48 John R. Paul and Stephen J. Tonsor
the impact of historical processes on the current
distribution and abundance of species.
EMPIRICAL TESTS OF AGE
AND AREA
There have been few explicit tests of the age-
and-area hypothesis. Two studies of marine
fossil fauna have found evidence that indirectly
supports a positive age and area relationship.
Jablonski (1987) documented a positive relation-
shi pbetween age (s pecies duration) and geo-
graphic range size in the beginning of fossil
mollusk species’ lifetimes, followed by long peri-
ods of stasis, but the focus of that study was
on the possibility of species-level selection, rather
than testing age and areaper se. Similarly, Miller
(1997) found that in Ordovician marine gen-
era, older genera had larger ranges. Studying
birds, Gaston and Blackburn (1997) found that
for the entire New World avifauna, there was
no relationshi pbetween mean range size of a
clade and clade age, but there was a weak pos-
itive relationshi pbetween evolutionary age and
total clade range size. In another study, Webb
and Gaston (2000) examined six clades of birds
and found various forms of the age and range
size relationship. Overall, roughly 20–50% of the
variance in range size could be accounted for
by species age (inferred from standard mitochon-
drial DNA molecular clock divergence estimates
of 2% divergence per million years, Ma), but only
one clade showed a positive age and area rela-
tionship; three showed a negative relationship
and two a hump-shaped relationship. A study on
Sylviawarblers found a weakly significant posi-
tive relationshi pbetween breeding range size and
species age, but in that study the relationship
could be better explained by older species gener-
ally having better dispersal abilities than younger
species (Böhning-Gaeseet al. 2006). Finally, Jones
et al. (2005) analyzed large molecular datasets
of primates and carnivores and found evidence
of a weakly negative age and area relationship
(see that paper for a more detailed overview of
Willis’s age-and-area hypothesis and approaches
to testing it).
Overall, a convincing positive age and area
relationship predicted by Willis is not supported
by these empirical data. However, a careful look
at the published data reveals two trends. First,
analyses that use fossil samples and measures of
species duration as a proxy for age tend to find
some evidence for a significant age and area rela-
tionshi p(e.g., Jablonski 1987, Miller 1997). In
contrast,studiesthatexamineextantspeciesusing
molecular divergence dates as a proxy for age
generally tend to find either no significant rela-
tionshi pbetween s pecies age and range size, or
a mixture of positive and negative relationships
(see table 7.1 in Joneset al. 2005). This discrep-
ancy may be due, at least partly, to the different
sampling methods. For example, a species’ fos-
sil record potentially allows sampling along the
entire history of a species’ range-size trajectory
over time (Figure 4.1a). This is the ideal situation,
in which the range size for a given species can be
estimated at multiple ages. In contrast, molecular
dating methods generally permit a single snapshot
of a species’ age and range size at a given point in
time, and by looking at multiple species we can
infer the general trend of the age and area rela-
tionshi pfor a grou pof organisms. Having only
snapshots of a species age and range size relation-
shi pcan introduce considerable variance into the
relationship, particularly if all species follow vary-
ing range-transformation trajectories over time
(even if the general shape of the relationship is
similar, e.g., hump-shaped; Figure 4.1b). However,
it is likely that the majority of future age estimates
for most taxa will be derived through molecular-
based inference; thus, understanding how these
measures can potentially bias relationships such
as age and area is critical to robust interpretation
of results.
In addition to the potential discrepancies intro-
duced through fossil versus molecular analysis
of age and area, studies on extant species sug-
gest that the phylogenetic level of the analysis
is important. In studies of large clades contain-
ing many well-defined and potentially divergent
subgroups (e.g., mammals, carnivores, or birds),
general analyses of age and area find no or weak
relationships ( Joneset al. 2005), while studies
of individual clades within these broad groups
often find significant, but inconsistent, relation-
ships (e.g., the six clades of birds studied by Webb
and Gaston 2000). This discrepancy suggests that
the signal of an age and area relationshi pmay be