EvoluTIonARy BIology 19
human history (see Chapter 3). Usually, however, we must infer
evolutionary history and its causes by interpreting less direct
evidence. Some historical events are inferred from fossils, the
province of paleontology (see Chapters 17 and 19). Other evo-
lutionary events are inferred from comparisons among living
organisms or by studying their phylogenetic relationships,
which provide a framework that enables us to draw conclusions
about the historical evolution of their phenotypic characteris-
tics and even their genes (see Chapters 2 and 16).
The causes of evolution, such as genetic drift and natu-
ral selection, are often studied by comparing data, such as
patterns of variation in genes, with theoretical models (see
Chapters 4–8). They are also studied by the methods of
experimental evolution, in which laboratory populations of rap-
idly reproducing organisms adapt to an environment (e.g., a
stressful temperature) designed by an investigator (see Chap-
ter 6). The adaptive reasons for certain characteristics (e.g.,
birdsong) may be inferred from experimental and other func-
tional studies, from their “fit” to a theoretical design (e.g., the
heart fits a “pump” design), or by comparing many popula-
tions or species to see if the characteristic is correlated with
a specific environmental factor or way of life (see Chapters
10 –13). Certain patterns of variation in DNA sequences can
tell us if natural selection has affected evolutionary changes
in genes (see Chapters 5, 7, and 14).
When we make inferences about history, or about past causes
of change such as natural selection, we do not see the changes
occurring, nor do we observe the causes in action. But throughout
science, causes are not seen; rather, they are inferred. All of chem-
istry, for example, concerns invisible atoms and orbitals that
govern the association of atoms into molecules. These theoreti-
cally postulated entities and their behavior have been confirmed
because the theory that employs them makes predictions (hypotheses)
that are matched by observed data. We know that DNA replicates
semiconservatively not because anyone has ever seen DNA do
that, but because the outcome of a famous experiment (and of later ones) matched
the prediction made by the hypothesis.
This hypothetico-deductive method, in which hypotheses are tested (and are
rejected, modified, or provisionally accepted), has been a powerful tool through-
out the sciences and is the basis of much evolutionary research. For example,
would you predict that the DNA in mitochondria carries more mutations that
are harmful to males than to females? There is no obvious biochemical reason
to expect this, but evolutionary theory makes such a prediction. The mitochon-
dria of both males and females are inherited from the mother; the mitochondria
in males are not inherited via sperm and are thus at a “dead end.” If a muta-
tion in mitochondrial DNA reduces the survival or reproduction of females, it
is unlikely to be transmitted to subsequent generations, but the transmission of
a mutation will not be affected if it is similarly harmful only to males, because
males do not transmit the DNA anyway. So, male-deleterious mitochondrial
mutations are expected to accumulate. This prediction, from the theory of natu-
ral selection at the level of the gene, has been verified: mitochondrial variants
commonly affect male, but not female, fertility in humans and other animals,
and they cause variation in reproductive gene expression in male fruit flies [6,Ernst Mayr, George C. Williams, John Maynard SmithWilliam D. Hamilton01_EVOL4E_CH01.indd 19 3/23/17 8:43 AM