Removal of a reproductive organ greatly increases

Overcoming an evolutionary conflict: Removal of
a reproductive organ greatly increases
locomotor performance
Margarita Ramos*, Duncan J. Irschick*†, and Terry E. Christenson‡
*Department of Ecology and Evolutionary Biology, 310 Dinwiddie Hall, and ‡Department of Psychology, 2007 Stern Hall, Tulane University, New Orleans,
LA 70118
One potential consequence of sexual size dimorphism is conflict
among characters. For example, a structure evolved for reproduction can impair performance during other activities (e.g., locomotion). Here we provide quantitative evidence for an animal overcoming an evolutionary conflict generated by differential scaling
and sexual size dimorphism by obligatorily removing an undamaged reproductive organ, and thus dramatically enhancing its
locomotor performance. The spider genus Tidarren (Araneae,
Theridiidae) is interesting because, within several species presenting extreme sexual size dimorphism (males representing ⬇1% of
the total mass of the female), males voluntarily remove one of their
two disproportionately large pedipalps (modified copulatory organs; a single one represents ⬇10% of the body mass in an adult)
before achieving sexual maturity. Whether the left or right pedipalp is removed appears to be random. Previous researchers have
hypothesized that pedipalp removal might enhance locomotor
performance, a prediction that has remained untested. We found
that, for male Tidarren sisyphoides, maximum speed increased
(44%) significantly and endurance increased (63%) significantly
after pedipalp removal. Furthermore, spiders with one pedipalp
moved ⬇300% greater distances before exhaustion and had a
higher survival after exertion than those with two pedipalps.
Removal of the pedipalp may have evolved in male Tidarren
because of enhanced abilities to search for females (higher endurance and survival after exertion) and to out-compete rival males on
the female’s web (higher maximum speed). Our data also highlight
how the evolution of conflicts can result in the evolution of a novel
central tenet of optimality theory is that natural selection
will optimize structure and performance resulting in an
overall highly fit organism (1, 2). Many studies, however, have
shown that the evolution of high performance in one task can
lead to decreased performance during another task (e.g., refs.
3–5). In extreme cases of such apparent conflicts, a structure
evolved for one activity can substantially impair performance
during another activity. For example, Darwin (6) described how
sexual selection for sex differences might lead to such functional
conflicts, particularly in males. He depicted how the relatively
elaborate feathers in some male birds result in enhanced reproductive success via female mate choice, yet also reduces or
constrains flight capacities, thus potentially making the animal
more susceptible to predators (7). A relatively unexplored area
is how organisms cope with these constraints imposed by factors
such as sexual selection, natural selection, or allometry. One
view of constraints is that they limit or hinder morphological or
behavioral change, but another possibility is that constraints can
result in a novel phenotype or behavior (8, 9). For example, as
plethodontid salamanders undergo evolutionary miniaturization
(become smaller), they shift from a terrestrial to an arboreal
lifestyle (10). This transition occurs because of developmental
constraints on limb growth, resulting in increased toe webbing,
and hence enhanced climbing ability in small salamanders.
Here, we examine a functional conflict imposed by differential
scaling and extreme sexual size dimorphism in a spider, and
describe how a novel behavior overcomes this conflict. We
demonstrate that male spiders gain a substantial locomotor
performance advantage through the removal of a structure that
is normally essential for reproduction (a pedipalp). Biologists
have shown that the removal of a structure occurs in a variety of
taxa [lizards and tail loss (11, 12); harvestmen and leg loss (13);
damselfly larvae and loss of lamellae (14)]. However, these
examples are cases where animals remove structures facultatively in relatively specific situations. Here, we document a case
in which all males of the species remove a reproductive organ in
an obligatory manner, thus representing a case of dramatic
evolutionary change in behavior.
Although sexual size dimorphism is common in spiders (15–
17), it is extreme in the cobweb spider, Tidarren (Araneae,
Theridiidae) (18–20). In this genus, sexual size dimorphism
results from a decrease in male size (19, 21, 22) and an increase
in female size (20). The male pedipalps (copulatory organs) are
considered large in relation to male body size (refs. 21, 23, and
24, but see ref. 25) (Fig. 1). Vollrath (21) has suggested that the
pedipalp is large to enable mechanical coupling with the reproductive organ of the relatively large female. In other words, as
males decreased in body size, their pedipalps did not decrease in
size at the same rate (unpublished data). In most spiders, males
use as copulatory organs a pair of modified appendages disconnected from the gonads. After loading or inducting them with
sperm, males search for females. During copulation, males
generally use both pedipalps in an alternating fashion to inseminate the paired spermathecae in the female. In Tidarren,
however, one pedipalp is removed (either left or right pedipalp,
seemingly at random, see ref. 26) before sexual maturation (20,
22, 25, 26), which has also been described for another spider of
similar size and closely related to Tidarren (Echinotheridion, ref.
27). Just after molting to the penultimate instar, the male secures
one of its pedipalps to a silk scaffold and then twists it off by
turning in circles and pushing the bulb with the third and fourth
pairs of legs (26).
Previous authors have suggested that having two extremely
large pedipalps presents a functional constraint (23), and that the
removal of one pedipalp would enhance locomotion. Because
adult male spiders actively move about in search of females
(28–31) and participate in often intense intermale competition
(29, 32), any decrement in locomotor ability would have important negative consequences for their reproductive fitness. This
proposed explanation for the origin of the pedipalp removal
behavior has remained untested; no studies have examined the
potential functional consequences of locomotion with these
large pedipalps.
whom correspondence should be addressed. E-mail: [email protected]
© 2004 by The National Academy of Sciences of the USA
PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14 兩 4883– 4887
Communicated by Thomas W. Schoener, University of California, Davis, CA, January 15, 2004 (received for review February 26, 2003)
Materials and Methods
Subjects. Male T. sisyphoides emerge from the egg sac in the
second instar and attain adulthood in three subsequent molts
(M.R., personal observation). Thirteen adult individuals and 10
egg sacs were collected at the F. Edward Hebert Center of
Tulane University in Belle Chase, LA, and maintained in the
laboratory at room temperature (all locomotion trials also took
place at 22°C). The 956 resulting spiderlings were housed
individually in 20-ml glass scintillation vials with several nylon
strands for web support. They were fed small mealworms (⬇5
mm long), wingless Drosophila flies, and newborn crickets, and
were spritzed with water twice a week. A total of 31 males in the
fourth instar (penultimate) and 38 males in the fifth instar
(adult) were used in this study.
Quantification of Pedipalp Mass. The male spiders’ mean weights
were estimated by weighing 15 adult specimens simultaneously.
The weight of the pedipalp was estimated by calculating the
volume of the different body segments of three adult male
spiders by approximating them to the closest geometric figure.
The bodies were divided into cephalothorax (cylinder), abdomen
(sphere), legs (cone), and pedipalp (ellipse), and then each
segment was photographed from two different angles and
Maximum Speed. Sixteen penultimate instar spiders were ob-
Fig. 1. A male T. sisyphoides before (A) and after (B) removing a pedipalp.
Note the pedipalps overlap in the two-pedipalp condition (A), whereas the
one pedipalp is carried in a central position after pedipalp removal (B). (The
scale bars represent 1 mm.)
In this study, we had several objectives. We first attempted to
measure the size of Tidarren pedipalps to estimate the loss of
mass after pedipalp removal. Second, we tested the hypotheses
that removal of the pedipalp by male Tidarren sisyphoides would
result in an increase in maximum speed (potentially linked to
intermale competition) and endurance (potentially linked to
enhanced mate searching abilities and intermale competition).
Third, we examined the effects of pedipalp removal on survivorship after the above taxing endurance trials. We hypothesized
that individual male spiders that have removed their pedipalp
will be significantly less likely to die after endurance trials
compared to males that have not yet removed their pedipalp. If
the presence of the two palps causes high mortality in males, then
pedipalp removal may thus provide a fitness advantage, and
could provide some insight into why this genus has evolved the
palp-removal behavior. To test these hypotheses, we measured
maximum speed in individual males before and after pedipalp
removal, measured the endurance of spiders with one or two
pedipalps by chasing them until exhaustion, and measured
survivorship after the endurance trials. Finally, we accounted for
the potential confounding variables of the molting process and
practice effects by measuring maximum speed several times in
two groups of spiders.
4884 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0400324101
served at least once before and after pedipalp removal. Pedipalp
removal usually occurs within a few hours of the molt to this
instar. Individual spiders were filmed while moving on a strand
of tightly stretched silk (from a virgin adult female) the observer
had placed between two 15-cm-long sticks placed 5 cm apart. A
small ruler (7 cm long) was glued to a second set of sticks parallel
to the first. A Redlake high-speed motionscope PCI camera (60
frames per second) filmed all locomotion and a PC digitally
recorded all trials. The spiders were removed from the vial by
using a brush, and were placed on one of the sticks. The trial
began once the animal moved onto the silk, and the trial ended
when it reached the second stick. Maximum speed was defined
as the maximum instantaneous velocity achieved (quantified by
using PEAK PERFORMANCE software) by an animal during the
trial while it was chased with a microcapillary tube. We analyzed
only locomotion in which the spider moved continuously to
ensure that we examined steady-state locomotion.
Potential Confounding Variables: Hardening of the Exoskeleton, Experience, and Instar Differences. After emerging from their old
exoskeleton, spiders are soft and relatively immobile (30). We
therefore filmed eight adult males to examine the effects of
hardening of the exoskeleton on maximum speed. Each spider
was run 20, 60, 120, 180, 240, and 300 min after molting. To
control for a practice effect in a manner such that there were no
interactions with the effects of a recent molt, we collected 13
adults in the field from female webs and filmed them twice (1-h
rest between trials) following the previously outlined procedure.
These males were filmed at least 5 days after molting, because
the process of sperm induction takes several days after the last
molt. Finally, to demonstrate that speed did not differ according
to instar, we compared maximum speed in penultimate instar
(n ⫽ 16) and adult (n ⫽ 8) males bearing only one pedipalp.
Endurance. Time to exhaustion was measured in 15 penultimate
instar subjects bearing two pedipalps (different from the subjects
used to measure maximum speed) and 17 adult subjects with one
pedipalp. The same spiders could not be examined under both
one- and two-pedipalp conditions because the endurance trials
were extremely taxing for the animals. Arthropods experience a
high rate of water loss for a period after the molt (33). All
Ramos et al.
Fig. 3. Box plot of maximum speed in a group of spiders (n ⫽ 16) before and
after pedipalp removal. The bottom of the box represents the 25th percentile,
the dotted line represents the mean, the solid line represents the median, the
top of the box represents the 75th percentile, and the whiskers represent the
5th and 95th percentile.
animals therefore were tested 4 h after molting to minimize a
potential confounding effect of water loss. Subjects were chased
with a small paintbrush on a sheet of paper until exhaustion, that
is, when the spider did not respond to 10 consecutive touches.
During the second minute of each test, the animal’s path was
traced by following the spider with a pencil to estimate the
distance moved during that minute, and to provide an estimate
of the average running speed (distance兾time) during that
minute. We quantified only the second minute because handling
and placement of the spider on the test arena precluded an
accurate trace of the path during the first minute of each test and
because tracing the entire path would have resulted in too
complex a record (375–1,930 s in range). The track was measured
by laying a piece of string over the path and measuring its total
length with a ruler. Mean speed per path was calculated by
dividing this total distance by 60 s. SYSTAT (SPSS Institute,
Chicago) was used for statistical analysis, and SIGMAPLOT was
used for creating graphs (SPSS Institute).
Endurance. For the endurance trials, spiders with two pedipalps
were able to move for an average of 17 min 30 s ⫾ 55 s, whereas
spiders with one pedipalp were able to move for an average of
28 min 30 s ⫾ 45 s (Fig. 4; two-sample t test, t ⫽ 9.3, df ⫽ 30,
P ⬍ 0.001). Thus, spiders with one pedipalp had endurance
capacities that were ⬇63% greater than spiders with two pedipalps. In addition, spiders with one pedipalp moved significantly
faster (1.4 ⫾ 0.04 cm兾s) than those with two pedipalps (0.8 ⫾
0.03 cm兾s) (t ⫽ 11.5, df ⫽ 24, P ⬍ 0.001). Assuming constant
speeds throughout the duration of the tests, we estimated that
Quantification of Pedipalp Mass. We estimate the adult male
pedipalp to account on average for 10% of body mass (Fig. 1).
Tidarren males have an average body length of 1.4 mm and
weight of 7.1 ⫻ 10⫺4 g, whereas females are on average 6 mm in
length and weigh 5.8 ⫻ 10⫺2 g (Fig. 2).
Maximum Speed. Pedipalp removal significantly increased maxi-
mum speed (Fig. 3) from an average of 2.7 ⫾ 0.2 (SE) cm兾s to
3.8 ⫾ 0.3 cm兾s (paired t test, t ⫽ 4.41, df ⫽ 15, P ⬍ 0.001). Thus,
spiders ran 44% faster after they had removed one pedipalp than
when they still possessed both pedipalps. We used analysis of
covariance (ANCOVA) to test whether the potentially confounding process of exoskeleton hardening (time after molt)
affected maximum speed. Individual spiders did not differ in
their slope between maximum speed (dependent variable) and
Ramos et al.
Fig. 4. Box plot of time until exhaustion in two different groups of spiders
with two (n ⫽ 15) and one (n ⫽ 17) pedipalp, respectively. The bottom of the
box represents the 25th percentile, the dotted line represents the mean, the
solid line represents the median, the top of the box represents the 75th
percentile, and the whiskers represent the 5th and 95th percentile.
PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14 兩 4885
Fig. 2. Male and female T. sisyphoides in copula. The minute male (indicated
by the arrow) on the female’s ventrum is ⬇1% of the female’s mass. (The scale
bar represents 1 mm.)
time since molting (independent variable) (homogeneity of
slopes test; F7,31 ⫽ 0.65, P ⬎ 0.50), and time since molting also
did not affect maximum speed (F1,38 ⫽ 0.26, P ⬎ 0.50). A linear
least-squares regression of the first (independent variable) and
second (dependent variable) runs also did not reveal a significant
relationship between them (F1,11 ⫽ 1.2, r ⫽ 0.31, df ⫽ 13, P ⬎
0.25, slope ⫽ 0.38). Thus, how spiders performed on their first
run did not predict how they would perform on their second run.
In addition, we did not find a significant difference in maximum
speed between penultimate instar and adult males with a single
pedipalp (two sample t test, t ⫽ 0.26, df ⫽ 22, P ⬎ 0.1).
animals with one pedipalp moved an average of 2,435 ⫾ 118 cm
or 17,393 body lengths (based on a mean body length of 1.4 mm
taken from ref. 23) before exhaustion, whereas those with two
pedipalps moved and average of 822 ⫾ 61 cm or 5,870 body
lengths. In other words, spiders with one pedipalp moved on
average 296% greater distances than those with two pedipalps.
Of 11 spiders that died immediately after an endurance trial,
significantly more spiders bore two pedipalps (eight individuals,
or 53% of the total number of spiders measured bearing two
pedipalps) than one pedipalp (two individuals, or 12% of the
total number of spiders measured bearing one pedipalp) (Fishers’ exact test, P ⫽ 0.021). A multiple regression using survival
(1 ⫽ spider died, 2 ⫽ spider lived) as the dependent variable and
endurance and average speed during the second minute as the
independent variables was statistically significant (F2,23 ⫽ 7.09,
P ⬍ 0.01), but endurance was the only statistically significant
(and positive) predictor of survival (P ⫽ 0.028, standardized
coefficient ⫽ 0.77). Thus, spiders with high endurance capacities
tended to more often survive after endurance trials. In addition,
those individuals with two pedipalps that survived the endurance
trial (n ⫽ 4) took up to 5 days to remove the pedipalp (mean ⫽
3.0 ⫾ 0.9 days), although this process typically takes place within
a few hours after the molt.
One potential consequence of extreme sexual size dimorphism is
conflict among characters (e.g., refs. 6 and 7). A central issue in
evolutionary biology concerns the evolution of how organisms
cope with such conflicts. In T. sisyphoides, male body size has
been reduced (19–22), whereas the reproductive organs might
not have changed evolutionarily in an equal manner (21). The
apparent cost of such conflict resulting from this differential
scaling is a relatively large pair of pedipalps (Fig. 1) that
substantially impair both maximum speed and endurance. Remarkably, rather than evolving one pedipalp to overcome this
functional conflict, male Tidarren have evolved the behavior of
pedipalp removal. In this regard, our data provide evidence for
an organism overcoming an evolutionary conflict by evolving a
novel behavior.
Which proximate factors could explain why Tidarren have
greater maximum speed and endurance with one compared to
two pedipalps? One explanation is that the observed performance difference is not caused by removal of the pedipalp, but
rather to gross anatomical, physiological, or behavioral changes
in male spiders as they molt from the fourth (penultimate) to the
fifth (adult) instar. We consider this possibility unlikely. First,
male size and overall morphology change very little between the
fourth and fifth instars in Tidarren (C. Linn, personal communication), and we found no difference in maximum speed
between spiders bearing one pedipalp in these two instars. In
fact, relative pedipalp size increases dramatically (⬇370%) from
the third to the fourth instar, whereas it increases only slightly
from the fourth to the fifth instar (⬇2%) (C. Linn, personal
communication). Therefore, the two pedipalps become a hindrance only after the molt into the fourth instar, and this might
explain why the pedipalp is not removed earlier. In terms of
physiological changes, spiders produce hormones (e.g., ecdysone) that play a vital role in the molting process (29). However,
we are not aware of any hormones that would restructure the
physiological system to provide the dramatic changes in performance that we observed after pedipalp removal. Moreover, the
oxygen transport system of spiders is not known to undergo
dramatic changes between instars (29). Finally, it is unlikely that
any differences in general natural history and behavior between
individuals in the fourth and fifth instars could explain the
observed performance increase after pedipalp removal, because
our subjects were in a period of relative quiescence, between the
4886 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0400324101
molting and pedipalp removal (fourth instar) or sperm induction
(fifth instar) events.
A more likely possibility to explain the large difference in
locomotor performance between spiders bearing one or two
pedipalps is that the presence of both pedipalps might mechanically interfere with locomotion, such as impairing limb movement or by dragging on the locomotor surface (Fig. 1). Locomotion would become more energetically expensive for spiders
moving long distances, and maximum speed would be reduced.
The structure of male spider pedipalps is complex, and dragging
when moving on flat surfaces could damage pedipalpal components needed for coupling with female genitalia. Another nonexclusive possibility is that the loss of mass caused by removal of
the pedipalp enables Tidarren to move both faster and longer
relative to their body size. Tidarren males are very small, so the
effects of gravity on both horizontal and vertical locomotion are
minor compared to larger animals (34, 35). Loss of mass will
likely have a lesser impact on the locomotion of a smaller animal
compared to the effect of a proportional loss in mass on a larger
animal. Ideally, this latter hypothesis would most profitably be
tested by adding or subtracting weights equaling the mass of a
pedipalp, but adding such a small mass would be extremely
difficult. However, further experiments that examine the position of the pedipalps during locomotion with one and two
pedipalps might differentiate among these hypotheses. In addition, pedipalp removal does not seem to have significant proximal costs for the spider. For instance, the animals twist in circles
to remove the pedipalp, thus effectively sealing the wound and
apparently preventing excessive fluid loss and infection.
The endurance capacities of male Tidarren are remarkably
high for a spider (36–42) and are comparable to many mammal
species (43, 44). This high endurance is especially intriguing in
the light of the lack of a sophisticated circulatory system in
spiders that would enable high rates of oxygen exchange, such as
found in birds and mammals (45). More remarkably, Tidarren
males ran for very long distances at relatively fast speeds, which
should lead to high levels of lactate in their tissues caused by
anaerobic processes (39, 40). Indeed, in cases where the spider
died after the endurance trails, the most likely cause was the
build-up of lactate (39, 40). The fact that males with two
pedipalps died significantly more often than males with one
indicates that long movements with two pedipalps are especially
The negative effect of pedipalp size on both aerobic and
anaerobic locomotion could have profound implications for the
reproductive ecology of Tidarren. Several lines of evidence
suggest that Tidarren males move long distances (relative to their
size). First, a large body of literature shows that male spiders, on
reaching adulthood, shift from being sedentary to actively
searching for females. Male spiders frequently travel long distances, often vertically, to find a mate, because females can be
dispersed widely (31, 46). Second, some simple considerations
show that Tidarren males may be forced to move long distances
because of their very small size. For example, if a Tidarren male
moves only 2 meters, that distance is equivalent to ⬇1,400 body
lengths, which is similar to a 3-m tetrapod (e.g., an alligator)
moving ⬇4.3 km. Added to this is intense intermale competition
that might result in males often moving from female to female
as is the case in other species of spiders. If Tidarren males do
indeed move long distances, then the significantly lower mortality of one-pedipalp males compared to two-pedipalp males
after endurance trials may provide an immediate fitness advantage for palp-removal. However, to test this idea rigorously, one
would need detailed field data on both movement distances and
speeds of Tidarren males during mate searching.
Another context in which locomotion might be important for
Tidarren is during scramble competition for a female (T.E.C.,
C. Linn, J. DuFatta, M.R., J. Feldman, and T. Bukowski,
Ramos et al.
unpublished data). For example, once a male has located a
female, maximum speed may be especially important for outcompeting rival males (47–50) on the female’s web. Up to 25
male T. sisyphoides have been observed at one time on a single,
unrestrained penultimate instar female’s web (T.E.C., C. Linn,
J. DuFatta, M.R., J. Feldman, and T. Bukowski, unpublished
data). Furthermore, in staged encounters with a newly molted
female, the first male to reach the female’s abdomen usually
copulates. Endurance may also be important for the prolonged
efforts expended attempting to dislodge a rival male in copula
(T.E.C., C. Linn, J. DuFatta, M.R., J. Feldman, and T. Bukowski,
unpublished data).
It is possible that ancestral male Tidarren spiders that first
exhibited pedipalp removal were at a selective advantage relative
to males that did not because of the enhanced locomotor
performance that a one-pedipalp condition provides. However,
one must also consider the possibility that the behavior of
pedipalp removal has evolved for reasons more related to
reproduction. For instance, as Tidarren and Echinotheridion
males die in copula (Tidarren cuneolatum, Tidarren argo, T.
sisyphoides, Echinotheridion gibberosum: refs. 25, 22, 20, and 27,
respectively), they can insert one pedipalp only once and insem-
inate only one of the female’s two spermathecae. Consequently,
removal of a pedipalp might result in conservation of gametes
because only one pedipalp is used in copulation. If both pedipalps were filled and only one used for copulation, the sperm in
the unused palp would be wasted. Males could simply leave one
palp unfilled at sperm induction. The fact that they do not
suggests that there are benefits to palp removal, as noted earlier.
Finally, it is not clear why Tidarren males have not evolved a
single-pedipalp condition, which would obviate the need for
removing a seemingly functional organ. Pedipalps might play an
important role in other functions during the early instars, such
as in prey handling. In addition, the genetic traits underlying the
development of two pedipalps may be more difficult to overcome
than simply evolving a behavior of pedipalp removal. Thus, data
that shed light on the developmental genetics of pedipalps in
spiders would be especially interesting within this evolutionary
1. Rosen, R. (1967) Optimality Principles in Biology (Plenum, New York).
2. Stephens, D. W. & Krebs, J. R. (1986) Foraging Theory (Princeton Univ. Press,
3. Huey, R. B. & Hertz, P. E. (1984) J. Exp. Biol. 110, 113–123.
4. Macrini, T. E. & Irschick D. J. (1998) Biol. J. Linn. Soc. 63, 579–591.
5. Vanhooydonck, B., Van Damme, R. & Aerts, P. (2001) Evolution (Lawrence,
Kans.) 55, 1040–1048.
6. Darwin, C. (1896) The Descent of Man and Selection in Relation to Sex
(Appleton and Company, New York), 2nd Ed.
7. Andersson, M. (1982) Nature 299, 818–820.
8. Rashevsky, N. (1960) Mathematical Biophysics: Physico-Mathematical Foundations of Biology (Dover, New York), 3rd Ed.
9. Gould, S. J. (2002) The Structure of Evolutionary Theory (Belknap, Cambridge,
10. Wake, D. B. (1991) Am. Nat. 138, 543–567.
11. Dial, B. E. & Fitzpatrick, L. C. (1984) Anim. Behav. 32, 301–302.
12. Daniels, C. B. (1983) Herpetologica 39, 162–165.
13. Guffey, C. (1999) Can. J. Zool.兾Rev. Can. Zool. 77, 824–830.
14. Stoks, R. (1999) Behav. Ecol. Sociobiol. 47, 70–75.
15. Elgar, M. (1991) Evolution (Lawrence, Kans.) 45, 444–448.
16. Prenter, J., Elwood, R. W. & Montgomery, W. I. (1998) Proc. R. Soc. London
B 265, 57–62.
17. Prenter, J., Elwood, R. W. & Montgomery, W. I. (1999) Evolution (Lawrence,
Kans.) 53, 1987–1994.
18. Levi, H. (1955) N.Y. Entomol. Soc. 63, 59–81.
19. Hormiga, G., Coddington, J. A. & Scharff, N. (2000) Sys. Biol. 49, 435–462.
20. Knoflach, B. & Benjamin, S. P. (2003) J. Arachnol. 31, 445–448.
21. Vollrath, F. (1998) TREE 13, 159–163.
22. Knoflach, B. & van Harten, A. (2001) J. Zool. Lond. 254, 449–459.
23. Chamberlain, R. V. & Ivie, W. (1934) Bull. Univ. Utah Biol. Ser. 24, 3–18.
24. Eberhard, W. G. (1985) Sexual Selection and Animal Genitalia (Harvard Univ.
Press, London).
25. Knoflach, B. & van Harten, A. (2000) J. Nat. Hist. 34, 1639–1659.
26. Branch, J. H. (1949) Bull. South. Cal. Acad. Sci. 41, 139–140.
27. Knoflach, B. (2002) in European Arachnology 2000, eds. Toft, S. & Sharff, N.
(Aarhus Univ. Press, Aarhus, Denmark), pp. 139–144.
28. Givens, R. P. (1978) Ecology 59, 309–321.
29. Foelix, R. F. (1996) Biology of Spiders (Oxford Univ. Press, Oxford).
30. Frameneau, V., Dieterich, M., Reich, M. & Plachter, H. (1996) Rev. Suisse
Zool. Hors serie, 223–234.
31. Moya-Laraño, J., Halaj, J. & Wise, D. H. (2002) Evolution (Lawrence, Kans.)
56, 420–425.
32. Robinson, M. H. (1982) Ann. Rev. Entomol. 27, 1–20.
33. Gnatzy, W. & Romer, F. (1984) in Biology of the Integument: Invertebrates, eds.
Bereiter-Hahn, J., Matolsy, A. G. & Richards, K. S. (Springer, Berlin), Vol. I,
pp. 638–684.
34. Huey, R. B. & Hertz, P. E. (1984) Evolution (Lawrence, Kans.) 38, 441–444.
35. Full, R. J. & Tullis, A. (1990) J. Exp. Biol. 149, 307–317.
36. Cloudsley-Thompson, J. L. (1957) Biol. J. Linn. Soc. (Zool.) 43, 134–152.
37. Wilson, R. S. (1970) Z. Morphol. Tiere. 68, 308–322.
38. Anderson, J. F. & Prestwich, K. N. (1985) J. Comp. Physiol. B 155, 529–539.
39. Prestwich, K. N. (1983) Physiol. Zool. 56, 112–121.
40. Prestwich, K. N. (1983) Physiol. Zool. 56, 122–132.
41. Prestwich, K. N. (1988) J. Comp. Physiol. B 158, 437–447.
42. Bromhall, C. (1987) J. Comp. Physiol. B 157, 451–460.
43. Taylor, C. R., Heglund, N. C., McMahon, T. A. & Looney, T. R. (1980) J. Exp.
Biol. 86, 9–18.
44. Alexander R. M. N. & Goldspink, G., eds. (1977) Mechanics and Energetics of
Animal Locomotion. (London, Chapman and Hall).
45. Schmidt-Nielsen, K. (1990) Animal Physiology (Cambridge Univ. Press, Cambridge, U.K.), 4th Ed.
46. Legrand R. S. & Morse, D. H. (2000) Biol. J. Linn. Soc. 71, 643–664.
47. Ghiselin, M. T. (1974) The Economy of Nature and the Evolution of Sex (Univ.
of California Press, Berkeley).
48. Partridge, L., Ewing, A. & Chandler, A. (1987) Anim. Behav. 35, 555–562.
49. Zamudio, K. R. (1998) Evolution (Lawrence, Kans.) 52, 1821–1833.
50. Able, D. J. (1999) Behav. Ecol. Sociobiol. 46, 423–428.
Ramos et al.
PNAS 兩 April 6, 2004 兩 vol. 101 兩 no. 14 兩 4887
We thank T. Sherry, Anthony Herrel, Bieke VanHooydonck, and three
anonymous reviewers for helpful comments on previous versions of this
manuscript. This work was supported by National Science Foundation
Grant IBN 9983003 (to D.J.I.).