Explaining Geographic Range Size by Species Age 57prediction of a positive species age and range size
relationship. Thus, the more important question
may be, when does a positive age and area rela-
tionshi pcease to exist, and why? Furthermore,
analyses that examine clades of species and ask
if on average rare species are younger than old
species, rather than simply looking for a positive
slope of an age and area relationship, may be more
informative.
Second, the positive age and area expectations
of most models of post-speciation transformation
are primarily driven by the assumption that new
species start with small population sizes. But do
they? It has been asserted that much speciation
in tropical woody plants arises through isolation
of small local populations (e.g., Ehrendorfer 1982,
Leighet al. 2004), but strong empirical evidence to
support this position is generally lacking. Since the
population sizes of new species cannot practically
be measured, inference must be used to estimate
the sizes of ranges and populations. For example,
fossil evidence supports African large-mammal
populations starting as small, narrowly ranging
populations (Vrba and DeGusta 2004). Unfortu-
nately, the sparse fossil record for many taxa,
particularly plants in the tropics, makes inference
based on fossil evidence rare. The data presented
here forPiperare certainly suggestive that newer
species have small range sizes, as evidenced by the
preponderance of young species with small range
sizes and the lack of young species with large
ones. Future analyses of age and area relation-
ships in tropical plants may help to fill in the gaps
of our knowledge of new species population and
range sizes that are unlikely to be filled by fossil
evidence.
Third, a practical difficulty arises from using
divergence times of species as proxies for ages.
When speciation is defined as a cladogenic
(splitting) event, such as on a dichotomously
branching phylogenetic tree, any speciation event
yields at least two new species, both assigned the
same age. These new species have range and pop-
ulation sizes defined by the boundaries of their
newly isolated gene pools (or lineages). Thus,
when speciation is viewed as a splitting process
with a geographical component, new species will
oftenhavesmallerrangeandpopulationsizesthan
their direct ancestor, because the ancestral range
(and the distribution of individuals defining it) is
subdivided. If the relative range and population
sizes of sister species are markedly skewed, there
will be considerable variance in the distribution
of population sizes of the new species. For exam-
ple, when a new species (B) is introduced via a
point-mutation model of speciation (where one
individual is assigned a new species status based
on some new defining character,sensuHubbell
2001a), its ancestor species (A) with population
sizeNmust also be deemed a new species (C), with
a population sizeN– 1. Since species B and C are
assigned the same age, the youngest species in the
community are represented by species with both
small (B) and large (C) population and range sizes.
In other words, when a widespread species gives
rise to a narrowly endemic sister species, but the
widespread species persists essentially unchanged
in its ecological and genetic attributes, both sister
species are assigned the same age. This is poten-
tially at odds with the meaning of species age in an
evolutionary sense. It also clearly creates difficulty
in analyzing age and area, as such a process will
obscure any expectation of a positive relationship
if such asymmetric range splits are commonplace
in a clade. In light of this potential source of noise
in the age and area relationship, it is all the more
remarkable that a positive relationship explaining
a good portion of the variance in range size was
found in our analysis ofPiperspecies.
Finally, molecular age estimates are potentially
subject to many different kinds of errors and
uncertainties (Arbogastet al. 2002). For example,
the model of molecular evolution used, the degree
of consensus between gene trees examined and
true species trees (Nichols 2001), the reliability of
any fossil ages used for calibration, and success of
an analytical model dealing with rate heterogene-
ity can all introduce potential errors in estimates
of ages (Sandersonet al. 2004, Renner 2005).
In summary, future studies on age and area
relationships in tropical plants have the potential
to provide insight into the role that the sim-
ple explanatory variable species age can play in
explaining patterns of rarity and endemism. Of
course, as Willis himself recognized, age by itself
cannot be the mechanistic driver of these patterns
we observe. Rather, age acts as a proxy for the
playing-out of various ecological interactions at