NaTUre HUmaN BeHavIoUr Articles
the white and melanistic carbonaria morphs of B. betularia in Caldy,
near Manchester. For this species, the melanistic carbonaria morph—
a classic case of directional evolution driven by natural selection^45 —
evolved at a rate of h 1 = 1.56 s.d. per year (Figs. 1 and 3e), but snail
shell colour polymorphisms, which fluctuate greatly in frequency
from one year to the next, evolved at a rate of h 1 > 3 s.d. per year.
The highest long-term rate of evolution was found in the three-spined
stickleback, G. aculeatus, of Loberg lake, Alaska^52 , whose armour
plating evolved at a rate of h 22 = 6.0 s.d. per year. Because these rates
are much higher than most published estimates of rates of organic
evolution, we also applied the log rate – log interval method6,7,
and found very similar results (Supplementary Fig. 2). All our
polymorphic traits showed h 1 rates in excess of 0.7 s.d. per year,
regardless of whether they were evolving directionally or not.
Our data also include several sets of continuous traits, the most
well known of which are from Peter and Rosemary Grant’s study
of Darwin’s finches, G. fortis and G. scandens, in the Galapagos46,
(Figs. 1 and 3h). Compared with polymorphisms, continuous traits
had low h 1 rates (Supplementary Table 2). Only two traits had
h 1 > 0.2 s.d. per year: age at maturity in Atlantic cod and horn size
in bighorn sheep, both of which were subject to strong fishing or
hunting pressure54–56.
Polymorphisms are due to alleles at a few loci and are under
strong genetic control, meaning that their Haldane rates are true
rates of evolution. By contrast, changes in continuous traits may
be caused partially, or even entirely, by phenotypic plasticity (for
example, in refs. 57,^58 ), so their Haldane rates are properly viewed
as rates of phenotypic change. Whereas genetic studies confirm
that the observed change in each of our continuous traits is par-
tially heritable46,53,55,59,^60 , we view them as upper bounds on rates
of evolution.
How do rates of artefact and organic evolution compare?
Biologists have long been keen to demonstrate evolution in action,
and many of these populations were studied precisely in the hope
that they might do so. For this reason, our data are surely not rep-
resentative of organic traits in general. They do, however, suggest
upper limits to the rate of organic evolutionary change. Notably, the
upper limits of the distribution of h 1 for all four artefact distribu-
tions are well below the upper limits of the distribution of organic
traits (Fig. 4a,c and Supplementary Table 3). This is mostly due to
our inclusion of polymorphisms, which have generally been ignored
in compilations of evolutionary rates6,8,9; however, even if we exclude
them, it is clear that the upper limits of cultural and organic traits
are of the same order of magnitude, 0.1 < h 1 < 1.0 s.d. per year. This
is also true whether we consider vertebrate or invertebrate species
alone. We have fewer 25 year interval data for organic populations,
but again no artefact trait has a h 25 greater than the most rapidly
evolving organic trait (Fig. 4b,d). Hunt has shown^29 that, even in
the absence of evolution, Haldane rates have a small but finite value
due to sampling error, which may dominate when sample sizes are
small. The lower bounds of the estimated Haldane confidence inter-
vals of most of our traits, however, exceed this minimum value, at
least in the long term (Supplementary Fig. 3); that is, our traits have
evolved. We conclude that rates of cultural and organic evolution
are surprisingly comparable and that some organic traits can evolve
much faster than any cultural trait we observed.
Although we focused on populations and traits with time-series
data, we did consider some point estimates of evolutionary rates
available in the literature. We were particularly interested in a study
on guppies evolving in response to a novel predation pressure^61 ,
because the observed rates in Darwins were said to be very high,
and, importantly, were based on animals raised in a common garden
and so reflect genuine evolutionary change. The highest rate in these
guppies (h 4 = 0.26 s.d. per year for male age to maturity in El Cedro
river fish) is about 4.5 times faster than the fastest Darwin’s finch
beak trait over comparable intervals, but slower than those of many
polymorphisms. We also reanalysed a dataset based on point esti-
mates of phenotypic change in 49 different species and traits^62 and
found that the mean Haldane rate was h≤ 4 =. 008 s.d. per year. The
fastest evolving trait in this dataset was a morphometric trait in a
population of sticklebacks from Iceland, where h 1 = 0.71 s.d. per year
(Supplementary Table 4).
0.
0.
0.
0.
0.
01020304050
0.
0.
0.
01020304050
0.
0.
0.
0.
01020304050
0.
0.
0.
0.
0 20 40 60
0
5
10
010203040
0
2
4
6
0510 15 20
0.
0.
0.
0.
010203040
h Gf (bh)
Gf (bz)
Gs (bh)
Gs (bz)
Os (hd)
Os (hl)
Gm (f)
Jh Gm (m)
Ga
Pm
g
Cn (banded)
Cn (brown)
Cn (pink)
Tp (A)
Tp (O)
f
Bb
Pd
0.
0.
1.
1.
2.
020406080
Interval (years)
e
h (s.d. per year)
a b c d
Fig. 3 | rates of evolution of artefact and organic populations. a – h , rates of evolution in absolute Haldanes ( h ) of artefact and organic populations.
Populations as in Fig. 1. Note that the scales of the axes vary among plots.
