How the Horned Lizard
Got Its Horns
Kevin V. Young,
Edmund D. Brodie Jr.,
Edmund D. Brodie III
Many descriptions of evolutionary adapta-
tions are criticized as “just-so stories” (1) that
are based more on intuition than on direct
tests of adaptive hypotheses. The elaborate
crowns of horns possessed by many species
of horned lizards (genus Phrynosoma) are
classic examples of intuitively adaptive fea-
tures that lack direct tests of function. The
bony horns that give horned lizards their
name are presumed to function as a defense
against predators (Fig. 1B). Here we present
data from the wild showing that natural se-
lection by loggerhead shrikes favors longer
horns (fig. S1) in the flat-tailed horned lizard
Predation is difficult to document in the
wild. Some predators, however, leave behind
explicit records of individual predation events
that can be exploited to assay natural selection.
Loggerhead shrikes (Lanius ludovicianus) often
impale their prey onto thorns, twigs, and even
barbed wire as a means of subduing their quarry
(2). When shrikes attack horned lizards, they
typically spear the lizard through the neck and
pull off the soft tissue. What remains is a record
of the successful shrike predation attempts
marked by desiccated skulls of horned lizards
hanging in trees and bushes (Fig. 1A).
We quantified selection (3, 4) on relative
horn lengths of flat-tailed horned lizards by
comparing the skulls (n ⫽ 29) of shrike-killed
lizards with the heads of live lizards (n ⫽ 155).
Our results showed predation by loggerhead
shrikes generated selection that favored longer
parietal and squamosal horns (Fig. 1, C and D).
The average parietal horn length of live horned
lizards was 10.0% longer (x ⫾ SE : 9.65 ⫾ 0.01
mm) than that of shrike-killed lizards (8.77 ⫾
0.21 mm), and the average squamosal horn
length was 10.4% greater in live lizards
(24.28 ⫾ 0.21 mm) than in those killed by
shrikes (21.99 ⫾ 0.49 mm). Visualization of the
selection function indicates that both traits ex-
perience positive directional selection with
threshold lengths above which predation is rare
or absent. Standardized selection gradients
[measured in standard deviation units (3)] sug-
gest that selection is stronger on the length of
squamosal (␤⬘ ⫽ 0.0945; P ⫽ 0.007) than on
the length of parietal horns (␤⬘ ⫽ 0.0549; P ⫽
0.055). These magnitudes of selection are less
than the median observed in most selection stud-
ies (␤⬘ ⫽ 0.15) (5) but nonetheless indicate that
constant selection with moderate heritability (0.5)
of horn length would change squamosal and pa-
rietal horn lengths a full standard deviation in 21
and 36 generations, respectively.
Modern methods for analyzing natural se-
lection have increased our understanding of
which traits experience selection (6). These
methods, however, typically cannot identify
agents of selection or reveal the functional re-
lations that result in natural selection (3). Even
most classic data sets demonstrating selection
in the wild, including Bumpus’s sparrows (7)
and Lande and Arnold’s pentatomid bugs (8),
did not reveal the agents responsible for the
observed patterns of survival. Our results
present a rare opportunity to link the statistical
form of selection to an identifiable agent, in this
case predation by shrikes. Our study does not
show that other agents and forms of selection
do not play a role in the evolution of horn
size, but clearly illustrates that defense
against shrike predation is one factor driving
the radical elongation of horns in some spe-
cies of horned lizards.
References and Notes
1. R. Kipling, Just So Stories (Doubleday, New York,
2. R. Yosef, Evol. Ecol. 6, 527 (1992).
3. E. D. Brodie III, A. J. Moore, F. J. Janzen, Trends Ecol.
Evol. 10, 313 (1995).
4. Materials and methods are available as supplemental
material on Science Online.
5. J. M. Hoekstra et al., Proc. Natl. Acad. Sci. U.S.A. 98,
6. J. G. Kingsolver et al., Am. Nat. 157, 245 (2001).
7. H. C. Bumpus, Biol. Lect. Woods Hole Mar. Biol. Sta.
6, 209 (1899).
8. R. Lande, S. J. Arnold, Evolution 37, 1210 (1983).
9. Funded by the Department of Defense Legacy Re-
source Management Program through the U.S. Ma-
rine Corps, Marine Corps Air Station, Yuma, and
administered by the Southwest Division Naval Facil-
ities Engineering Command, Natural Resources
Branch, and by the Bureau of Reclamation. Fieldwork
was facilitated by P. Cutler, W. Fisher, B. Morrill, R.
Palmer, R. Pearce, and A. Young.
Supporting Online Material
Materials and Methods
17 December 2003; accepted 11 February 2004
Department of Biology, Utah State University, Logan,
UT 84322–5305, USA.
Department of Biology, Indi-
ana University, Bloomington, IN 47405–3700, USA
*To whom correspondence should be addressed. E-
Fig. 1. (A) Flat-tailed horned lizard skull and dorsal skin impaled on a branch. [Photo, E. D. Brodie
Jr.] (B) Live ﬂat-tailed horned lizard in defensive posture. [Photo, K. V. Young] The live lizard in this
photo had unhealed wounds anterior to the rear legs, consistent with an unsuccessful attack by a
predator. Selection surfaces showing relations between survival probability and (C) relative parietal
horn length and (D) relative squamosal horn length. Bars show means and 95% conﬁdence intervals
for shrike-killed and live lizards.
www.sciencemag.org SCIENCE VOL 304 2 APRIL 2004 65