Charles Lyell developed an ingenious technique to estimate extinction rate by charting the gradual diminution of living species as one went back in geological time.. Problems in Measurin
Trang 1We could calculate the number of taxa that become extinct or
the percent of the former pool of taxa that became extinct
Time could be measured in years, but often we only have
segments of relative geological time units such as geological
stages (A set of stages comprises a geological series, and a set
of series comprises a geological period (e.g., Cambrian,
Cret-aceous).) In many parts of the fossil record, the absolute time
represented by a stage is not accurately known, and different
geological stages are often of great difference in temporal
extent
Charles Lyell developed an ingenious technique to estimate
extinction rate by charting the gradual diminution of living
species as one went back in geological time This type of
an-alysis can give longevities and extinction rates Such Lyellian
curves demonstrate, for example, that the diminution of
bi-valve species on the Pacific coast of the USA is at a steady pace
whereas a more precipitous extinction occurred in the Atlantic
Problems in Measuring Extinction Rates
Taxon-Level Bias
We typically think of extinction rate as a measure of the loss of
species To create a database for paleontology, the species level
is very difficult to trust with any degree of confidence; most
paleontologists tend to trust the genus and higher taxonomic
levels in identifications In recent years, more and more effort
has been directed toward accounting for the ranges of all
named species in the fossil record, but most analyses have
been done at the family or genus level The large-scale
data-base we now employ owes its existence to the dedicated work
of David Raup and especially the late Jack Sepkoski, who
continuously sought to produce a more complete database of
the geological ranges of all fossil groups Initially, the
com-pilation was at the level of taxonomic order but subsequent
analyses have moved down the taxonomic hierarchy to the
family and generic levels
Can extinctions of higher-level taxa be used to estimate
species-level extinctions? To estimate species richness using
numbers of higher-level taxa (e.g., orders), we assume that
taxonomic diversity at higher taxonomic levels is correlated
with species richness, but they are not necessarily correlated in
this manner This can be seen clearly where changes in ratios
of one taxonomic level to another occur over broad spans of
time If the ratios change, then higher taxonomic units might
be flawed estimators of changes in species diversity For
ex-ample, the ratio of taxonomic orders to families decreased
significantly from the Paleozoic to the Mesozoic Era
Over short periods of time, the number of taxa at a higher
taxonomic level (e.g., level of family) might have a regular
relationship with a lower taxonomic level, such as species To
estimate species-level extinction from family-level extinction,
David M Raup used a rarefaction technique based on the
sampling curve that relates the number of species collected at
random to the number of families recovered This method has
been modified and refined to be used in many estimates of
standing fossil diversity
The rarefaction approach is the best we have so far
Nevertheless, we must be careful in applying it The biggest
problem is the potential change in the relationship over
geological time For example, the ratio of families to species decreases by a factor of two from the Mesozoic to the Ceno-zoic, and other cases are known of changing ratios of species
to genera Selective extinction can also bias our conclusions For example, certain families may be much more prone to extinction, due to their presence in a particularly vulnerable habitat (e.g., coral reefs during a cooling event) This might overestimate total extinction, if these are added to a larger species list It is also difficult to get sufficient data to calculate good rarefaction curves for all but the most abundant fossil groups
Raw Data for Analyses
Any set of diversity data involves a count of the number of taxa But this number is strongly modulated by the intensity of sampling Thus diversity could be inflated simply because more studies have been done in a given locality at a given time interval Thus raw data must be corrected for sampling in-tensity In most instances, the genus is the level used for di-versity estimates, so the number of genera must be estimated
by some sort of sampling process that corrects for the total number of studies, the number of localities, and the number
of fossils that are collected at any one locality and time hori-zon Two examples of these potential biases are the possible monograph effect of studying intensively one part of geo-logical time and a bias that inflates the diversity of parts of the geological record just before the present, because species are still alive and can be sampled and studied more completely than their antecedents
Biased Preservation and Convergence
Estimates of extinction rates may be biased by preservation and abundance at the time of extinction Preservation of ap-propriate habitats during an extinction may be greatly re-duced Thus a species might have survived, but there are no opportunities to see it because its usual facies of occurrence has not been preserved
A common change in probability of preservation occurs when a systematic change in rock preservation occurs, as in the reduction of deposition during a regression phase of the sea,
as at the end of the Permian and just before the end of the Cretaceous (regression refers to a lowering of sea level in a given area; transgression refers to a rise in sea level) Suppose the ranges of a group of species all ended at the very terminus
of the Cretaceous A gradual reduction of deposition would,
by sampling error alone, give the impression of a gradual disappearance of the fossil species Even if deposition does not decline, previously rarer species would be difficult to sample for presence during a general decline in abundance during extinction, just because we would be unlikely to find them These biases have come to be known as the Signor–Lipps ef-fect, or ‘‘backward smearing,’’ because a sharp extinction might appear to be gradual from fossil sampling Only abundant forms would be sufficiently ‘‘findable’’ that we could assess their total geological range with confidence, especially up to the time of their extinction Even at smaller levels of geological resolution, the problem of a spurious component of estimated diversity due to the preservation of different amounts of rock
of a given age is a major source of bias The problem is compounded by cases where some environmental factors
414 Extinction in the Fossil Record