Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Herpetologists' League is collaborating with JSTOR to digitize, preserve and extend access to Herpetologica.
http://www.jstor.org
ABSTRACT: Horned lizards,genus Phrynosoma, show a range in dietary habits, such that some
species are more strongly myrmecophagous (ant-eating) than others. The relationship between
skeletalmorphologyand dietary specializationwas examinedby correlationand stepwise regression
analyses.Five of seven morphologicalvariablesfrom the mandibleand epipterygoidare significantly
correlatedwith the mean percentage of ants in the diet. Epipterygoidlength, height of the coronoid
process, vertical diameter of the mandible, and length of the mandible posteriorto the coronoid
process are all negatively correlated; length of the tooth row is positively correlated. Stepwise
regressionanalysisselected epipterygoid length as the single best predictorof diet in Phrynosoma.
The best two-variablemodel includes epipterygoid length and vertical diameter of the mandible,
and has an R2= 0.8377.
Key words: Lacertilia;Phrynosoma; Morphology;Foraging; Diet; Ants
HORNED lizards of the genus Phryno- some species being more strongly myr-
soma are primarily ant-eating reptiles, mecophagous than others. Those species
whose dietary habitsare well known (Mon- with a comparatively lower percentage of
tanucci, 1981; Pianka and Parker, 1975, ants in the diet consume a greater variety
and references cited therein; Powell and of terrestrialinsects and other arthropods.
Russell, 1984; Rissing, 1981; Turner and When the mandibles of various species of
Medica, 1982). The apparent relationship Phrynosoma were compared, I noted ap-
between dentitional morphology and diet parent differences in robustnessand avail-
in Phrynosoma was examined by Hotton able area (on the coronoid and posterior
(1955) as part of a broader study of the to it) for muscle attachment. Furthermore,
adaptive relationships of dentition to diet the epipterygoids, which are thought to
in North American iguanids. He examined brace the palatal complex of the reptilian
three species (P. cornutum, P. douglassii, skull (Romer, 1956), are reduced or absent
and P. platyrhinos), concluding that they in Phrynosoma (Axtell, 1986; Presch,
were all similar in having blunt, non-cus- 1969). These observations led me to hy-
pate, peg-like teeth, and that their diet pothesize that in highly myrmecophagous
consisted almost entirely of ants. Presch species of horned lizards, the mandible
(1969) reiteratedHotton'sobservations,but should be relatively slender, area for mus-
also noted that in three species (P. asio, P. cle attachment reduced, teeth reduced in
mcallii, and P. coronatum), the posterior size, and epipterygoids greatly reduced or
teeth show a hint of a tricuspid condition. absent. These anticipated relationshipsbe-
I could not confirm the difference between tween morphology and diet in Phrynoso-
anteriorand posteriorteeth in these species, ma are herein examined statistically.
but the dentitional morphology of all
members of the genus should be reviewed. METHODS AND MATERIALS
Pianka and Parker (1975) determined Mandiblesand epipterygoids from skel-
the percentage of ants (by number and etons of Phrynosoma were examined un-
volume) in the diet of each species of Phry- der a Wild M-5 dissecting microscope.The
nosoma. In reviewing Pianka and Parker's left or right element from each specimen
data, I concluded that the species of Phry- was randomly selected for measurements,
nosoma are not all alike in their feeding which were made with an ocular microm-
habits. Rather, there appears to be a range eter. Adult specimens were selected in or-
of dietary habits within the genus, with der to minimize ontogenetic variationwith-
208
with the proportionof ants in the diet were lare (10), x = 21.6, 19-23; P. platyrhinos
then subjected to stepwise regressionanal- (32), xf = 17.4, 13-27; P. solare (14), x =
ysis in order to confirm the anticipated 19.8, 17-23; P. taurus (4), x = 22.2, 21-
relationships between some of the inde- 24. In terms of actual numbers then, P.
pendent (morphological)variablesand the mcallii has the lowest average number of
dependent variable (percentage of ants). tooth positions and P. ditmarsi has the
The programused was the StatisticalAnal- highest average number. However, the lat-
ysis System (SAS) STEPWISE procedure ter value may be an artifact of small sam-
(SAS Institute Inc., 1985). Results of the ple size; note that four other species have
stepwise regression analysis allowed selec- limits that overlap those of P. ditmarsi.
tion of a model that best explains the re- Associationsamong the seven standard-
lationship between morphology and diet ized morphological variables and the di-
in Phrynosoma. etary percentage of ants are revealed by
correlation analysis (Table 2). Epiptery-
RESULTS goid length (EP*) shows a strong, positive
Descriptive statistics for the seven mor- correlationwith the height of the coronoid
phological variablesare presented in Table process (CP*), and it is less strongly cor-
1. Note that Phrynosoma ditmarsi is dis- related with the vertical diameter of the
tinctive, compared with the other species, mandible (MD*) and tooth height (TH*).
in having a very long epipterygoid (EP*), Epipterygoid length is also negatively cor-
extensive vertical expansion of the man- related with the length of the tooth row
dible (MD*), tall coronoid process (CP*), (TR*). The vertical diameter of the man-
and the greatest distance from the coro- dible (MD*) shows highly significant pos-
noid processto the articularcondyle (LC*). itive correlation with height of the coro-
Phrynosoma solare is notable in having the noid process (CP*)and with distance from
shortestteeth and the smallest mean value the coronoid process to the articular con-
for epipterygoid length. In some speci- dyle (LC*). Tooth height (TH*) is posi-
mens of P. solare, the elements are entirely tively correlated with height of the coro-
lacking. Phrynosoma solare, as well as P. noid process.Length of the tooth row (TR*)
modestum and P. mcallii, have the short- is negatively correlatedwith distance from
est coronoid process. The mandible is most the coronoid process to the articular con-
slender in P. modestum and P. solare, and dyle (LC*), and with diameter of the man-
the shortest distance from the coronoid to dible (MD*).
the articular process is seen in P. mcallii The percentage of ants in the diet shows
and P. platyrhinos. Representative man- very significant negative correlationswith
dibles from the 12 species of Phrynosoma epipterygoid length (EP*), height of the
are illustrated for comparison (Fig. 2). coronoid process (CP*), and vertical di-
The number of tooth positions in Table ameter of the mandible (MD*) and is neg-
1 represents standardized data. Thus, rel- atively correlated, though less so, with dis-
ative to jaw length, P. modestum has the tance between the coronoid process and
greatest number of teeth and P. asio has the articular condyle (LC*). Ant con-
the fewest teeth. The actual number of sumption is positively correlated with
tooth positions, based on a larger sample length of the tooth row (TR*).
(including left and right mandibles for Stepwise regression analysis of the data
some specimens), is given below. The was performed in order to obtain a model
species, sample size, mean, and observed that best predicts the diet of the species of
limits follow: P. asio (16), X = 20.8, 18-24; Phrynosoma. Two morphological vari-
P. braconnieri (5), X = 20.8, 15-23; P. cor- ables, TP* and TH*, were deleted from
nutum (31), X= 19.2,13-26; P. coronatum the analysis because of the lack of signif-
(34),X = 21.3, 16-27; P. ditmarsi (2), Xt = icant correlationswith ants in the diet. Sev-
25.5, 25-26; P. douglassii (27), X = 20.5, eral methods of model selection were em-
17-26; P. mcallii (13), X = 14.9, 11-18; P. ployed to find the variables among the
modestum(9), x = 23.2, 20-25; P. orbicu- remaining five that contribute most im-
asio 0.170 ? 0.004 0.059 ? 0.001 0.320 ? 0.006 0.538 ? 0.006 0.123 ? 0.003 0.171
0.158-0.196 0.050-0.067 0.302-0.354 0.500-0.560 0.107-0.135 0.15
braconnieri 0.258 ? 0.025 0.052 ? 0.003 0.342 ? 0.002 0.511 ? 0.017 0.127 ? 0.003 0.143
0.182-0.289 0.042-0.060 0.337-0.349 0.463-0.540 0.120-0.135 0.10
cornutum 0.190 ? 0.006 0.057 ? 0.001 0.338 ? 0.003 0.518 ? 0.004 0.115 ? 0.002 0.182
0.130-0.260 0.043-0.070 0.313-0.365 0.489-0.560 0.097-0.137 0.16
coronatum 0.200 ? 0.005 0.063 ? 0.001 0.325 ? 0.002 0.525 ? 0.003 0.120 ? 0.003 0.150
0.151-0.240 0.051-0.079 0.303-0.343 0.500-0.556 0.088-0.142 0.12
ditmarsi 0.224 0.062 0.353 0.491 0.146 0
douglassii 0.194 ? 0.007 0.072 ? 0.001 0.338 ? 0.006 0.531 ? 0.006 0.133 ? 0.002 0.177
0.145-0.250 0.059-0.084 0.298-0.385 0.496-0.566 0.114-0.150 0.15
mcallii 0.192 ? 0.005 0.067 ? 0.001 0.304 ? 0.005 0.533 ? 0.005 0.099 ? 0.005 0.143
0.164-0.211 0.061-0.074 0.280-0.329 0.506-0.554 0.082-0.121 0.13
modestum 0.276 ? 0.032 0.047 ? 0.001 0.321 ? 0.007 0.538 ? 0.006 0.084 ? 0.002 0.123
0.149-0.333 0.042-0.052 0.302-0.346 0.512-0.552 0.076-0.092 0.10
orbiculare 0.196 ? 0.008 0.073 ? 0.005 0.330 ? 0.005 0.530 ? 0.002 0.130 ? 0.007 0.173
0.165-0.211 0.053-0.086 0.316-0.345 0.523-0.538 0.115-0.156 0.15
platyrhinos 0.206 ? 0.008 0.056 ? 0.001 0.304 ? 0.004 0.520 ? 0.003 0.108 ? 0.003 0.130
0.152-0.300 0.042-0.070 0.258-0.330 0.490-0.552 0.073-0.126 0.11
solare 0.208 ? 0.008 0.031 ? 0.002 0.316 ? 0.008 0.539 ? 0.006 0.085 ? 0.002 0.129
0.163-0.234 0.021-0.040 0.285-0.351 0.516-0.574 0.070-0.092 0.11
taurus 0.207 ? 0.015 0.058 ? 0.002 0.325 ? 0.013 0.533 ? 0.018 0.127 ? 0.001 0.152
0.192-0.223 0.056-0.060 0.312-0.339 0.514-0.552 0.126-0.128 0.14
ii Ui u u r nri
a U v | 1-m1 " | | uuriii
I a .
onG S~~~~~~~~~~~~~~~~~~
g ~n ~~ H
~ ~ ~ ~ ~I | ~~~~1
| |. I N I 11 II II It Ii U II
FIG. 2.-Mandibles of Phrynosoma: (A) P. asio (AMNH 72636); (B) P. braconnieri (AMNH 90833); (C)
P. cornutum (AMNH77117);(D) P. coronatum (AMNH99685);(E) P. ditmarsi (UAZ 35511); (F) P. douglassii
(AMNH 70639); (G) P. mcallii (AMNH 77042); (H) P. modestum (AMNH 74499); (I) P. orbiculare (RRM
2379); (J) P. platyrhinos (RRM 2323); (K) P. solare (RRM 2324); (L) P. taurus (CAS 141361). Scale is in
millimeters.
TABLE 2.-Pearson correlation coefficients (below the diagonal) and the probabilities (above the diagonal)
that the observed r values are significantly different from zero. See Table 1 for abbreviations.
portantly to the predictive power of the MD*, and CP*, has an R2of 0.8435, which
model. is not much higher than the R2 value of
The forward selection procedure added the best two-variable model. The equation
variables one by one to the model until no for the regressionline in the best two-vari-
remaining variable produced a significant able model is Y = -1.205(EP*) -
F statistic. Two variables, EP* and MD*, 207.544(MD*) + 122.683. The results of
entered the model at the 0.10 level of sig- the three model selection procedures are
nificance. The first variable which entered identical and are summarized in Table 3.
the model was EP*, with an R2 = 0.7555.
The backward elimination procedure DISCUSSION
deleted variables(all were initially entered Significantintercorrelationsexist among
into the model) one by one until all vari- some of the morphological variables, as
ables remaining in the model produced F previously mentioned (Table 2). Epipter-
statisticssignificant at the 0.10 level. Three ygoid length (EP*), which is the single best
variables were removed in the following predictor of diet in Phrynosoma, shows
sequence: TR*, LC*, and CP*, and only strong positive correlation with height of
EP* and MD* were retained. The final R12 the coronoid process (CP*), and it is also
for the two-variable model, using either correlated,though less strongly, with tooth
the backward elimination or forward-se- height (TH*) and diameter of the man-
lection procedures, was R2 = 0.8377. dible (MD*). Tooth height is not signifi-
The maximum R2 improvement pro- cantly correlated with ants, but variables
cedure was used to find the best one-vari- CP* and MD* are. The strong intercor-
able model, the best two-variable model, relation of EP* and CP* apparently pre-
and so forth. This procedure selected vari- cluded the entry of the latter variable into
able EP* as having the highest R12. The the models obtained in the stepwise regres-
second variable added to the model was sion analysis.In other words, CP* was con-
MD*, as it produced the greatest increase tributing very little additional variance to
in R2= 0.8377 (which is the same as in the the model not already explained by EP*.
previously mentioned procedures). Under Note the small increase in R2 between the
the maximum R2improvement procedure, best two-variable and the best three-vari-
other variables could be substituted after able models obtained by the maximum R2
comparisons with the two variables al- improvement procedure. This assertion is
ready in the model. The comparisonswould also supported in that when EP* was de-
determine if removing one variable and leted from the analysis, CP* was chosen
replacing it with another increased the R2. by stepwise regressionas the variable with
Such comparisons did not produce an in- the highest R2 (= 0.7121). A two-variable
crease in the R2, and the best two-variable model with CP* and MD* had an R2 =
model, with EP* and MD*, stands. The 0.7514, which is substantially less than R2
best three-variablemodel, containing EP*, = 0.8377 of the previously obtained two-
of the epipterygoids would be required to items were treated as though they were
brace the pterygoids when prey items are ants, and such attempts to ingest prey were
being subdued and crushed. The vertical usually unsuccessful. The prey was simply
diameter of the mandible (MD*)and tooth too large to be held by the sticky tongue
height (TH*) are also positively correlated without assistance from the jaws. Even-
with the height of the coronoid process, as tually, some lizards learned to catch the
would be expected. The variable MD* is large prey by using the jaws. Phrynosoma
a measure of the robustness of the man- solare also has the smallest teeth of any
dible, and TH* may reflect the grasping member of the genus (Table 1). The small
and masticating capabilities of the teeth. size of the teeth is probably a consequence
The distance from the coronoid process to of diet, although there is no significantcor-
the articular condyle (LC*) is negatively relation between tooth height (TH*) and
correlated with the length of the tooth row ant consumption across species (Table 2).
(TR*). A slight anterior shift of the coro- The remaining species appear to be less
noid process would diminish TR*, but in- morphologically specialized for ant-eat-
crease LC*, thus providing more area for ing. Phrynosoma ditmarsi appears to con-
muscle attachment. Two muscles, the M. sume the lowest proportion of ants (Table
pseudotemporalis profundus and the M. 1), but comparatively little is known of its
adductor mandibularis posterior, insert feeding habits in nature. Roth (1971, per-
onto this area (Jenkinsand Tanner, 1968). sonal communication) found weevils, oth-
A slight anterior shift of the coronoid er beetles, true bugs, grasshoppers,spiders,
process would perhaps also enhance the and scorpions, in addition to ants, in the
lever action of the mandibles by moving stomachs of specimens. Phrynosoma dit-
the main point of muscle pull (coronoid marsi has enormous vertical expansion of
process) closer to the area bearing the the mandible, a tall coronoid process,more
weight (TR*), and farther from the ful- extensive area posteriorto that process,and
crum (articularcondyle). Note that TR* is a long epipterygoid. The teeth are not ex-
also negatively correlated with MD*, CP*, ceptionally tall; four other species surpass
and EP*. P. ditmarsi in relative tooth height. How-
Specialization for myrmecophagy in ever, the crowns of the teeth are quite
Phrynosoma appears to have involved the flattened. This suite of morphological fea-
following morphological changes: (1) a re- tures suggests greater crushing power in
duction or loss of the epipterygoid; (2) re- P. ditmarsi compared with other Phry-
duction of the coronoid process; (3) re- nosoma. Behavioralobservationslend sup-
duction in the diameter of the mandibular port to this idea. P. ditmarsi is more effi-
ramus; (4) reduction in the area posterior cient in capturing large prey than is P.
to the coronoid process, and a concomitant douglassii or P. orbiculare (Montanucci,
increase in the length of the tooth row, unpublished data). The two latter species
possibly accomplished by a slight shift in frequently beat the prey against the sub-
the position of the coronoid. Such mor- strate with lateral movements of the head,
phological changes have evolved, in var- apparently to subdue it; P. ditmarsi sel-
ious combinations,in Phrynosoma mcallii, dom, if ever, does this. Moreover, P. dit-
P. modestum, P. platyrhinos, and most marsi masticates its prey somewhat more
notably P. solare. Phrynosoma solare has than do the other two species. Although P.
the highest consumption of ants compared ditmarsi may be regarded as a dietary gen-
with other members of the genus (Table eralist, its morphology suggests specializa-
1). The specialization for myrmecophagy tion for eating large, chitinous arthropods.
is corroborated not only by morphology I speculate that such prey items may be
and dietary information,but by behavioral periodicallyabundantand importantin the
observationsas well. When fed large prey diet of this horned lizard, but field studies
items such as crickets, adult P. solare at- are needed to provide concrete data on
tempted to catch the insects by extruding various aspects of the natural history of
the tongue. Regardless of size, the prey this rare and interesting species.
Acknowledgments.-I thank the following cura- MCDOWELL, S. B., JR., AND C. M. BOGERT. 1954.
tors and museum staff for the loan of skeletal speci- The systematic position of Lanthanotus and the
mens used in this study (museumabbreviationsfollow affinities of the anguinomorphan lizards. Bull. Am.
Leviton et al., 1980): G. W. Foley, R. G. Zweifel Mus. Nat. Hist. 105:1-142.
(AMNH);R. C. Drewes, A. E. Leviton, J. V. Vindum, MONTANUCCI, R. R. 1979. Notes on systematics of
L. Wishmeyer(CAS);R. F. Inger, H. Marx,A. Resetar horned lizards allied to Phrynosoma orbiculare
(FMNH); W. E. Duellman, S. K. Dutta, D. R. Frost (Lacertilia: Iguanidae). Herpetologica 35:116-124.
(KU);R. L. Bezy, J. W. Wright (LACM);P. Alberch, . 1981. Habitat separation between Phry-
J. P. Rosado (MCZ);D. A. Good, H. W. Greene, D. nosoma douglassi and P. orbiculare (Lacertilia:
B. Wake (MVZ);C. H. Lowe (UAZ); D. M. Harris, Iguanidae) in Mexico. Copeia 1981:147-153.
A. G. Kluge(UMMZ);R. W. McDiarmid,A. D. Yoder, PIANKA, E. R., AND W. S. PARKER. 1975. Ecology
G. R. Zug (USNM);J. A. Campbell (UTA);T. R. Van of horned lizards: A review with special reference
Devender (TRV, personal collection). My personal to Phrynosomaplatyrhinos.Copeia 1975:141-162.
specimens (RRM)used in this study will be deposited POWELL, G. L., AND A. P. RUSSELL. 1984. The diet
in the LACM. Computer time and facilities were of the eastern short-horned lizard (Phrynosoma
made available through the Clemson University douglassi brevirostre)in Alberta and its relation-
Computer Center; William C. Bridges,Jr. and Louis ship to sexual size dimorphism. Can. J. Sci. 62:428-
Deluca kindlyprovidedtechnicalassistanceand help- 440.
ful discussions. PRESCH, W. 1969. Evolutionary osteology and re-
lationships of the horned lizard genus Phrynosoma
LITERATURE CITED (family Iguanidae). Copeia 1969:250-275.
RISSING, S. W. 1981. Prey preferences in the desert
AXTELL, R. W. 1986. The epipterygoids in Phry- horned lizard: Influence of prey foraging method
nosoma solare, still present, but disappearing. J. and aggressive behavior. Ecology 62:1031-1040.
Herpetol. 20:79-81. ROMER, A. S. 1956. Osteology of the Reptiles. Uni-
CAMP, C. L. 1923. Classification of the lizards. Bull.
versity of Chicago Press, Chicago.
Am. Mus. Nat. Hist. 48:289-481. ROTH, V. D. 1971. Food habits of Ditmars' horned
HOTTON, N., III. 1955. A survey of adaptive rela-
lizard with speculations on its type locality. J. Ar-
tionships of dentition to diet in the North American izona Acad. Sci. 6:278-281.
Iguanidae. Am. Midl. Nat. 53:88-114. SAS INSTITUTE INC. 1985. SAS User's Guide: Sta-
JENKINS, R. L., AND W. W. TANNER. 1968. Os-
tistics, 1985 ed. SAS Institute, Cary, North Carolina.
teology and myology of Phrynosoma p. platyrhinos TURNER, F. B., AND P. A. MEDICA. 1982. The dis-
Girard and Phrynosoma d. hernandesi Girard. tribution and abundance of the flat-tailed horned
Brigham Young Univ. Sci. Bull. Biol. Ser. 9(4):1- lizard (Phrynosomamcallii). Copeia 1982:815-823.
34.
LEVITON, A. E., R. McDIARMID, S. MOODY, M. Accepted: 22 August 1988
NICKERSON, J. ROSADO, 0. SOKOL, AND H. VORIS. Associate Editor: William Cooper, Jr.
1980. Museum acronyms-Second edition. Her-
petol. Rev. 11:93-102.