ISSN 1022-7954, Russian Journal of Genetics, 2007, Vol. 43, No. 3, pp. 338–340. © Pleiades Publishing, Inc., 2007.
Explaining the remarkable >80 000-fold variation in
haploid genome size among eukaryotes is certainly
among the most important long-standing problems in
evolutionary biology . This variation is particularly
puzzling given the absence of an obvious association
between genome size and gene number or organismal
complexity, a phenomenon commonly known as the C-
value paradox [2, 3]. Although several hypotheses have
been put forth to explain the C-value paradox, one of
them received considerable attention in the past decade:
the suggestion that biases in the rates of insertions and
deletions (indels) might account for the observed dif-
ferences in genome size among organisms, the DNA
loss hypothesis [4–7]. According to Petrov , the
mechanism for variation in genome sizes among organ-
isms lies in the thermodynamics of replication slippage.
An insertion requires the melting and rereplication of a
segment of previously duplicated DNA, whereas dele-
tions involve only a skipping of unreplicated bases.
These processes would cause deletions to be more fre-
quent than insertions. Variation in the magnitude of this
bias would lead to genomes of different sizes. The main
evidence put forth by proponents of the DNA loss
hypothesis is a highly signiﬁcant negative correlation
between genome size and relative DNA loss rate mea-
sured as bases lost per base pair substitution, with more
than 96% of the variance in the dataset being explained
by this relationship .
Before this relationship can be accepted as valid evi-
dence for the DNA loss hypothesis, it should pass yet
another test. The advent of modern comparative meth-
ods indicates that interspeciﬁc data cannot be analyzed
using regular statistical methods because of phyloge-
netic non-independence, which can increase (often sub-
stantially) the probability of type-I errors [8–10].
Evolution of Genome Size: A Phylogenetic Test
of the DNA Loss Hypothesis*
M. R. Pie
Universidade Federal do Paraná, Departamento de Zoologia, Curitiba, PR, 81531 Brazil;
Received March 27, 2006
—It has been recently suggested that the C-value paradox, the lack of an obvious association between
organismal complexity and genome size, can result simply from biases in insertion and deletion rates—the DNA
loss hypothesis. This hypothesis has been heavily criticized, particularly because its evidence, a negative relation-
ship between genome size and DNA loss rate, is based on a highly selective use of the available data. In this study
it is shown that the even the most favorable interpretation of the data favoring the DNA loss hypothesis is largely
an artifact of phylogenetic nonindependence, supporting the assertion made by other authors that the mechanisms
underlying genome size evolution might be more complex than envisioned by the DNA loss hypothesis.
* The text was submitted by the author in English.
Hypothesis of phylogenetic relationships between the taxa
used in the present study for the independent contrasts anal-
yses. Branch lengths are calibrated using various sources:
Human-mice divergence time (Hedges, 2002);
(Powell and DeSalle, 1995; Russo et al.,
1995); Spiny pufferfish-smooth pufferfish (Brainerd et al.,
2001; Tyler and Santini, 2002; Neafsey and Palumbi,
2003); Human-fish divergence time (Hedges, 2002); Origin
of nematodes (Nei et al., 2001); fly-grasshopper divergence
(Gaunt and Miles, 2002); Base of tree (Hedges, 2002);
(Labandera, 1994) (given that no molecular
estimate of this divergence was available, the date of the
earliest Acridid fossil was used instead).