Is Natural Selection a Population-Level Causal Process? - Mechanism and Causality in Biology and Economics - page 166

The idea that drift is equivalent to an indiscriminate sampling process, as shown
in Millstein's account, is problematic. According to Millstein, natural selection is a
causal process in the sense that physical differences among organisms are causally
relevant to differences in reproductive success. The strength of natural selection
depends accordingly on the degree of causal relevance. By contrast, drift is
characterized in terms of causal irrelevance . Can the strength of drift depend on
the degree of causal irrelevance? Note that on Millstein's account, causal irrele-
vance, unlike causal relevance, cannot be a matter of degree. Causal irrelevance is a
matter of all or nothing, not a matter of more or less. If causal irrelevance were a
matter of degree, granting that “highly causally irrelevant” is equivalent to “barely
causally relevant” and that “moderately causally irrelevant” to “moderately caus-
ally relevant,” the clear-cut distinction between natural selection and drift would be
blurred. As a result, there might be cases where one and the same sampling process
is both natural selection and drift, with a variation in degree of strength. Such cases
would threaten to undermine Millstein's efforts to distinguish conceptually natural
selection from drift. In short, causal irrelevance, unlike causal relevance, cannot be
a matter of degree. Hence, although the strength of natural selection varies with the
degree of causal relevance, the strength of drift can vary neither with the degree of
causal irrelevance nor with the degree of causal relevance. In other words, causal
relevance is a factor of strength in the case of natural selection, but not in the case of
drift.
Let's consider a second factor that affects, as Millstein (personal communica-
tion) agrees, not only the strength of natural selection but also the strength of drift,
namely, the effective population size. Indeed, the effective population size, as a
factor of strength, affects any sampling process, discriminate or not . Now, is there
any connection between the two factors, namely, the size of population and causal
relevance? Here is Millstein's favorite example of drift:
To use Hartl and Clark's example, imagine shellfish that 'produce vast numbers of pelagic
larvae that drift about in the sea' (Hartl and Clark 1989 , p. 70). Although Hartl and Clark do
not elaborate, the image is of virtually identical larvae, subject to the vagaries of tides and
predators (i.e., indiscriminate sampling). (Millstein and Skipper 2007 )
Note that an indiscriminate sampling process can operate in any population,
whatever its size. Pelagic larvae among a large population would drift about in the
sea no less than larvae among a small population. Conversely, the size of population
has no impact on how indiscriminate a sampling process would be when it operates
in the population. The sampling process occurring in a large population of larvae
would be as indiscriminate as the sampling process occurring in a small population.
Now, if one reads “indiscriminate” as “causal irrelevance,” then causal irrelevance
has nothing to do with the size of population. A similar inference would lead to the
conclusion that causal relevance also has nothing to do with the size of population.
Not surprisingly, for any information about the size of population would be
unnecessary to tell whether or not physical differences among organisms are
causally relevant to differences in reproductive success. It follows that the two