Natural Selection and Darwin's Finches
by Martin Wikelski, University of
One of the classic studies in the evolution of natural
populations was conducted by Rosemary and Peter Grant and
coworkers on Darwin's finches. It is among the firstand is
certainly the most elegantstudy to document evolution in a wild
population of vertebrates. Rosemary and Peter investigated
Darwin's finches for almost 3 decades and conducted most of their
field work on two small islands, Daphne Major and Genovesa, in
the Galápagos archipelago, Ecuador.
About 30 years ago, Peter Grant started
from the question of how interspecific competition influenced the
formation of animal communities. He decided to study the
'classic' case of character release. David Lack suggested in 1947
that the bill sizes of Darwin's ground finches differ between
islands, depending on the presence or absence of other
(competing) species of Darwin's finches. However, there was
hardly any knowledge of why this should be so. Nevertheless, the
data are very clear-cutthe beak sizes of medium ground finches
(Geospiza fortis) from Daphne Major are smaller than the
ones from the same species on the nearby island of Santa Cruz
Island, although the islands are only about 10 kilometers apart.
The articles by Boag and Grant (1984) summarize the enormous
body of their work.
P. T. Boag and P. R. Grant. 1984.
The classical case of character release: Darwin's finches
(Geospiza) on Isla Daphne Major, Galápagos. Biological
Journal of the Linnean Society 22: 243-287.
The prediction of evolutionary change is the real goal in
evolutionary analysis. To make clear how the measurements from
previous studies can be used to predict microevolutionary change,
P. R. Grant and B. R . Grant.
1995. Predicting microevolutionary responses to directional
selection on heritable variation. Evolution 49: 241-251.
To truly appreciate the scope and detail of their research,
consider these points:
- The study of evolution is necessarily a long-term goal,
especially in natural vertebrate populations. Observing
the finch population of Daphne Island for three decades
revealed that natural selection can push traits of
animals into different directions, depending on the
- Rosemary and Peter followed individual birds and
determined both their fate, as well as, the fate of
their offspring. It cannot be emphasized enough how
important data on individuals are to an understanding of
evolutionary change. We can only observe evolution if we
know how the offspring of individually known parents
succeed. Furthermore, nearly all finches of the entire
island of Daphne Major were individually known. This is
unique for a natural population of this size. It shows
that not only a subset of a population has been studied,
but the entire population.
- There are very strong episodes of selection in the
Galápagos islands, caused by both unusually wet (El
Niño) and unusually dry (La Niña) weather conditions.
These selection episodes were witnessed by Peter and
Rosemary, who analyzed the quantitative responses of
birds to those events. Thus, although the Grant et al.'s
studies are not experimental studies in a strict sense,
the environmental perturbations are the equivalent of
- The best test to see whether we really understand
evolution is if we can predict evolutionary change under
a set of environmental conditions. To achieve a
prediction, we need to understand how natural
selection works. This implies that one has to measure all
the ecological details like the force it takes to crack
seeds, the availability of seeds in the environment and
Boag and Grant examined four hypotheses to explain the evolution
of the intermediate beak sizes of G. fortis on Daphne
- Genetic drift and/or founder effect contributed to the
intermediate beak sizes.
- Hybridization with the small ground finch (Geospiza
fuliginosa), which is also present on the island,
contributed to the intermediate beak sizes.
- Local adaptation to food supply was influential in
contributing to intermediate beak size.
- Character release in allopatry (i.e., in a situation
where only one of several species exists).
Boag and Grant captured (with mistnets) and measured the beaks
of hundreds of finches on Daphne and Santa Cruz islands to find
out whether the formerly observed character release was still
present. Captured birds were banded with individually numbered
aluminum and colored plastic rings. Nestlings were banded each
year during the breeding season. Ecological parameters like plant
cover, seed abundance, seed diversity, feeding behavior, rainfall
etc. were measured during the same periods.
Most importantly, the data confirmed that indeed the medium
ground finches from Daphne were smaller than the conspecifics
from nearby Santa Cruz. With regard to the four hypotheses, Boag
and Grant could show that:
- There was no strict isolation of the Daphne population
from the Santa Cruz population, because birds fly between
the large (Santa Cruz) and the small island (Daphne).
Furthermore, immigrating birds mated and reproduced on
Daphne. Although immigrants occur at a very low
frequency, immigration is enough to counteract the
effects of drift. Thus, genetic drift is unlikely to
account for the differences in beak sizes between the two
populations of the same species.
- Hybridization does occur between the medium and the small
ground finch--between 1.1 and 3.5% of breeding pairs
were hybrid pairs. Because the variation in beak size has
a genetic component, i.e. beak size is heritable, they
expected the hybrid offspring between the medium and the
small ground finch to have smaller beaks than pure medium
ground finches. However, compared to the large selection
pressures exerted by changes in environmental conditions,
the influence of hybridization on beak sizes is small.
Please also note that hybridization between the medium
ground and the cactus finch pulls beak size upwards.
- Natural selection acts through the size of food items
that are available. This was the most important factor
determining beak size for each population. If there are
only large food items available, birds with large beaks
can crack them better and thus survive better. To the
contrary, if the seed size is very small, then small
birds survive better. During the current study, large
birds survived better.
- Boag and Grant also found evidence for character
releasemedium ground finches expanded their diet during
periods of food shortages. Boag and Grant could also show
that there was indirect competition between the medium
ground finch and the cactus finch by determining the time
course of diet similarities between different finch
In summary, the last three hypotheses probably
contribute to the difference in morphology between the Daphne
and the St. Cruz island populations.
The main conclusion from this study was that the medium ground
finch phenotype "... represents a balance between
introgression with G. fuliginosa, selection for larger ...
size in dry years and selection for smaller ... size in wet
years." The major importance of this study was that the
evidence for all hypotheses could be quantified in a natural
situation. From these (and many other) results, Rosemary and
Peter Grant could go ahead and predict evolutionary
change. The importance of a prediction cannot be emphasized
enough. A quote from the famous evolutionist J. B. S. Haldane
might make this clear (after Grant and Grant 1995): "No
scientific theory is worth anything unless it enables us to
predict something which is actually going on. Until that is done,
theories are a mere game of words, and not such a good game as
A sample data set is
available for this case study. The file is an Excel spreadsheet
This exercise will allow you to verify the conclusions arrived at
by Boag and Grant. The data file contains representative data
from the Boag and Grant study. The following procedures provide a
general guide to help you analyze the data. (Bold
letters refer to the variables in the data file.)
- Classify the beak depths (Daphne Island
and Santa Cruz Island) into
classes that range 0.2 mm (thus, one class would contain
all values from 8.01 to 8.2 mm beak depth, the next class
all values from 8.21 to 8.4 mm beak depth).
- Determine the number of animals in each class. Relate
this number to the total number of animals (i.e.,
determine the percentage of animals in each class).
- Use these values to plot histograms of the beak depths of
medium ground finches from Santa Cruz, Daphne birds in
1976, and Daphne birds in 1978 (those individuals that
survived, i.e. Survived = 1).
- Determine the mean beak depth of the parents (Mother
Beak and Father Beak) for each offspring in
- Plot this mean value (the "midparent" beak
depth) on the x-axis against the offspring beak depth on
- Fit a regression line to those data and determine whether
there is a significant relationship. If this is the case,
determine the slope of the regression. The slope is a
measure of the heritability of the trait 'beak depth'
between parents and offspring, reflecting the degree to
which resemblance is due to shared genes.
What does your analysis of this data suggest? Does it support
the conclusions drawn by Boag and Grant?
- Look at the extremely high values for beak depths for
medium ground finches from Daphne. These values exceed
the values for the medium ground finches from St. Cruz.
Why is this so? Discuss what this could mean for the
species concept. Do you think these birds are hybrids
between medium and large ground finches? Also consider
the fact that different numbers of birds were measured on
both islands (see also point 2).
- Determine the coefficient of variation (calculate the
standard deviation, multiply by 100 then divided by the
mean) for each group of birds. What does it tell you with
regard to the force of natural selection between 1976 and
- Plot the histograms both as percentages and as absolute
values. Does this demonstrate the strength of natural
- Plot the means and standard errors of beak depths of
birds from St. Cruz, Daphne 1976 and Daphne 1978. If you
use two standard errors in each direction (which then is
the 95% confidence interval), you are able to estimate
whether two means are different at the 0.05% level. As an
approximation, when the 95% confidence interval does not
overlap with the mean of another value, the two values
differ on the 0.05% level. Also compare the absolute
difference in beak depth before and after the episode of
selection. By how many millimeters does the mean beak
- Determine the residuals of the regression of midparent
beak depth against offspring beak depth. How large are
the residuals and what does the amount of residual
variation tell you?
by Prentice Hall, Inc.
A Pearson Company Legal Notice