK; Dunhill et al., in review) 14

VALIDITY OF SAMPLING PROXIESNON-INDEPENDENCE & MULTIVARIATE MODELLING

Triassic-Jurassic UK; Dunhill et al., in review

15

Sampling proxies can predict paleodiversity when included in multivariate models.

Sampling and facies effects are non-independent in their influence on paleodiversity.

Complexity of biasing factors confirms that singular sampling proxies (i.e. outcrop area) should not be used to correct the fossil record.

CONCLUSIONS

1. Sampling proxies representing rock volume are not good representations of amount of rock available for sampling.

2. Proxies for different aspects of sampling do not share a common pattern.

3. Sampling proxies and paleodiversity do not consistently correlate across:

(i) Different geographical and stratigraphical scales.

(ii) Different facies.

(iii)Different lithologies.

4. Multivariate models better predict paleodiversity – complex.

The use of a singular sampling proxy to identify and correct for bias in the fossil record is poorly supported. 16

ACKNOWLEDGEMENTS

Michael Benton (Bristol)Richard Twitchett

(Plymouth)Andrew Newell (BGS)

Bjarte Hannisdal (Bergen)Manabu Sakamoto (Bristol)

Graeme Lloyd (Oxford)Felix Marx (Otago)

Phil Donoghue (Bristol)Emily Rayfield (Bristol)Marcello Ruta (Bristol)

Alistair McGowan (Glasgow)

[email protected]@AlexDunhill

17